RESEARCH Open Access
Change in body mass index du ing
ansi ion o s a u o y e i emen : an
occupa ional coho s udy
Sa i S enholm
1,2*
, S e lana Solo ie a
3
, Ei a Viika i-Jun u a
3
, Ville Aal o
3
, Mika Ki imäki
3,4,5
and Jussi Vah e a
1
Abs ac
Backg ound: Re i emen is a majo li e ansi ion a ec ing heal h beha io s. The aim o his s udy was o examine
wi hin-indi idual changes in body mass index (BMI) du ing ansi ion om ull- ime wo k o s a u o y e i emen by
sex and physical wo k cha ac e is ics.
Me hods: A mul iwa e coho s udy epea ed e e y 4 yea s and da a linkage o eco ds om e i emen egis e s.
Pa icipan s we e 5426 Finnish public-sec o employees who e i ed on a s a u o y basis in 2000–2011 and who epo ed
hei body weigh one o h ee imes p io o (w
−3
,w
−2
,w
−1
), and one o h ee imes a e (w
+1
,w
+2
,w
+3
) e i emen .
Resul s: Du ing he 4-yea e i emen ansi ion (w
+1
, s.w
−1
) men showed decline in BMI, which was mos ma ked
among men wi h seden a y wo k (−0.18 kg/m
2
, 95% CI −.30 o −0.05). In con as , BMI inc eased du ing e i emen
ansi ion in women and was mos ma ked among women wi h di e se (0.14 kg/m
2
, 95% CI 0.08 o 0.20) o physically
hea y wo k (0.31 kg/m
2
, 95% CI 0.16 o 0.45). Physical ac i i y du ing leisu e ime o commu ing o wo k, alcohol
consump ion o smoking did no explain he obse ed changes du ing e i emen ansi ion.
Conclusions: In his s udy s a u o y e i emen was associa ed wi h small changes in BMI. Weigh loss was mos isible
in men e i ing om seden a y jobs and weigh gain in women e i ing om di e se and physically hea y jobs.
Keywo ds: Body weigh , BMI, Weigh change, Wo k exposu e, Re i emen , Aging, Coho s udy
Backg ound
T ansi ion o e i emen is an impo an u ning poin in
li e accompanied by subs an ial inc ease in ime a ailabil-
i y and changes in daily ou ines [1]. P e ious esea ch has
shown ha leisu e- ime physical ac i i y [2, 3] and o al al-
cohol consump ion [4, 5] end o inc ease du ing e i e-
men ansi ion. Since changes in weigh a e ela ed o
changes in ene gy expendi u e (physical ac i i y) and en-
e gy in ake ( ood and alcohol in ake) o hei combina ion,
e i emen may also con ibu e o weigh change among
olde adul s. Examining changes in weigh and ac o s p e-
disposing such changes in his li e s age is impo an since
bo h weigh gain and weigh loss may a ec heal h a
olde ages. I has been shown, o example, ha weigh
gain can inc ease isk o mo bidi y and disabili y [6–9]
while unin en ional weigh loss may inc ease he isk o
mo ali y [10]. In en ional weigh loss accompanied by in-
c eased physical ac i i y, on he o he hand, is associa ed
wi h imp o ed physical unc ioning in olde adul s [11].
To da e, ela i ely li le is known abou he e ec s o
e i emen on weigh , al hough i is likely ha he e ec s
a e he e ogeneous and in pa depend on he job a pe -
son is e i ing. I is possible, o example, ha e i emen
om physically s enuous wo k leads o educ ion in
o al physical ac i i y and hus induces weigh gain. On
he o he hand, pe sons who e i e om seden a y jobs
may be able o inc ease ene gy consump ion by engaging
mo e physical ac i i y h oughou he day. In ag eemen
wi h his, some p e ious s udies sugges ha weigh gain
is mo e common in people e i ing om physically
s enuous job o blue-colla occupa ions [12–18]. How-
e e , he esul s o seden a y wo k and weigh change
* Co espondence: [email p o ec ed]
1
Depa men o Public Heal h, Uni e si y o Tu ku and Tu ku Uni e si y
Hospi al, Tu ku, Finland
2
Facul y o Social Sciences (Heal h Science), Uni e si y o Tampe e, Tampe e,
Finland
Full lis o au ho in o ma ion is a ailable a he end o he a icle
© The Au ho (s). 2017 Open Access This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (h p://c ea i ecommons.o g/licenses/by/4.0/), which pe mi s un es ic ed use, dis ibu ion, and
ep oduc ion in any medium, p o ided 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 license, and indica e i changes we e made. The C ea i e Commons Public Domain Dedica ion wai e
(h p://c ea i ecommons.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.
