Re iew a icle
The in luence o educa ion in p edic ing con e sion om Subjec i e
cogni i e decline (SCD) o objec i e cogni i e impai men : A sys ema ic
e iew and me a-analysis
Sonali A o a
a,b
, Sco B. Pa en
c
, Sabela C. Mallo
a,b
, C is ina Lojo-Seoane
a,b
, Alba Felpe e
a
,
Da id Facal-Mayo
a,b
, A u o X. Pe ei o
a,b,*
a
Depa men o De elopmen al Psychology, Uni e si y o San iago de Compos ela, San iago de Compos ela, Xos´
e Ma ía Su´
a ez Nú˜
nez, Campus Su , San iago de
Compos ela, Galicia 15782, Spain
b
Applied Cogni i e Neu oscience and Psychoge on ology g oup, Heal h Resea ch Ins i u e o San iago (IDIS), Uni e sidade de San iago de Compos ela, Rúa Xos´
e Ma ía
Su´
a ez Nú˜
nez, s/n, Campus Vida, San iago de Compos ela 15782, Spain
c
Depa men o Psychia y, Uni e si y o Calga y, 2500 Uni e si y D . NW, Calga y, AB T2N 1N4, Canada
ARTICLE INFO
Keywo ds:
Educa ion
Cogni i e ese e
Subjec i e cogni i e decline
P og ession
Risk
Cogni i e decline
Me a-analysis
ABSTRACT
Backg ound: Subjec i e cogni i e decline (SCD) is conside ed a p e-symp oma ic s age o demen ia cha ac e ized
by cogni i e complain s. The abili y o educa ion o educe he isk o demen ia is well known. Ou objec i e is o
in es iga e he in luence o educa ion on he isk o p og ession om SCD o MCI o demen ia.
Me hods: P ospec i e longi udinal s udies o adul s (≥50 yea s) wi h SCD e alua ing p og ession o objec i e
cogni i e decline, MCI, o demen ia we e selec ed. Pooled es ima es ( andom e ec s model) and 95 % con idence
in e als we e calcula ed, explo ing he e ogenei y. S anda dized educa ion di e ences, Odds Ra io, o Haza d
Ra io be ween con e e s and non-con e e s we e es ima ed.
Resul s: The sys ema ic e iew ca ied ou showed ha high educa ion, as well as o he cogni i e ese e p oxies,
delays cogni i e decline. The i s me a-analysis showed a signi ican associa ion o SCD wi h con e sion in bo h
high and low educa ion s a a. A second me a-analysis conside ing educa ion as a con inuous a iable ound ha
SCD con e e s showed wo yea s less educa ion han non-con e e s.
Conclusions: Ou esul s sugges ha educa ion has a delaying e ec agains cogni i e decline p og ession. The
p esumed imp o emen in accu a ely de ec ing cogni i e decline associa ed wi h be e me acogni i e skills in
highe -educa ed SCD pa icipan s does no seem o neu alize he inc emen al isk o objec i e cogni i e decline
associa ed wi h lowe educa ional a ainmen .
1. In oduc ion
Subjec i e Cogni i e Decline (SCD) is conside ed a p e-symp oma ic
s age o Alzheime ’s Disease (AD) cha ac e ized by pe sis en Subjec i e
Cogni i e Complain s (SCCs) ha canno be explained by medical con-
di ions o subs ance use and wi hou e idence o objec i e cogni i e
impai men (Jessen e al., 2014, 2020). SCD has been p oposed as he
ansi ional s age be ween Cogni i ely Unimpai ed (CU) and Mild
cogni i e impai men (MCI) (Jack e al., 2018). The p esence o SCCs is a
ele an c i e ion o MCI diagnosis (Albe e al., 2011; Pe e sen e al.,
1999) and a isk ac o o cogni i e decline and demen ia (Mendonca
e al., 2016; Pe ei o e al., 2021). G ea e SCCs we e ound o be linked
o a heigh ened isk o objec i e cogni i e decline (Du ouil e al., 2005;
Numbe s e al., 2021). Acco dingly, me a-analy ical e idence suppo s
ha SCD pa icipan s show an inc eased isk o p og ession o MCI and
demen ia (Mi chell e al., 2014; Pike e al., 2021).
Cu en esea ch on demen ia isk ac o s is also in e es ed in he
a iables ha educe he chances o cogni i e symp oms associa ed wi h
his disease (Sil a e al., 2019). Cogni i e Rese e (CR) e e s o he
a iables capable o delaying he onse o symp oms e en in he p es-
ence o neu odegene a ion (S e n e al., 2020). Educa ion is he p oxy
ha has p o en o be he mos ele an wi hin his cons uc (Roe e al.,
* Co espondence o: Facul y o Psychology, Uni e si y o San iago de Compos ela, Xos´
e Ma ía Su´
a ez Nu˜
nez S/N, Campus Su , San iago de Compos ela 15782,
Spain.
E-mail add esses: [email p o ec ed] (S. A o a), [email p o ec ed] (S.B. Pa en), [email p o ec ed] (S.C. Mallo), [email p o ec ed] (C. Lojo-
Seoane), [email p o ec ed] (A. Felpe e), [email p o ec ed] (D. Facal-Mayo), [email p o ec ed] (A.X. Pe ei o).
Con en s lis s a ailable a ScienceDi ec
Ageing Resea ch Re iews
jou nal homepage: www.else ie .com/loca e/a
h ps://doi.o g/10.1016/j.a .2024.102487
Recei ed 3 No embe 2023; Accep ed 31 Augus 2024
Ageing Resea ch Re iews 101 (2024) 102487
A ailable online 5 Sep embe 2024
1568-1637/© 2024 The Au ho (s). Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license ( h p://c ea i ecommons.o g/licenses/by-
nc-nd/4.0/ ).
2007; Valenzuela, Sachde , 2006).
