Epidemiology of dementia – With particular focus on time trends and methodology Hanna Wetterberg Department of Psychiatry and Neurochemistry, Institute of Neuroscience and Physiology at Sahlgrenska Academy University of Gothenburg Gothenburg, Sweden, 2023 Cover and illustrations by Ida Nordigårds Epidemiology of dementia – With particular focus on time trends and methodology © Hanna Wetterberg 2023 hanna.wetterberg@gu.se hanna.wetterberg@gmail.se ISBN: 978-91-8069-199-4 (PRINT) ISBN: 978-91-8069-200-7 (PDF) http://hdl.handle.net/2077/74511 Printed in Borås, Sweden 2023 Printed by Stema Specialtryck AB Till Oliver iv Abstract Abstract ABSTRACT Dementia is a clinical syndrome characterised by deterioration in cognitive Dfuenmcteionntisa, iws hai cchli nciacuasl essy npderrosomnea lc hsuafrfaecrtienrgis eadn db ys odceietetarilo crhatailolenn gine sc. oSgtnuidtiivees finuvnecsttioignast,i nwg hthiceh i nccaiudseensc ep,e prsroevnaalle nsucfef,e arnindg m anordt asloitcyi eotfa dl ecmhaelnletniag aerse. Snteueddieeds finovr etshtieg autnindge rtshtea nindcinidge nocfe t, hper esvoaclieentacle ,a anndd emcoonrtoamlitiyc obfu drdemene notfi at haree dnieseeadseed. Efopri dtheme iuonlodgericsatal nsdtuindgie os fo tfh ed esmoceinettiaal afancde emcoenthoomdiocl obguircdael nc hoafl lethneg edsi.s eTahseis. Ethpeisdise meioxlaomgiicnaeld stuthdeie s tiomf ed emtreenndtisa fianc e dmemetehnotdiao loegpicidael mchioallolegnyg esa.m Tohnigs othcetosigse neaxraiamnsin, eadn tahgee tgirmoeu p trreanpdidsl y inin cdreeamsienngt iaa nde piwdhemerieo lodgeym eanmtiao nigs coocmtomgeonna.r iaMnse,t haond oalogeg icgarl ocuopn sriadpeirdaltyio ninsc, resuascihn ga sa nsdel ewcthioenre bdiaesm aenndti at hies icmompamcto onf. tMhee tchhoodicoelo ogfi cdaila gcnoonsstiidce troaotilosn, sw, esreu cahls oa ss tusedlieecdt.i oDna tbai auss eadn idn tthhee pimappearcst owf atsh ed cehrioviecde offr odmia gnthoes tipco tpouollast,i owne-rbea aslesdo sGtuodtiheedn. bDuartga uHs7ed0 inB itrhthe Cpaopheorrst wSatsu ddieesr,i vaendd frtohme Pthroe sppeocptuivlea tioPno-pbualsaetdio nG oStthuednyb uorfg WHo7m0 eBni rtinh CGoohthoertn bSutrugd, iSews, edaennd. Tthhee cPorhoosrptse citnivcleu dPeodp wuelartei obno rnS tiund 1y 90o1f- 0W2, o1m92e3n- 2i4n, aGnodt h1e9n3b0u. rFgi,n Sdwinegdse fnr.o Tmh eP acopheor rIt ss hinocwluedde dth wate prea rbtiocripna innt s1 9in0 1g-e0n2e, r1a9l, 2h3a-d2 4a, alonwde 1r 9p3r0e.v Falienndcien gosf fdriosmor dPearps,e hr igI hsehro ewdeudc atthioatn apla lretviceilp, aanntds wine gree nmeroarle, hoaftde na mlowarerrie pdr ethvaanle nrecfeu osaf lsd.i sTohrdereer sw, heirgeh feerw eedru dciaftfieorneanlc leesv einl, acnodm wpaerries omno wreit ohf tthene tmaragrerite pdo tphuanla trieofnu soafl ss. aTmhee-raeg ewde rine dfievwideura dlsi fifne rGenoctehse ninb ucorgm. pIna rPisaopne wr iIthI, twhee ftaorugnedt pthoaptu tlhatei odnia gonf ossatmice c-arigteedri ai nind iIvCidDua-1ls0 i nyi eGldoetdh ethneb ulorgw. eIsnt pPraepvealre nIIc,e w oef dfoeumnedn tthiaa ta nthde IdCiaDg-n1o1s ttihc ec rhitigehrieas itn, fIoCllDow-1e0d y bieyl dtehde tDheS Mlow-5e. sTt hpere avgarleenemcee onft bdeemtweenetnia t haen dD ISCMD-5-1 a1n dth IeC hDig-1h1e swt,a fso slulobwsteadn tbiayl . tIhne P DapSMer- I5I. IT, whee faogurenedm tehnatt tbheetw seuernv tivhael DtiSmMe- 5i nacnrde aIsCeDd -1b1o wtha si nsu bthstoasnet iawl. iIthn Paanpde rw IiItIh,o wuet fdoeumnde nthtiaat. Dtheem seunrtviaiv wala st itmhee minocsrte iamsepdo rtbaontth p riend itchtoors eo f wdietaht ha innd b owthit hcoouhto rdtse.m Leansttliya,. Din ePmaepnetria I Vw,a sw teh feo munodst a i mdepcorertaasnetd p prreedviaclteonrc oef o dfe daethm ienn btioa taht caogehso 8rt5s .a Lnads 8tl8y., Win eP aalpsoe rf oIVun, dw ae dfoecurnedas ae dine ctrheea fsoedu rp-yreevara liennccidee onfc de eomf ednemtiae antt iaag. eTsh 8e5 f ainnddi n8g8s. Wfroem al stoh ifso tuhnedsi sa pdreocrveidasee i inns itghhet fso iunrt-oy ethare itnimcidee tnrceen dosf dine mtheen etipai.d Tehmei ofilnodgiyn ogsf dfreomme nthtiias, tahse wsies llp aros vinidtoe iinmspigohrttsa nint taos pthecet sti mofe m treetnhdosd oinlo tghiec aelp ciodnemsidioelroagtiyo nosf idne mdeemnteina,t iaas rweseelal racsh i.n to important aspects of methodological considerations in dementia research. KKeeyywwoordrsd s KDemywenotriad, sE pidemiology, Selection Bias, Mortality, Time trends Dementia, Epidemiology, Selection Bias, Mortality, Time trends SAMMANFATTNING PÅ SVENSKA Demens är ett paraplybegrepp för många olika sjukdomar som har gemensamt att de försämrar den kognitiva förmågan. I takt med den ökande medellivslängden ökar antalet personer med demens, vilket kommer sätta prov på samhällets förmåga att möta dessa behov. Under det senaste decenniet har ett mönster av minskande nyinsjuknande (incidens) och förekomst (prevalens) av demens framkommit. Det är ännu oklart om den här minskningen även går att se i åldersgruppen 85–90-åringar. Därför ville vi undersöka om incidensen och prevalensen av demens har förändrats i den här åldersgruppen. Det är också oklart om ökningen i medellivslängd även omfattar personer med demens. Vi undersökte därför också om mortaliteten i relation till demens hade förändrats. Resultat från populationsbaserade studier om demens kan påverkas av selektivt bortfall, det vill säga om personer som tackar nej skiljer sig från deltagarna. Därför undersökte vi hur representativa deltagarna i våra studier var. Hur diagnosticeringen av demens görs har också betydelse för resultaten av studier på demens. De kriterier som vanligtvis används är baserade på olika upplagor av the International Classification of Diseases (ICD) och Diagnostic and Statistical Manual of Mental Disorders (DSM). Vi ville undersöka vilken påverkan valet av dessa har för beräkningen av prevalens av demens. Data som användes i den här avhandlingen kommer från den populations- baserade H70-studien och Kvinnostudien i Göteborg. Deltagarna som ingår i den här avhandlingen kommer från tre kohorter födda 1901-02, 1923-24 och 1930. Vi fann att deltagarna i jämförelse med de som tackade nej hade lägre prevalens av olika sjukdomar, högre utbildningsnivå och oftare var gifta. Det var färre skillnader mellan deltagarna och den totala populationen av jämnåriga göteborgare. Vi såg också att det diagnostiska kriteriet ICD-10 resulterade i lägst prevalens av demens, medan ICD-11 och DSM-5 gav de högsta. Överensstämmelsen mellan de två nyaste kriterierna ICD-11 och DSM-5 var hög. Överlevnadstiden från 85-års ålder ökade både hos dem med och utan demens, och demens var den viktigaste faktorn kopplad till död i båda kohorterna. Prevalensen av demens vid 85- och 88-års ålder och incidensen mellan 85-89 år minskade. Sammanfattningsvis bidrar den här avhandlingen med ökad kunskap om tidstrender i demensepidemiologi och metodfrågor gällande detta, så som selektivt bortfall och vikten av val av diagnostiska kriterier. LIST OF PAPERS List of papers This thesis is based on the following papers, referred to in the text by their Roman numerals. I . Wetterberg H*, Rydén L*, Ahlner F, Falk Erhag H, Gudmundsson P, Guo X, Joas E, Johansson L, Kern S, Mellqvist Fässberg M, Najar J, Ribbe M, Rydberg Sterner T, Samuelsson J, Sacuiu S, Sigström R, Skoog J, Waern M, Zettergren A, Skoog I. Representativeness in population-based studies of older adults: five waves of cross-sectional examinations in the Gothenburg H70 Birth Cohort Study. BMJ Open 2022;12:e068165. *HW and LR are joint first authors I I . Wetterberg H, Najar J, Rydberg Sterner T, Skoog I. The effect of diagnostic criteria on dementia prevalence – A population-based study from Gothenburg, Sweden. Submitted. I I I . Wetterberg H, Najar J, Rydén L, Ribbe M, Rydberg Sterner T, Zettergren A, Guo X, Falk Erhag H, Sacuiu S, Kern S, Skoog I. Dementia remains the major predictor of death among octogenarians. A study of two population cohorts of 85-year-olds examined 22 years apart. European Journal of Epidemiology. 2021;36(5):507-17. IV. Wetterberg H, Najar J, Rydberg Sterner T, Rydén L, Falk Erhag H, Sacuiu S, Kern S, Zettergren A, Skoog I. Decreasing incidence and prevalence of dementia among octogenarians. A population-based study on three cohorts born 30 years apart. The Journals of Gerontology: Series A, 2023; glad071. TABLE OF CONTENT Page 01 INTRODUCTION 01 Dementia 03 Alzheimer’s disease 05 Vascular dementia 06 Other dementia subtypes 07 Mixed dementia 07 Risk factors for dementia 08 Diagnostic criteria 13 Epidemiology 16 Epidemiology of dementia 17 Potential biases in epidemiological studies of dementia 18 02 AIMS 21 Specific aims of the included papers 23 Paper I 23 Paper II 23 Paper III 24 Paper IV 24 03 METHODS AND MATERIALS 27 Participants 29 Birth cohort 1901-02 30 Birth cohort 1923-24 31 Birth cohort 1930 32 Study populations by paper 33 Variables and outcome measures 34 Neuropsychiatric examinations and key informant interviews 35 Dementia diagnosis 37 Register data 41 Data analyses 42 Ethical considerations 46 Page04 MAIN RESULTS 49 Main results of Paper I 51 Main results of Paper II 52 Main results of Paper III 53 Main results of Paper IV 54 05 DISCUSSION 57 Methodological discussion 59 Selection bias 59 Measurement bias 61 Use of register data 63 Paper I 64 Paper II 65 Paper III 65 Paper IV 66 General discussion 66 Representativeness 67 Choice and use of diagnostic criteria 68 Mortality, incidence, and prevalence of dementia 70 Potential explanations for time trends in dementia epidemiology 75 0 6 CONCLUSION 79 Future Perspective 82 08 ACKNOWLEDGEMENT 85 09 REFERENCES 91 LIST OF F IGURES AND BOXES Figure 1. Disorders included in the umbrella term dementia 4 Figure 2. The Alzheimer’s disease pathological cascade 6 Figure 3. Conceptual diagram over the relationship between AD and VaD pathology and mixed dementia 8 Figure 4. Prevalence of dementia in men and women by age 9 Figure 5. Suggested mechanisms for promoting cognitive reserve and risk reduction 11 Figure 6. The bathtub analogy 16 Figure 7. The Gothenburg H70 Birth cohort studies 30-31 Figure 8. Workflow of setting dementia diagnoses according to the DSM-III-R 38 Figure 9. Figures over potential changes in survival with dementia 73 Figure 10. Different birth cohorts’ life courses in the view of important societal events 77 Box 1. Diagnostic criteria in the ICD-systems 14 Box 2. Diagnostic criteria in the DSM-systems 15 Box 3. Summary of the study designs in the papers 33 Box 4. Characteristics of key informant interviews 36 Box 5. The classification systems used in the papers 41 Box 6. Analyses used in Paper I 42 Box 7. Analyses used in Paper II 43 Box 8. Analyses used in Paper III 44 Box 9. Analyses used in Paper IV 45 Abbreviation ABBREVIATIONS AD Alzheimer’s disease VaD Vascular dementia DLB Lewy bodies DSM Diagnostic and Statistical Manual of Mental Disorders ICD International Classification of Diseases WHO World Health Organization OR Odds Ratio IRR Incidence Rate Ratio HR Hazard Ratio SD Standard Deviation CI Confidence Interval MMT Mini Mental Test ADL Activities of daily living iADL Instrumental activities of daily living ECG Electrocardiogram PAR Population Attributable Risk IPR National Inpatient Register CDREG Cause of death register Study names H70 The Gothenburg H70 Birth Cohort Study PPSW The Prospective Population Study of Women MRC CFAS the Cognitive Function and ageing study PAQUID French Personnes Agées Quid MoVIES Monongahela Valley Independent Elders Survey MYHAT Monongahela-Youghiogheny Healthy Ageing Team HRS Health and Retirement study KP The Swedish study Kungsholmen Project SNAC-K Swedish National Study on Aging and Care in Kungsholmen ABBREVI ATI O N 17 Definitions in short DEFINITIONS IN SHORT Cognitive domains Aphasia Inability to comprehend or formulate language due to damage to specific regions of the brain1 Apraxia Inability to plan and perform previously learned skills of common motorial movements1 Agnosia Inability to process sensory information, such as recognising people, objects, sounds, shapes or smells1 Executive function Cognitive functions that control complex, goal-directed thought and behaviour, such as working memory, flexible thinking, and planning1 Episodic memory Declarative memory consist of personal memory, in contrast to general knowledge1 Complex attention Ability to focus on multiple things at once and deliberately choose what to pay attention to2 Learning and Ability to record new information and then retrieve it2 memory Perceptual-motor Ability to coordinate the bodies’ movement in response function to what is happening2 Social cognition Ability to process and use information in social contexts, such as control impulses, express empathy, and recognize social cues and facial expressions2 Epidemiology Time trend A change that occurred over time, often slowly and observable only after a certain amount of time Birth cohort Persons classified by a particular year of birth1 Incidence The number of new cases of disease during a given period in a specified population1 Prevalence The total number of cases of a disease in a specified population at a designated time1 Mortality All deaths reported in a given population1 Representativeness The degree to which the characteristics of participants in a study are similar to the target population3 DEF IN IT IO NS I N SHO RT 1 01 INTRODUCTION I0n1troduction INTRODUCTION DDeemmeenntitaia The focus of this thesis is dementia, a syndrome currently affecting 57 million people worldwide3 and is globally the seventh leading cause of death.4 As a result of growing populations and increasing survival into high ages, the number of dementia cases is expected to increase to 153 million worldwide by the year 2050. Today in Sweden, around 130 to 150 000 people are estimated to have dementia.5 The prevalence increases with age and almost doubles every five years,6,7 from around 1.0-1.5% in 60-64-year-olds to 25- 39% in those older than 90.4,8-11 Dementia is considered one of the major causes of disability and dependency among older adults12 and has a major impact on the affected individuals and their relatives. Besides this, the syndrome is associated with high societal costs and a high burden of informal care. Globally, the yearly estimated cost is $1.3 trillion.4 Nearly 50% of the cost is accounted for by informal care provided by families. Women are disproportionally affected by dementia, with higher incidence, especially in populations older than 85,11,13,14 as well as more often providing informal care.13 In recent years, epidemiological studies of dementia have indicated a decline in dementia incidence in Western countries,15-20 and more and more manageable risk factors are being identified. A recent Lancet report showed that as much as 40% of dementia cases might be preventable.21 Being a major societal challenge for the future but also offering hopeful prospects of preventive strategies, the research of dementia distribution and determinants is crucial. Dementia is an umbrella term for a range of diseases that affect the brain with loss of vital cognitive functions, influencing mood and behaviour, resulting in impaired function in activities of daily living (ADL) and instrumental activities of daily living (IADL) (Figure 1). Memory is commonly affected, as well as the ability to plan and perform everyday activities, language, judgment, orientation, and perception of time. The symptoms, course of the disease, and underlying pathology vary depending on the type of dementia. The most common type of dementia is Alzheimer’s disease (AD), accounting for IaNrToRuOnDdU C6T0I-O8N0 % of all dementia cases, the second most common is vascula3r dementia (VaD), and the third most common is dementia with Lewy bodies (InDtrLoBdu)c.2t2io Tn he different diseases causing dementia are grouped into primar3y and secondary dementia. Figure 1 Disorders included in the umbrella term dementia Dementia is an umbrella term covering a range of disorders. The most common type of dementia is AD, the second most common VaD, and the third most common DLB. In addition to these types, there is a range of less common disorders. It is also common with mixed dementia, with more than one condition causing dementia simultaneously. Primary dementias are most often caused by progressive and degenerative neuronal loss. Symptoms of these diseases have, in general, a slow onset with a gradual worsening of cognitive function. Most cases of dementia are primary and include, for example, AD, DLB, frontotemporal dementia, Huntington’s disease, and Creutzfeldt-Jakob’s disease. Although VaD is a vascular disease, not a neurodegenerative condition, it is usually grouped together with primary dementias. Secondary dementia can occur as a result of a disease or injury. They are generally progressive, however, sometimes chronic or even reversible. Examples of conditions that can cause secondary dementia or symptoms resembling dementia are normal-pressure hydrocephalus, multiple sclerosis, 4 H A N N A W E T T E R B E R G well as the ability to plan and perform everyday activities, language, judgment, orientation, and perception of time. The symptoms, course of the disease, and underlying pathology vary depending on the type of dementia. The most common type of dementia is Alzheimer’s disease (AD), accounting for around 60-80% of all dementia cases, the second most common is vascular wdeemll eans ttihae ( VabaiDlit)y, taon dp lathne a tnhdi rpde rmforsmt c eovmermydoany iasc dtievmitiens,t ila nwgiuthag Le,e jwuyd gbmodeniets, o(DrieLnBt)a.t2i2o Tn,h aen di fpferrecnetp tdioisne aosef st icmaues. iTngh ed esmymenpttioam asr,e cgoruoruspee do fi nthtoe pdrisimeaasrey, and usencdoenrldyainryg dpeamtheonltoiga.y vary depending on the type of dementia. The most common type of dementia is Alzheimer’s disease (AD), accounting for around 60-80% of all dementia cases, the second most common is vascular dementia (VaD), and the third most common is dementia with Lewy bodies (DLB).22 The different diseases causing dementia are grouped into primary and secondary dementia. DEMENTIA Umbrella term for loss of memory and other thinking abilities severe enough to interfere with daily life. Alzheimer´s: Vascular Lewy Body Frontotemporal Others: 60-80% Dementia: Dementia: Dementia: Parkinson´s 5-10% Figure 1 Disorders included15 i-n30 t%he umbrell5a- 1t0e%rm dementia Huntington´s Dementia is an umbrella term covering a range of disorders. The most common type of dementia is AD, the second most common VaD, and the third most common DLB. In addition to these types, there is a range of less common disorders. It is also common with mixed dementia, with more than one conMdiixteido nde cmaeunstina:g dementia simultaneously. Dementia from more than one cause. Figure 1 Disorders included in the umbrella term dementia DPreimeanrtiya ids eamn eunmtibarse lla rtee rm ocsotv eorifntge na cranugsee dof bdyis oprdroergs.r eTshseiv me oasnt dco dmemgoen etyrpatei voef dneemuernotniaa lis l oAsDs., Sthyem spetcoonmds m oofs tth ceosme mdoisne aVsaeDs ,h aanvde ,t hine tgheirnde rmaol,s at csolomwm oonn sDeLt Bw. itIhn addition to these types, there is a range of less common disorders. It is also common with ma ixgerda ddeumale nwtiao, wrsiethn minogr e othfa nc oongen citoinvdei tiofun nccautisoinng. deMmoenstti a csaimseuslt anoefo udsleym. entia are primary and include, for example, AD, DLB, frontotemporal dementia, Huntington’s disease, and Creutzfeldt-Jakob’s disease. Although VaD is a Pvarsimcualrayr deismeaesnet,i ans oatr ea mnoeustr oodfetegnen cearuatsievde bcyo npdriotigorne,s siitv eis aunsdu adlelyg egnreoruatpievde ntoeguertohnearl wloisths. pSryimpartoy mdesm oef nthtiease. diseases have, in general, a slow onset with a gradual worsening of cognitive function. Most cases of dementia are pSericmonardya rayn de minecnlutdiae ,c afno ro cecxuarm apsl ea, rAeDsu,l t DoLf Ba , dfirsoeansteo toemr ipnojurarly . dTehmeeyn tairae, Hgeunnetrianllgyt onp’rso gdriessesaisvee, , anhdo wCerveeurt, zfseoldmt-eJtaimkoebs’ s cdhirsoenasice . oArl theovuengh rVevaeDrs iibsl ea. vEaxsacmulparle sd iosefa sceo,n ndoittio an sn ethuarto dceagne nceaurasteiv see condaitriyo nd,e mit eins tiuas uoarl lys ygmrpotuopmeds troesgemthbelri nwgi tdhe pmreimntairay adrem neonrmtiaasl.- p ressure hydrocephalus, multiple sclerosis, Secondary dementia can occur as a result of a disease or injury. They are g4 enerally progressive, however, sometimes chronic orH AeNvNeAn W EreTvTeErRsBibE RleG. Examples of conditions that can cause secondary dementia or symptoms resembling dementia are normal-pressure hydrocephalus, multiple sclerosis, 4 H A N N A W E T T E R B E R G 4 Hanna Wetterberg ddeepprreessssiioonn,, meettaabboolliicc ddiissoorrddeerrss ((ssuucchh aass vviittaamiinn B1122 ddeeffiicciieennccyy)),, hheeaadd iinnjjuurriieess,, bbrraaiinn ttuumoouurrss,, aanndd cchhrroonniicc aallccoohhooll aabbuussee..2233,,2244 Allzzzhheehiimeieemrr’’sse rddii'ssee daassiseee ase AD ttyyppiiccaallllyy hhaass aa sslloow oonnsseett ooff ssyymppttoomss,, uussuuaallllyy ssttaarrttiinngg wiitthh aa pprrooggrreessssiivvee lloossss ooff eeppiissooddiicc meemoorryy aanndd ddiiffffiiccuullttiieess iinn lleeaarrnniinngg nneew iinnffoorrmaattiioonn.. AD ppaattiieennttss ccoommoonnllyy rreeppeeaatt qquueessttiioonnss aanndd ccoonnvveerrssaattiioonnss..2255 Otthheerr ssiiggnnss ooff AD iinncclluuddee aapprraaxxiiaa,, aaggnnoossiiaa,, aanndd aapphhaassiiaa,, aass weellll aass iimppaaiirreedd jjuuddggmeenntt aanndd ddeecciissiioonn--maakkiinngg,, aanndd oorriieennttaattiioonn..2266 AD waass ffiirrsstt ddeessccrriibbeedd bbyy Drr.. Allooiiss Allzzhheeiimeerr iinn 11990077,, whhoo hhaadd iiddeennttiiffiieedd tthhee ccoombbiinnaattiioonn ooff ccooggnniittiivvee ssyymppttoomss wiitthh sseenniillee ppllaaqquueess aanndd nneeuurrooffiibbrriillllaarryy ttaanngglleess..2277 Afftteerr oovveerr 111155 yyeeaarrss,, tthhiiss iiss ssttiillll ccoonnssiiddeerreedd ttoo bbee tthhee hhaallllmaarrkk ooff tthhee ddiisseeaassee..2288 Thhee sseenniillee ppllaaqquueess aarree ddeeppoossiittss ooff tthhee pprrootteeiinn aamyyllooiidd bbeettaa ((Aββ)),, whhiicchh aaccccuumuullaatteess oouuttssiiddee aanndd bbeettweeeenn tthhee nneeuurroonnss.. Thhee nneeuurrooffiibbrriillllaarryy ttaanngglleess aarree hhyyppeerrpphhoosspphhoorryyllaatteedd ttaauu pprrootteeiinn tthhaatt aaggggrreeggaatteess aanndd aaccccuumuullaatteess wiitthhiinn tthhee nneeuurroonnss.. Thhiiss sspprreeaadd pprrooggrreessssiivveellyy,, uullttiimaatteellyy lleeaaddiinngg ttoo ssyynnaappttiicc aanndd nneeuurroonnaall lloossss,, ccaauussiinngg tthhee bbrraaiinn ttoo sshhrriinnkk..2266 Thhiiss ppaatthhoollooggyy ssttaarrttss maannyy yyeeaarrss,, eevveenn ddeeccaaddeess,, pprriioorr ttoo tthhee cclliinniiccaall ssyymppttoomss aappppeeaarr ((FFiigguurree 22))..2299 Whhaatt ccaauusseess tthhee ddiisseeaassee iiss ssttiillll nnoott kknnoownn,, aanndd tthheerree aarree sseevveerraall ssuuggggeesstteedd tthheeoorriieess,, ssuucchh aass tthhee aamyyllooiidd ccaassccaaddee hhyyppootthheessiiss,, tthhee cchhoolliinneerrggiicc hhyyppootthheessiiss,, ggeenneettiicc ssuusscceeppttiibbiilliittyy,, aacccceelleerraatteedd aaggeeiinngg,, nneeuurrooiinnffllaammaattiioonn aanndd iimmuunnee ddyyssrreegguullaattiioonn,, ssyynnaappttiicc ddyyssffuunnccttiioonn,, nneeuurroovvaassccuullaarr ddyyssffuunnccttiioonn,, tthhee miittoocchhoonnddrriiaall ccaassccaaddee hhyyppootthheessiiss,, aanndd eennvviirroonnmeennttaall rriisskk ffaaccttoorrss..3300 Onnee ooff tthhee moosstt ddoomiinnaanntt tthheeoorriieess ttooddaayy iiss tthhee aamyyllooiidd ccaassccaaddee hhyyppootthheessiiss..3311 Thhee aamyyllooiidd ccaassccaaddee hhyyppootthheessiiss ssuuggggeessttss tthhaatt aabbnnoorrmaall ddeeppoossiittiioonn ooff Aββ iinniittiiaatteess aa sseeqquueennccee ooff eevveennttss.. Whheenn tthhee pprroodduuccttiioonn aanndd cclleeaarraannccee ooff Aββ aarree iimbbaallaanncceedd,, tthhee pprrootteeiinn aaggggrreeggaatteess iinn tthhee bbrraaiinn,, aanndd tthhee ffoorrmaattiioonn ooff oolliiggoomeerrss aaffffeeccttss nneeuurroonnaall aanndd ssyynnaappttiicc ffuunnccttiioonnss iinn tthhee bbrraaiinn..2266 Aββ ffoorrmss ppllaaqquueess ggrraadduuaallllyy,, whhiicchh aaccttiivvaattee miiccrroogglliiaa aanndd aassttrrooccyytteess,, ccaauussiinngg aann iinnffllaammaattoorryy rreessppoonnssee.. Thhee ttooxxiicc aaccccuumuullaattiioonn ooff Aββ aallssoo iinndduucceess tthhee hhyyppeerrpphhoosspphhoorryyllaattiioonn aanndd aaggggrreeggaattiioonn ooff ttaauu..2266 Hooweevveerr,, aass AD hhaass aa llaarrggee hheetteerrooggeenneeiittyy iinn rreeggaarrdd ttoo cclliinniiccaall maanniiffeessttaattiioonn,, ppaatthhoollooggyy,, aanndd ddiisseeaassee progression, there might be multiple pathways that cause the disease.32 AD is often classified by the age of symptom onset as either early-onset AD (<65 years) or late-onset AD (>65 years). The disease is often more aggressive in early-onset AD, with a faster progression and more severe pathology. Among II NNTTRROthosO DDUCT I ON 5e wUCitThI O Nla te-onset AD, the severity varies, and sometimes the level of5 pInatrthodoulcotgioyn i n AD patients can even be the same as in cognitively unimpaired5 controls.33 Figure 2 The Alzheimer’s disease pathological cascade. Measurable changes in biomarkers for AD, shown as the thin lines, start many years prior to the clinic appearance of cognitive impairment, shown by the light-green zone. The high and low-risk borders illustrate differentiated risks for cognitive impairment, indicating that two individuals with the same biomarker profile can present with different cognitive levels. Source: Jack et al., 2013.29 Vascular dementia The main feature of VaD is cerebrovascular pathology damaging the brain and impairing cognition.34 There are many different underlying vascular pathologies of VaD, such as large infarcts and haemorrhages, and small vessel disease.34,35 The research field of VaD has expanded during the past decades, moving from the hypothesis that only large cortical infarcts caused dementia to the understanding that there are many potential causes.35 In conjunction with this, vascular cognitive impairment (VCI) was suggested as a broader term since cerebrovascular effects impact the brain in more ways than previously appreciated.35,36 It is, for example, common with related vascular pathology in AD cases, and its contribution to dementia with other 6 H A N N A W E T T E R B E R G pprrooggrreessssiioonn,, tthheerree miigghhtt bbee muullttiippllee ppaatthhwaayyss tthhaatt ccaauussee tthhee ddiisseeaassee..3322 AD iiss oopffrttoeegnnr eccsllasaissossniiff,ii eetddh ebbryey mtthhieeg haatgg eeb eoo fmf ssuyylmti ppltteoo pma toohnnwssaeeytts a atssh eaetiitt hchaeeurr seeaa trrhllyye-- oodnnissseeeatts eA.3D2 A ((e 66o55f yyseyeamarrssp))t..o Tmhh eeo nddsiisseeeta aasssee eiissit hooeffttree enna rmlyoo-roreen asaeggtgg rrAeeDssssi ivv(ttee6rr5 pp ryrooeaggrrsee)ss.ss Tiioohnne aa dnnidds e masooerr eeis ss oeevvfteeerrnee ppmaaotthhreoo llaooggggyyr..e Assmivooen niggn ttehhaoorlssyee- owniisttehht AllaaDttee-,- oownnistsheett a AfaDst,,e rtt hhpeer ossgeervveeesrrsiiittoyyn v vaaanrrdiiee sms,, oaarnnedd s essvooemreee tptiiamtheesos lotthhgeey . llAeevvmeello noogff ppthaaotthhsooe lloowggiyyth ii nnl aAteD-o pnpasatetiitee nnAttssD cc, aatnnh eev vseeennv ebbreei tttyhh eev assraaimesee, aasns diinn sccooomggnnetiititmiivveellsyy tuuhnneii mleppvaaeiilrr eeoddf ccpooannthttrrooolollssg..33y33 in AD patients can even be the same as in cognitively unimpaired controls.33 FF iigguurree 22 TThhee Allzzhheeiimeerr’’ss ddiisseeaassee ppaatthhoollooggiiccaall ccaassccaaddee.. MFieegaaussuurerraa 2bb llee Tcchhheaa nnAggeess iinn bbiioomaarrkkeerrss ffoorr AD,, sshhoownn aass tthhee tthhiinn lliinneess,, ssttaarrtt maannyy yyeeaarrss pprriioorr ttoo ttMhheee accslluiinnriiaccb alaepp pcpheeaaarnraagn lzchee oimnecse i no f ecr’s disease pathological cascade. fb cooggnnlo -risk borders illustrateio mdifa iirttkiivveeer s ii mfoppra aAiirrDme,e nnshtt,,o sswhhnoo wasnn tbbhyye ttthhheein ll iilgginhhett--sgg, rrseeteeannrt zzmooannneey.. TTyehhaeer shh ipiggrhhio aarn ntddo l it ohwndei - vc risk ildinuiac b aoprderls piteha sr ainllcues toraf tec odginfffieteirrveeenn titmiiaattpeeaddi r rmriissekknsst ,ff ooshrr occwoonggn nbiittyiiv vteeh eii mlipgphaaiitrrm the sa e bio arker profile can present ith differe-ngr eeennetnt,, ziinnoddniiecc.aa ttTiinnhgge tthhhigaatht ttawnood ilnodwiv-riidsuka blso wrditehr st hiell ussatmraet eb idoimffearrkenert ipatreodf ilrei sckasn fporre sceongtn witiitvhe dimiffpeareirnmtt ceconogtg,nn iiinttiidvvieec alleetvivneegllss ..t SSoouurrccee:: JJianaccdkkiv eeitdt uaalla.., 29 hat two l,s 22 w0011it33h.. 2t9h e same biomarker profile can present with different cognitive levels. Source: Jack et al., 2013.29 Vassscccuullalrr aderm deenmttiiae ntia Vhaes cualianr fdeeatmureen otifa The main feature of VaaD iiss cceerreebbrroovvaassccuullaarr ppaatthhoollooggyy ddaamaaggiinngg tthhee bbrraaiinn aaTnnhdde iimappinaaii rrfiiennaggt uccroeo ggonnfii ttViiooannD..334 4 isT chheeerrreeeb raaorrveea smcuaalnnayyr pddaiifftfhfeeorrleeonngtty uudnnaddmeearrlglyyiiinnggg thvvaeas sbccuurallaainrr ppanaattdhh ooilmloogpgiiaeeissr ioonffg VcaaoDg,n, ssituuiocchhn .aa3s4s llTaarrhggeer iienn ffaarrecc ttssm aaannnddy h hadaeeifmfeoorrerrrnhhtaa ggueenssd,, eaanrnlddyi snsmg aavlllla vvsecesusssleaerll ddpiiassteehaaosseelo..33g44,i,33e55s T ohhf eeV rraeeDssee,a asrruccchhh ff aiieesll ddla roogffe V inaafDar hchtaasss a eenxxdpp ahannadedmeeddo ddrruuhrraiinnggegs tt,hh aeen pdp aasssmtt dadleel ccvaaeddseessse,,l mdioosevvaiinsnegg. 3ff4rr,3oo5 mT htthhe eer ehhsyyeppaoroctthh eefssieiissl d tt hhoaaftt VoonanDllyy lhlaaarrsgg ee xccpooarrnttiidccaaelld ii nndffuaarrriccnttgss tcchaaeuu sspeeadds tdd deeemceeandntetiisaa, ttmoo ottvhhieen guu nnfrddoeemrrss tttaahnnedd hiinnyggp otthhthaatet sttihhs eetrrheea taa rroeen mly aalnanryyg epp ooctoteernntittciiaall iccnaafuuassreecsst.s.33 55c aIIunns cecodon ndjjueumnnccettniiootinna wtoiit ththh etthh uiissn,, dvveaarsscctauunllaadrri nccgoo ggtnhniaitttii vvteeh eiirme ppaaariierr mmeeannntt y (( VpoCtIIe))n twiaalss cssauuggsggeeess.s3tt5ee ddIn aa css o aan jbburrnoocaatddioeenrr ttweerritmh tsshiinniscc, eev accseecrrueelbbarroo cvvoaagsscncuuitlliaavrre eeimffffeepccattissr miimenpptaa (ccVtt CtthhI)ee wbbarrsaa iisnnu giingne smteoodrr eea sw aaa yybssr ottahhdaaennr ppterrreemvvii oosuuissnllcyye aa ppcpperreecbciiraaotteveddas..33c55u,,336l6a IIrt t eiissf,,f effcootrrs eeixxmaampappclltee ,, tcchooem bmraooinn winiitt hhm rroeelrlaaett eewdd avvyaasss cctuuhllaanrr ppaarttehhvooiolloouggsylyy aiinnp prAeDcia tcecadass.3ee5ss,3,,6 Iaatnn dids , ifittossr eccoxoannmttrrpiibbleuu,tt iicooonnm mttooo ndd eewmiteehnn trtiieaal atwediitt hhv asoocttuhhleearr pathologiye s ini s AaDck ncoawseles,d gaendd.3 6 itCs ocmomntorinbluyt, iotnh e top rdeveamleenncteia owf ithV aDot heisr estimated to be 15-30% of all dementia cases, but if also counting dementia c66 ases due to mixed pathology and white matter hyperintensHHiAAtiNNeNNs,AA t WhWeEE TTeTTstEEiRRmBBaEEtRReGGs w6 ould rise to 50-70% of all dementia cases.37 H A N N A W E T T E R B E R G Compared to other types of dementia, the symptoms are more variable as it 6d epends on the type and location of the underlying path 37Hoanlonga yW.et teIrtb eirsg, however, common with subcortical vascular pathology, which often leads to deficits in attention, executive functions, and information processing, not always with a clear impact on memory.35 Depression and apathy are also common symptoms of VaD.35 Other dementia subtypes Beyond the most common types of dementia is a range of less common types, such as Lewy body dementia, frontotemporal dementia, and disorders linked to dementia, such as Huntington’s disease, traumatic brain injury (TBI), and Creutzfeldt-Jakob disease. Lewy body dementia (dementia with Lewy bodies or Parkinson’s disease dementia) is like AD progressive and with deposits of proteins α-synuclein, forming clumps called Lewy bodies within the neurons.38 Typical symptoms of Lewy body dementias are deficits in attention, executive function, visuospatial ability, fluctuation cognition, spontaneous parkinsonism, sleep problems, and recurrent visual hallucinations. Frontotemporal dementia is a collection of neurodegenerative dementias that affect the frontal and temporal lobes, leading to deficits in behaviour, executive function, and language.39 Mixed dementia Mixed dementia is the combination of more than one underlying cause of the disease co-occurring in the brain (Figure 3). It was long believed that dementia was due to one single cause, but studies have shown that there is a range of neuropathological abnormalities besides the AD neuropathological changes I NTRO DUCT I ON 7 ppaatthhopathoo llogloog iiegiee ss is s iiss aacknoacckknnoo wwlleedgewleddggee dd..3d.3 66 CCoommmonl36 Commmoonnl y, lyy, tthhe p, thee pp rrevalreevvaal eennccee ooff VVaaDD iiss eessttiimmaatteedd ttoo bbee 1155--3300% ooff aallll ddeemmeennttiiaa ccaasseess,, bbuutt iiff aallssoo c leonucnet inogf VaD is estimated to be 15-30% of all dementia cases, but if also ccoouunnttiinng d g dd eemem eenntmentt iia iaa ccaasses due caseess dduue to mixed path e ttoo mmiixxeedd ppaatthh oollooggyy aannd whiteology andd wwhhiitte mmaatttteerr hhyyppeerriintensities, thee matter hyperinntteennssiittiieess,, tthhee eessttiimmaatteess wwoouulldd rriissee ttoo 5500--7700% ooff aallll ddeemmeennttiiaa cases. estimates 37 would rise to 50-70% of all dementia ccaasseess..3377 CCoommppaarreedd ttoo ootthheerCompared to otherr types of dementia, the symptoms are more ttyyppeess ooff ddeemmeennttiiaa,, tthhee ssyymmppttoommss aarree mmoorree vvaarriiaabbllee aass iit depends variable as it t ddeeppeenndds oonn tthhee type and locas on the ttyyppee aanndd lloocca ttion of thatiioonn ooff tth ee underlhe uunnddeerrl y ly i yi n in g ng g pa ppa tthhoollooggyy..3377 37 IItt iiss,, hhoowwev athology. It is, howeev eerr,, ccoommmmoon with subcorticver, commonn wwiitthh ssuubbccoorrttiic aall vvaassccuullaarr cal vascular ppaatthhop o llooggyy,, wwhhiicchh oofftteenn lleeaaddss ttoo ddeeffiicciittss iinn aatttteennttiioonn,, eexxeeccuuttiivvee ffuunnccttiioonnss,, aannd a tihnofolormgya, twi hich often leads to deficits in attention, executive functions, andd iinnffoorrmmaattii oononn p p r prr ooccessingoceessssiinng ,, not alway g, nnoott aallwwaay ss wwiitth a clear imys withh aa cclleeaarr iim pac mppaac t on memor ctt oonn mmeemmoorr yy.35 Deprey..3355 DDeepprre ssssiioonn and apathy are alession aanndd aappaatthhyy aarree aal ssoo ccoommmmoonn ssyymmppttooms of VaD. lso 35 common symptommss ooff VVaaDD..3355 OOOttthhhheeerrre dddreee mmdeeennnmtttiiieaaa n sssutuubbbitattyyy pppseeeusssb types BBeeyyoonndd tthhee mmoosstt ccoommmmoonn ttyyppeess ooff dementia iBeyond the most common types of ddeemmeennttiiaa i ss aa rraannggee ooff lleessss ccoommmmoonn ttyyppeess,, ssuch as Lewy body dementia, frontotemporal i sd ae mraenngteia o, fa lnedss d ciosomrdmeorsn ltinypkeesd, ts uch as Lewy body dementia, frontotemporal dementia, and disorders linked to u o ch as Lewy body ddeemmeennttiiaa,, ssuucchh a sd eHmuennttiniag, tforontotemas Huntingtonn’’ss ddiisseeaas peo, rtarl dementise, traauummaattiicc b ar, aainn di ndjuisroyr d(TerBsI l)i,n aknedd Ctor eduetmzfeenldtita-J, askuocbh dasis eHausen. tington’s disease, traumatic b brraaiinn iinnjjuurryy ((TTBBII)),, aanndd CCrreeuuttzzffeellddtt--JJaakkoobb ddiisseeaassee.. LLeewwyy bbooddyy ddeemmeenntia (dementia with Lewy bodies or Parkinson’s disease dLeemweyn btioa)d yis dliekme eAn ti Dtia a p ((ddeerogm meennttiiaa wwiitthh LLeewwyy bodies or Parkinson’s disease de ressive and with de bpoosdiitess oof rp rPoatrekininss oαn-s’ys nduicsleeainse, fdoe mmeennttiiaa)) iiss lliikkee AADD pprrooggressiverming clumps calle ressive aanndd wwiitthh ddeeppoossiittss ooff pprrootteeiinnss αα--ssyynnuucclleeiinn,, ffoorrmmiinngg cclluummppss ccaallllee dd LLeewwyy bbooddiieess wwiitthhiinn tthhee nneeuurroonnss..3388 TTyyppical symptoms of Lewy body demde nLtieaws y abroe dideesf wiciittsh in the neurons. 38 Typiiccaall ssyymmppttoommss ooff LLeewwyy bbooddyy ddeemmentias are deficits iinn aatttteennttiioonn,, eexxeeccuuttiivvee ffuunnccttiioon, visuospatial ability, entias are deficits in atte n, visuospatial ability, fflluuccttuuaattiioonn ccooggnniittiioonn,, ssppoonntta nnteiooun, exeaneouss ppaarrk ciuntsivoekinson fun niissmm, cstlieon, sleeep , p pvrispro ubolespmast,i aloble an adb irleitcyu, rfrleunctt uvaistuioanl hcaollgunciitnioanti,o nsps.o Fntraonnetooutesm ppaorkrainl sdoenmisemnt, ias liese pa cporollebcletim mss,, aanndd rreeccuurrrreenntt vviissuuaall hhaalllluucciinnaattiioonnss.. FFrroonnttotemporal dementia is a collectioonn ooff nneeuurrooddeeggeenneerraattiivvee ddeemmeennttiiaass tthhaatt a oftfeemcporal dementia iaffectt tthhee ffrroonnttaall aan s da tceomllection of neurodegenerative dementias that affect the frontal a nndd tteem ppoorraall lloobbeess,, lleeaaddiinngg ttoo ddeeffiicciittsm s in behaviour, executive function, and langpuoagrael. 39l o bes, leading to deficits i inn bbeehhaavviioouurr,, eexxeeccuuttiivvee ffuunnccttiioonn,, aanndd llaanngguuaaggee..3399 MMiii ixxxxeeededd ddd eed demmeemeennntettiiiaana tia Miixxeedd ddeemmeennttiiaa iiss tthhee ccoommbbiinnaattiioonn ooff mmoorree tthhaan one underlying cause of the dMisiexaesde dceom-oecncutirar iins gt hine cthoem bbrianinat i(oFnig ourfe m 3o).r eIt t whaan n oonnee uunnddeerrllyyiinngg ccaauussee ooff tthhee ddiisseeaassee ccoo--ooccccuurrrriinngg iinn tthhee bbrraaiinn ((FFiigguurree 33)).. IItt wwa ss lloonngg bbeelliieved that das long belieevveedd tthhaatt dd eemmeennttiiaa wwaass dduuee ttoo oonne single cause, but ementia was due to onee ssiinnggllee ccaauussee,, bbuutt ssttuuddiieess hhaavvee sshhoowwnn tthhaatt tthheerree iiss aa rraannggee ooff nneeuurrooppaatthhoollooggiiccaall aabbnormaliti studies have shown that there is a range of neuropathological abnnoorrmmaalliittii eess bbesides es beessiiddeess tthhee AADD nneeuurrooppaatthhoollooggiiccaall cchhaannggeess that are associated with changes in cognthiteio An,D40 -4n4 eaunrodp tahthato ltohgei csayl ncehragnisgteics effect of multiple types of disease processes increases the likelihood of severe cognitive impairment or dementia.44,45 This becomes more evident with advancing age. For example, in the Religious Orders Study and Rush Memory IaINNnTTdRR OOADDgUUiCCnTTgII OOPNNr oject, two longitudinal cohort studies with participants with a 77 mI NTeRanO DaUgCT I ON e of 89 years at death, as much as 95% of those with a clinical7 diagnosis of probable AD had mixed pathologies at autopsy.46 Most common to coexist was a vascular disease, which was present in 90% of cases, and other pathologies were present in 65%. In the longitudinal 90+ Study from California, pathological evaluations found that 45% of dementia cases had mixed pathologies,42 and results from the same cohort found microinfarcts in 5In1tr%od oucft tiohne participants.47 7 Figure 3 Conceptual diagram over the relationship between AD and VaD pathology and mixed dementia. Other changes, such as DLB, also often coexists with AD or VaD, causing mixed dementia.37,48 It is, therefore, difficult to determine the type of dementia based on only clinical symptoms and medical history. To correctly diagnose the type of dementia requires sophisticated biomarkers and is sometimes impossible without a pathological examination, which, in many cases, is difficult and expensive in the research setting. It is instead common to perform analyses on all cases of dementia without stratifying by dementia subtype.6 This is especially true for those older than 85, and this is the age group of the main population included in this thesis. In Paper II and Paper IV, we only used the umbrella term dementia. In Paper III, we used the umbrella term dementia as the main outcome, but we also performed subanalyses stratified by AD, VaD, and mixed dementia, to evaluate the results further.49 Risk factors for dementia Throughout the life course, the accumulation of risk and protective factors affects the risk of dementia.21,36,50 Today there are a wide range of acknowledged risk factors for dementia, both non-modifiable as well as modfiable.6,51 Among non-modifiable risk factors are age, genetics, and biological sex. 8 H A N N A W E T T E R B E R G that are associated with changes in cognition,40-44 and that the synergistic effect of multiple types of disease processes increases the likelihood of severe cognitive impairment or dementia.44,45 This becomes more evident with advancing age. For example, in the Religious Orders Study and Rush Memory and Aging Project, two longitudinal cohort studies with participants with a mean age of 89 years at death, as much as 95% of those with a clinical diagnosis of probable AD had mixed pathologies at autopsy.46 Most common to coexist was a vascular disease, which was present in 90% of cases, and other pathologies were present in 65%. In the longitudinal 90+ Study from California, pathological evaluations found that 45% of dementia cases had tmthhaiaxtt e adar rpee a atahsssosooloccigiaaitteeesdd, 4 2 w wainitthdh rccehhsauanlntgsge efssr oiimnn ctchooegg nsnaiittmiiooenn c, 40-44 ,4o0-h44o raatnn fddo uttnhhdaat t m tthihceer ossiynynnfeaerrrgcgtiisst tiiincc t e5e h f1f af%et are associated with4 7changes in cognition, 40-44 and that the synergistic efffeec c tto oof fft hmmeuu pllttaiiprptllieec ittpyyappneetsss o.off ddiisseeaassee pprocesses increases the likelihood of sct of multiple types of disease prroocceesssseess iinnccrreeaasseess tthhee lliikkeelliihhoooodd ooff ssee evveerree c cog 44,45 vere cooggn niittiivveenitive iimmppaairmentimpaiirrmmeenntt or d oorr ddee emmeenntimenttiia a. a..4444,,4 455 TT T hh hiissis bb beeec coocomm meeess s mmmo ore orre e evviiddeenntt wwiitthh a addvvaanncciinngg aaggee.. FFoorr eexxaammppllee, in the Religious Orders Study and e Reuvsidhe Mnte mwoitrhy aFa“dniPgvduua rrnAeec ”3gi nAingDCg ao gPnecre.o pFjteuocartl ,ed txiawagmora mplol eMo, ,v iiennr ttthhe Religious Orders ngixiteude id n rReaelmlea ltcieiogonnihotsiohuairspt Obsetrtudwdeiresn SSAttDuudd ayyn aadnn Vdd “a PDRRusu urpsahehth M” MVoloeemaDgmy ooanrryyd aamnnixdde dAA dggeiimnneggn tPPiarr. oo jjeecctt,, ttwwoo lloonnggiittuuddiinnaall ccoohhoorrtt ssttuuddiie ess wwiitthh ppaarrttiicciippaannttss wwiitthh aa mOmteehaaenrn c haaaggneeg eoso,f fs u88c9h9 aysy eeDaaLrrsBs , aaatlts odd oeefaattethnh ,,c oaaess muc es with participants with x ismtsu wcihth AaasDs 9o95r5 V%%a D oo, fcf a tuthhsionogss eem iwwxeiidtth hd eama eccnlltiininai.i3cc a 7a,a4l8lm ddii eaagnn oagagnossiis e o s ooff f p 8ro9b yabealer sA aDt hdaedat mh, ixaesd m pautchho laosg i9e5s %at aouft otphsoys.e46 wMioths a clinical probable AD had mixed pa t common dtoia gcnooesxiiss to fw parso ab avbalse AD had mixed pat thhoollooggiieess aatt aauuttooppssyy..4466 MMoosstt ccoommmmoonn ttoo ccooeexxiisstt wwaass aa vvaassc cuullaarr ddiisseeaassee,, wwhhiicchh wwaass pprreesent in 90%cular disease, which was presseenntt iinn 9900%% of cases, and Iott hise,r tphaetrheofoloreg,i edsi fwfiecruel tp troes ednet eirnm 6in5e% t. hIen ttyhpee l onf gditeumdiennatli a9 0b o off ccaasseess,, aanndd ootthheerr ppaatthhoollooggiieess wweerree pprreesseenntt iinn 65%. In the longitudinal +as eSdtu doyn foronmly cClianliifcoarl nsiyam, ppattohmolso gaincda l meveadliucatli ohn i6s5%. In the longitudinal 9 900++ SSttuuddyy ffrroomm CCaalliiffoorrnniiaa,, ppaatthhoollooggiiccaall eevvaalluuaattiioonnss t fofooruyu.nn dTd otth hcaaott r44r5e5c%%tl y oo fdf idadegemnmoeesnnett iiatah ceca astseyesps ehh aoaddf mdmeiimxxeeedn tpia threoqlougiries , 4s2d pathologies,42 oaanpndhd i rsreteiscualttesd f rboim ofmo tuhanerk dse artmhs aeat c n45% of dementia cases had mixed pathologies,42 and ressuullttss ffrroomm the same co odhh ooisrr tt sffoomuunneddti mmeiiccsr rooimiinnpffaaorrsccsttissb iilnne 5w511i%t%ho ooufft tthah eep ppaatahrrttoiiclcoiippgaaic 47 nnattlss ..4e7x amination, twhhe iscahm, ein c omhaonryt fcoausneds, misi cdroififnicfuarltc tasn ind 5eA x1lzp%hee noismfi vteher e’is n pd atihrsteica rispesa 47 epnatarstc.hho lsoegttying. It is instead commoVna stcou plaerr pfoartmho laongaylyses o n all cases of dementia without stratifying by dementia subtype. 