S enholm e al. In e na ional Jou nal o Beha io al Nu i ion
and Physical Ac i i y (2017) 14:85
DOI 10.1186/s12966-017-0539-2
a e con lic ing, some s udies epo ing no weigh change
and o he s udies weigh loss du ing e i emen ansi-
ion [12, 14]. Whe he he changes in body weigh di e
in men and women is no p ope ly examined. Mo eo e ,
mos p e ious esea ch is based on a single da ase , he
Heal h and Re i emen S udy om he US. The US ep-
esen s a coun y in which pension schemes a y and ex-
ended labo -ma ke pa icipa ion is o en due o
inancial easons which di e s om many Eu opean
coun ies [19]. I is he e o e impo an o examine hese
ques ions also in o he popula ions and se ings. Acco d-
ingly, he aim o he cu en s udy is o examine changes
in body mass index du ing yea s p eceding e i emen ,
du ing e i emen ansi ion and a e e i emen wi h
pa icula ocus on po en ial di e ences in weigh de el-
opmen by sex and physical wo k cha ac e is ics.
Me hods
S udy popula ion
The da a we e om he Finnish Public Sec o s udy, an
ongoing p ospec i e occupa ional coho s udy wi h
iden i iable ques ionnai e su eys. The eligible popula-
ion o he o iginal coho included all employees who
had been wo king o a minimum o 6 mon hs in he
pa icipa ing o ganiza ions, which included en owns
and six hospi al dis ic s, be ween 1991 and 2005
(n= 151,901) [20]. Nes ed su ey coho s included all
hose who we e employed by he pa icipa ing o ganiza-
ions a he ime o su eys o had le he o ganiza ions
a e pa icipa ing in an ea lie su ey. The i s su ey
o he o al pe sonnel in all pa icipa ing o ganiza ions
was conduc ed in 2000–2002 and he ea e epea ed a
4-yea in e als. The i s su ey o lea e s was con-
duc ed in 2005 and he ea e epea ed a 4-yea in e -
als. Fo his s udy we used da a om su eys
pe o med o employees in 2000–2002, 2004 and 2008
and lea e s in 2005, 2009 and 2013. Su ey da a o co-
ho membe s we e success ully linked o eco ds o he
Finnish Cen e o Pension’s egis e ( e i emen da e
and ype), employe s’ eco ds (bi h da e, sex, occupa-
ional i le) and comp ehensi e na ional heal h egis e s
(diseases and medica ion) h ough unique pe sonal iden-
i ica ion codes, which a e assigned o all ci izens in
Finland. The FPS s udy was app o ed by he E hics
Commi ee o he Hospi al Dis ic o Helsinki and
Uusimaa.
O all he FPS coho membe s, we i s iden i ied
hose who we e employed and esponded o a leas one
su ey in 2000–2002, 2004 o 2008 (n= 81,587). O
hese employees, 19,058 we e awa ded hei i s pension
by Decembe 31, 2011, and o hese, 9787 pe sons had
esponded o a leas one su ey be o e and a e e i e-
men . We ocused on hose pe sons who had e i ed a
he s a u o y e i emen age (i.e. old age e i emen ) as
hei i s awa ded pension scheme (n= 5898). Pa ici-
pan s who e i ed pa - ime (n= 1619) o on a heal h
g ounds (n= 2185) o because o unemploymen
(n= 85) we e excluded om his s udy, because hese
ypes o e i emen a e endogenous and po en ially e-
la ed o he causes ha may a ec weigh (e.g. disease)
and hus subjec o e e se causa ion bias.
