Wange al. BMC Public Heal h (2024) 24:2239
h ps://doi.o g/10.1186/s12889-024-19734-w
RESEARCH Open Access
© The Au ho (s) 2024. Open Access This a icle is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional License, which
pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long as you gi e app op ia e c edi o he
o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons licence, and indica e i changes we e made. The images o
o he hi d pa y ma e ial in his a icle a e included in he a icle’s C ea i e Commons licence, unless indica ed o he wise in a c edi line
o he ma e ial. I ma e ial is no included in he a icle’s C ea i e Commons licence and you in ended use is no pe mi ed by s a u o y
egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om he copy igh holde . To iew a copy o his
licence, isi h p://c ea i ecommons.o g/licenses/by/4.0/. The C ea i e Commons Public Domain Dedica ion wai e (h p://c ea i ecom‑
mons.o g/publicdomain/ze o/1.0/) applies o he da a made a ailable in his a icle, unless o he wise s a ed in a c edi line o he da a.
BMC Public Heal h
Age a sepa a ion, esiden ial mobili y,
anddep essi e symp oms among wins
inla e adolescence andyoung adul hood:
aFinnTwin12 coho s udy
Zhiyang Wang1, Alyce M. Whipp1,2, Ma ja Heinonen‑Guzeje 1 and Jaakko Kap io1,2*
Abs ac
Backg ound Sepa a ing wi h close siblings and lea ing he pa en al home a an ea ly age ep esen s a majo li e
e en o an adolescen ( e lec ed by age a sepa a ion in a win pai ) and may p edispose hem o poo men al
heal h. This s udy aims o examine he associa ion o age a sepa a ion and esiden ial mobili y on dep essi e symp‑
oms in la e adolescence and young adul hood and o explo e possible unde lying gene ic e ec s.
Me hods Residen ial mobili y consis ed o he numbe and o al dis ance o mo es be o e age 17. Based on 3071
wins om he FinnTwin12 coho , we used linea eg ession o assess he associa ion o age a sepa a ion and esi‑
den ial mobili y wi h Gene al Beha io In en o y (GBI) sco es a age 17 and in young adul hood. A highe GBI sco e
indica ed mo e dep essi e symp oms occu ed. Then, he mixed model o epea ed measu es (MMRM) was used
o isualize he sco es’ ajec o y and es he associa ions, con olling o "baseline" s a e. Twin analyses wi h a bi a i‑
a e c oss‑lagged pa h model we e pe o med be ween he di e ence in GBI sco es, be ween co wins, and sepa a ion
s a us o he po en ial gene ic in luence.
Resul s Compa ed o wins sepa a ed be o e age 17, wins who sepa a ed la e had signi ican ly lowe GBI sco es
a age 17 and in young adul hood. In MMRM, sepa a ion a a la e age and a highe numbe o mo es we e associa ed
wi h a highe GBI sco e in young adul hood. A small gene ic e ec was de ec ed whe ein GBI wi hin‑pai di e ences
a age 17 we e associa ed wi h sepa a ion s a us be o e age 22 (coe icien : 0.01).
Conclusion The s udy p o ides alid e idence abou he in luence o siblings and amily on dep essi e symp oms
in la e adolescence and young adul hood while inding some e idence o a e e se di ec ion e ec . This sugges s
mo e cau ion in he in e p e a ion o esul s. A s ong associa ion be ween esiden ial mobili y and dep essi e symp‑
oms was a i med, al hough u he de ailed esea ch is needed.
Keywo ds Twins, Residen ial mobili y, Dep essi e symp om, Adolescen
Backg ound
The global bu den o majo dep essi e diso de (MDD)
is ge ing p og essi ely wo se since he las cen u y [1].
High-income a eas had a highe p e alence o MDD [2].
In 2019, acco ding o he Global Bu den o Diseases,
he age-s anda dized p e alence o dep essi e diso de
was 3851 pe 100,000 people in Wes e n Eu ope and was
*Co espondence:
Jaakko Kap io
jaakko.kap [email p o ec ed]
1 Ins i u e o Molecula Medicine Finland, Helsinki Ins i u e o Li e
Science, Uni e si y o Helsinki, PL 20 (Tukholmanka u 8), Helsinki FI‑00014,
Finland
2 Depa men o Public Heal h, Uni e si y o Helsinki, Helsinki, Finland
Page 2 o 12
Wange al. BMC Public Heal h (2024) 24:2239
highe han he global le el [3]. Da a o he gene al US
popula ion shows he p e alence o dep ession in ado-
lescen s and young adul s has been inc easing since 2005
[4, 5]. Mul iple isk ac o s o dep ession in childhood
and adolescence ha e been iden i ied such as low-quali y
educa ion, s ess ul li e e en s, and amily in luences [6].
Du ing childhood, mo ing o a new loca ion exposes
child en o g ea e social ne wo k loss han adul s and
hei pa en s a e less able o gi e hem he a en ion hey
need du ing he p ocess. Mo ing o a new loca ion has
been ega ded as a ype o nega i e li e expe ience ha
can be associa ed wi h la e dep ession [7, 8]. Residen ial
ins abili y o mobili y desc ibes hose child en who ha e
an uns able li e and a e mo e equen ly exposed o his
dele e ious expe ience. Nume ous s udies ha e sugges ed
ha high esiden ial ins abili y is linked o a la ge isk
o many psychological o psychia ic ou comes, including
dep ession [7, 9–11]. Howe e , he measu e o esiden ial
mobili y is o en limi ed by ecall bias om child en’s o
hei pa en s’ ques ionnai es [9]. Longi udinal esiden ial
eco ds om he na ional sys em could quan i y he
cumula i e e ec o mobili y o enhance he assessmen
o he s eng h o his ela ionship [11].