The e is conside able e idence abou he associa ion be ween highe
educa ional a ainmen and lowe isk o MCI (Liu e al., 2022; Roe e al.,
2007; Te o e al., 2004) and demen ia (Mungas e al., 2018; Robi aille
e al., 2018; S e n e al., 2020; Xu e al., 2016). A me a-analy ic s udy
in ol ing 69 s udies showed ha he incidence and p e alence o AD
and demen ia signi ican ly dec eased in olde adul s wi h high educa-
ional le els compa ed o hose wi h a low le el o educa ion (Meng,
D’A cy, 2012). This is highly consis en wi h he CR hypo hesis, which
p oposes ha indi iduals wi h mo e cogni i e abili ies and
b ain-s imula ing expe iences can be e cope wi h neu ological damage
and aging- ela ed changes (S e n, 2009).
No wi hs anding, ecen s udies ha e shown a s age-dependen e ec
o CR on cogni i e unc ioning. Acco ding o Soldan e al., (2017),
highe CR is associa ed wi h imp o ed cogni i e pe o mance and
delayed onse o symp oms in indi iduals wi h MCI. Ano he s udy
(Zahodne e al., 2019) e ealed ha he posi i e impac o educa ion on
cogni i e p o ec ion migh be s onge in he ini ial s ages o age- ela ed
b ain issues such as CU o SCD. Howe e , in la e s ages, he nega i e
cogni i e e ec s o b ain damage in people wi h high ese e could be
inc eased due o he c i ical o e low o he compensa o y capaci y o
CR. Acco dingly, o he s udies (Amie a e al., 2014; Mungas e al., 2018;
Ye e al., 2013; Yu e al., 2012) ha e shown ha wi h inc easing neu-
odegene a ion, he compensa o y e ec s o highe educa ion a e low-
e ed as mo e de eloped unc ional ne wo ks ha suppo esilience
could become mo e sensi i e o b ain damage, po en ially leading o
as e cogni i e decline.
Resea ch on he ole o educa ion in SCD p og ession o MCI and
demen ia has been sca ce (Mon emu o e al., 2021). Mos o he
a ailable e idence shows ha highe educa ion in SCD pa icipan s can
lowe he isk o p og ession o objec i e cogni i e decline (C umley
e al., 2014; Hao e al., 2017, 2019; Yang e al., 2020). Complemen a y
o his, some s udies ha e shown ha pa icipan s wi h a lowe le el o
educa ion a e mo e likely o epo SCCs (S ewa e al., 2008; Bolla
e al.; 1991). Howe e , some inconsis en esul s we e ound and o he
s udies ha e in o med ha highe educa ion in SCD pa icipan s in-
c eases he isk o AD (Aghjayan e al., 2016; Jonke e al., 2000; Rabin
e al., 2017), by a be e es ima ion o cogni i e changes associa ed o
inc eased me acogni i e skills in highe educa ional le els (He zog
e al., 2008).
Ou objec i es we e o conduc : a) a sys ema ic e iew o he lon-
gi udinal associa ion be ween cogni i e wo sening, p og ession o MCI
o demen ia, and educa ion (alone o wi h o he CR p oxies); and b) a
me a-analysis on he in luence o educa ion on he es ima ion o he isk
o p og ession o SCD o MCI o demen ia.
2. Me hods
2.1. PICO e alua ion
The e iew ques ion was o med using he PICO s a emen
(Cumps on e al., 2021):
−P (Popula ion): cogni i ely unimpai ed pa icipan s aged 50 yea s o
mo e wi h SCD.
−I (In e en ion): no in e en ion/exposu e, obse a ional (coho
s udy).
−C (Compa ison): le el o educa ion o he pa icipan s who p og ess
o MCI and/o demen ia e sus hose who do no .
−O (Ou come): isk o p og ession o MCI and/o demen ia.
Re iew ques ion: How much educa ion a ec s he alidi y o SCD in
p edic ing he p og ession o MCI o con e sion o demen ia in olde
adul s?
2.2. Sea ch s a egy
Ou sys ema ic e iew was conduc ed acco ding o he P e e ed
Repo ing I ems o Sys ema ic Re iew and Me a-analyses s a emen
(PRISMA 2020) (Page e al., 2021) (eTable1 in he supplemen ) and
MOOSE (S oup e al., 2000) guidelines (eTable2 in he supplemen ). A
p o ocol was also egis e ed wi h he In e na ional P ospec i e Regis e
o Sys ema ic Re iews (PROSPERO) on 24 Oc obe 2022, wi h he
egis a ion numbe CRD42022362750 (Na ional Ins i u e o Heal h
Resea ch, n.d.).
Fou da abases (Web o Science - WOS, PubMed, SCOPUS, and Psy-
cINFO) we e sea ched o published a icles be ween Sep embe 1993
( he ea lies s udy ound) and June 14, 2022. The keywo ds used we e:
("subjec i e memo y complain s" OR "Subjec i e Cogni i e decline" OR
"Memo y complain s" OR "Cogni i e complain s") AND (educa ion OR
"cogni i e ese e" OR "educa ional le el" OR "yea s o schooling") AND
(aging OR elde ly OR "olde adul s" OR ageing) AND (p og ession OR
con e * OR e olu ion OR Risk) (eAppendix in he supplemen ). Ti les
we e sc eened i s , ollowed by abs ac s and ull- ex a icles. Re e -
ences we e compiled using Mendeley Desk op 1.19.8, wi h duplica es
emo ed using he same so wa e.
2.3. S udy selec ion
Two e iewe s (S.A. and C.L.-S.) independen ly sc eened i les, ab-
s ac s, and ull- ex a icles. Rayyan QCRI (Ouzzani e al., 2016) was
used o sc eening i les and abs ac s. An o e all consensus o 84 % was
eached be ween he e iewe s. Any disag eemen s ega ding eligibili y
we e esol ed by consensus mee ings, o i necessa y, wi h a hi d
e iewe (S.C.M.).
Inclusion c i e ia we e: (1) Analyse he isk o objec i e cogni i e
decline, p og ession o MCI o con e sion o demen ia o any impac on
cogni i e pe o mance due o le el o educa ion in SCD pa icipan s; (2)
Pa icipan s aged 50 yea s o mo e; (3) SCD diagnosed by any p ocedu e
(ques ion/s, ques ionnai es/spon aneous sel - epo s on he p esence o
memo y complain s); (4) Educa ion in o med by ange o numbe o
yea s; (5) Empi ical longi udinal o ollow-up s udies.