We cen e ed he da a a ound he e i emen da e. The e
we e h ee s udy wa es be o e e i emen (w
−3
,w
−2
,w
−1
),
and h ee wa es a e e i emen (w
+1
,w
+2
,w
+3
). Each suc-
cessi e wa e was on a e age 4 yea s apa om each o he .
To be included in his s udy, he pa icipan s had o epo
hei body weigh and heigh in a leas wo su eys, one
immedia ely be o e and a e ansi ion o s a u o y e i e-
men (i.e., w
−1
and w
+1
)(n= 5458). Then we excluded
unde weigh pe sons a p e- e i emen (n= 22) and hose
wi h missing occupa ional i les (n= 10) esul ing in an
analy ic sample o 5426 pe sons. Thus, depending on he
e i emen da e, pa icipan s’obse a ions came om one
o he ollowing al e na i e se o wa es: 1) w
−3
,w
−2
,w
−1
,
w
+1
,2)w
−2
,w
−1
,w
+1
,w
+2
,o 3)w
−1
,w
+1
,w
+2
,w
+3
.The
ela ion o he su ey yea s o he s udy wa es a ound
e i emen is demons a ed in supplemen al ma e ial
(Addi ional ile 1: Table S1). On a e age, hese pa ic-
ipan s p o ided in o ma ion on body weigh a 3.6
( ange 2–4) o he possible ou s udy wa es du ing a
ollow-up o 8–12 yea s.
Assessmen o e i emen
Da a on ype and da e o e i emen we e ob ained om
he Finnish Cen e o Pensions, which coo dina es all
ea nings- ela ed pensions o pe manen esiden s in
Finland [21]. All gain ul employmen is insu ed in a pen-
sion plan and acc ues a pension; hus he pension da a
wi h success ul linkage we e a ailable o all pa icipan s.
The s a da es o any pension we e ob ained o all
pa icipan s om 2000 h ough 2011, i espec i e o he
pa icipan s’employmen s a us o wo kplace a ollow-
up. Acco ding o he public sec o Employees’Pension
Ac , he s a u o y e i emen age was gene ally om 63
o 65 yea s un il 2005 and 63 o 67 yea s om 2005 on-
wa ds, al hough some indi iduals had kep hei ea lie
e i emen age om he p e ious pension ac in which
pension ages in some occupa ions we e below 63 yea s
(e.g., 60 yea s o p ima y school eache s, 58 o p ac-
ical nu ses).
Assessmen o body weigh and body mass index
Body weigh and heigh we e inqui ed a each s udy wa e
and was eco ded in kilog ams and cen ime e s, espec -
i ely. Based on his in o ma ion body mass index (BMI)
was calcula ed a each s udy wa e and BMI was ca ego-
ized in o no mal weigh (18.5 kg/m
2
≤BMI < 25.0 kg/m
2
),
o e weigh (25–29.9 kg/m
2
≤BMI < 30 kg/m
2
)andobesi y
S enholm e al. In e na ional Jou nal o Beha io al Nu i ion and Physical Ac i i y (2017) 14:85 Page 2 o 8
class I (30 kg/m
2
≤BMI < 35 kg/m
2
) and obesi y class II I
(35 kg/m
2
≤BMI) [22].
Assessmen o physical wo k cha ac e is ics
Occupa ional i les we e ob ained om he employe s’
egis e s and coded acco ding o he S anda d Classi ica-
ion o Occupa ions 2001 by S a is ics Finland [23]. In-
o ma ion on he las occupa ion p eceding e i emen
(w
−1
) was used and i his was no a ailable, hen in o -
ma ion om w
−2
was used. We linked each occupa ional
code o alida ed gende -speci ic job exposu e ma ix
(JEM) [24], which was de eloped by using exposu e in-
o ma ion om a la ge na ionally ep esen a i e su ey
(Heal h 2000 S udy [25]). The ma ix includes exposu e
in o ma ion o mo e han 401 occupa ions o occupa-
ional g oups coded acco ding o he Classi ica ion o
Occupa ions 2001 by S a is ics Finland. The classi ica-
ion is based down o he 4-digi le el, co esponding o
he EU’s classi ica ion o occupa ions (ISCO-88(COM)).