In la e adolescence and young adul hood, people will
g adually pa wi h hei siblings and mo e o a new
place due o a ious easons. Because o he simila age
and same gene a ion, siblings spend mo e ime oge he
and ha e mo e mu ual in luence han pa en s. Long-
e m daily companionship om siblings p omo es o a
s ong emo ional connec ion, which has a signi ican
impac on child en’s de elopmen [12]. Suppo i e and
bene icial sibling ela ionships, such as spending mo e
ime wi h each o he , appea s o be in e connec ed wi h
ewe dep essi e symp oms and highe well-being le els
[13]. Posi i e sibling bonding is also capable o elie ing
de imen al pa en al in luences [14]. Fo a ela i ely
ex eme example in China, AIDS o phans who we e
sepa a ed om hei siblings su e ed a highe le el o
anxie y, dep ession, ange , and dissocia ion [15]. As a
special case o siblings, in gene al, wins ha e a close
ela ionship and ecip oci y wi h each o he han non-
win siblings [16, 17]. The sepa a ion be ween co wins,
al e na i ely e med “ ela ional shi ”, b eaks he in ense
in imacy and dependency and can be accompanied by
emo ional u moil and con usion due o indi idua ion
[18]. Sepa a ion is an impo an challenge o wins
du ing childhood and adolescence.
Mo eo e , sepa a ing wi h siblings/ wins in adoles-
cence o young adul hood may imply lea ing om he
pa en al home. Anxie y d i en by ea ly sepa a ion om
pa en s has been connec ed o a se ies o nega i e men-
al heal h condi ions such as complica ed g ie , mo e
se e e symp oms o dep ession, o la e panic diso de
[19–21]. Pa en al suppo ac s like a “sa e y ne ” o
enhance he esilience in coping wi h challenges and o
sus ain he success ul ansi ion o he child o adul -
hood [22, 23]. Family closeness, suppo , a en i eness,
and o he posi i e ac o s a e associa ed wi h a la e im-
ing o lea ing, while eenage s li ing wi h s ep-pa en s
o incomple e amilies end o lea e ea ly due o a lack o
esou ces [24–26].
The age a sepa a ion be ween co wins e lec s he
leng h o ime ha wins s ay oge he and also p obably
wi h hei pa en s. I is likely associa ed wi h he amoun
o amily suppo , deg ee o men al ma u i y, and p epa-
a ion ime o wins o add ess he sepa a ion challenge,
connec ing o la e men al heal h. On he o he hand, we
also assume he ime o sepa a ion is when a leas one
o he co wins lea es he pa en al home. Unlike ques ions
abou he amily en i onmen in su eys, age a sepa a-
ion by esiden ial his o y in wins could be used as an
objec i e measu e o desc ibe he sepa a ion ime, as well
as amilial in luence, alidly wi h less in o ma ion bias.
The e o e, in his s udy, we hypo hesize ha olde age a
sepa a ion is a p o ec i e ac o and esiden ial mobil-
i y is a isk ac o and aim o de e mine: 1) he associa-
ion be ween age a sepa a ion and dep essi e symp oms
and he sex-speci ic e ec , 2) he associa ion be ween
esiden ial mobili y be o e age 17 and dep essi e symp-
oms in la e adolescence and young adul hood, and 3)
he gene ic in luence behind he associa ion be ween age
a sepa a ion and dep essi e symp oms, based on he
FinnTwin12 win coho .
Me hods
This is a longi udinal win coho s udy.
Pa icipan s
The s udy pa icipan s we e om he FinnTwin12 coho ,
which is a popula ion-based p ospec i e coho among all
Finnish wins bo n be ween 1983 and 1987. Pa icipan s
we e di ided in o wo pa s: he o e all coho and he
in ensi e s udy. A o al o 5184 ini ial wins en olled a
wa e one ( he o e all coho ) when hey eached age 11,
and wins illed in he gene al age 14 (wa e wo), age 17
(wa e h ee), and young adul (wa e ou ) ques ionnai es
wi h 92%, 75%, and 66% e en ion, espec i ely. In he
in ensi e s udy, 1035 amilies we e addi ionally in i ed
when he win was a age 14 (wa e wo) o psychia ic
in e iews, some biological samples, and addi ional
ques ionnai es, and 1854 wins pa icipa ed. They we e
in i ed back o new in e iews and o he measu emen s
as young adul s (wa e ou ). In e iews (n = 1347) in
young adul hood we e comple ed o 73% o wins in
he in ensi e s udy. The wins in he in ensi e s udy we e
mos ly andomly selec ed om he o e all epidemiology
Page 3 o 12
Wange al. BMC Public Heal h (2024) 24:2239
s udy, while en iched wi h wins a ele a ed amilial isk
o alcoholism [27]. The e was no e idence o selec ion
o amily ype, pa en al age, esidence, win sex, and
zygosi y a e ca e ully assessing he non- esponde s a
each s age [28]. The lowcha o he gene al FinnTwin12
coho is p esen ed in supplemen al Fig.1. The upda ed
e iew o his coho was published in 2019 [29].
Measu es
Age a sepa a ion
Age a sepa a ion was de ined as he age ha , o he i s
ime, he wins no longe esided a he same add ess,
wi h one o bo h wins mo ing o a new add ess as
desc ibed elsewhe e [30]. Reuni ica on migh occu
some imes a e sepa a ion bu was no conside ed in his
s udy. Age a sepa a ion was gene a ed om he esiden-
ial eco ds linking o he egis e o na ional Digi al and
Popula ion Da a Se ice Agency (DVV) in Finland since
bi h. The DVV, an o icial go e nmen agency, p omo es
digi aliza ion, secu es he a ailabili y o in o ma ion,
and p o ides a ious se ices including add ess se ices
and mo ing no ices. I upda es he esiden ial eco ds
apidly ensu ing accu acy. Sepa a ion indica ed ha he
DVV esiden ial eco ds di e ed be ween co wins. The
esiden ial eco ds included he in o ma ion on he no h
and eas coo dina es (EUREF-FIN), and he age mo ing
in and ou o each add ess. Then, we ca ego ized i in o
ou g oups: sepa a ed be o e age 17, 17 o < 19.5, 19.5
o < 22, and age 22 o olde . Subsequen ly, we c ea ed
wo bina y a iables o sepa a ion s a us (yes, no) be o e
ages 17 and 22 based on he age a sepa a ion o show
whe he wins sepa a ed a ha age. Twins wi hou alid
in o ma ion on age a sepa a ion (44 indi iduals) o win
pai s who ne e sepa a ed by he end o he ollow-up in
Decembe 2020 (13 pai s) we e excluded om he analy-
sis wi h age a sepa a ion.
