In e na ional Jou nal o
En i onmen al Resea ch
and Public Heal h
A icle
A No el Sco e o P edic ing Alzheime ’s Disease Risk om
La e Li e Psychopa hological and Heal h Risk Fac o s
Ja ie San abá ba a 1,2,3 , Juan Bueno-No i ol 4,* , Da en M. Lipnicki 5, Concepción de la Cáma a 2,3,6,7,
Raúl López-An ón2,3,8, An onio Lobo 2,3,7 and Pa icia G acia-Ga cía2,3,4,7


Ci a ion: San abá ba a, J.;
Bueno-No i ol, J.; Lipnicki, D.M.; de
la Cáma a, C.; López-An ón, R.; Lobo,
A.; G acia-Ga cía, P. A No el Sco e
o P edic ing Alzheime ’s Disease
Risk om La e Li e
Psychopa hological and Heal h Risk
Fac o s. In . J. En i on. Res. Public
Heal h 2021,18, 1802. h ps://
doi.o g/10.3390/ije ph18041802
Academic Edi o : Ma yam Vase i
Recei ed: 20 Decembe 2020
Accep ed: 8 Feb ua y 2021
Published: 12 Feb ua y 2021
Publishe ’s No e: MDPI s ays neu al
wi h ega d o ju isdic ional claims in
published maps and ins i u ional a il-
ia ions.
Copy igh : © 2021 by he au ho s.
Licensee MDPI, Basel, Swi ze land.
This a icle is an open access a icle
dis ibu ed unde he e ms and
condi ions o he C ea i e Commons
A ibu ion (CC BY) license (h ps://
c ea i ecommons.o g/licenses/by/
4.0/).
1Depa men o P e en i e Medicine and Public Heal h, Uni e sidad de Za agoza, 50001 Za agoza, Spain;
jsan aba ba a@uniza .es
2Ins i u o de In es igación Sani a ia de A agón (IIS A agón), 50001 Za agoza, Spain;
[email p o ec ed] (C.d.l.C.); lan on@uniza .es (R.L.-A.); alobo@uniza .es (A.L.);
[email p o ec ed] (P.G.-G.)
3
Cen o de In es igación Biomédica en Red de Salud Men al (CIBERSAM), Minis y o Science and Inno a ion,
28029 Mad id, Spain
4Psychia y Se ice, Hospi al Uni e si a io Miguel Se e , 50009 Za agoza, Spain
5Cen e o Heal hy B ain Ageing, School o Psychia y, Uni e si y o New Sou h Wales Medicine,
2052 Randwick, Aus alia; [email protected]
6Psychia y Se ice, Hospi al Clínico Uni e si a io Lozano Blesa, 50009 Za agoza, Spain
7Depa men o Medicine and Psychia y, Uni e sidad de Za agoza, 50001 Za agoza, Spain
8Depa men o Psychology and Sociology, Uni e sidad de Za agoza, 50001 Za agoza, Spain
*Co espondence: [email p o ec ed]; Tel.: +34-659-743-354
Abs ac :
Wi h he inc easing size o he aging popula ion, demen ia isk educ ion has become a
main public heal h conce n. Demen ia isk models o indices may help o iden i y indi iduals in he
communi y a high isk o de elop demen ia. We ha e aimed o de elop a no el demen ia isk index
ocused on he la e-li e (65 yea s o mo e) popula ion, ha add esses isk ac o s o Alzheime ’s
disease (AD) easily iden i iable a p ima y ca e se ings. These isk ac o s include some shown o be
associa ed wi h he isk o AD bu no ea u ed in exis ing indices, such as hea ing loss and anxie y.
Ou index is also he i s o accoun o he compe ing isk o dea h. The Za agoza Demen ia and
Dep ession P ojec (ZARADEMP) Alzheime Demen ia Risk Sco e p edic s an indi idual
´
s isk o
de eloping AD wi hin 5 yea s. The p obabili y o la e onse AD signi ican ly inc eases in hose wi h
isk sco es be ween 21 and 28 and, u he mo e, is almos 4- old highe o hose wi h isk sco es o
29 o highe . Ou index may p o ide a p ac ical ins umen o iden i y subjec s a high isk o AD
and o design p e en i e s a egies a ge ing he con ibu ing isk ac o s.
Keywo ds: isk index; demen ia; psychopa hological isk ac o s; ZARADEMP; compe ing isk
1. In oduc ion
The e we e 50 million people wo ldwide li ing wi h demen ia in 2018, a an es ima ed
inancial cos o socie y o $1 billion. Wi h he apidly g owing global popula ion o olde
indi iduals, he p e alence o demen ia and i s associa ed cos a e expec ed o double by
2030 [
1
]. Acco dingly, demen ia is ecognized as a Public Heal h P io i y by he Wo ld
Heal h O ganiza ion (WHO) [
2
] and demen ia isk educ ion is one o he main a ge s in
he Global Ac ion Plan on he Public Heal h Response o Demen ia 2017–2025 [
3
]. The mos
common o m o demen ia is Alzheime
´
s Disease (AD), which may con ibu e o 60–70%
o cases [4].
Gi en he conside able implica ions o demen ia o a ec ed indi iduals, hei amilies
and socie y, as well as he e being no e ec i e ea men , a p e en i e app oach is c ucial.
P e en ion s a egies could bo h delay he onse o demen ia and educe i s p e alence [
5
].