6 This is FeFsiigpguuerrceei a 33l lyC Ctoornuncceee ppftotuuraa ll t dhdiioaaggsrreaa mmo l odoveveerrr ttthhheae n rre e8llaa5tti,ioo nannshdi pt hbiest wise etnh eA Dag aeFmigixuerde d 3e mCenotniac.e ptual diagram over the relationsshhiipp bbeettwweeeenn AADD aann n dgd r VVoauaDDp ppoaaftt hhtohd VaD pathool loeog gmlogyy y aa aannind ndd mmpOoiitxxpeeddu lddOthheerr cc aheetmmaioneegnne ttsiii,aan s.. cu lcuhd aed in this thesis. In Paper II and Paper IV,hanges, such ass DDLLBB,, aallssoo oofften coexists with AD or VaD, causing mix wede d eomnelyn tuias.3e7,d48 Othteh eru cmhabnrgeesll, as uctehr ams D dLBem, alesno toiaftt.ee nnI ccnoo eePxxiissattpss wweriitt hhI AAIDDI, oowrr VVeaa DDu,,s cceaaduu ssiitnnhgge mm iiuxxmeedd bddreeemmlleeann ttiitaae..3 377,rm, 4488 dementia as the main outcome, but we also performed subanalyses stratified IbItty iiAss,, D tthh, eVerreaefDfoo,rr eae,n, ddi ifmfffiiiccxuuelldtt dttooe m ddeetnteetriramm, itinnoee e tvthhaeleu tatyytpep eet h ooef f r eddseeumltse nfutirat hbears. 4e9 It is, therefore, difficult to determine the type of demmeennttiiaa bbaasseed d oonn oonnllyy cclliinniiccaall ssyymmppttoommss aanndd mmeeddiiccaall hhiissttoorry. To correctly diagnose th de otny poen olyf cdleinmiceanl symptoms and medical histor yy.. TToo ccoorrrreeccttllyy ddiiaaggnnoossee tthhee ttyyppee ooff ddeemmeenntt tiiaa requires sophisticated biomarkers and is sometimes impossible Rwiisthko fuatic aa to r re pe qquirrsatu hfior eess ssoopphhiissor t tiiccaatteedd bbiioommaarrkkeerrss aanndd iiss ssoommeettiimmeess iimmpossible without a pathollo odggeiiccmaall e eenxxtaaimam iinnaattiioonn,, wwhhiicchh,, iinn mmaannyy ccaasseess,, iiss ddiiffffii pcuolsts iabnled wexitpheonusti va pa cult and eTxhpreonusgivheoe uiinnt tthhologicttheee rese aalr cehx asmetitninagti.o Int , isw ihniscthe,a din c ommanmyo cna tsoes p, eisr fodrifmfi caunlta lyasneds eoxnp eanlls icvaes eins ot fe lrrieefessee caaorrccuhhr ssseee, ttttthiinnegg ..a cIct uism iunlsatteioadn coofm rimsko ann tdo ppreortfeocrtmiv ea nfaadementia withou tI ts tirsa tinifsytienagd bcyo mdemmoenn ttioa psuerbftoyrpme. 6a nTa cllyytosseersss on all cases o his is oafnf eacts the rifs espelcli aclalys etsr uoef f kdd eeommfee nndtteiiaam wweniittthhiaoo.2uu1tt,3 6ss,5tt0rr aattTiiffoyyiidnnaggy bbtyyh eddreem eanretor those older than 85, and this is ement iiaaa ssuuwbbittdyyepp ee..r66a nTTghheii ss oiissf eeasscppkeenccoiiaawlllllyye dttgrruueeed ffrooirrs ktt hhfooassceet ooorllsdd eefrro ttrhh aadnne m8855e,,n aatinnadd, btthhoiitssh ii ssn tt tohhnee - amaggoee d ggirfroioauubpple o ofaf s tt hhwee e mmll aaiain po he age group of the main ns ppmooop pu pduufll liaaaat tbiioloenn.6 ,i5inn1 cc lluuddeedd iinn tthhiiss tthheessiiss.. IInn PPaappeerr IIII aanndd Paper IVtion included in this thesis. In Paper Paper IV, , wwee oonnllyy uusseedd tthhee uummbbrreellllaa tteerrmm ddeemmeennttiiaa.. IInn PPaappeerr IIII II,I awned uPsaepde rth IeV ,u wmeb roenlllay utesremd tdheem uemntbiar ealsla t htee rmma ind eomuetcnotima.e In Paper III I,, wwee uusseedd tthhee uummbbrella term dAemoengti an aosn t-hmeo mdia , but we also performed subanalyse rse slltar attiefrimed dbeym AeDnt,i a as the ma fiiinnab ooleuu ttccrioosmmk eef,,a bbctuuottr wws eea raaellss ooa gppeee, rrffgooerrnmmeteeicdds ss, uuabbnaadnn aabllyyiosseelos g4s9 ssittcrraaltt iiffsiieexdd. bbyy AADD,, VV VaD, and mixe aaDD,, aanndd mmiixxeedd d dementia, ddeemmeennttiiaa,, ttoo eevvaalluuaattee tthhee results further. to evaluate the rreessuullttss ffuurrtthheerr..4499 8 H A N N A W E T T E R B E R G RRiiissskkk ffa afccattoocrrsts offoorrr sdd eefmmoeernn tdtiiaa Risk factors for dementiae mentia TThhrroouugghhoouutt tthhee lliiffee ccoouurrssee,, tthhe accumulation of risk and protecThroughout the life course, thee aaccccuummuullaattiioonn ooff rriisskk aanndd pprrootteecctt tiivvee factors affects the risk of dementia.21,36,50 Today there are a wide ivrea nf faaccttoorrss aaffffeeccttss tthhee risk of dementia.2211,,3366,,5500 Today there are a wide rang gee ooff aacckknnoowwlleedge rdis kri sof dementia. Today there are a wide range of acknowleddggeedd rriisskk k ffaaccttoorrss ffoorr ddeemmeennttiiaa,, bbootthh nnoonn--mmooddiiffiiaabbllee aas well as mod 6,51 factors for dementia, both non-modifiable a ss wweellll aass mmooddff fiab iiaabbll l ee e...66,51 ,51 AAmmoonngg nnoonn--mmooddiiffiiaabbllee rriisskk ffaacctorAmong non-modifiable risk facttoorrs s s a a ar r re e e a aag gee,, genetics, and biologge, ggeenneettiiccss,, aanndd bbiioollooggi ical s iccaall sse exx.. ex. 88 H A N N A W E T T E R B E R G 88 H A Hanna Wetterberg H A NN NN AA WW EE TT TT EE RR BB EE RR GG that are associated with changes in cognition,40-44 and that the synergistic effect of multiple types of disease processes increases the likelihood of severe cognitive impairment or dementia.44,45 This becomes more evident with advancing age. For example, in the Religious Orders Study and Rush Memory and Aging Project, two longitudinal cohort studies with participants with a mean age of 89 years at death, as much as 95% of those with a clinical diagnosis of probable AD had mixed pathologies at autopsy.46 Most common to coexist was a vascular disease, which was present in 90% of cases, and other pathologies were present in 65%. In the longitudinal 90+ Study from California, pathological evaluations found that 45% of dementia cases had mixed pathologies,42 and results from the same cohort found microinfarcts in 51% of the participants.47 Figure 3 Conceptual diagram over the relationship between AD and VaD pathology and mixed dementia. Other changes, such as DLB, also often coexists with AD or VaD, causing mixed dementia.37,48 It is, therefore, difficult to determine the type of dementia based on only clinical symptoms and medical history. To correctly diagnose the type of dementia requires sophisticated biomarkers and is sometimes impossible without a pathological examination, which, in many cases, is difficult and expensive in the research setting. It is instead common to perform analyses on all cases of dementia without stratifying by dementia subtype.6 This is especially true for those older than 85, and this is the age group of the main population included in this thesis. In Paper II and Paper IV, we only used the umbrella term dementia. In Paper III, we used the umbrella term dementia as the main outcome, but we also performed subanalyses stratified by AD, VaD, and mixed dementia, to evaluate the results further.49 Risk factors for dementia Throughout the life course, the accumulation of risk and protective factors affects the risk of dementia.21,36,50 Today there are a wide range of acknowledged risk factors for dementia, both non-modifiable as well as modfiable.6,51 Among non-modifiable risk factors are age, genetics, and biological sex. Although dementia is not a normal part of ageing,36 age is considered the most crucial risk factor as the prevalence increases almost exponentially with a8g e (Figure 4).6,7 Genetic factors are associated with theH rAiNsNk Ao Wf EdTeTmE ReBnEtRiaG. However, less than five percent of all dementia cases are caused by an autosomal dominant mutation, labelled as familial AD.30 The common type of AD is considered to be sporadic. Still, a family history of sporadic AD increases the risk of dementia, and the heritability of sporadic dementia is estimated to be 60% to 80%.52 Genome-wide association studies have shown that the strongest genetic risk factor for sporadic AD is the ɛ4 allele of the apolipoprotein E (APOE).53,54 The APOE gene codes the apolipoprotein E protein (apoE) and has three polymorphic alleles, ɛ2, ɛ3, and ɛ4. While the ɛ4 allele is a risk factor for AD, the ɛ2 is considered to be protective, and the ɛ3 neutral.55 Having one APOE ɛ4 allele increases the risk of AD 3-4 compared toh eh mavein gr enaochnien, ga nthde hhaivgihn ga gtews oin c wophiiecsh idnecmreeansetsia tihs ec orimskm 9o-n1,5 wtiomuelds .5h6a vIne apdrdoitteicotniv teo fAacPtoOrEs mɛ4a, ktihnegr et haerem o tlehsesr gveunlneetirca brleeg ioton sd reemlaetendti at.o TAhDer er isakr.e57, Thohweseev ear,e onthoet r afsa cstororsn ge xrpislaki nfiancgt otrhse asse xth dei fAfePreOnEce .ɛ F4 oarn edx amrep cleo, ma moentaly- garnoaulypseisd wtoitghe tdhaetra ionn p aollmygoesnti c5 8r is0k0 0s cpoarretsic iinp asntutsd ifeosu.5n8 d that women with the APOEɛ3/ɛ4 genotype in the age group 65 to 75 years had a higher risk of Bdeiovleologipcianl gs eAx Dis athlsaon comnesnid ewreitdh at hries ks afamcteo rg efnoro tdyepme.e61n tSiaim,59i lasr lyw, opmrevni oaures asftufedcted b60ies havy e dfeomuenndt ial owtoe r a coggrenaittievre efxutnenctti otnh aanm monegn wino msenv erwali thw atyhse. Agepnportoyxpiem tahtealny tmweon-,t6h2 iradnsd o af ahlilg dheemr ernistkia ocfa sceosn avrer wsionm efrno, man dm wildo mcoegnn hitaive aim hp5iga0ihremr elnifte ttoim Ae Dri.s6k3 Mofo rdeeovveelro,p tihneg sAuDdd.3e2n Are dpuarctt ionf tohfi so eis treoxgpelani ndeude btoy wmoemnoepna ulisvein hga slo bnegeenr sthugange mstedn ,a sa nad f acst oarg ea fifse ctthien gs trrioskn gfeosrt AriDsk infa cwtomr feonr. d m40Oesternotgiae,n m hoarse, iwn oamniemn alli vaen dlo ncegl l emnooudgehls tsoh doewvne lotop bthe en deuisreoapsero.1t4e,5c9,t6i0v Iet iins aslesvo3e 0hraylp woathyse.s1i3z Tedh eth raotl ed uoef otoe stthreo gloewnse ro mn tohreta rliistyk ionf wAoDm iesn, h, tohwereev eisr ,a d seubravtievdal. ePfrfevcito aums onbgse mrveanti orneacl hstinugd iehsig rhe paogrets .a T rhedisu scuerdv irvisakl eofff eAcDt w ino uwldo menan u sthinagt exo2g0enous estrogen (i.e., hormonal replacement therapy),64,65 while others rep1o0rt an increased risk.66,67 Further, studies examining endogenous estrogen 6 Age groupsexpo0sure (e.g., reproductive period [age at menarche to age at menopause]) als 5o-0 report divergent results, with some showing that a longer reproductive pe r4i0od in4c0r-e5a9ses 6th0-e6 4risk6 o5f-6 A9D,7680,6-79 4whi7le5 -o79ther8s0 r-e8p4ort8 a5 -r8e9duc9ed0+ risk a9m5+ong wo 3m0en with a shorter reproductive Apgeer igordo.u7p0 sIn addition to the sex-specific ris k2 0factors, there are sociocultural aspects that relate to the below-described Men Women mo 1d0ifiable risk factors for dementia.71 - Figure 4 40P-r5e9va6le0n-c6e4 of6 d5e-m69en7ti0a -i7n4 m7e5n- a7n9d 8w0o-m84en8 b5y- 8ag9e. 90+ 95+ Prevalence data based on data fromM Kodesh et.al., 2019, the World Health Organisation, 2021, Crimmins et.al., 2018, Cao et.al., 202e0n, and BöWrjeossmone-nHanson et.al., 2004.4,8-11 Interacting with non-modifiable risk factors is a wide range of modifiable risk factors for dementia that have been identified in the last decades.21,36 A meta- IINntTrRoOdDucUCTanalysist iosn I O N 9 howed that up to 40% of all dementia cases theoretically are9 accounted for by modifiable risk factors, meaning that a large proportion of dementia cases could be prevented or delayed.21,72 These include twelve specified risk factors, being lower education, hearing impairment, depression, physical inactivity, infrequent social contact, head injury, hypertension, smoking, diabetes, obesity, excessive alcohol consumption, and air pollution (Figure 5). The mechanisms of how prevention or managing of these risk factors could prevent or delay dementia are proposed to be mediated in two different paths; reduced neuropathological damage and increased and 10 H A N N A W E T T E R B E R G Prevalence of dementia Prevalence of dementia Although dementia is not a normal part of ageing,36 age is considered the most crucial risk factor as the prevalence increases almost exponentially with age (Figure 4).6,7 Genetic factors are associated with the risk of dementia. However, less than five percent of all dementia cases are caused by an autosomal dominant mutation, labelled as familial AD.30 The common type of AD is considered to be sporadic. Still, a family history of sporadic AD increases the risk of dementia, and the heritability of sporadic dementia is estimated to be 60% to 80%.52 Genome-wide association studies have shown that the strongest genetic risk factor for sporadic AD is the ɛ4 allele of the apolipoprotein E (APOE).53,54 The APOE gene codes the apolipoprotein E protein (apoE) and has three polymorphic alleles, ɛ2, ɛ3, and ɛ4. While the ɛ4 allele is a risk factor for AD, the ɛ2 is considered to be protective, and the ɛ3 neutral.55 Having one APOE ɛ4 allele increases the risk of AD 3-4 compared to having none, and having two copies increases the risk 9-15 times.56 In addition to APOE ɛ4, there are other genetic regions related to AD risk.57 These are not as strong risk factors as the APOE ɛ4 and are commonly grouped together in polygenic risk scores in studies.58 Biological sex is also considered a risk factor for dementia,59 as women are affected by dementia to a greater extent than men in several ways. Approximately two-thirds of all dementia cases are women, and women have a higher lifetime risk of developing AD.32 A part of this is explained by wthoem meenn l irveiancgh ilnogn gthere thhiagnh mageens , inan wdh aicsh a gdee mise nthtiea sist rcoonmgemsto rni,s kw ofaucltdo rh afvoer dementia, more women live long enough to develop the disease.14,59,60 protective factors making them less vulnerable to dementia. There Iatr eis, ahlosow hevyepro, tohtehseizre fda tchtoatr sd uexe ptola itnhien lgo wtheer mseox rdtaiflifteyr ienn wceo. mFoenr , etxhaemrep isle a, sau mrveivtaa-l eafnfaelcyts iasm woitnhg dme atan o rne aaclmhionsgt h5i8g h0 0a0g epsa.rTtihciisp asnutrsv ifvoaul nedf ftehcat t wwooumlde nm weainth tthhaet tAhPe OmEeɛn3 r/eɛa4c hgienngo tthype eh iingh t haeg easg ein g wrohuicph 6 d5e tmo e7n5ti ay eisa rcso hmadm oa nh, igwhoeur ldri shka voef pdreovteeloctpivineg faAcDto rsth amna kminegn twheitmh tlheses svamulne ergaebnloe tytpoe .d61e mSiemnitliaar. lyT, hperree 60 v iaorues, hstouwdieevse rh, aovteh erf ofuanctdo rslo exp Alagine ignrgo upswer cognittihvee sefuxn dcitfifoenr enacmeo. nFor example, a 50 g women wit hm etthae- agnenaloytsyisp ew itthha nd amtae onn,6 2a lamndo sat 5h8ig 0h0e0r priasrkt icoifp acnotnsv feorusinodn tfhraotm w ommiledn c wogitnhi ttihvee A 40imPpOaiErmɛ3e/nɛt 4t og eAnoDt.y6p3 eM ino rtehoev earg,e thgreo suupd 6d5en t or e7d5u cyteioarns ohfa do eas thrioggheenr rdiuske otof dmee v3ne0oloppaiunsge hAaDs bteheann sumgegne stwedit ha st ha ef ascatmore a fgfeenctointygpe. 61 risk Sfoimr iAlaDrly i,n pwreovmioeuns. sOtue 2ds0tireos gehna vhea sf, oiunn adn imlowal ear ndco cgenlli timveo dfeulsn csthioonw na mtoo nbge nweourmoepnro twecitthiv et hine gseev n1eo0rtaylp we atyhsa.1n3 Tmheen r,o62l ea onfd oae shtrigohgeern sr ioskn tohfe croisnkv oefr sAioDn ifsr,o hmow meviledr ,c doegbnaitteivde. iPmrepv a-iiormuse notb steor vAaD 63 tio.n aMl soturedoievse rr,e pthoer t sau dreddeunc eredd ruisckti oonf AofD o iens twroogmene nd uuesi ntog menopa4u0s-e5 9ha6s0 -b6e4e65-69 70-74 75-79 80-84 85-89 90+ 95+exogenous estrogenn (si.ueg.,g ehsotermd oasn aal fraecptloarc eamffeecntti ntgh errisakp yf)o,6r4 ,6A5 Dw hinil ew oomtheenrs. Orepesotrrto agnen in hcares,a siend arnisimk.6a6l an Mde ncell moWdomen,67 Further, studeiless sehxoawmnin tion gb een dnoeugreonporuost eecsttirvoeg einn s 13eexvpeorsaul rwea (yes..g., Trehper roodleu cotfiv oee spterroigoedn [sa ogne taht em riesnka orcf hAeD to i sa, gheo wate mveern, odepbauatseed]). Palrseov iroeupso rotb dsievrevragteionnt arle sstuuldtsi,e sw ritehp osortm ae r esdhuocweidn gr isthka ot fa A loDn giner w reopmroend uucstiinvge epxeorigoedn ionucsr eeassterso tgheen ri(si.ke .o, fh AoDrm,6o8,n69a wl rheiplel aoctehmeresn rte pthoerrta ap rye),d64u,6c5e dw rhiislke aomthoenrgs IrwNeoTpRmoOreDtn Ua CnwT iIintOhcN 66,67 r ae assheodr rteisrk r.epr oFduurctthivere, psteurdioieds. 7e0 xIanm aidndinitgi oenn dtoo gtehneo suesx e-ssptreocgifeinc9 erixspko fsaucrteo r(se,. gth.,e rreep arroed suocctiiovec uplteurrioald a [sapgeec tast tmhaetn raerlcahtee ttoo tahgee b aetl omwe-ndoepscaruibsee]d) amlsood irfeipabolret rdisivke fragcetnotr sr efsourl tdse, mweitnht isao.7m1 e showing that a longer reproductive p eriod increases the risk of AD,68,69 while others report a reduced risk among wFiogmuree n4 wPirtehv aale nshceo orft edre mreepntrioa dinu mcteinv ea npde wrioomde.n70 bIyn a gaed. dition to the sex-specific rPirsekvCrim afleance micntso er data ts.a,l .t hbaesed on da, 201re8, aCraeo seo tac firoocmu lKtuordaest.al., 2020, andl ha sept.aelc., t2019, th Börjessso nt-hHaatn r e World seolna teet. atlo Health Org ., 2th00e4 .b4,e8-l1o awni1 - sdateiosnc,r 2ib02e1d, modifiable risk factors for dementia.71 Figure 4 Prevalence of dementia in men and women by age. PInretvearlaencctien dga twa bitahse ndo onn- dmatoa dfriofmia bKleo dreisshk efta.acl.t,o 2r0s1 9is, tah ew Widoer lrda Hnegaelt oh fO mrgoandisifaitaiobnl,e 2 r0i2s1k, Cfarcimtomrisn sf oetr. adl.e, 2m01e8n, tCiaa oth eat.ta lh., a2v0e20 b, eanedn Bidörejnestsiofine-dH ains tohne e tl.aasl.t, 2d0e0c4a.4d,8e-1s1 .21,36 A meta- analysis showed that up to 40% of all dementia cases theoretically are accounted for by modifiable risk factors, meaning that a large proportion of Idnetmereancttiian gc wasieths ncoonu-ldm obdei fpiarbelvee rnistekd f aoctro rdse ilsa yae wd.i2d1,e7 2r aTnhgee soef imncolduidfiea btlwe erlivske fsapcetcoirfise fdo rri sdke mfaecntotiras ,t hbaetin hga lvoew beere end iudceanttioifnie,d h iena rtihneg liamstp daeircmadeenst.,2 1d,3e6p Are mssieotan-, apnhaylsyisciasl sihnoacwteivdi tyt,h aitn furepq utoen t4 0s%oc ioalf caoll ndtaecmt, enhteiaa dc aisnejus ryth, ehoyrpeteirctaelnlys ioarne, asmccookuinntge,d d fiaobr ebteys m, oobdeisfiiatyb,l ee xrciseks sfiavcet oalrcso, hmoela cnoinngsu tmhaptt iao lna,r gaen dp raoirp poortlilounti oonf d(Feimguernet ia5 ).c aTshese mcoeuclhda nbies mpsr eovfe nhtoewd porre vdeenlatyioend .2o1,r7 2 mTahneasgein ign colfu dthe estwe erlivske sfapcetcoifrise dco ruislkd fparcetvoerns,t boeri ndge llaoyw deer medeunctiaat iaorne, phreoaprionsge idm tpoa birem menetd, idateepdr einss itowno, pdhifyfesirceanlt inpaacthtisv;i tyr,e dinufcreedq uneneut rosopcaitahlo lcoognictaalc t,d ahmeaadg e inajnudry , inhcyrpeearsteedn siaonnd, smoking, diabetes, obesity, excessive alcohol consumption, and air pollution 1 0 Hanna Wetterberg (Figure 5). The mechanisms of how prevention or managing of these risk factors could prevent or delay dementia are proposed to be mediated in two different paths; reduced neuropathological damage and increased and 10 H A N N A W E T T E R B E R G 10 H A N N A W E T T E R B E R G Prevalence of dementia the men reaching the high ages in which dementia is common, would have protective factors making them less vulnerable to dementia. There are, however, other factors explaining the sex difference. For example, a meta- analysis with data on almost 58 000 participants found that women with the APOEɛ3/ɛ4 genotype in the age group 65 to 75 years had a higher risk of developing AD than men with the same genotype.61 Similarly, previous studies have found lower cognitive function among women with the genotype than men,62 and a higher risk of conversion from mild cognitive impairment to AD.63 Moreover, the sudden reduction of oestrogen due to menopause has been suggested as a factor affecting risk for AD in women. Oestrogen has, in animal and cell models shown to be neuroprotective in several ways.13 The role of oestrogens on the risk of AD is, however, debated. Previous observational studies report a reduced risk of AD in women using exogenous estrogen (i.e., hormonal replacement therapy),64,65 while others report an increased risk.66,67 Further, studies examining endogenous estrogen exposure (e.g., reproductive period [age at menarche to age at menopause]) also report divergent results, with some showing that a longer reproductive period increases the risk of AD,68,69 while others report a reduced risk among women with a shorter reproductive period.70 In addition to the sex-specific risk factors, there are sociocultural aspects that relate to the below-described modifiable risk factors for dementia.71 Figure 4 Prevalence of dementia in men and women by age. Prevalence data based on data from Kodesh et.al., 2019, the World Health Organisation, 2021, Crimmins et.al., 2018, Cao et.al., 2020, and Börjesson-Hanson et.al., 2004.4,8-11 Interacting with non-modifiable risk factors is a wide range of modifiable risk factors for dementia that have been identified in the last decades.21,36 A meta- analysis showed that up to 40% of all dementia cases theoretically are accounted for by modifiable risk factors, meaning that a large proportion of dementia cases could be prevented or delayed.21,72 These include twelve specified risk factors, being lower education, hearing impairment, depression, physical inactivity, infrequent social contact, head injury, hypertension, smoking, diabetes, obesity, excessive alcohol consumption, and air pollution (Figure 5). The mechanisms of how prevention or managing of these risk factors could prevent or delay dementia are proposed to be mediated in two different paths; reduced neuropathological damage and increased and maintained cognitive reserve. Neuropathological damage can be increased by diabetes, hypertension, o10besity, head injury, smoking, and air pollution. Diabetes, hypertension, and H A N N A W E T T E R B E R G obesity are all vascular risk factors that increase the risk of cognitive decline.73 Smoking and air pollution have a vascular and toxic effect on the brain, and head injuries have recently been recognized to increase the risk of neuropathological damage. If the exposure is minimized and adequately treated, the neuropathological damage could be reduced. Tmhaein thayinpeodth ceosgisn iotifv ec roegsneritvive.e reserve is based on evidence of individual variability of cognitive symptoms at the same levels of neuropathological cNheaunrgoeps,a tihnodliocagtiicnagl rdeasmilieangec e cinan s obme e iinncdrievaisdeuda lsb.7y4 Tdhiaisb erteessil,i ehnycpe ecratenn bsiootnh, boeb emsitayn, ihfeesatde din ajusr ya, scmogonkiitnivg,e anreds earivr ep oalnludt ioan .b Draiianb erteesse,r vhey.p eIrnt enshsioornt,, atnhde ob 73cogensiittyiv aer ere aslel rvvaes csutalanrd rsi sfko fra tchtoe rasd tahpatta ibniclirteya soef tthhee rbisrka ionf tcoo gmnoitriev ef udnecctliinoen.al bSrmaionk pinrgo caensdse asi,r a pnodl tluhtiiso and ahpatvaeb ail ivtya sccaunl abre ainmdp trooxviecd e bffye lcifte ostny lteh fea cbtroarins., Tanhde bhreaaidn riensjeurrviees i s hthave en eruercoebnitolylo gbicealn c arpeictaolg, nmizeeadn intgo thien cnrueamseb ert hoef nreisukr onosf annedu rosypnaathposelosg, icwahl icdha mcaagne . bIef itnhcer eeaxsepdo suorre misa inmtianinimediz etdh roanudgh adceoqgunaittievley strtiematuelda,t itohne. 7n4 eHuriogphaetrh eodlougciactaiol nd aamnadg ef rceoquuledn bt es orecdiaul cceodn. t act are thought to increase the cognitive reserve, while untreated hearing impairment reduces tThhee cohgynpiottivhee srise seorfv ec,o glinkeitliyv ed uree setorv ea viosi dbaansceed oof ns teimviudleanticneg oafc tiivnidtiievsid.2u1,7a2l Pvahryiasibcialilt iyn aocft ivcoitgyn, diteivper essysmionp,t oamnds eaxtc etshsei vsea amlceo hleovle ulss eo afr en tehuorougphatth tool oafgfieccatl tchhea ngreiss,k indoifc atidnegm reesnitliiae nceth irno usoghm e binodtihv iduthal 74 es . c Toghnisi triveesi liernecsee rcvaen baontdh nbeeu rmoapnaitfheosltoedgi caasl daa mcoaggne.i t ive reserve and a brain reserve. In short, the cognitive reserve stands for the adaptability of the brain to more functional brain processes, and this adaptability can be improved by lifestyle factors. The Fbirgauinre r5e seSruvgeg eisste tdh me encheaunriosmbsio floor gpircoaml octainpgi tcaol,g nmitievaen reinsegr vteh aen dn urimsk bredru octfio n.e urons Aan sdum smyanray posf ethse, twwehlvice hm ocdaifnia bblee r isikn fcarcetaosrse adn do trh e msuagignetsateidn epdat htwharyos uing whh iccho gthneiyt iavree modelled to change the risk of dementia. Source: Livingston, et.al. 2020.21 stimulation.74 Higher education and frequent social contact are thought to increase the cognitive reserve, while untreated hearing impairment reduces the cognitive reserve, likely due to avoidance of stimulating activities.21,72 Physical inactivity, depression, and excessive alcohol use are thought to affect ItNhTeR O DrUisCkT I ONo f dementia through both the cognitive reserve an1d1 neuropathological damage. Figure 5 Suggested mechanisms for promoting cognitive reserve and risk reduction. A summary of the twelve modifiable risk factors and the suggested pathways in which they are modelled to change the risk of dementia. Source: Livingston, et.al. 2020.21 Introduction 1 1 I NTRO DUCT I ON 11 • Minimise diabetes • Treat hypertension • Prevent head injury • Stop smoking • Reduce air pollution Reduced neuropathological damage • Reduce midlife obesity (amyloid or tau-mediated, vascular or inflammatory) • Maintain frequent exercise • Reduce occurrence of depression Preventing dementia • Avoid exessive alcohol Increased and maintained cognitive reserve • Treat hearing impairment • Maintain frequent social contact • Attain high level of education maintained cognitive reserve. Neuropathological damage can be increased by diabetes, hypertension, obesity, head injury, smoking, and air pollution. Diabetes, hypertension, and obesity are all vascular risk factors that increase the risk of cognitive decline.73 Smoking and air pollution have a vascular and toxic effect on the brain, and head injuries have recently been recognized to increase the risk of neuropathological damage. If the exposure is minimized and adequately treated, the neuropathological damage could be reduced. The hypothesis of cognitive reserve is based on evidence of individual variability of cognitive symptoms at the same levels of neuropathological changes, indicating resilience in some individuals.74 This resilience can both be manifested as a cognitive reserve and a brain reserve. In short, the cognitive reserve stands for the adaptability of the brain to more functional brain processes, and this adaptability can be improved by lifestyle factors. The brain reserve is the neurobiological capital, meaning the number of neurons and synapses, which can be increased or maintained through cognitive stimulation.74 Higher education and frequent social contact are thought to increase the cognitive reserve, while untreated hearing impairment reduces the cognitive reserve, likely due to avoidance of stimulating activities.21,72 Physical inactivity, depression, and excessive alcohol use are thought to affect the risk of dementia through both the cognitive reserve and neuropathological damage. Figure 5 Suggested mechanisms for promoting cognitive reserve and risk reduction. A summary of the twelve modifiable risk factors and the suggested pathways in which they are modelled to change the risk of dementia. Source: Livingston, et.al. 2020.21 I NTRO DUCT I ON 11 1 2 Hanna Wetterberg maintained cognitive reserve. Diiaaggnnoosstticic c crirtietreirai a Neuropathological damage can be increased by diabetes, hypertension, There are two major sets of classification systems to diagnose dementia:36 the obesity, head injury, smoking, and air pollution. Diabetes, hypertension, and International Statistical Classification of Diseases and Related Health obesity are all vascular risk factors that increase the risk of cognitive decline.73 Problems (ICD) system, developed by the World Health Organization75 Smoking and air pollution have a vascular and toxic effect on the brain, and (WHO), and the Diagnostic and Statistical Manual (DSM) developed by the head injuries have recently been recognized to increase the risk of American Psychiatric Association.76 The ICD system is used globally within neuropathological damage. If the exposure is minimized and adequately the health care system and death certificate data to code diseases and is treated, the neuropathological damage could be reduced. accompanied by diagnostic guidance. The DSM is the most frequently used classification system in clinical research to categorize psychiatric diseases.36 In The hypothesis of cognitive reserve is based on evidence of individual the 50s, “senility and ill-defined diseases” was first introduced in the ICD- variability of cognitive symptoms at the same levels of neuropathological 7,77and in the two first published editions of DSM in 1952 and 1968, chronic changes, indicating resilience in some individuals.74 This resilience can both brain syndrome due to arteriosclerosis and senile brain disease were included. be manifested as a cognitive reserve and a brain reserve. In short, the With every new edition, the description and diagnostic criteria have been cognitive reserve stands for the adaptability of the brain to more functional updated within both systems. The current version of the ICD (ICD-11) was brain processes, and this adaptability can be improved by lifestyle factors. The published in May 2019 and is being translated into Swedish but is yet to be brain reserve is the neurobiological capital, meaning the number of neurons implemented, and the DSM-5 was published in 2013.76 As both classification and synapses, which can be increased or maintained through cognitive systems are widely used, there have been efforts to harmonize the systems stimulation.74 Higher education and frequent social contact are thought to since the preparation of the ICD-10 and DSM-IV, released in the mid 90s.78 increase the cognitive reserve, while untreated hearing impairment reduces Despite this, the criteria differ in many ways (see Box 1 and Box 2 for brief the cognitive reserve, likely due to avoidance of stimulating activities.21,72 descriptions of the diagnostic criteria). The main reason for this is because of Physical inactivity, depression, and excessive alcohol use are thought to affect different priorities and use of the criteria between the two organizations.78 the risk of dementia through both the cognitive reserve and neuropathological damage. As the criteria differ, the choice of diagnostic criteria will affect the number of individuals diagnosed with dementia. The older versions are generally seen as more Alzheimer’s oriented, as deficits in memory are mandatory for Figure 5 Suggested mechanisms for promoting cognitive reserve and risk reduction. A summary of the twelve modifiable risk factors and the suggested pathways in which they are diagnosing dementia, whilst the newer versions accept deficiencies in any modelled to change the risk of dementia. Source: Livingston, et.al. 2020.21 cognitive domain in order to capture other types of dementia.2 There has been some research comparing the different systems, indicating that the DSM systems generally capture more cases than the ICD criteria.79-81 To our knowledge, there are no studies comparing the latest DSM-5 and the ICD-11 criteria, which is the aim of Paper II. I NTRO DUCT I ON 11 12 H A N N A W E T T E R B E R G Introduction 1 3 Box 1. Diagnostic criteria in the ICD-systems ICD-10 ICD 11 Impairment of short-term or long-term Impairment in two or more cognitive memory domains: memory impairment, executive functioning, attention, language, social Significant decline in other cognitive cognition and judgment, psychomotor abilities characterized by deterioration in speed, and visuoperceptual or judgment and thinking, such as planning visuospatial functioning. and organizing Information obtained from the Significant impairment of emotional individual, informant, or clinical control or motivation or change in social observation behaviour, presenting in at least one of the following: emotional lability, Substantial impairment in memory irritability, apathy, coarsening of social performance as demonstrated by behaviour, representing a decline standardized neuropsychological or cognitive testing or, in its absence, another quantified clinical assessment. Behavioural changes (e.g., changes in personality, disinhibition, agitation, irritability) may also be present The symptoms result in significant impairment in personal, family, social, educational, occupational, or other important areas of functioning and represent a decline The cognitive deficits do not occur The symptoms are not better accounted exclusively in the context of a delirium for by disturbance of consciousness The symptoms are not better accounted for by altered mental status Source: World Health Organization. International Classification of Diseases, Eleventh Revision (ICD-11). 2019/2021,75 World Health Organization. The ICD-10 Classification of Mental and Behavioural Disorders: Diagnostic Criteria for Research. 1993.82 1 4 Hanna Wetterb1e3rg I NTRO DUCT I ON Box 2. Diagnostic criteria in the DSM-systems DSM-III-R DSM-IV DSM-5 Concern of self or informant of significant cognitive decline in one or more cognitive domains Impairment of short-term Impairment of short-term Significant impairment in and long-term memory or long-term memory cognitive performance in one or more cognitive Impairment of abstract Significant impairment in domains: learning and thinking, impaired at least one of the memory, language, judgment, aphasia, apraxia, following domains: executive function, agnosia, constructional aphasia, apraxia, agnosia, complex attention, difficulties, and personality and disturbance in perceptual-motor, social change executive functioning cognition The above criteria each The above criteria each The cognitive deficits cause significant social/ cause significant social/ interfere with occupational dysfunction occupational dysfunction independence in everyday and represent a decline and represent a decline activity and represent a decline The cognitive deficits do The cognitive deficits do The cognitive deficits do not occur exclusively in the not occur exclusively in the not occur exclusively in context of a delirium context of a delirium the context of a delirium The cognitive deficits are The cognitive deficits are The cognitive deficits are not better explained by not better explained by not better explained by another mental disorder another mental disorder another mental disorder Source: American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 3rd ed., revised. 