Residen ial mobili y
Residen ial mobili y consis ed o wo a iables: he num-
be o mo es (ne e , once, wice, h ee o mo e imes)
and he o al dis ance o mo es (in kilome e s) be o e age
17 accumula ed ac oss all mo es. Bo h we e gene a ed
om he esiden ial eco ds. I wins did no ha e add ess
in o ma ion due o missing add esses o li ing ab oad o
any pe iod, hei o al dis ance o mo es was in alid and
excluded om he co esponding analysis (301 indi idual
wins). O e all, he mean o al dis ance o mo es be o e
age 17 was 48.4 (SD: 153.2) kilome e s, and he o al dis-
ance o mo es was dis ibu ed in a igh skewed ashion.
Dep essi e symp oms
We used he sho - e sion Gene al Beha io In en o y
(GBI) o e alua e he dep essi e symp oms among wins
[31]. I is a sel - epo ed in en o y designed o iden i y
mood- ela ed beha io s such as dep essi e and manic
symp oms [32]. The sho -scale e sion has 10 i ems and
uses a 4-poin Like scale om 0 (ne e ) o 3 ( e y o en)
o que y he occu ence o dep essi e symp oms. The
Fig. 1 T ajec o y o log‑ ans o med GBI sco e by he age a sepa a ion
Page 4 o 12
Wange al. BMC Public Heal h (2024) 24:2239
o al sco e anges om 0 o 30. This eliable measu e has
been used in many Finnish s udies including FinnTwin12
[31, 33, 34]. To alida e he GBI, we compa ed i o a
DSM-IV diagnosis o MDD assessed by he Semi-S uc-
u ed Assessmen o he Gene ics o Alcoholism om
he in ensi e s udy [35]. In a logis ic eg ession analysis
based on 1232 indi idual wins (mean age: 22.4), he GBI
sco e in young adul hood s ongly p edic ed MDD, wi h
an a ea unde he ecei e ope a ing cha ac e is ic cu e
(AUC) o 0.83. The GBI was assessed a age 17 and in
young adul hood. We also calcula ed he change o GBI
sco e be ween he wo s ages.
Ano he wo a iables o he wi hin-pai di e ences in
GBI sco e we e c ea ed a bo h s ages. The a iables we e
cons uc ed by sub ac ing he GBI sco e o one co win
om he sco e o he o he co win, and hen we ook he
absolu e alues. Highe alues indica ed a highe le el o
wi hin-pai di e ence.
Co a ia es
The e we e se en co a ia es de ined a p io i: sex (male,
emale), zygosi y (monozygo ic (MZ), dizygo ic (DZ),
unknown), smoking (ne e , qui , occasional, cu en )
in young adul hood, wo k s a us ( ull ime, pa - ime,
i egula , no wo king) in young adul hood, seconda y
le el school (senio high school, oca ional, none) in
young adul hood, pa e nal educa ion, ma e nal educa ion
and age when wins p o ided he GBI assessmen in
young adul hood. Sex was assigned a bi h because o
biological dimensions. Pa e nal and ma e nal educa ion
(missing, low, medium, high) was epo ed by pa en s
in wa e one. These co a ia es we e selec ed due o hei
ela ionship o ei he age a sepa a ion, men al heal h,
o bo h, acco ding o p e ious li e a u e [30, 36, 37].
Addi ionally, he mean age when wins p o ided he
GBI assessmen in young adul hood was 24.18 yea s
(SD: 1.68), which was only adjus ed o in he analysis
in ol ing GBI in young adul hood.
S a is ical analysis
A o al o 3071 indi idual wins was included in analyses
a inal, o whom 2769 p o ided da a a he age 17 su ey
and 3045 in young adul hood. Because o he di e en ial
missing in o ma ion pa e n, sample sizes in di e en
analyses we e sligh ly di e en .
Due o he skewness o he GBI sco es a bo h ages, we
added one o he GBI sco e and log- ans o med i . Then,
linea eg ession was used o examine he associa ion o
age a sepa a ion and esiden ial mobili y wi h log- ans-
o med GBI sco es a ages 17 and in young adul hood and
change o GBI sco e. The p e-speci ied co a ia es we e
adjus ed in he models. We used he uni a ia e analysis o
a iance o de e mine he s a is ical di e ences be ween
he mean GBI sco e o he g oups o ca ego ical co a i-
a es. Addi ionally, he sepa a ion be o e age 17 (no/no
in o ma ion/ ea ed apa , yes) was u he adjus ed o in
he analysis wi h esiden ial mobili y.
Fu he , since we had wo measu es o GBI sco e,
he mixed model o epea ed measu es (MMRM) was
chosen o isualize he ajec o y o GBI sco e by age a
sepa a ion o wins a age 17 and in young adul hood.
The MMRM me hod is a ype o di ec -likelihood
analysis and uses longi udinal da a o es ima e he
co a iance be ween da a om di e en ime poin s
[38]. This echnique is able o conside he in luence on
he ou come (GBI in young adul hood) by he exposu e
o in e es ( ixed e ec ) and “baseline” GBI a age 17
( andom e ec ) [39]. I also allowed us o handle he
une enness o missing in o ma ion a di e en ages
o inc ease he powe . Co a ia es we e adjus ed o .
S a i ica ion analyses by sex we e also pe o med, gi en
ha dep ession is mo e common in emales han in males
a e pube y.
To e alua e he po en ial gene ic in luence be ween
sepa a ion and dep essi e symp oms, in aclass co ela-
ion coe icien s ( ) we e calcula ed a i s , as an explo a-
ion. This was done o decompose o e all a iance in o
he gene ic a iances: addi i e gene ic (A) and dominan
gene ic (D) componen s and en i onmen al a iances:
common en i onmen al (C), and unique en i onmen al
(E) componen s o dep essi e symp oms a bo h ages p e-
limina ily. Based on he di e ence o be ween he MZ and
DZ win pai s, he mos app op ia e model (ACE o ADE)
would be selec ed, and hen we calcula ed a iance compo-
nen s based on in aclass co ela ions [40]. Di e ences in
be ween zygosi y indica ed he di e en ial con ibu ion
o gene ic and en i onmen al a iances o he o e all a i-
ance in GBI sco es, and s a i ica ion by age a sepa a ion
showed i s e ec s, such as modi ica ion, on he gene ic and
en i onmen al a iances. Fu he mo e, a bi a ia e c oss-
lagged pa h model was i ed wi h sepa a ion s a us and
wi hin-pai di e ences, simila i y, in GBI sco e a bo h ages
wi h he maximum likelihood es ima o ( win pai s we e
analyzed as he uni o obse a ion). This gauged he s uc-
u al ela ionships o epea edly measu ed a iables and
es ed bo h di ec ions be ween di e ences and sepa a ion,
which would could sugges a po en ial ole o gene ics [41].
