Using an Overlapping Time Interval Strategy to Study Diagnostic Instability in Mild Cognitive Impairment Subtypes

b ain sciences A icle Using an O e lapping Time In e al S a egy o S udy Diagnos ic Ins abili y in Mild Cogni i e Impai men Sub ypes Da id Facal 1,* , Joan Guà dia-Olmos 2, A u o X. Pe ei o 1, C is ina Lojo-Seoane 1, Ma ibel Pe ó2and Onésimo Juncos-Rabadán1 1Depa men o De elopmen al Psychology, Uni e si y de San iago de Compos ela, 15782 San iago de Compos ela, Galicia, Spain; a u oxose.pe [email p o ec ed] (A.X.P.); [email p o ec ed] (C.L.-S.); [email p o ec ed] (O.J.-R.) 2Depa men o Me hodology o Beha iou al Sciences, Uni e si y o Ba celona, 08035 Ba celona, Ca alunya, Spain; jgua [email p o ec ed] (J.G.-O.); [email p o ec ed] (M.P.) *Co espondence: [email p o ec ed] Recei ed: 21 Augus 2019; Accep ed: 18 Sep embe 2019; Published: 19 Sep embe 2019   Abs ac : (1) Backg ound: Mild cogni i e impai men (MCI) is a diagnos ic label in which s abili y is ypically low. The aim o his s udy was o examine empo al changes in he diagnosis o MCI sub ypes by using an o e lapping- ime s a egy; (2) Me hods: The s udy included 435 pa icipan s aged o e 50 yea s wi h subjec i e cogni i e complain s and who comple ed a leas one ollow-up e alua ion. The p obabili y o ansi ion was es ima ed using Bayesian odds a ios; (3) Resul s: Wi hin he di e en ime in e als, he con ols wi h subjec i e cogni i e complain s ep esen ed he la ges p opo ion o pa icipan s, ollowed by sda-MCI a baseline and in he i s i e in e als o he ollow-up, bu no in he las eigh in e als. The odds a ios indica ed highe odds o con e sion o demen ia in sda-MCI and mda-MCI g oups ela i e o na-MCI (e.g., in e al 9–15 mon hs—sda-MCI OR =9 and mda-MCI OR =3.36; in e al 27–33—sda-MCI OR =16 and mda-MCI =5.06; in e al 42–48—sda-MCI OR =8.16 and mda-MCI =3.45; in e al 45–51—sda-MCI OR =3.31 and mda-MCI = 1); (4) Conclusions: No able pa e ns o ins abili y consis en wi h he cu en li e a u e we e obse ed. The limi a ions o a p ospec i e app oach in he s udy o MCI ansi ions a e discussed. Keywo ds: cogni i e aging; mild cogni i e impai men ; subjec i e cogni i e complain s; con e sion o demen ia; Bayesian odds a ios; ime o e lapping in e als; sc eening and diagnosis 1. In oduc ion Mild cogni i e impai men (MCI) is he diagnos ic en i y used o desc ibe a condi ion in middle-aged and old adul s who expe ience cogni i e decline bu wi hou impai ed daily unc ioning [ 1 ]. MCI has been indica ed as a ocal concep o unde s anding p e-demen ia s ages in ela ion o cogni i e ageing [ 2 ]. Despi e some con o e sy, he ollowing a e gene ally accep ed as he co e MCI c i e ia: (i) E idence o subjec i e cogni i e complain s om pa ien s o hei close con ac s; (ii) e idence o cogni i e impai men in one o mo e cogni i e domains ha is g ea e han expec ed o he pa ien ’s age and educa ional backg ound; (iii) p ese a ion o minimal impai men o ins umen al ac i i ies o daily li ing; and (i ) non- ul illmen o diagnos ic c i e ia o demen ia [ 3 – 7 ]. This consensus also ex ends o he di e en sub ypes o MCI (amnes ic and non-amnes ic, single and mul i-domain) p oposed by Pe e sen and colleagues [ 3 , 6 ]. Apa om i s impo ance in esea ch on cogni i e impai men , MCI is also a mul i ace ed clinical label, cha ac e ized by complex cogni i e changes and diagnos ic ins abili y in MCI sub ypes, con e sion o demen ia and eco e y o no mal cogni i e aging [8–10]. B ain Sci. 2019,9, 242; doi:10.3390/b ainsci9090242 www.mdpi.com/jou nal/b ainsci