1987,83 American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders: DSM-IV. 1994,84 American Psychiatric Association. Diagnostic and statistical manual of mental disorders 5th ed. 2013.76 14 H A N N A W E T T E R B E R G Introduction 1 5 Epidemiology Epidemiology Epidemiology is the study of the distribution and determinants of public health concerns in a population.85 Descriptive epidemiology examines the dEisptriidbuetmioni oolfo dgisye ases in regard to population, place, and time, while analytical epidemiology examines the cause or aetiology of the disease. When s tudying the distribution of disease, measures used include incidence, prevalence, aEnpdid memoirotaloligtyy. iIsn cthidee nsctue drye folefc ttsh eth ed isnturmibubteiro no f annedw d ceatesrems iwniatnhitns oa fd pefuinbelidc phoeapluthla tcioonn cdeurrnins gi na gai vpeonp tuimlateio fnra.8m5 eD, ewshcriliep ttihvee perpeivdaelmenicoelo igs yt heex apmroinpeosr titohne odfis itnridbiuvtidiouna lso wf itdhis tehaes edsi seinas er eagt aar gdi vteon tpimope.u Tlahtieosne, copnlacceep, tsa nindt ertiamcte a, nwd hairlee oanftaelnyt idcaels cerpibideedm wioiltohg tyh eex a“mbaitnhetsu tbh ea ncaaluosgey o”r (aFeitgiourloe g6y) o. fT thhee dinisceidaesen.c We, hthene nsteuwdylyin dgi athgen odsisetdri bcuatsieosn, iosf sdhisoewasne ,b my etahseu rneesw u sweda tienrc lfulodwe iinncgi dinentoc et,h per ebvaatlhetnucbe., Tanhde pmreovrtaalelintyc.e ,I tnhceid pernocpe orretfiolenct hs atvhien gn tuhme bdeisr eoasf en, iesw r ecparseesse nwtietdh ians ath de ewfianteedr aplorepaudlayt iionn t hdeu rbinatgh atu gbi.v Tenh et immoe rftraalimtye i,s w shhiolew tnh aes p trheev dalreaninc,e wisi tthh ec apsreos pleoarvtiionng tohfe i nbdaitvhitduuba.l sT wheit hp rtehvea dleisnecaes eis a tin af lguiveennc etdim bey. Tthhee srea tceo onfc nepewts cinatseersa cotc acnudrr ianrge (oifntceind ednecsec)r bibuetd a lwsoit bhy t thhee “mboarthtatluitby . aCnhaalonggye”s i(nF tighuer ien c6i)d. eTnchee ainndc ipdreenvcael,e nthcee onfe wdliys edaisaeg ncoosueldd cmaosevse, iins sdhioffwerne nbty d tihreec tnioewns .w Iaft ethr ef loinwciindegn icnet oo ft hteh eb adtihsetuasbe. dTehcer eparseevs,a tlhene cper,e tvhael epnrcoep coortuioldn s htiallv iinncgr ethasee d iifs ethasee m, ios rrteaplirteys oenf ttehde adsi stehaes ew aaltseor dalerceraedays ein.8 6t hSein bcaet hthtueb a. sTphecet ms aofrfteaclitt ye aisc hs hootwhenr ,a sit tihse i mdrpaoinr,t awnitt hto c assteusd lye athveinmg athlle t ob authntduebr.s tTahned pthreev eapleindceem iiso ilnofglyu eonfc tehde b dyi stehaes era. tIen o Pf anpewer cIaIsIe sa nodc cPuarrpinegr I(iVnc, iwdee nincev)e bstuigt aatles ot hbey ttihmee m troerntadlsit iyn. Cthhea ningceisd einn cthe,e pinrecvidaelennccee a, nadn dp rmevoarlteanlictye ooff ddeismeaesnet iac oaumldo nmgo ovcet oing edniaffreiarnens.t directions. If the incidence of the disease decreases, the prevalence could still increase if the mortality of the disease also decrease.86 Since the aspects affect each other, it is important to study them all to understand the epidemiology of the disease. In Paper III and Paper IV, we investigate the time trends in the incidence, prevalence, and mortality of dementia among IoNcCtoIgDeEnaNriCanEs. PREVALENCE F igure 6 The bathtub analogy. T he bathtub analogy describes the relationship between incidence (newly diagnosed cases shown as water flowing into the tub), prevalence (number of cases having the disease, shown a s water already in the tub), and mortality (shown as water leaving the tub). MORTALITY Figure 6 The bathtub analogy. The bathtub analogy describes the relationship between incidence (newly diagnosed cases shown as water flowing into the tub), prevalence (number of cases having the disease, shown as water already in the tub), and mortality (shown as water leaving the tub). I NTRO DUCT I ON 15 1 6 Hanna Wetterberg I NTRO DUCT I ON 15 EEppiiddeemmioilooglyo ogf yde omfe ndteiam entia The prevalence of dementia is low among those younger than 65, ranging 1.0- 1.5% in 60-64-year-olds4,8 and is most often related to familial dementia.87 The prevalence then increases rapidly with age, doubling every five years.6,7 Among octogenarians, the reported prevalence has ranged between 10- 35%.8,9,88,89 Also, the incidence of dementia increases almost exponentially with age and doubles every 6 years, from 4 per 1,000 person-years at age 60- 64 to 105 per 1,000 person-years at age 90+.90 The estimated incidence of dementia among octogenarians reported from various studies also varies, from 27/1,000 up to 70/1,000.16,18,91,92 The wide range of estimates is likely to a great extent due to varying methods and diagnostic criteria applied,10 however, the prevalence and incidence of dementia also vary between regions. A meta-analysis presented age- standardized prevalence among those aged >85 years of 24-26% in Europe, Asia, and North Africa, 33% in Latin America, and 9-16% in Sub-Saharan Africa regions.93 In the recent decade, numerous population-based studies indicate that the prevalence and incidence of dementia vary over time and between birth cohorts, with a trend of declining rates in North America and Europe, both in prevalence88,89,94,95 and incidence.15-17,19,20,92 Problematically, some studies indicate an increase in East Asian and African countries, such as Japan,96 China,97 as well as in Nigeria.92 These increases have been suggested to be associated with the rapid increase in cardiovascular risk factors.86 If the declining prevalence and incidence of dementia in North America and Europe results from the age of onset being pushed into higher ages, or if the decline remains into the high ages, such as above 85 years, is still unknown. There are some studies reporting estimates for octo- and nonagenarians, but they are scarce. In Paper IV, we specifically investigate the time trends of prevalence and incidence of dementia among 85-90-year-olds. Dementia strongly influences life expectancy.98-100 The population attributable risk (PAR) of death from dementia was 31% in men and 50% in women in a cohort of 85-year-olds born 1901-02 examined within the Gothenburg H70 studies (H70).101 An onset of dementia at higher ages has a lower effect on survival time than an onset at younger ages, predicting shorter survival of dementia.99,102,103 However, among those with late-onset dementia, a higher age predicts shorter survival,100 as do more severe dementia.102 Reported average survival time varies from three to 12 years, and a systematic review showed that most studies find survival times of seven to ten years.99 16 H AN N A W E T T E R B E R G I t is difficult to compare results from studies as it is often uncertain whether Itnhter osdurcvtiiovna l time should be calculated from the onset of symptoms or th1e7 time of diagnosis.102 As described in a previous section, the relationship between the incidence and prevalence of a disease is related to mortality. With reported time trends in the incidence of dementia, it is interesting to study potential changes in the mortality of dementia. As the length of survival in dementia is not only related to the age of onset but also midlife sociodemographic factors, and cardiovascular risk factors (i.e., factors that are modifiable), it is likely that the length of survival in dementia is not static but can fluctuate or change over time. There are, however, few studies on time trends in mortality in dementia,86 and even fewer among octogenarians. In Paper III, we aimed to examine if there were any time trends in the mortality of dementia among octogenarians, as well as the importance of dementia in relation to other common diseases to predict mortality.49 Potential biases in epidemiological studies of dementia Being defined as “the study of the distribution and determinants of disease frequency,” the field of epidemiology might be perceived as straightforward. However, epidemiological studies come with various methodological issues that are important to consider when interpreting results.85 If a study result is biased, it means that the estimated association, such as the risk ratio, odds ratio, or hazards ratio, deviates from the true association in the population. When performing epidemiological studies, two major types of systematic errors can affect the validity of the study, i.e. selection bias and measurement bias.85 Selection bias occurs when the participants and the population of interest I NTRO DUCT I ON 17 ssuurrvviivvaall ooff ddeemmeennttiiaa..9999,,110022,,110033 HHoowweevveerr,, aammoonngg tthhoossee wwiitthh llaattee--oonnsseett ddeemmeennttiiaa,, aa hhiigghheerr aaggee pprreeddiiccttss sshhoorrtteerr ssuurrvviivvaall,,110000 aass ddoo mmoorree sseevveerree ddeemmeennttiiaa..110022 RReeppoorrtteedd aavveerraaggee ssuurrvviivvaall ttiimmee vvaarriieess ffrroomm tthhrreeee ttoo 1122 yyeeaarrss,, aanndd aa ssyysstteemmaattiicc rreevviieeww sshhoowweedd tthhaatt mmoosstt ssttuuddiieess ffiinndd ssuurrvviivvaall ttiimmeess ooff sseevveenn ttoo tteenn yyeeaarrss..9999 IItt iiss ddiiffffiiccuulltt ttoo ccoommppaarree rreessuullttss ffrroomm ssttuuddiieess aass iitt iiss oofftteenn uunncceerrttaaiinn wwhheetthheerr tthhee ssuurrvviivvaall ttiimmee sshhoouulldd bbee ccaallccuullaatteedd ffrroomm tthhee oonnsseett ooff ssyymmppttoommss oorr tthhee ttiimmee ooff ddiiaaggnnoossiiss..110022 AAss ddeessccrriibbeedd iinn aa pprreevviioouuss sseeccttiioonn,, tthhee rreellaattiioonnsshhiipp bbeettwweeeenn tthhee iinncciiddeennccee aanndd pprreevvaalleennccee ooff aa ddiisseeaassee iiss rreellaatteedd ttoo mmoorrttaalliittyy.. WWiitthh rreeppoorrtteedd ttiimmee ttrreennddss iinn tthhee iinncciiddeennccee ooff ddeemmeennttiiaa,, iitt iiss iinntteerreessttiinngg ttoo ssttuuddyy ppootteennttiiaall cchhaannggeess iinn tthhee mmoorrttaalliittyy ooff ddeemmeennttiiaa.. AAss tthhee lleennggtthh ooff ssuurrvviivvaall iinn ddeemmeennttiiaa iiss nnoott oonnllyy rreellaatteedd ttoo tthhee aaggee ooff oonnsseett bbuutt aallssoo mmiiddlliiffee ssoocciiooddeemmooggrraapphhiicc ffaaccttoorrss,, aanndd ccaarrddiioovvaassccuullaarr rriisskk ffaaccttoorrss ((ii..ee..,, ffaaccttoorrss tthhaatt aarree mmooddiiffiiaabbllee)),, iitt iiss lliikkeellyy tthhaatt tthhee lleennggtthh ooff ssuurrvviivvaall iinn ddeemmeennttiiaa iiss nnoott ssttaattiicc bbuutt ccaann fflluuccttuuaattee oorr cchhaannggee oovveerr ttiimmee.. TThheerree aarree,, hhoowweevveerr,, ffeeww ssttuuddiieess oonn ttiimmee ttrreennddss iinn mmoorrttaalliittyy iinn ddeemmeennttiiaa,,8866 aanndd eevveenn ffeewweerr aammoonngg ooccttooggeennaarriiaannss.. IInn PPaappeerr IIIIII,, wwee aaiimmeedd ttoo eexxaammiinnee iiff tthheerree wweerree aannyy ttiimmee ttrreennddss iinn tthhee mmoorrttaalliittyy ooff ddeemmeennttiiaa aammoonngg ooccttooggeennaarriiaannss,, aass wweellll aass tthhee iimmppoorrttaannccee ooff ddeemmeennttiiaa iinn rreellaattiioonn ttoo ootthheerr ccoommmmoonn ddiisseeaasseess ttoo pprreeddiicctt mmoorrttaalliittyy..4499 Potential biases in epidemiological sPPtooutteednnttiiiaaells bb oiiaafss eedsse iinnm eeppniiddteeimmaiioollooggiiccaall ssttuuddiieess ooff ddeemmeennttiiaa BBeeiinngg ddeeffiinneedd aass ““tthhee ssttuuddyy ooff tthhee ddiissttrriibbuuttiioonn aanndd ddeetteerrmmiinnaannttss ooff ddiisseeaassee ffrreeqquueennccyy,,”” tthhee ffiieelldd ooff eeppiiddeemmiioollooggyy mmiigghhtt bbee ppeerrcceeiivveedd aass ssttrraaiigghhttffoorrwwaarrdd.. HHoowweevveerr,, eeppiiddeemmiioollooggiiccaall ssttuuddiieess ccoommee wwiitthh vvaarriioouuss mmeetthhooddoollooggiiccaall iissssuueess tthhaatt aarree iimmppoorrttaanntt ttoo ccoonnssiiddeerr wwhheenn iinntteerrpprreettiinngg rreessuullttss..8855 IIff aa ssttuuddyy rreessuulltt iiss bbiiaasseedd,, iitt mmeeaannss tthhaatt tthhee eessttiimmaatteedd aassssoocciiaattiioonn,, ssuucchh aass tthhee rriisskk rraattiioo,, ooddddss rraattiioo,, oorr hhaazzaarrddss rraattiioo,, ddeevviiaatteess ffrroomm tthhee ttrruuee aassssoocciiaattiioonn iinn tthhee ppooppuullaattiioonn.. WWhheenn ppeerrffoorrmmiinngg eeppiiddeemmiioollooggiiccaall ssttuuddiieess,, ttwwoo mmaajjoorr ttyyppeess ooff ssyysstteemmaattiicc eerrrroorrss ccaann aaffffeecctt tthhee vvaalliiddiittyy ooff tthhee ssttuuddyy,, ii..ee.. sseelleeccttiioonn bbiiaass aanndd mmeeaassuurreemmeenntt bbiiaass..8855 SSeelleeccttiioonn bbiiaass ooccccuurrss wwhheenn tthhee ppaarrttiicciippaannttss aanndd tthhee ppooppuullaattiioonn ooff iinntteerreesstt differ in non-random ways. This can happen if, for example, the procedure to select individuals for the study influences participation.85 It can also occur if there are non-random differences in who accepts or refuses participation in the study. There are several characteristics that are known to be associated with the choice to participate in studies, such as being healthier (the healthy I NTRO D I vNoTlRuOnDte UeCrT I OeNff UCT I ON ect),104-107 being married,106-109 having higher education an11d77 socioeconomic status,104,108,109 being younger,106,110 having lower mortality rates,104,106 and being an immigrant.107,109 In studies of older adults, specific factors might affect recruitment. These include sensory deficits such as 1h8e aring and visual impairment, lowering the willingness to be interviewed or Hanna Wetterberg tested. Cognitive slowing or dementia could make understanding the invitation information or the study procedures difficult. Multiple comorbidities, common in high age groups, often involve frequent hospitalizations making individuals difficult to contact or averse to further examinations.111 In epidemiological studies of dementia, the risk of selection bias is prominent due to differences in study participation, attrition, or survival in individuals with and without cognitive deficits.112 Previous studies show that cognitive decline causes attrition,110,113 as do frailty and illness.110 To answer questions of the distribution of disorders, or as in this thesis, time trends of dementia, it is important that the sample examined is representative of the target population.114 In Paper I, we further explore the selection bias in the H70 studies. Measurement bias is introduced into a study when the information collected within the study is wrong, or the measurement of the key study variable is inaccurate.85 If cases are placed in the wrong category, this bias is called misclassification. In studies of dementia, this is a potential bias that could affect the validity of the dementia diagnoses. Studies have, for example, shown that the use of brief cognitive assessment commonly used within the clinical setting often leads to misclassification of dementia. For example, higher education in participants can cause false-negative misclassification, and visual impairment can cause false-positive classification.115 As the classification often relies on reports from the key informant interview, recall bias also poses a challenge. If the key informants of those with cognitive decline report differently than those without due to a higher awareness of the symptoms, the differences between the groups could be inflated.85 Participants in a 18 H AN N A W E T T E R B E R G differ in non-random ways. This can happen if, for example, the procedure to select individuals for the study influences participation.85 It can also occur if there are non-random differences in who accepts or refuses participation in the study. There are several characteristics that are known to be associated with the choice to participate in studies, such as being healthier (the healthy volunteer effect),104-107 being married,106-109 having higher education and socioeconomic status,104,108,109 being younger,106,110 having lower mortality rates,104,106 and being an immigrant.107,109 In studies of older adults, specific factors might affect recruitment. These include sensory deficits such as hearing and visual impairment, lowering the willingness to be interviewed or tested. Cognitive slowing or dementia could make understanding the invitation information or the study procedures difficult. Multiple comorbidities, common in high age groups, often involve frequent hospitalizations making individuals difficult to contact or averse to further examinations.111 In epidemiological studies of dementia, the risk of selection bias is prominent due to differences in study participation, attrition, or survival in individuals with and without cognitive deficits.112 Previous studies show that cognitive decline causes attrition,110,113 as do frailty and illness.110 To answer questions of the distribution of disorders, or as in this thesis, time trends of dementia, it is important that the sample examined is representative of the target population.114 In Paper I, we further explore the selection bias in the H70 studies. Measurement bias is introduced into a study when the information collected within the study is wrong, or the measurement of the key study variable is inaccurate.85 If cases are placed in the wrong category, this bias is called misclassification. In studies of dementia, this is a potential bias that could affect the validity of the dementia diagnoses. Studies have, for example, shown that the use of brief cognitive assessment commonly used within the clinical setting often leads to misclassification of dementia. For example, higher education in participants can cause false-negative misclassification, and visual impairment can cause false-positive classification.115 As the classification often relies on reports from the key informant interview, recall bias also poses a challenge. If the key informants of those with cognitive decline report differently than those without due to a higher awareness of the symptoms, the differences between the groups could be inflated.85 Participants in a longitudinal study could also be misclassified due to practice effects as a result of repeated administration of the same test resulting in higher performance.112 Besides these biases, the different diagnostic criteria used within the field by default classify cases differently.79 In Paper II, we compare the prevalence of 1d8e mentia based on which diagnostic criteria are used, incluHdAiNnNgA t hWeE TtwT EoR mB EoRsGt recent editions: the DSM-5 and th e ICD-11. Introduction 1 9 I NTRO DUCT I ON 19 02 AIMS Aims 0A2ims A IMS The overarching aim of this thesis was to study the epidemiology of dementia, wThiteh oav epraarrctihciunlgar a ifmo coufs t hoisn thtiemsies wtraesn tdos stiund tyh teh ei necpiiddeenmcieo, lopgrye voafl ednecme,e natniad, wmiothrt aali tpy aortfi cduelmare nfoticau, sa nodn m tiemtheo dtroelnodgsic ainl asthpee citns criedgeanrcdein, gp rdeavtaal ecnocllee,c tainond amnodr tdailaitgyn oosf tdicesm. entia, and methodological aspects regarding data collection and diagnostics. Specific aims of the included papers Specific aims of the included papers Specific aims of the included papers Paperr I I RPeappreesr eIn tativeness of the population being studied is essential in Repeipdreemseionltoatgiivceanl estsus dieosf whtheen thpeo apiumla itsi oton debsecirnibge tshteu ddiiestdr ibiust ioens soefn dtiisael asien, eanpdid seemleioctlioognic baila sst ucdanie sa fwfehcetn t hthee i natimer pisr etota dtieosnc roibf er etshuel tdsi.s Ttrhibeu atiimon o off Pdaispeeasr eI, wanads steol edcetisocnri bbeia tsh cea nc raofsfse-cste tchtieo innatle rspamretpalteiso no fo ft hree s1u9lt3s0. -Tchoeh oairmt o off tPhae pHer7 0I swtuasd iteos , daensdcr itboe etxhaem cirnoes st-hsee cdtiioffnearle nsacmesp bleest woef etnh ep 1ar9t3ic0i-pcaonhtos,r tr eoffu stahles , Han7d0 sstaumdeie-as,g eadn din tdoiv eidxuaamlsi nine Gthoet hdeifnfberuerngc (etsh eb etatrwgeeet np oppaurtliactiipoann)t as,n dre Sfuwseadlse, na. nd same-aged individuals in Gothenburg (the target population) and Sweden. Paper I II I TPhaepree ra IrIe mainly two diagnostic systems to diagnose dementia, the ICD and Tthhee DreS aMre, mboatihn lwy ittwho s edviaegranlo esdtiicti osynsst.e Pmrse vtioo udsia sgtnuodsiees d heamveen sthiao, wthne t hICatD th aensde ethdeit iDonSsM v,a bryo tinh twhietihr cselavsesrifailc eadtiiotinosn, sy.i ePldreinvigo duisf fsetruedniet se shtaimvea tsihoonws onf t dheamt tehnetsiea epdreitvioalnesn vcea.r y Tinh tihs eirm calaksessi ficcaotmiopnasr, iysioenldsi nbg edtiwffeeernen ts etustdimiesa tiounsisn ogf ddeimffeernetniat pdiraegvnaolesnticce s. ysTtehmiss pmroakbelesm actoicm. pIna rtihsoe nlass t btwetow eedeinti osntsu odfie dsi agunsoinsgti c dsiyfsfteermenst, edfiafogrntos shtiacv sey bsteeemn sm pardoeb tleom readtiucc. eIn t hthise p lraostb tlwemo .e Tdoit ioounrs konf odwialgendogest, inc os yssttuedmiess, ceoffmorptsa rhea vthee b neeenw emsta deed ittoio rneds uince b tohtish psryosbtelemms:. Tthoe oIuCrD kn-1o1w alenddg eth, ne oD sStuMdi-e5s. cTohme paaimre othf eP napewere sItI ewdaitsi otnos c ionm bpoatrhe sfyivsete dmifsf:e trhene tI eCdDit-io1n1 sa onfd ththee I DCDSM an-5d. TDhSeM a ismys otefm P, aapse wr eIlIl awsa tsh teo c clionmicpala rceo nfisveen dsuifsf edrieangtn oedsiisti obnasse odf otnh et hICe D SaMnd- IDIIS-MR usysestde mwi,t hasin w tehlel aHs 7t0h es tculdinieicsa. l consensus diagnosis based on the DSM- III-R used within the H70 studies. 20 H AN N A W E T T E R B E R G 20 H AN N A W E T T E R B E R G Aims 2 3 Paperrr IIII II II Dementtiia iiss one off tthe ssttrrongesstt prrediicttorrss off morrttalliitty among ollderr adullttss.. Mean lliiffe expecttancy hass iincrreassed glloballlly durriing tthe passtt decadess.. Howeverr,, whettherr lliiffe expecttancy hass iincrreassed among tthosse wiitth dementtiia iiss nott cllearr.. The aiim off Paperr III wass tto examiine iiff tthe eiightt--yearr morrttalliitty hass changed bettween ttwo cohorrttss off 85--yearr--olldss borrn 22 yearrss aparrtt.. A ssecondarry aiim wass tto examiine iiff tthe popullattiion attttrriibuttablle rriissk off deatth due tto dementtiia iin rrellattiion tto ottherr common diissorrderrss hass changed.. Paperrr IIV I V Prreviiouss ssttudiiess have iindiicatted a declliine iin dementtiia prrevallence and iinciidence among ollderr adullttss iin Norrtth Amerriica and Eurrope.. Howeverr,, ffew ssttudiiess have examiined iiff tthiiss declliine perrssiissttss iintto hiigherr agess.. Morreoverr,, among avaiillablle publliisshed worrk,, tthe rressullttss arre iinconcllussiive.. The aiim off Paperr IV wass tto examiine iiff tthe prrevallence and iinciidence off dementtiia have changed bettween tthe cohorrttss off 85--yearr--olldss borrn 30 yearrss aparrtt.. A ssecondarry aiim wass tto examiine tthe ssenssiittiiviitty and sspeciiffiiciitty off dementtiia diiagnossess iin tthe rregiisstterrss ((ii..e.. tthe Nattiionall IInpattiientt Regiisstterr [[IIPR]] and tthe Causse off Deatth Regiisstterr [[CDREG]])),, whiich wass ussed fforr tthosse llosstt tto ffollllow--up.. AA II MSS 2211 2 4 Hanna Wetterberg Aims 2 5 03 METHODS AND MATERIALS M03ethMoEdTsH aOnDdS m AaNteDr iMalAsT ERIALS Parrttiicciippaanntsts This thesis is based on data from the population-based Gothenburg H70 birth cohort studies (H70) and the Prospective Population Study of Women (PPSW). The H70 studies are longitudinal multidisciplinary studies initiated in 1971 by Professor Alvar Svanborg to examine the general health status of 70-year-olds in Gothenburg.116 To yield representative samples, the selection of individuals was systematic and based on pre-specified birth dates of each month. The participants were followed up continuously with examinations and medical records retrieval. New waves of 70-year-olds have been added throughout the decades to update the knowledge of health status and to facilitate cohort comparisons (Figure 7). In the autumn of 2022, the seventh and most recent wave of 70-year-olds started, examining a cohort born in 1952-53. In 2009, a cohort of 85-year-olds was added to compare the health of octogenarians of today with the previous cohort. The study procedures have been kept as similar as possible throughout the decades to ensure the study of time trends related to age-related disorders and their risk and protective factors.117 Changes to the study protocol have mainly been restricted to adding new instruments and questionnaires. In this thesis, we used data from the cohorts born in 1901-02, 1923-24, and 1930. The PPSW is also a population-based longitudinal multidisciplinary study, starting in 1968 under the lead of Professor Calle Bengtsson.118 Five age groups were selected, with individuals sampled based on birth dates, similar to the H70 studies. The cohort has been followed with examinations and medical records for over 50 years, regardless of residence, and individuals have also been added throughout the decades.68,118 One of the selected age groups was born in 1930, with selection birth dates overlapping the fifth wave of 70-year-olds in H70, examined in 2000. The two studies were therefore combined, and thus, participants from the PPSW are also included in this thesis. See Box 3 for an overview of which cohorts are included in each paper. 22 H AN N A W E T T E R B E R G Methods and Materials 2 9 Biirrtht cho hcoorth 19o0r1-t0 21 9 01-02 The birth cohort of 1901-02 was the first cohort examined within the H70 studies, and, thus also the first cohort of 85-year-olds to be examined. In 1985, all individuals born between July 1, 1901 and June 30, 1902 and registered as living in the municipality of Gothenburg were invited to a health examination (n=1502).119 They were all given, consecutively after the date of birth, a number between 1-5 or 11-15. Those with numbers 1, 2, 11, 12, or 14 were further selected for the psychiatric examination (n=826). From these, 43 died before the examination, leaving an effective sample of 783 individuals, among which 14 (1.8%) had moved or could not be traced, 229 (29.2%) declined participation in the studies altogether, and 46 (5.9%) declined the neuropsychiatric examination. This resulted in a final sample of 494 individuals (143 men and 351 women) and a response rate of 64.4%. IFnig tuhree f7i rsTt hfeo Glloowth-eunbpu, r2g4 H87 808 B-iyrteha rc-oohlodrst sptuadrtieics.i pated (response rate 70.7%, 185 wThoem figeunr ea snhdo w6s3 t hme eexna)m. iInna taiodnd years on the y-axis, the birth cohort year on the x-axis, and the target age of the participants at eitxiaomni,n attwioenl vweit hpianr tthice ifpigaunrets. T(h7e wcoohmoretsn i nacnludd e5d min ethnis) wthehsois dareec hliingheldig hthteed bwaitshe lbilnacek p bsoyxcehs.i aStoruicrc e:x Oamrigiinalt ipoicnt uprea rctriecaitpeadt beyd Tinho tmhaes fMolalroloww-, umpo diefixeadm anind aptuioblnis.h eIdn b yt Mhee llqsveicsot Fnäds sbfeorlgl oetw a-l.u 2p0,1 91,15205 a n9d0 f-uyrtehaerr- moloddsi fiepda rbtyi caiupthaoter.d (response rate 61.0%, 117 women and 38 men). In addition, 45 participants wBhiroth ecitohhEeorx armdt iencaltiionne yde arobaseline took p19ar0t1 a-t0 t2h r were not part of the psychiatric examination at Birth e age of 90, giving a total number of 200 participants. Thyee abr irth cohort of 1901-02 was the first cohort examined within the H70 1901-02 70 - 75 79 - 81 82 83 - 85 - 88 - 90 - 92 - - 95 97 - 99 100 101 102 103 104 105 stu1d90i3es, and, thus also the first cohort of 85-year-olds to be examined. I95n - 97 - 99 100 101 102 103 104 105 106 107 108 109 B19ir819t50h4, caollh ionrdt i1v9id2u3a-ls2 4b o rn between July 1, 1901 and June 30, 1902 and 100 101 102 103 104 105 106 107 1081905 99 100 101 102 103 104 105 106 107 reg19is06t-0e7red as living i7n0 th- e m75 un- ici-pali-ty 7o9f Gothenburg were invited to a health 95 - 97 - 99 100 101 102 103 104 105 106 107 Texha1e9m0 8binirathti o6c0no h(-no=rt66 1o5f0 -129)7.221139- -2T4h w-eya sw- ienr-vei tae-lld g fi-ovre nt-h, ec- ofinrss- et ctui-mtiev84 ealty - tahfet- earg -teh oe- fd 8a5-te. Ao- ft - 92 - 95 - 97 - 99 100 101 102 103 104 105 106 the19 0b9ase 97 - 99 100 101 102 103 104 105bir1t9h10, a nlinume ebxearm bientawtieoenn i1n- 250 o0r8 -1210-1105,. iTndhiovsied uwailtsh b nournm Jbuelrys 1 1, ,1 29,2 31,1 t, o1 2Ju, noer 100 101 102 103 104 105 2194, 1 w911e9-1r2e4 f, uornth dera tseesl eecntdedin fgo7 0wr tiht-he 17p2,s y3c-, h5i,a- 7tr,i c-o re x796a, maninda rtieognis (tner=e8d2 r6e)s.i Fdents in the 100 101 102 103 104 105 mu19n14icipali5t4y o- f G60ot-hen66bu-rg, -we-re i-nvi-ted- to -a h-ealt-h e-xa7m8 in-atio- n - rom- th- ese- , - 86 - - - - 91 - - - 95 - - - - - 101 43 191d5-1i6ed before the examination, leaving an 7e5ffe-ctiv- e - sam- p8l(0eN =o1f 01738)3. Finod1r9it1v8yi din1922 u dali54s v0,i damu- aol5sn6 dg i-ewdh 6ib2cehf- o1r4e- (t1h.-e8 %e-x)a hm-aidn -amtioo-vne, d-1 9o -rc ocou- uldl7d 4n no-ot ts pb- eea t-kr aSc-wede,d- 2is2h-9, - 82 - - - - 87 - - - 91 - - - - - 97 - - 100 f(2o9u19.r22 3-%h24a 6 )d deemcilg - ra5t2ed -ou5t8sid-e S- - - - - - - - - 70 -ined participatiowne dienn ,t hane ds stuixd iceosu ladl tnooget tbhee rt,race - d, -lea-vin- - - 78 - - - - 83 - - - 87 - - - - - 93 - - 96 and 46 (5.g9 %an) 85 - 88 - 90 - 95 - - - - - 97 edfefc1e9l3ci0ntievde tsh3a8em n-pelue4 r4oofp -s9y4c540h iian-tdriicv- iedxu-aamls-i,n oa-fti own-h. iTc-hh i5s- 7r1e- s(u6l-0te.5d6% 2in, -a3 5f9i-n awl- osmam-enp- lea no-df - 70 - - - - 75 - - - 79 - - - - - 85 - - 88 - - 90 2 194192419 44 5mi2n-5e3dni)v tiodouka lpsa (r1t 4in3 tmhee sntu adnyd. I 3n5 t1h ew fiorsmt feonll)o awn-du pa, 3re2s2p 8o8n-yseea rra-oteld os fp 6ar4t.i4c%ipa. ted 70 - - - - 75 - - - 70 - ( response rate 73.5%, 205 women and 117 men). In the second follow-up, 2F5ig0u r9e0 7- yeTahr-eo Gldosth penabrtuircgi pHa7t0e dB i(rtrhe scpohoonrst es truadtiees .7 3.7%, 160 women, 90 men). The figure shows the examination years o n the y-axis, the birth cohort year on the x-axis, and the target age of the participants at examination within the figure. The cohorts included in this thesis are highlighted with black boxes. Source: Original picture created by Thomas Marlow, modified and published by Mellqvist Fässberg et al. 2019,120 and further modified by author. MET HO DS AND MAT ERI AL S 23 3 0 Hanna Wetterberg 24 H AN N A W E T T E R B E R G MET HO DS AND MAT ERI AL S 23 1968 1971 1974 1976 1980 1981 1982 1983 1984 1985 1986 1987 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 IInn tthhee ffiirrsstt ffoollllooww--uupp,, 224488 8888--yyeeaarr--oollddss ppaarrttiicciippaatteedd ((rreessppoonnssee rraattee 7700..77%%,, 118855 wwoommeenn aanndd 6633 mmeenn)).. IInn aaddddiittiioonn,, ttwweellvvee ppaarrttiicciippaannttss ((77 wwoommeenn aanndd 55 mmeenn)) wwhhoo ddeecclliinneedd tthhee bbaasseelliinnee ppssyycchhiiaattrriicc eexxaammiinnaattiioonn ppaarrttiicciippaatteedd iinn tthhee ffoollllooww-- uupp eexxaammiinnaattiioonn.. IInn tthhee sseeccoonndd ffoollllooww--uupp,, 115555 9900--yyeeaarr--oollddss ppaarrttiicciippaatteedd ((rreessppoonnssee rraattee 6611..00%%,, 111177 wwoommeenn aanndd 3388 mmeenn)).. IInn aaddddiittiioonn,, 4455 ppaarrttiicciippaannttss wwhhoo eeiitthheerr ddeecclliinneedd oorr wweerree nnoott ppaarrtt ooff tthhee ppssyycchhiiaattrriicc eexxaammiinnaattiioonn aatt bbaasseelliinnee ttooookk ppaarrtt aatt tthhee aaggee ooff 9900,, ggiivviinngg aa ttoottaall nnuummbbeerr ooff 220000 ppaarrttiicciippaannttss.. BBiiirrrtthht cchoo hhcooorrtt h1199o22r33t--22 441 9 23-24 TThhee bbiirrtthh ccoohhoorrtt ooff 11992233--2244 wwaass iinnvviitteedd ffoorr tthhee ffiirrsstt ttiimmee aatt tthhee aaggee ooff 8855.. AAtt tthhee bbaasseelliinnee eexxaammiinnaattiioonn iinn 22000088--22001100,, iinnddiivviidduuaallss bboorrnn JJuullyy 11,, 11992233,, ttoo JJuunnee 2299,, 11992244,, oonn ddaatteess eennddiinngg wwiitthh 11,, 33,, 55,, 77,, oorr 99,, aanndd rreeggiisstteerreedd rreessiiddeennttss iinn tthhee mmuunniicciippaalliittyy ooff GGootthheennbbuurrgg,, wweerree iinnvviitteedd ttoo aa hheeaalltthh eexxaammiinnaattiioonn ((NN==11001133)).. FFoorrttyy iinnddiivviidduuaallss ddiieedd bbeeffoorree tthhee eexxaammiinnaattiioonn,, 1199 ccoouulldd nnoott ssppeeaakk SSwweeddiisshh,, ffoouurr hhaadd eemmiiggrraatteedd oouuttssiiddee SSwweeddeenn,, aanndd ssiixx ccoouulldd nnoott bbee ttrraacceedd,, lleeaavviinngg aann eeffffeeccttiivvee ssaammppllee ooff 994444 iinnddiivviidduuaallss,, ooff wwhhiicchh 557711 ((6600..55%%,, 335599 wwoommeenn aanndd 221122 mmeenn)) ttooookk ppaarrtt iinn tthhee ssttuuddyy.. IInn tthhee ffiirrsstt ffoollllooww--uupp,, 332222 8888--yyeeaarr--oollddss ppaarrttiicciippaatteedd ((rreessppoonnssee rraattee 7733..55%%,, 220055 wwoommeenn aanndd 111177 mmeenn)).. IInn tthhee sseeccoonndd ffoollllooww--uupp,, 225500 9900--yyeeaarr--oollddss ppaarrttiicciippaatteedd ((rreessppoonnssee rraattee 7733..77%%,, 116600 wwoommeenn,, 9900 mmeenn)) . . Examination year Birth year 1901-02 70 - 75 79 - 81 82 83 - 85 - 88 - 90 - 92 - - 95 97 - 99 100 101 102 103 104 105 1903 95 - 97 - 99 100 101 102 103 104 105 106 107 108 109 1904 100 101 102 103 104 105 106 107 108 1905 99 100 101 102 103 104 105 106 107 1906-07 70 - 75 - - - 79 95 - 97 - 99 100 101 102 103 104 105 106 107 1908 60 - 66 - 72 - - - - - - - - - - 84 - - - - - - - 92 - 95 - 97 - 99 100 101 102 103 104 105 106 1909 97 - 99 100 101 102 103 104 105 1910 100 101 102 103 104 105 1911-12 70 - 72 - - - 76 100 101 102 103 104 105 1914 54 - 60 - 66 - - - - - - - - - - 78 - - - - - - - 86 - - - - 91 - - - 95 - - - - - 101 1915-16 75 - - - - 80 1918 50 - 56 - 62 - - - - - - - - - - 74 - - - - - - - 82 - - - - 87 - - - 91 - - - - - 97 - - 100 1922 46 - 52 - 58 - - - - - - - - - - 70 - - - - - - - 78 - - - - 83 - - - 87 - - - - - 93 - - 96 1923-24 85 - 88 - 90 - 95 - - - - - 97 1930 38 - 44 - 50 - - - - - - - - - - 62 - - - - - - - 70 - - - - 75 - - - 79 - - - - - 85 - - 88 - - 90 1944 70 - - - - 75 - - - 1952-53 70 - 24 H AN N A W E T T E R B E R G 24 H AN N A W E T T E R B E R G Methods and Materials 3 1 1968 1971 1974 1976 1980 1981 1982 1983 1984 1985 1986 1987 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 Biirrtht cho hcoorth 19o3r0t 1930 The 1930 birth cohort is more complex since it is used 1) as five cross- sectional studies with a new sampling at each wave (with one exception), 2) as two longitudinal studies (from age 70 and from age 75, where the latter included a large baseline sample), and 3) in combination with the PPSW. At all examination years, the cross-sectional inclusion criteria were based on pre- specified birth dates each month and residential addresses within the municipality of Gothenburg. The cross-sectional samples at ages 70, 75, 79, 85, and 88 are described in detail in Paper I.121 The longitudinal inclusion criteria were based on individuals previously examined within the H70 studies or PPSW, even if they had moved outside of Gothenburg (applied in the H70 studies from the examination year 2009). At the 85-year examination in 2015-2017, individuals born on dates ending with 2, 3, 5, 6, 11, 12, 16, 18, 20, 21, 24, 27, or 30, and registered residents in the municipality of Gothenburg were invited to a health examination (N=764). Forty-two individuals died before the examination, 31 could not speak Swedish, six could not be traced, and 13 were not invited due to technical issues (misclassified as not living in Gothenburg), leaving an effective sample of 672 individuals, of which 416 (61.9%, 251 women and 165 men) took part in the study. In addition, 102 individuals (effective sample n=95) previously examined within the PPSW or H70 living in other parts of Sweden were invited to a follow-up study, of which 75 (78.9%) participated. At age 88, based on the cross-sectional inclusion criteria, of the 555 individuals invited (effective sample n=505), 258 individuals participated (51.1%, 162 women, and 96 men). In addition, 99 individuals (effective sample n=82) previously examined within the PPSW or H70, living in other parts of Sweden, were invited to the longitudinal study, of which 75 (52.4%) participated. This cohort was planned to be examined at age 90 during 2020, but due to the Covid-19 pandemic, the wave was postponed and is thus not part of this thesis. MET HO DS AND MAT ERI AL S 25 3 2 Hanna Wetterberg BBoxo x3 .3 .S uSmummarayr yo fo tfh teh es tsutduyd yd edseisgingsn sin i nth teh ep appaepresr s PaPpaepre rD Deseigsing n BiBrtihrt hco choohrotsr ts AgAeg e OOutucotcmome e I I CrCorsos-ssse-sceticotnioanl al H7H07: 01:9 13903 0 707, 07,5 7, 5, DiDffiefrfeenrecnecs ebse btweteweene n 797, 98,5 8, 5, paprtairctiipcaipnatns,t sr,e fruefsuersse,r s, anadn 8d8 8 8 samsaem-aeg-aegde idn dinivdiidvuidaulsa lisn in GoGtohtehnebnubrugr agn adn Sdw Sewdeedne. n. III I CrCorsos-ssse-sceticotnioanl al H7H07 a0n adn d 858 5 NuNmubmebr eorf o dfe dmeemnetniat icaa cseass es PPPSPWSW: 1:9 12932-3- byb dyi fdfiefrfeenretn ctl acslsaisfsicifaictiaotnio n 242, 41,9 13903 0 sysstyesmtesm. s. IIIII I CrCorsos-ssse-sceticotnioanl,a tli,m tiem e H7H07: 01:9 10910-102-0, 2, 858 5 CoChoohrot rdti fdfiefrfeenrecnecse isn in tretnredn d 1912932-234-2 4 momrotarltiatyli toyf o dfe dmeemnetniat.i a . IVI V CrCorsos-ssse-sceticotnioanl,a l, H7H07: 01:9 10910-102-0, 2, 858, 58,8 8, 8, CoChoohrot rdti fdfiefrfeenrecnecse isn in lonlognitguitduindainl,a tli,m tiem e 1912932-234-2, 41,9 13903 0 909 0 prpevreavleanlecnec aen adn idn cinidceidnecnec e tretnredn d ofo dfe dmeemnetniat.i a . SStutuduydd yyp op ppoouplupalutaiotlinoasnt sbi yob ypn apspa beprey r paper InI nP aPpaepre rI ,I w, we eu suesde dc rcorsoss-ss-escetciotinoanl adl adtaat afr formom th teh e1 913903 0b ibrtihrt hc ochoohrot rat ta fti vfiev e exeaxmaminiantaiotino nw wavaevse, su, suesde das a csr corsoss-ss-escetciotinoanl adl adtaat. aA. As tsh teh ree rseesaeracrhc hq uqeusetsiotino nw wasa s tot oe xeaxmaminien eth teh er erperperseesnetnattaivtievneensess so fo tfh teh eH H707 0s tsutduidesie, sw, we ein icnlculduedde do nolnyl yth teh e paprat rot fo tfh teh esa smamplpel ere rseidsiidnign gin i nth teh em munuinciicpiapliatlyit yo fo Gf Gotohtehnebnubrugr.g . InI nP aPpaepre rI II,I w, we eu suesde dd adtaat afr formom th teh e1 912932-32-42 4a nadn d1 913903 0b ibrtihrt hc ochoohrotsrt sa ta tth teh e exeaxmaminiantaiotinosn ast a atg aeg e8 58.5 I.n I nth tihs ips appaepre, rt,h teh ere rseesaeracrhc hq uqeusetsiotinosn ws wereer feo fcoucsuesde do no n dedmemenetniat iad idagiangonsotsicti ct otoolos,l s,m makaiknign gt htihs iss asmamplpe lew withit ha ah ihgihg hp rperveavlealnecnec eo fo f dedmemenetniat ias usiutaitbalbel. eT. oT oin icnrceraesaes et hteh es asmamplpel es izsiez, ew, we ein icnluclduedde d7 57 5in idnidviivdiudaulsa ls frformom th teh elo lnognigtuitduidnianl asl asmamplpe len on olo lnognegre rr erseidsiidnign gin i nG Gotohtehnebnubrugr.g T. hTeh es asmame e exeaxmaminiantaiotino nb abtatettreyr yw wasa as papplpielide din i nb obtoht hg rgoruopusp. s . InI nP aPpaepre rI IIII,I w, we eu suesde dd adtaat afr formom th teh e1 910910-10-20 2a nadn d1 912932-32-42 4b ibrtihrt hc ochoohrotsrt sa ta t thteh ea gaeg eo fo 8f 58.5 A. As sth teh ere rseesaeracrhc hq uqeusetsiotino nw wasa sto t oe xeaxmaminien e8 -8y-eyaera mr morotartliatlyit, yw, we e onolnyl yin icnlculduedde dth teh fei rfsirts ttw two oco choohrotsrt os fo 8f 58-5y-eyaera or lodlsd ass a ssu sfufficfiiceinetn tti mtime sei nsicnec eth teh e exeaxmaminiantaiotino nh ahda dp apsassesde.d . InI nP aPpaepre Ir VIV, w, we eu suesde dd adtaat afr formom th teh e1 910910-10-20,2 1, 912932-32-42,4 a, nadn d1 913903 0c ochoohrotsrt sa ta t thteh ea gaegse s8 58,5 8, 88,8 a, nadn d9 09.0 I. nI nth tihs isp appaepre, rt,h teh ere rseesaeracrhc hq uqeusetsiotino nw wasa sto t oe xeaxmaminien e timtime et retrnednsd si ni nb obtoht ht hteh ei nicnicdiednecnec ea nadn dp rperveavlealnecnec eo fo fd edmemenetniat iaa mamonogn g octogenarians, which is why we used a longitudinal sample from age 85 and cross-sectional samples at all ages. Due to the covid-19 pandemic, the 1930 cohort could not be examined at the age of 90. 