I wi hin-pai di e ence (ins ead o a he indi idual le el)
leads o sepa a ion, gene ic e ec s a e sugges ed, while i
sepa a ion leads o wi hin-pai di e ence, we should con-
side ed he mo e en i onmen al oles on he pheno ype.
Th ee co a ia es we e adjus ed o : sex and zygosi y com-
bina ion (male monozygo ic (MMZ), emale monozygo ic
(FMZ), male dizygo ic (MDZ), emale dizygo ic (FDZ),
and opposi e-sex dizygo ic (OSDZ)), pa e nal educa ion,
and ma e nal educa ion. Twin pai s whose zygosi y was
Page 5 o 12
Wange al. BMC Public Heal h (2024) 24:2239
unknown o who had missing in o ma ion on sepa a ion
s a us and wi hin-pai di e ences in GBI sco e a bo h ages
we e excluded om he c oss-lagged pa h model.
Due o he possibili y o e e se causali y, whe ein
men al heal h condi ions migh in luence wins o sepa-
a e o lea e home, we conduc ed a sensi i i y linea
eg ession o examine he associa ion o age a sepa a ion
wi h se e al men al heal h condi ions a age 14, adjus ing
o sex, zygosi y, pa e nal educa ion, and ma e nal edu-
ca ion. The e we e i e subscales –– dep ession, social
anxie y, hype ac i i y, agg ession, and ina en ion ––
om he Mul idimensional Pee Nomina ion In en o y
(MPNI) was used o assess wins’ men al heal h a age
14. Each subscale con ained a a ying numbe o i ems
(4-poin Like scale), and we used he mean o i em
esponses as he ou come. The MPNI scale has been ali-
da ed and showed p edic i e powe in hep e ious s udy
[27]. Due o di e en ins umen s o measu e dep essi e
symp oms be ween di e en ages, dep essi e symp oms
om he MPNI a age 14 we e no added o he MMRM.
Then, we conduc ed ano he sensi i i y analysis based
on he linea eg ession o he po en ial con ounding by
men al heal h a age 14. The dep ession om MPNI was
addi ionally adjus ed as a co a ia e.
The i s wo hypo heses in he s udy we e es ed in he
same ypes o models, bu he age a sepa a ion and esi-
den ial mobili y we e es ed independen ly in models as
he exposu e o in e es . Addi ionally, we con olled o
he clus e e ec by win pai in he eg ession analyses
h ough “ obus ” s anda d e o s. Reg ession coe icien s
and 95% con idence in e als (CIs) we e epo ed. A wo-
ailed P < 0.05 was conside ed s a is ically signi ican .
All s a is ical analyses we e pe o med using S a a 17.0
(S a aCo p, College S a ion, TX, USA) and Mplus 8.6
(Mu hen & Mu hen, Los Angeles, CA, USA). The code
used in his esea ch can be ound a Gi hub (h ps://
gi hub. com/ doge73/ age- a - sepa a ion- GBI).
Resul s
Demog aphic cha ac e is ics
A o al o 3071 wins we e included in he inal analysis
including 1284 win pai s wi h bo h wins included and
ano he 503 indi idual wins (i.e. hei co win had no
pa icipa ed o o he wise excluded as de ined abo e).
As shown in Table1, he mean sco e o GBI assess-
men a age 17 was 5.1, and in young adul hood was 4.4.
The majo i y o wins we e emale (56.4%), dizygo ic
(60.6%), and epo ed ne e smoking in young adul -
hood (53.5%). Fo socioeconomic s a us, wo k s a us,
seconda y le el school, ma e nal educa ion had signi i-
can be ween-g oup di e ences in ei he o bo h GBI
sco es a age 17 and young adul hood. Supplemen al
Table1 shows he compa ison o some cha ac e is ics
be ween indi idual wins included in he analyses and
was a ailable a baseline (n = 5104, consen ed o pa -
icipa e in FinnTwin12 cu en ly), and a simila dis i-
bu ion in bo h g oups sugges ed a low isk o selec ion
bias.
Table 1 Demog aphic, socioeconomic cha ac e is ics, and
dep essi e symp oms o indi idual wins
a P alue was calcula ed om he analysis o a iance
b 302 indi idual wins we e missing because hey did no p o ide GBI
assessmen a age 17
c 26 indi idual wins we e missing because hey did no p o ide GBI assessmen
in young adul hood
Cha ac e is ics Indi idual n. (%) Mean (SD) o P- aluea
GBI a age 17bGBI in
young
adul hoodc
O e all 3071 5.1 (4.9) 4.4 (4.68)
Sex < 0.0001 < 0.0001
Male 1339 (43.6) 3.7 (3.9) 3.6 (4.3)
Female 1732 (56.4) 6.2 (5.2) 5.0 (4.9)
Zygosi y 0.00 0.06
Monozygo ic 1066 (34.7) 4.6 (4.6) 4.1 (4.5)
Dizygo ic 1861 (60.6) 5.3 (5.0) 4.5 (4.7)
Unknown 144 (4.7) 5.7 (5.2) 4.8 (5.6)
Smoking < 0.0001 < 0.0001
Ne e 1642 (53.5) 4.4 (4.4) 3.7 (4.1)
Qui 340 (11.1) 6.2 (5.4) 4.7 (5.0)
Occasional 309 (10.1) 5.7 (5.0) 4.5 (4.7)
Cu en 780 (25.4) 5.9 (5.3) 5.7 (5.4)
Wo k 0.11 < 0.0001
Full‑ ime wo k 1582 (51.5) 4.9 (4.8) 4.0 (4.3)
Pa ‑ ime wo k 394 (12.8) 5.1 (4.4) 4.5 (4.3)
I egual wo k 372 (12.1) 5.1 (4.6) 4.4 (4.4)
No wo king 723 (23.5) 5.5 (5.2) 5.3 (5.7)
Seconda y le el
school < 0.0001 < 0.0001
Senio high school 1850 (60.2) 4.9 (4.6) 4.1 (4.4)
Voca ional 1042 (33.9) 5.0 (4.8) 4.4 (4.6)
None 179 (5.8) 7.9 (7.2) 7.7 (6.7)
Pa e nal educa ion 0.06 0.39
Missing 496 (16.2) 5.5 (5.4) 4.6 (4.9)
Low 678 (22.1) 5.4 (5.0) 4.5 (4.8)
Medium 1496 (48.7) 4.9 (4.7) 4.3 (4.5)
High 401 (13.1) 5.0 (4.7) 4.5 (4.7)
Ma e nal educa ion 0.00 0.03
Missing 225 (7.3) 6.1 (5.3) 5.4 (6.2)
Low 517 (16.8) 5.7 (5.3) 4.8 (4.9)
Medium 1863 (60.7) 5.0 (4.7) 4.3 (4.5)