B ain Sci. 2019,9, 242 2 o 13 Al hough majo ad ances in esea ch ha e led o co esponding changes in he diagnos ic c i e ia o MCI, i emains unclea how cogni i e de ici s inc ease o e ime and how hese de ici s a ec p og ession o demen ia in di e en MCI sub ypes [ 1 , 11 ]. Acco ding o Ge s eneke and Mas [ 1 ], he ela ionship be ween MCI sub ypes and ansi ion and con e sion pa e ns is no as linea as ini ially p oposed. Amnes ic MCI is conside ed o be he MCI sub ype mos likely o p og ess o AD. The diagnos ic guidelines p oduced by he Na ional Ins i u e on Aging- Alzheime ’s disease wo king g oups [ 5 ] indica e ha he ‘MCI, due o AD’, e e s o he symp oma ic p e-demen ia phase o AD and is cha ac e ized by impai men in one o mo e cogni i e domains, ypically including episodic memo y. Di e en s udies ha e al eady shown ha mul i-domain amnes ic MCI (mda-MCI) is he mos eliable MCI sub ype, wi h a lowe chance o e e sion and a highe isk o con e sion o demen ia o e ime [12,13]. Al hough MCI is unde s ood o be an uns able s age and he passage o he ime is ele an in i s cogni i e and clinical mani es a ions, scan a en ion has been gi en in he li e a u e o he ime be ween assessmen s -bu see [ 9 , 14 ]. Es ablishing his in e al may ha e a conside able impac on he indings o MCI esea ch. Fo example, longi udinal s udies ha e de eloped non-linea , pla eau models o decline ha a e dependen on he ime o measu emen [ 15 ]. In he cu en li e a u e on MCI, he da e o ini ial diagnosis and subsequen measu emen poin s a e de e mined as a unc ion o he s udy s a da e; his is p oblema ic as he momen a which pa icipan s epo subjec i e cogni i e complain s (SCCs) is in insically a iable [ 6 ]. To add ess his p oblem, Clou ie e al. [ 11 ] sugges ed aligning he momen e ospec i ely acco ding o he yea in which pa icipan s ecei ed hei diagnosis o AD. Facal e al. [ 9 ] p oposed an al e na i e, complemen a y app oach in ol ing an o e lapping in e al s a egy ha conside s di e en mid-poin s ages depending on he ime be ween he baseline and ollow-up assessmen . The aim o he p esen s udy was o es he capaci y o an o e lapping ime in e al s a egy o add ess changes in MCI, by including supplemen a y ollow-up e alua ions and b oade ime in e als. The o e lapping in e al s a egy was applied in he cu en sample o each ime in e al o explo e whe he changes (decline, s abili y, eco e y) a e empo ally s able o a y acco ding o di e en ime in e als, and also o examine he ajec o ies o cogni i e unc ion in di e en MCI sub ypes wi hin di e en in e als. 2. Ma e ials and Me hods 2.1. Pa icipan s The cu en esea ch included 435 pa icipan s o e 50 yea s old ( ange 50–88) included in he on-going Compos ela aging s udy and who comple ed baseline and a leas one ollow-up e alua ion [ 16 ]. All pa icipan s we e e e ed o us by gene al p ac i ione s a e a ending p ima y ca e heal h cen e s wi h subjec i e cogni i e complain s, bu wi h no p io diagnosis o demen ia, psychia ic o neu ological diso de s. A baseline, he sample comp ised 40 pa icipan s in he mda-MCI g oup (mul i-domain amnes ic MCI), 34 pa icipan s in he na-MCI g oup (no amnes ic MCI), 68 pa icipan s in he sda-MCI, and 293 pa icipan s in he con ol g oup wi h subjec i e cogni i e complain s (SCCs). Age, yea s o educa ion and esul s o he cogni i e assessmen a baseline a e shown in Table 1. G oup di e ences we e calcula ed using non-pa ame ic es s (K uskal-Wallis and Mann-Whi ney es ) gi en he di e ences in sample size be ween he g oups. Diagnoses we e made ollowing