Howe e , o op imize he e ec i eness o isk educ ion p og ams, i is necessa y o know
he modi iable isk ac o s o demen ia and ha e eliable es ima es o hei e ec size [
6
].
In . J. En i on. Res. Public Heal h 2021,18, 1802. h ps://doi.o g/10.3390/ije ph18041802 h ps://www.mdpi.com/jou nal/ije ph
In . J. En i on. Res. Public Heal h 2021,18, 1802 2 o 13
This can acili a e he de elopmen o a demen ia isk sco e o iden i y indi iduals a high
isk o de eloping demen ia o a ge ing wi h p e en ion s a egies.
Se e al models o p edic ing demen ia ha e been de eloped (see Tang e al. [
7
]
o a e iew). The CAIDE (Ca dio ascula Risk Fac o s, Aging and Demen ia) s udy
de eloped a isk sco e o p edic ing la e-li e demen ia in middle-aged people [
8
], bu he
model showed poo anspo abili y o olde aged coho s [
7
] and low and middle income
coun ies [
9
]. O he models ha e included isk a iables ha can be expensi e o assess
and a e no uni e sally a ailable, such as b ain imaging [
10
] o Apo-E4 geno ype [
11
].
Mo e ecen ly de eloped isk models ha e educed complexi y [
12
] and inco po a e sel -
epo ed a iables, such as he Aus alian Na ional Uni e si y Alzheime
´
s Disease Risk
Index (ANU-ADRI) [
13
] and he LI es yle o BRAin heal h (LIBRA) index [
14
], o include
a iables easily accessible in p ima y ca e, such as he B ie Demen ia Sc eening Indica o
(BDSI) [15] and he F amingham Hea S udy isk sco e [16].
An ex e nal alida ion s udy o ou demen ia p edic ion models (including he
CAIDE, ANU-ADRI and BDSI) in an elde ly, communi y-dwelling sample ound a ying
accu acies o p edic ing demen ia (C-s a is ics and 95%CI a 5-yea ollow-up: CAIDE 0.54
(0.50–0.58), BDSI 0.80 (0.76–0.84), ANU-ADRI 0.78 (0.76–0.81) and DRS 0.82 (
0.78–0.86
)), as
well as all models pe o ming simila ly o p edic ions based on age alone [
17
]. The au ho s
ecommended ha new o e ined demen ia p edic ion models a e needed. Demen ia
p edic ion models may be made mo e accu a e by including isk ac o s no ea u ed
p e iously. To his end, dep ession and anxie y ha e been ecen ly ecognized as po en ially
use ul [
18
]. Few demen ia isk sco es de eloped om popula ion-based coho s ha e
included dep ession [
13
–
15
], and hese used sel - epo ed i ems o symp oma ic scales
a he han mo e s ingen clinical c i e ia o dep ession known o ha e a highe associa ion
wi h demen ia isk [
19
]. Sys ema ic e iews and me a-analysis suppo anxie y as a majo
isk ac o o demen ia [
20
,
21
] and AD [
22
]. Fu he mo e, we ha e ecen ly epo ed
ha clinically ele an anxie y is a isk ac o o o e all demen ia [
23
] and AD [
24
] in he
elde ly. Howe e , o he bes o ou knowledge, no demen ia isk model has ye included
anxie y. Hea ing loss is ano he isk ac o o demen ia, as iden i ied by he 2020 Lance
Commission epo [25], which has no been included in p e ious demen ia isk sco es.
Dea h is a compe ing isk o he de elopmen o nume ous diseases in old age, and i
is ecommended his mo ali y e ec be conside ed in incidence s udies [
26
]. While isk
models o diabe es and co ona y a e y disease ha e used a compe ing isk analysis model
in he p esence o dea h [
26
,
27
], his does no appea o ha e been done o demen ia. I
has been acknowledged ha no doing so migh ha e biased he demen ia isk p edic ions
o p e ious models [
14
,
17
], and ha u u e models should ake he mo ali y e ec in o
accoun [17].
In his s udy, we aimed o de elop a new demen ia isk sco e ha includes dep ession,
anxie y, and hea ing loss in addi ion o he isk ac o s mo e commonly used. We in ended
o all included ac o s o be sel - epo able o accessible o p ima y ca e doc o s. Ou
model is u he dis inguished by aking he compe ing isk o dea h in o accoun .
2. Ma e ials and Me hods
This wo k ollows S eng hening he Repo ing o Obse a ional S udies in Epidemi-
ology (STROBE) [
28
] and he S a is ical Analysesand Me hods in he Published Li e a u e
(SAMPL) [
29
] guidelines o epo ing obse a ional s udies in epidemiology and s a is-
ics, espec i ely.
2.1. Sample and P ocedu e
We used da a om he Za agoza Demen ia and Dep ession (ZARADEMP) P ojec ,
a longi udinal, popula ion-based s udy in ended o documen he incidence and isk
ac o s o soma ic and psychia ic diseases, speci ically demen ia, in adul s aged
≥
55
yea s. The E hics Commi ee o he Ins i u ional Re iew Boa d (CEICA) app o ed he
In . J. En i on. Res. Public Heal h 2021,18, 1802 3 o 13
s udy, and p inciples o w i en in o med consen , p i acy, and con iden iali y ha e been
main ained h oughou .