Exclusion c i e ia we e: (1): S udies ocused on demen ed o MCI
pa ien s; (2) Pa icipan s om a special popula ion we e included (e.g.,
pa icipan s wi h a his o y o s oke, people wi h cance , Fab y disease,
cogni i e ail y); (3) The c i e ia o e alua ion o de e mine SCD, MCI
o demen ia we e no speci ied; (4) Educa ion was no conside ed as CR
measu e; (5) S udies lacking sepa a e educa ion da a o assess he isk o
onse o objec i e cogni i e decline, MCI p og ession, o demen ia
con e sion in SCD pa icipan s.
Fu he mo e, o he me a-analysis, hose s udies ha do no sepa-
a ely p o ide quan i a i e da a on he educa ional le el o SCD pa -
icipan s who emain s able and hose who p og ess owa ds mo e o less
se e e o ms o objec i e cogni i e impai men we e excluded.
2.4. Da a abs ac ion
A s anda dized Excel sp eadshee was designed including he
ollowing a iables: s udy in o ma ion (i.e., i s au ho , yea ), sub-
s an i e cha ac e is ics (i.e., coun y, mean age, pe cen age o women,
yea s o educa ion, dep essi e symp oms, pe sonali y ai s), me hodo-
logical cha ac e is ics (i.e., mon hs o ollow up, sample ype, c i e ia
used o SCD/MCI/demen ia and esul s. This sec ion includes he mean
and S anda d De ia ion (SD) o yea s o educa ion o con e e s and
non-con e e s SCD pa icipan s. Besides, pa ame e s such as Haza d
Ra io (HR) and Odds Ra io (OR) o con e e s s. non-con e e s
s a i ied by educa ional le el we e also included. One e iewe (S.A.)
ex ac ed he da a om all included s udies. The abs ac ed da a we e
discussed wi h a second e iewe (S.C.M.) and any disag eemen s we e
esol ed by consensus o by a hi d esea che (A.X.P.).
In he s udy, con e e s e e o pa icipan s wi h SCD who p og ess
S. A o a e al.
Ageing Resea ch Re iews 101 (2024) 102487
2
o any cogni i e decline, MCI, o demen ia du ing he ollow-up pe iod.
Con e sely, non-con e e s e e o pa icipan s wi h SCD who do no
p og ess o objec i e cogni i e decline, MCI, o demen ia du ing he
ollow-up pe iod.
2.5. Quali y scale used
S udy quali y was assessed using he quali y assessmen ool o
obse a ional coho and c oss-sec ional s udies om he Na ional
Hea , Lung and Blood Ins i u e (eTable3 in supplemen ). This ool
consis s o 14 ques ions on which quali y is de e mined. Each i em was
esponded as “yes”, “no”, o “canno de e mine” and an o e all a ing
o each s udy was p o ided as “good”, “ ai ” o “poo ”. Fo a icles
classi ied as “good”, 11 o mo e “yes” answe s (11/14 ha is, 78.5 %
posi i e answe s) we e equi ed. Fu he mo e, o a “ ai ” a ing, 10 o
mo e “yes” answe s (10/14 ha is, 71 % posi i e answe s) we e
equi ed. Finally, o a “poo ” a ing, 9 o ewe “yes” answe s o he
scale i ems we e conside ed. Two e iewe s (S.A.) and (S.C.M.) inde-
penden ly assessed he quali y o all included pape s. Disag eemen s
we e econciled wi h he o iginal e iewe s and i equi ed, wi h an
independen pa y (A.X.P.).
2.6. Me a-analysis
The me a-analysis was conduc ed using he “me a” command in
STATA 17 (College S a ion, TX) (Ha is e al., 2008). A isual summa y
was gene a ed using o es plo s. The analysis used a andom e ec s
model implemen ed using he De Simonian-Lai d me hod. He e oge-
nei y was quan i ied using he au-squa ed (T
2
) and I-squa ed (I
2
) pa-
ame e s (Higgins e al., 2003), and es ed agains he null hypo hesis o
no he e ogenei y using he Q- es .
The p ima y analysis es ima ed he ela i e isk o SCD o con-
e sion wi h s a i ica ion by educa ion le el as a bina y a iable (low
e sus high educa ion). The agg ega e ou come is labeled as “con e -
sion” in he analysis and includes he onse o documen ed objec i e
cogni i e impai men , MCI, o demen ia in SCD pa icipan s. This me a-
analysis was based on he pooling o epo ed a io-based measu es (one
s udy epo ed haza d a ios and h ee o he s epo ed odds a ios),
again using he De Simonian-Lai d me hod. The es ima es we e hen
back- ans o med a e i ing he me a-analy ic model.
A seconda y analysis compa ed he mean educa ion le el in samples
wi h SCD who subsequen ly demons a ed objec i e cogni i e decline,
MCI, o demen ia e sus hose who did no . These me a-analy ic models
es ima ed he mean di e ence in educa ion be ween con e e s and non-
con e e s g oups.
3. Resul s
3.1. Iden i ica ion o s udies
A o al o 657 a icles we e iden i ied om he ini ial da abase sea ch
and 1 om he manual sea ch. A lowcha o he p ocess acco ding o
PRISMA is shown in Fig. 1.
Ou o hose 657 a icles iden i ied om da abases, 247 we e
excluded as duplica es, and 365 we e emo ed a e sc eening o i le
and abs ac based on he inclusion and exclusion c i e ia: a) Being a
e iew, sys ema ic e iew, me a-analysis, o case s udies (1.13 %; N=
4); b) S udies ocused on demen ed o MCI pa ien s (30.95 %; N=113);
c) Pa icipan s belonging o a special popula ion a e included (10.73 %;
N=39); d) S udies no speci ying c i e ia o e alua ion me hods o
de e mining SCD, MCI, and/o demen ia (30.95 %; N=113); e) Edu-
ca ion is no conside ed as CR measu e (14.24 %; N=52); and ) S udies
lacking sepa a e educa ion da a o assess he isk o onse o objec i e
cogni i e decline, MCI p og ession, o demen ia con e sion in SCD
pa icipan s (12 %; N=44).