In addi ion, na ional ci cums ances ha e been aken in o
accoun by adding 5-digi occupa ional g oups, when ne-
cessa y. In his s udy popula ion 244 di e en occupa-
ional i les we e p esen which all could be linked o
JEM.
Fo he cu en s udy, wo physical wo k cha ac e is-
ics we e used, namely “physical hea iness o wo k”and
“si ing a wo k”. The dicho omized exposu es we e
based on he ollowing wo ques ions a he Heal h 2000
Su ey: “Does you cu en job in ol e hea y physical
wo k, in which you ha e o li o ca y hea y i ems, o
dig, sho el o pound?”(yes/no) and “Do you in you
cu en wo k need o si (wo k machine o ca d i ing
no included) on an a e age a leas i e hou s a day?”
(yes/no). By using in o ma ion om bo h i ems, a com-
posi e wo k indica o was c ea ed and labelled as “Sed-
en a y”(si ing bu no hea y wo k), “Di e se”(no si ing
and no hea y wo k”and “Physically hea y”(no si ing
bu hea y wo k). The e was no one in jobs wi h bo h si -
ing and hea y wo k.
In addi ion, socioeconomic posi ion (SES) was de ined
based on ISCO and ca ego ized in o high (ISCO ca ego -
ies 1–2 / e.g. eache s, physicians), in e media e (ISCO
ca ego ies 3–4 / e.g. egis e ed nu ses, echnicians) and
low (ISCO ca ego ies 5–9 / e.g. cleane s, main enance
wo ke s) [23].
Assessmen o co a ia es
To measu e non-occupa ional physical ac i i y, espon-
den s we e asked o es ima e hei a e age weekly hou s
o leisu e- ime physical ac i i y (including commu ing)
wi hin he p e ious yea in walking, b isk walking, jog-
ging, and unning, o hei equi alen ac i i ies [26]. The
ime spen on ac i i y a each in ensi y le el in hou s pe
week was mul iplied by he a e age ene gy expendi u e
o each ac i i y, exp essed in me abolic equi alen
(MET) [2, 27]. Alcohol use was based on sel - epo ed
habi ual equency and amoun o bee , wine and spi i s
consump ion, which was con e ed in o g ams o alco-
hol pe week [28]. Due o i s dis ibu ional skewness, a
loga i hmic ans o ma ion is used in he analyses.
Smoking s a us was ca ego ized in o ne e , o me and
cu en smoke s. Fo he analyses, physical ac i i y, alco-
hol use and smoking we e modelled as a ime- a ian
a iables since hey may in luence on weigh change.
Disease s a us was cons uc ed by aking in o accoun
ch onic diseases in all p e- e i emen wa es a ailable. In-
o ma ion on ch onic illnesses was ob ained on na ion-
wide egis e s: as hma, diabe es, heuma oid a h i is
and co ona y hea disease based on he Social Insu -
ance Ins i u ion o Finland’s (SII) D ug Reimbu semen
Regis e ; dep ession based on he Finnish P esc ip ion
Regis e kep by SII (ATC code N06A) and cance based
on he Finnish Cance Regis y. In addi ion in o ma ion
on os eoa h i is was ob ained om he ques ionnai es.
Fo he analyses, pa icipan s we e ca ego ized as ha ing
no disease, one disease and wo o mo e diseases be o e
e i emen .
An addi ional co a ia e possibly ela ed o body weigh
is a job s ain, which was asce ained using ques ions
om he Job Con en Ques ionnai e (JCQ) [29]. The
FPS s udy su eys included job con ol and job demands
scales om he sho e e sion o he JCQ [30]. The
p esence o job s ain a p e- e i emen (w
−1
) was de-
ined as ha ing high demands and a low con ol sco e
based on he median alues om he yea 2000 su ey.