High 466 (15.2) 4.7 (4.7) 4.4 (4.8)
Page 6 o 12
Wange al. BMC Public Heal h (2024) 24:2239
Associa ion o age a sepa a ion wi hGBI
The sample sizes o each linea eg ession analysis a e
p esen ed in Table 2. A e adjus men o co a ia es,
compa ed o wins sepa a ed be o e age 17, he e we e
signi ican lowe log- ans o med GBI sco es a age 17 in
all h ee g oups o wins, hose who sepa a ed be ween
ages 17 and 19.5 (coe icien : -0.16, 95% CI: -0.31, -0.02),
be ween ages 19.5 and 22 (coe icien : -0.22, 95% CI:
-0.37, -0.08), and a e age 22 (coe icien : -0.44, 95% CI:
-0.60, -0.28). Mo eo e , wins who sepa a ed be ween
ages 19.5 and 22 and a e age 22 also had lowe log-
ans o med GBI sco es in young adul hood (coe icien :
-0.19, 95% CI: -0.37, -0.02 and coe icien : -0.25, 95% CI:
-0.44, -0.06, espec i ely), compa ed o wins sepa a ed
be o e age 17.
Figu e1 depic s he ajec o y o he p edic i e ma -
ginal mean o log- ans o med GBI sco e om age 17
o young adul hood. The h ee age-a -sepa a ion g oups
be o e age 22 dec eased in hei p edic i e ma ginal
mean o log- ans o med GBI sco es, bu he la es
age a sepa a ion g oup seems no o change no ably.
The esul s o he MMRM a e p esen ed in Table3 and
included 2712 wins a age 17 and 2962 wins in young
adul hood. A e con olling o log- ans o med GBI
sco e a age 17 and p e-speci ied co a ia es, wins who
sepa a ed be ween ages 17 and 19.5 (coe icien : -0.19,
95% CI: -0.34, -0.04), be ween ages 19.5 and 22 (coe i-
cien : -0.21, 95% CI: -0.36, -0.05), and a e age 22 (coe -
icien : -0.26, 95% CI: -0.42, -0.09) sco ed signi ican ly
lowe on he GBI (log- ans o med) in young adul hood
han hose who sepa a ed be o e age 17.
Then we pe o med sex-s a i ied MMRM o age a
sepa a ion (Supplemen al Table 2). Rega dless o he
age a sepa a ion, males’ mean GBI sco es we e no ably
lowe han hose o emales. In emales, a e con olling
o log- ans o med GBI sco e a age 17 and p e-
speci ied co a ia es, wins who sepa a ed a e age 22 had
lowe log- ans o med GBI sco es in young adul hood
(coe icien : -0.29, 95% CI: -0.52, -0.06), compa ed o
wins sepa a ed be o e age 17.
Table 2 Associa ion o age a sepa a ion and esiden ial mobili y wi h dep essi e symp oms (GBI) using linea eg ession
a Adjus ed o sex, zygosi y, smoking, wo k s a us, seconda y le el school, pa e nal educa ion, and ma e nal educa ion; o numbe o mo es and o al dis ance o
mo ing be o e 17, u he adjus ed o sepa a ion be o e age 17
b Adjus ed o sex, zygosi y, smoking, wo k s a us, seconda y le el school, pa e nal educa ion, ma e nal educa ion, and age when wins p o ided he GBI assessmen
in young adul hood; o numbe o mo es and o al dis ance o mo ing be o e 17, u he adjus ed o sepa a ion be o e age 17
* P < 0.05
Cha ac e is ics log- ans o med GBI sco e a age 17 log- ans o med GBI sco e in young
adul hood Change o GBI
Indi idual
n. (%) Unadjus ed
coe icien
(95% CI)
Adjus ed
coe icien
(95% CI)a
Indi idual
n. (%) Unadjus ed
coe icien
(95% CI)
Adjus ed
coe icien
(95% CI)b
Indi idual
n. (%) Unadjus ed
coe icien
(95% CI)
Adjus ed
coe icien
(95% CI)b
Age a
sepa a ion 2725 2976 2700
ze o o less han
17 110 (4.04) Re Re 131 (4.40) Re Re 107 (3.96) Re Re
17 o less han
19.5 967 (35.49) ‑0.18 (‑0.34,
‑0.01)* ‑0.16 (‑0.31,
‑0.02)* 1078 (36.22) ‑0.21 (‑0.40,
‑0.02)* ‑0.17 (‑0.35,
0.00) 959 (35.52) ‑0.41 (‑1.41,
0.60) ‑0.22 (‑1.19,
0.75)
19.5 o less han
22 1146 (42.06) ‑0.34 (‑0.50,
‑0.18)* ‑0.22 (‑0.37,
‑0.08)* 1228 (41.26) ‑0.31 (‑0.50,
‑0.13)* ‑0.19 (‑0.37,
‑0.02)* 1133 (41.96) ‑0.05 (‑1.03,
0.94) 0.04 (‑0.93,
1.00)
22 o mo e 502 (18.42) ‑0.62 (‑0.79,
‑0.44)* ‑0.44 (‑0.60,
‑0.28)* 539 (18.11) ‑0.43 (‑0.63,
‑0.23)* ‑0.25 (‑0.44,
‑0.06)* 501 (18.56) 0.65 (‑0.36,
1.66) 0.67 (‑0.34,
1.67)
Numbe o
mo es
be o e age 17
2769 3045 2743
None 736 (26.58) Re Re 776 (25.48) Re Re 728 (26.54) Re Re
Once 733 (26.47) 0.05 (‑0.05,
0.15) 0.05 (‑0.04,
0.13) 809 (26.57) 0.06 (‑0.03,
0.16) 0.06 (‑0.03,
0.15) 726 (26.47) ‑0.10 (‑0.58,
0.38) ‑0.07 (‑0.55,
0.40)
Twice 549 (19.83) 0.01 (‑0.09,
0.12) ‑0.00 (‑0.10,
0.10) 602 (19.77) 0.12 (0.01,
0.23)* 0.09 (‑0.01,
0.19) 544 (19.83) 0.56 (0.04,
1.08)* 0.52 (‑0.00,
1.04)
Th ee
imes o mo e 751 (27.12) 0.07 (‑0.03,
0.16) 0.00 (‑0.09,
0.10) 858 (28.18) 0.15 (0.06,
0.25)* 0.08 (‑0.01,
0.18) 745 (27.16) 0.37 (‑0.16,
0.89) 0.40 (‑0.15,
0.94)
To al dis ance
o mo ing (pe
100km) be o e
age 17
2505 0.01 (‑0.02,
0.03) 0.01 (‑0.02,
0.03) 2747 0.01 (‑0.02,
0.03) 0.01 (‑0.02,
0.03) 2482 0.00 (‑0.12,
0.12) 0.00 (‑0.12,
0.12)
Page 7 o 12
Wange al. BMC Public Heal h (2024) 24:2239
The sensi i i y analysis e ealed signi ican co ela-
ions o age a sepa a ion wi h dep ession, hype ac i -
i y, and ina en ion a age 14 (Supplemen al Table3),
sugges ing he p esence o e e se causali y. A e u -
he adjus ing dep ession a age 14, compa ed o hose
who sepa a ed be o e age 17, wins who sepa a ed
a e age 22 had lowe log- ans o med GBI sco es a
age 17 and in young adul hood (Supplemen al Table4).