Fo y- i e s udies we e selec ed o a subsequen ull- ex e iew. O
hose, 34 we e excluded o (a) Absence o empi ical longi udinal o
ollow-up s udies (38.23 %; N=13); (b) S udies ocused on demen ed o
MCI pa ien s (14.70 %; N=5); (c) Pa icipan s belonging o a special
popula ion a e included (8.82 %; N=3); (d) Educa ion no in o med by
ange o numbe o yea s (32.34 %; N=11); (e) Being a sys ema ic e-
iew o me a-analysis (2.94 %; N=1); and ( ) No speci ying he c i e ia
o e alua ion o de e mine SCD, MCI o demen ia (2.94 %; N=1).
Finally, 12 s udies (11 iden i ied om da abases and 1 om manual
sea ch) we e included in he sys ema ic e iew (Bessi e al., 2018; Cha y
e al., 2013; Jia e al., 2021; Lojo-Seoane e al., 2018; 2020; Mazzeo
e al., 2019; Mondini e al., 2022; Qi e al., 2018; Sil a e al., 2014; Van
den Komme e al., 2008; Van Oijen e al., 2007; Yun e al., 2020).
Howe e , 10 a icles we e selec ed o he me a-analysis a e he
exclusion o he manusc ip s ha do no sepa a ely p o ide quan i a i e
da a on he educa ional le el o SCD pa icipan s who emain s able and
hose who p og ess owa ds mo e o less se e e o ms o objec i e
cogni i e impai men (Van den Komme e al., 2008; Yun e al., 2020).
3.2. Cha ac e is ics o he s udies
S udies included in he sys ema ic e iew we e published be ween
2007 and 2022. Th ee o he 12 s udies we e om I aly, wo om Spain,
wo om China, wo om he Ne he lands, and one om Po ugal,
F ance, and Ko ea. Fu he , he ollow-up pe iods anged om 18 o 512
mon hs and he mean age o he s udy sample anged om 61.3 o 78.2.
All o hem included mo e women han men (pe cen age o emales
anging om 51.20 % o 70.77 %). Se en s udies we e clinic-based and
i e we e communi y-based. The main cha ac e is ics o he s udies
selec ed a e summa ized in Table 1.
All s udies included neu opsychological e alua ion and he Mini-
Men al S a e Examina ion (MMSE; Fols ein e al., 1975) was he mos
equen ins umen o e alua e global cogni ion.
Rega ding SCD iden i ica ion, se en s udies used single ques ions
(Cha y e al., 2013; Jia e al., 2021; Mondini e al., 2022; Qi e al., 2018;
Van den Komme e al., 2008; Van Oijen e al., 2007; Yun e al., 2020);
h ee s udies used a speci ic ques ionnai e (Bessi e al., 2018; Mazzeo
e al., 2019; Sil a e al., 2014); and wo used spon aneous SCC
sel - epo s (Lojo-Seoane e al., 2018, 2020). Only wo o he a o e-
men ioned s udies (Bessi e al., 2018; Mazzeo e al., 2019) e i ied
compliance wi h he SCD-I Wo king G oup c i e ia (Jessen e al., 2014).
Rega ding demen ia diagnosis, he ollowing c i e ia we e used in
he selec ed s udies: a) he Diagnos ic and S a is ical Manual o Men al
Diso de s-Thi d Edi ion-Re ised (APA, 1987) (Van Oijen e al., 2007; Jia
e al., 2021; Cha y e al., 2013); b) he Diagnos ic and S a is ical Manual
o Men al Diso de s-Thi d Edi ion-Fou h Edi ion Diagnos ic and S a-
is ical Manual o Men al Diso de s (APA, 2000) (Sil a e al., 2014); c)
Alzheime ’s Disease and Rela ed Diso de s Associa ion (NINCDS-A-
DRADA; McKhann e al., 1984) and/o he Na ional Ins i u e o
Neu ological and Communica i e Diso de s and S oke and/o he Na-
ional Ins i u e o Neu ological and Communica i e Diso de s and
S oke and he Alzheime ’s Disease and Rela ed Diso de s Associa ion
(NINCDS-ADRADA; McKhann e al., 1984) (Lojo-Seoane e al., 2018,
2020); d) he Na ional Ins i u e o Neu ological Diso de s and S oke
and he Associa ion In e na ionale pou la Reche che e l
′
Enseignemen
en Neu osciences (NINDS-AIREN; E kinjun i, 1994) (Van Oijen e al.,
2007); e) he NIA-AA c i e ia (Jack e al., 2018) (Bessi e al., 2018;
Mazzeo e al., 2019; Qi e al., 2018); and e) he Diagnos ic and S a is ical
Manual o Men al Diso de s-Fi h Edi ion (DSM-5; Ame ican Psychia ic
Associa ion, 2013) (Yun e al., 2020).
Eigh s udies included a sample o SCD pa icipan s who p og essed
o MCI and/o demen ia (Bessi e al., 2018; Cha y e al., 2013; Jia e al.,
2021; Mazzeo e al., 2019; Qi e al., 2018; Sil a e al., 2014; Van Oijen
e al., 2007; Yun e al., 2020), while he emaining ou s udies es ed
whe he SCD pa icipan s expe ienced objec i e cogni i e decline a
ollow-up (Lojo-Seoane e al., 2018, 2020; Mondini e al., 2022; Van den
Komme e al., 2008).
S. A o a e al.
Ageing Resea ch Re iews 101 (2024) 102487
3
Fig. 1. Flowcha o he p ocess o a icle selec ion and inclusion.
S. A o a e al.
Ageing Resea ch Re iews 101 (2024) 102487
4
Table 1
. Summa y o cha ac e is ics o included s udies.
S udy Coun y
(Con inen )
N Mean
age
% o
emales
Le el o
Educa ion o
CR measu e
Sample Follow-
up
(mon hs)
SCC
measu emen
Diagnos ic
C i e ia
(MCI/
Demen ia)
Ou come Resul s
Bessi e al.