he Pe e sen c i e ia [ 3 , 4 ] upda ed by Albe e al. [ 5 ] and we e eached by consensus a esea ch eam diagnos ic mee ings. B ain Sci. 2019,9, 242 3 o 13 Table 1. Mean alues and s anda d de ia ions (in pa en heses) o he demog aphic and cogni i e measu es in each g oup a baseline. mda-MCI na-MCI sda-MCI SCCs G oup Di e ences χ2(gl) G oup Compa isons Age 72.42 (8.33) 67.12 (8.82) 69.29 (9.36) 65.49 (9.05) 24.83 ** mda-MCI >naMCI, SCCs; sda-MCI>SCCs Yea s o educa ion 9.68 (8.82) 8.29 (3.75) 9.40 (4.17) 9.92 (4.70) 1.81 Memo y complain s—pa icipan 19.43 (4.69) 20.35 (3.45) 19.03 (4.69) 18.81 (4.51) 8.53 * naMCI >SCCs Memo y complain s—p oxy 17.97 (4.54) 18.03 (4.76) 16.84 (4.46) 15.49 (4.23) 17.03 ** mda-MCI, naMCI > sdaMCI, SCCs; sda-MCI >SCCs CVLT Sho Delay F ee Recall 3.10 (2.03) 9.18 (2.21) 3.88 (2.05) 10.34 (2.70) 228.58 ** SCCs, na-MCI > mda-MCI, sda-MCI; sda-MCI >mda-MCI CVLT Long Delay F ee Recall 3.82 (3.19) 9.76 (2.69) 5.12 (3.06) 11.14 (2.83) 184.80 ** SCCs, sda-MCI, na-MCI >mdaMCI; SCCs>na-MCI, sda-MCI; na-MCI>sda-MCI CAMCOG-R memo y 15.50 (3.80) 18.56 (2.87) 18.53 (3.88) 21.26 (2.78) 99.18 ** SCCs, sda-MCI, na-MCI >mdaMCI; SCCs >na-MCI, sda-MCI CAMCOG-R o ien a ion 7.98 (1.46) 9.29 (0.80) 9.31 (0.83) 9.64 (0.63) 79.19 ** SCCs, sda-MCI, na-MCI >mdaMCI; SCCs >na-MCI, sda-MCI CAMCOG-R language 22.77 (2.36) 23.53 (2.00) 24.90 (2.43) 25.60 (2.40) 54.55 ** SCCs, sda-MCI > mdaMCI, na-MCI CAMCOG-R a en ion - calcula ion 5.03 (2.21) 4.29 (1.73) 7.25 (1.70) 7.05 (1.97) 100.90 ** SCCs, sda-MCI > mdaMCI, na-MCI CAMCOG-R p axis 9.10 (2.53) 10.03 (1.71) 10.71 (1.47) 11.14 (1.19) 48.01 ** SCCs, sda-MCI > mdaMCI, na-MCI CAMCOG-R pe cep ion 5.97 (1.64) 6.50 (1.28) 6.53 (1.48) 7.06 (1.44) 21.68 ** SCCs, sda-MCI > mdaMCI; SCCs > na-MCI, sda-MCI CAMCOG-R execu i e unc ion 13.22 (3.94) 15.35 (4.20) 15.93 (4.28) 18.42 (4.20) 58.84 ** SCCs, sda-MCI > mdaMCI; SCCs > na-MCI, sda-MCI ** p<0.01 * p<0.05. No e: CVLT =cali o nia e bal lea ning es ; CAMCOG-R =camb idge cogni i e assessmen - e ised; mda-MCI =mul i-domain amnes ic Mild Cogni i e Impai men ; na-MCI =non-amnes ic mild cogni i e impai men ; sda-MCI =single-domain amnes ic mild cogni i e impai men ; SCCs =subjec i e cogni i e complain s. In he inal sample ha comple ed a leas one ollow up e alua ion, 415 pa icipan s comple ed he i s ollow-up a ound 1 1/2 yea a e baseline (mean =17.92 mon hs, S.D. =3.87, ange 10–31 mon hs), and 330 pa icipan s comple ed he second ollow up assessmen a ound 3 yea s a e baseline (mean =37.13 mon hs, S.D. =5.43, ange 27–51 mon hs; his includes 20 pa icipan s who did no comple e he i s ollow-up assessmen , bu who did pa icipa e in he second ollow-up). Reasons o de ia ion om he e e ence imes include holidays, heal h issues and wo k issues. 2.2. P ocedu e The baseline e alua ions we e made be ween 2 Janua y 2008 and 11 No embe 2012. Pa icipan s unde ook wide- anging cogni i e and neu opsychological e alua ions, including he Spanish e sions o he Cali o nia Ve bal Lea ning Tes (CVLT) [ 17 ], he Mini-Men al S a e Examina ion (MMSE) [ 18 ] and he Camb idge Cogni i e Examina ion—Re ised (CAMCOG-R), which include subscales in se e al cogni i e domains (memo y, o ien a ion, language, a en ion and calcula ion, p axis, pe cep ion and execu i e unc ioning) [ 19 ]. Wi hin he ollow-up assessmen s, all pa icipan s again unde wen he assessmen and we e e-diagnosed by he same esea ch eam, again by consensus a speci ic mee ings. A all h ee e alua ion imes, he cu -o was 1.5 s anda d de ia ions (SDs) below age and educa ion no ms on he co esponding es s. All pa icipan s who displayed no mal cogni i e pe o mance (sco es B ain Sci. 2019,9, 242 4 o 13 highe han cu -o sco es on cogni i e s a us and cogni i e unc ioning, including memo y) we e included in he SCCs g oup, as hey all epo ed subjec i e memo y complain s. Fo all MCI pa icipan s, he gene al c i e ia ou lined by Albe e al. [ 5 ] we e applied: (a) P esence o complain s co obo a ed by an in o man ; (b) impai men in one o mo e cogni i e unc ions; (c) independence in daily li ing wi h minimum suppo o help; and (d) absence o demen ia acco ding o he NINCDS-ADRDA and DSM-IV s anda ds. Fo he speci ic diagnosis o MCI sub ypes, he c i e ia p oposed by Pe e sen and colleagues [ 3 , 6 ] we e applied, including (b1) a sco e o 1.5 SDs below age s anda ds o sho - e m and long- e m ecall measu es o he Spanish e sion o he CVLT o amnes ic MCI (aMCI). Two di e en ypes o aMCI we e diagnosed: Sda-MCI in hose pa icipan s (b2) who pe o med 1.5 SDs below age no ms in he Spanish e sion o he CVLT, bu pe o med no mally in he MMSE and he CAMCOG-R subscales; and mda-MCI in hose pa icipan s (b3) who pe o med 1.5 SDs below no ms in he MMSE and a leas wo sub-sco es o he CAMCOG-R ep esen ing a leas wo cogni i e unc ions. Finally, na-MCI was diagnosed in hose pa icipan s (b4) who pe o med wi hin he no mal ange in he Spanish e sion o he CVLT, bu 1.5 SDs below a e age in a leas one o he o he subscales o he CAMCOG-R. Single- and mul i-domain na-MCI we e no di e en ia ed, because o he small sample size in he single-domain na-MCI g oup. P obable AD o o he ypes o demen ia we e diagnosed acco ding o he NINCDS-ADRDA and DMS-IV c i e ia. P og ession o demen ia was con i med by consul a ion o he medical his o y, and he da e o neu ological diagnosis was eco ded. The s udy ecei ed app o al om he E hics in Clinical Resea ch Commi ee o he Galician Go e nmen and was conduc ed in acco dance wi h he p o isions o he Decla a ion o Helsinki as e ised in B azil 2013. W i en in o med consen was ob ained om all pa icipan s. 2.3. O e lapping In e als P ocedu e Acco ding o he a iabili y in he ime be ween he baseline and ollow-up e alua ions, and he po en ial e ec o ime in he s udy o MCI acco ding o i s a iabili y, a ime-o e lapping in e al app oach was aken [ 20 ] conside ing 13 mid-poin ime in e als wi h a ange o 6 mon hs ac oss each in e al. This app oach enabled a wide ange o ime dis ibu ions o be co e ed be ween assessmen s and also conside a ion o all pa icipan s e alua ed, i espec i e o he ime a which he ollow-up was ca ied ou and whe he hey pa icipa ed in he i s ollow up assessmen (abou one yea and a hal a e he baseline assessmen ) and/o he second ollow up (abou h ee yea s a e he baseline). Se en in e als we e included wi hin he ime-lapse es ablished o he i s ollow up assessmen (9–15, 12–18, 15–21, 18–24, 21–27, 24–30, 27–33), and se en in e als we e also included wi hin he ime-lapse es ablished o he second ollow up assessmen (27–33, 30–36, 33–39, 36–42, 39–45, 42–48, 45–51). Acco ding o his o e lapping-in e al design, and in o de o maximize sample sizes and empo al dis ibu ions ac oss successi e e alua ions, he sample size was di e en o each in e al (see Table 2). Acco ding o he e e ence imes o he i s (18 mon hs) and he second (36 mon hs) ollow-up assessmen s, he sample sizes ended o be la ge in he in e als close o hese imes and smalle in he in e ening in e als. This app oach allows o e lapping coun s be ween ime in e als, a oiding