Wa e I (ZARADEMP I) is a baseline, c oss-sec ional s udy, in ended o iden i y a
coho o indi iduals wi hou demen ia, as well as he p e alence and dis ibu ion o
he hypo hesized isk ac o s o demen ia. The ield wo k o his baseline s udy was
comple ed by well- ained lay in e iewe s (senio medical s uden s). In e iews we e con-
duc ed a pa icipan s home o , in he case o ins i u ionalized subjec s, a ins i u ion. The
in e iews las ed 25–90 min and inco po a ed alida ed Spanish e sions o he ollowing
in e na ional assessmen ins umen s: he Mini-Men al S a us Examina ion (MMSE); Ge i-
a ic Men al S a e B (GMS-B); Au oma ed Ge ia ic Examina ion o Compu e -Assis ed
Taxonomy (AGECAT); His o y and Ae iology Schedule (HAS); Ka z’s Index o basic
ac i i ies o daily li ing (ADL), he Law on and B ody scale o ins umen al ADL, and
he Eu opean S udies o Demen ia (EURODEM) Risk Fac o s Ques ionnai e. The blood
p essu e, heigh and weigh o each indi idual we e also checked and eco ded ou inely
in his phase o he s udy. Medical epo s we e used in some cases when a ailable o help
in he diagnos ic p ocess [30,31].
The Ge ia ic Men al S a e (GMS) is a well-known semis uc u ed s anda dized clini-
cal in e iew o assessing he men al s a e o elde ly pe sons. I is also a synd ome case
inding ins umen o communi y subjec s, and he compu e ized p og am AGECAT
can be applied wi h his pu pose [
32
]. The His o y and Ae iology Schedule (HAS) is a
s anda dized me hod o collec ing his o y and ae iology da a om an in o man , o di ec ly
om he esponden when hey a e judged o be eliable. I concen a es on hose ea u es
expec ed o be ele an o psychia ic diagnosis in olde people and is c ucial o comple e
he GMS and acili a e a diagnos ic p ocess. The Risk Fac o s Ques ionnai e used in his
s udy was designed by he EURODEM Wo kg oup [
33
]. The ins umen is in ended o
include in o ma ion ela ed o he ollowing po en ial isk ac o s o demen ia, Azheime ’s
Disease and ascula demen ia: his o y o medical diseases, including ca dio ascula
disease, auma ic b ain inju y, epilepsy, Down’s Synd ome, Pa kinson’s disease, diabe es
melli us, hy oid disease, abuse o alcohol o smoking; menopause; psychia ic his o y,
in pa icula dep ession; use o medica ions; his o y o gene al. Each i em in he in e -
iew has been ope a ionally de ined, acco ding o p e iously ag eed EURODEM c i e ia
[30,31]
. Indi iduals we e nomina ed as “p obable cases” on he basis o GMS h eshold
“global” sco es (1/2) and/o MMSE s anda d cu -o poin s (23/24), decided on he bases
o adequa e nega i e p edic i e alue. Howe e , he da a on each elde ly we e ho oughly
e iewed by he esea ch psychia is s supe ising indi idually he lay in e iewe s. In he
inal s ep o Wa e I, he psychia is s eco ded a diagnosis o “demen ia”, “dep ession”,
“cases”. “Subcases” o “demen ia” we e also nomina ed on he basis o bo de line sco es
on he same ins umen s. Fo a diagnosis o demen ia, documen ed de e io a ion in ADL
due o cogni i e de e io a ion was equi ed [30,31].
In he ZARADEMP s udy, he ep esen a i e sample was d awn om Spanish o icial
census lis s, s a i ied wi h p opo ional alloca ion by age and sex, and included ins i u-
ionalized indi iduals. The baseline assessmen in 1994 included 4803 indi iduals. He e,
we epo esul s om baseline (Wa e I) and wo ollow-up wa es (Wa es II and III). As
we we e in e es ed in cogni i ely in ac indi iduals, we excluded subjec s conside ed o be
cases o subcases o demen ia a baseline (n= 746) o he ollow-up, esul ing in an ini ial
sample o 4057 pa icipan s, o which 2704 pa icipa ed in Wa e II and hen 2258 in Wa e
III. Fo an easy compa ison wi h he exis ing li e a u e, we ha e selec ed he age g oup o
o e 65 yea s (n= 3044).
2.2. Diagnosis o Inciden AD a Wa es II and III
A wo-phase, sc eening p ocedu e was implemen ed in each o he wa es, II and III,
using he Spanish e sions o he in e na ional assessmen ins umen s de ined o Wa e I.
Pa icipan s we e classi ied in phase I as “p obable cases” o demen ia based on he GMS
h eshold “global” sco e (1/2) and/o MMSE s anda d cu -o poin (23/24). In phase II,
In . J. En i on. Res. Public Heal h 2021,18, 1802 4 o 13
all p obable cases o demen ia we e eassessed in hei place o esidence by a esea ch
psychia is using he same ins umen s in phase I, as well as Hachinski’s scale [
34
], and a
b ie , p e iously s anda dized neu ological examina ion. Inciden demen ia and ype (AD,
ascula , o he ) we e ini ially diagnosed by he esea ch psychia is , bu a inal diagnosis
based on he Diagnos ic and S a is ical Manual o Men al Diso de s-IV (DSM-IV) was
made by a consensus panel (a leas h ee o ou psychia is s in ag eemen ). Ou p e ious
s udies ha e suppo ed he alidi y o his diagnos ic p ocess [
35
]. Mo eo e , o documen
he accu acy o he panel, a p opo ion o cases we e in i ed o a hospi al diagnos ic wo k-
up, which inco po a ed a neu opsychological ba e y and neu oimaging, and he Na ional
Ins i u e o Neu ological and Communica i e Diso de s and he Alzheime ‘s Disease
and Rela ed Diso de s Associa ion (NINCDS-ADRDA) [
36
] c i e ia used o diagnose AD.
