1
Does High Op imism P o ec Agains he In e -Gene a ional T ansmission o
High BMI? The Ca dio ascula Risk in Young Finns S udy
Se lachius, Anna; Pulkki-Raback, Lau a; Juonala, Ma kus; Sabin, Ma hew; Leh imäki, Te ho;
Rai aka i, Olli; Elo ainio, Ma ko
2
Abs ac
Objec i e: The ansmission o o e weigh om one gene a ion o he nex is well es ablished,
howe e li le is known abou wha psychosocial ac o s may p o ec agains his amilial isk.
The aim o his s udy was o examine whe he op imism plays a ole in he in e gene a ional
ansmission o obesi y.
Me hods: Ou sample included 1043 pa icipan s om he p ospec i e Ca dio ascula Risk in
Young Finns S udy. Op imism was measu ed in ea ly adul hood (2001) when he coho we e
aged 24-39 yea s. BMI was measu ed in 2001 (baseline) and 2012 when hey we e aged 35-50
yea s. Pa en al BMI was measu ed in 1980. Hie a chical linea eg ession and logis ic
eg ession we e used o examine he associa ion be ween op imism and u u e BMI/obesi y,
and whe he an in e ac ion exis ed be ween op imism and pa en al BMI when p edic ing
BMI/obesi y 11 yea s la e .
Resul s: High op imism in young adul hood demons a ed a nega i e ela ionship wi h high
BMI in mid-adul hood, bu only in women (β= = -0.127, p=0.001). The op imism × ma e nal
BMI in e ac ion e m was a signi ican p edic o o u u e BMI in women (β = -0.588, p=0.036).
The logis ic eg ession esul s con i med ha high op imism p edic ed educed obesi y in
women (OR=0.68, 95% CI, 0.55-0.86), howe e he op imism × ma e nal obesi y in e ac ion
e m was no a signi ican p edic o (OR=0.50, 95% CI, 0.10-2.48).
Conclusions: Ou indings suppo ed ou hypo hesis ha high op imism mi iga ed he
in e gene a ional ansmission o high BMI, bu only in women. These indings also p o ided
e idence ha posi i e psychosocial ac o s such as op imism a e associa ed wi h long- e m
p o ec i e e ec s on BMI in women.
Keywo ds: op imism, BMI, obesi y
3
In oduc ion
Pa en al o e weigh and obesi y a e well-known isk ac o s o o sp ing obesi y, ye ew
s udies ha e examined p o ec i e psychosocial ac o s ha may mi iga e he sp ead o obesi y
om one gene a ion o he nex . Psychosocial ac o s which a e likely implica ed in he
in e gene a ional ansmission o o e weigh include childhood socioeconomic s a us (SES),
dep ession, psychosocial s ess, and social suppo (1-4). Ano he p o ec i e psychosocial
ac o , which may play a simila ole, is op imism, gene ally de ined as ha ing a posi i e
ou look abou he u u e (5).
Op imism has shown o posi i ely associa e wi h a heal hy die and a low body mass
index (BMI) in a popula ion-based s udy o young adul s (6), and wi h heal hy beha iou s
including inc eased physical ac i i y and ea ing a heal hie die in elde ly men (7). Ano he
s udy ound ha in a coho o middle-aged adul s, op imism was associa ed wi h a heal hie
lipid p o ile, media ed pa ly by heal h beha iou s such as die as well as BMI (8). Despi e
eme ging esea ch assessing he impac o op imism on BMI and weigh gain, esea ch
examining op imism in he con ex o he in e gene a ional ansmission o obesi y is lacking.
Boehm and Kubzansky (9) ecen ly de eloped a model o explain he associa ion
be ween posi i e psychological ac o s (including op imism) and physical heal h. The model
sugges s ha he associa ion be ween posi i e psychological wellbeing and ca dio ascula
disease is media ed by es o a i e p ocesses, which help o bu e agains ill heal h (e.g. ea ing
a heal hy die ), as well as by he absence o de e io a i e p ocesses ha nega i ely a ec heal h
(e.g. smoking). These es o a i e and de e io a i e p ocesses include heal h beha iou s as well
as biological unc ions (e.g. in lamma ion o choles e ol).