(2018)
I aly
(Eu ope)
109 64.6 67.88 % Highe
educa ion
(+13 yea s)
CR: TIB es
CR p oxy:
In ellec ual
ac i i ies
Clinical 84 Memo y
Assessmen
Clinics-
Ques ionnai e (20
i ems), C i e ia:
SCD-I Wo king
G oup
MCI and
Demen ia:
NIA-AA
c i e ia
Bo h MCI
and
Demen ia
SCD
pa icipan s
who
con e ed o
MCI/
demen ia had
lowe
a e age
educa ion
le els
compa ed o
hose who
did no
con e .
Cha y e al.
(2013)
F ance
(Eu ope)
2882 74.7 58.70 % Two g oups:
LEL
(<elemen a y
school
diploma) and
HEL
(Seconda y
school o
Uni e si y)
Communi y 240 "Do you
equen ly ha e 1.
o ge ulness in
ac i i ies o daily
li ing; 2.
di icul ies in
e aining o
emembe ing new
simple
in o ma ion; 3.
di icul ies in
e ie ing o
emembe ing old
memo ies and 4.
di icul ies in
inding wo ds"
Demen ia:
A, B, and C,
DSM-III-R,
AD,
ascula
demen ia,
o o he
ypes
Demen ia SCD
pa icipan s
wi h high
educa ion
le els
showed a
sligh ly
inc eased
likelihood o
con e ing o
demen ia
compa ed o
hose wi h
low
educa ion
le els.
Jia e al.
(2021)
China (Asia) 940 72 51.20 % Th ee g oups:
Lowe (≤10),
In e media e
(>10 and
<12) and
Highe
educa ion
(≥12)
CR p oxies:
Ea ly li e
educa ion,
mid-li e
occupa ion,
and la e-li e
cogni i e
ac i i ies
Communi y 24 “Ha e you ended
o o ge hings
ecen ly?” and
“Do you ha e o
make mo e e o
o emembe
hings han you
used o?
Demen ia:
DSM-III-R
Demen ia SCD
pa icipan s
wi h low
educa ion
le els had a
subs an ially
highe isk o
con e ing o
demen ia
compa ed o
hose wi h
high
educa ion
le els.
Lojo-Seoane
e al.
(2018)
Spain
(Eu ope)
189 66.68 67.67 % Range om
basic
schooling o
uni e si y
s udies (mean:
9.63 ±4.45)
CR: CRIq
ques ionnai e
Clinical 18.67 Pa icipan s
spon aneously
epo ed ha hei
memo y was no
as good as be o e
Demen ia:
DSM IV
and/o
NINCDS-
ADRADA
Cogni i e
decline
SCD
pa icipan s
who
con e ed o
any cogni i e
decline had
lowe
a e age
educa ion
le els
compa ed o
hose who
did no
con e .
Lojo-Seoane
e al.
(2020)
Spain
(Eu ope)
164 65.86 68.40 % Range om
basic
schooling o
uni e si y
s udies (mean:
9.64 ±4.58)
CR: CRIq
ques ionnai e
Clinical 36 Pa icipan s
spon aneously
epo ed ha hei
memo y was no
as good as be o e
Demen ia:
DSM IV
and/o
NINCDS-
ADRADA
Cogni i e
decline
SCD
pa icipan s
who
con e ed o
any cogni i e
decline had
lowe
a e age
educa ion
le els
compa ed o
(con inued on nex page)
S. A o a e al.
Ageing Resea ch Re iews 101 (2024) 102487
5
Table 1 (con inued)
S udy Coun y
(Con inen )
N Mean
age
% o
emales
Le el o
Educa ion o
CR measu e
Sample Follow-
up
(mon hs)
SCC
measu emen
Diagnos ic
C i e ia
(MCI/
Demen ia)
Ou come Resul s
hose who
did no
con e .
Mazzeo e al.
(2019)
I aly
(Eu ope)
154 61.3 70.77 % Highe
educa ion
(+13 yea s)
CR: TIB es
CR p oxy:
P emo bid
in elligence
Communi y 84 Memo y
Assessmen
Clinics-
Ques ionnai e (20
i ems), C i e ia:
SCD-I Wo king
G oup
MCI and
Demen ia:
NIA-AA
c i e ia
Bo h MCI
and
Demen ia
SCD
pa icipan s
who
con e ed o
MCI/
demen ia had
lowe
a e age
educa ion
le els
compa ed o
hose who
did no
con e .
Mondini
e al.
(2022)
I aly
(Eu ope)
507 78.2 65.23 % Mean yea s o
educa ion:
6.99 ±3.85
CR p oxies:
Educa ion and
occupa ion
Clinical 36+Iden i ied based
on psychome ic
da a, anamnes ic
and clinical
in o ma ion,
Esame
Neu opsicologico
B e e 2 (ENB−2)
ba e y
assessmen
NR Cogni i e
decline
SCD
pa icipan s
who
con e ed o
any cogni i e
decline had
highe
a e age
educa ion
le els
compa ed o
hose who
did no
con e .
Qi e al.
(2018)
China (Asia) 1713 71.1 69.60 % Th ee g oups:
Illi e a e,
P ima y
school, and
Junio high
school o
abo e
Clinical 512 Pa icipan s "Do
you hink ha you
ha e any
p oblems wi h
you memo y?”,
In o man s “Do
you belie e he
subjec has any
p oblems wi h
memo y?”
MCI and
Demen ia:
NIA-AA
Bo h MCI
and
Demen ia
SCD
pa icipan s
wi h high
educa ion
had a highe
likelihood o
p og essing
o MCI and
con e ing o
demen ia.
Low-
educa ion
pa icipan s
had a simila
likelihood o
bo h MCI and
demen ia.
Sil a e al.
(2014)
Po ugal
(Eu ope)
133 68.2 58.64 % Fo mally
educa ed
pa icipan s
Clinical 24 SMC (25 i ems) Demen ia:
DSM-IV-TR
Demen ia SCD
pa icipan s
who
con e ed o
demen ia had
lowe
educa ion
le els
compa ed o
non-
con e e s.
Van den
Komme
e al.