Ag eemen be ween panel membe s on he diagnosis o demen ia and AD was eached in
95.8% and 86.7% o cases, espec i ely [31].
2.3. Risk Fac o s Assessed
The po en ial isk ac o s we e e alua ed a baseline by lay-in e iewe s (indi idually
supe ised by esea ch psychia is ), and a e classi ied as socio-demog aphic (age, sex,
educa ional le el, and ma i al s a us), psychological (anxie y and dep ession), beha io al
( obacco and alcohol use, and obesi y), and medical (hea ing loss, hype ension, diabe es,
and his o y o angina, acu e myoca dial in a c ion o s oke).
2.3.1. Socio-Demog aphic Risk Fac o s
To simpli y he sco ing me hod, con inuous a iables we e ca ego ized. Fo compa -
ison wi h p e ious isk sco es, we only include indi iduals aged 65 o mo e yea s, and
ca ego ized ages in o h ee g oups: 65–74, 75–84 and 85+ yea s. Educa ional s a us was
classi ied as: “illi e a e (unable o ead and w i e, o wi h less han 2 yea s o o mal educa-
ion)”, “p ima y s udies (comple e o incomple e)” and “seconda y educa ion o abo e”.
Ma i al s a us was de ined as “single”, “ma ied o li ing as a couple” and “ o me ly
ma ied (di o ced, sepa a ed o widowed)”.
2.3.2. Psychological Risk Fac o s
The diagnosis o anxie y and dep ession was based on he GMS-AGECAT sys em.
A e symp om assessmen by he GMS-B Scale, a compu e p og am compa ed synd ome
clus e s (e.g., demen ia, dep ession, anxie y) o each a inal diagnosis, o his s udy we
conside ed “case” le els o anxie y o dep ession (con idence le els
≥
3). The ecommended
cu -o
≥
3 has shown a good sensi i i y (0.91) and speci ici y (0.89) o diagnosis o dep es-
sion clinically signi ican ; ha means cases wi h cu en signs and symp oms o dep ession
a ed by clinicians se e e enough o equi e an idep essan in e en ion [37].
2.3.3. Beha io al Risk Fac o s
Alcohol and obacco use we e sel - epo ed, and bo h ca ego ized as use , non-use ,
o o me use . Body mass index (BMI) was calcula ed as weigh in kilog ams di ided by
heigh in me e s squa ed. A BMI g ea e han 30 kg/m2was de ined as “obesi y”.
2.3.4. Medical Risk Fac o s
Hea ing loss was sco ed when he subjec was almos o comple ely dea . His o y o
ca dio ascula isk ac o s (angina, myoca dial in a c ion o s oke) and diabe es was based
on da a om he EURODEM ques ionnai e [
33
]. A posi i e his o y o diabe es was based on
a p e ious medical diagnosis and/o ecei ing ea men o diabe es. Blood p essu e (BP)
was calcula ed as he mean alue o 2 measu emen s using a s anda d sphygmomanome e .
Hype ension was de ined as a sys olic BP g ea e han 140 mmHg, a dias olic BP g ea e
han 90 mmHg, and/o use o blood p essu e-lowe ing d ugs.
In . J. En i on. Res. Public Heal h 2021,18, 1802 5 o 13
2.4. S a is ical Analysis
We assessed demen ia isk on cogni i ely in ac subjec s a baseline, acco ding o isk
o de eloping demen ia a ollow-up o e e y s udied isk ac o .
The ollow-up pe iod was he ime om baseline (Wa e I) o whiche e o he ollow-
ing occu ed i s : demen ia, dea h, loss o ollow-up, o he end o ollow-up (Wa e III)
a e nea ly 5 yea s. We used a compe i i e isk eg ession [
38
] o subdis ibu ion haza ds
eg ession o adjus es ima es o inciden demen ia isk aking in o accoun he compe i i e
mo ali y isk [
39
] con eyed by he o e all du a ion o ollow-up. The subdis ibu ion
haza d a io (SHR) o assessing he isk o de eloping demen ia o an indi idual ac o
was calcula ed using he cmp sk lib a y in he Rpackage, which allows adjus men o
socio-demog aphic and clinical a iables. The SHR is a way o exp essing he ins an aneous
isk o de eloping a gi en e en in an indi idual who has no ye expe ienced such an
e en (a isk), when aking in o accoun a compe ing isk, such as dea h [
40
]. A highe
SHR would mean a highe isk o de eloping AD o his ac o , aking in o accoun dea h
as a compe ing isk.
We chose an app oach simila o ha o Li e al. [
16
]. Fi s , all po en ial isk ac o s we e
included sepa a ely in subdis ibu ion haza ds eg ession models. Fac o s ha eached
p≤0.1
we e hen simul aneously included in a mul i a ia e p edic ion model and as
a iables o he isk sco e. Fo each model, he SHR and 95% con idence in e al we e
compu ed. To examine he p opo ional haza ds assump ion, he ime- a ying e ec o
each co a ia e was es ed using he Scheike and Zhang es [41].