double-coun ing o pa icipan s by conside ing he pa icipan s only once in hose in e als in which mon hs be ween assessmen s a e included. The ange a ound each mid-poin in e al was six mon hs, so ha adjacen in e als di e in mean ime be ween assessmen s, bu o e lap in he ime ange. The e o e, he mid-poin o he 9–15 mon h g oup was 12, he mid-poin o he 12–18 mon h g oup was 15, and so on. The esul s can, hus, be consul ed wi h he maximum empo al con inuum (13 ime g oups) o wi h he minimum o e lap (se en and six ime g oups, 9–15, 15–21, 21–27, 27–33, 33–39, 39–45, 45–51 mon hs and 12–18, 18–24, 24–30, 30–36, 36–42, 42–48, espec i ely). O e lapping in e als a e used as independen sample es ima es. B ain Sci. 2019,9, 242 5 o 13 Table 2. Sample size in each o e lapped ime in e al. In e al mda-MCI na-MCI sda-MCI SCCs 9–15 mon hs 6 3 10 35 12–18 mon hs 14 15 30 125 15–21 mon hs 16 16 31 155 18–24 mon hs 13 11 22 105 21–27 mon hs 5 8 13 49 24–30 mon hs 4 4 7 25 27–33 mon hs 6 3 11 47 30–36 mon hs 9 7 21 98 33–39 mon hs 10 15 21 98 36–42 mon hs 8 12 14 56 39–45 mon hs 8 8 9 51 42–48 mon hs 6 3 9 27 45–51 mon hs 4 4 9 18 2.4. S a is ical Analysis The p obabili y o ansi ion om one s a e o ano he was es ima ed using Bayesian es ima es, calcula ed as ollows: P(Ai|B)=p(B|Ai)P(Ai)/Σk=1P(B|Ak)P(Ak), whe e P(Ai) ep esen s he a p io i p obabili y, P(B|Ai) ep esen s he p obabili y o ansi ion om Ai o B and, inally, P(Ai|B) ep esen s he a-pos e io i p obabili y. In he p esen s udy, e en Ai assumes he s a e o he diagnosis a a gi en ime ( ), and B is he s a e o diagnosis a ime ( +1). E iden ly, ansi ions om s a es Ai o Ak and hen o he nex s a e B d aw all combina ions exis ing be ween he diagnos ic ca ego ies a a gi en momen and he same ca ego ies a he nex ime. Models o hese ansi ions we e gene a ed using P(Ai) and P(Ai|B) om es ima es o he p opo ions obse ed in he o iginal dis ibu ions. In addi ion, as he demands o he Bayes’ heo em equi e exhaus i eness, he odds a io was ob ained om he con as s, as ollows: ORij =P(Ai|B)/P(Aj|B), wi h he aim o e alua ing he mos p obable ansi ions o he pai (i,j) as in he p e ious exp ession. Thus, apa om he es ima ed alues o he Bayesian p obabili ies, mo e applied alues o highe clinical alue we e ob ained. De ails o all o he calcula ions, including hose made o ob ain Odds Ra ios in each in e al, a e a ailable o eade s in Supplemen a y File S1. All he ma hema ical p ocedu es we e gene a ed by Excel ou ines and speci ic p og amming in Ma Lab. 3. Resul s The diagnos ic p obabili ies o he di e en MCI sub ypes a e p esen ed in Figu es 1and 2. Figu e 3shows hose pa icipan s whose diagnosis did no change in he ollow-up assessmen s and hose pa icipan s in he di e en g oups who p og essed o demen ia. The numbe o pa icipan s in he mda-MCI g oup who p og essed o demen ia was 4 a he i s ollow-up assessmen and 7 a he second ollow-up assessmen ; he co esponding numbe s in he na-MCI g oup we e 0 a he i s ollow-up and 1 a he second ollow-up assessmen ; he numbe s in he sda-MCI g oup we e 3 a he i s ollow-up and 5 a he second ollow-up; and inally, he co esponding numbe s in he SCCs g oup we e 0 a he i s ollow-up and 4 a he second ollow-up. Cases eco ded a he i s ollow-up we e also coun ed a he second ollow-up, so ha he numbe s a e cumula i e. B ain Sci. 2019,9, 242 6 o 13 SCCs Figu e 1. Diagnos ic p obabili ies a baseline (inne ci cle) and a i s ollow-up (ou e ci cle). Diagnos ic p obabili ies a e p esen ed nume ically, as p opo ions, wi h he o al sum o p opo ions