Besides, wins who sepa a ed be ween age 17 and 19.5
also had lowe log- ans o med GBI sco es in young
adul hood. A e u he adjus ing dep ession a age
14, MMRM showed ha wins who sepa a ed be ween
ages 17 and 19.5 (coe icien : -0.22, 95% CI: -0.42,
-0.02), and a e age 22 (coe icien : -0.27, 95% CI:
-0.50, -0.05) had lowe log- ans o med GBI sco es in
young adul hood han hose who sepa a ed be o e age
17 (Supplemen al Table5).
Associa ion o esiden ial mobili y wi hGBI
In he linea eg ession, a e adjus men o co a ia es,
we did no obse e any signi ican associa ion o he
numbe and o al dis ance o mo es be o e age 17
wi h any o he GBI a iables (Table2). Supplemen al
Fig.2 depic s he ajec o y o he p edic i e ma ginal
mean o log- ans o med GBI sco e om age 17 o
young adul hood by he numbe o mo es be o e age
17. The p edic i e ma ginal means dec eased in all
numbe so mo es g oups and educ ions we e mo e
subs an ial in wins who ne e mo ed o mo ed once
han in o he g oups. In MMRM, a e con olling o
baseline alues and co a ia es, wins who mo ed wice
o mo ed h ee o mo e imes had highe GBI (log-
ans o med) sco es in young adul hood han hose
who ne e mo ed (coe icien : 0.10, 95% CI: 0.01, 0.19
and coe icien : 0.09, 95% CI: 0.00, 0.17, espec i ely)
(Supplemen al Table 6). A e u he adjus ing
dep ession a age 14, associa ions a enua ed o null
(Supplemen al Table4).
Rela ionship be weensepa a ion andwi hin-pai
di e ence
The o e all in aclass co ela ion coe icien s o GBI
sco es among MZ and DZ pai s we e 0.56 and 0.14 a
age 17 and 0.52 and 0.22 in young adul hood, espec-
i ely (Supplemen al Table 7). The ADE model was
selec ed (as MZ > 2 DZ). A age 17, he gene ic (A + D)
and en i onmen al (E) a iances we e 0.56 and 0.44,
espec i ely. In young adul hood, he gene ic and en i-
onmen al a iances we e 0.52 and 0.48, espec i ely.
No ably, o he GBI sco e in young adul hood, coe i-
cien s did no di e be ween MZs (0.55) compa ed o
DZs (0.53) in win pai s who sepa a ed be o e age 17.
Supplemen al Table 8 p esen s he dis ibu ion o
1280 win pai s, including in he c oss-lagged pa h
model, o e he sepa a ion s a us and demog aphic
cha ac e is ics. The wi hin-pai di e ence in GBI
sco es dec eased om age 17 o age 22 o e all and in
almos e e y subg oup, excep MMZ and FDZ. Wi hin-
pai di e ences in GBI a age 17 we e signi ican ly asso-
cia ed wi h sepa a ion s a us be o e age 22 (coe icien :
0.01), wi h a small e ec size. Mo eo e , he e we e
o he signi ican associa ions be ween he wi hin-pai
di e ences in GBI sco e a ages 17 and in young adul -
hood (coe icien : 0.22) and be ween sepa a ion s a us
be o e age 17 and age 22 (coe icien : 0.19) (Supplemen-
al Fig.3).
Discussion
Wi h a o al o 3071 wins om he FinnTwin12 coho
included in his s udy, we ound ha la e sepa a ion in
win pai s was associa ed wi h ewe dep essi e symp-
oms in la e adolescence and young adul hood. In
addi ion, highe esiden ial mobili y was linked o mo e
dep essi e symp oms in young adul hood. Al hough
emales had mo e dep essi e symp oms han males,
he e we e no ob ious sex di e ences in he associa ion
be ween age a sepa a ion and dep essi e symp oms.