Since all a iables we e ca ego ical, hei es ima ed con ibu ion o he isk o demen ia
could be exp essed by simpli ied sco es assigned o each ca ego y. We assigned a isk sco e
o each ac o using he
β
-coe icien s o mul i a ia e subdis ibu ion haza d models. To
acili a e in e p e a ion,
β
coe icien s we e s anda dized by di iding hem by he lowes
obse ed alue (i.e., he lowes hen had a alue o 1) and ounding o he closes in ege .
Since he lowes
β
alue was 0.15 and i s mul iplica ion by 6.7 makes i app oxima ely 1, all
β
alues we e mul iplied by 6.7 and we e ounded o he closes in ege [
8
]. Fo addi ional
analyses, we summed hese sco es as p edic o s o he isk o AD incidence o e a 5-yea
ollow-up pe iod. This was pe o med o each pa icipan based on a cumula i e incidence
unc ion (CIF), aking in o accoun he compe ing e en (dea h) as ime p og essed [
42
], as
we ha e p e iously done [31].
3. Resul s
Ou inal sample included 3044 pa icipan s aged 65+ wi hou demen ia a baseline
(median 4.4 yea s; in e qua ile ange: 2.9–4.9 yea s). Du ing he ollow-up pe iod, 663
(21.8%) indi iduals died, 582 (19.1%) we e los (by e usal o ake pa , changing esidence
o being impossible o con ac ), 85 (2.8%) we e inciden AD cases and 47 (1.5%) we e
inciden cases o o he demen ias. Using a compe i i e isk eg ession model, hey we e all
included in he isk calcula ion [38].
Table 1shows baseline demog aphic cha ac e is ics acco ding o AD incidence s a-
us. Pa icipan s wi h inciden AD we e signi ican ly olde , mo e likely o be emale,
o me ly ma ied, and o ha e lowe educa ional le el and highe anxie y han pa icipan s
wi hou AD.
Table 1.
Baseline cha ac e is ics acco ding o inciden AD s a us and associa ions be ween indi idual isk ac o s and AD isk.
Va iables
Follow-Up AD S a us Uni a ia e Reg ession Model
No AD
(N= 2959)
Inciden AD
(N= 85) p-Value SHR (95% CI) ap-Value
Sociodemog aphic
cha ac e is ics
Age (yea s) 75.4 (7.7) 84.2 (6.3) <0.001 1.14 (1.12–1.17) <0.001
Female sex 1645 (55.6%) 59 (69.4%) 0.016 1.81 (1.14–2.87) 0.012
Educa ion (yea s) 7.3 (3.8) 5.9 (3.8) 0.001 0.89 (0.82–0.96) 0.003

In . J. En i on. Res. Public Heal h 2021,18, 1802 6 o 13
Table 1. Con .
Va iables
Follow-Up AD S a us Uni a ia e Reg ession Model
No AD
(N= 2959)
Inciden AD
(N= 85) p-Value SHR (95% CI) ap-Value
Ma i al s a us ( e . single)
<0.001
Ma ied/in couple 1690 (57.1%) 25 (29.4%) 2.06 (0.49–8.68) 0.320
Fo me ly ma ied 980 (33.1%) 58 (68.2%) 8.18 (2.00–33.39) 0.003
Psychological isk ac o s
Dep ession 306 (10.3%) 14 (16.4%) 0.101 1.44 (0.81–2.55) 0.210
Anxie y 66 (2.2%) 6 (7.0%) 0.011 3.28 (1.43–7.54) 0.005
Beha io al isk ac o s
Alcohol o Smoking 758 (25.6%) 20 (23.5%) 0.757 0.90 (0.54–1.49) 0.680
BMI 26.9 (7.3) 26.3 (5.3) 0.463 0.63 (0.44–0.90) 0.012
Medical isk ac o s
Diabe es 395 (13.5%) 10 (11.9%) 0.803 0.87 (0.45–1.68) 0.680
Hype ension 2112 (71.5%) 55 (64.7%) 0.214 0.73 (0.47–1.15) 0.170
Hea ing loss 22 (0.7%) 2 (2.3%) 0.304 3.17 (0.80–13.20) 0.022
Angina 177 (6.1%) 4 (4.7%) 0.776 0.75 (0.27–2.04) 0.570
Myoca dial in a c ion 86 (3.0%) 4 (4.8%) 0.527 1.61 (0.59–4.38) 0.350
S oke 171 (5.8%) 5 (5.9%) 0.845 1.04 (0.80–2.57) 0.930
No es: Da a a e gi en as mean (s anda d de ia ion) o numbe (%); AD: Alzheime ’s disease; CI: con idence in e al; SHR: subdis ibu ion
haza d a io.
a
Repo ed SHR o AD is ela ed o non-cases, CIs and p alues ela ed o SHR we e om “no mal app oxima ion” o Wald
χ2
es wi h 1 d .
O he isk ac o s assessed, hose associa ed wi h AD isk in uni a ia e eg ession
models we e age, sex, educa ion, ma i al s a us, anxie y, BMI, and hea ing loss (
Table 1
).