262 6 H AHNANNAN AW EWTE T T E R B E R G VMe athroidas banlde Msat earinalsd outcome measures T E R B E R G 3 3 After the systematic selection was made in each cohort, the potential respondents were approached with an invitation letter to participate in a health examination. A few days later, a research nurse or administrative staff member contacted them by telephone. The examination battery in the H70 studies is extensive. Although the exact included examination parts have to some extent, varied over the decades, the examination has always included semi-structured health interviews, neuropsychiatric examinations, physical examinations, and psychometric testing.117 The semi-structured health interviews included questions regarding somatic and psychiatric disorders, social and sociodemographic factors, medications, lifestyle factors, and ADL/iADL. The physical examinations included anthropometric measures, blood pressure, blood sampling, lung function, and electrocardiogram (ECG). In addition, the participants in H70 were invited to a range of additional examinations such as ophthalmologic and hearing examinations, diet history interview, key informant interview, MRI and CT scanning, DXA scanning, and dental examination. Although the examinations have been kept as similar as possible to ensure comparability over time, some changes have been implemented that are worth noting. In cohort 1901-02, the examination had three major parts.119 The first part consisted of a nurse home visit to the participant to collect basic information through semi-structured interviews. The participant was then in the second part examined at an outpatient department of the Geriatric hospital, which included a physical examination by a geriatrician, neuropsychological tests administered by a psychologist, and laboratory tests such as ECG, blood tests, and chest X-ray.116,122 In the final and third step, a psychiatrist made a home visit to perform the neuropsychiatric examination. In cohorts 1923-24 and 1930, the examination was generally performed in one step where the basic information, health examination, and MET HO DS AND MAT ERI AL S 27 octogenarians, which is why we used a longitudinal sample from age 85 and cross-sectional samples at all ages. Due to the covid-19 pandemic, the 1930 cohort could not be examined at the age of 90. Vaarriiaabbleless a anndd o uotuctocmome me emaesuarseusr e s After the systematic selection was made in each cohort, the potential respondents were approached with an invitation letter to participate in a health examination. A few days later, a research nurse or administrative staff member contacted them by telephone. The examination battery in the H70 studies is extensive. Although the exact included examination parts have to some extent, varied over the decades, the examination has always included semi-structured health interviews, neuropsychiatric examinations, physical examinations, and psychometric testing.117 The semi-structured health interviews included questions regarding somatic and psychiatric disorders, social and sociodemographic factors, medications, lifestyle factors, and ADL/iADL. The physical examinations included anthropometric measures, blood pressure, blood sampling, lung function, and electrocardiogram (ECG). In addition, the participants in H70 were invited to a range of additional examinations such as ophthalmologic and hearing examinations, diet history interview, key informant interview, MRI and CT scanning, DXA scanning, and dental examination. Although the examinations have been kept as similar as possible to ensure comparability over time, some changes have been implemented that are worth noting. In cohort 1901-02, the examination had three major parts.119 The first part consisted of a nurse home visit to the participant to collect basic information through semi-structured interviews. The participant was then in the second part examined at an outpatient department of the Geriatric hospital, which included a physical examination by a geriatrician, neuropsychological tests administered by a psychologist, and laboratory tests such as ECG, blood tests, and chest X-ray.116,122 In the final and third step, a psychiatrist made a home visit to perform the neuropsychiatric examination. In cohorts 1923-24 and 1930, the examination was generally performed in one step where the basic information, health examination, and neuropsychiatric examination were performed either at the outpatient clinic or in the participants’ home, if requested by the participant. However, as the examination took about six hours to perform, the participants were offered to divide the examination into two or more parts to reduce the burden for those who wished. MET HO DS AND MAT ERI AL S 27 Also, in cohort 1901-02, the same psychiatrist (Ingmar Skoog, today PI of the H70 studies) performed all neuropsychiatric examinations at the baseline examination. In the more recent cohorts, the neuropsychiatric examinations were performed by experienced psychiatric research nurses that were trained by Ingmar Skoog. The kappa value for the agreement in rating symptoms and 3s4i gns of dementia between psychiatrists and nurses in the H7H0a nsntau Wdieetste hrbaevreg been reported to be high (kappa values 0.74-1.00).88 Neuropsychiatric examinations and key informant interviews The neuropsychiatric examinations included assessments of psychiatric and cognitive symptoms according to the Comprehensive Psychopathological Rating Scale (CPRS),123 structured assessments of clinical symptoms and signs of dementia, the Gottfries-Bråne-Steen Scale (GBS),124 and the Clinical Dementia Rating score (CDR).125 The participants performed several tests of mental functioning, such as short and long-term memory (naming the current and former Prime minister of Sweden, remembering objects shown earlier in the interview), orientation (identifying the current place and time), abstract thinking (understanding proverbs), aphasia (naming objects), verbal functioning (Word Fluency, naming objects), apraxia (following commands, perform how one would send a letter to oneself in five steps), agnosia (naming fingers), constructional difficulties and executive abilities (copying drawings of shapes), and complex attention (mental arithmetic, correctly identifying the number of letters in a list read out with both letters and numbers). These tests are part of the Mini-Mental State Examination (MMSE)126 and the Alzheimer’s Disease Assessment Scale Cognitive Subscale – ADAS-Cog,127 as well as a few tests specific to the H70 study protocol. At every examination, the participants were asked to name a close relative or friend that would be able to conduct a comprehensive key informant 28 H AN N A W E T T E R B E R G nneeuurrooppssyycchhiiaattrriicc eexxaamiinnaattiioonn weerree ppeerrffoorrmeedd eeiitthheerr aatt tthhee oouuttppaattiieenntt cclliinniicc oorr iinn tthhee ppaarrttiicciippaannttss’’ hhoomee,, iiff rreeqquueesstteedd bbyy tthhee ppaarrttiicciippaanntt.. Hooweevveerr,, aass tthhee eexxaamiinnaattiioonn ttooookk aabboouutt ssiixx hhoouurrss ttoo ppeerrffoorrm,, tthhee ppaarrttiicciippaannttss weerree ooffffeerreedd ttoo ddiivviiddee tthhee eexxaamiinnaattiioonn iinnttoo ttwoo oorr moorree ppaarrttss ttoo rreedduuccee tthhee bbuurrddeenn ffoorr tthhoossee whhoo wiisshheedd.. Allssoo,, iinn ccoohhoorrtt 11990011--0022,, tthhee ssaamee ppssyycchhiiaattrriisstt ((IInnggmaarr SSkkoooogg,, ttooddaayy PPII ooff tthhee H7700 ssttuuddiieess)) ppeerrffoorrmeedd aallll nneeuurrooppssyycchhiiaattrriicc eexxaamiinnaattiioonnss aatt tthhee bbaasseelliinnee eexxaamiinnaattiioonn.. IInn tthhee moorree rreecceenntt ccoohhoorrttss,, tthhee nneeuurrooppssyycchhiiaattrriicc eexxaamiinnaattiioonnss weerree ppeerrffoorrmeedd bbyy eexxppeerriieenncceedd ppssyycchhiiaattrriicc rreesseeaarrcchh nnuurrsseess tthhaatt weerree ttrraaiinneedd bbyy IInnggmaarr SSkkoooogg.. Thhee kkaappppaa vvaalluuee ffoorr tthhee aaggrreeeemeenntt iinn rraattiinngg ssyymppttoomss aanndd ssiiggnnss ooff ddeemeennttiiaa bbeettweeeenn ppssyycchhiiaattrriissttss aanndd nnuurrsseess iinn tthhee H7700 ssttuuddiieess hhaavvee bbeeeenn rreeppoorrtteedd ttoo bbee hhiigghh ((kkaappppaa vvaalluueess 00..7744--11..0000))..8888 Neuropsychiatric examinations aNneeuudrr ookppessyycc ihhniiaafttrroiiccr eemxxaamniitnn aaittniiootnness raavnnddie kkweeyys iinnffoorrmaanntt iinntteerrvviieewss Thhee nneeuurrooppssyycchhiiaattrriicc eexxaamiinnaattiioonnss iinncclluuddeedd aasssseessssmeennttss ooff ppssyycchhiiaattrriicc aanndd ccooggnniittiivvee ssyymppttoomss aaccccoorrddiinngg ttoo tthhee Coompprreehheennssiivvee PPssyycchhooppaatthhoollooggiiccaall Raattiinngg SSccaallee ((CPPRSS)),,112233 ssttrruuccttuurreedd aasssseessssmeennttss ooff cclliinniiccaall ssyymppttoomss aanndd ssiiggnnss ooff ddeemeennttiiaa,, tthhee Goottttffrriieess--Brråånnee--SStteeeenn SSccaallee ((GBSS)),,112244 aanndd tthhee Clliinniiccaall Deemeennttiiaa Raattiinngg ssccoorree ((CDR))..112255 Thhee ppaarrttiicciippaannttss ppeerrffoorrmeedd sseevveerraall tteessttss ooff meennttaall ffuunnccttiioonniinngg,, ssuucchh aass sshhoorrtt aanndd lloonngg--tteerrm meemoorryy ((nnaamiinngg tthhee ccuurrrreenntt aanndd ffoorrmeerr PPrriimee miinniisstteerr ooff SSweeddeenn,, rreemeembbeerriinngg oobbjjeeccttss sshhoownn eeaarrlliieerr iinn tthhee iinntteerrvviieew)),, oorriieennttaattiioonn ((iiddeennttiiffyyiinngg tthhee ccuurrrreenntt ppllaaccee aanndd ttiimee)),, aabbssttrraacctt tthhiinnkkiinngg ((uunnddeerrssttaannddiinngg pprroovveerrbbss)),, aapphhaassiiaa ((nnaamiinngg oobbjjeeccttss)),, vveerrbbaall ffuunnccttiioonniinngg ((Woorrdd FFlluueennccyy,, nnaamiinngg oobbjjeeccttss)),, aapprraaxxiiaa ((ffoolllloowiinngg ccoommaannddss,, ppeerrffoorrm hhoow oonnee woouulldd sseenndd aa lleetttteerr ttoo oonneesseellff iinn ffiivvee sstteeppss)),, aaggnnoossiiaa ((nnaamiinngg ffiinnggeerrss)),, ccoonnssttrruuccttiioonnaall ddiiffffiiccuullttiieess aanndd eexxeeccuuttiivvee aabbiilliittiieess ((ccooppyyiinngg ddrraawiinnggss ooff sshhaappeess)),, aanndd ccoompplleexx aatttteennttiioonn ((meennttaall aarriitthhmeettiicc,, ccoorrrreeccttllyy iiddeennttiiffyyiinngg tthhee nnuumbbeerr ooff lleetttteerrss iinn aa lliisstt rreeaadd oouutt wiitthh bbootthh lleetttteerrss aanndd nnuumbbeerrss)).. Thheessee tteessttss aarree ppaarrtt ooff tthhee Miinnii--Meennttaall SSttaattee Exxaamiinnaattiioonn ((MMSSE))112266 aanndd tthhee Allzzhheeiimeerr’’ss Diisseeaassee Asssseessssmeenntt SSccaallee Cooggnniittiivvee SSuubbssccaallee –– ADASS--Coogg,,112277 aass weellll aass aa ffeew tteessttss ssppeecciiffiicc ttoo tthhee H7700 ssttuuddyy pprroottooccooll.. Att eevveerryy eexxaamiinnaattiioonn,, tthhee ppaarrttiicciippaannttss weerree aasskkeedd ttoo nnaamee aa cclloossee rreellaattiivvee oorr ffrriieenndd tthhaatt woouulldd bbee aabbllee ttoo ccoonndduucctt aa ccoompprreehheennssiivvee kkeeyy iinnffoorrmaanntt interview. Questions during the interview included changes in memory, behaviour, personality, mood, language, psychiatric symptoms, intellectual functioning, and if the key informant reported prevalent dementia, questions regarding the age of onset and disease course were asked.128 2288 HH AANN NN AA WW EE TT TT EE RR BB EE RR GG A telephone interview at age 85 in cohort 1901-02 was performed with 451 key informants (91%) with a median length of the interview of 31 minutes (min 9, max 95). In the 1923-24 cohort, an interview was performed with 439 informants (77%), and the median length was 45 minutes (min 5, max 180). In cohort 1930, an interview was performed with 340 informants (82%), and the median length was 65 minutes (min 15, max 200). The key informants were often a spouse or a child (Box 4). Methods and Materials 3 5 MET HO DS AND MAT ERI AL S 29 Box 4. Characteristics of key informant interviews Cohort 1901-02 1923-24 1930 Number 450 (91%) 439 (77%) 340 (82%) Median interval between examination and interview 6 (0-12) 10 (0-33) 2 (0-17) Months, (min-max) Median length Minutes, (min-max) 31 (9-95) 45 (5-180) 70 (15-200) Key informant Spouse 7% 23% 26% Child 58% 63% 58% Living together 12% 21% 25% Participant MMT, median (95 % CI) Interview performed 27 (26-27) 27 (27-28) 28 (28-28) No key informant interview 28 (28-29) 28 (27-28) 28 (27-29) 3 6 Hanna Wetterberg 30 H AN N A W E T T E R B E R G Deemmeenntiat idaia gdnioasgisn osis Dementia diagnoses used in the H70 studies are based on the DSM-III-R criteria,83 using data from the neuropsychiatric examination and key informant interview. An important difference between the H70 diagnose, and the DSM-III-R criteria are that in the H70 diagnose, impairment in short- term or long-term memory is enough to fulfil the criteria of impairment in memory, whereas in the DSM-III-R criteria, it is mandatory with impairment in both short-term and long-term memory. The clinical expertise decided this diverging from the original criteria, finding the criteria too strict.129 This decision was strengthened by the release of the DSM-IV, where the same change had been applied.84 Similarities and differences between the different diagnostic tools and the clinical consensus diagnosis are further examined in Paper II. The procedure for classifying dementia cases has been kept identical over the cohorts to ensure comparability (see Figure 8). First, an algorithm based on the neuropsychiatric examination and the key informant interview produced two separate diagnoses, respectively. The symptoms included in the algorithm had to attain a level causing significant difficulties in social life to generate an indication of dementia. Second, the information from both sources were combined, and at least two clinical experts separately reviewed the output of the algorithms. In cases where the algorithm output was inconclusive, the research file was reviewed in full. To classify a dementia case, four levels of ascertainment were followed: 1. Dementia, according to both the psychiatric examination and the informant interview 2. Dementia, according to either the psychiatric examination or the informant interview, supported by the other examination 3. Dementia, according to the informant interview, confirmed by MMSE 4. In cases of no informant interview, severe dementia, according to the neuropsychiatric examination In the third and last step, the clinical experts held a consensus conference to decide on the final diagnoses. In 2019, to ensure comparability over the cohorts, the diagnoses for cohorts M19E0T H1O-0D2S aAnNdD 1M9A2T3E-R2I A4L wS ere re-evaluated during the same time-period in whic3h1 t Mheeth doidasg annod sMinatge roiafl sc ohort 1930 was initiated. The only change made was on3e7 case of dementia in the 1923-24 cohort that was re-coded to no dementia.49 In Paper III, we also classified the severity of dementia and etiological subgroups based on the likely causes of dementia. Probable or possible AD was diagnosed in accordance with the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s disease and Related Disorders Association (NINCDS-ADRDA) criteria.130 We diagnosed vascular dementia based on the National Institute of Neurological Disorders and Stroke–Association Internationale pour la Recherché et l’Enseignement en Neurosciences (NINDS-AIREN) criteria,131 meaning a temporal connection (within one year) between the first symptoms of dementia and a stroke or transient ischaemic attack (TIA). If there was no clear temporal connection between the onset of dementia and a stroke/TIA, we diagnosed mixed dementia. The information on stroke/TIA was based on self-report and key informant interviews and was only diagnosed when clear focal neurological symptoms (such as aphasia or hemiparesis) were reported, with a duration of symptoms of <24h for TIA and >24h for stroke.88 We also used information from the IPR, which has shown to have good sensitivity and specificity of stroke.132 A few cases were diagnosed with other causes when other disorders in temporal connection to the dementia onset and of sufficient degree to cause dementia were identified. Other causes included NPH, alcohol dementia, Parkinson’s disease, brain tumour, head injury, and unspecified dementia when no clear aetiology was identified.88 MET HO DS AND MAT ERI AL S 33 DATA COLLECTION DATA CLEANING BUILDING ALGORITHMS • One using data from neuropsychiatric tests and assessments • One based on key informant interviews REVIEW OF RESULTS FROM ALGORITHMS BY TWO CLINICAL EXPERTS, DIAGNOSIS MADE ON THE BASIS OF FOUR STEPS 1. Dementia, according to both the examination and the informant interview. 2. Dementia, according to either the examination or the informant. 3. Dementia, according to the informant interview, confirmed by MMSE. 4. Severe dementia, accordning to the psychiatic examination; no informant interview. CONSENSUS CONFERENCE BETWEEN THE TWO EXPERTS Figure 8 Workflow of setting dementia diagnoses according to the DSM-III-R. The workflow of setting the clinical consensus diagnoses has been kept identical in all cohorts included in the H70-studies. 3 8 Hanna Wetterberg 32 H AN N A W E T T E R B E R G In the third and last step, the clinical experts held a consensus conference to decide on the final diagnoses. In 2019, to ensure comparability over the cohorts, the diagnoses for cohorts 1901-02 and 1923-24 were re-evaluated during the same time-period in which the diagnosing of cohort 1930 was initiated. The only change made was one case of dementia in the 1923-24 cohort that was re-coded to no dementia.49 In Paper III, we also classified the severity of dementia and etiological subgroups based on the likely causes of dementia. Probable or possible AD was diagnosed in accordance with the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s disease and Related Disorders Association (NINCDS-ADRDA) criteria.130 We diagnosed vascular dementia based on the National Institute of Neurological Disorders and Stroke–Association Internationale pour la Recherché et l’Enseignement en Neurosciences (NINDS-AIREN) criteria,131 meaning a temporal connection (within one year) between the first symptoms of dementia and a stroke or transient ischaemic attack (TIA). If there was no clear temporal connection between the onset of dementia and a stroke/TIA, we diagnosed mixed dementia. The information on stroke/TIA was based on self-report and key informant interviews and was only diagnosed when clear focal neurological symptoms (such as aphasia or hemiparesis) were reported, with a duration of symptoms of <24h for TIA and >24h for stroke.88 We also used information from the IPR, which has shown to have good sensitivity and specificity of stroke.132 A few cases were diagnosed with other causes when other disorders in temporal connection to the dementia onset and of sufficient degree to cause dementia were identified. Other causes included NPH, alcohol dementia, Parkinson’s disease, brain tumour, head injury, and unspecified dementia when no clear aetiology was identified.88 MET HO DS AND MAT ERI AL S 33 Methods and Materials 3 9 Severity of dementia was determined based on the criteria in the DSM-III-R.83 • Mild dementia: social activities were significantly impaired, but the capacity for independent living remained with adequate personal hygiene and relatively intact judgment. • Moderate dementia: independent living would be hazardous, and some degree of supervision would be necessary • Severe dementia: ADL would be so impaired that supervision would be continually required, such as minimal personal hygiene would be unable to maintain or the person being incoherent or mute. The classification system for diagnosing dementia is used slightly differently in the four papers included in this thesis (see Box 5). In Paper I, dementia was one of the variables used for investigating selection bias. In this paper, we collected dementia diagnoses from the IPR, hence using the ICD-10 classification system. In Paper II, the aim was to compare the different classification systems, which is why five different editions of the systems were included (DSM-III-R, DSM-IV, DSM-5, ICD-10, and ICD-11), as well as the clinical consensus diagnosis. In Paper III, we only used the clinical consensus diagnosis based on the DSM-III-R (as described above). In Paper IV, we mainly used the clinical consensus diagnosis and added register data from the IPR and CDREG for cases lost to follow-up. As data from the first cohort was collected in the late 1980s, the register data covers three versions of the ICD: 8, 9, and 10. Box 5. The classification systems used in the papers 34 H AN N A W E T T E R B E R G 4 0 Hanna Wetterberg Severity of dementia was determined based on the criteria in the DSM-III-R.83 • Mild dementia: social activities were significantly impaired, but the capacity for independent living remained with adequate personal hygiene and relatively intact judgment. • Moderate dementia: independent living would be hazardous, and some degree of supervision would be necessary • Severe dementia: ADL would be so impaired that supervision would be continually required, such as minimal personal hygiene would be unable to maintain or the person being incoherent or mute. The classification system for diagnosing dementia is used slightly differently in the four papers included in this thesis (see Box 5). In Paper I, dementia was one of the variables used for investigating selection bias. In this paper, we collected dementia diagnoses from the IPR, hence using the ICD-10 classification system. In Paper II, the aim was to compare the different classification systems, which is why five different editions of the systems were included (DSM-III-R, DSM-IV, DSM-5, ICD-10, and ICD-11), as well as the clinical consensus diagnosis. In Paper III, we only used the clinical consensus diagnosis based on the DSM-III-R (as described above). In Paper IV, we mainly used the clinical consensus diagnosis and added register data from the IPR and CDREG for cases lost to follow-up. As data from the first cohort was collected in the late 1980s, the register data covers three versions of the ICD: 8, 9, and 10. Box 5. The classification systems used in the papers Paper I ICD 10 PaperP Ia per II ICD 1C0l inical consensus diagnosis (DSM-III-R) DSM-III-R Paper II ClinicDal ScMon-IsVen sus diagnosis (DSM-III-R) 34 DSM-DIISIM-R- 5 H AN N A W E T T E R B E R G DSM-IICVD -10 DSM-I5C D-11 ICD-10 Paper III ICD-1C1l in ical consensus diagnosis (DSM-III-R) PaperP IaIpI er IV ClinicCall icnoicnasle cnosnuse dnisaugsn odsiaisg n(DosSisM (-DIISIM-R-)I II-R) ICD-8 Paper IV ClinicIaCl Dco-n9s ensus diagnosis (DSM-III-R) ICD-8IC D-10 ICD-9 ICD-10 Register data RegiIsints eatrd edriatit oadn a toa the data collected during the examinations, we used register data in different ways in Paper I, Paper III and Paper IV. In addition to the data collected during the examinations, we used register data iInn dPifafpereern It, wwaey cso ilnle cPtaepd edra tIa, Pona pdearte IsI oI fa dneda tPha fproemr I tVh.e Swedish Tax Agency, demographics from Statistics Sweden, and data on discharge diagnoses from In Patphee rN I,a wtioe ncaoll lBecotaerdd doaft Ha oenal tdha taensd o Wf deelafathre f r(othme IthPeR S) wfoerd pisahr tTicaixp aAngtse,n rceyf,u sals, demoagnrda pshaimcse f-raogmed Sitnadtiisvtiidcsu aSlws eind eGn,o atnhde ndbautar go nan ddis cinh aSrwgee ddeiang.n Toose sp frrootemc t the the Naantoionnyaml Bityo aorfd t ohfe Hpaeratlitchip aanndt sW ase lwfaerlel a(tsh teh IeP rRef)u fsoarl sp, atrhteic dipaatan tws,a rse rfeucseailvse, d on and saanm aeg-gargeegda tiendd lievvideul.a ls in Gothenburg and in Sweden. To protect the anonymity of the participants as well as the refusals, the data was received on an agIgnre Pgaatpede rl eIvIeIl,. dates of death from the Swedish Tax Agency and data on discharge diagnoses from the IPR were used for comparisons between In Papparetric IipIaIn, tdsa atnesd onfo nd-epaatrht icfripoamn tsth. Te hSew deidsicshha rTgea xd iAaggneonsceys awnedr ed aaltsao ounse d to dischcaorgme pdleimagennots eths e firnofmor mthaeti oInP Ron wdeisreea suesse din cflourd ecdo minp tahreis omnosd eblest wpreeedni cting particmipoarnttasli tayn. d non-participants. The discharge diagnoses were also used to complement the information on diseases included in the models predicting mortaInlit Py.a per IV, we used register data to collect information on the date of death from the Swedish Tax Agency. We used the IPR and the CDREG to collect In Paipnfeorr ImVa, twioen u osned i nrecgidisetnetr cdaasteas t oo fc doellmecetn intifao irnm tahtoiosne olons tt-htoe dfoatlleo owf- duepa, tahs well froma tsh teo S ewxeadmisinhe T tahxe Asegnesnitcivy.i tWy aen uds sepde tchifei cIiPtyR o afn dde mtheen CtiDa dRiaEgGno tsoe sc oinll ethcte IPR informanadt ioCnD oRnE iGnc. ident cases of dementia in those lost-to follow-up, as well as to examine the sensitivity and specificity of dementia diagnoses in the IPR and CDREG. Methods and Materials 4 1 MET HO DS AND MAT ERI AL S 35 MET HO DS AND MAT ERI AL S 35 Data analyses Data analyses Box 6. Analyses used in Paper I Aim: To describe the representativeness of participants in relation to three levels of representativeness; refusals, same-aged individuals in Gothenburg, and same-aged individuals in Sweden. Variables: Sociodemographic variables included: marital status, highest attained educational level, country of birth, paid labour, and average monthly income. The hospital discharge diagnoses included: cancer, alcohol-related-, and neuropsychiatric disorders (including dementia), cardiovascular-, ischemic heart-, cerebrovascular-, and chronic obstructive pulmonary diseases, diabetes mellitus, unipolar depression, and osteoarthritis. Survival time. Statistical methods: Persons Chi-square or Fisher’s exact test, independent samples t-test, Cox proportional hazards model (using age as time-scale) presented as hazard ratios (HR), and 95% confidence intervals (CI). Analyses: 1) The differences in response rates and the proportion excluded between the examination years. 2) Mortality in participants and refusals. 3) The difference in the prevalence of sociodemographic variables and the hospital discharge diagnoses between: a) Participants and refusals b) Participants and same-aged individuals in Gothenburg c) Participants and same-aged individuals in Sweden 4 2 36 H AHaNnNnAa WeEtTtTerEbReBrgE R G Box 7. Analyses used in Paper II Aim: To examine how the prevalence of dementia varies between different diagnostic tools, including the recent DSM-5 and ICD-11. Statistical methods: Persons Chi-square test, Cohen’s kappa coefficient, and McNemar’s test. Analyses: 1) Overlap between the different dementia diagnoses. 2) Agreement (Cohen’s kappa) between the different dementia diagnoses. 3) Difference in proportions of dementia cases. 4) Difference in the severity of dementia in the different dementia diagnoses. 5) Sensitivity analyses: a) Comparison of the prevalence of dementia according to the clinical consensus diagnoses in those with and without missing data. b) Investigating the impact of key informant interview by performing 1), 2), and 3) after excluding this information. Methods and Materials 4 3 MET HO DS AND MAT ERI AL S 37 Box 8. Analyses used in Paper III BoAxi m8.: TAon eaxlaymseins eu isfe tdh ein 8 P-yaepare rm IoIIr tality in relation to dementia Aamimon: gT 8o5 e-yxeaamr-ionled sif htahde c8h-ayneagre dm oovretarl ittwy oin c orehloatritosn b otorn d 2e2m yeenatrias aapmaortn. gW 8e5 -aylesaor -eoxladms hinaedd c thhaen gimedp oorvtearn tcwe oo fc odheomrtesn btioa rinn 2r2e layteiaorns atop aortth. eWr ed iaselsaos eesx taom pinreeddi ctth me iomrtpaolitryta. n ce of dementia in relation to other diseases to predict mortality. Statistical methods: Pearson’s Chi-square, independent samples St-ttaetsits, tMicaaln mn–eWthhoitdnse:y P eUa rsToens’ts, CKhaip-slqanu–arMe,e iinedr espuernvdiveanl t asnamalypsliess, at-ntdes Ct, oMx parnonp–oWrtihointnale yh aUza rTdse,s ut,s inKga tpimlane-–oMn-esiteurd ys uarsv aiv taiml ea-nsaclaylsei.s , and Cox proportional hazards, using time-on-study as a time-scale. Covariates: Dementia severity, cerebrovascular disorders, Ccoonvgaersitaivtees :h eaDrte mfaeinlutriae , dsieavbeertietys, mceelrlietbusro, vcahsrcounlaicr bdroisnocrhdietriss,, catorniagle sftiibvreil lahteioarnt, faanilguirnea, dpieacbteotreiss , mmeylloitcuasr,d icahl roinnfiacr cbtiroonn, chtoittiasl, acthroialle sfteibrroill,l atiocnh, oalensgtienrao l pecttroeraistm, emnyto, cardhiyapl erintefnarsciotino,n , toantadl chhyopleerstteenrsoiol, n trecahtomleesntet.r ol treatment, hypertension, and hypertension treatment. Models: Mod1e)l s:A ge and sex 12)) AAggee , asnedx ,s aenxd dementia 23)) AAggee,, sseexx,, aanndd ddeemmeennttiiaa severity 34)) AAggee,, sseexx,, adnedm denemtiae, natniad seedvuecriattyi onal level 45)) AAggee,, sesxex, ,d edmeemnetinat, iaa,n de edduuccaatitoionnaal l lleevveell , and relevant 5) Adigseea, sesse x(a, s dseelmecetnedti ab, y epdruimcaatriyo nanala lylesevse)l , and relevant diseases (as selected by primary analyses) Analyses: Analy1s) eDs: ifferences in mortality between men and women within 1) Deaicfhfe croenhcoerst .i n mortality between men and women within 2) eDaicfhfe croenhcoerst .i n mortality between those with and without 2) Ddeimffeernetniac ewsi tihni nm eoarctahl ictyo hboetrwt. een those with and without 3) dDeimffeernetniac ewsi tinh imn oearctahl itcyo bheotrwt. een the cohorts, total group 3) Danidff estrreantcifeies din b my soerxta, luitsyi nbge tmwoeedne ltsh 1e- c5o. horts, total group 4) aInntde rsatcratitoifnie odf b sye xs*ecxo, huosirnt ga nmdo eddeulsc a1t-i5o.n al level*cohort in 4) Irenltaetrioacnt itoon m ofo rsteaxl*itcyo. hort and educational level*cohort in 5) rDeliaffteiorenn tcoe sm ionr tamliotyr.t ality between cohorts stratified by 5) Ddeimffeernetniac essta tiuns ,m uosirntgal imtyo dbeeltsw 1e-e5n. cohorts stratified by 6) dTehme eenffteiac ts otaf teuasc, hu osifn tgh em ionddeelpse 1n-d5e. nt predictors for 8-year 6) Tmhoer teaflifteyc wt oasf ceaalcchu loatfe tdh aes i pnodpepuleantidoenn att ptrribeduitcatbolers r ifsokr ( 8P-AyeRa)r. mortality was calculated as population attributable risk (PAR). Box 9. Analyses used in Paper IV Box 9. Analyses used in Paper IV 4 4 38 H AHNaNnAna W WEeTttTeErRbeBrEgR G 38 H AN N A W E T T E R B E R G Box 8. Analyses used in Paper III Aim: To examine if the 8-year mortality in relation to dementia among 85-year-olds had changed over two cohorts born 22 years apart. We also examined the importance of dementia in relation to other diseases to predict mortality. Statistical methods: Pearson’s Chi-square, independent samples t-test, Mann–Whitney U Test, Kaplan–Meier survival analysis, and Cox proportional hazards, using time-on-study as a time-scale. Covariates: Dementia severity, cerebrovascular disorders, congestive heart failure, diabetes mellitus, chronic bronchitis, atrial fibrillation, angina pectoris, myocardial infarction, total cholesterol, cholesterol treatment, hypertension, and hypertension treatment. Models: 1) Age and sex 2) Age, sex, and dementia 3) Age, sex, and dementia severity 4) Age, sex, dementia, and educational level 5) Age, sex, dementia, educational level, and relevant diseases (as selected by primary analyses) Analyses: 1) Differences in mortality between men and women within each cohort. 2) Differences in mortality between those with and without dementia within each cohort. 3) Differences in mortality between the cohorts, total group and stratified by sex, using models 1-5. 4) Interaction of sex*cohort and educational level*cohort in relation to mortality. 5) Differences in mortality between cohorts stratified by dementia status, using models 1-5. 6) The effect of each of the independent predictors for 8-year mortality was calculated as population attributable risk (PAR). Box 9. Analyses used in Paper IV Aim: To examine if the incidence and prevalence of dementia among 85-year-olds decreased over three cohorts born 30 years apart. We also examined the sensitivity and specificity of the IPR 3a8n d CDREG. H AN N A W E T T E R B E R G Statistical methods: Logistic regression and Poisson regression with a natural log person-years follow-up offset term. Models: 1) Sex 2) Sex and educational level Analyses: 1) Comparing the prevalence of dementia at ages 85, 88 and 90 between the cohorts, using model 1-2. 2) Comparing the prevalence of dementia between men and women within the cohorts. 3) Comparing the difference in four-year incidence of dementia between the cohorts, using model 1-2. 4) Comparing the difference in four-year incidence of dementia between men and women within the cohorts. 5) Examining the sensitivity and specificity of dementia diagnoses in the IPR and CDREG. MeMthEoTdHsO aDnSd MAaNtDe rMiaAlTs ERI AL S 4 5 39 EEtthhiiccaall c coonnssidideerartaiotinosn s The Regional Ethical Review Board in Gothenburg has approved all studies. Two amendments for Paper I regarding the use of aggregated register data were approved by the Swedish Ethical Review Authority. Informed consent was obtained before the examinations according to the Helsinki declaration, primarily by the participants, still in some cases when this was not possible (e.g. due to severe dementia), a family member or close relative consented to the participation. The participant was informed before the examination of the expected duration of the examination, that lunch and a snack would be provided, a general overview of the content of the examination, as well as data management and storage. The participants were also informed on the potential risks (i.e. no risk, but potential discomfort during blood sampling), and the potential benefits (information on current health status regarding e.g. blood pressure, cholesterol, and glucose levels, ECG, hearing, and visual status). They were also informed that withdrawal from the study would be possible (including deletion of already collected data) at any time without declaring a reason and that declining participation would not affect their contact with health care. A medical doctor on-site reviewed the results from the examination to determine if there were potential medical implications. If no acute situation was identified, participants were referred to an appropriate clinic if previously unknown diseases or pathologies were detected. As the examinations were extensive, participation could be demanding, especially in the high-age groups included in this thesis. To minimize this burden, participants were offered home visits and to split up the examination into more than one time point. However, most participants took part in all examination parts, and the follow-ups had high response rates. 40 H AN N A W E T T E R B E R G 4 6 Hanna Wetterberg Methods and Materials 4 7 04 MAIN RESULTS M04ain r MAeIsNu lRtsE SULTS A ssummarry off tthe maiin rressullttss off tthe ffourr iinclluded paperrss iiss prressentted iin tthe ffollllowiing ssecttiionss.. To rread tthe ffullll rressullttss,, plleasse ssee tthe rre--prriintted publliicattiionss and manusscrriiptt att tthe end off tthe tthessiiss.. Maiiinn r reessuulltlst so fo Pf aPpaepre II r I Hanna Wetttterrberrg*,, Liina Rydén* ett all.. Reeprreesseenttattiivveeneessss iin poopullattiioon--basseed ssttudiieess ooff oolldeerr adullttss – Fiivvee wavveess ooff ccrroossss--sseeccttiioonall eexamiinattiioonss iin tthee Goottheenburrgg H70 Biirrtth Coohoorrtt Sttudyy.. BMJJ Open 2022;;12::e068165..112211 * HW and LR arre jjoiintt ffiirrsstt autthorrss.. IIn Paperr I,, we ffound tthatt tthe rressponsse rratte rranged ffrrom 51..1% tto 69..6% and wass hiigherr att age 70 and llowerr att age 88 comparred tto allll ottherr examiinattiionss.. The rressponsse rratte wass hiigherr among tthosse wiitth hiigherr educattiion and tthosse who werre marrrriied,, butt ttherre werre no diifffferrencess bettween men and women.. Morrttalliitty wass llowerr iin parrttiiciipanttss tthan rreffussallss,, and tthe prrevallence off a rrange off diissorrderrss iin tthe IIPR wass llowerr,, ssuch ass carrdiiovasscullarr diisseasse,, neurropssychiiattrriic--,, and allcoholl--rrellatted diissorrderrss.. The prrevallence off osstteoarrtthrriittiiss wass hiigherr iin parrttiiciipanttss,, tthey had hiigherr educattiionall llevell,, and werre morre offtten marrrriied,, comparred tto rreffussallss.. Comparred tto ssame--aged iindiiviiduallss iin Gotthenburrg,, tthe parrttiiciipanttss had hiigherr educattiionall llevellss and werre morre offtten borrn iin Sweden.. Att age 70,, tthey had a llowerr prrevallence off cerrebrrovasscullarr and neurropssychiiattrriic diissorrderrss iin tthe IIPR.. IIn comparriisson tto ssame--aged iindiiviiduallss iin Sweden,, tthe educattiionall llevell wass hiigherr iin parrttiiciipanttss,, tthey werre lleessss offtten borrn iin Sweden,, had hiigherr averrage iincome,, werre morre offtten diivorrced,, and had a llowerr prrevallence off a rrange off diissorrderrss,, ssuch ass cancerr,, carrdiiovasscullarr--,, cerrebrrovasscullarr--,, and iisschemiic heart disease. MAAIIaNi RnEE SSrUeLL TTsSSu lts of Paper II 4411 Hanna Wetterberg, et al. The effect of diagnostic criteria on dementia prevalence – A pMoapinu lraetsiuonlt-sb ased study from Gothenburg, Sweden. 5 1(Submitted). As Paper II includes unpublished results, the summary of results is condensed. The results show that the prevalence of dementia varies depending on the choice of the edition of the classification system. The classification with the lowest prevalence was based on the ICD-10 and the highest on ICD-11, followed by the DSM-5. The kappa values of agreement between the classification systems were overall high, with the highest being between the ICD-11 and DSM-5, and DSM-IV and the clinical consensus diagnosis, which is based on the DSM-III-R criteria. Although the agreement between ICD-11 and DSM-5 was high, small differences in the criteria yielded differences in the number of dementia cases. We also found that symptoms of cognitive decline, as well as reported concern regarding cognitive decline, were common in octogenarians. The most common symptom of cognitive decline was a “decline in other cognitive abilities characterized by deterioration in judgment and thinking” as defined by ICD-10 and “disturbance in executive functioning” as defined by DSM- IV. Less frequent were the symptoms of “impaired judgment” and “coarsening of social behaviour”, as defined by the DSM-III-R and DSM-IV. 42 H AN N A W E T T E R B E R G heart disease. h eart disease. Main results of Paper II MHaaninna Wreesttuerlbtesrg o, eft Pal.a Tphee ref feIcIt of diagnostic criteria on dementia prevalence – A popaulaitnio nr-beassedu sltutdsy foromf PGoathpeneburrg ,I ISweden. (Submitted). H anna Wetterberg, et al. The effect of diagnostic criteria on dementia prevalence – A populati on-based study from Gothenburg, Sweden. (Submitted). As Pa per II includes unpublished results, the summary of results is conden sed. The results show that the prevalence of dementia varies Adesp ePnadpinegr oInI tihnec lucdheosi ceu nopfu tbhlies heeddi tiorens uoltfs ,t hteh ec lassusmifimcaatrioy n osfy srtesmul.t sT hise colanssdiefnicsaetdio. n Twhieth rtehseu lltosw sehsto wpr evthaalet ncthe ew apsr ebvaasledn coen othf e dIeCmDe-n1t0ia anvda rtihese dheigpheensdt ionng IoCnD t-h1e1 ,c fhoolliocew eodf btyh et hed DitiSoMn -o5f. Tthe kclaapspsiaf ivcatluioens osfy satgermee. mTehnet cbleatswsiefeicna ttihoen cwlaisthsi ftihceat ilonw essyts tpermevs awleenrcee owvaesr ablla hseigdh o, nw itthe tIhCe Dhi-g1h0e astn db etihneg hbiegthweesetn o nth IeC IDC-D11-1, 1fo allnodw eDdS bMy -t5h, ea DndS MD-S5M. T-IhVe kaanpdp tah vea lculiensi coafl acgorneesemnesnuts bdeiatgwneoesni st, hweh ciclahs sisif ibcasteiodn o sny sthteem Ds SwMe-rIeI Io-vRe rcarlilt ehriigah. A, wltihtho utghhe thieg haegsrte ebmeienngt between ItCheD I-C11D a-n1d1 DanSdM D-5S wMa-s5 h, iagnhd, s mDaSlMl d-iIfVfe raenndc etsh ien ctlhine icrailt ecroian yseienldsuesd diaffgenroensicse, sw ihni cthe i sn buamsebde ro onf t hde mDeSnMtia-I cIIa-sRes c. r iteria. Although the agreement between ICD-11 and DSM-5 was high, small differences in the criteria yielded dWifef earelsnoc efso iunn tdh et hnautm sbyemr potfo dmems eonf ticao cgansietsiv. e decline, as well as reported concern regarding cognitive decline, were common in octogenarians. The Wmoe sta clsoom fmouond s ytmhaptt osmym opf tcoomgns itoivfe cdoegcnliintiev we ads eac “lidnec, lianse iwne ollt haesr creopgonritievde caobnilicteiersn crheagraarcdtienrgiz ecdo gbnyi tdiveete rdieocrlaintieo,n w ine rjeu dcgomemnot nan idn tohcintokginegn”a raias ndse. fTinheed mbyo IsCt cDo-m10m aond s y“mdipstuormb aonfc ceo ignn ietxiveec udteicvlein feu wncatsi oan “idnegc”l inase dine foitnheedr cboyg DniStiMve- aIVbi.l itiLees scsh afrraecqtuereinzte d wbeyr ed ettehreio rsaytimonp tionm jus dgomf en“itm apnadi rtehdin kjiundgg”m aesn dt”ef inaend b“cyo IaCrsDen-1in0g a onfd s o“dciastl ubrebhaanvcieo uinr” e, xaesc duetfivinee fdu bnyc ttihoen iDngS”M a-sI IdIe-Rfi naendd bDyS DMS-MIV-. I V. Less frequent were the sy mptoms of “impaired judgment” and “coarsening of social behaviour”, as defined by the DSM-III-R and DSM-IV. 42 H AN N A W E T T E R B E R G 42 H AN N A W E T T E R B E R G 5 2 Hanna Wetterberg Maaiinn rreessuultlst so of Pf aPpaepre IrII III Hanna Wetterberg, et al. Dementia remains the major predictor of death among octogenarians. A study of two population cohorts of 85-year-olds examined 22 years apart. European Journal of Epidemiology, 2021. 36(5): p. 507-517.49 We found that the median survival time among 85-year-olds increased from 4.9 (95% CI 4.4-5.5) years in cohort 1901-02 to 5.7 (95% CI 5.2-6.2) years in cohort 1923-24. Participants with dementia had higher mortality than those without in both cohorts, but the mortality decreased between cohorts among those with dementia after adjusting for dementia severity and common diseases (HR 0.7; 95% CI 0.5–0.99) (Table 1). The PAR of dementia for death was higher than the other diseases examined, such as cerebrovascular disorders, myocardial infarction, and congestive heart failure, in both cohorts, meaning that dementia was the most important predictor of death. However, the relative risk of death from dementia did not change between the cohorts. The mortality increased with the increasing severity of dementia. For the purpose of this thesis, the effect of the interaction between cohort and dementia severity on mortality was tested with Cox proportional hazards model. After adjusting for sex and age at baseline, there was no interaction between the dementia severity and cohort (p=0.846), indicating that mortality in relation to dementia declined in all severity groups. T able 1. Change in 8-year mortality between birth cohorts 1901–02 and 1923–24, stratified by dementia sta tus Model 1 Model 5 Deceased (%) HR (95% CI) HR (95% CI) Dementia at baseline Cohort 1901-02 95.2 1.0 (Ref.) 1.0 (Ref.) Cohort 1923-24 93.5 0.7 (0.5-1.0) 0.7 (0.5-0.99) Dementia-free at baseline Cohort 1901-02 69.2 1.0 (Ref.) 1.0 (Ref.) Cohort 1923-24 64.1 0.7 (0.5-0.9) 0.7 (0.5-0.9) Hazards ratios derived from Cox proportional hazards model. Bolded P-values and hazard ratios have a P-value < 0.05. Model 1: adjusted for age and sex. Model 5: adjusted for age, sex, baseline dementia severity (in the dementia group), education, and relevant diseases. See reprinted paper for Models 2-4. Source: Table 2 in Wetterberg et al. 2021.49 MAI N RESUL T S 43 Main results 5 3 44 H AN N A W E T T E R B E R G Maiinn r reessuultlst so of Pf aPpaepre IrV IV Hanna Wetterberg et al. Decreasing incidence and prevalence of dementia among octogenarians. A population-based study on three cohorts born 30 years apart. The Journals of Gerontology: Series A, 2023; glad071.133 When comparing 85-year-olds, the prevalence of dementia decreased from 29.8% in cohort 1901-02 to 21.5% in cohort 1923-24. The prevalence of 24.5% in cohort 1930 was not significantly different from cohort 1901-02 nor 1923-24. Among 88-year-olds, the prevalence decreased from 41.9% in cohort 1901-02 to 28.0% in cohort 1923-24 and 21.7% in cohort 1930. At age 90, the prevalence of dementia was 41.5% in cohort 1901-02 and 37.2% in cohort 1923-24, which did not represent a significant decline (cohort 1930 was not examined at age 90 due to the Covid-19 pandemic). We also found that the four-year cumulative incidence of dementia from age 85 declined from 49/1000 person-years in cohort 1901-02 to 23/1000 person-years in cohort 1930. The 38/1000 person-years incidence rate in cohort 1923-24 did not differ from cohort 1901-02 nor 1930. The decrease in prevalence and incidence of dementia was more accentuated among women. Women had a higher prevalence of dementia than men at age 88 in cohorts 1901-02 and 1923-24 but not in cohort 1930. The IPR and CDREG had moderate sensitivity and a high specificity, and was similar for all three cohorts. For this thesis, additional analyses were performed to investigate potential varying attrition bias between the cohorts. The response rate at follow-up among survivors in relation to dementia status at baseline was tested with logistic regression (Table 2). The response rate was lower among those with dementia in cohort 1930 compared to cohort 1901-02, but no differences between those without dementia were found. Also, the response rate by educational level was tested in a logistic regression, showing that individuals with a lower educational level more often were lost to follow-up in cohort 1923-24 than in 1930 (Table 2). MAI N RESUL T S 45 5 4 Hanna Wetterberg Table 2. Potential attrition bias at follow-up Response rate* Model 1 Model 2 OR (95% CI) OR (95% CI) Dementia at base line Cohort 1901-02 78 1.0 (Ref.) 2.3 (1.1-5.0) Cohort 1923-24 73 0.8 (0.4-1.7) 1.8 (0.8-3.9) Cohort 1930 60 0.4 (0.2-0.9) 1.0 (Ref.) Dementia- free at base l ine Cohort 1901-02 69 1.0 (Ref.) 0.8 (0.6-1.2) Cohort 1923-24 74 1.3 (0.9-1.9) 1.0 (0.7-1.5) Cohort 1930 73 1.2 (0.9-1.8) 1.0 (Ref.) Elementary educat ion Cohort 1901-02 70 1.0 (Ref.) 0.7 (0.5-1.1) Cohort 1923-24 65 0.8 (0.5-1.2) 0.6 (0.4-0.9) Cohort 1930 77 1.4 (0.9-2.2) 1.0 (Ref.) More than e lementary educat ion Cohort 1901-02 73 1.0 (Ref.) 1.1 (0.6-1.9) Cohort 1923-24 81 1.5 (0.9-2.7) 1.6 (1.0-2.6) Cohort 1930 72 0.9 (0.5-1.7) 1.0 (Ref.) Note. Odds ratio derived from logistic regression model, stratified first by dementia status and then by educational level. In model 1, cohort 1901-02 is used as reference. In model 2, cohort 1930 is used as reference, as this shows differences also between cohorts 1923-24 and 1930. Bolded P-values and hazard ratios have a P-value < 0.05. *Response rates at follow-up at age 88 46 H AN N A W E T T E R B E R G Main results 5 5 05 DISCUSSION Discussion 0D7iscussion D ISCUSSION In this thesis, methodological aspects of epidemiological studies of dementia, Ians wtheilsl tahse tsiims, em terethnodds oinlo tghicea el pasidpeemctiso olofg eyp oidfe dmeimoleongtiicaa, lh satvued ibese eonf sdteumdieendt iian, faos uwr elsle apsa rtiamtee ptraepnedrss .i nI nth eth eips idseemctiioonlo, gym oefth doedmoelongtiiac,a lh acvoen bsiedeenr asttiuodnise da rine dfoisucru ssseepda, raast ew eplal paes rtsh. e Isnt retnhgist hsse acntido nli,m mitaettihoonds oolfo tghicea iln ccloundseidd epraaptieornss. Tahree dseiscctiuosnse da,l saos wineclll uads etsh ea s tgreenngetrhasl adnisdc luimssiitoanti oonfs othf eth efi nindcilnugdse. dI pna ptehres . eTnhde, sseucgtgioesnt ioanlsso f oirn cfulutdurees rae segaerncehr adli redcisticounsss iaorne porfo pthoese dfi.n dings. In the end, suggestions for future research directions are proposed. Methodological discussion Meetthhooddoolologgicicaal dl idsicsucsussisoino n Seleecctiotni obinas b ias SIne pleocptuiolant iobnia-bsa sed studies that aim to describe the distribution of disease, sInel epcotipounl abtiioasn -pboasseesd a s tthudreieast ttoh atth ea imex tteor ndaels vcarilbidei ttyh oe fd tihsetr irbeusutil 85 otsn. o fS edliescetaiosen, sbeialesc ctioounl db icaasu psoe saens ais sthuere bato ttoh taht eb eaxsetelirnnea la vs awlidelilt ya so fin t hloe nrgeistuuldtsin.8a5 lS seelettcitnigosn. bInia ss tcuoduields ocafu tsiem aen t risesnudes ,b iot this aets pbeasceialilnlye iams pwoerltla nast itno lcoonngsitiudderin tahl es ertitsikn gosf. Idnif fsetruednite ss eolefc ttiiomne btiraesneds so, ri ta titsr ietisopne cbiaeltlwy eiemnp tohret acnoth toor tsc.o nsider the risk of dTihffee crehnoti csee lteoc tuisoen b biriathse ds aoter sa ttotr isteiolenc tb ientdwiveeidnu tahlse icno thhoer Hts7. 