4
In line wi h his model, we wan ed o explo e whe he op imism bu e s agains he
amilial ansmission o high BMI. Using da a om he Ca dio ascula Risk in Young Finns
S udy (Young Finns S udy), a p ospec i e s udy design ini ia ed 37 yea s ago; we examined
whe he : (1) op imism in young adul hood p edic s BMI in mid-adul hood, and (2) op imism
in e ac s wi h pa en al BMI o mi iga e he in e gene a ional ansmission o high BMI.
Me hods
The Young Finns S udy is a p ospec i e coho s udy ini ia ed in 1980 o examine
ca dio ascula isk ac o s in a andomly selec ed coho o 3,596 Finnish child en aged 3-18
yea s old (see Rai aka i e al.) (10). The cu en s udy included an analy ic sample o 1043
pa icipan s wi h no missing da a on key a iables: pa en al BMI, pa en al socioeconomic
s a us, BMI in adul hood, and op imism in adul hood. Pa icipan s p o ided w i en in o med
consen and he s udy was app o ed by he e hics commi ees o he pa icipa ing uni e si ies.
Op imism was measu ed in 2001 when pa icipan s we e 24-39 yea s, using he Li e
O ien a ion Tes -Re ised (11, 12), a sel - epo ques ionnai e ha measu es posi i e
expec ancies abou he u u e. The ques ionnai e has six i ems o which h ee a e wo ded
posi i ely and h ee nega i ely. I ems a e a ed on a 5-poin Like scale (0=no a all and
4= e y much so). The sco e can ange om 0-24 (whe eby he nega i e i ems a e e e sed
sco ed), so ha a high sco e ep esen s high op imism. The scale demons a es good alidi y
and eliabili y (12). In he p esen sample, he C onbach alpha was α = 0.78.
BMI was measu ed bo h in 2001 ( o adjus o baseline BMI) as well as in 2012 when
pa icipan s we e aged 35-50 yea s. Weigh and heigh we e measu ed by ained s a . BMI
was calcula ed wi h he o mula: BMI=weigh (kg)/[heigh (m)]2. Ma e nal and pa e nal heigh
and weigh we e assessed using sel - epo ques ionnai es in 1980 and pa en al BMI was
calcula ed using he abo e o mula.
5
Pa en al occupa ion was assessed wi h sel - epo ques ionnai es in 1980. Pa en al
occupa ion was coded acco ding o he ca ego ies used by he Cen al S a is ical O ice o
Finland om 1 o 3 (1=manual, 2=lowe non-manual, 3=uppe non-manual) and we e used as
a p oxy o SES.
S a is ical Analyses
A e conduc ing a i ion analyses, we used hie a chical eg ession o examine he associa ion
be ween pa en al BMI (ma e nal and pa e nal sepa a ely) and BMI in he coho in mid-
adul hood, wi h all models adjus ed o age and pa en al SES (Model 1). We hen sequen ially
adjus ed o op imism (Model 2) and he in e ac ion e m be ween ma e nal BMI and op imism,
and pa e nal BMI and op imism (Model 3). In ou supplemen a y analysis we also adjus ed o
BMI a baseline (measu ed a he same ime as op imism in 2001), in o de o examine he
di ec ion o he ela ionship be ween op imism and BMI. All con inuous a iables we e
s anda dized (Mean=0, SD=1). Due o he signi ican in e ac ion be ween op imism and sex
(p=.033) when p edic ing BMI in 2012, we sepa a ed all he analyses by sex. We epea ed he
main analyses using logis ic eg ession, wi h a bina y dependen a iable (BMI>30). We used
odds a ios (ORs) wi h 95% con idence in e als (CI) o examine how he odds o being obese
in adul hood we e associa ed wi h pa en al obesi y a baseline, and adjus ed o he abo e
co a ia es. Las ly, in o de o isualize he condi ional e ec o pa en al BMI on BMI in mid-
adul hood a low (-1 SD om mean), medium (mean) and high le els (+ 1 SD om mean) o
op imism, we used a simple slopes analysis using he PROCESS mac o (13). All he analyses
we e pe o med using SPSS (23).