(2008)
Ne he lands
(Eu ope)
169 73 NR Two g oups:
Low educa ion
(≤Elemen a y
educa ion)
and High
educa ion
(>elemen a y
school)
Communi y 36 Ques ion asking i
he pa ien s ha e
memo y
complain s o no .
NR Cogni i e
decline
SCD
pa icipan s
wi h highe
educa ion
had a
signi ican ly
inc eased
isk o
p og essing
o any
cogni i e
decline,
while hose
wi h lowe
(con inued on nex page)
S. A o a e al.
Ageing Resea ch Re iews 101 (2024) 102487
6
Six s udies use, in addi ion o educa ion, o he p oxy measu es o CR.
Among hese, one s udy conside ed in elligence (Mazzeo e al., 2019) o
in ellec ual ac i i ies as addi ional measu es o CR (Bessi e al.,2018).
Jia e al. (2021) combined ea ly-li e educa ion, mid-li e occupa ional
complexi y, and la e-li e cogni i e ac i i ies as measu es o CR, simila
o Mondini e al. (2022) and Lojo-Seoane e al. (2018, 2020) ha used
educa ion and occupa ion as main p oxies o CR.
In he me a-analysis, educa ion was ope a ionalized as a ca ego ical
a iable in ou s udies (Cha y e al., 2013; Jia e al., 2021; Qi e al.,
2018; Van Oijen e al., 2007) and as a con inuous a iable (i.e., yea s o
schooling) in six s udies (Bessi e al., 2018; Lojo-Seoane e al., 2018,
2020; Mazzeo e al., 2019; Mondini e al., 2022; Sil a e al., 2014).
Among he s udies ha conside ed educa ion as a ca ego ical a iable:
(a) one conside ed a bina y classi ica ion app oach (i.e., low educa ion:
below he elemen a y school and high educa ion: seconda y school o
uni e si y) (Cha y e al., 2013); (b) wo adop ed a h ee- ie ed classi i-
ca ion sys em (i.e., lowe educa ion: ≤10 yea s, in e media e educa ion:
>10 and <12 yea s, and highe educa ion: ≥12 yea s in Jia e al., 2021)
and illi e a e, p ima y school, and junio high school o abo e in Qi
e al., 2018); and (c) one employed an e en highe numbe o ca ego ies
(i.e., low, low-in e media e, high-in e media e, and high educa ion
le els) (Van Oijen e al., 2007).
3.3. Quali y assessmen
Rega ding he quali y o he selec ed s udies o he sys ema ic e-
iew, eigh s udies (66.67 %) we e a ed as good quali y (Bessi e al.,
2018; Cha y e al., 2013; Lojo-Seoane e al., 2018, 2020; Mazzeo e al.,
2019; Sil a e al., 2014; Van den Komme e al., 2008; Yun e al., 2020),
while wo s udies (16.67 %) we e quali ied as ai (Qi e al., 2018; Van
Oijen e al., 2007) and wo (16.67 %) as poo quali y (Jia e al., 2021;
Mondini e al., 2022) (see eTable3 in he supplemen ).
All he s udies me he quali y c i e ia e alua ed h ough he
ollowing ques ions: (a) “Was he esea ch ques ion o objec i e in his
pape clea ly s a ed?”; (b) “Was he s udy popula ion speci ied and
de ined?”; (c) “Was he pa icipa ion a e o eligible pe sons a leas
50 %?”; (d) “We e all he subjec s selec ed o ec ui ed om he same o
simila popula ions (including he same pe iod?”; (e) We e inclusion and
exclusion c i e ia o being in he s udy p especi ied and applied uni-
o mly o all pa icipan s?”; ( ) “Fo he analyses in his pape , we e he
exposu e(s) o in e es measu ed be o e he ou come(s) being
measu ed?”; and (g) “Was he exposu e(s) assessed mo e han once o e
ime?”.
Only one s udy (8.33 %) p o ided a sample size jus i ica ion, powe
desc ip ion, o a iance and e ec es ima es (Cha y e al., 2013), and
none epo ed blinding s a egies o conceal pa icipan s’ s a us om
ou come assesso s.
3.4. Sys ema ic e iew
Almos all selec ed s udies concluded ha educa ion alone o
oge he wi h o he CR p oxies dec eases he isk o p og ession owa d
objec i e cogni i e impai men (Bessi e al., 2018; Jia e al., 2021;
Lojo-Seoane e al., 2018, 2020; Mazzeo e al., 2019; Mondini e al.,
2022; Qi e al., 2018; Sil a e al., 2014; Van den Komme e al., 2008;
Yun e al., 2020). Some o hem epo ed ha SCD pa icipan s wi h
Table 1 (con inued)
S udy Coun y
(Con inen )
N Mean
age
% o
emales
Le el o
Educa ion o
CR measu e
Sample Follow-
up
(mon hs)
SCC
measu emen
Diagnos ic
C i e ia
(MCI/
Demen ia)
Ou come Resul s
educa ion
had a lowe
isk o no
con e ing o
any cogni i e
decline.
an Oijen
e al.
(2007)
Ne he lands
(Eu ope)
1309 69.5 60 % Fou g oups:
"Low, Low-
in e media e,
High-
in e media e,
and High
educa ion"
Communi y 108 Single ques ion
"Do you ha e
memo y
complain s?"
Demen ia:
DSM-III-R,
AD:
NINCDS-
ADRADA,
ascula
NINDS-
AIREN
Demen ia SCD
pa icipan s
wi h high
educa ion
le els had a
highe isk o
con e ing o
demen ia
compa ed o
hose wi h
low
educa ion
le els.
Yun e al.
(2020)
Ko ea (Asia) 107 66.6 37.38 % Mean yea s o
educa ion:
8.56 ±4.75
Clinical 60 CDR Assessmen :
“Do you ha e
p oblems wi h
you memo y o
hinking?”
Demen ia:
DSM 5
Demen ia Educa ion
le el was no
signi ican ly
associa ed
wi h
demen ia
con e sion in
SCD
pa icipan s.