Table 3 Associa ion o age a sepa a ion wi h dep essi e symp oms (GBI) using MMRM
a Adjus ed o sex, zygosi y, smoking, wo k s a us, seconda y le el school, pa e nal educa ion, ma e nal educa ion, and age when wins p o ided he GBI assessmen
in young adul hood
* P < 0.05
Age a sepa a ion Mean (SD) Adjus ed coe icien (95% CI)a
GBI a age 17
(indi idual n. = 2725) GB1 in young adul hood
(indi idual n. = 2976) Log- ans o med GBI sco e
in young adul hood
ze o o less han 17 6.77 (5.02) 6.20 (5.81) Re
17 o less han 19.5 5.91 (5.28) 4.81 (4.89) ‑0.19 (‑0.34, ‑0.04)*
19.5 o less han 22 4.91 (4.63) 4.17 (4.37) ‑0.21 (‑0.36, ‑0.05)*
22 o mo e 3.65 (3.98) 3.79 (4.50) ‑0.26 (‑0.42, ‑0.09)*
Page 8 o 12
Wange al. BMC Public Heal h (2024) 24:2239
Howe e , his assessmen was shadowed by e e se
causali y o some ex en , and in e p e a ion and ans-
la ion should be mo e cau ious. Fu he mo e, we ound
a small gene ic in luence on he sepa a ion and ai ly
mino and inconsis en e ec s o sepa a ion on he di -
e ence be ween co wins, indica ing ha age a sepa-
a ion is no a majo modi ie o sibling di e ence in
dep essi e symp oms.
Pa ing om he amily o o igin is conside ed asa hall-
ma k o becoming an adul , in which he you h is sepa-
a ed om hei pa en s and siblings, loses hei o iginal
amilial ne wo k, and en e s a comple ely new li ing en i-
onmen . The majo i y o he age a sepa a ion o he
wins in his s udy was be ween ages 17 and 22, which
also implied li e ansi ions such as en e ing uni e si y o
some ex en . The ansi ion una oidably leads o po en-
ial nega i e o posi i e men al heal h e ec s, which a e
complica ed by sibling and amilial con ex s. Indi iduals
wi h wa m and encou aging siblings ela ionships may
eel small loss, while siblings wi h ela ionships ma ked by
con lic may eel mu ual elie a e sepa a ion [42]. Sei e-
K enke ound ha he a e o amily con lic was highe in
adolescen s who le home a he age o 21 and 23yea s
o emales and males, espec i ely, han hose s ill in
he nes , and he in- ime lea e pa e n co ela ed wi h a
wo se le el o psychological heal h [43]. Highe indi id-
ual and amily income we e associa ed wi h highe odds
o home lea ing [44, 45], and lowe inancial s ess was
mu ually co ela ed wi h ewe dep essi e symp oms a
he s age o eme ging adul hood [46]. Wi h ou esul s, we
obse ed ha longe o mo e amily suppo , indica ed by
an olde age a sepa a ion, as well as mo e mu ual suppo
by co wins, could help adolescen s cope wi h he men al
heal h bu den o sepa a ion om pa en s and siblings
(same-aged win sibling in ou case) and lea ing home.
Sepa a ion is ega ded as a pa o he social/ amily expo-
some [47]. Ex ended amily suppo including mo e ela-
i es has been associa ed wi h lowe a es o MDD and
ewe dep essi e symp oms [48, 49]. Longe and mo e
suppo could be ex apola ed o be e suppo . Be e
amily in ol emen and sibling ela ionships including
wa m h, a en ion, and p aise om pa en s p edic ed
less MDD in adul hood and a e aspec s o co espond-
ing in e en ions [50–52]. Familial cohesion (emo ional
bonding) and sa is ac o y social suppo we e also
shown o be help ul in educing dep essi e symp oms
in lowe -middle-class communi ies in he US [53]. A
a o able amily backg ound and ela ionships could be
p o ec i e ac o s o suppo ea ly adul hood ansi ion,
and s able ansi ions we e associa ed wi h ewe dep es-
si e symp oms [54].
Gi en he e idence o a e e se e ec o men al heal h
on age a sepa a ion in his s udy, we u he p opose a
mo e dynamic e ec be ween amily, sepa a ion, and
men al heal h. Fo example, a p e ious s udy showed, a
la e age o lea ing was associa ed wi h close esidence
o aging pa en s and mo e equen con ac , compa ed
wi h siblings who mo ed ou on ime, depic ing a posi-
i e e ec om ex ended co esidence [55]. A UK longi-
udinal s udy sugges ed ha pa en s’ ela ionship quali y
and ea ly childhood beha io al di icul ies we e ecip o-
cally linked [56]. Addi ionally, he e was a bidi ec ional
in luence o pa en ing beha io and conduc p oblems
om childhood o adolescence among boys shown in a
US s udy [57]. Fu he mo e, S ice and Ba e a ound a
ull ecip ocal ela ionship be ween adolescen subs ance
use, bu no ex e nalizing, and amily suppo and con ol
[58], while null e idence also exis s [59]. On he o he
hand, age a sepa a ion e lec ed he sha ed en i onmen
be ween co wins ins ead o he sepa a ion e en i sel ,
hus, o some deg ee, i also included he amilial en i-
onmen induced by wins’ men al heal h. This means
mo e complex in e ac ions be ween indi iduals, siblings,
and amilies. Fu u e s udies equi e mo e sophis ica ed
models and a la ge numbe o amilial a iables, and a
li ecou se pe spec i e is ecommended.
Residen ial mobili y in ol es he cons an change o
he buil en i onmen and could be ega ded as a pa o
he physical exposome. Ins able esidence educes amily
in e ac ion, agmen s educa ion, cu s social connec ions,
and leads o emo ional s ess [60]. A solid body o
e idence has been es ablished ha a s ong associa ion
be ween highe childhood esiden ial mobili y and poo e
men al heal h a la e ages ac oss coun ies [7, 9, 11]. One
s udy in No h I eland used census-based eco d linkage
[61], while ou indings could s eng hen his associa ion
h ough longi udinal analysis wi h apid upda es. Solís
e al. de eloped he biological plausibili y ha b oad ea ly
nega i e expe ience inc eased physiological wea -and-
ea , measu ed by allos a ic load, which la gely explained
he heal h beha io s and educa ion le el in la e ages
[62]. Ne e heless, we should be awa e ha he e ec
could be complica ed by child en’s emo ionali y, amilial
con ex , and eason o mo es. Ou null esul s ega ding
he dis ance o mo es may be due o he e ogeneous
mo i a ion, whe ein a sho e dis ance could help o
main ain he social ne wo k and longe dis ances could
be ela ed o wo k eposi ioning [7]. Physical mo ing
i sel does no necessa ily co ela e wi h he sepa a ion o
amily, and company wi h amily is pe sis ing. In a sample
o 70 child en om u al a eas o Geo gia, US., child en
wi h in ense emo ions had a su p ising educ ion
in dep essi e symp oms as a unc ion o inc eased
esiden ial mobili y, while child en wi h non-in ense
emo ions expe ienced he opposi e [10]. Addi ionally,
amilial suppo was able o bu e he ha m ul e ec o
Page 9 o 12
Wange al. BMC Public Heal h (2024) 24:2239
mobili y on child en’s educa ion and ca ee achie emen
[63]. In ou s udy, we had o exclude wins who mo ed
ab oad a leas once, limi ing ou analysis by selec ion
bias o some ex en . Mul iple s udies ha e shown ha
non-Wes e n immig an adolescen s ha e a highe isk
o men al heal h p oblems compa ed o na i e Eu opean
adolescen s [64, 65].