Resul s
A i ion analyses showed ha pa icipan s in he analy ic sample had highe op imism
(p<0.001, Cohen’s d=0.18) as well as lowe adul hood BMI (p=0.013, Cohen’s d=0.11) and
baseline BMI measu ed in 2001 (p=0.003, Cohen’s d=0.12), compa ed o pa icipan s om he

6
b oade sample. The e we e no di e ences in ma e nal BMI (p=0.265, Cohen’s d=0.04) o
pa e nal BMI (p=0.997, Cohen’s d=0.00). The pa en s om he analy ic sample we e also mo e
likely o ha e a highe SES (p<0.001).
Ou o he 1043 pa icipan s, 645 we e women (61.8%). A e age BMI o women in
adul hood was 25.9 kg/m2 (± 5.3), ou o which 125 had a BMI>30. A e age BMI o men in
adul hood was 26.6 kg/m2 (± 4.2), and 77 had a BMI >30. Mean op imism in adul hood was
17.5 (± 3.7) o women and 17.2 (± 3.8) o men. The coho ’s mo he s’ mean BMI was 24
kg/m2 (± 3.7), ou o which 70 had a BMI >30 a baseline. The a he s’ BMI was 25.5 kg/m2
(± 3.0) and 85 had a BMI>30. Pa en al SES was g ouped as low (n=259, 24.8%), medium
(n=634, 60.8%), and high (n=150, 14.4%).
Due o he signi ican sex × op imism in e ac ion e m (p=0.033) when p edic ing BMI
in 2012, he ollowing analyses we e sepa a ed by sex. In he eg ession models, high op imism
in young adul hood demons a ed a nega i e ela ionship wi h high BMI in mid-adul hood, bu
only in women (β= = -0.127, p=0.001, Table 1). Op imism was no a signi ican p edic o o
u u e BMI in men (β= -0.017, p=0.728), he e o e he esul s ocus on women. Ma e nal BMI
was posi i ely associa ed wi h BMI in mid-adul hood ac oss all models. Only he ma e nal (no
he pa e nal) BMI × op imism in e ac ion e m was signi ican in p edic ing BMI in 2012 (β =
-0.588, p=0.036 e sus β = -0.148, p=0.676). Once we adjus ed o he ma e nal BMI ×
op imism in e ac ion e m, op imism was no longe a signi ican p edic o o BMI in women
(β = 0.503, p=0.180), indica ing mode a ion.
-INSERT TABLE 1 HERE-
The simple slopes analysis con i med hese esul s, and indica ed ha low op imism in
young adul hood inc eased he isk o ha ing a high BMI i he mo he had a high BMI (Figu e
1). As a supplemen a y analysis (see Supplemen a y Table 1), we also epea ed he linea
eg ession analysis (Models 1-3) and adjus ed o BMI a baseline (measu ed in 2001), in o de
7
o examine he empo al ela ionship be ween op imism and BMI in women. Op imism
emained a signi ican p edic o o BMI in 2012, e en a e adjus ing o baseline BMI (β = -
0.068, p=0.001). Howe e , he in e ac ion be ween ma e nal BMI and op imism was no longe
a p edic o o BMI in 2012 (β = -0.048, p=0.755).
Las ly, he logis ic eg ession also ound ha high op imism p edic ed educed u u e
obesi y in women bu no men (OR=0.68, 95% CI, 0.55-0.86, p=0.001 compa ed o OR=0.98,
95% CI, 0.75-1.29, p=0.947, see Supplemen a y Table 2). Howe e , he in e ac ion e m
be ween op imism × ma e nal obesi y was no a signi ican p edic o o obesi y in women
(OR=0.50, 95% CI, 0.10-2.48, p=0.396).