S a is ical analyses
Cha ac e is ics o he s udy popula ion be o e e i emen
(w
−1
) o men and women a e p esen ed as mean alues
o con inuous a iables and as p opo ions o ca ego -
ical a iables. Changes in BMI we e assessed using linea
eg ession analyses wi h gene alized es ima ion equa-
ions (GEE). The GEE models con ol o he in a-
indi idual co ela ion be ween epea ed measu emen s
using an exchangeable co ela ion s uc u e and is no
sensi i e o measu emen s missing comple ely a
andom [31, 32].
We s a ed by examining whe he BMI change was
di e en among men and women and since he in e -
ac ion ‘sex x ime’was s a is ically signi ican
(p< 0.0001), all analyses we e conduc ed sepa a ely o
men and women. In u he analyses he in e es was in
di e ences by physical wo k cha ac e is ics and hus he
models included ‘wo k cha ac e is ics x ime’in e ac ion
e m. Time a iable in ou models indica ed ime poin
in espec o e i emen .
To examine changes in BMI a ound he ansi ion o
e i emen , we cons uc ed h ee consecu i e pe iods:
S enholm e al. In e na ional Jou nal o Beha io al Nu i ion and Physical Ac i i y (2017) 14:85 Page 3 o 8
he p e- e i emen pe iod ( om w
−3
o w
−2
), he e-
i emen ansi ion ( om w
−1
o w
+1
) and he pos -
e i emen pe iod ( om w
+2
o w
+3
). The pe iods
we e non-o e lapping o allow es ing whe he changes
in BMI di e ed be ween he p e- e i emen pe iod, he
e i emen ansi ion, and he pos - e i emen pe iod.
The s a is ical signi icance o hese changes we e es ed
using ‘Pe iod x Time’in e ac ion e m whe e Time was
ea ed as con inuous a iable.
Adjus ed mean es ima es and hei 95% con idence in-
e als we e calcula ed o ep esen an a e age o 4-yea
change o BMI a di e en pe iods. In addi ion, isk o
obesi y (BMI ≥30 kg/m
2
) was examined wi h log-
binomial GEE models by calcula ing isk a ios (RR)
wi hin each pe iod by compa ing p e alence o obesi y
a he la e s udy wa e o he p e ious s udy wa e (e.g.
wa e
+1
s. wa e
−1
). The analyses we e adjus ed o e-
i emen age, SES, ime- a ying physical ac i i y, alcohol
consump ion and smoking as well as ma i al s a us,
numbe o ch onic diseases, job s ain and BMI be o e
e i emen (w
−1
).
Finally, we conduc ed a sensi i i y analyses among
hose whom da a on BMI we e a ailable om all ou
measu emen s (n= 3683). This was done o add ess he
ques ion on whe he including hose pa icipan s who
had missing da a in one o wo measu emen poin s a -
ec ed he esul s. The SAS 9.4 S a is ical Package was
used o all o he analyses (SAS Ins i u e Inc., Ca y,
NC).
Resul s
The a e age e i emen age di e ed be ween sexes be-
ing 62.2 (SD 2.3) yea s in men and 61.8 (SD 1.9) yea s
in women. A he su ey be o e e i emen (w
−1
),
p e alence o no mal weigh was 35% in men and 46%
in women, o e weigh 47% in men and 39% in women
and obesi y 18% in men and 15% in women. O he pa -
icipan s 13% o men and 16% o women we e engaged
in hea y physical wo k and 37% o men and 14% o
women had seden a y wo k. The de ailed cha ac e is-
ics in men and women be o e e i emen (w
−1
)a e
showninTable1.