No e: N: Numbe o pa icipan s wi h Subjec i e Cogni i e Decline; Basic schooling: 0–4 yea s o educa ion; Uni e si y s udies: +13 yea s o educa ion; Low educa ion:
p ima y educa ion; Low-in e media e educa ion: p ima y educa ion plus no comple ed highe educa ion wi h pe sons wi h lowe oca ional educa ion; High-
in e media e educa ion: in e media e oca ional educa ion wi h pe sons wi h gene al seconda y educa ion; High educa ion: highe oca ional educa ion o uni e -
si y aining; MCI: Mild Cogni i e Impai men ; NR: No epo ed; SMC: Subjec i e Memo y Complain s Scale; NIA-AA: Na ional Ins i u e on Aging-Alzheime ’s As-
socia ion; DSM -III-R: Diagnos ic and s a is ical manual o men al diso de s-Thi d edi ion-Re ised; AD: Alzheime ’s Disease; DSM IV: Diagnos ic and S a is ical Manual
o Men al Diso de s, 4 h edi ion; NINCDS/ADRDA: Na ional Ins i u e o Neu ological and Communica i e Diso de s and S oke and he Alzheime ´s Disease and Rela ed
Diso de s Associa ion; DSM-IV-TR: Diagnos ic and S a is ical Manual o Men al Diso de s, 4 h Edi ion, Tex Re ision; NINDS-AIREN: Na ional Ins i u e o Neu ological
Diso de s and S oke and he Associa ion In e na ionale pou la Reche che e l
′
Enseignemen en Neu osciences; DSM 5: Diagnos ic and S a is ical Manual o Men al
Diso de s, 5 h edi ion; TIB: Tes di In elligenza B e e; CRIq: Cogni i e Rese e Index ques ionnai e
S. A o a e al.
Ageing Resea ch Re iews 101 (2024) 102487
7
highe CR delayed he onse o MCI (Mazzeo e al., 2019) o demen ia
(Bessi e al., 2018, Jia e al., 2021) by 2–9 yea s in compa ison o hose
wi h lowe CR. Consis en wi h he p e ious esul s, o he s udies ound
ha : a) a high le el o CR p edic s be e le els o cogni i e pe o mance
a e h ee yea s compa ed o pa icipan s wi h a low le el o CR
(Lojo-Seoane e al., 2018, 2020; Mondini e al., 2022) and b) lowe
educa ion in SCD pa icipan s inc eased he isk o p og ession o
objec i e cogni i e decline mo e han h ee yea s o ollow-up (Van den
Komme e al., 2008; Qi e al., 2018) o con e sion o demen ia a e
mo e han wo yea s o ollow-up (Sil a e al., 2014; Yun e al., 2020).
Only 2 s udies showed disc epan esul s poin ing ou ha a highe
educa ional le el in SCD inc eased he isk o demen ia a e h ee yea s
(Cha y e al., 2013) o p og ession o AD a e nine mon hs o ollow-up
(Van Oijen e al., 2007).
3.5. Me a-analyses
Ou o he 10 s udies included in he me a-analysis, h ee epo ed
su icien da a o me a-analyze con e sion isk sepa a ely in low and
high-educa ion g oups (Cha y e al., 2013; Jia e al., 2021; Van Oijen
e al., 2007). Besides, an addi ional s udy by Qi e al. (2018) did no
epo an o e all con e sion a e. Ins ead, i epo ed a es o con e sion
o MCI and demen ia sepa a ely. The e o e, sepa a e me a-analyses we e
conduc ed wi h he inclusion o Qi and colleagues’ s udy (2018) o
con e sion o MCI and demen ia. The esul o his analysis showed a
consis en associa ion o SCD wi h con e sion in bo h educa ion s a a.
In he low educa ion g oup, he pooled Risk Ra ios (RRs) o SCD we e,
espec i ely, 1.61 (95 % CI 1.04–2.49, p =0.03) and 1.62 (95 % CI 1.0 –
2.62, p =0.05) depending on whe he he esul s o he s udy by Qi e al.
(2018) we e inco po a ed o no . Some he e ogenei y (I
2
app oxima ely
63 % and he Q- es p =0.04) was obse ed in bo h analyses o he low
educa ion s a um. Fig. 2 p esen s a o es plo o he low-educa ion
g oup ha includes demen ia con e sion da a om Qi e al. (2018).
Simila ly, in he high educa ion g oup, he combined RRs o SCD
we e espec i ely 1.40 (95 % CI 1.12 – 1.75, p <0.001) and 1.58 (95 %
CI 1.24 – 2.01, p <0.001), depending on whe he he esul s o he s udy
by Qi e al. (2018) we e inco po a ed o no (see Fig. 3).
Ou o he en s udies included in he me a-analysis, six epo ed
mean educa ion le el and he associa ed SD among con e e s and non-
con e e s, hough Bessi e al. (2018) sepa a ely epo ed mean educa-
ion o hose con e ing o MCI o demen ia. Mean educa ion le el and
SD we e calcula ed as a weigh ed mean and weigh ed SD o Bessi e al.
(2018), weigh ing hese es ima es by he numbe o con e e s o MCI
and demen ia. The mean di e ences a e p esen ed in he o es plo
(Fig. 4). Despi e he e e sed di ec ion o e ec obse ed in he s udy
epo ed by Mondini e al. (2022), he e is no e idence o s a is ical
he e ogenei y, wi h bo h T
2
and I
2
ounding o ze o and he I
2
es ima e
no being s a is ically signi ican acco ding o he Q es (Q =3.86, d =
5, p =0.57). As displayed in Fig. 4, he con idence in e als associa ed
wi h his s udy (Mondini e al., 2022) a e consis en wi h a ange o
popula ion alues o e lapping wi h hose o he o he s udies. The
pooled es ima e o he mean di e ence was −2.0 (z = − 5.34, p <0.001)
indica ing ha , on a e age, he con e e s had wo ewe yea s o ed-
uca ion han he non-con e e s in hese s udies. Judging by he con i-
dence limi s, he es ima es in his li e a u e a e consis en wi h a
di e ence as small as 1.3 yea s o as la ge as 2.7 yea s a he 95 % le el
o con idence.
Me hodologically, he s udy de eloped by Mondini and colleagues
(2022) was cha ac e ized by high a i ion (95 %), c ea ing he possi-
bili y o me hodological he e ogenei y, bu also leading o imp ecision.