-INSERT FIGURE 1 HERE-
Discussion
To ou knowledge, his is he i s s udy o demons a e he p o ec i e e ec s o high op imism
in he in e gene a ional ansmission o high BMI, he eby also adding o he li e a u e on
posi i e psychological ac o s and heal h ou comes (9). Ano he no el inding was ha
op imism was ela ed o change in BMI o e ime, wi h he supplemen a y analysis (whe e we
adjus ed o baseline BMI) sugges ing ha high op imism is d i ing low BMI. Bo h o hese
indings we e only signi ican in women. Al hough ou main esul s demons a ed ha
op imism in young adul hood in e ac s wi h pa en al BMI o mode a e he in e gene a ional
ansmission o high BMI, when we adjus ed o baseline BMI, his in e ac ion was no longe
signi ican . This seems o sugges ha pe haps ins ead o p e en ing he g ow h o high BMI,
high op imism may p e en ini ially high le els o BMI in women wi h high ma e nal BMI.
As we did no se ou o es possible mechanisms o he p o ec i e e ec o high
op imism, we can only specula e on he easons why high op imism in women seems o be
8
p o ec ing agains u u e weigh gain. High op imism has been linked o p o ec i e heal h
beha iou s, sugges ing ha op imis ic people may be mo e likely o adop and main ain heal hy
beha iou s (14, 15). Ano he possible pa hway be ween high op imism and a heal hie BMI is
ha op imism may p o ec agains psychosocial s ess, which has been associa ed wi h long-
e m weigh gain and inc eased cen al adiposi y (16, 17).
I was unexpec ed ha op imism did no independen ly p edic u u e weigh gain in
men. The li e a u e on sex-speci ic e ec s o p o ec i e psychosocial ac o s, such as op imism,
on heal h ou comes a e sca ce. Howe e , o he s udies ha e ound ha maladap i e
psychosocial ac o s such as s ess can ha e sex-speci ic e ec s on heal h ou comes including
BMI and Type 2 diabe es (18, 19). One possibili y o he sex-speci ic ou comes in ou s udy
could be ha ha ing a posi i e ou look has a s onge associa ion wi h ce ain heal h beha iou s
ha p o ec agains weigh gain in women compa ed o men. P e ious s udies ha e ound ha
low mood and psychological dis ess was associa ed wi h di e en heal h beha iou s in men
and women, including poo die a y beha iou s and lack o exe cise in women (20, 21).
Al e na i ely, pe haps op imism p o ec s women mo e han men agains ch onic psychosocial
s ess and i s impac on heal h, o which women may be mo e ulne able (22), and which may
exace ba e weigh gain (17).
S eng hs o ou s udy include a longi udinal design ha allowed us o examine he
associa ion be ween op imism and BMI 11 yea s la e . The coho is also unique in ha we
ha e da a om wo gene a ions, allowing us o in es iga e in e gene a ional phenomena.
Limi a ions include a ela i ely homogenous popula ion, sel - epo ed pa en al BMI, and
a i ion o e he mo e han h ee decades; howe e , i has no been signi ican enough o bias
indings (23, 24). Las ly, we only measu ed op imism a one ime poin , making i di icul o
ule ou ha op imism did no change du ing he ollow-up pe iod. Howe e , op imism is
9
gene ally seen as a s able ai (5), wi h app oxima ely 25% o op imism hough o be due o
gene ic ac o s (25).
In summa y, ou esul s sugges ha high op imism p o ec s agains high BMI in
women o e ime, and ha high op imism mode a es he in e gene a ional ansmission o BMI
in women. The empo al na u e o he la e ela ionship wa an s u he in es iga ion. Ou
indings a e consis en wi h eme ging e idence sugges ing ha posi i e psychosocial ac o s,
bo h in childhood and in adul hood, a e associa ed wi h long- e m p o ec i e e ec s on
me abolic unc ioning (26, 27).