De elopmen o BMI be o e, du ing and a e e i e-
men in men and women a e illus a ed in Fig. 1. The
magni ude o change in BMI wi hin each pe iod di e ed
be ween p e- e i emen , e i emen ansi ion and pos -
e i emen pe iods bo h in men and women (pe iod x
ime in e ac ion p= 0.0009 o men and p< 0.0001 o
women). In men BMI inc eased du ing p e- e i emen
pe iod by 0.29 kg/m
2
(95% con idence in e al (CI) 0.13
o 0.46), dec eased du ing e i emen ansi ion by
0.11 kg/m
2
(95% CI -0.22 o −0.01) and became s eady
du ing pos - e i emen pe iod (−0.03 kg/m
2
, 95% CI
-0.23 o 0.17) (Table 2). In women BMI inc eased du ing
Table 1 Cha ac e is ics o he s udy popula ion be o e
e i emen in men and women (n= 5426)
Men Women
n= 1116 n= 4310 p- alue
Mean SD Mean SD
Re i emen age (yea s) 62.2 2.3 61.8 1.9 <.0001
n%n%
Re i emen age
< 60 159 14 505 12 <.0001
60–64 758 68 3253 75
> 64 202 18 559 13
Socioeconomic s a us <.0001
High 537 48 1537 36
In e media e 212 19 1229 29
Low 367 33 1528 36
Wo k cha ac e is ics <.0001
Seden a y 414 37 623 14
Di e se 555 50 2984 69
Physically hea y 147 13 703 16
Body mass index <.0001
No mal weigh 384 35 1980 46
O e weigh 522 47 1673 39
Obese class I 162 15 520 12
Obese class II 41 4 125 3
Low leisu e- ime physical ac i i y
a
No 647 58 2491 58 0.99
Yes 463 42 1787 42
Alcohol use
b
<.0001
None 85 8 767 18
Mode a e 908 82 3223 75
Hea y 113 10 298 7
Smoking s a us <.0001
Ne e 586 55 3361 79
Fo me 337 32 546 13
Cu en 149 14 322 8
P esence o ch onic diseases <.0001
0 46 4 224 5
1 671 60 2220 51
> 1 399 36 1866 43
Job s ain <.0001
No 945 85 3086 73
Yes 164 15 1159 27
a
Low physical ac i i y a leisu e ime o du ing commu ing o wo k is de ined
as less han 14 MET-hou s pe week.
b
Mode a e alcohol use co esponds
<192 g / week in women and <288 g / week in men and hea y alcohol use
≥192 g / week in women and ≥288 g / week in men.
S enholm e al. In e na ional Jou nal o Beha io al Nu i ion and Physical Ac i i y (2017) 14:85 Page 4 o 8
p e- e i emen pe iod by 0.41 kg/m
2
(95% CI 0.31 o
0.51) and du ing he e i emen ansi ion by 0.15 kg/m
2
(95% CI 0.10 o 0.20), bu became s eady du ing pos -
e i emen pe iod (0.10 kg/m
2
, 95% CI -0.02 o 0.22)
(Table 2). Fu he adjus men s o ime- a ying physical
ac i i y, alcohol consump ion and smoking as well as
ma i al s a us, numbe o ch onic diseases, body mass
index and job s ain be o e e i emen (w
−1
) did no al e
he esul s.
Changes in BMI by wo k cha ac e is ics, we e also ex-
amined. Men wi h physically hea y wo k had sligh ly
highe BMI (27.18 kg/m
2
, 95% CI 26.64 o 27.72) com-
pa ed o men wi h seden a y (26.69 kg/m
2
, 95% CI 26.31
o 27.08) and di e se wo k (26.87 kg/m
2
, 95% CI 26.55
o 27.20) be o e e i emen (w
−1
) (Table 2, Fig. 2a). A e
adjus ing o co a ia es, men in seden a y wo k in-
c eased hei BMI du ing p e- e imen pe iod (0.26 kg/
m
2
, 95% CI 0.003 o 0.52) and dec eased du ing e i e-
men ansi ion (−0.19 kg/m
2
, 95% CI -0.32 o −0.07),
bu no change was obse ed du ing pos - e i emen
pe iod. Among men wi h di e se and physically hea y
wo k, no signi ican changes we e obse ed du ing e-
i emen ansi ion and pos - e i emen pe iods.
The esul s o women a e shown in Table 3 and Fig. 2b.
Be o e e i emen (w
−1
), women wi h physically hea y
wo k had highe BMI (26.57 kg/m
2
, 95% CI 26.28 o
26.86) han women wi h seden a y wo k (26.16 kg/m
2
,
95% CI 25.82 o 25.50) and wi h di e se wo k (25.85 kg/
m
2
, 95% CI 25.71 o 26.00). BMI inc eased in all wo k
cha ac e is ics g oups du ing p e- e i emen pe iod by
abou 0.40 kg/m
2
. Howe e , only women wi h di e se and
physically hea y wo k, showed inc ease in BMI du ing e-
i emen ansi ion pe iod (0.16 kg/m
2
, 95% CI 0.09 o
0.22 and 0.30 kg/m
2
, 95% CI 0.15 o 0.46). No s a is ically
signi ican change in BMI in pos - e i emen pe iod was
obse ed in any o he wo k cha ac e is ics g oups.