These ac o s we e included in a mul i a ia e model, wi h Table 2showing he
β
-coe icien s
and SHRs o AD inciden cases and he isk sco es assigned o each isk ac o . Dep ession
was no signi ican ly associa ed wi h AD isk, bu was kep in he inal model because we
p e iously ound in a me a-analy ic s udy including ou sample ha pa icipan s wi h
clinically signi ican dep ession had a wo- old highe isk o AD [
43
] and ha se e e
dep ession was associa ed wi h a 4- old isk o inciden AD in ou sample [
44
]. In addi-
ion, emo al o dep ession om he analysis did no change he sco es de i ed o he
o he ac o s.
Based on selec ed ac o s, he o al sco e anged om 0 o 56 (Table 2). Fo each
one-poin inc emen in he isk scale, he AD isk inc eased signi ican ly by 16% (SHR:
1.16; 95% CI: 1.12–1.19; p< 0.001). Table 3shows ha he isk o AD inc eased ac oss isk
sco e ca ego ies, de ined by di iding ou sample in o qua iles. Using ou isk scale, an
85-yea -old male, wi h highe educa ion, single, wi hou dep ession o anxie y, o e weigh
and wi hou hea ing loss has 17 isk poin s in o al and a 1.78- old isk o AD. Howe e , an
85-yea -old woman, wi h p ima y educa ion, ma ied, anxie y, dep ession, no mal weigh ,
and hea ing loss has 48 isk poin s and a 22.6- old isk o AD. Finally, Table 4shows he
sco e shee de eloped o p edic Alzheime ’s disease.
Table 2. Risk ac o associa ions wi h AD isk in a mul i a ia e eg ession model.
Va iables βCoe icien SHR (95% CI) ap-Value Risk Sco e
Sociodemog aphic cha ac e is ics
Age
65–74 (n= 1584) 0 ( e e ence) 1 ( e e ence) 0
75–84 (n= 902) 1.64 5.16 (2.32–11.50) <0.001 11
O e 85 (n= 558) 2.54 12.66 (5.71–28.06) <0.001 17
Sex
Male (n= 1340) 0 ( e e ence) 1 ( e e ence) 0
Female (n= 1704) 0.46 1.59 (0.92–2.74) 0.096 3
Educa ion (yea s)
Seconda y o highe (n= 479) 0 ( e e ence) 1 ( e e ence) 0
P ima y (n= 2275) 0.23 1.26 (0.59–2.71) 0.550 2
Illi e a e (n= 266) 1.08 2.95 (1.23–7.10) 0.015 8
In . J. En i on. Res. Public Heal h 2021,18, 1802 7 o 13
Table 2. Con .
Va iables βCoe icien SHR (95% CI) ap-Value Risk Sco e
Ma i al s a us
Single (n= 284) 0 ( e e ence) 1 ( e e ence) 0
Ma ied/in couple (n= 1715) 1.26 3.51 (0.80–15.41) 0.096 9
Fo me ly ma ied (n= 1038) 1.66 5.28 (1.27–22.00) 0.022 11
Psychological isk ac o s
Dep ession
No case (n= 2674) 0 ( e e ence) 1 ( e e ence) 0
Case (n= 370) 0.15 1.16 (0.64–2.12) 0.630 1
Anxie y
No case (n= 2972) 0 ( e e ence) 1 ( e e ence) 0
Case (n= 72) 1.20 3.32 (1.39–7.94) 0.007 8
Beha io al isk ac o s
BMI
O e weigh /Obesi y (n= 2051) 0 ( e e ence) 1 ( e e ence) 0
No mal (n= 993) 0.50 1.65 (1.04–2.63) 0.034 4
Medical isk ac o s
Hea ing loss
No case (n= 3014) 0 ( e e ence) 1 ( e e ence) 0
Case (n= 24) 0.49 1.63 (0.35–7.47) 0.530 4
No es: AD: Alzheime ’s disease; CI: con idence in e al; SHR: subdis ibu ion haza d a io.
a
Repo ed SHR o AD is ela ed o non-cases,
CIs and p alues ela ed o SHR we e om “no mal app oxima ion” o Wald χ2 es wi h 1 d .
Table 3. Fine and G ay eg ession model ela ing he sco e qua iles wi h isk o AD.
Risk Sco e Uni a ia e Reg ession Model
No. a Risk aInciden AD Cases (%) SHR (95% CI) bp-Value
0–14 954 4 (0.4%) 1 ( e e ence)
15–20 676 5 (0.7%) 1.78 (0.48–6.63) 0.390
21–28 714 17 (2.4%) 5.78 (1.95–17.16) 0.002
29+ 663 59 (8.9%) 22.61 (8.23–62.12) <0.001
No es: AD: Alzheime ’s disease; CI: con idence in e al; SHR: subdis ibu ion haza d a io.
a
O he 3044
pa icipan s in he baseline, 27 had missing isk sco e alues and we e excluded, lea ing a o al o 3007 a isk.
b
Repo ed SHR o AD is ela ed o non-cases, CIs and p alues ela ed o SHR we e om “no mal app oxima ion”
o Wald χ2 es wi h 1 d .
Table 4. Sco e anges and p obabili y o AD wi hin 5 yea s o indi iduals aged o e 65 yea s.