Ex ensi e esea ch has in es iga ed he di ec ion
be ween simila i y and con ac , and i seems ha bo h
di ec ions exis , bu he magni ude a ies by age, sex, and
ai s [41, 66–68]. Sepa a ion is a u ning poin o con ac ,
since he deg ee o con ac ine i ably dec eases a e
sepa a ion. Based on he small e ec size in he c oss-
lagged longi udinal analyses, he gene ic in luence on he
sepa a ion was minimally obse ed. Age a sepa a ion
migh modi y he simila i y o dep essi e symp oms o
a mino deg ee. The in aclass co ela ion coe icien s
we e almos always highe in MZs han in DZs a all ages
a sepa a ion in ou s udy, bu he di e ence dec eased
in young adul hood and wi h la e sepa a ion, showing
a sligh ly dec easing gene ic e ec . Be ween ages 17
and 22, he en i onmen al a iance inc eased, which
was consis en wi h inc easing E wi h age epo ed in a
p e ious s udy [69]. This was also co obo a ed by he
nea ly iden ical coe icien s in he ea lies sepa a ion
g oup be ween MZ and DZ in young adul hood.
Mo eo e , as abo emen ioned, age a sepa a ion was
e lec ed as a p oxy o he en i e amilial exposome.
Gi en he he i abili y o 37% o dep essi e symp oms
in he p e ious analysis [70], he e was as oom o
en i onmen al e ec s including he amilial exposome,
and ou esul s sugges ed ha genes play a ole in he
in e ac ion o en i onmen al e ec s wi h dep essi e
symp om. A p e ious win s udy in he UK highligh ed
he gene ic in luence in shaping a dep essogenic
en i onmen in middle childhood as a gene–en i onmen
co ela ion [71]. Th ough DNA me hyla ion, Ca e as-
Gallo e al. illus a ed a gene–en i onmen in e ac ion
mechanism h ough DNA me hyla ion, elucida ing i s
connec ion wi h he ea ly-li e exposome [72]. Thus,
ins ead o implying any causali y, ou indings on he
ole o gene ics sugges ed ha age a sepa a ion should
be conside ed in e ining he assessmen o bo h gene ic
and en i onmen al in luences, e ealing he in ica e
ela ionships in u u e s udies.
Ou s udy was s eng hened by he objec i e meas-
u e o he leng h ha wins s ayed in hei pa en al
home by age a sepa a ion h ough he na ional egis y,
which helps o a oid ecall bias and p o ide mo e accu-
a e in o ma ion such as he da e o mo es. Mo eo e ,
in ou p e ious s udy, he sensi i i y and speci ici y o
sel - epo ed sepa a ion s a us we e no ideal compa ed
o he age a sepa a ion p o ided h ough he na ional
sys em [30]. In addi ion, he wo a iables on esiden ial
mobili y helped us quan i y i s cumula i e and longi u-
dinal e ec on he de elopmen o wins. Fu he mo e,
he analysis, including epea ed measu es, could cha ac-
e ize he c i ical phase o dep essi e symp om de elop-
men om la e adolescence o young adul hood.
The e we e also se e al limi a ions in his s udy.
Fi s , he sensi i i y analysis e ealed he p obabili y
o e e se causali y, sugges ing a complex ela ionship
be ween sepa a ion and men al heal h, especially con-
side ing he ole o gene ics and complex in e pe sonal
ne wo k. Al hough, a e adjus ing o mid-adolescence
dep ession, ou main inding abou age a sepa a ion
was s ill alid, we should be cau ious in in e p e ing
and ansla ing ou indings. Second, he 10-i em GBI
was a non-diagnos ic sel - epo measu e o dep es-
si e symp oms. Howe e , p e ious s udies ha e shown
high eliabili y and alidi y o he ull-scale GBI om
which his sho e e sion is de i ed [73]. As we had
in e iew-based diagnos ic in o ma ion on MDD on
he same pa icipan s, we could show a high alidi y
o his sho GBI. Thi d, he win s udy may lead o he
conce n o low gene alizabili y o ep esen a i eness.
Mos ai s in adul wins including adap i e beha io ,
dep ession, and anxie y did no di e om single ons,
which we e sugges ed o be gene alizable [74–76].
Fou h, when Finnish males a e doing hei manda o y
mili a y se ice, hei esiden ial eco ds do no change
o hei se ice loca ion, which may lead o in o ma-
ion bias. Addi ionally, mili a y se ice could possibly
a ec men al heal h [77], which sugges s po en ial oles
as a con ounde o modi ie . Compa ed o many o he
coun ies, he du a ion o Finnish mili a y se ice is
ela i ely sho , bu i is obliga o y o all men. Co wins
usually s a simul aneously and can spend weekends
and holidays oge he a home. We belie e ha he
in luence o mili a y se ice was nei he andom no
e y subs an ial, and pe o med he sex-s a i ica ion
analysis o in es iga e his u he , bu he e was no
ob ious sex-speci ic e ec . Fi h, some po en ial con-
ounde s we e no included due o a ailabili y such as
amily weal h o pe sonali y. Physical en i onmen al
componen s, such as g eene y, as a e lec ion o social
seg ega ion, du ing adolescence should be also consid-
e ed in a li ecou se pe spec i e [78]. Al hough pa en ’s
educa ion could e lec he gene al amily condi ion o
some ex en , u u e access o Finnish social and edu-
ca ional egis e s may help o add ess his conce n.
Finally, he o e lap be ween pe sonali y, pe haps some
o he pheno ypes, and dep essi e symp oms may in e -
e e wi h he es ima ion [79], and we would like o u -
he disen angle hei complica ed ela ionships wi h
sepa a ion in he u u e.