In addi ion o changes in absolu e BMI alues, he isk
o obesi y wi hin each pe iod was examined. In men, e-
i emen did no inc ease isk o becoming obese, bu
women in di e se (RR 1.15, 95% CI 1.07 o 1.22) and
physically hea y wo k had inc eased isk o obesi y du -
ing e i emen ansi ion (RR 1.20, 95% CI 1.07 o 1.34)
(Addi ional ile 1: Table S2).
Finally, we epea ed he analyses including only pa ic-
ipan s who had all ou body weigh measu emen s
a ailable. The esul s we e e y simila o hose o he
ac ual analyses wi h only mino di e ences in he es i-
ma es.. In gene al, hese indings sugges no majo selec-
ion by sex o physical wo k cha ac e is ics (Addi ional
ile 1: Table S3).
Discussion
The esul s o his longi udinal s udy o Finnish public
sec o wo ke s sugges ha ansi ion o s a u o y e i e-
men is associa ed wi h a small dec ease in BMI in men
and a sligh inc ease in women. In men he decline du -
ing e i emen ansi ion was mos ma ked among hose
wi h seden a y jobs. In con as , in women he weigh
Fig. 1 Changes in body mass index du ing e i emen ansi ion in men
and women. Adjus ed o e i emen age and socioeconomic s a us
Table 2 Change in body mass index be o e, du ing and a e e i emen ansi ion by physical wo k cha ac e is ics in men
Time in ela ion o e i emen
P e- e i emen (w
−2
s. w
−3
) Re i emen ansi ion (w
+1
s. w
−1
) Pos - e i emen (w
+3
s. w
+2
)
Mean change
a
95% CI Mean change
a
95% CI Mean change
a
95% CI
To al Model 1 0.29 0.13 0.46 −0.11 −0.22 −0.01 −0.03 −0.23 0.17
(n= 1116) Model 2 0.30 0.12 0.44 −0.11 −0.22 −0.01 0.07 −0.14 0.28
Seden a y Model 1 0.21 −0.04 0.46 −0.18 −0.30 −0.05 −0.0004 −0.29 0.29
(n= 414) Model 2 0.26 0.003 0.52 −0.19 −0.32 −0.07 0.08 −0.23 0.38
Di e se Model 1 0.44 0.20 0.68 −0.05 −0.20 0.11 0.01 −0.28 0.31
(n= 555) Model 2 0.40 0.14 0.65 −0.03 −0.18 0.12 0.13 −0.16 0.43
Physically hea y Model 1 0.07 −0.41 0.55 −0.20 −0.63 0.23 −0.34 −1.01 0.32
(n= 147) Model 2 0.06 −0.51 0.63 −0.19 −0.64 0.26 −0.22 −0.91 0.46
Da a a e cen e ed a e i emen : w
−1
,w
−2
, and w
−3
e e o su ey wa es be o e e i emen , and w
+1
,w
+2
and w
+3
e e o su ey wa es a e e i emen .
a
Change is es ima ed o e 4 yea s o ime. Model1: adjus ed o e i emen age and socioeconomic posi ion Model 2: Model 1 + adjus ed o physical ac i i y,
alcohol use and smoking as ime- a ying co a ia es and ma i al s a us, body mass index, numbe o ch onic diseases and job s ain be o e e i emen .
S enholm e al. In e na ional Jou nal o Beha io al Nu i ion and Physical Ac i i y (2017) 14:85 Page 5 o 8

inc ease was mos p onounced among hose wi h di e se
and physically hea y wo k. They we e also mo e likely
become obese du ing e i emen ansi ion.