Risk Sco e 5 y AD P obabili y (%)
0–5 0.11
6–10 0.24
11–15 0.49
16–20 1.02
21–25 2.13
26–30 4.41
31–35 9.01
36–40 17.92
41–45 33.82
46–50 57.81
50+ 83.54
No es: AD: Alzheime ’s disease; y : yea .
4. Discussion
The “ZARADEMP Alzheime Demen ia Risk Sco e” p edic s an indi idual’s isk o
de eloping AD wi hin 5 yea s based on selec ed isk ac o s easily accessible in p ima y
ca e se ings: age, sex, educa ion, ma i al s a us, dep ession, anxie y, BMI, and hea ing
loss. Mos a iables a e assessed by di ec ques ions, and obesi y and clinically signi ican
anxie y and dep ession a e egula ly app oached by p ima y ca e doc o s in hei daily
ou ine p ac ice. Ou index could be easily applied by any clinician aiming o assess he isk
o AD in hei ou inely p ac ice (p ima y ca e doc o s, neu ologis , psychia is , clinical
psychologis , o ge ia is ). Mo eo e , isk index could be calcula ed in a simple way: he
In . J. En i on. Res. Public Heal h 2021,18, 1802 8 o 13
speci ic isk sco es (RS) o any o he a iables a e summed up o each indi idual. Fo
example, a 76-yea -old (RS 11) woman (RS 3), widowed (RS 11), wi h p ima y s udies (RS
2), wi h clinically signi ican anxie y (RS 8) bu no clinically dep essed (RS 0), non-obese
(RS 4), and wi hou hea ing loss (RS 0), would ha e a o al isk sco e o 39 o de eloping
AD a 5 yea s o ollow-up. The p obabili y o la e-onse AD was signi ican ly high o isk
sco es be ween 21 and 28, bu almos 4- old highe han his o isk sco es 29+.
In ou inal model, age was he ac o mos s ongly associa ed wi h AD isk, as ex-
pec ed, since age is consis en ly he g ea es isk ac o o o e all demen ia [
25
]. P e alence
o AD inc eases con inuously and exponen ially wi h age, being epo ed 3% in subjec s
om 65 o 69 yea s old and 32% a age o 85 o olde [
45
]. While p e ious s udies ha e
consis en ly shown women as ha ing highe p e alence o AD han men, esul s abou
di e ences in he isk o de eloping AD o men and women o he same age a e mixed [
45
].
Consis en ly, women in ou sample showed a signi ican inc eased isk o AD ela ed o
men in he uni a ia e eg ession model, bu esul s we e no s a is ically signi ican in he
mul i a ia e eg ession model. None heless, a endency o inc eased isk o AD in women
was obse ed, and his a iable ep esen s 3 poin s in ou inal isk sco e. The g ea e isk
o AD associa ed wi h illi e acy is consis en wi h p e ious esul s ha sugges an in e se
associa ion be ween educa ional achie emen and isk o demen ia, [
46
,
47
] suppo ing he
cons uc o “cogni i e ese e”. “Cogni i e ese e”, o “ ese e” [
48
], e e s o he b ain
´
s
abili y o de elop cogni i e ne wo ks ha enable a pe son o con inue o pe o m cogni i e
ask despi e degene a i e b ain changes [
45
,
48
]. Besides yea s o o mal educa ion, o e en
gene ic o o he en i onmen al ac o s [
48
], engaging in s imula ing men al ac i i ies may
also help o build cogni i e ese e [
45
], so ha his could be a modi iable ac o o AD. As
o ci il s a us, indi iduals o me ly ma ied had a highe isk o AD han single o ma ied
indi iduals, maybe because o g ea e loneliness [
49
] which has been shown o con ibu e
o demen ia in a p e ious me a-analysis [
50
]. In line wi h he li e a u e, we ound a highe
incidence o AD in subjec s wi h hea ing loss. I has been widely associa ed in he li e a u e
wi h he isk o AD [
51
] and all-cause demen ia [
52
], and se e al hypo heses abou such
causal ela ionship ha e been p oposed. Among hem, i is hypo hesized ha i could lead
o social isola ion, and his o demen ia [
51
]. In addi ion, i has been shown ha he gene ic
isk o AD also in luences he hea ing o speech in noise, wi hou hese hea ing de ici s
being ela ed o u he cogni i e impai men [
53
]. Rega ding obesi y, we ound in ou
elde ly sample ha highe BMI had p o ec i e e ec s on demen ia isk. Howe e , obesi y
in midli e has been iden i ied as a isk ac o o demen ia [
25
]. Ou esul s a e consis en
wi h hose o Li e al. [
16
], and suppo p e ious s udies ha ound age-dependen e ec s
o obesi y on demen ia isk [
54
]. Simila age-dependen e ec s ha e been desc ibed o
hype ension [
55
]. I would be in e es ing o s udy he e ec o o he beha io al ac o s
such as he habi ual consump ion o speci ic p oduc s, as e ec s on memo y ha e been
obse ed [
56
] and as his should be easily a ailable in o ma ion in a p ima y ca e in e -
iew. In e es ingly, we ound ha clinically signi ican anxie y showed a much s onge
associa ion wi h AD isk han clinically signi ican dep ession, and ha hese psychological
a iables con ibu ed o AD isk mo e han ca dio- ascula isk ac o s and diabe es.