in he inne and ou e ci cles equaling 1. The naMCI g oup is absen om he ou e ci cle, bu co esponds o he mdaMCI g oup in he inne ci cle as no ansi ions om mdaMCI o naMCI we e eco ded. SCCs = Subjec i e cogni i e complain s; MCI =mild cogni i e impai men ; mda =mul idomain amnes ic; na =non-amnes ic; sda =single domain amnes ic. B ain Sci. 2019,9, 242 7 o 13 SCCs Figu e 2. Diagnos ic p obabili ies a i s ollow-up (inne ci cle) and a second ollow-up (ou e ci cle). Diagnos ic p obabili ies a e p esen ed nume ically, as p opo ions, wi h he o al sum o p opo ions in he inne and ou e ci cles equaling 1. The naMCI g oup is absen om he ou e ci cle, bu co esponds o he mdaMCI g oup in he inne ci cle as no ansi ions om mdaMCI no naMCI we e eco ded. SCCs =Subjec i e cogni i e complain s; MCI =mild cogni i e impai men ; mda = mul i-domain amnes ic; na =non-amnes ic; sda =single domain amnes ic. B ain Sci. 2019,9, 242 8 o 13 Figu e 3. The numbe o pa icipan s whose diagnosis did no change a he ollow-up assessmen s, and he numbe o pa icipan s who con e ed o demen ia. SCCs =Subjec i e cogni i e complain s; MCI =mild cogni i e impai men ; mda =mul i-domain amnes ic; na =non-amnes ic; sda =single domain amnes ic. Acco ding o he o e lapping ime in e al s a egy, he diagnos ic p obabili ies o he di e en MCI sub ypes a di e en imes a e p esen ed in Figu e 4. Simila ends we e obse ed o he di e en ime in e als, as ollows: (a) SCCs ep esen ed he la ges p opo ion o pa icipan s, and he p opo ion ended o inc ease a ollow-up assessmen s; (b) sda-MCI pa icipan s comp ised he second la ges g oup a baseline and in he i s i e in e als o he ollow-up, bu no in he las eigh in e als o he ollow-up con inuum; (c) he p obabili ies o diagnosis o mda-MCI and na-MCI we e simila a bo h baseline and ollow-up, wi h sligh a ia ions ac oss he ime in e als; and (d) he e was a small, bu inc easing, p obabili y o con e sion o demen ia, ep esen ed in he igh column o he ollow-up sec ion in Figu e 3. In his igu e, only he cases o con e sion o demen ia eco ded du ing each ime in e al a e coun ed. Acco ding o he highe , bu empo ally a iable, p obabili y o sda-MCI a ollow-up and o es he heo e ical ele ance o mda-MCI as he closes poin o demen ia wi hin he MCI con inuum, odds a ios we e calcula ed o each ime in e al and o sda-MCI and mda-MCI ela i e o na-MCI. This enabled he p obabili y o con e sion o demen ia o be de e mined by compa ing he wo di e en amnes ic sub ypes and conside ing he non-amnes ic sub ype as he e e ence sub ype. The con e sion odds (Table 3) e ealed a highe p obabili y o con e sion o demen ia in sda-MCI han in na-MCI, which would be expec ed acco ding o he highe p opo ion o sda-MCI pa icipan s a baseline, and also a highe p obabili y o con e sion o demen ia in mda-MCI han in na-MCI. B ain Sci. 2019,9, 242 9 o 13 SCCs Figu e 4. Diagnos ic p obabili ies a baseline P(A) and a ollow-up P(B) . Mon hs be ween baseline and ollow-up assessmen s a e ep esen ed on he igh -hand side o he diag am. Each line ep esen s one ime in e al, wi h b acke s in he column on he igh showing he o e lapping na u e o he in e als (e.g., ime in e al 1 is indica ed by he b acke encompassing mon hs 9 o 15 and o e laps wi h in e al 2, indica ed by he b acke encompassing mon hs 12 o 18). Diagnos ic p obabili ies a e ep esen ed a baseline in ligh g ey and a ollow-up in da k g ey. SCCs =subjec i e cogni i e complain s; MCI =mild cogni i e impai men ; mda =mul idomain amnes ic; na =non-amnes ic; sda =single domain amnes ic.