0 studies increases Tthhee c chhaoniccee otof ruesteri ebviritnhg daa rteeps rteos esenlteacttiv ine dsaivmidpulael.s I nin c tohme Hpa7r0is ostnu,d sioems ien csrtueadsieess othfe t icmhaen tcree nodfs r eintr ideevminegn at irae phraevsee nsatmatipvlee dsa imnd 15,19 pilvei.d Iuna clso fmropmar iesloenc,t osoraml ero sltlus dies oorf ptiamtiee n 16,134 135 trte nds ionr dinemsuernanticae h raevgei sstaemrs.ple dA innodtihveidr umalest hfroodm a pelpelcietodr ianl rtohlel sH 157,109 ostru pdaietsie tnot r16e,d13u4 coer tihnes urrisakn coef rseegleisctteiorsn.1 3b5i aAsn woaths etro mofeftehro hdo ampep lvieisdi tisn. Tthhei sH h7a0s pstruedviieosu tsoly rbeedeunc es hthoew rni stko oinfc sreela 136 escet iroens pboianss ew raast etso. of fIenr phaortmiceu lvairs,i tist .i nTchriesa sheads pthree vrieosupsolyn sbee erant esh ino wa ng rtoo uinpc wreiathse h riegshp omnosreb riadtietys.,1 3r6e lIunc ptaanrtt itcou lvairs, iitt tihnec rcelainseicd. Hthoe wreesvpeor,n swee r astheo iwn ead girno uPpa pweitrh I h tighhat m thoerrbei dwiteyr, er ecluhcatraanctte trois vtiicssit atshseo ccilainteicd. wHiothw elvoewre, rw pe asrhtiociwpeatdi oinn Praatepse ri nI tthhea t 1t9h3e0r ec woheorer tc, hsaurcahc tearsi sltoicws ears sleovceial teodf ewdituhc altoiowne,r bpeainrtgi cbipoartnio onu trsaitdees oinf Stwhee d1e9n3, 0n octo hboeirnt,g smucahrr iaesd , loasw were llle avse le .ogf. hedavuicnagti ocanr,d bioevinags cbuolarrn d oisuetassidese, onfe uSrwoepdseynch, inaotrti cb deiisnogr dmearrsr, ioedr ,a lacso whoelll- raesla ete.gd. disorders 121having ca.rdi oAvass ctuhleasre d cisheaarsaecst,e nriesutircosp soyvcehrilaatpr icw ditihso rkdneorws, no rr iaslkc ofhaoclt-orresla tfeodr ddeismorednetrias,. 12a1 loAws etrh epsaer ticchiparaaticotner irsattices ino vtehrelsaep gwroituhp sk ncoowulnd rhisakv ef accatuosresd foanr underestimation of the prevalence and incidence of dementia in Papers III and IV. Although this selection bias might affect the estimates, both papers aimed to investigate time trends. That is why the most important concern relates to how much the selection bias varied between the cohorts. The rDeI SspCoUSnSsIeO N rate at age 85 was consistent over the three birth cohorts include 4d7 iDnI StChUisS StIhOeNs is (65%, 61%, and 62%). However, there are some differences 4in7 s election bias. For example, the non-participants in cohorts 1923-24 and 1930 hDiasdcu sas ihoing her 3-year mortality rate compared to participants. This differen5c9e was not found in cohort 1901-02. This could have inflated the difference in mortality between the two cohorts (1901-02 and 1923-24) in Paper III if participants in the later-born cohort included a healthier sample than the first cohort. It could also have influenced the findings in Paper IV if the differences in mortality rate were related to dementia. Shown in this thesis as additional results, there was an interaction between cohort and dementia status in relation to attrition between ages 85 and 88. This indicated that individuals with dementia more often refused at follow-up in cohort 1930 compared to both previous cohorts. This is an important insight, not at least in regards to Paper IV, since it might have inflated the decline in dementia prevalence we found when comparing cohorts 1930 and 1901-02 at age 88. However, there was no difference in refusal at follow-up based on dementia status between cohorts 1923-24 and 1901-02, which strengthens the conclusion of a declining prevalence of dementia. The attrition rate between ages 85 and 88 in those without dementia did not differ between cohorts, but it is impossible to know if those with incident dementia declined participation to a greater extent than previous cohorts. However, previous studies have shown that incidence is not as sensitive to non-response as prevalence.16,137 Another finding presented as additional results in this thesis was that the attrition at the age 88 follow-up differed by educational level between cohort 1923-24 and 1930, with the former having a larger dropout among those with only elementary education. Knowing that the risk of dementia is larger in the group with lower educational levels, we might have underestimated the incidence more in cohort 1923-24 than in cohort 1930. An emerging problem in population-based studies is the increasing 48 H AN N A W E T T E R B E R G dementia, a lower participation rate in these groups could have caused an underestimation of the prevalence and incidence of dementia in Papers III and IV. Although this selection bias might affect the estimates, both papers aimed to investigate time trends. That is why the most important concern relates to how much the selection bias varied between the cohorts. The response rate at age 85 was consistent over the three birth cohorts included in this thesis (65%, 61%, and 62%). However, there are some differences in selection bias. For example, the non-participants in cohorts 1923-24 and 1930 had a higher 3-year mortality rate compared to participants. This difference was not found in cohort 1901-02. This could have inflated the difference in mortality between the two cohorts (1901-02 and 1923-24) in Paper III if participants in the later-born cohort included a healthier sample than the first cohort. It could also have influenced the findings in Paper IV if the differences in mortality rate were related to dementia. Shown in this thesis as additional results, there was an interaction between cohort and dementia status in relation to attrition between ages 85 and 88. This indicated that individuals with dementia more often refused at follow-up in cohort 1930 compared to both previous cohorts. This is an important insight, not at least in regards to Paper IV, since it might have inflated the decline in dementia prevalence we found when comparing cohorts 1930 and 1901-02 at age 88. However, there was no difference in refusal at follow-up based on dementia status between cohorts 1923-24 and 1901-02, which strengthens the conclusion of a declining prevalence of dementia. The attrition rate between ages 85 and 88 in those without dementia did not differ between cohorts, but it is impossible to know if those with incident dementia declined participation to a greater extent than previous cohorts. However, previous studies have shown that incidence is not as sensitive to non-response as prevalence.16,137 Another finding presented as additional results in this thesis was that the attrition at the age 88 follow-up differed by educational level between cohort 1923-24 and 1930, with the former having a larger dropout among those with only elementary education. Knowing that the risk of dementia is larger in the group with lower educational levels, we might have underestimated the incidence more in cohort 1923-24 than in cohort 1930. An emerging problem in population-based studies is the increasing proportion of the population that is hard to contact. It is challenging to analyse the potential selection bias from this group, as we do not know if they differ in specific ways from those we get in contact with. Previous studies have suggested that this group might have worse psychological and physical health.11048 In previous H70 examinations performed in theH A1N9N7A0 sW aEnTdT E1R9B8E0RsG, o nly 0.4%-1.4% were coded as “not traceable”.116,128 In Paper I, we showed that the proportion of those we were unable to contact in cohort 1930 ranged from 0.5% to 3.6%.121 One reason for the increase could be the rapid decline in the use of fixed landline telephones. Even though a fixed landline is still the most common among older Swedes, mobile phones are increasingly replacing the landline,138 and mobile phone numbers are generally more difficult to find.139 Another reason could be that it is now common to have a 6 0 Hanna Wetterberg caller ID, identifying who is calling. The ID that was shown when the research team called was “unknown.” This could have affected the responses, as previous research has shown that respondents are affected by the familiarity of the organisation calling.140,141 As these changes have occurred during the decades included in the time trend analyses in Paper III and Paper IV, this also poses a threat of differences regarding the selection bias between the compared cohorts. However, the proportion who we were unable to get in contact with was low and it is therefore unlikely that this affected the results of the papers. Measurement bias As dementia is the primary outcome in this thesis, the diagnostic procedures must be discussed. A measurement bias in the dementia diagnostic procedure could cause misclassification, i.e. categorising participants into the wrong category. At the data collection phase of the study, measurement biases could be introduced that, in a later stage, cause misclassification. For example, there is a risk that some diverging or sliding in how the interview is performed occurs. Several of the variables included in the dementia classification are clinical assessments based on the interview and tests altogether. If the assessments of symptoms and signs of dementia have changed, the results in Paper III and IV might have been affected. To minimise this risk, the D I SCUSSIO N 49 pprrooppoorrttiioonn ooff tthhee ppooppuullaattiioonn tthhaatt iiss hhaarrdd ttoo ccoonnttaacctt.. IItt iiss cchhaalllleennggiinngg ttoo aannaallyyssee tthhee ppootteennttiiaall sseelleeccttiioonn bbiiaass ffrroomm tthhiiss ggrroouupp,, aass wwee ddoo nnoott kknnooww iiff tthheeyy ddiiffffeerr iinn ssppeecciiffiicc wwaayyss ffrroomm tthhoossee wwee ggeett iinn ccoonnttaacctt wwiitthh.. PPrreevviioouuss ssttuuddiieess hhaavvee ssuuggggeesstteedd tthhaatt tthhiiss ggrroouupp mmiigghhtt hhaavvee wwoorrssee ppssyycchhoollooggiiccaall aanndd pphhyyssiiccaall 110 hheeaalltthh..110 IInn pprreevviioouuss HH7700 eexxaammiinnaattiioonnss ppeerrffoorrmmeedd iinn tthhee 11997700ss aanndd 11998800ss,, oonnllyy 00..44%%--11..44%% wweerree ccooddeedd aass ““nnoott ttrraacceeaabbllee””.. 116,128 116,128 IInn PPaappeerr II,, wwee sshhoowweedd tthhaatt tthhee pprrooppoorrttiioonn ooff tthhoossee wwee wweerree uunnaabbllee ttoo ccoonnttaacctt iinn ccoohhoorrtt 11993300 rraannggeedd ffrroomm 00..55%% ttoo 33.6%. 121 .6%.121 OOnnee rreeaassoonn ffoorr tthhee iinnccrreeaassee ccoouulldd bbee tthhee rraappiidd ddeecclliinnee iinn tthhee uussee ooff ffiixxeedd llaannddlliinnee tteelleepphhoonneess.. EEvveenn tthhoouugghh aa ffiixxeedd llaannddlliinnee iiss ssttiillll tthhee mmoosstt ccoommmmoonn aammoonngg oollddeerr SSwweeddeess,, mmoobbiillee pphhoonneess aarree iinnccrreeaassiinnggllyy replacing the landline,138replacing the landline,138 aanndd mmoobbiillee pphhoonnee nnuummbbeerrss aarree ggeenneerraallllyy mmoorree ddiiffffiiccuulltt ttoo ffiinnd 139 d..139 AAnnootthheerr rreeaassoonn ccoouulldd bbee tthhaatt iitt iiss nnooww ccoommmmoonn ttoo hhaavvee aa ccaalllleerr IIDD,, iiddeennttiiffyyiinngg wwhhoo iiss ccaalllliinngg.. TThhee IIDD tthhaatt wwaass sshhoowwnn wwhheenn tthhee rreesseeaarrcchh tteeaamm ccaalllleedd wwaass ““uunnkknnoowwnn..”” TThhiiss ccoouulldd hhaavvee aaffffeecctteedd tthhee rreessppoonnsseess,, aass pprreevviioouuss rreesseeaarrcchh hhaass sshhoowwnn tthhaatt rreessppoonnddeennttss aarree aaffffeecctteedd bbyy tthhee ffaammiilliiaarriittyy ooff tthhee oorrggaanni 140,141 issaattiioonn ccaalllliinngg..140,141 AAss tthheessee cchhaannggeess hhaavvee ooccccuurrrreedd dduurriinngg tthhee ddeeccaaddeess iinncclluuddeedd iinn tthhee ttiimmee ttrreenndd aannaallyysseess iinn PPaappeerr IIIIII aanndd PPaappeerr IIVV,, tthhiiss aallssoo ppoosseess aa tthhrreeaatt ooff ddiiffffeerreenncceess rreeggaarrddiinngg tthhee sseelleeccttiioonn bbiiaass bbeettwweeeenn tthhee ccoommppaarreedd ccoohhoorrttss.. HHoowweevveerr,, tthhee pprrooppoorrttiioonn wwhhoo wwee wweerree uunnaabbllee ttoo ggeett iinn ccoonnttaacctt wwiitthh wwaass llooww aanndd iitt iiss tthheerreeffoorree uunnlliikkeellyy tthhaatt tthhiiss aaffffeecctteedd tthhee rreessuullttss ooff tthhee ppaappeerrss.. MMeeeaaasssuuurreermmeeemnntte bbniiaatsss bias AAss ddeemmeennttiiaa iiss tthhee pprriimmaarryy oouuttccoommee iinn tthhiiss tthheessiiss,, tthhee ddiiaaggnnoossttiicc pprroocceedduurreess mmuusstt bbee ddiissccuusssseedd.. AA mmeeaassuurreemmeenntt bbiiaass iinn tthhee ddeemmeennttiiaa ddiiaaggnnoossttiicc pprroocceedduurree ccoouulldd ccaauussee mmiissccllaassssiiffiiccaattiioonn,, ii..ee.. ccaatteeggoorriissiinngg ppaarrttiicciippaannttss iinnttoo tthhee wwrroonngg ccaatteeggoorryy.. AAtt tthhee ddaattaa ccoolllleeccttiioonn pphhaassee ooff tthhee ssttuuddyy,, mmeeaassuurreemmeenntt bbiiaasseess ccoouulldd bbee iinnttrroodduucceedd tthhaatt,, iinn aa llaatteerr ssttaaggee,, ccaauussee mmiissccllaassssiiffiiccaattiioonn.. FFoorr eexxaammppllee,, tthheerree iiss aa rriisskk tthhaatt ssoommee ddiivveerrggiinngg oorr sslliiddiinngg iinn hhooww tthhee iinntteerrvviieeww iiss ppeerrffoorrmmeedd ooccccuurrss.. SSeevveerraall ooff tthhee vvaarriiaabblleess iinncclluuddeedd iinn tthhee ddeemmeennttiiaa ccllaassssiiffiiccaattiioonn aarree cclliinniiccaall aasssseessssmmeennttss bbaasseedd oonn tthhee iinntteerrvviieeww aanndd tteessttss aallttooggeetthheerr.. IIff tthhee aasssseessssmmeennttss ooff ssyymmppttoommss aanndd ssiiggnnss ooff ddeemmeennttiiaa hhaavvee cchhaannggeedd,, tthhee rreessuullttss iinn PPaappeerr IIIIII aanndd IIVV mmiigghhtt hhaavvee bbeeeenn aaffffeecctteedd.. TToo mmiinniimmiissee tthhiiss rriisskk,, tthhee psychiatrist (Professor Ingmar Skoog) performing the interviews in the first cohort included in this thesis trained the psychiatric nurses performing the interviews in the later-born cohorts. IS continuously oversees the data collection to keep the interviews as close as possible to previous examinations. Although the inter-rater agreement between psychiatrists and DpIsSyCcUhSiaStIrOiNc research nurses has been high within the study,88 the agreemen4t9 D I SCUSSIO N 49 between waves of examination years is impossible to test. However, objective measures such as performance in cognitive tests has also improved in later- born cohorts.142 This strengthens the conclusion that the decline in dementia prevalence and incidence we found is not an artefact due to measurement bias. Another potential risk of misclassification arises from the choice of diagnostic tool. As the methods have been kept the same since the late 1980s, the diagnostic criteria used at that time, the DSM-III-R, has been used throughout the studies. As we showed in Paper II, the choice of diagnostic criteria has major implications on the estimated prevalence of dementia. DHisocuwsesivoenr , when diagnosing dementia in the H70 study in the 1980s, it w6a1s decided to acknowledge deficits in short- or long-term memory as fulfilling the requirement of memory impairment, whilst DSM-III-R requires deficits in both. This was decided as the requirement of deficits in both was considered as too strict.129 This decision makes the diagnoses of dementia in the H70 study more similar to the newer versions of diagnostic criteria, such as DSM- IV. It does, however, affect the comparability of prevalence and incidence estimates with other studies. Performing interviews with key informants provide additional information important for diagnosing dementia. Although some population-based studies do not have access to key informant interviews,19,143 it is commonly used. However, the procedure for inclusion varies. For example, in the Swedish National Study on Aging and Care in Kungsholmen (SNAC-K) studies, a key informant interview was performed in cases when the participant was unable to answer themselves, and in cases lost to follow-up due to death.95,144 In the Dutch Rotterdam study and the British MRC-CFAS study, key informant interviews were performed among participants below pre-specified cut-offs in dementia screening tests.89,145 In the H70-studies, all participants were asked to name a close relative or friend to participate in the extensive semi- structured interview. This was a strength, as Paper II showed that the 50 H AN N A W E T T E R B E R G ppssyycchhiiaattrriisstt ((PPrrooffeessssoorr IInnggmmaarr SSkkoooogg)) ppeerrffoorrmmiinngg tthhee iinntteerrvviieewwss iinn tthhee ffiirrsstt ccoohhoorrtt iinncclluuddeedd iinn tthhiiss tthheessiiss ttrraaiinneedd tthhee ppssyycchhiiaattrriicc nnuurrsseess ppeerrffoorrmmiinngg tthhee iinntteerrvviieewwss iinn tthhee llaatteerr--bboorrnn ccoohhoorrttss.. IISS ccoonnttiinnuuoouussllyy oovveerrsseeeess tthhee ddaattaa ccoolllleeccttiioonn ttoo kkeeeepp tthhee iinntteerrvviieewwss aass cclloossee aass ppoossssiibbllee ttoo pprreevviioouuss eexxaammiinnaattiioonnss.. AAlltthhoouugghh tthhee iinntteerr--rraatteerr aaggrreeeemmeenntt bbeettwweeeenn ppssyycchhiiaattrriissttss aanndd ppssyycchhiiaattrriicc rreesseeaarrcchh nnuurrsseess hhaass bbeeeenn hhiigghh wwiitthhiinn tthhee ssttuuddyy,,8888 tthhee aaggrreeeemmeenntt bbeettwweeeenn wwaavveess ooff eexxaammiinnaattiioonn yyeeaarrss iiss iimmppoossssiibbllee ttoo tteesstt.. HHoowweevveerr,, oobbjjeeccttiivvee mmeeaassuurreess ssuucchh aass ppeerrffoorrmmaannccee iinn ccooggnniittiivvee tteessttss hhaass aallssoo iimmpprroovveedd iinn llaatteerr-- bboorrnn ccoohhoorrttss..114422 TThhiiss ssttrreennggtthheennss tthhee ccoonncclluussiioonn tthhaatt tthhee ddeecclliinnee iinn ddeemmeennttiiaa pprreevvaalleennccee aanndd iinncciiddeennccee wwee ffoouunndd iiss nnoott aann aarrtteeffaacctt dduuee ttoo mmeeaassuurreemmeenntt bbiiaass.. AAnnootthheerr ppootteennttiiaall rriisskk ooff mmiissccllaassssiiffiiccaattiioonn aarriisseess ffrroomm tthhee cchhooiiccee ooff ddiiaaggnnoossttiicc ttooooll.. AAss tthhee mmeetthhooddss hhaavvee bbeeeenn kkeepptt tthhee ssaammee ssiinnccee tthhee llaattee 11998800ss,, tthhee ddiiaaggnnoossttiicc ccrriitteerriiaa uusseedd aatt tthhaatt ttiimmee,, tthhee DDSSMM--IIIIII--RR,, hhaass bbeeeenn uusseedd tthhrroouugghhoouutt tthhee ssttuuddiieess.. AAss wwee sshhoowweedd iinn PPaappeerr IIII,, tthhee cchhooiiccee ooff ddiiaaggnnoossttiicc ccrriitteerriiaa hhaass mmaajjoorr iimmpplliiccaattiioonnss oonn tthhee eessttiimmaatteedd pprreevvaalleennccee ooff ddeemmeennttiiaa.. HHoowweevveerr,, wwhheenn ddiiaaggnnoossiinngg ddeemmeennttiiaa iinn tthhee HH7700 ssttuuddyy iinn tthhee 11998800ss,, iitt wwaass ddeecciiddeedd ttoo aacckknnoowwlleeddggee ddeeffiicciittss iinn sshhoorrtt-- oorr lloonngg--tteerrmm mmeemmoorryy aass ffuullffiilllliinngg tthhee rreeqquuiirreemmeenntt ooff mmeemmoorryy iimmppaaiirrmmeenntt,, wwhhiillsstt DDSSMM--IIIIII--RR rreeqquuiirreess ddeeffiicciittss iinn bbootthh.. TThhiiss wwaass ddeecciiddeedd aass tthhee rreeqquuiirreemmeenntt ooff ddeeffiicciittss iinn bbootthh wwaass ccoonnssiiddeerreedd aass ttoooo ssttrriicctt..112299 TThhiiss ddeecciissiioonn mmaakkeess tthhee ddiiaaggnnoosseess ooff ddeemmeennttiiaa iinn tthhee HH7700 ssttuuddyy mmoorree ssiimmiillaarr ttoo tthhee nneewweerr vveerrssiioonnss ooff ddiiaaggnnoossttiicc ccrriitteerriiaa,, ssuucchh aass DDSSMM-- IIVV.. IItt ddooeess,, hhoowweevveerr,, aaffffeecctt tthhee ccoommppaarraabbiilliittyy ooff pprreevvaalleennccee aanndd iinncciiddeennccee eessttiimmaatteess wwiitthh ootthheerr ssttuuddiieess.. PPeerrffoorrmmiinngg iinntteerrvviieewwss wwiitthh kkeeyy iinnffoorrmmaannttss pprroovviiddee aaddddiittiioonnaall iinnffoorrmmaattiioonn iimmppoorrttaanntt ffoorr ddiiaaggnnoossiinngg ddeemmeennttiiaa.. AAlltthhoouugghh ssoommee ppooppuullaattiioonn--bbaasseedd ssttuuddiieess ddoo nnoott hhaavvee aacccceessss ttoo kkeeyy iinnffoorrmmaanntt iinntteerrvviieewwss,,1199,,114433 iitt iiss ccoommmmoonnllyy uusseedd.. HHoowweevveerr,, tthhee pprroocceedduurree ffoorr iinncclluussiioonn vvaarriieess.. FFoorr eexxaammppllee,, iinn tthhee SSwweeddiisshh NNaattiioonnaall SSttuuddyy oonn AAggiinngg aanndd CCaarree iinn KKuunnggsshhoollmmeenn ((SSNNAACC--KK)) ssttuuddiieess,, aa kkeeyy iinnffoorrmmaanntt iinntteerrvviieeww wwaass ppeerrffoorrmmeedd iinn ccaasseess wwhheenn tthhee ppaarrttiicciippaanntt wwaass uunnaabbllee ttoo aannsswweerr tthheemmsseellvveess,, aanndd iinn ccaasseess lloosstt ttoo ffoollllooww--uupp dduuee ttoo ddeeaatthh..9955,,114444 IInn tthhee DDuuttcchh RRootttteerrddaamm ssttuuddyy aanndd tthhee BBrriittiisshh MMRRCC--CCFFAASS ssttuuddyy,, kkeeyy iinnffoorrmmaanntt iinntteerrvviieewwss wweerree ppeerrffoorrmmeedd aammoonngg ppaarrttiicciippaannttss bbeellooww pprree--ssppeecciiffiieedd ccuutt--ooffffss iinn ddeemmeennttiiaa ssccrreeeenniinngg tteessttss..8899,,114455 IInn tthhee HH7700--ssttuuddiieess,, aallll ppaarrttiicciippaannttss wweerree aasskkeedd ttoo nnaammee aa cclloossee rreellaattiivvee oorr ffrriieenndd ttoo ppaarrttiicciippaattee iinn tthhee eexxtteennssiivvee sseemmii-- ssttrruuccttuurreedd iinntteerrvviieeww.. TThhiiss wwaass aa ssttrreennggtthh,, aass PPaappeerr IIII sshhoowweedd tthhaatt tthhee prevalence of dementia increases when including the information from key informants. However, the use of key informant interviews also poses potential issues. For example, the coverage of key informant interviews was not perfect and varied slightly between cohorts. The proportion of p5500a rticipants with a key informant interview varied from 77HH%AANN iNNnAA c WWohEEToTTTrEtE RR1BB9EE2RR3GG- 2 4 to 82% in 1930 and 91% in cohort 1901-02. As the interview provides information that captures more cases of dementia, the varying proportion could have affected the estimates. The difference in dementia prevalence between cohort 1923-24 and 1901-02 was mainly seen among the mild cases of dementia,88 in which the information provided by the key informant likely is more important to capture the subtle changes than among more severe cases. It is possible that the difference in coverage of key informant interviews led to a lower detection rate of mild cases of dementia in the newer cohorts. Another issue with key informant interviews was that the median time between the neuropsychiatric evaluation and the interview varied between and within the cohorts, potentially affecting the information given. 6U2 se of register data Hanna Wetterberg In Papers I, III and IV, register data from Statistics Sweden, the Swedish Tax Agency, and the National Board of Health and Welfare was used in different ways. Dates of death, retrieved from the Swedish Tax Agency, catch virtually all deaths.146 The exception is individuals that have emigrated from Sweden. This was the case for a few individuals in cohort 1930. As the status of these individuals was unknown, they were excluded from the mortality analyses. Since the number was low, this would not have affected the results. In Paper I, sociodemographic characteristics and discharge diagnoses were retrieved from Statistics Sweden and the National Board of Health and Welfare. A strength of using register data was that the same data was used for all groups and collected during the same period. As it is mandatory for all physicians to report data to the IPR, the coverage of hospital visits has been almost 100% since 1987.147 However, data on this level might be crude and D I SCUSSIO N 51 prevalence of dementia increases when including the information from key informants. However, the use of key informant interviews also poses potential issues. For example, the coverage of key informant interviews was not perfect and varied slightly between cohorts. The proportion of participants with a key informant interview varied from 77% in cohort 1923- 24 to 82% in 1930 and 91% in cohort 1901-02. As the interview provides information that captures more cases of dementia, the varying proportion could have affected the estimates. The difference in dementia prevalence between cohort 1923-24 and 1901-02 was mainly seen among the mild cases of dementia,88 in which the information provided by the key informant likely is more important to capture the subtle changes than among more severe cases. It is possible that the difference in coverage of key informant interviews led to a lower detection rate of mild cases of dementia in the newer cohorts. Another issue with key informant interviews was that the median time between the neuropsychiatric evaluation and the interview varied between and within the cohorts, potentially affecting the information given. Use of fre rgeisgteirs dtaetar data In Papers I, III and IV, register data from Statistics Sweden, the Swedish Tax Agency, and the National Board of Health and Welfare was used in different ways. Dates of death, retrieved from the Swedish Tax Agency, catch virtually all deaths.146 The exception is individuals that have emigrated from Sweden. This was the case for a few individuals in cohort 1930. As the status of these individuals was unknown, they were excluded from the mortality analyses. Since the number was low, this would not have affected the results. In Paper I, sociodemographic characteristics and discharge diagnoses were retrieved from Statistics Sweden and the National Board of Health and Welfare. A strength of using register data was that the same data was used for all groups and collected during the same period. As it is mandatory for all physicians to report data to the IPR, the coverage of hospital visits has been almost 100% since 1987.147 However, data on this level might be crude and not sensitive enough to reveal actual differences.104 A Danish study investigating the selection bias in a population-based cohort study compared the use of register data with previously collected clinical data. They found that the more detailed data, collected 20 years prior to the follow-up, confirmed more differences between participants and refusals compared to the use of rDeI SgCisUtS er S-IbOaNs ed data.104 Another limitation with register-based data is tha5t1 hospital discharge diagnose codes only are proxies for disease, with varying sensitivity and specificity.148 These considerations are also valid for Paper III, where register data, to some extent, was used for the data collection of diseases included as covariates in the models. Another concern in Paper III and IV was that we compared two cohorts examined 20 years apart. The diagnosing of some of the included diseases might have been influenced by time trends in awareness and diagnostic procedures. The diagnoses based on registers from the hospitals are also linked to changes in policies in the hospitals. However, in Paper IV, we found that the sensitivity of dementia in the IPR and CDREG did not differ between the three cohorts. Another limitation regarding registers is that the sensitivity for dementia is low or moderate, particularly in this high age group.133,149 This directly affects the incidence estimates of Paper IV, as the source of information on dementia among those lost to follow-up only was based on register data. Paper I DPisacpuesrs ioIn w as based on a series of cross-sectional examinations, and we use6d3 aggregated data to compare the characteristics of participants in the H70 studies with refusals and same-aged individuals in Gothenburg and Sweden.121 The sociodemographic variables included in the paper were selected based on previously known characteristics associated with declining participation. The hospital discharge diagnoses were chosen based on both disorders that are common in this age group, such as cardiovascular disease, and disorders associated with participation, such as neuropsychiatric or alcohol-related diseases. The nature of the data makes adjustment of confounders impossible, which limits the interpretation of the individual characteristics. Instead, we used chi- 52 H AN N A W E T T E R B E R G not sensitive enough to reveal actual differences.104 A Danish study investigating the selection bias in a population-based cohort study compared the use of register data with previously collected clinical data. They found that the more detailed data, collected 20 years prior to the follow-up, confirmed more differences between participants and refusals compared to the use of register-based data.104 Another limitation with register-based data is that hospital discharge diagnose codes only are proxies for disease, with varying sensitivity and specificity.148 These considerations are also valid for Paper III, where register data, to some extent, was used for the data collection of diseases included as covariates in the models. Another concern in Paper III and IV was that we compared two cohorts examined 20 years apart. The diagnosing of some of the included diseases might have been influenced by time trends in awareness and diagnostic procedures. The diagnoses based on registers from the hospitals are also linked to changes in policies in the hospitals. However, in Paper IV, we found that the sensitivity of dementia in the IPR and CDREG did not differ between the three cohorts. Another limitation regarding registers is that the sensitivity for dementia is low or moderate, particularly in this high age group.133,149 This directly affects the incidence estimates of Paper IV, as the source of information on dementia among those lost to follow-up only was based on register data. PPaappeerr I I Paper I was based on a series of cross-sectional examinations, and we used aggregated data to compare the characteristics of participants in the H70 studies with refusals and same-aged individuals in Gothenburg and Sweden.121 The sociodemographic variables included in the paper were selected based on previously known characteristics associated with declining participation. The hospital discharge diagnoses were chosen based on both disorders that are common in this age group, such as cardiovascular disease, and disorders associated with participation, such as neuropsychiatric or alcohol-related diseases. The nature of the data makes adjustment of confounders impossible, which limits the interpretation of the individual characteristics. Instead, we used chi- square tests to analyse differences between participants, refusals, and same- aged individuals in Gothenburg and Sweden, for each sample characteristic. This ended up with many tests, which increased the risk of type I errors (false positives). One might argue that correction for multiple testing, such as B52o nferroni correction, would lower this risk. This iHsA NtNruAe W, EbTuT Et R BsuEcRhG corrections are also known to be very conservative and increase the risk of type II errors (false negatives).150 Another limitation in Paper I was that the comparison groups of same-aged individuals in Gothenburg and Sweden include the sampled population. This was considered necessary as the participants represent 15-25% of the population in its age group in Gothenburg, so excluding them might skew the comparison group.121 Paper II In Paper II, we used cross-sectional data from two cohorts of 85-year-olds, an age group where dementia is common, to investigate the impact of the choice of diagnostic criteria on the prevalence of dementia. Although the different algorithms were constructed to correspond to different editions of the DSM and ICD-systems, we could not validate the algorithms against a gold-standard clinical diagnosis corresponding to the correct criteria. The usefulness of the algorithms would have increased if their sensitivity and specificity had been analysed. Another issue with algorithmic diagnoses is the difficulty of handling missing data. Previous studies have shown that those with too much missing data for 6t4h e algorithm to classify, often have severe dementia, making tHhaenmna uWnetatberlebe trog answer questions or perform tests.151 We did not standardise the cognitive tests included in the algorithms based on educational level, which previously have been shown to increase the accuracy of algorithmic diagnoses.152 This could potentially have led to a higher prevalence of dementia in the group with lower education. However, D I SCUSSIO N 53 square tests to analyse differences between participants, refusals, and same- aged individuals in Gothenburg and Sweden, for each sample characteristic. This ended up with many tests, which increased the risk of type I errors (false positives). One might argue that correction for multiple testing, such as Bonferroni correction, would lower this risk. This is true, but such corrections are also known to be very conservative and increase the risk of type II errors (false negatives).150 Another limitation in Paper I was that the comparison groups of same-aged individuals in Gothenburg and Sweden include the sampled population. This was considered necessary as the participants represent 15-25% of the population in its age group in Gothenburg, so excluding them might skew the comparison group.121 Paper I II I In Paper II, we used cross-sectional data from two cohorts of 85-year-olds, an age group where dementia is common, to investigate the impact of the choice of diagnostic criteria on the prevalence of dementia. Although the different algorithms were constructed to correspond to different editions of the DSM and ICD-systems, we could not validate the algorithms against a gold-standard clinical diagnosis corresponding to the correct criteria. The usefulness of the algorithms would have increased if their sensitivity and specificity had been analysed. Another issue with algorithmic diagnoses is the difficulty of handling missing data. Previous studies have shown that those with too much missing data for the algorithm to classify, often have severe dementia, making them unable to answer questions or perform tests.151 We did not standardise the cognitive tests included in the algorithms based on educational level, which previously have been shown to increase the accuracy of algorithmic diagnoses.152 This could potentially have led to a higher prevalence of dementia in the group with lower education. However, aass maannyy ooff tthhee vvaarriiaabblleess iinncclluuddeedd iinn tthhee aallggoorriitthhm aarree bbaasseedd oonn cclliinniiccaall aasssseessssmeennttss,, wwhheerree tthhee cclliinniicciiaann ttaakkeess aallll iinnffoorrmaattiioonn iinnttoo ccoonnssiiddeerraattiioonn,, iitt iiss uunnlliikkeellyy tthhaatt tthhiiss wwoouulldd bbee aa maajjoorr iissssuuee.. PDPPI a aSaCpppUeeeSrrrS IIO IIIIIN I I 53 PPaappeerr IIIIII iiss aa pprroossppeeccttiivvee ccoohhoorrtt ssttuuddyy ooff ttwwoo ccoohhoorrttss ooff 8855--yyeeaarr--oollddss wwhheerree ttiimee ttrreennddss iinn moorrttaalliittyy aamoonngg tthhoossee wwiitthh aanndd wwiitthhoouutt ddeemeennttiiaa wweerree eexxaamiinneedd.. Wee aallssoo eexxaamiinneedd tthhee PPAARR ooff ddeemeennttiiaa aanndd ootthheerr ccoommoonn ddiisseeaasseess oonn ddeeaatthh.. TThhee cchhooiiccee ooff ddiiaaggnnoosseess iinncclluuddeedd wwaass bbaasseedd oonn pprreevviioouuss kknnoowwlleeddggee aabboouutt ccoommoonn ddiisseeaasseess iinn tthhiiss aaggee ggrroouupp.. Hoowweevveerr,, uussiinngg aa vvaalliiddaatteedd iinnddeexx ssccoorree ooff ddiisseeaasseess,, ssuucchh aass tthhee CChhaarrllssoonn CCoomoorrbbiiddiittyy IInnddeexx,, wwoouulldd hhaavvee iinnccrreeaasseedd tthhee ccoomppaarraabbiilliittyy wwiitthh ootthheerr ssttuuddiieess.. TThhiiss wwaass,, hhoowweevveerr,, nnoott ffeeaassiibbllee iinn tthhiiss pprroojjeecctt,, aass tthhee ccoolllleecctteedd ddaattaa iinn tthhee ttwwoo ccoohhoorrttss lliimiittss tthhee aapppplliiccaattiioonn ooff ssttaannddaarrddiisseedd iinnddeexxeess.. TThhiiss wwaass bbeeccaauussee wwee hhaadd ttoo pprriioorriittiissee vvaarriiaabblleess tthhaatt wweerree aass ssiimiillaarr aass ppoossssiibbllee ttoo eennssuurree ccoomppaarraabbiilliittyy oovveerr tthhee ttwwoo ccoohhoorrttss,, wwhhiicchh eennddeedd uupp wwiitthh vvaarriiaabblleess nnoott maattcchhiinngg aa ssttaannddaarrddiisseedd iinnddeexx.. AAnnootthheerr lliimiittaattiioonn wwaass tthhaatt aass tthhee iinncclluuddeedd ddiisseeaasseess rreepprreesseenntt ddiisseeaasseess pprreevvaalleenntt aatt aaggee 8855,, wwee ccaannnnoott eevvaalluuaattee tthhee eeffffeecctt oonn moorrttaalliittyy ooff iinncciiddeenntt ddiisseeaasseess tthhaatt Doicsccuussion 6 5occurrrreedd aafftteerr tthhee eexxaamiinnaattiioonn.. TToo oobbttaaiinn tthhee ddiiaaggnnoosseess ooff ddiisseeaasseess iinncclluuddeedd,, ddaattaa ffrroom sseellff--rreeppoorrtt,, llaabboorraattoorryy rreessuullttss,, meeddiiccaattiioonn lliissttss,, aanndd rreeggiisstteerr ddaattaa wweerree ttrriiaanngguullaatteedd.. TThhiiss iinnccrreeaasseedd tthhee ppootteennttiiaall ttoo ccaappttuurree ddiisseeaasseess.. PPaappeerr IIVV PPaappeerr IIVV wwaass aa ccrroossss--sseeccttiioonnaall ssttuuddyy ttoo eexxaamiinnee ttiimee ttrreennddss iinn tthhee pprreevvaalleennccee ooff ddeemeennttiiaa aatt aaggeess 8855,, 8888,, aanndd 9900,, aanndd aa lloonnggiittuuddiinnaall ssttuuddyy ttoo iinnvveessttiiggaattee tthhee ffoouurr--yyeeaarr iinncciiddeennccee ooff ddeemeennttiiaa.. 