Compa ed o p e ious indices, ou “ZARADEMP Alzheime Demen ia Risk Sco e”
includes mainly socio-demog aphical and psychological a iables and is he i s o include
hea ing loss and anxie y. No ice ha anxie y had a mode a e-high weigh on o al isk
sco e (8 poin s) and con ibu es o AD isk as much as illi e acy. Mo eo e , we assessed
cu en and clinically signi ican anxie y and dep ession, whe eas p e ious indices assessed
dep ession using sel - epo symp oma ic scales. A la ge me a-analysis o demen ia isk
es ima es o dep ession [
19
] demons a ed highe isk es ima es o dep ession assessed
by mo e s ingen , and p e iously alida ed agains clinical c i e ia han hose using a
milde cu -o in symp oma ic scales. In his sense, dep ession acco ding o GMS-AGECAT
c i e ia has shown an accep able o e all ag eemen wi h dep ession acco ding o DSM
c i e ia and highe sensi i i y o de ec clinically signi ican dep ession in elde people [
37
].
While we did no ind clinically signi ican dep ession o be signi ican ly associa ed wi h
In . J. En i on. Res. Public Heal h 2021,18, 1802 9 o 13
he isk o AD in ou sample, we decided o include i in ou index because o dep ession
being a well-es ablished isk ac o o demen ia and AD [
19
,
25
] and p e iously epo ed
associa ions based on [
44
] o including ou sample [
43
]. We p e iously ound signi ican
esul s in ou sample only o se e e dep ession [
44
], bu in a u he me a-analysis, wi h
mo e powe o de ec an e ec han ou indi idual s udy, we ound signi ican esul s
o clinically signi ican dep ession [
43
]. Howe e , dep ession only added 1 poin o ou
isk sco e.
Some o he isk ac o s included in ou index, such as anxie y and dep ession, migh
no be independen o AD and could be p od omal symp oms o he disease p io o
clinical diagnosis. Despi e he exclusion o subjec s wi h cogni i e impai men , he lack
o bioma ke s o AD in ou sample could ha e led o p eclinical AD being unde es i-
ma ed. Howe e , we assessed clinically signi ican dep ession and anxie y, excluding
mild/ subsynd omal symp oms, and we hink ha he highe speci ici y in diagnosis o
dep ession and anxie y could suppo he hypo hesis o he eal isk in e p e a ion, as
opposed o p od omal symp oms o emo ional dys egula ion desc ibed as pa o he Mild
Beha io al Impai men cons uc [
57
]. This is cu en ly a con o e sial ques ion and u he
s udies a e needed, because he ew s udies ha ha e explo ed longi udinal ajec o ies o
dep ession in p eclinical phase o demen ia [
58
,
59
] sugges ha dep ession is mo e likely
o be a p od omal symp om and ela ed o demen ia- ela ed b ain changes.
In addi ion, a ecen sys ema ic e iew s a es ha subpopula ions wi h di e en isk
p o iles need o be conside ed and ailo ed scales c ea ed [
60
]. In his sense, i has been
obse ed ha p e ious models ca ied ou on middle-aged coho s ha e shown poo
ans e abili y o olde coho s [
7
], making age-speci ic models such as ou s highly ele an
o imp o e demen ia p edic ion.
S eng hs and Limi a ions
The main highligh s o ou index a e ha i includes anxie y as a modi iable isk ac o
o AD and accoun s o compe ing isk o dea h o he i s ime. P e ious indices did no
con ol o mo ali y, and su i al bias may ha e a ec ed hei esul s.
Ou s udy includes ele an a iables de i ed om me a-analyses o coho s udies
epo ing isk ac o s o demen ia and AD [
25
]. Howe e , o he po en ial isk ac o s
o demen ia, such as physical inac i i y [
25
] o cogni i e engagemen [
61
], we e no
assessed in he ZARADEMP s udy and, he e o e, we e no included in ou index. The
a iable “li ing alone”, which has been shown o be associa ed wi h an inc eased isk o
demen ia [
62
], was no analyzed sepa a ely om ma i al s a us. The use o hea ing aids,
which ha e been shown o delay diagnosis o demen ia and AD in indi iduals wi h hea ing
loss [63], was also no speci ically collec ed in ou s udy.
The ac o s included in ou isk sco e a e based on esul s om a ep esen a i e, la ge
sample o he gene al popula ion olde han 65 yea s, excluding pa icipan s wi h demen ia
a baseline. Howe e , being based on a single coho limi s he gene alizabili y o ou
index [
14
]. The applicabili y could also be limi ed i using he index o p edic AD isk
in la e li e (olde han 65) o in a ela i ely sho e m (5 yea s). Ou model could no
be applied a ea lie ages and o longe ollow up, his is a limi a ion because he b ain
changes in demen ia can s a up o 10 yea s be o e he ini ial symp oms appea . Fu he
alida ion o he “ZARADEMP Alzheime Demen ia isk sco e” is equi ed.
5. Conclusions
The “ZARADEMP Alzheime Demen ia Risk Sco e” may inc ease ou unde s anding
o he weigh ha speci ic isk ac o s, mos o hem modi iable, ha e on AD bu den a he
popula ion le el. Ou isk sco e includes, o he i s ime, cu en , clinically signi ican
anxie y and he a iable hea ing loss. Mo eo e , i may p o ide a p ac ical ins umen
o iden i y subjec s a high isk in ou ine p ima y ca e p ac ice, and owa ds whom
p e en i e s a egies a ge ing he con ibu ing ac o s could be di ec ed.