In a sensi i i y analysis wi h his s udy excluded he e was li le change
in he obse ed mean di e ence, which became −2.1 yea s (95 % CI:
−1.4 – −2.9) in his analysis.
4. Discussion
To he bes o ou knowledge, his is he i s sys ema ic e iew and
me a-analysis aimed a in es iga ing he in luence o educa ion on he
es ima ion o he isk o SCD p og ession owa ds objec i e cogni i e
impai men .
Among he selec ed s udies o he sys ema ic e iew, almos all
s udies (Bessi e al., 2018; Jia e al., 2021; Lojo-Seoane e al., 2018,
2020; Mazzeo e al., 2019; Mondini e al., 2022; Qi e al., 2018; Sil a
e al., 2014; Van den Komme e al., 2008; Yun e al., 2020) consis en ly
suppo he idea ha highe educa ion and o he CR p oxies a e asso-
cia ed o la e onse and educed isk o objec i e cogni i e decline,
including p og ession o MCI o con e sion o demen ia. Fu he mo e,
ou esul s suppo ha people wi h SCD wi h highe le els o educa ion
compa ed o hose wi h lowe educa ion appea ed o delay he p o-
g ession o objec i e cogni i e decline o e a pe iod anging om wo o
nine yea s.
These esul s a e consis en wi h hose hypo hesized om S e n’s CR
model (S e n e al., 2020), which p oposes ha highe educa ion le els
may enhance CR, which enables indi iduals o be e coun e ac he
g adual b ain changes associa ed wi h aging and neu odegene a i e
diseases. Thus, hey could be mo e esilien agains he cogni i e e ec s
o AD and o he demen ias, po en ially delaying he onse o he disease
o educing he se e i y o i s symp oms.
Ou esul s also coincide wi h p e ious longi udinal s udies ca ied
ou in pa icipan s wi hou cogni i e impai men and wi h hose ha
Fig. 2. Associa ion o SCD wi h con e sion o MCI o demen ia: low educa ion g oup. No e: All he s udies epo ed OR excep an Oijen e al. (2007) which e-
po ed HR.
S. A o a e al.
Ageing Resea ch Re iews 101 (2024) 102487
8
implemen c oss-sec ional designs in SCD. Thus, Amie a e al. (2014)
showed ha cogni i ely no mal indi iduals wi h highe educa ion
exhibi ed a slowe decline o e 15 yea s be o e mee ing he c i e ia o
demen ia, whe eas hose wi h lowe educa ion expe ienced a mo e apid
cogni i e decline o 7 yea s be o e demen ia onse . C oss-sec ional e i-
dence also sugges s ha highe educa ion in pa icipan s wi h SCD can
educe he isk o p og ession o objec i e cogni i e decline, MCI, o
demen ia (C umley e al., 2014; Hao e al., 2017, 2019; Yang e al.,
2020).
Howe e , he indings we e no uni e sally consis en . Two s udies
poin ed ou ha highe educa ion could, pa adoxically, aise he isk o
ansi ion o AD o demen ia (Cha y e al., 2013; Van Oijen e al., 2007).
These esul s ag ee wi h hose epo ed in p e ious s udies (Aghjayan
e al., 2016; Jonke e al., 2000; Rabin e al., 2017) and could gi e some
suppo o he hypo hesis ha p edic s a be e es ima e o cogni i e
changes in olde adul s wi h highe educa ion (He zog e al., 2008).
The i s me a-analysis calcula ed he isk o p og ession sepa a ely
in low and high-educa ion g oups and poin ed ou ha SCD is associa ed
wi h a highe equency o con e sion i espec i e o educa ion le el.
The con idence in e al leans mo e a o ably owa d hose wi h lowe
educa ion le els, sugges ing a compa a i ely highe likelihood o p o-
g ession in his subg oup. These esul s align wi h he CR hypo hesis
(S e n e al., 2020) and he a ailable e idence on he educ ion in he
isk o p og ession o delay in he cogni i e symp oms onse associa ed
wi h educa ion (Amie a e al., 2014; Ye e al., 2013; Yu e al., 2012),
al hough i does no elimina e he possibili y o su e ing om demen ia.
A second me a-analysis conduc ed by calcula ing s anda dized mean
educa ion di e ences be ween con e e s and non-con e e s indica ed
ha con e e s had wo yea s ewe o educa ion han non-con e e s.
Again, hese indings a e consis en CR hypo hesis (S e n e al., 2020)
showing ha lowe educa ion/CR is associa ed wi h an inc eased isk
and ea ly onse o cogni i e symp oms linked o demen ia compa ed o
highe educa ion. Howe e , ou esul s do no suppo he hypo hesis
ha he highe he educa ional le el in SCD, he g ea e he isk o
con e sion o demen ia (Aghjayan e al., 2016; Jonke e al., 2000;
Rabin e al., 2017). This e ec could be pa icula ly e iden in he mo e
ad anced s ages o he disease when CR is exhaus ed, leading o a mo e
apid p og ession, especially among indi iduals wi h highe educa ion
le els (Mungas e al., 2018; Zahodne e al., 2019). Howe e , his phe-
nomenon does no seem o mani es i sel in SCD, a s age in which
complain s a e in e p e ed as possible ea ly symp oms o demen ia.
In conclusion, ou s udy p o ides aluable insigh s in o he link be-
ween SCD and educa ion and i s p og ession o objec i e cogni i e
decline, MCI, o demen ia. Howe e , due o he limi ed s udies a ail-
able, ou indings a e en a i e. Fu u e esea ch is necessa y o ully
unde s and he mul i ace ed and in ica e ela ionship be ween
Fig. 3. Associa ion o SCD wi h con e sion o MCI o demen ia: high educa ion g oup. No e: All he s udies epo ed OR excep an Oijen e al. (2007) which
epo ed HR.
Fig. 4. Mean di e ences o yea s o educa ion in SCD pa icipan s con e e s and non-con e e s in six longi udinal s udies.
S. A o a e al.
Ageing Resea ch Re iews 101 (2024) 102487
9