5544 HHAANNNNAA WWEETTTTEERRBBEERRGG as many of the variables included in the algorithm are based on clinical assessments, where the clinician takes all information into consideration, it is unlikely that this would be a major issue. as many of the variables included in the algorithm are based on clinical assessments, where the clinician takes all information into consideration, it is uPnalpikeerly I ItIh at this would be a major issue. Paper III is a prospective cohort study of two cohorts of 85-year-olds where tPimape etrr eIInId s in mortality among those with and without dementia were examined. We also examined the PAR of dementia and other common dPiaspeears eIsI Io ins da epartohs. p ective cohort study of two cohorts of 85-year-olds where time trends in mortality among those with and without dementia were Texhaem cinheodic. e Wofe dailasgon oesxeasm ininceludd ethde wPaAs Rba soefd doenm pernetviaio auns dk noothweler dcgoem abmoount dcoismeamseosn o dni sdeeaastehs. i n this age group. However, using a validated index score of diseases, such as the Charlson Comorbidity Index, would have increased Tthhee c cohmopicaer aobfi lidtyia gwnitohs eost hinecrl ustdueddi ews.a Ts hbiass weda so, nh opwreevieoru, sn oknt ofewalseidbglee ianb tohuist cporomjemcto, na sd itsheaes ecso lilne cttheids adgaet ag rino utph.e H towwoe cvoerh,o urstsin gli ma ivtsa ltidhaet eadp pinlidceaxti osnco oref ostfa nddisaeradsiesse,d s uincdhe axse st.h eT hCihs awrlasos nb eCcoamusoer bwide ithya Idn dtoe xp, rwiooruitlidse h vaavrei aibnlcerse atsheadt twheer ec oams psaimraiblailri tays w pitohs soibthl e rt ost uedniseusr. eT choism wpasr,a bhioliwtye voevre, rn otht ef etawsiob lceo ihno trhtiss, pwrhoijcehc te, nadse dth eu pc owlliethc tvedar idaabtlae si n otth em atwtcoh incog hao srtsa nldimaridtsi stehde inadpepxli.c Aatnioonth oefr slitmanitdaatirodnis ewda si nthdaetx aes .t hTeh iins cwluadse db edcisaeuassee sw re phraedse ntot dpirsieoarsietiss ep rveavrailaebnlte sa tt ahgaet w85e,r ew aes csainmniloatr eavs apluoastsei btlhe et oe fefencstu roen c ommoprtaarlaitbyi liotyf oinvceird ethnet dtwisoe acsoesh otrhtast, owchciuchrr eedn daeftde ru tph ew eitxha mvainriatbiolens. n ot matching a standardised index. Another limitation was that as the included diseases represent diseases prevalent at age 8T5o, woeb taciann ntohte evdaialugantoes etsh e oef ffdeciste aosne s mionrctlaulditeyd ,o fd iantac idfernotm d isseealfs-erse ptohratt, olacbcourrarteodr ya frteesru tlhtse, emxaemdicinaatitoionn l.i s ts, and register data were triangulated. This increased the potential to capture diseases. To obtain the diagnoses of diseases included, data from self-report, l a boratory results, medication lists, and register data were triangulated. This increased the potential to capture diseases. P aperr IV I V Paper IV was a cross-sectional study to examine time trends in the prevalence of dementia at ages 85, 88, and 90, and a longitudinal study to investigate the fPoaupr-eyre aIrV i ncidence of dementia. Paper IV was a cross-sectional study to examine time trends in the prevalence oAAfss dttehhmee eppnaatrrittaii ccaiitpp aagnnettss 8ww5ee, rr8ee8 oo, aff naadd vv9aa0nn, ccaeenddd aagg leeo,,n ttghhieetu rrdiiissnkka loo sfft ubbdiiaayss eetodd irrneevsseuusllttssig addtuueee t httooe fssoeellueercc-ttyiivveaeer ssiuunrrcvvidiivveaanll caaenn oddf ccdooemppeneettiinna.gg rriisskk ooff ddeeaatthh aarree pprroomiinneenntt.. HHoowweevveerr,, aass tthhee moorrttaalliittyy wwaass hhiigghheerr iinn tthhee ffiirrsstt ccoohhoorrtt,, moorree ccaasseess wweerree lliikkeellyy ttoo bbee miisssseedd dduuee ttoo tthhee ccoomppeettiinngg rriisskk ooff ddeeaatthh iinn tthhiiss ccoohhoorrtt.. TThhiiss wwoouulldd uu54nn ddeerreessttiimaattee tthhee ddeecclliinnee iinn ddeemeennttiiaa iinncciiddeennccee aanndd ppHrrAeeNvvaaNllAee nnWccEeeT TttEhhRaattB EwwReeG ffoouunndd rraatthheerr tthhaann iinnffllaattiinngg iitt.. AA54ss tthhee nnuumbbeerr ooff ppaarrttiicciippaannttss iinn eeaacchh ccoohhoorrtt wwaass nnoott vveerryyH AhhNiiggNhhA,, WtthhEeeT TaaEnnRaaBllyyEssReeGss hhaadd llooww ppoowweerr ttoo ddeetteecctt ttrruuee ddiiffffeerreenncceess iinn tthhee pprreevvaalleennccee aanndd iinncciiddeennccee ooff ddeemeennttiiaa bbeettwweeeenn tthhee ccoohhoorrttss.. TThhiiss ccoouulldd eexxppllaaiinn wwhhyy wwee ddiidd nnoott ffiinndd aa ddiiffffeerreennccee iinn ddeemeennttiiaa pprreevvaalleennccee aatt aaggee 8855 bbeettwweeeenn ccoohhoorrttss 11990011--0022 aanndd 11993300.. IInn aa ppoosstthhoocc ppoowweerr aannaallyyssiiss,, wwee ffoouunndd tthhaatt wwiitthh tthhee nnuumbbeerr ooff 991100 iinncclluuddeedd iinn tthhee aannaallyyssiiss,, tthheerree wwaass aa ppoowweerr ooff 00..5566,, aatt aallpphhaa 00..0055,, ttoo ddeetteecctt aann ORR ooff 00..7755.. GGeeennneeerrraaall ldd diissiscccuuussssssiiooionnn TThhee maaiinn ffiinnddiinnggss ooff PPaappeerr II wweerree tthhaatt ppaarrttiicciippaannttss hhaadd hhiigghheerr eedduuccaattiioonnaall lleevveellss aanndd meeaann iinnccoomee,, aanndd lloowweerr pprreevvaalleennccee ooff ddiissoorrddeerrss iinn tthhee IIPPRR,, iinn ccoomppaarriissoonn ttoo rreeffuussaallss,, aanndd tthhaatt tthhee rreessppoonnssee rraattee ddeecclliinneedd wwiitthh aaggee.. TThhee ppaarrttiicciippaannttss wweerree,, hhoowweevveerr,, moorree ssiimiillaarr ttoo tthhee ttaarrggeett ppooppuullaattiioonn ooff ssaamee-- aaggeedd iinnddiivviidduuaallss iinn Gootthheennbbuurrgg aanndd ttoo ssaamee--aaggeedd iinnddiivviidduuaallss iinn SSwweeddeenn,, wwhheerree tthheeyy maaiinnllyy ddiiffffeerreedd iinn tthhaatt ppaarrttiicciippaannttss hhaadd hhiigghheerr eedduuccaattiioonnaall lleevveellss.. TThhee maaiinn ffiinnddiinnggss ooff PPaappeerr IIII wweerree tthhaatt tthhee IICCD--1111 aanndd DSSM--55 ccllaassssiiffiiccaattiioonn ssyysstteemss ffoorr ddeemeennttiiaa ggeenneerraatteedd hhiigghheerr pprreevvaalleennccee ccoomppaarreedd ttoo oollddeerr eeddiittiioonnss ooff bbootthh IICCD aanndd DSSM,, aanndd tthhaatt ddeemeennttiiaa ccllaassssiiffiiccaattiioonn hhaass bbeeccoomee moorree ssiimiillaarr bbeettwweeeenn tthhee ttwwoo aass aa rreessuulltt ooff iinntteennttiioonnaall wwoorrkk ttoo 6h6haa rrmoonniissee tthhee ssyysstteemss.. PPaappeerrss II aanndd IIII pprroovvide imp Hanna Wetterberg ide i poorrttaanntt iinnssiigghhttss oonn meetthhooddoollooggiiccaall ccoonnssiiddeerraattiioonnss iinn ssttuuddiieess ooff oollddeerr aadduullttss,, ssuucchh aass hhooww tthhee rreepprreesseennttaattiivveenneessss iiss aaffffeecctteedd bbyy sseelleeccttiioonn bbiiaass,, aanndd ffuurrtthheerr eempphhaassiissee tthhee iimppoorrttaannccee ooff ccoomppaarriinngg ddeemeennttiiaa pprreevvaalleennccee uuttiilliissiinngg tthhee ssaamee oorr ssiimiillaarr ccrriitteerriiaa.. IInn PPaappeerr IIIIII,, wwee sshhoowweedd tthhaatt 8855--yyeeaarr--oollddss wwiitthh ddeemeennttiiaa ssuurrvviivveedd lloonnggeerr iinn aa ccoohhoorrtt bboorrnn 11992233--2244,, ccoomppaarreedd ttoo aa ccoohhoorrtt bboorrnn 11990011--0022,, aafftteerr aaddjjuussttiinngg ffoorr ddeemeennttiiaa sseevveerriittyy,, bbuutt tthhaatt ddeemeennttiiaa ssttiillll rreemaaiinnss tthhee moosstt iimppoorrttaanntt ffaaccttoorr ooff ddeeaatthh.. IInn PPaappeerr IIVV,, wwee sshhoowweedd tthhaatt tthhee pprreevvaalleennccee aanndd DD II SSCCUUSSSSIIOO NN 5555 As the participants were of advanced age, the risk of biased results due to selective survival and competing risk of death are prominent. However, as the mortality was higher in the first cohort, more cases were likely to be missed due to the competing risk of death in this cohort. This would underestimate the decline in dementia incidence and prevalence that we found rather than inflating it. As the number of participants in each cohort was not very high, the analyses had low power to detect true differences in the prevalence and incidence of dementia between the cohorts. This could explain why we did not find a difference in dementia prevalence at age 85 between cohorts 1901-02 and 1930. In a posthoc power analysis, we found that with the number of 910 included in the analysis, there was a power of 0.56, at alpha 0.05, to detect an OR of 0.75. General discussion The main findings of Paper I were that participants had higher educational levels and mean income, and lower prevalence of disorders in the IPR, in comparison to refusals, and that the response rate declined with age. The participants were, however, more similar to the target population of same- aged individuals in Gothenburg and to same-aged individuals in Sweden, where they mainly differed in that participants had higher educational levels. The main findings of Paper II were that the ICD-11 and DSM-5 classification systems for dementia generated higher prevalence compared to older editions of both ICD and DSM, and that dementia classification has become more similar between the two as a result of intentional work to harmonise the systems. Papers I and II provide important insights on methodological considerations in studies of older adults, such as how the representativeness is affected by selection bias, and further emphasise the importance of comparing dementia prevalence utilising the same or similar criteria. In Paper III, we showed that 85-year-olds with dementia survived longer in a cohort born 1923-24, compared to a cohort born 1901-02, after afodjuurs-tyienagr ifnocri ddeenmcee notfia d esemveenrittiya, dbeuctr etahsaetd damemoenngt itah es teilxl armeminaeidn s8 5t-h8e8 -myeoasrt- iomldpso ortvaenrt tfhacet opra sot f3 d0e ayteha.r sI.n A P asepceorn IdVa,r yw fei nsdhionwg ewd atsh atht atth eth per esvenalseinticveit ya nodf fdoeumr-eynetaiar iinn ctihdee nIPceR oafn dd eCmDeRntEiaG d ewcarse amseodd earmatoen, gb utht eth eex sapmecinifeidci t8y5 w-8a8s- hyeigahr-. oInldtesr eosvteinr gtlhye, tphaes ste 3n0si tyiveaitrys .a nAd ssepceocnifdicairtyy dfiindd ninogt dwifafse rt hbaett wtheee ns ethnes ictiovhitoyr tosf. dementia in the IPR and CDREG was moderate, but the specificity was high. Interestingly, the sensitivity and specificity did not differ between the cohorts. DRI SeCpUrSeSsIOeNn tativeness 55 R Reepprreseesnetnattivaetne iv sse ness The results from Paper I align with previous studies indicating that those choosing to participate in health examination studies generally have higher Tedhuec aretisounlt, s fferwoemr dPiasepaesre sI, aalnigdn hwigithhe rp rienvcioomuse ,s tcuodmiepsa irneddi ctaot inregf uthsaalts .t1h07o-1s0e9 cRheospoosinnsge troa tpesa ritnic hipeaatlteh i nex hameailntha teioxna mstiundaiteios nh asvtue dinie sg egneenrearla, ldlye chraevasee dh igohveerr ethdeu cpataisotn ,d efceawdeers .d15i3s,1e5a4s eTs,h iasn dtr ehnidgh eisr winocrormyien,g ,c oams pita rmedi gthot rdeefcurseaalss.e1 07t-h10e9 Rreepsrpeosennseta rtiavteesn einss h iena lethp idexeammioinloagtiiocnal ssttuuddiieess .h aFvoer ienx gaemnpelrea,l ,a d Fecinrenaissehd s otuvdeyr tehxea mpiansitn gd ethcea dreesp.1r5e3s,1e5n4 taTthivise ntersesn idn ias hweaolrthry ienxga,m aisn aitti omn isguhrtv edye ccroenadseu ctthede roevperre s2e5n ytaetairvse nshesosw iend e tphiadte mevieonlo tghicoaulg sht utdhiee sp. aFrtoicri peaxtaimonp lrea,t ea dFeinclninisehd sitnu dalyl esoxacmioiencionngo tmheic r gerporuepsesn sttautdivieedn,e sths ei nd eac lhineeal wtha se xlaarmgeinr ainti othne s gurrovuepy wcoitnhd luocwteedr oovcceur p2a5t iyoenaarls schlaosws edan tdh ate deuvceant itohnoaul ghle vtheel. 1p53a rtSicuigpgaetsiotend r arteea dsoecnlsin efdo ri nt halel sdoecciloineicnogn roamtesic i ngrcoluudpes pstaurtdiiceidp,a ttihoen d eexchlianues wtioans ldarugee tro i nth teh ein gcrroeuaspi nwgi tnhu lmowbeerr oocf ceuxpaamtiionnaatilo ncsla asns d asnudrv eeydsu pceartfioonrmal edle avnedl.1 l5o3 wSeur gvgoelsutnetde erriesamso onvs erfaollr. 15t5h e declining rates include participation exhaustion due to the increasing number oTfh ex raemspinoantsioen rsa aten din s uthrvee yHs7 p0e-rsftourdmieesd a amnodn lgo w7e0r- yveoalru-notldees riwsmas o8v5e%ra liln.1 55t h e early 1970s,116 a high response rate which declined to 81% in the mid 1970s, Ttoh e7 7r%es pionn sthe er aetaer liyn 1th98e 0Hs, 7a0n-sdt utdoi e7s0 a%m oinn g2 07000-y. eTarh-oisl dtsr ewndas o8f5 %de cilnin tinhge ereasrplyo 1n9s7e 0rsa,t1e16 w aa hs ibghro rkeesnp oinn steh er amteo wsth riecche ndte ccloinheodr tt oo f8 710%-y iena rt-hoel dms iidn 12907104s–, t1o6 , 7w7%ith ina rthesep oeanrsley r1a9t8e0 os,f a7n2d% t.o11 77 I0n% Pianp 2e0r 0I0,. wTeh isd idtr esnede oa f ddeecclilninei ning rreessppoonnssee rraattee owvaesr b trhoek yeena rins itnh ec omhoosrtt r1e9c3e0n,t b cuoth woert inotfe 7rp0-ryeeteadr- othldiss mina i2n0l1y 4a–s 1an6 , agwei tehf fae crt.e1s2p1 oDnessep irtaet eth oe fd e7c2l%ini.n11g7 rIens pPoanpsee rr aIte, , wthee dreidfu ssaeles ha adde ac lhinigeh einr rpersepvoanlesnec rea itne omvoerre t hdeis oyerdaresr sin t hcaonh opratr t1ic9i3p0a,n btsu ta tw aeg ien 7te0r pthreatne dat t ahgise m85a ionrly 8 a8s. aTnh iasg eh eigfhfelicgth.1t2s1 Dthees pnitoet itohne dtehcalti nrinesgp roenspseo nrsaete r aitse , ntohte rneefcuessaslsa rhilayd tah he igbheesrt pinrdeivcaalteinocne oinf mstourdey dqisuoarlditeyr.s Itth aisn mpaortriec ipimanptosr atat natg et o7 0u tnhdaenr satta angde t8h5e onr o8n8-. TDreishscipusos snhiosigne h bliigahs.t sI nt haed dnitoiotino,n i t thhaast breeesnp osnusgeg ersatteed itsh ant otth enseec tewssoa rciolyn ctehpet sb ae6rse7t ionndliyc awteioank lyo afs ssotucdiayt eqdu.1a5l6i tTyh. iIst r eisla tmeso troe tihme pfionrdtainngt tthoa tu tnhdee prastratincdip athnets nwoenre- rmesoproe nssime bilaiars t. oI nth aed dtaitrigoent , pito hpausla btieoenn, sinu glignees twedit hth raets tuhltess es htwowo nc oinn cperpetvsi oaures ostnuldy iwese.1a0k4 lyH aoswsoecvieart,e dth.1e5 6n Tohni-sr reeslpaotenss teo b tihaes fisin gdeinnge rtahllayt dthifef ipcaurltti ctiop aansstes sws,e ares more similar to the target population, in line with results shown in previous studies.104 However, the non-response bias is generally difficult to assess, as 56 H AN N A W E T T E R B E R G 56 H AN N A W E T T E R B E R G four-year incidence of dementia decreased among the examined 85-88-year- olds over the past 30 years. A secondary finding was that the sensitivity of dementia in the IPR and CDREG was moderate, but the specificity was high. Interestingly, the sensitivity and specificity did not differ between the cohorts. Representativeness The results from Paper I align with previous studies indicating that those choosing to participate in health examination studies generally have higher education, fewer diseases, and higher income, compared to refusals.107-109 Response rates in health examination studies have in general, decreased over the past decades.153,154 This trend is worrying, as it might decrease the representativeness in epidemiological studies. For example, a Finnish study examining the representativeness in a health examination survey conducted over 25 years showed that even though the participation rate declined in all socioeconomic groups studied, the decline was larger in the group with lower occupational class and educational level.153 Suggested reasons for the declining rates include participation exhaustion due to the increasing number of examinations and surveys performed and lower volunteerism overall.155 The response rate in the H70-studies among 70-year-olds was 85% in the early 1970s,116 a high response rate which declined to 81% in the mid 1970s, to 77% in the early 1980s, and to 70% in 2000. This trend of declining response rate was broken in the most recent cohort of 70-year-olds in 2014– 16, with a response rate of 72%.117 In Paper I, we did see a decline in response rate over the years in cohort 1930, but we interpreted this mainly as an age effect.121 Despite the declining response rate, the refusals had a higher prevalence in more disorders than participants at age 70 than at age 85 or 88. This highlights the notion that response rate is not necessarily the best indication of study quality. It is more important to understand the non- response bias. In addition, it has been suggested that these two concepts are only weakly associated.156 This relates to the finding that the participants were more similar to the target population, in line with results shown in previous tshtued bieiass.1 0i4s Ha orewseuvlte ro, ft hthe en uonnk-rneospwonn sree sbpioans sies gpernoebraablliyli tdyi fifnic iunltte troa catsiosens sw, iatsh tthhee rbeisaps oisn sae rreastue lto fo tfh teh est uudnyk.n15o6 w n response probability in interaction with the response rate of the study.156 Having a representative study sample is important in studies where the pHuarvpionsge ais rteop rdeessecnrtiabteiv teh es ttuadrgye ts apmoppluel aits io nim. pIto irst aanlts oi nim sptuodrtieasn t winh esrteu dtihees p56u rpose is to describe the target population. It is also imH ApNoNrtAa nWtE TinT EsRtuBdE RieG w here the outcome of interest might vary between different subgroups. Fosr wexhaemrep lteh, e tohuet corimske ooff indteemreesnt tmia igihs t vsuagryg ebsetetwd eetno dvifafreyr enbty sua bgnruomupbse.r Foorf ecxhaamrapctleer, istthices , rissukc ho f ads emedeunctiaat ioisn als uglegveeslt,e de tthon icviatyry, sbeyx , a annudm bmearr itoafl csthaatruasc.1t3e,1r5i7s,t1i5c8s T, hseusceh c haarsa cteedruisctaictiso anlaslo cleovinecl,i dee twhnitihc iftayc, tosresx a, ssaoncdia temd awriittahl sretaftuussin.1g3, 1p57a,1r5t8i cTiphaetsieo nc.h Aarsa cstheoriwstnic isn a slsuop pcoleimnceindeta wryi tfhig fuarcet oSr1s ians sPoacipaeterd I ,w tihthe rgerfouuspin wg iptha rtthicei ploatwioens.t Aress pshoonwsen r ainte s autp apllle emxeanmtainrya tfioignusr ew Sa1s uinn mPaarpreierd I ,m thene wgriothu pl owwitehr tehdeu lcoawtieosnta rl eslepvoenlss.e Trahties act oaulll de xhamavien aatfiofencst ewda st huen mesatrimrieadte ms eonf wmiothrt alloitwy,e irn ceidduecnacteio, annadl plerveevlasl.e nTchei so fc doeumlde nhtaiav ef oaufnfedc itne dP atphee rse sItIimI aanteds IVof. mIno Prtaaplietyr, Iin, wcide eanlscoe ,f aonudn dp rienv tahleant cthe oosfe d reemfuesnintiga tfoo upnadrt iicni pPaatep eart sa gIIeI 7 a0n md IoVre. Ionft Pena pheard Id, ewme eanlstoia faocucnodrd iinn gth taot tthhoe sIeP rRe.f uHsionwg etvoe pr,a ratsi csihpoatwe na ti na gPe a7p0e mr IoVre, tohfete sne nhsaidti vdietym oefn ttihae a IcPcoRr dainndg CtoD tRheE IGP Rw.a Hs oonwleyv 4e3r,% a,s mshaokwinng iint dPiaffpiceurl tI Vto, itnhtee rspernesti titvhiitsy foinf dthineg .I PIRt isa nadls Co DwRoErtGh nwoatsin ogn tlyh a4t3 t%he, mhiagkhinerg fitr edqiuffeincucylt otof dinetmerepnrteita tdhiiasg nfionsdeisn ga.m Iot nisg arelsfou swalos rmthi gnhot tbineg a t hreastu tlth eo fh tihghise gr rforuepqu beenicnyg oinf ldeesms heneatilath d, ihaagvnionsge sm aomreo hnogs rpeiftuals avliss imts,i gahntd b teh uas rae shuiglth oefr cthhiasn gcreo oufp r beceeinivgi ning ale sdse mheeanltthia, dhiaavginnog sme oinr et hheo rsepgitiastl evriss.i t s, and thus a higher chance of receiving a dementia diagnose in the registers. Chhooiciec aen da nusde uosf edi aogfn odsitaicg cnriotesrita ic criteria Choice and use of diagnostic criteria We found in Paper II that the choice of diagnostic criteria for dementia has aWne i mfopuonrdt ainnt Peaffpeecrt IoIn t hthate thpere cvhaoleincec eo fo fd icaagsneoss tidice ncrtiitfeierdia. fTohr ids eims ennotti ao hnalys saunp ipmoprtoerdta nbty epffreecvti oouns tshtued pierse,v79a,1le59n,1c60e bouft caalsseos, itdoe nstoifmieed . eTxtheins t,i s enxopte cotnedly. Dsuuprpinogrt ethde bpya sptr ethvrioeeu st os tfuoduier sd,7e9,c15a9d,1e60s ,b tuhte arlessoe, atroch s oomf de emexetenntita, hexasp emctaedde. Dmuajroinr gp trhoeg rpesass t inth ruened teors ftaonudr indge ctahdee sd, itshoer dreers,e aarncdh tohfe ddeimagennotsiat ich assy smteamdes hmaavjeo r bpereong reuspsd iant eudn daecrcsotardnidnignlgy . thTeh de isocrridteerri,a anthd atth ey iedlidaegdn osthtiec shyisgthemesst phraevvea lebneceen wuepred attheed laatcecsto rIdCinDg-l1y.1 aTnhde DcSriMte-r5ia crtihtearti a,y iwelhdiecdh ctohrer ehspigohnedsst pwreelvl atloen cthe ew aetrtee mthpet sl atteos tm ICakDe -1th1e a ncrdi tDeriSaM m-5o rcer itienrcialu, swivhei ctho wcoarrrdess pootnhdesr dweemll etnot iatsh eth aantt epmrimptasr itlyo AmDa.k2 eT hthe ed icffreitreerniac ems ionr cer iitnecrilau seimvep htoaswisaer dths e onteheedr fdoerm heanrtmiaso nthisaant iporni manardil ya wAaDre.2n Teshse wdhifefenr ecnocmesp ianr icnrgi tererisau eltms. pFhoars iseex athmep nlee,e da Fforre nhcahr mstoundiys arteiopno ratendd aanw ianrcerneeassse winh celnin cicoaml cpoanrisnegn sruess-udltiasg. nFoosre de xdaemmpelne,t iaa oFvreenr cah t esntu-dyeya rre ppeorritoedd. 1a5 nH ionwcreevaesre, itnh ec ltiwnioca cl ochoonrstesn tshuast- dwieagren ocosemdp daermede nhtaida 6o8v er a ten-year period.15 However, the two cohorts that were Hcaonmnap Waertetder hbeardg D I SCUSSIO N 57 DI SCUSSIO N 57 the bias is a result of the unknown response probability in interaction with the response rate of the study.156 Having a representative study sample is important in studies where the purpose is to describe the target population. It is also important in studies where the outcome of interest might vary between different subgroups. For example, the risk of dementia is suggested to vary by a number of characteristics, such as educational level, ethnicity, sex, and marital status.13,157,158 These characteristics also coincide with factors associated with refusing participation. As shown in supplementary figure S1 in Paper I, the group with the lowest response rate at all examinations was unmarried men with lower educational levels. This could have affected the estimates of mortality, incidence, and prevalence of dementia found in Papers III and IV. In Paper I, we also found in that those refusing to participate at age 70 more often had dementia according to the IPR. However, as shown in Paper IV, the sensitivity of the IPR and CDREG was only 43%, making it difficult to interpret this finding. It is also worth noting that the higher frequency of dementia diagnoses among refusals might be a result of this group being in less health, having more hospital visits, and thus a higher chance of receiving a dementia diagnose in the registers. Choice and use of diagnostic criteria We found in Paper II that the choice of diagnostic criteria for dementia has an important effect on the prevalence of cases identified. This is not only supported by previous studies,79,159,160 but also, to some extent, expected. During the past three to four decades, the research of dementia has made major progress in understanding the disorder, and the diagnostic systems have been updated accordingly. The criteria that yielded the highest prevalence were the latest ICD-11 and DSM-5 criteria, which corresponds well to the attempts to make the criteria more inclusive towards other dementias than primarily AD.2 The differences in criteria emphasise the need for harmonisation and awareness when comparing results. For example, a French study reported an increase in clinical consensus-diagnosed dementia over a ten-year period.15 However, the two cohorts that were compared had been diagnosed based on the DSM-III-R in the first cohort and DSM-5 in the second cohort. As we found in Paper II, DSM-5 yields a considerably higher prevalence than DSM-III-R. The finding in the French study of an increasing incidence of dementia could, therefore, rather be a result of the differences in two editions of the DSM. The French study also reported that D I SCUSSIO N 57 w hen diagnosing dementia with the same algorithmic approach in both cohorts, the incidence decreased. This additionally highlights the impact choice of diagnostic criteria has on the results. In Paper IV, those lost to follow-up were followed by register data, where the first cohort was coded in accordance with the ICD-9 and the second with ICD-10. As ICD-10 has been found to yield a higher prevalence than the ICD-9,79 there might be a risk that more cases were detected in the latter cohort. However, in Paper IV, we found that the sensitivity of register data was similar in all three cohorts. Many large population-based cohort studies on older adults were initiated in the late 1980s or early 1990s, which is why the DSM-III-R criteria for dementia is commonly used.15,16,18,92,144 The next edition, the DSM-IV, contained important differences, such as the change to allow for short-term or long-term memory deficits, as opposed to the requirement of deficits in both. In the most recent DSM-5, the differences compared to previous editions are even larger, as memory impairment is no longer required. These large differences put into question how future diagnosing of dementia in population-based studies from the 1980-1990s should be applied. Keeping the DSM-III-R diagnostic procedures enables cohort comparisons but reduces the possibility of estimating dementia prevalence that corresponds to modern criteria. One option would be to create two diagnoses, one based on modern criteria and one corresponding to the historically used criteria. This process is however, labour intensive and might not be possible in large population-based studies. Instead, algorithmic diagnoses based on different diagnostic criteria might be more feasible. There have been several attempts to produce algorithms with varying results.152,161,162 When comparing the overall accuracy of five different algorithms produced within the context of the Health and Retirement Study Discussion 6 9 58 H AN N A W E T T E R B E R G in the US, the researchers concluded that it was high enough to justify the use of algorithms.161 However, the performance varied across subgroups such as min itnhoer iUtiSes, ,t heed urecsaetaiorcnhael rgsr coounpcsl,u danedd tahgaet igt rwoausp hs.i gAh erneoceungth Atou jsutrsatilfiayn t hset uudsey ofof uanlgdo trhitahtm ths.e1i6r1 aHlgoowrietvhemr,s tfhoer pdeermfoernmtiaan accec voarrdieindg a tcor oDssS sMub-5g raonudp Ds sSuMch-I aVs hmaidn ohriigthie sa,c ceudruaccayt icoonmalp garreodu ptos, calinndic aalg ceo gnrsoeunpsus.s Adi argencoenset sA.16u2 sItnrateliraens tsintugdlyy, ifno utnhdis tphaapt etrh, eair saulgboseritt homf sc afsoers dweams ednitaiag naocsceodr dbinyg t wtoo D clSinMic-i5a nasn dto D tSesMt -tIhVe ihnatde rh-rigathe ra crceuliraabciyli tcyo (mκ =pa 0re.7d9 t, o9 5c%lin iCcaI;l 0c.o5n4s–e1n.s0u) sw dhiiacghn wosaess f.1o62u nIndt etore bstei nognllyy, sinli gthhtilsy phaigphere,r ath saunb tsheet oagf rceaesmese nwt abse tdwiaegenno tsheed cbliyn itcwiaon calnindi ctihaen sa lgtoo rtiethstm t h(κe =in0te.7r-2r,a t9e5r %re liCabI il0it.y6 2(κ– 0=.8 00.)7.9 ,T 9h5is% i sC Ii;n 0 .l5in4e– 1w.0i)t hw ha icshtu wdays ffrooumn dt thoe b1e9 o8n0lsy, srelipgohrtltyin hgi gthhaetr tthhea na gtrheee magerneet mbeetnwt ebeent wtheee ncl itnhiec acll icnoicnisaenn asunsd dthiaeg nalogsoisri tahnmd a(nκ a=lg0o.7r2it,h 9m5i%c d iCagI n0o.s6i2s –w0a.8s 0o)n. Tthhei ss aims ei nle vlienle a sw bitehtw ae esntu cdliyn ifcraol mra tethrse. 16139 T8h0iss, rtoepgeotrhtienrg w thitaht tthhee angorteioemn ethnat tb tehtwe eaelgno trhiteh cmli nhicoaldl sc ohnigsehn rseulsia dbiialigtyn oassi si ta insd n aont aalfgfeocrtitehdm bicy dsieacgunlaors itsr ewnadss o onr t hinet rsaapmeer sloevneall acsh banetgwese eans cmlinigichatl hraatpeprse.n16 3i nT hains taossgeestshoerr, wfiothrm thse narogtuiomne nthtsa t tthoew aarlgdo ruitshimng hoalnd s ahlgigohr itrhelmiaibci litayp pasr oita cish noint apfofpecutleadti obny- bsaesceudla srt utrdeinesd.s15 1o,1r6 3i nTthraepree rasroen aalsl oc hhaynbgrieds daess imgnigsh atv haialapbpleen, winh earne aa sssmesasoller,r sfuobrmsest oafr gthuem penarttsi citpoawnatsr dis urasnindgo mainz eda lgtoo rbiteh mexiacm ainpepdr owacithh ainn epxotpeunlsaitvieo ncl-ibnaicsaeld p srtoutdoiceos.l1,5 a1,n16d3 fTrhoemre t hairse carlesaot ihnygb mrido ddeelssi,g tnos p arveadiilcatb dlee,m wehnetriea aam smonagll etrh es uobthseetr opfa rtthicei ppaanrttsic.1i6p4 a nts is randomized to be examined with an extensive clinical protocol, and from this creating models, to predict dementia aAmmoonngg t hthee o dthiseard pvaarnttiacgipeasn otsf. 1u64s i ng algorithms are that the validity could be lowered as not all information can be taken into account.163 In comparison, iAnm cloinnigc atlh ceo dnisseandsvuasn tdaigaegsn oosfe us,s itnhge arlegsoerairtchhm fsi laer he atsh aotf ttehne bveaelind irteyv cieowuledd bine flouwll.e Irend a adsd intiootn a, lla incofonrcmerant iorani sceadn abgea intaskt eanl gionrtioth amccico udniat.g1n63o Isne sc iosm thpaatr itshoeny, min igclhint ihcaalv ec ognosoedn ssuesn dsiitaigvnitoy saens,d t hspe erceisfeicairtcyh f ofirl ep hreavs aolefntetn c baseeesn orfe vdieemweedn tiina bfuultl. Ilonw aedrd iatmiono,n ag cionncicdeernnt radiesemde nagtiaai ncsta saelsg.o16r2i,t1h65m Iict dwiaags naorsgeus eids tthhaatt ththeye mpriegvhatl ehnatv cea gseoso dar es eenassitieivr ittyo adnedte scpt ebcyif iacligtyo rfiothrm psr,e vaasl ethnet yc atesensd otfo dbeem menotriea bseuvte rleo wcoerm pamareodn gt oi ninccidideennt t ddeemmeennttiiaa ccaasseess.. 1A62l,1t6h5o uItg hw tahse aarimgu eind Pthapate rt hIeI pwraesv atole nidt ecnatsifeys parree veaalesinetr ctaos edse, teitc tq ubeys tailognosr ihthomws ,h iagsh t htheey steenndsi ttiovi tbye o mf othree saelgvoerriet hcmoms wpaarse fdo rto m iinldciedre cnats edse mofe dnetima ecnastieas.. Although the aim in Paper II was to identify prevalent cases, it questions how high the sensitivity of the algorithms was for milder cases of dementia. Mortality, incidence, and prevalence of dementia Mortality, incidence, and prevalence of dementia IMno Prtaaplietyr ,I IinI caindde nPcaep,e ar nIVd ,p wree vinavleenstcigea toef tdhee mtimene ttiare nds in the mortality, i ncidence, and prevalence of dementia among octogenarians. We found that tIhne P mapoertra lIiItyI aanmdo nPga ptheor sIeV w, withe ipnrveevsatliegnatt ed tehme etnimtiae tarte angdes 8in5 twhea sm loowrtearli tiyn, icnochidoertn 1ce9,2 a3n-2d4 p croemvapleanrecde otof cdoehmoernt t1ia9 0a1m-0o2n.g49 oPcrteovgieonuasr siatunds.i eWs ien vfoeustnigda tthinagt time trends in survival time with dementia has shown varying results. Results from the Swedish study KP and SNAC-K showed that the mortality among those with dementia decreased between the late 1980s and the 2000s,144 as D I SCUSSIO N 59 d id the French PAQUID comparing cohorts during the same time period.166 There are also studies showing the opposite, with higher mortality after dDI S em CUeSnStiIaO No nset, such as the Health and Retirement study (HRS) comparin5g9 7c0o horts examined in 2000 and 2010,167 the Cognitive FunctiHoan nan Wde tategrebienrg study (MRC CFAS) comparing cohorts examined between the early 1990s and 2010s,168, and the Framingham study comparing cohorts examined between the late 1970s and mid-2000s.169 The results from the Framingham study also showed an increased age of onset of dementia. The shorter survival time in these studies has been interpreted as a compression of morbidity, i.e. shorter periods of disease. It has been suggested that compression of morbidity in dementia could be a result of an increasing cognitive reserve, making the brain withstand a greater pathological load before clinical symptoms become detectable.74 However, once the onset of clinical symptoms occurs, the dementia is more severe and the disease course will progress more rapidly. In Paper III, we adjusted the mortality analyses for dementia severity as we previously have shown that the severity at age 85 decreased between the two cohorts.88 However, we did not specifically examine if there had been different changes in survival rate by dementia severity. Additional analyses performed for the purpose of this thesis shows that there was no interaction between the dementia severity and cohort in relation to mortality, indicating that the mortality decreased evenly in mild, moderate, and severe dementia. 60 H AN N A W E T T E R B E R G the mortality among those with prevalent dementia at age 85 was lower in cohort 1923-24 compared to cohort 1901-02.49 Previous studies investigating time trends in survival time with dementia has shown varying results. Results from the Swedish study KP and SNAC-K showed that the mortality among those with dementia decreased between the late 1980s and the 2000s,144 as did the French PAQUID comparing cohorts during the same time period.166 There are also studies showing the opposite, with higher mortality after dementia onset, such as the Health and Retirement study (HRS) comparing cohorts examined in 2000 and 2010,167 the Cognitive Function and ageing study (MRC CFAS) comparing cohorts examined between the early 1990s and 2010s,168, and the Framingham study comparing cohorts examined between the late 1970s and mid-2000s.169 The results from the Framingham study also showed an increased age of onset of dementia. The shorter survival time in these studies has been interpreted as a compression of morbidity, i.e. shorter periods of disease. It has been suggested that compression of morbidity in dementia could be a result of an increasing cognitive reserve, making the brain withstand a greater pathological load before clinical symptoms become detectable.74 However, once the onset of clinical symptoms occurs, the dementia is more severe and the disease course will progress more rapidly. In Paper III, we adjusted the mortality analyses for dementia severity as we previously have shown that the severity at age 85 decreased between the two cohorts.88 However, we did not specifically examine if there had been different changes in survival rate by dementia severity. Additional analyses performed for the purpose of this thesis shows that there was no interaction between the dementia severity and cohort in relation to mortality, indicating that the mortality decreased evenly in mild, moderate, and severe dementia. 60 H AN N A W E T T E R B E R G Discussion 7 1 We examined survival time from the examination, meaning that we cannot know if the survival time from the onset of the disease also changed. As depicted in Figure 9, there are a number of potential scenarios of changes in mortality in relation to dementia severity. The different scenarios depend on potential cohort changes occurring prior to the baseline examination at age 85. On the x-axis in Figure 9 are years prior to and after the baseline examination shown. The bars represent the median survival time from dementia onset, stratified by dementia severity and cohort. In a) shows that the median survival from the baseline examination was shorter by dementia severity in both cohorts, but was longer in cohort 1923-24 compared to 1901-02. We do however not know if this represents a compression of morbidity or longer survival with dementia, since we don’t know when the disease onset occurred. In b), it is assumed that the onset of dementia has not changed between the cohorts. If this was the case, the number of years lived with dementia would have increased. In c), the dementia onset is assumed to be postponed for the sWame ee xtiammei naesd t hseu srvuirvvaivl atli mafet efrr oagme 8th5e. Tehxaism winoautlido nm, emane atnhiantg t hteh astu rwveiv caal ntinmoet know if the survival time from the onset of the disease also changed. As depicted in Figure 9, there are a number of potential scenarios of changes in mortality in relation to dementia severity. The different scenarios depend on potential cohort changes occurring prior to the baseline examination at age 85. On the x-axis in Figure 9 are years prior to and after the baseline examination shown. The bars represent the median survival time from dementia onset, stratified by dementia severity and cohort. In a) shows that the median survival from the baseline examination was shorter by dementia severity in both cohorts, but was longer in cohort 1923-24 compared to 1901-02. We do however not know if this represents a compression of morbidity or longer survival with dementia, since we don’t know when the disease onset occurred. In b), it is assumed that the onset of dementia has not changed between the cohorts. If this was the case, the number of years lived with dementia would have increased. In c), the dementia onset is assumed to be postponed for the same time as the survival after age 85. This would mean that the survival time with dementia remained the same. In d), it is assumed that the onset of dementia has been postponed closer to the examination point. If this was the case, the survival with dementia would be shorter, and the number of years lived with the disease lower, i.e., the morbidity would be compressed. D I SCUSSIO N 61 with dementia remained the same. In d), it is assumed that the onset of dementia has been postponed closer to the examination point. If this was the case, the survival with dementia would be shorter, and the number of years lived with the disease lower, i.e., the morbidity would be compressed. D I SCUSSIO N 61 7 2 Hanna Wetterberg a) -4 -3 -2 -1 0 1 2 3 4 5 6 b) Severe - cohort 1901-02 Severe - cohort 1923-24 -4 -3 -2 -1 0 1 2 3 4 5 6 Moderate - cohort 1901-02 c) Moderate - cohort 1923-24 Mild - cohort 1901-02 Mild - cohort 1923-24 -4 -3 -2 -1 0 1 2 3 4 5 6 d) -4 -3 -2 -1 0 1 2 3 4 5 6 Figure 9 Figures over potential changes in survival with dementia Results from Paper III shows a lower mortality in both those with and without dementia from the baseline examination at age 85. Depicted in this figure are three potential scenarios of what happened with the timing of dementia onset prior to the examination. DAis cpuostseionnt ial effect of a lower mortality rate in individuals with dementia cou7ld3 be an increasing prevalence. However, in Paper IV, we show that the prevalence of dementia was decreasing between ages 85 and 88 in these two cohorts. We also show that the incidence declined between cohorts 1930 and 1901-02. However, the prevalence at age 85 between cohorts 1930 and 1901- 02, and the incidence between cohorts 1923-24 and 1901-02, did not change statistically, although the estimates were lower in the more recent cohorts. As the sample sizes were relatively small, it is difficult to evaluate whether there was no change or if it was a result of low power. In some studies where information on incidence of dementia was not available, changes in incidence was inferred based on mortality and prevalence data only.95,144 One of these studies compared populations of 78+ examined in 1995-97 and 2001-03. The results showed a decline in prevalence of dementia among men, who also had a lower mortality. Another study compared populations of 75+ examined in 1987-89 and 2001-2004. They did not find a difference in prevalence of dementia, but they found that the survival time with dementia increased.144 In both studies, this led to the conclusion that a decline in the incidence of dementia explained the lower and stable prevalence. The prevalence of dementia reported in Paper IV decreased from 30% to 25% at age 85, from 42% to 22% at age 88, and 42% to 37% at age 90. The prevalences we report are in line with previous studies; however, ours are somewhat higher. For example, in the Swedish SNAC-K study, the prevalence at ages 85-89 was reported to be around 18% in both the late 1980s and 2010s’.144 In the Spanish ZARADEMP-projects, the prevalence among participants >85 years was 16% and 18% in the late 1980s and mid- 1990s.170 In the British MRC-CFAS study the prevalence in the late 1980s was 24%, and in the 2010’s 16%, in participants aged 85-89.89 In this study, the identification of dementia cases was based on an algorithmic approach, which might be one explanation for the lower estimates. Another explanation for 62 H AN N A W E T T E R B E R G Figure 9 Figures over potential changes in survival with dementia Results from Paper III shows a lower mortality in both those with and without dementia from the baseline examination at age 85. Depicted in this figure are three potential scenarios of what happened with the timing of dementia onset prior to the examination. A potential effect of a lower mortality rate in individuals with dementia could be an increasing prevalence. However, in Paper IV, we show that the prevalence of dementia was decreasing between ages 85 and 88 in these two cohorts. We also show that the incidence declined between cohorts 1930 and 1901-02. However, the prevalence at age 85 between cohorts 1930 and 1901- 02, and the incidence between cohorts 1923-24 and 1901-02, did not change statistically, although the estimates were lower in the more recent cohorts. As the sample sizes were relatively small, it is difficult to evaluate whether there was no change or if it was a result of low power. In some studies where information on incidence of dementia was not available, changes in incidence was inferred based on mortality and prevalence data only.95,144 One of these studies compared populations of 78+ examined in 1995-97 and 2001-03. The results showed a decline in prevalence of dementia among men, who also had a lower mortality. Another study compared populations of 75+ examined in 1987-89 and 2001-2004. They did not find a difference in prevalence of dementia, but they found that the survival time with dementia increased.144 In both studies, this led to the conclusion that a decline in the incidence of dementia explained the lower and stable prevalence. The prevalence of dementia reported in Paper IV decreased from 30% to 25% at age 85, from 42% to 22% at age 88, and 42% to 37% at age 90. The prevalences we report are in line with previous studies; however, ours are somewhat higher. For example, in the Swedish SNAC-K study, the prevalence at ages 85-89 was reported to be around 18% in both the late 1980s and 2010s’.144 In the Spanish ZARADEMP-projects, the prevalence among participants >85 years was 16% and 18% in the late 1980s and mid- 1990s.170 In the British MRC-CFAS study the prevalence in the late 1980s was 24%, and in the 2010’s 16%, in participants aged 85-89.89 In this study, the identification of dementia cases was based on an algorithmic approach, which might be one explanation for the lower estimates. Another explanation for the lower rates in previous studies, compared to Paper IV, is that all studies, including previous studies from our research team, based the dementia diagnosis on the DSM-III-R criteria. However, as we found in Paper II, the H70 clinical consensus diagnosis has a higher agreement with the DSM-IV c62 ri teria, likely due to the change in memory requirement oHf AaNcNceAp WtiEnTgT dE RefBiEciRtGs in short-term or long-term memory. We also show that the prevalence of dementia presents higher prevalence when applying the DSM-IV compared to the DSM-III-R. This highlights the importance of the choice of diagnostic criteria in epidemiological studies on dementia. 7I4n Paper IV, we also found a decline in four-year demeHnatninaa iWnectitderebnecreg between ages 85-89, from IR 49/1,000 person-years to 23/1,000 person- years. The incidence rates we report are similar to findings in the Rotterdam study, where the five-year incidence rate in the age group 85-89 years was 31/1,000 person-years in the early 90s, and 26/1,000 person-years in year 2000.18 Other studies show higher incidence numbers in this age group. For example, the SNAC-K study presented rates of ten-year incidence in this age group from 108/1,000 person-years to 70/1,000 person years.91 The MRC- CFAS presented two-year incidence rates of 62/1,000 person-years to 49/1,000 person-years, in the age group 85+.16 Comparing incidence rates is not only difficult due to the point made above regarding how dementia is diagnosed. Previous studies also vary largely in the time of follow-up, including age groups in the age bands presented, and in methods of follow- up. For example, in the MRC CFAS study, a likelihood model to calculate the incidence of dementia among dropouts at follow-up was used.16 In PAQUID, information about refusals and deaths was collected from close informants and medical practitioners.15 The Rotterdam study had a similar approach, but only in persons with low cognition at baseline, and used information from the regional institute for outpatient mental health care for the total group.18 In the SNAC-K study, a similar approach as us ours were used, where information was collected from medical records and key informants in those who died between waves.91 Whether these different approaches explain the different results between studies is not clear. D I SCUSSIO N 63 the lower rates in previous studies, compared to Paper IV, is that all studies, including previous studies from our research team, based the dementia diagnosis on the DSM-III-R criteria. However, as we found in Paper II, the H70 clinical consensus diagnosis has a higher agreement with the DSM-IV criteria, likely due to the change in memory requirement of accepting deficits in short-term or long-term memory. We also show that the prevalence of dementia presents higher prevalence when applying the DSM-IV compared to the DSM-III-R. This highlights the importance of the choice of diagnostic criteria in epidemiological studies on dementia. In Paper IV, we also found a decline in four-year dementia incidence between ages 85-89, from IR 49/1,000 person-years to 23/1,000 person- years. The incidence rates we report are similar to findings in the Rotterdam study, where the five-year incidence rate in the age group 85-89 years was 31/1,000 person-years in the early 90s, and 26/1,000 person-years in year 2000.18 Other studies show higher incidence numbers in this age group. For example, the SNAC-K study presented rates of ten-year incidence in this age group from 108/1,000 person-years to 70/1,000 person years.91 The MRC- CFAS presented two-year incidence rates of 62/1,000 person-years to 49/1,000 person-years, in the age group 85+.16 Comparing incidence rates is not only difficult due to the point made above regarding how dementia is diagnosed. Previous studies also vary largely in the time of follow-up, including age groups in the age bands presented, and in methods of follow- up. For example, in the MRC CFAS study, a likelihood model to calculate the incidence of dementia among dropouts at follow-up was used.16 In PAQUID, information about refusals and deaths was collected from close informants and medical practitioners.15 The Rotterdam study had a similar approach, but only in persons with low cognition at baseline, and used information from the regional institute for outpatient mental health care for the total group.18 In the SNAC-K study, a similar approach as us ours were used, where information was collected from medical records and key informants in those who died between waves.91 Whether these different approaches explain the different results between studies is not clear. PPootteentniatl ieaxpll aenxaplanationsin dementia etpioindse fmori otimloe t rfeondrs tini mdeem tgy e rnetian edpsi demiology In the papers published during the past years indicating a decline in dementia prevalence and incidence, the most common explanatory models for the declining trend include better control of vascular risk factors, higher and better quality of education, and higher living standards.16,17,19,20,171 These DI SCUSSIO N 63 protective factors also overlap with factors associated with overall better health and lower mortality. However, the decline is yet to be fully understood, as studies have not been able to identify causal mechanisms to fully account for the decline.20 For example, The Framingham Heart study showed a decrease of 44% in the incidence of dementia between 1975 and early 2010, simultaneously with a decrease of vascular risk factors.17 However, the reduction in vascular risk factors did not fully explain the decreased incidence of dementia. Similarly, pooled data from the MoVIES and the MYHAT Dshisocuwsseidon t hat the decreasing incidence of 77%, when comparing a cohort bor7n5 in 1932-41 with a cohort born in 1902-1911, could not be explained by an increasing educational level.19 The PAQUID and Three-city studies compared populations examined in the 1990s and 2000s and showed that education and vascular factors explained only a small part of the 35% decrease in dementia incidence.15 The KP and SNAC-K showed a 30% decline in dementia incidence, comparing cohorts examined in the 2010s to the 1980s, and improvement in lifestyle factors, vascular disorders, education, and work conditions, only in part explained this decline.91 Simultaneously as the trends of declining dementia incidence and prevalence have occurred, large societal and medical events and developments have happened.171,172 In Figure 10, the widely different life courses lived by the cohorts included in the H70 studies are shown in relation to examples of important events and developments that have occurred during the past century.173 Individuals in the cohort born 1901-02 were born in a time of widespread poverty, and survived the Flu pandemic in their 20s. They were in their 40’s when penicillin became available, and treatment of hypertension was not available until they were in their 60s. The individuals in the latest cohort were born in 1930 during the Great depression and they were young during the World War II. However, governmental funding for maternal and perinatal care had been initiated, public food service in schools was developed 64 H AN N A W E T T E R B E R G Potential explanations for time trends in dementia epidemiology In the papers published during the past years indicating a decline in dementia prevalence and incidence, the most common explanatory models for the declining trend include better control of vascular risk factors, higher and better quality of education, and higher living standards.16,17,19,20,171 These protective factors also overlap with factors associated with overall better health and lower mortality. However, the decline is yet to be fully understood, as studies have not been able to identify causal mechanisms to fully account for the decline.20 For example, The Framingham Heart study showed a decrease of 44% in the incidence of dementia between 1975 and early 2010, simultaneously with a decrease of vascular risk factors.17 However, the reduction in vascular risk factors did not fully explain the decreased incidence of dementia. Similarly, pooled data from the MoVIES and the MYHAT showed that the decreasing incidence of 77%, when comparing a cohort born in 1932-41 with a cohort born in 1902-1911, could not be explained by an increasing educational level.19 The PAQUID and Three-city studies compared populations examined in the 1990s and 2000s and showed that education and vascular factors explained only a small part of the 35% decrease in dementia incidence.15 The KP and SNAC-K showed a 30% decline in dementia incidence, comparing cohorts examined in the 2010s to the 1980s, and improvement in lifestyle factors, vascular disorders, education, and work conditions, only in part explained this decline.91 Simultaneously as the trends of declining dementia incidence and prevalence have occurred, large societal and medical events and developments have happened.171,172 In Figure 10, the widely different life courses lived by the cohorts included in the H70 studies are shown in relation to examples of important events and developments that have occurred during the past century.173 Individuals in the cohort born 1901-02 were born in a time of widespread poverty, and survived the Flu pandemic in their 20s. They were in their 40’s when penicillin became available, and treatment of hypertension was not available until they were in their 60s. The individuals in the latest cohort were born in 1930 during the Great depression and they were young during the World War II. However, governmental funding for maternal and perinatal care had been initiated, public food service in schools was developed during their childhood, and they were in their 30s when the working week was changed from six to five days. Moreover, during the past century, large changes have occurred in food availability and diet, working conditions, access to information, and access and quality of education and care.171,174 The d64if ferences in the birth cohorts' life courses have affected HriAsNkN aAn Wd EpTrToEtReBctEiRveG f actors in diverse ways, which in turn could have influenced the risk of dementia.172 7 6 Hanna Wetterberg Figure 10 Different birth cohorts’ life courses in the view of important societal events. The risk of dementia should be considered in the broader historical and geographical context in which the cohort was born and lived their life. Source: Skoog et.al.,173 Statistics Sweden.175 D I SCUSSIO N 65 during their childhood, and they were in their 30s when the working week was changed from six to five days. Moreover, during the past century, large changes have occurred in food availability and diet, working conditions, access to information, and access and quality of education and care.171,174 The differences in the birth cohorts' life c urses have affected risk and protective factors in diverse ways, which in turn could have influenced the risk of dementia.172 Figure 10 Different birth cohorts’ life courses in the view of important societal events. The risk of dementia should be considered in the broader historical and geographical context in which the cohort was born and lived their life. Source: Skoog et.al.,173 Statistics Sweden.175 Discussion 7 7 D I SCUSSIO N 65 100 90 80 Life expectancy 70 60 Age 50 40 1 3 190 192 193 0 orn n n30 se b ose bor bor o th thos e Men rse o f th rse o f f 20 u u urs e o Women Life co Life co Life co 10 0 1900 1905 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 Increasing levels of education, better perinantal care, improved working conditions, access to healtcare and the wellfare state Widespread WWII Freezers Manned Fall of the poverty, poor WWI The Great Famine InFamine Depression Germany and Polio moonepidemics landings Berlin Wallperinantal care Holland Flu Pencillin Treatment of Treatment ofpandemic hypertension Cholesterol Radio Television Internet 06 CONCLUSION C06onCclOuNsiCoLnU S ION Epidemiological studies investigating the incidence, prevalence, and mortality of dementia are needed for the understanding of the societal and economic burden of the disease, as well as for policy makers to plan for the consequences of dementia on the public health and welfare system. Several studies published during the last decade have shown an overall trend toward a decline in the incidence and prevalence of dementia.15-17,19,91 Results from Paper IV showed that this trend might be continued also among octogenarians, an age group few studies have reported on. This age group is rapidly increasing, and a decline in the incidence and prevalence of dementia could slow down the alarming projected increase in the burden of dementia. In Paper III, we found that the survival time from age 85 has increased both in those with and without dementia. However, the relative risk of dementia on mortality remained similar between the cohorts. These findings indicate that individuals live with dementia to higher ages than in previous cohorts. Results from Paper III and IV can be used to further understand the time trends of dementia incidence, prevalence, and mortality. Moreover, to adequately plan for health care and welfare, it is imperative for policy makers to have up-to-date data to base the forecasting of the future epidemiology of dementia on. In Paper II, we found that the diagnostic criteria that have been used during the past decades differ, resulting in varying prevalences of dementia. The newest editions, ICD-11 and DSM-5, generated the highest prevalence of dementia compared to the ICD-10, DSM-III-R, DSM-IV, and the clinical consensus diagnosis. This finding highlights the importance of the choice of diagnostic tools, as it has a direct effect on the estimates. We also found that the agreement between the ICD-11 and DSM-5 was high, suggesting that the work to harmonize the two systems has to some extent been successful. Finally, in Paper I, we showed that participants in the H70-study cohort born in 1930 to some degree, differed from refusals, as they had higher educational 66 H AN N A W E T T E R B E R G Conclusion 8 1 Conclusion Epidemiological studies investigating the incidence, prevalence, and mortality of dementia are needed for the understanding of the societal and economic burden of the disease, as well as for policy makers to plan for the consequences of dementia on the public health and welfare system. Several studies published during the last decade have shown an overall trend toward a decline in the incidence and prevalence of dementia.15-17,19,91 Results from Paper IV showed that this trend might be continued also among octogenarians, an age group few studies have reported on. This age group is rapidly increasing, and a decline in the incidence and prevalence of dementia could slow down the alarming projected increase in the burden of dementia. In Paper III, we found that the survival time from age 85 has increased both in those with and without dementia. However, the relative risk of dementia on mortality remained similar between the cohorts. These findings indicate that individuals live with dementia to higher ages than in previous cohorts. Results from Paper III and IV can be used to further understand the time trends of dementia incidence, prevalence, and mortality. Moreover, to adequately plan for health care and welfare, it is imperative for policy makers to have up-to-date data to base the forecasting of the future epidemiology of dementia on. In Paper II, we found that the diagnostic criteria that have been used during the past decades differ, resulting in varying prevalences of dementia. The newest editions, ICD-11 and DSM-5, generated the highest prevalence of dementia compared to the ICD-10, DSM-III-R, DSM-IV, and the clinical consensus diagnosis. This finding highlights the importance of the choice of diagnostic tools, as it has a direct effect on the estimates. We also found that the agreement between the ICD-11 and DSM-5 was high, suggesting that the work to harmonize the two systems has to some extent been successful. Finally, in Paper I, we showed that participants in the H70-study cohort born in 1930 to some degree, differed from refusals, as they had higher educational lleevveellss aanndd mmeeaann iinnccoommee,, aanndd lloowweerr pprreevvaalleennccee ooff sseevveerraall ddiissoorrddeerrss.. TThhee ppaarrttiicciippaannttss wweerree mmoorree ssiimmiillaarr ttoo tthhee ttaarrggeett ppooppuullaattiioonn ooff ssaammee--aaggeedd iinnddiivviidduuaallss iinn GGootthheennbbuurrgg tthhaann ttoo rreeffuussaallss.. TThhee sseelleeccttiioonn bbiiaass mmiigghhtt hhaavvee rreessuulltteedd iinn lloowweerr eessttiimmaatteess ooff ddeemmeennttiiaa pprreevvaalleennccee aanndd iinncciiddeennccee,, aass tthhee rr6ee6ff uussaallss hhaadd mmoorree iillllnneessss aanndd lloowweerr eedduuccaattiioonnaall lleevveellss tthhaannH AtthhNeeN App aaWrrEttiiTccTiippEaaRnnBttEssR.. G FFuuttuurree PPpeerrrssspppeeecccttiitvviveee FFeeww ssttuuddiieess hhaavvee eexxaammiinneedd ttiimmee ttrreennddss iinn ddeemmeennttiiaa iinncciiddeennccee aanndd pprreevvaalleennccee aammoonngg ooccttooggeennaarriiaannss.. IItt iiss uunncclleeaarr wwhheetthheerr tthhee iinnccrreeaassiinngg lliiffee eexxppeeccttaannccyy iiss aassssoocciiaatteedd wwiitthh aa ccoommpprreessssiioonn ooff mmoorrbbiiddiittyy,, oorr iiff tthhee ttiimmee ssppeenntt iinn ddiisseeaassee wwiillll eexxppaanndd.. IItt iiss ddiiffffiiccuulltt ttoo aacchhiieevvee aa hhiigghh eennoouugghh nnuummbbeerr ooff ppaarrttiicciippaannttss iinn tthhiiss aaggee ggrroouupp ttoo eennssuurree tthhee ppoowweerr ttoo ddeetteecctt ssmmaalllleerr cchhaannggeess iinn ttiimmee ttrreennddss iinn ddeemmeennttiiaa eeppiiddeemmiioollooggyy.. FFuuttuurree ssttuuddiieess ccoommbbiinniinngg ddaattaa ffrroomm sseevveerraall lloonnggiittuuddiinnaall ppooppuullaattiioonn--bbaasseedd ssttuuddiieess ccoouulldd hhaavvee tthhee ppootteennttiiaall ttoo vvaalliiddaattee tthhee ttrreennddss iinn iinncciiddeennccee,, pprreevvaalleennccee,, aanndd mmoorrttaalliittyy ooff ddeemmeennttiiaa ffoouunndd iinn tthhiiss tthheessiiss.. TToo ffuurrtthheerr tthhee uunnddeerrssttaannddiinngg ooff hhooww sseelleeccttiioonn bbiiaass aaffffeeccttss tthhee eessttiimmaatteess ooff ddeemmeennttiiaa wwiillll bbee iimmppoorrttaanntt ffoorr rreettrriieevviinngg aaccccuurraattee eessttiimmaatteess ooff ddeemmeennttiiaa iinncciiddeennccee aanndd pprreevvaalleennccee.. IInn aaddddiittiioonn,, ffuuttuurree ppooppuullaattiioonn--bbaasseedd ssttuuddiieess mmiigghhtt bbeenneeffiitt ffrroomm aappppllyyiinngg mmeetthhooddss ttoo iinnccrreeaassee tthhee rreepprreesseennttaattiivveenneessss iinn ggrroouuppss kknnoowwnn ttoo rreeffuussee ppaarrttiicciippaattiioonn ttoo aa ggrreeaatteerr eexxtteenntt.. OOnnee ssuucchh aapppprrooaacchh ccoouulldd bbee wweeiigghhtteedd ssaammpplliinngg.. SSeevveerraall lloonngg--rruunnnniinngg ppooppuullaattiioonn--bbaasseedd ssttuuddiieess ddiiaaggnnoossee ddeemmeennttiiaa bbaasseedd oonn tthhee DDSSMM--IIIIII--RR,, ppuubblliisshheedd iinn 11998877.. AAss tthhee nneewweerr eeddiittiioonnss aarree mmoorree wwiiddeellyy ddeeffiinneedd ttoo ccoovveerr ssyymmppttoommss aassssoocciiaatteedd wwiitthh ootthheerr ddeemmeennttiiaass tthhaann AADD,, ffuuttuurree ssttuuddiieess wwoouulldd bbeenneeffiitt ffrroomm aappppllyyiinngg tthheessee ccrriitteerriiaa.. HHoowweevveerr,, ppeerrffoorrmmiinngg cclliinniiccaall ccoonnsseennssuuss ddiiaaggnnoosseess iiss ttiimmee--ccoonnssuummiinngg aanndd llaabboouurr iinntteennssiivvee.. AAnn aalltteerrnnaattiivvee wwoouulldd bbee ttoo aappppllyy aallggoorriitthhmmiicc ddiiaaggnnoosseess aanndd mmaacchhiinnee lleeaarrnniinngg mmeetthhooddss.. WWee ddoo,, hhoowweevveerr,, nneeeedd mmoorree kknnoowwlleeddggee aabboouutt tthhee aaddvvaannttaaggeess aanndd ttrraaddee--ooffffss ooff uussiinngg tthheessee mmeetthhooddss ffoorr ddiiaaggnnoossiinngg ddeemmeennttiiaa iinn ppooppuullaattiioonn-- bbaasseedd ssttuuddiieess.. TThhee eevvoolluuttiioonn ooff ddeemmeennttiiaa eeppiiddeemmiioollooggyy sshhoouulldd bbee ffoolllloowweedd cclloosseellyy,, aass tthhee 8 2 Hanna Wetterberg CCOO NNCCLL UUSSIIOO NN 6677 levels and mean income, and lower prevalence of several disorders. The participants were more similar to the target population of same-aged individuals in Gothenburg than to refusals. The selection bias might have resulted in lower estimates of dementia prevalence and incidence, as the refusals had more illness and lower educational levels than the participants. Future Perspective Few studies have examined time trends in dementia incidence and prevalence among octogenarians. It is unclear whether the increasing life expectancy is associated with a compression of morbidity, or if the time spent in disease will expand. It is difficult to achieve a high enough number of participants in this age group to ensure the power to detect smaller changes in time trends in dementia epidemiology. Future studies combining data from several longitudinal population-based studies could have the potential to validate the trends in incidence, prevalence, and mortality of dementia found in this thesis. To further the understanding of how selection bias affects the estimates of dementia will be important for retrieving accurate estimates of dementia incidence and prevalence. In addition, future population-based studies might benefit from applying methods to increase the representativeness in groups known to refuse participation to a greater extent. One such approach could be weighted sampling. Several long-running population-based studies diagnose dementia based on the DSM-III-R, published in 1987. As the newer editions are more widely defined to cover symptoms associated with other dementias than AD, future studies would benefit from applying these criteria. However, performing clinical consensus diagnoses is time-consuming and labour intensive. An alternative would be to apply algorithmic diagnoses and machine learning methods. We do, however, need more knowledge about the advantages and trade-offs of using these methods for diagnosing dementia in population- based studies. The evolution of dementia epidemiology should be followed closely, as the decline in prevalence and incidence is not necessarily a stable downward trend. For example, better treatment of vascular disorders is believed to have had a suppressing effect on the risk of dementia. The frequency of obesity and diabetes type II is, however, increasing,19 and this could, in turn, have a negative effect on the declining trend in dementia incidence and prevalence. TCOhNeCreLfUoSrIOe,N it is important to continuously examine new cohorts of olde6r7 a dults to update estimates of dementia incidence and prevalence. It is estimated that 40% of all dementia cases could be prevented,21 but despite this, no study has been able to fully explain the decline in dementia prevalence and incidence. Future studies examining the mechanisms of known risk factors for dementia, as well as studies searching for additional risk factors, are needed. In addition, studies on interventions aiming at reducing the risk of dementia at a population level are needed, as such interventions have the potential to, in a cost-effective way, promote brain- healthy lifestyles.176 68 H AN N A W E T T E R B E R G Conclusion 8 3 07 ACKNOWLEDGEMENT A0c7knowledgement ACKNOWLEDGEMENT The completion of this thesis would not have been possible without the help and support of many people whom I would like to take the opportunity to acknowledge, and to express my sincere gratitude towards. First and foremost, I want to thank all the participants of the Gothenburg H70 Birth Cohort study and the Prospective Population Study of Women for taking their time to contribute to science. I would like to express my deepest appreciation to my main supervisor, Ingmar Skoog. Thank you for your never ending patience and for believing in me, and for generously sharing your knowledge and expertise. To my supervisor, Anna Zettergren. Thank you for always providing the highest quality of feedback. Whether it was an abstract for a conference or the discussion part of this thesis, you have always taken the time to thoroughly consider my texts. Having your support has kept me floating through many rough times. To my supervisor, Silke Kern. Thank you for always believing in me. You have always made me feel that I can do it (“it” being anything). To my supervisor, Hanna Falk Erhag. Thank you for being such a strong role model. From the very first time we met when you supervised me in my master’s thesis, I have been inspired by your focus and dedication towards research. I also want to express my gratitude towards all my co-authors: Lina Rydén, Felicia Ahlner, Hanna Falk Erhag, Pia Gudmundsson, Xinxin Guo, Erik Joas, Lena Johansson, Silke Kern, Madeleine Mellqvist Fässberg, Jenna Najar, Mats Ribbe, Therese Rydberg Sterner, Jessica Samuelsson, Simona Sacuiu, Robert Sigström, Johan Skoog, Margda Waern, Anna Zettergren, and Ingmar Skoog. Thank you for your input and discussions, which not only improved the papers, but made me grow as a researcher. I have appreciated every comment and suggested change, and the more comments I received, the more I felt your engagement in my work. I want to acknowledge and thank Valter Sundh for the many times you’ve assisted me in statistical queries. ACKNOW L EDG EMENT 69 Acknowledgement 8 7 I received a great deal of help in the process of writing this thesis frame. Thank you Jessica Samuelsson, Anna Zettergren, Therese Rydberg Sterner, Lina Rydén, Pia Gudmundsson, Hanna Falk Erhag, and my sister, Nordigårds Emma Olsson (everyone should have a practically native English speaking sibling). I am not even sure that there would be a thesis frame without your suggestions, editing, and most importantly, support and encouragement. Thank you Felicia Ahlner, for sharing the burden of finalising a thesis. Another couple of weeks and we’re both Dr’s! To my current and former colleagues in the research group EPINEP. I want to thank you for all great discussions, insightful seminars during Friday breakfasts, and light-hearted lunches. Jenna Najar, working with you throughout these years have taught me so much. You have shared most generously not only your medical competence, your energy and drive, your time and thoughts, but most importantly, your friendship. Therese Rydberg Sterner, having you in my corner makes me feel stronger. I know that I can turn to you for advice both in research and “in real life”, and you will always make time (even though I don’t understand how?) to share your insightful thoughts. Felicia Ahlner, from showing me how to serve breakfast to our participants to discussing appropriate statistical methods, you have always provided a sound voice that has kept me grounded and on track. Jessica Samuelsson, our relationship has grown from chats at the copying machine every Friday, to co-PhD students, office room-mates, to friendship. During the past year, you have more and more become like a mentor to me. Pia Gudmundsson. Your warmth has kept my spirit high even during low days. Also, how many has a colleague who helped them prepare for giving birth (to a baby [no, not this thesis, an actual baby])? Lina Rydén, what started out as an organization project ended up in a peer reviewed article and a brand new interest for methodological research. I’ve learned so much during this process and the discussions we’ve had throughout the work. To Tina Jacobsson. Thank you for being the solid rock in our group. Your long experience of handling researchers and students in panic has been vital for me many times. To Ida Nordigårds. Thank you for spending evenings and weekends making the thesis more beautiful than I ever dreamed. Your dedication and interest in my work kept my inspiration going throughout this process. 70 H AN N A W E T T E R B E R G 8 8 Hanna Wetterberg To my parents, Carina and Erik Olsson. Thank you for always cheering me on and encouraging me. Your support has made me feel safe to walk new paths in life and gave me strength to pursue my doctoral studies. To my parents-in-law, Tomas and Kia Wetterberg. Thank you for believing in me, and for all stimulating dinner conversations we’ve had over the past eleven years. You’ve kept me growing and are a part of who I am today. To my son, Oliver. Thank you for being the light of my life. Your presence constantly reminds me of what really matters in life. I love you. Finally, to my beloved husband Sebastian Wetterberg. Thank you for leaving your hometown to come with me to Gothenburg for my internship, which eventually turned into my doctoral studies. This journey would not have been possible without your support, and your patience in listening and discussing both my frustrations and my sometimes manic ideas. Everything is much more fun with you by my side. ACKNOW L EDG EMENT 71 Acknowledgement 8 9 08 REFERENCES 0R8ef eRrEeFnEcReEs NCES 1. Karolinska Institutet. Svensk MeSH. Accessed 22, February, 2023. https://mesh.kib.ki.se/ 2. Sachdev PS, Blacker D, Blazer DG, et al. Classifying neurocognitive disorders: the DSM-5 approach. Nature Reviews Neurology. Nov 2014;10(11):634-642. 3. Nichols E, Steinmetz JD, Vollset SE, et al. Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease 2019. Lancet Public health. Feb 2022;7(2):E105-E125. 4. World Health Organization. Global status report on the public health response to dementia. 2021. 5. Socialstyrelsen. Vård och omsorg vid demenssjukdom. 2017. 6. van der Flier WM, Scheltens P. Epidemiology and risk factors of dementia. Journal of Neurology, Neurosurgery & Psychiatry. 2005;76(suppl 5):v2. 7. Jorm AF, Jolley D. The incidence of dementia - A meta-analysis. Neurology. Sep 1998;51(3):728-733. 8. Kodesh A. Prevalence and comorbidities of dementia in Israel: A nationally representative cohort study. Int Psychogeriatr. Jul 2019;31(7):1059-1063. 9. Crimmins EM, Saito Y, Kim JK, Zhang YS, Sasson I, Hayward MD. Educational Differences in the Prevalence of Dementia and Life Expectancy with Dementia: Changes from 2000 to 2010. J Gerontol B Psychol Sci Soc Sci. Apr 16 2018;73(suppl_1):S20-S28. 10. Cao Q, Tan CC, Xu W, et al. The Prevalence of Dementia: A Systematic Review and Meta-Analysis. J Alzheimers Dis. 2020;73(3):1157-1166. 11. Borjesson-Hanson A, Edin E, Gislason T, Skoog I. The prevalence of dementia in 95 year olds. Neurology. Dec 28 2004;63(12):2436-2438. 12. World Health Organization. The Global Dementia Observatory Reference Guide. 2018. 13. Mielke MM, Vemuri P, Rocca WA. Clinical epidemiology of Alzheimer's disease: assessing sex and gender differences. Clin Epidemiol. 2014;6:37-48. 14. Beam CR, Kaneshiro C, Jang JY, Reynolds CA, Pedersen NL, Gatz M. Differences Between Women and Men in Incidence Rates of Dementia and Alzheimer's Disease. J Alzheimers Dis. 2018;64(4):1077-1083. 15. Grasset L, Brayne C, Joly P, et al. Trends in dementia incidence: Evolution over a 10-year period in France. Alzheimers Dement. Mar 2016;12(3):272-80. 72 H AN N A W E T T E R B E R G 1 6. Matthews FE, Stephan BC, Robinson L, et al. A two decade dementia i ncidence comparison from the Cognitive Function and Ageing Studies I Referencesand II. Nat Commun. Apr 19 2016;7:11398. 9 3 17. Satizabal CL, Beiser AS, Chouraki V, Chene G, Dufouil C, Seshadri S. Incidence of Dementia over Three Decades in the Framingham Heart Study. N Engl J Med. Feb 11 2016;374(6):523-32. 18. Schrijvers EM, Verhaaren BF, Koudstaal PJ, Hofman A, Ikram MA, Breteler MM. Is dementia incidence declining? Trends in dementia incidence since 1990 in the Rotterdam Study. Neurology. 2012;78:1456-1463. 19. Sullivan KJ, Dodge HH, Hughes TF, et al. Declining Incident Dementia Rates Across Four Population-Based Birth Cohorts. J Gerontol A Biol Sci Med Sci. Oct 12 2018;74(9):1439-1445. 20. Wolters FJ, Chibnik LB, Waziry R, et al. Twenty-seven-year time trends in dementia incidence in Europe and the United States: The Alzheimer Cohorts Consortium. Neurology. Aug 4 2020;95(5):e519-e531. 21. Livingston G, Huntley J, Sommerlad A, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. Aug 8 2020;396(10248):413-446. 22. World Health Organization. Global action plan on the public health response to dementia 2017 - 2025. 2017. 23. Yousuf RM FA, Wai KT, Amran M, Akter SFU, Ramli M. . Potentially reversible causes of dementia. International Journal of Collaborative Research on Internal Medicine & Public Health. 2010;2(8):258-265. 24. Bello VME, Schultz RR. Prevalence of treatable and reversible dementias: A study in a dementia outpatient clinic. Dement Neuropsychol. Jan-Mar 2011;5(1):44-47. 25. Arvanitakis Z, Shah RC, Bennett DA. Diagnosis and Management of Dementia: Review. JAMA. Oct 22 2019;322(16):1589-1599. 26. Blennow K, de Leon MJ, Zetterberg H. Alzheimer's disease. Lancet. Jul-Aug 2006;368(9533):387-403. REF ERENCES 73 14. Beam CR, Kaneshiro C, Jang JY, Reynolds CA, Pedersen NL, Gatz M. Differences Between Women and Men in Incidence Rates of Dementia and Alzheimer's Disease. J Alzheimers Dis. 2018;64(4):1077-1083. 15. Grasset L, Brayne C, Joly P, et al. Trends in dementia incidence: Evolution over a 10-year period in France. Alzheimers Dement. Mar 2016;12(3):272-80. 16. Matthews FE, Stephan BC, Robinson L, et al. A two decade dementia incidence comparison from the Cognitive Function and Ageing Studies I and II. Nat Commun. Apr 19 2016;7:11398. 17. Satizabal CL, Beiser AS, Chouraki V, Chene G, Dufouil C, Seshadri S. Incidence of Dementia over Three Decades in the Framingham Heart Study. N Engl J Med. Feb 11 2016;374(6):523-32. 18. Schrijvers EM, Verhaaren BF, Koudstaal PJ, Hofman A, Ikram MA, Breteler MM. Is dementia incidence declining? Trends in dementia incidence since 1990 in the Rotterdam Study. Neurology. 2012;78:1456-1463. 19. Sullivan KJ, Dodge HH, Hughes TF, et al. Declining Incident Dementia Rates Across Four Population-Based Birth Cohorts. J Gerontol A Biol Sci Med Sci. Oct 12 2018;74(9):1439-1445. 20. Wolters FJ, Chibnik LB, Waziry R, et al. Twenty-seven-year time trends in dementia incidence in Europe and the United States: The Alzheimer Cohorts Consortium. Neurology. Aug 4 2020;95(5):e519-e531. 21. Livingston G, Huntley J, Sommerlad A, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. Aug 8 2020;396(10248):413-446. 22. World Health Organization. Global action plan on the public health response to dementia 2017 - 2025. 2017. 23. Yousuf RM FA, Wai KT, Amran M, Akter SFU, Ramli M. . Potentially reversible causes of dementia. International Journal of Collaborative Research on Internal Medicine & Public Health. 2010;2(8):258-265. 24. Bello VME, Schultz RR. Prevalence of treatable and reversible dementias: A study in a dementia outpatient clinic. Dement Neuropsychol. Jan-Mar 2011;5(1):44-47. 25. Arvanitakis Z, Shah RC, Bennett DA. Diagnosis and Management of Dementia: Review. JAMA. Oct 22 2019;322(16):1589-1599. 26. Blennow K, de Leon MJ, Zetterberg H. Alzheimer's disease. Lancet. Jul-Aug 2006;368(9533):387-403. 27. Maurer K, Volk S, Gerbaldo H. Auguste D and Alzheimer's disease. Lancet. May 24 1997;349(9064):1546-1549. 28. Knopman DS, Amieva H, Petersen RC, et al. Alzheimer disease. Nat Rev Dis Primers. May 13 2021;7(1) 29. Jack CR, Knopman DS, Jagust WJ, et al. Tracking pathophysiological REF ERENCpErSo cesses in Alzheimer's disease: an updated hypothetical model of dynam7ic3 biomarkers. Lancet Neurol. Feb 2013;12(2):207-216. 30. Armstrong RA. What causes alzheimer's disease? Folia Neuropathol. 2013;51(3):169-88. 9341 . Breijyeh Z, Karaman R. Comprehensive Review on AlzhHeaimnnear 'Ws eDttiesrebaesreg: Causes and Treatment. Molecules. Dec 8 2020;25(24) 32. Nebel RA, Aggarwal NT, Barnes LL, et al. Understanding the impact of sex and gender in Alzheimer's disease: A call to action. Alzheimers Dement. Sep 2018;14(9):1171-1183. 33. Blennow K, Zetterberg H. Biomarkers for Alzheimer's disease: current status and prospects for the future. J Intern Med. Dec 2018;284(6):643-663. 34. Bir SC, Khan MW, Javalkar V, Toledo EG, Kelley RE. Emerging Concepts in Vascular Dementia: A Review. J Stroke Cerebrovasc Dis. Aug 2021;30(8):105864. 35. O'Brien JT, Thomas A. Vascular dementia. The Lancet. 2015/10/24/ 2015;386(10004):1698-1706. 36. Borenstein A, Mortimer J. Alzheimer's disease. Life course perspectives on risk reduction. 1st edn. ed. Academic Press; 2016. 37. van der Flier WM, Skoog I, Schneider JA, et al. Vascular cognitive impairment. Nat Rev Dis Primers. 2018/02/15 2018;4(1):18003. 38. Walker Z, Possin KL, Boeve BF, Aarsland D. Lewy body dementias. Lancet. Oct 24 2015;386(10004):1683-1697. 39. Hogan DB, Jetté N, Fiest KM, et al. The Prevalence and Incidence of Frontotemporal Dementia: a Systematic Review. Canadian Journal of Neurological Sciences / Journal Canadien des Sciences Neurologiques. 2016;43(S1):S96-S109. 40. Montine TJ, Corrada MM, Kawas C, et al. Association of Cognition and Dementia With Neuropathologic Changes of Alzheimer Disease and Other Conditions in the Oldest Old. Neurology. 2022;99(10):e1067. 74 H AN N A W E T T E R B E R G 27. Maurer K, Volk S, Gerbaldo H. Auguste D and Alzheimer's disease. Lancet. May 24 1997;349(9064):1546-1549. 28. Knopman DS, Amieva H, Petersen RC, et al. Alzheimer disease. Nat Rev Dis Primers. May 13 2021;7(1) 29. Jack CR, Knopman DS, Jagust WJ, et al. Tracking pathophysiological processes in Alzheimer's disease: an updated hypothetical model of dynamic biomarkers. Lancet Neurol. Feb 2013;12(2):207-216. 30. Armstrong RA. What causes alzheimer's disease? Folia Neuropathol. 2013;51(3):169-88. 31. Breijyeh Z, Karaman R. Comprehensive Review on Alzheimer's Disease: Causes and Treatment. Molecules. Dec 8 2020;25(24) 32. Nebel RA, Aggarwal NT, Barnes LL, et al. Understanding the impact of sex and gender in Alzheimer's disease: A call to action. Alzheimers Dement. Sep 2018;14(9):1171-1183. 33. Blennow K, Zetterberg H. Biomarkers for Alzheimer's disease: current status and prospects for the future. J Intern Med. Dec 2018;284(6):643-663. 34. Bir SC, Khan MW, Javalkar V, Toledo EG, Kelley RE. Emerging Concepts in Vascular Dementia: A Review. J Stroke Cerebrovasc Dis. Aug 2021;30(8):105864. 35. O'Brien JT, Thomas A. Vascular dementia. The Lancet. 2015/10/24/ 2015;386(10004):1698-1706. 36. Borenstein A, Mortimer J. Alzheimer's disease. Life course perspectives on risk reduction. 1st edn. ed. Academic Press; 2016. 37. van der Flier WM, Skoog I, Schneider JA, et al. Vascular cognitive impairment. Nat Rev Dis Primers. 2018/02/15 2018;4(1):18003. 38. Walker Z, Possin KL, Boeve BF, Aarsland D. Lewy body dementias. Lancet. Oct 24 2015;386(10004):1683-1697. 39. Hogan DB, Jetté N, Fiest KM, et al. The Prevalence and Incidence of Frontotemporal Dementia: a Systematic Review. Canadian Journal of Neurological Sciences / Journal Canadien des Sciences Neurologiques. 2016;43(S1):S96-S109. 40. Montine TJ, Corrada MM, Kawas C, et al. Association of Cognition and Dementia With Neuropathologic Changes of Alzheimer Disease and Other Conditions in the Oldest Old. Neurology. 2022;99(10):e1067. 41. Boyle PA, Yu L, Leurgans SE, et al. Attributable risk of Alzheimer's dementia attributed to age-related neuropathologies. Ann Neurol. 2019/01/01 2019;85(1):114-124. 42. Kawas CH, Kim RC, Sonnen JA, Bullain SS, Trieu T, Corrada MM. Multiple pathologies are common and related to dementia in the oldest-old: The 90+ Study. Neurology. Aug 11 2015;85(6):535-42. 4743 . Kovacs GG, Alafuzoff I, Al-Sarraj S, et al. Mixed BH AraNinN AP WatEhToTloEgRiBesE RinG Dementia: The BrainNet Europe Consortium Experience. Dement Geriatr Cogn Disord. 2008;26(4):343-350. 44. James BD, Bennett DA, Boyle PA, Leurgans S, Schneider JA. Dementia from Alzheimer disease and mixed pathologies in the oldest old. JAMA. May 2 2012;307(17):1798-800. 45. White LR, Edland SD, Hemmy LS, et al. Neuropathologic comorbidity and cognitive impairment in the Nun and Honolulu-Asia Aging Studies. Neurology. Mar 15 2016;86(11):1000-8. References 9 5 46. Kapasi A, DeCarli C, Schneider JA. Impact of multiple pathologies on the threshold for clinically overt dementia. Acta Neuropathol. Aug 2017;134(2):171-186. 47. Corrada MM, Sonnen JA, Kim RC, Kawas CH. Microinfarcts are common and strongly related to dementia in the oldest-old: The 90+study. Alzheimers & Dementia. Aug 2016;12(8):900-908. 48. Valenzuela M, Esler M, Ritchie K, Brodaty H. Antihypertensives for combating dementia? A perspective on candidate molecular mechanisms and population-based prevention. Translational psychiatry. Apr 2012;2 49. Wetterberg H, Najar J, Ryden L, et al. Dementia remains the major predictor of death among octogenarians. A study of two population cohorts of 85-year-olds examined 22 years apart. Eur J Epidemiol. May 2021;36(5):507-517. 50. Hugo J, Ganguli M. Dementia and cognitive impairment: epidemiology, diagnosis, and treatment. Clin Geriatr Med. Aug 2014;30(3):421-42. 51. Chen J-H, Lin K-P, Chen Y-C. Risk Factors for Dementia. J Formos Med Assoc. 2009/10/01/ 2009;108(10):754-764. 52. Rabinovici GD. Late-onset Alzheimer Disease. Continuum (Minneap Minn). 2019;25(1):14-33. REF ERENCES 75 41. Boyle PA, Yu L, Leurgans SE, et al. Attributable risk of Alzheimer's dementia attributed to age-related neuropathologies. Ann Neurol. 2019/01/01 2019;85(1):114-124. 42. Kawas CH, Kim RC, Sonnen JA, Bullain SS, Trieu T, Corrada MM. Multiple pathologies are common and related to dementia in the oldest-old: The 90+ Study. Neurology. Aug 11 2015;85(6):535-42. 43. Kovacs GG, Alafuzoff I, Al-Sarraj S, et al. Mixed Brain Pathologies in Dementia: The BrainNet Europe Consortium Experience. Dement Geriatr Cogn Disord. 2008;26(4):343-350. 44. James BD, Bennett DA, Boyle PA, Leurgans S, Schneider JA. Dementia from Alzheimer disease and mixed pathologies in the oldest old. JAMA. May 2 2012;307(17):1798-800. 45. White LR, Edland SD, Hemmy LS, et al. Neuropathologic comorbidity and cognitive impairment in the Nun and Honolulu-Asia Aging Studies. Neurology. Mar 15 2016;86(11):1000-8. 46. Kapasi A, DeCarli C, Schneider JA. Impact of multiple pathologies on the threshold for clinically overt dementia. Acta Neuropathol. Aug 2017;134(2):171-186. 47. Corrada MM, Sonnen JA, Kim RC, Kawas CH. Microinfarcts are common and strongly related to dementia in the oldest-old: The 90+study. Alzheimers & Dementia. Aug 2016;12(8):900-908. 48. Valenzuela M, Esler M, Ritchie K, Brodaty H. Antihypertensives for combating dementia? A perspective on candidate molecular mechanisms and population-based prevention. Translational psychiatry. Apr 2012;2 49. Wetterberg H, Najar J, Ryden L, et al. Dementia remains the major predictor of death among octogenarians. A study of two population cohorts of 85-year-olds examined 22 years apart. Eur J Epidemiol. May 2021;36(5):507-517. 50. Hugo J, Ganguli M. Dementia and cognitive impairment: epidemiology, diagnosis, and treatment. Clin Geriatr Med. Aug 2014;30(3):421-42. 51. Chen J-H, Lin K-P, Chen Y-C. Risk Factors for Dementia. J Formos Med Assoc. 2009/10/01/ 2009;108(10):754-764. 52. Rabinovici GD. Late-onset Alzheimer Disease. Continuum (Minneap Minn). 2019;25(1):14-33. 53. Liu CC, Liu CC, Kanekiyo T, Xu H, Bu G. Apolipoprotein E and Alzheimer disease: risk, mechanisms and therapy. Nat Rev Neurol. Feb 2013;9(2):106- 18. 54. Corder EH, Saunders AM, Strittmatter WJ, et al. Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer's disease in late onset families. Science. Aug 13 1993;261(5123):921-3. 55. Husain MA, Laurent B, Plourde M. APOE and Alzheimer's Disease: From REF ERENCLEipS id Transport to Physiopathology and Therapeutics. Front Neurosc7i5. 2021;15:630502. 56. Farrer LA, Cupples LA, Haines JL, et al. Effects of age, sex, and ethnicity on the association between apolipoprotein E genotype and Alzheimer disease. A meta-analysis. APOE and Alzheimer Disease Meta Analysis Consortium. JAMA. Oct 22-29 1997;278(16):1349-56. 57. Bellenguez C, Küçükali F, Jansen IE, et al. New insights into the genetic etiology of Alzheimer’s disease and related dementias. Nat Genet. 2022/04/01 2022;54(4):412-436. 58. van der Lee SJ, Wolters FJ, Ikram MK, et al. The effect of APOE and other common genetic variants on the onset of Alzheimer's disease and dementia: a community-based cohort study. Lancet Neurol. May 2018;17(5):434-444. 59. Mauvais-Jarvis F, Bairey Merz N, Barnes PJ, et al. Sex and gender: modifiers of health, disease, and medicine. The Lancet. 2020;396(10250):565-582. 9 6 Hanna Wetterberg 60. Michelle M. Mielke. Sex and Gender Differences in Alzheimer Disease Dementia Psychiatr Times. Nov 2018;35(11):14-17. 61. Neu SC, Pa J, Kukull W, et al. Apolipoprotein E genotype and sex risk factors for Alzheimer disease: A meta-analysis. Article. JAMA Neurology. 2017;74(10):1178-1189. 62. Lehmann DJ, Refsum H, Nurk E, et al. Apolipoprotein E epsilon4 and impaired episodic memory in community-dwelling elderly people: a marked sex difference. The Hordaland Health Study. J Neurol Neurosurg Psychiatry. Aug 2006;77(8):902-8. 63. Andre A, Lu T, W. HV, D. GM. Sex modifies the APOE-related risk of developing Alzheimer disease. Ann Neurol. 2014;75(4):563-573. 64. Kawas C, Resnick S, Morrison A, et al. A prospective study of estrogen replacement therapy and the risk of developing Alzheimer's disease: the Baltimore Longitudinal Study of Aging. Neurology. Jun 1997;48(6):1517-21. 65. Matyi JM, Rattinger GB, Schwartz S, Buhusi M, Tschanz JT. Lifetime estrogen exposure and cognition in late life: the Cache County Study. 76 H AN N A W E T T E R B E R G 53. Liu CC, Liu CC, Kanekiyo T, Xu H, Bu G. Apolipoprotein E and Alzheimer disease: risk, mechanisms and therapy. Nat Rev Neurol. Feb 2013;9(2):106- 18. 54. Corder EH, Saunders AM, Strittmatter WJ, et al. Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer's disease in late onset families. Science. Aug 13 1993;261(5123):921-3. 55. Husain MA, Laurent B, Plourde M. APOE and Alzheimer's Disease: From Lipid Transport to Physiopathology and Therapeutics. Front Neurosci. 2021;15:630502. 56. Farrer LA, Cupples LA, Haines JL, et al. Effects of age, sex, and ethnicity on the association between apolipoprotein E genotype and Alzheimer disease. A meta-analysis. APOE and Alzheimer Disease Meta Analysis Consortium. JAMA. Oct 22-29 1997;278(16):1349-56. 57. Bellenguez C, Küçükali F, Jansen IE, et al. New insights into the genetic etiology of Alzheimer’s disease and related dementias. Nat Genet. 2022/04/01 2022;54(4):412-436. 58. van der Lee SJ, Wolters FJ, Ikram MK, et al. The effect of APOE and other common genetic variants on the onset of Alzheimer's disease and dementia: a community-based cohort study. Lancet Neurol. May 2018;17(5):434-444. 59. Mauvais-Jarvis F, Bairey Merz N, Barnes PJ, et al. Sex and gender: modifiers of health, disease, and medicine. The Lancet. 2020;396(10250):565-582. 60. Michelle M. Mielke. Sex and Gender Differences in Alzheimer Disease Dementia Psychiatr Times. Nov 2018;35(11):14-17. 61. Neu SC, Pa J, Kukull W, et al. Apolipoprotein E genotype and sex risk factors for Alzheimer disease: A meta-analysis. Article. JAMA Neurology. 2017;74(10):1178-1189. 62. Lehmann DJ, Refsum H, Nurk E, et al. Apolipoprotein E epsilon4 and impaired episodic memory in community-dwelling elderly people: a marked sex difference. The Hordaland Health Study. J Neurol Neurosurg Psychiatry. Aug 2006;77(8):902-8. 63. Andre A, Lu T, W. HV, D. GM. Sex modifies the APOE-related risk of developing Alzheimer disease. Ann Neurol. 2014;75(4):563-573. 64. Kawas C, Resnick S, Morrison A, et al. A prospective study of estrogen replacement therapy and the risk of developing Alzheimer's disease: the Baltimore Longitudinal Study of Aging. Neurology. Jun 1997;48(6):1517-21. 65. Matyi JM, Rattinger GB, Schwartz S, Buhusi M, Tschanz JT. Lifetime estrogen exposure and cognition in late life: the Cache County Study. Menopause-the Journal of the North American Menopause Society. Dec 2019;26(12):1366-1374. 66. Savolainen-Peltonen H, Rahkola-Soisalo P, Hoti F, et al. Use of postmenopausal hormone therapy and risk of Alzheimer's disease in Finland: nationwide case-control study. BMJ. Mar 6 2019;364:l665. 76 H AN N A W E T T E R B E R G 67. Whitmer RA, Quesenberry CP, Zhou JF, Yaffe K. Timing of Hormone Therapy and Dementia: The Critical Window Theory Revisited. Ann Neurol. Jan 2011;69(1):163-169. 68. Najar J, Östling S, Waern M, et al. Reproductive period and dementia: A 44-year longitudinal population study of Swedish women. Alzheimers Dement. 2020/06/23 2020;n/a(n/a) 69. Geerlings MI, Ruitenberg A, Witteman JCM, et al. Reproductive period and risk of dementia in postmenopausal women. Jama-J Am Med Assoc. Mar 21 2001;285(11):1475-1481. 70. Gilsanz P, Lee C, Corrada MM, Kawas CH, Quesenberry CP, Whitmer RA. Reproductive period and risk of dementia in a diverse cohort of health care members. Neurology. Apr 23 2019;92(17):E2005-E2014. 71. Mielke MM, Aggarwal NT, Vila-Castelar C, et al. Consideration of sex and gender in Alzheimer's disease and related disorders from a global perspective. Article. Alzheimers Dement. 2022;18(12):2707-2724. 72. Livingston G, Sommerlad A, Orgeta V, et al. Dementia prevention, Referencesi ntervention, and care. The Lancet. 2017/12/16/ 2017;390(10113):267937- 2734. 73. Skoog I, Lernfelt B, Landahl S, et al. 15-year longitudinal study of blood pressure and dementia. Lancet. Apr 27 1996;347(9009):1141-5. 74. Stern Y, Arenaza-Urquijo EM, Bartres-Faz D, et al. Whitepaper: Defining and investigating cognitive reserve, brain reserve, and brain maintenance. Alzheimers Dement. Sep 14 2018; 75. World Health Organization. International Classification of Diseases, Eleventh Revision (ICD-11). 2019/2021. https://icd.who.int/browse11 76. American Psychiatric Association. Diagnostic and statistical manual of mental disorders 5th ed. Am Psychiatric Assoc. 2013; 77. Alharbi MA, Isouard G, Tolchard B. Historical development of the statistical classification of causes of death and diseases. Cogent Medicine. 2021/01/01 2021;8(1):1893422. REF ERENCES 77 Menopause-the Journal of the North American Menopause Society. Dec 2019;26(12):1366-1374. 66. Savolainen-Peltonen H, Rahkola-Soisalo P, Hoti F, et al. Use of postmenopausal hormone therapy and risk of Alzheimer's disease in Finland: nationwide case-control study. BMJ. Mar 6 2019;364:l665. 67. Whitmer RA, Quesenberry CP, Zhou JF, Yaffe K. Timing of Hormone Therapy and Dementia: The Critical Window Theory Revisited. Ann Neurol. Jan 2011;69(1):163-169. 68. Najar J, Östling S, Waern M, et al. Reproductive period and dementia: A 44-year longitudinal population study of Swedish women. Alzheimers Dement. 2020/06/23 2020;n/a(n/a) 69. Geerlings MI, Ruitenberg A, Witteman JCM, et al. Reproductive period and risk of dementia in postmenopausal women. Jama-J Am Med Assoc. Mar 21 2001;285(11):1475-1481. 70. Gilsanz P, Lee C, Corrada MM, Kawas CH, Quesenberry CP, Whitmer RA. Reproductive period and risk of dementia in a diverse cohort of health care members. Neurology. Apr 23 2019;92(17):E2005-E2014. 71. Mielke MM, Aggarwal NT, Vila-Castelar C, et al. Consideration of sex and gender in Alzheimer's disease and related disorders from a global perspective. Article. Alzheimers Dement. 2022;18(12):2707-2724. 72. Livingston G, Sommerlad A, Orgeta V, et al. Dementia prevention, intervention, and care. The Lancet. 2017/12/16/ 2017;390(10113):2673- 2734. 73. Skoog I, Lernfelt B, Landahl S, et al. 15-year longitudinal study of blood pressure and dementia. Lancet. Apr 27 1996;347(9009):1141-5. 74. Stern Y, Arenaza-Urquijo EM, Bartres-Faz D, et al. Whitepaper: Defining and investigating cognitive reserve, brain reserve, and brain maintenance. Alzheimers Dement. Sep 14 2018; 75. World Health Organization. International Classification of Diseases, Eleventh Revision (ICD-11). 2019/2021. https://icd.who.int/browse11 76. American Psychiatric Association. Diagnostic and statistical manual of mental disorders 5th ed. Am Psychiatric Assoc. 2013; 77. Alharbi MA, Isouard G, Tolchard B. Historical development of the statistical classification of causes of death and diseases. Cogent Medicine. 2021/01/01 2021;8(1):1893422. 78. First MB, Gaebel W, Maj M, et al. An organization- and category-level comparison of diagnostic requirements for mental disorders in ICD-11 and DSM-5. World Psychiatry. Feb 2021;20(1):34-51. 79. Wancata J, Borjesson-Hanson A, Ostling S, Sjogren K, Skoog I. Diagnostic criteria influence dementia prevalence. Am J Geriatr Psychiatry. Dec 2007;15(12):1034-45. REF ERENCES 77 8 0. Henderson AS, Jorm AF, Mackinnon A, et al. A survey of dementia in the Canberra population: experience with ICD-10 and DSM-III-R criteria. Psychol Med. May 1994;24(2):473-82. 81. Riedel-Heller SG, Busse A, Aurich C, Matschinger H, Angermeyer MC. Incidence of dementia according to DSM-III-R and ICD-10 - Results of the Leipzig Longitudinal Study of the Aged (LEILA75+) Part 2. Br J Psychiatry. Sep 2001;179:255-260. 82. World Health Organization. The ICD-10 Classification of Mental and Behavioural Disorders: Diagnostic Criteria for Research. 1993. 83. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 3rd ed., revised. American Psychiatric Association. 1987. 84. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders: DSM-IV. 1994. 85. Rothman KJ. Epidemiology : an introduction. 2nd ed. Oxford University Press; 2012. 86. Prince M, Ali GC, Guerchet M, Prina AM, Albanese E, Wu YT. Recent global trends in the prevalence and incidence of dementia, and survival with dementia. Alzheimers Res Ther. Jul 30 2016;8(1):23. 9 8 Hanna Wetterberg 87. Sorbi S, Forleo P, Tedde A, et al. Genetic risk factors in familial Alzheimer's disease. Mech Ageing Dev. 2001/11/01/ 2001;122(16):1951-1960. 88. Skoog I, Borjesson-Hanson A, Kern S, et al. Decreasing prevalence of dementia in 85-year olds examined 22 years apart: the influence of education and stroke. Sci Rep. Jul 21 2017;7(1):6136. 89. Matthews FE, Arthur A, Barnes LE, Bond J, Jagger C, Robinson L. Medical Research Council Cognitive Function and Ageing Collaboration. A two- decade comparison of prevalence of dementia in individuals aged 65 years and older from three geographical areas of England: results of the Cognitive Function and Ageing Study I and II. Lancet. 2013;382(9902):1405-12. 90. Alzheimer’s Disease International. World Alzheimer Report 2015 2015. 78 H AN N A W E T T E R B E R G 78. First MB, Gaebel W, Maj M, et al. An organization- and category-level comparison of diagnostic requirements for mental disorders in ICD-11 and DSM-5. World Psychiatry. Feb 2021;20(1):34-51. 79. Wancata J, Borjesson-Hanson A, Ostling S, Sjogren K, Skoog I. Diagnostic criteria influence dementia prevalence. Am J Geriatr Psychiatry. Dec 2007;15(12):1034-45. 80. Henderson AS, Jorm AF, Mackinnon A, et al. A survey of dementia in the Canberra population: experience with ICD-10 and DSM-III-R criteria. Psychol Med. May 1994;24(2):473-82. 81. Riedel-Heller SG, Busse A, Aurich C, Matschinger H, Angermeyer MC. Incidence of dementia according to DSM-III-R and ICD-10 - Results of the Leipzig Longitudinal Study of the Aged (LEILA75+) Part 2. Br J Psychiatry. Sep 2001;179:255-260. 82. World Health Organization. The ICD-10 Classification of Mental and Behavioural Disorders: Diagnostic Criteria for Research. 1993. 83. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 3rd ed., revised. American Psychiatric Association. 1987. 84. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders: DSM-IV. 1994. 85. Rothman KJ. Epidemiology : an introduction. 2nd ed. Oxford University Press; 2012. 86. Prince M, Ali GC, Guerchet M, Prina AM, Albanese E, Wu YT. Recent global trends in the prevalence and incidence of dementia, and survival with dementia. Alzheimers Res Ther. Jul 30 2016;8(1):23. 87. Sorbi S, Forleo P, Tedde A, et al. Genetic risk factors in familial Alzheimer's disease. Mech Ageing Dev. 2001/11/01/ 2001;122(16):1951-1960. 88. Skoog I, Borjesson-Hanson A, Kern S, et al. Decreasing prevalence of dementia in 85-year olds examined 22 years apart: the influence of education and stroke. Sci Rep. Jul 21 2017;7(1):6136. 89. Matthews FE, Arthur A, Barnes LE, Bond J, Jagger C, Robinson L. Medical Research Council Cognitive Function and Ageing Collaboration. A two- decade comparison of prevalence of dementia in individuals aged 65 years and older from three geographical areas of England: results of the Cognitive Function and Ageing Study I and II. Lancet. 2013;382(9902):1405-12. 90. Alzheimer’s Disease International. World Alzheimer Report 2015 2015. 91. Ding M, Qiu C, Rizzuto D, Grande G, Fratiglioni L. Tracing temporal trends in dementia incidence over 25 years in central Stockholm, Sweden. Alzheimers Dement. May 2020;16(5):770-778. 92. Gao S, Ogunniyi A, Hall KS, et al. Dementia incidence declined in African- Americans but not in Yoruba. Alzheimers Dement. Mar 2016;12(3):244-51. 9783 . Prince M, Bryce R, Albanese E, Wimo A, Ribeiro W, HFAeNrrNi AC PW.E ThT Ee RgBloEbRaGl prevalence of dementia: A systematic review and metaanalysis. Alzheimers Dement. 2013;9(1):63-75.e2. 94. Langa KM, Larson EB, Karlawish JH, et al. Trends in the prevalence and mortality of cognitive impairment in the United States: is there evidence of a compression of cognitive morbidity? Alzheimers Dement. Mar 2008;4(2):134-44. 95. Wimo A, Sjolund BM, Skoldunger A, et al. Cohort Effects in the Prevalence and Survival of People with Dementia in a Rural Area in Northern Sweden. J Alzheimers Dis. 2016;50(2):387-96. 96. Ohara T, Hata J, Yoshida D, et al. Trends in dementia prevalence, incidence, and survival rate in a Japanese community. Neurology. May-16 88(20):1925- 1932. 97. Li S, Yan F, Li G, et al. Is the dementia rate increasing in Beijing? Prevalence and incidence of dementia 10 years later in an urban elderly population. Acta Psychiatr Scand. Jan 2007;115(1):73-79. 98. Tschanz JT, Corcoran C, Skoog I, et al. Dementia: the leading predictor of death in a defined elderly population: the Cache County Study. Neurology. Apr 13 2004;62(7):1156-62. 99. Todd S, Barr S, Roberts M, Passmore AP. Survival in dementia and predictors of mortality: a review. Int J Geriatr Psychiatry. 2013;28(11):1109- 1124. 100. Roehr S, Luck T, Bickel H, et al. Mortality in incident dementia - results Referencesf rom the German Study on Aging, Cognition, and Dementia in Prima9r9y Care Patients. Acta Psychiatr Scand. 2015;132(4):257-269. 101. Aevarsson O, Svanborg A, Skoog I. Seven-year survival rate after age 85 years: relation to Alzheimer disease and vascular dementia. Arch Neurol. Sep 1998;55(9):1226-32. 102. Brodaty H, Seeher K, Gibson L. Dementia time to death: a systematic literature review on survival time and years of life lost in people with dementia. Int Psychogeriatr. Jul 2012;24(7):1034-45. REF ERENCES 79 91. Ding M, Qiu C, Rizzuto D, Grande G, Fratiglioni L. Tracing temporal trends in dementia incidence over 25 years in central Stockholm, Sweden. Alzheimers Dement. May 2020;16(5):770-778. 92. Gao S, Ogunniyi A, Hall KS, et al. Dementia incidence declined in African- Americans but not in Yoruba. Alzheimers Dement. Mar 2016;12(3):244-51. 93. Prince M, Bryce R, Albanese E, Wimo A, Ribeiro W, Ferri CP. The global prevalence of dementia: A systematic review and metaanalysis. Alzheimers Dement. 2013;9(1):63-75.e2. 94. Langa KM, Larson EB, Karlawish JH, et al. Trends in the prevalence and mortality of cognitive impairment in the United States: is there evidence of a compression of cognitive morbidity? Alzheimers Dement. Mar 2008;4(2):134-44. 95. Wimo A, Sjolund BM, Skoldunger A, et al. Cohort Effects in the Prevalence and Survival of People with Dementia in a Rural Area in Northern Sweden. J Alzheimers Dis. 2016;50(2):387-96. 96. Ohara T, Hata J, Yoshida D, et al. Trends in dementia prevalence, incidence, and survival rate in a Japanese community. Neurology. May-16 88(20):1925- 1932. 97. Li S, Yan F, Li G, et al. Is the dementia rate increasing in Beijing? Prevalence and incidence of dementia 10 years later in an urban elderly population. Acta Psychiatr Scand. Jan 2007;115(1):73-79. 98. Tschanz JT, Corcoran C, Skoog I, et al. Dementia: the leading predictor of death in a defined elderly population: the Cache County Study. Neurology. Apr 13 2004;62(7):1156-62. 99. Todd S, Barr S, Roberts M, Passmore AP. Survival in dementia and predictors of mortality: a review. Int J Geriatr Psychiatry. 2013;28(11):1109- 1124. 100. Roehr S, Luck T, Bickel H, et al. Mortality in incident dementia - results from the German Study on Aging, Cognition, and Dementia in Primary Care Patients. Acta Psychiatr Scand. 2015;132(4):257-269. 101. Aevarsson O, Svanborg A, Skoog I. Seven-year survival rate after age 85 years: relation to Alzheimer disease and vascular dementia. Arch Neurol. Sep 1998;55(9):1226-32. 102. Brodaty H, Seeher K, Gibson L. Dementia time to death: a systematic literature review on survival time and years of life lost in people with dementia. Int Psychogeriatr. Jul 2012;24(7):1034-45. 103. Koedam ELGE, Pijnenburg YAL, Deeg DJH, et al. Early-Onset Dementia Is Associated with Higher Mortality. Dement Geriatr Cogn Disord. 2008;26(2):147-152. 104. Drivsholm T, Eplov LF, Davidsen M, et al. Representativeness in population-based studies: A detailed description of non-response in a Danish cohort study. Scandinavian Journal of Public Health. 2006/12/01 REF ERENC2E0S0 6;34(6):623-631. 79 105. Knudsen AK, Hotopf M, Skogen JC, Overland S, Mykletun A. The Health Status of Nonparticipants in a Population-based Health Study The Hordaland Health Study. Am J Epidemiol. Dec 11 2010;172(11):1306-1314. 106. Bergholdt HKM, Bathum L, Kvetny J, et al. Study design, participation and characteristics of the danish general suburban population study. Dan Med J. 2013;Sep;60(9):A469. 107. Lundberg I, Damström Thakker K, Hällström T, Forsell Y. Determinants of non-participation, and the effects of non-participation on potential cause-effect relationships, in the PART study on mental disorders. Soc Psychiatry Psychiatr Epidemiol. 2005/06/01 2005;40(6):475-483. 108. Korkeila K, Suominen S, Ahvenainen J, et al. Non-response and related factors in a nation-wide health survey. Eur J Epidemiol. 2001;17(11):991-999. 109. Bergstrand R, Vedin A, Wilhelmsson C, Wilhelmsen L. Bias Due to Non- Participation and Heterogenous Subgroups in Population Surveys. J Chronic Dis. 1983;36(10):725-728. 110. Chatfield MD, Brayne CE, Matthews FE. A systematic literature review of attrition between waves in longitudinal studies in the elderly shows a consistent pattern of dropout between differing studies. J Clin Epidemiol. Jan 2005;58(1):13-9. 111. Mody L, Miller DK, McGloin JM, et al. Recruitment and Retention of Older Adults in Aging Research. J Am Geriatr Soc. 2008/12/01 2008;56(12):2340- 2348. 112. Weuve J, Proust-Lima C, Power MC, et al. Guidelines for reporting methodological challenges and evaluating potential bias in dementia 1 0 0 research. Alzheimers & Dementia. Sep 2015;11(9):1098-1109H. a nna Wetterberg 113. Matthews FE, Chatfield M, Brayne C, Cfas M. An investigation of whether factors associated with short-term attrition change or persist over ten years: data from the Medical Research Council Cognitive Function and Ageing Study (MRC CFAS). BMC Public Health. Jul 18 2006;6 114. Rothman KJ, Gallacher JEJ, Hatch EE. Why representativeness should be avoided. Int J Epidemiol. Aug 2013;42(4):1012-1014. 80 H AN N A W E T T E R B E R G 103. Koedam ELGE, Pijnenburg YAL, Deeg DJH, et al. Early-Onset Dementia Is Associated with Higher Mortality. Dement Geriatr Cogn Disord. 2008;26(2):147-152. 104. Drivsholm T, Eplov LF, Davidsen M, et al. Representativeness in population-based studies: A detailed description of non-response in a Danish cohort study. Scandinavian Journal of Public Health. 2006/12/01 2006;34(6):623-631. 105. Knudsen AK, Hotopf M, Skogen JC, Overland S, Mykletun A. The Health Status of Nonparticipants in a Population-based Health Study The Hordaland Health Study. Am J Epidemiol. Dec 11 2010;172(11):1306-1314. 106. Bergholdt HKM, Bathum L, Kvetny J, et al. Study design, participation and characteristics of the danish general suburban population study. Dan Med J. 2013;Sep;60(9):A469. 107. Lundberg I, Damström Thakker K, Hällström T, Forsell Y. Determinants of non-participation, and the effects of non-participation on potential cause-effect relationships, in the PART study on mental disorders. Soc Psychiatry Psychiatr Epidemiol. 2005/06/01 2005;40(6):475-483. 108. Korkeila K, Suominen S, Ahvenainen J, et al. Non-response and related factors in a nation-wide health survey. Eur J Epidemiol. 2001;17(11):991-999. 109. Bergstrand R, Vedin A, Wilhelmsson C, Wilhelmsen L. Bias Due to Non- Participation and Heterogenous Subgroups in Population Surveys. J Chronic Dis. 1983;36(10):725-728. 110. Chatfield MD, Brayne CE, Matthews FE. A systematic literature review of attrition between waves in longitudinal studies in the elderly shows a consistent pattern of dropout between differing studies. J Clin Epidemiol. Jan 2005;58(1):13-9. 111. Mody L, Miller DK, McGloin JM, et al. Recruitment and Retention of Older Adults in Aging Research. J Am Geriatr Soc. 2008/12/01 2008;56(12):2340- 2348. 112. Weuve J, Proust-Lima C, Power MC, et al. Guidelines for reporting methodological challenges and evaluating potential bias in dementia research. Alzheimers & Dementia. Sep 2015;11(9):1098-1109. 113. Matthews FE, Chatfield M, Brayne C, Cfas M. An investigation of whether factors associated with short-term attrition change or persist over ten years: data from the Medical Research Council Cognitive Function and Ageing Study (MRC CFAS). BMC Public Health. Jul 18 2006;6 114. Rothman KJ, Gallacher JEJ, Hatch EE. Why representativeness should be avoided. Int J Epidemiol. Aug 2013;42(4):1012-1014. 115. Ranson JM, Kuźma E, Hamilton W, Muniz-Terrera G, Langa KM, Llewellyn DJ. Predictors of dementia misclassification when using brief cognitive assessments. Neurol Clin Pract. Apr 2019;9(2):109-117. 116. Rinder L, Roupe S, Steen B, Svanborg A. Seventy-year-old people in 80 Gothenburg. A population study in an industrialized Sw H AedNiNsAh cWitEyT. TAEcRtaB ME ReGd Scand. Nov 1975;198(5):397-407. 117. Rydberg Sterner T, Ahlner F, Blennow K, et al. The Gothenburg H70 Birth cohort study 2014-16: design, methods and study population. Eur J Epidemiol. Feb 2019;34(2):191-209. 118. Bengtsson C, Blohme G, Hallberg L, et al. The study of women in Gothenburg 1968-1969--a population study. General design, purpose and sampling results. Acta Med Scand. Apr 1973;193(4):311-8. 119. Skoog I. Mental disorders in the elderly. A population study in 85-year-olds. University of Gothenburg; 1993. 120. Fässberg MM, Vanaelst B, Jonson M, et al. Epidemiology of suicidal feelings in an ageing Swedish population: from old to very old age in the Gothenburg H70 Birth Cohort Studies. Epidemiol Psychiatr Sci. Apr 1 2019;29:e26. 121. Wetterberg H, Rydén L, Ahlner F, et al. Representativeness in population- based studies of older adults: five waves of cross-sectional examinations in the Gothenburg H70 Birth Cohort Study. Bmj Open. 2022;12(12):e068165. 122. Svanborg A. Seventy-year-old people in Gothenburg a population study in an industrialized Swedish city. II. General presentation of social and medical conditions. Acta Med Scand Suppl. 1977;611:5-37. 123. Asberg M, Montgomery SA, Perris C, Schalling D, Sedvall G. A comprehensive psychopathological rating scale. Acta Psychiatr Scand Suppl. 1978;(271):5-27. 124. Gottfries CG, Brane G, Steen G. A New Rating-Scale for Dementia Syndromes. Gerontology. 1982;28:20-31. 125. Hughes CP, Berg L, Danziger WL, Coben LA, Martin RL. A New Clinical- Scale for the Staging of Dementia. Br J Psychiatry. 1982;140(Jun):566-572. 126. Folstein MF FS, McHugh PR. "Mini-mental state". A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1 975;12(3):189. References 1 0 1 127. Rosen WG, Mohs RC, Davis KL. A new rating scale for Alzheimer's disease. Am J Psychiatry. Nov 1984;141(11):1356-64. REF ERENCES 81 115. Ranson JM, Kuźma E, Hamilton W, Muniz-Terrera G, Langa KM, Llewellyn DJ. Predictors of dementia misclassification when using brief cognitive assessments. Neurol Clin Pract. Apr 2019;9(2):109-117. 116. Rinder L, Roupe S, Steen B, Svanborg A. Seventy-year-old people in Gothenburg. A population study in an industrialized Swedish city. Acta Med Scand. Nov 1975;198(5):397-407. 117. Rydberg Sterner T, Ahlner F, Blennow K, et al. The Gothenburg H70 Birth cohort study 2014-16: design, methods and study population. Eur J Epidemiol. Feb 2019;34(2):191-209. 118. Bengtsson C, Blohme G, Hallberg L, et al. The study of women in Gothenburg 1968-1969--a population study. General design, purpose and sampling results. Acta Med Scand. Apr 1973;193(4):311-8. 119. Skoog I. Mental disorders in the elderly. A population study in 85-year-olds. University of Gothenburg; 1993. 120. Fässberg MM, Vanaelst B, Jonson M, et al. Epidemiology of suicidal feelings in an ageing Swedish population: from old to very old age in the Gothenburg H70 Birth Cohort Studies. Epidemiol Psychiatr Sci. Apr 1 2019;29:e26. 121. Wetterberg H, Rydén L, Ahlner F, et al. Representativeness in population- based studies of older adults: five waves of cross-sectional examinations in the Gothenburg H70 Birth Cohort Study. Bmj Open. 2022;12(12):e068165. 122. Svanborg A. Seventy-year-old people in Gothenburg a population study in an industrialized Swedish city. II. General presentation of social and medical conditions. Acta Med Scand Suppl. 1977;611:5-37. 123. Asberg M, Montgomery SA, Perris C, Schalling D, Sedvall G. A comprehensive psychopathological rating scale. Acta Psychiatr Scand Suppl. 1978;(271):5-27. 124. Gottfries CG, Brane G, Steen G. A New Rating-Scale for Dementia Syndromes. Gerontology. 1982;28:20-31. 125. Hughes CP, Berg L, Danziger WL, Coben LA, Martin RL. A New Clinical- Scale for the Staging of Dementia. Br J Psychiatry. 1982;140(Jun):566-572. 126. Folstein MF FS, McHugh PR. "Mini-mental state". A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12(3):189. 127. Rosen WG, Mohs RC, Davis KL. A new rating scale for Alzheimer's disease. Am J Psychiatry. Nov 1984;141(11):1356-64. 128. Skoog I, Nilsson L, Palmertz B, Andreasson LA, Svanborg A. A Population-Based Study of Dementia in 85-Year-Olds. N Engl J Med. Jan 21 1993;328(3):153-158. 129. Ingmar Skoog. Personal communication. November, 24, 2022. 1R3E0F.E RENCMEcSK hann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM81. Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer's Disease. Neurology. Jul 1984;34(7):939- 44. 131. Roman GC, Tatemichi TK, Erkinjuntti T, et al. Vascular dementia: diagnostic criteria for research studies. Report of the NINDS-AIREN International Workshop. Neurology. Feb 1993;43(2):250-60. 132. Lindblad U, Rastam L, Ranstam J, Peterson M. Validity of register data on acute myocardial infarction and acute stroke: the Skaraborg Hypertension Project. Scand J Soc Med. Mar 1993;21(1):3-9. 133. Wetterberg H, Najar J, Rydberg Sterner T, et al. Decreasing incidence and prevalence of dementia among octogenarians. A population-based study on three cohorts born 30 years apart. The Journals of Gerontology: Series A. 2023:glad071. 134. Harrison SL, Lang C, Whitehead C, et al. Trends in Prevalence of Dementia for People Accessing Aged Care Services in Australia. J Gerontol A Biol Sci Med Sci. Jan 20 2020;75(2):318-325. 135. Manton KC, Gu XL, Ukraintseva SV. Declining prevalence of dementia in the U.S. elderly population. Adv Gerontol. 2005;16:30-7. 136. Lissner L, Skoog I, Andersson K, et al. Participation bias in longitudinal studies: experience from the Population Study of Women in Gothenburg, Sweden. Scand J Prim Health Care. Dec 2003;21(4):242-7. 137. Knopman DS, Roberts RO, Pankratz VS, et al. Incidence of Dementia Among Participants and Nonparticipants in a Longitudinal Study of Cognitive Aging. Am J Epidemiol. Aug 15 2014;180(4):414-423. 138. Höstner J, Celin M. The Swedish population’s use of the Internet and Telephones—an individual survey 2017. Stockholm: Swedish National Post and Telecom agency. 2017;Report No: 1819-1701 139. Bexelius C, Merk H, Sandin S, et al. SMS versus telephone interviews for epidemiological data collection: feasibility study estimating influenza vaccination coverage in the Swedish population. Eur J Epidemiol. Feb 2009;24(2):73-81. 1 0 2 Hanna Wetterberg 82 H AN N A W E T T E R B E R G 128. Skoog I, Nilsson L, Palmertz B, Andreasson LA, Svanborg A. A Population-Based Study of Dementia in 85-Year-Olds. N Engl J Med. Jan 21 1993;328(3):153-158. 129. Ingmar Skoog. Personal communication. November, 24, 2022. 130. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer's Disease. Neurology. Jul 1984;34(7):939- 44. 131. Roman GC, Tatemichi TK, Erkinjuntti T, et al. Vascular dementia: diagnostic criteria for research studies. Report of the NINDS-AIREN International Workshop. Neurology. Feb 1993;43(2):250-60. 132. Lindblad U, Rastam L, Ranstam J, Peterson M. Validity of register data on acute myocardial infarction and acute stroke: the Skaraborg Hypertension Project. Scand J Soc Med. Mar 1993;21(1):3-9. 133. Wetterberg H, Najar J, Rydberg Sterner T, et al. Decreasing incidence and prevalence of dementia among octogenarians. A population-based study on three cohorts born 30 years apart. The Journals of Gerontology: Series A. 2023:glad071. 134. Harrison SL, Lang C, Whitehead C, et al. Trends in Prevalence of Dementia for People Accessing Aged Care Services in Australia. J Gerontol A Biol Sci Med Sci. Jan 20 2020;75(2):318-325. 135. Manton KC, Gu XL, Ukraintseva SV. Declining prevalence of dementia in the U.S. elderly population. Adv Gerontol. 2005;16:30-7. 136. Lissner L, Skoog I, Andersson K, et al. Participation bias in longitudinal studies: experience from the Population Study of Women in Gothenburg, Sweden. Scand J Prim Health Care. Dec 2003;21(4):242-7. 137. Knopman DS, Roberts RO, Pankratz VS, et al. Incidence of Dementia Among Participants and Nonparticipants in a Longitudinal Study of Cognitive Aging. Am J Epidemiol. Aug 15 2014;180(4):414-423. 138. Höstner J, Celin M. The Swedish population’s use of the Internet and Telephones—an individual survey 2017. Stockholm: Swedish National Post and Telecom agency. 2017;Report No: 1819-1701 139. Bexelius C, Merk H, Sandin S, et al. SMS versus telephone interviews for epidemiological data collection: feasibility study estimating influenza vaccination coverage in the Swedish population. Eur J Epidemiol. Feb 2009;24(2):73-81. 140. Callegaro M, McCutcheon AL, Ludwig J. Who's Calling? The Impact of Caller ID on Telephone Survey Response. Field Method. May 2010;22(2):175- 191. 141. Ravanam MS, Skalland B, Zhao Z, Yankey D, Smith C. An Evaluation of the Impact of Using an Alternate Caller ID Display in the National 82 Immunization Survey. Proceedings American Statistical H AANsNsoAci aWtiEoTn T EARnBnEuRalG Meeting. 2018;73 142. Thorvaldsson V, Karlsson P, Skoog J, Skoog I, Johansson B. Better Cognition in New Birth Cohorts of 70 Year Olds, But Greater Decline Thereafter. J Gerontol B-Psychol. Jan 1 2017;72(1):16-24. 143. Rajan KB, Weuve J, Barnes LL, Wilson RS, Evans DA. Prevalence and incidence of clinically diagnosed Alzheimer's disease dementia from 1994 to 2012 in a population study. Alzheimers Dement. 2019;15(1):1-7. 144. Qiu C, von Strauss E, Backman L, Winblad B, Fratiglioni L. Twenty-year changes in dementia occurrence suggest decreasing incidence in central Stockholm, Sweden. Neurology. May 14 2013;80(20):1888-94. 145. Hofman A, van Duijn CM, Franco OH, et al. The Rotterdam Study: 2012 objectives and design update. Eur J Epidemiol. Aug 2011;26(8):657-86. 146. Brooke HL, Talback M, Hornblad J, et al. The Swedish cause of death register. Eur J Epidemiol. Sep 2017;32(9):765-773. 147. Ludvigsson JF, Andersson E, Ekbom A, et al. External review and validation of the Swedish national inpatient register. BMC Public Health. 2011;11:450-450. 148. Ryden L, Sigstrom R, Nilsson J, et al. Agreement between self-reports, proxy-reports and the National Patient Register regarding diagnoses of cardiovascular disorders and diabetes mellitus in a population-based sample of 80-year-olds. Age Ageing. Jul 2019;48(4):513-518. 149. Rizzuto D, Feldman AL, Karlsson IK, Dahl Aslan AK, Gatz M, Pedersen NL. Detection of Dementia Cases in Two Swedish Health Registers: A Validation Study. J Alzheimers Dis. 2018;61(4):1301-1310. 150. Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology. Jan 1990;1(1):43-6. 151. Grasset L, Matthews FE, Peres K, et al. Evolution of dementia diagnosis over time (1988-2013): Evidence from French and English cohorts. Implication for secular trends analyses. Alzheimers Dement (Amst). 2018;10:490-497. References 1 0 3 REF ERENCES 83 152. Tschanz JT, Welsh-Bohmer KA, Skoog I, et al. Dementia diagnoses from clinical and neuropsychological data compared: the Cache County study. Neurology. Mar 28 2000;54(6):1290-6. 153. Reinikainen J, Tolonen H, Borodulin K, et al. Participation rates by educational levels have diverged during 25 years in Finnish health examination surveys. Eur J Public Health. Apr 2018;28(2):237-243. 154. Mindell JS, Giampaoli S, Goesswald A, et al. Sample selection, recruitment and participation rates in health examination surveys in Europe--experience from seven national surveys. BMC Med Res Methodol. Oct 5 2015;15:78. 155. Galea S, Tracy M. Participation Rates in Epidemiologic Studies. Ann Epidemiol. 2007/09/01/ 2007;17(9):643-653. 156. Hedlin D. Is there a 'safe area' where the nonresponse rate has only a modest effect on bias despite non-ignorable nonresponse? International Statistical Review. 2020/12/01 2020;88(3):642-657. 157. Najar J, Aakre JA, Vassilaki M, et al. Sex Difference in the Relation Between Marital Status and Dementia Risk in Two Population-Based Cohorts. Journal of Alzheimers Disease. 2021;83(3):1269-1279. 158. Kornblith E, Bahorik A, Boscardin WJ, Xia F, Barnes DE, Yaffe K. Association of Race and Ethnicity With Incidence of Dementia Among Older Adults. JAMA. 2022;327(15):1488-1495. 159. Pioggiosi P, Forti P, Ravaglia G, Berardi D, Ferrari G, De Ronchi D. Different classification systems yield different dementia occurrence among nonagenarians and centenarians. Dement Geriatr Cogn Disord. 2004;17(1- 2):35-41. 160. Erkinjuntti T, Ostbye T, Steenhuis R, Hachinski V. The effect of different diagnostic criteria on the prevalence of dementia. N Engl J Med. Dec 4 1997;337(23):1667-1674. 161. Gianattasio KZ, Wu Q, Glymour MM, Power MC. Comparison of Methods for Algorithmic Classification of Dementia Status in the Health and Retirement Study. Epidemiology. Mar 2019;30(2):291-302. 162. Eramudugolla R, Mortby ME, Sachdev P, Meslin C, Kumar R, Anstey KJ. Evaluation of a research diagnostic algorithm for DSM-5 neurocognitive disorders in a population-based cohort of older adults. Alzheimers Research & Therapy. Mar 4 2017;9 163. Copeland JRM, Dewey ME, Griffithsjones HM. A Computerized Psychiatric Diagnostic System and Case Nomenclature for Elderly Subjects - Gms and Agecat. Psychol Med. Feb 1986;16(1):89-99. 164. Mayeda ER, Shaw C. Algorithmic dementia classification: promises and challenges. Am J Epidemiol. Jan 5 2023; 165. Nichols E, Ng DK, James BD, Deal JA, Gross AL. The application of cross-sectionally derived dementia algorithms to longitudinal data in risk 84 factor analyses. Ann Epidemiol. 2023/01/01/ 2023;77:7H8A-8N4N.A W E T T E R B E R G 1 10646 . Grasset L, Peres K, Joly P, et al. Secular trends of mortalitHya annda Wdeetmteernbteirag- free life expectancy over a 10-year period in France. Eur J Epidemiol. Feb 2019;34(2):115-123. 167. Crimmins EM, Saito Y, Kim JK. Change in Cognitively Healthy and Cognitively Impaired Life Expectancy in the United States: 2000-2010. SSM Popul Health. Dec 2016;2:793-797. 168. Jagger C, Matthews FE, Wohland P, et al. A comparison of health expectancies over two decades in England: results of the Cognitive Function and Ageing Study I and II. Lancet. Feb 20 2016;387(10020):779- 86. 169. Dufouil C, Beiser A, Chêne G, Seshadri S. Are Trends in Dementia Incidence Associated With Compression in Morbidity? Evidence From The Framingham Heart Study. The Journals of Gerontology: Series B. 2018;73(suppl_1):S65-S72. 170. Lobo A, Saz P, Marcos G, et al. Prevalence of dementia in a southern European population in two different time periods: the ZARADEMP Project. Acta Psychiatr Scand. Oct 2007;116(4):299-307. 171. Rocca WA. Time, Sex, Gender, History, and Dementia. Review. Alzheimer Dis Assoc Disord. 2017;31(1):76-79. 172. Skoog I, Najar J, Wetterberg H. The Importance of Birth Year for the Incidence of Dementia. J Am Geriatr Soc. 2019;67(7):1330-1332. 173. Skoog I. Dementia: Dementia incidence - the times, they are a-changing. Nat Rev Neurol. Jun 2016;12(6):316-8. 174. Rydberg Sterner T. Depression among Swedish 70-year-olds: sex differences from a gender perspective. University of Gothenburg; 2020. 175. Statistics Sweden - Statistical database. Life expectancy 1751–2021. Accessed 7, March, 2023. https://www.scb.se/en/finding- statistics/statistics-by-subject-area/population/population- composition/population-statistics/pong/tables-and-graphs/births-and- deaths/life-expectancy-17512021/ REF ERENCES 85 164. Mayeda ER, Shaw C. Algorithmic dementia classification: promises and challenges. Am J Epidemiol. Jan 5 2023; 165. Nichols E, Ng DK, James BD, Deal JA, Gross AL. The application of cross-sectionally derived dementia algorithms to longitudinal data in risk factor analyses. Ann Epidemiol. 2023/01/01/ 2023;77:78-84. 166. Grasset L, Peres K, Joly P, et al. Secular trends of mortality and dementia- free life expectancy over a 10-year period in France. Eur J Epidemiol. Feb 2019;34(2):115-123. 167. Crimmins EM, Saito Y, Kim JK. Change in Cognitively Healthy and Cognitively Impaired Life Expectancy in the United States: 2000-2010. SSM Popul Health. Dec 2016;2:793-797. 168. Jagger C, Matthews FE, Wohland P, et al. A comparison of health expectancies over two decades in England: results of the Cognitive Function and Ageing Study I and II. Lancet. Feb 20 2016;387(10020):779- 86. 169. Dufouil C, Beiser A, Chêne G, Seshadri S. Are Trends in Dementia Incidence Associated With Compression in Morbidity? Evidence From The Framingham Heart Study. The Journals of Gerontology: Series B. 2018;73(suppl_1):S65-S72. 170. Lobo A, Saz P, Marcos G, et al. Prevalence of dementia in a southern European population in two different time periods: the ZARADEMP Project. Acta Psychiatr Scand. Oct 2007;116(4):299-307. 171. Rocca WA. Time, Sex, Gender, History, and Dementia. Review. Alzheimer Dis Assoc Disord. 2017;31(1):76-79. 172. Skoog I, Najar J, Wetterberg H. The Importance of Birth Year for the Incidence of Dementia. J Am Geriatr Soc. 2019;67(7):1330-1332. 173. Skoog I. Dementia: Dementia incidence - the times, they are a-changing. Nat Rev Neurol. Jun 2016;12(6):316-8. 174. Rydberg Sterner T. Depression among Swedish 70-year-olds: sex differences from a gender perspective. University of Gothenburg; 2020. 175. Statistics Sweden - Statistical database. Life expectancy 1751–2021. Accessed 7, March, 2023. https://www.scb.se/en/finding- statistics/statistics-by-subject-area/population/population- composition/population-statistics/pong/tables-and-graphs/births-and- deaths/life-expectancy-17512021/ 176. Walsh S, Govia I, Peters R, et al. What would a population-level approach to dementia risk reduction look like, and how would it work? Alzheimers Dement. 2023/02/15 2023;n/a(n/a) REF ERENCES 85 References 1 0 5 86 H AN N A W E T T E R B E R G