These indings add o exis ing e idence on e i emen
and weigh . An ad an age o he p esen in es iga ion
o e p e ious s udies is ha we examined changes in
weigh in men and women aking in o accoun bo h
physical wo k cha ac e is ics and SES. In addi ion, we
con olled o a numbe o unde lying ac o s o weigh
change, including leisu e- ime physical ac i i y, alcohol
consump ion and smoking. Since mos o he p e ious
s udies ha e analyzed men and women oge he [13, 14],
ou compa ison o o he s udies is limi ed o s udies
showing esul s sepa a ely o men o women.
Ou indings a e consis en wi h he Heal h and Re-
i emen S udy (HRS) [15] om he US in which e-
i emen was associa ed wi h weigh gain among
women om blue colla occupa ions, whe eas no
change was obse ed in men. The magni ude o a e -
age weigh change was ela i ely simila among
women in he Finnish Public Sec o s udy (0.8 kg
o e 4 yea s) and ha HRS s udy (abou 0.5 kg o e
2 yea s). Ou indings a e also in ag eemen wi h an
analysis based on he US HRS s udy by Goldman
e al. [12] sugges ing ha men e i ing om seden a y
jobs los abou 0.12 BMI-uni s (kg/m
2
) o e 2 yea s
compa ed o 0.2 BMI-uni s o e 4 yea s in he
p esen s udy. Ou indings a e in con as o hose
epo ed by Goldman e al. [12], Goda d e al. [17]
and Nooyens e al. [16] as in hese s udies weigh
gain among men e i ing om physically hea y wo k
was obse ed. In ac , in ou s udy he e was, i any-
hing, a endency owa ds dec easing BMI among
hose wi h physically hea y wo k. Taken oge he , he
changes in BMI we obse ed du ing e i emen ansi-
ion a e s a is ically signi ican bu in absolu e e ms
ela i ely small, less han ±1 kg pe 4 yea s. A 5%
weigh loss o weigh gain (3.5 kg o a pe son who
weigh 70 kg) is o en conside ed clinical ele an
[33]; wi h his c i e ion e i emen appea s no o
a ec weigh s a us in a clinically ele an way. A
he popula ion le el, howe e , e en ela i ely small
Fig. 2 Changes in body mass index du ing e i emen ansi ion in men and women by physical wo k cha ac e is ics. Adjus ed o e i emen age
and socioeconomic s a us. a) Men, b) Women
Table 3 Change in body mass index be o e, du ing and a e e i emen ansi ion by physical wo k cha ac e is ics in women
Time in ela ion o e i emen
P e- e i emen (w
−2
s. w
−3
) Re i emen ansi ion (w
+1
s. w
−1
) Pos - e i emen (w
+3
s. w
+2
)
Mean change
a
95% CI Mean change
a
95% CI Mean change
a
95% CI
To al Model 1 0.41 0.31 0.51 0.15 0.10 0.20 0.10 −0.02 0.22
(n= 4310) Model 2 0.41 0.30 0.52 0.16 0.11 0.22 0.11 −0.01 0.23
Seden a y Model 1 0.45 0.23 0.67 0.02 −0.12 0.16 0.03 −0.37 0.43
(n= 623) Model 2 0.47 0.24 0.71 0.02 −0.12 0.16 0.03 −0.38 0.43
Di e se Model 1 0.38 0.26 0.50 0.14 0.08 0.20 0.10 −0.04 0.23
(n= 2984) Model 2 0.40 0.27 0.52 0.16 0.09 0.22 0.09 −0.05 0.23
Physically hea y Model 1 0.47 0.16 0.78 0.31 0.16 0.45 0.14 −0.12 0.41
(n= 703) Model 2 0.38 0.04 0.71 0.30 0.15 0.46 0.25 −0.04 0.53
Da a a e cen e ed a e i emen : w
−1
,w
−2
, and w
−3
e e o su ey wa es be o e e i emen , and w
+1
,w
+2
and w
+3
e e o su ey wa es a e e i emen .
a
Change is es ima ed o e 4 yea s o ime. Model 1: adjus ed o e i emen age and socioeconomic posi ion. Model 2: Model 1 + adjus ed o physical ac i i y,
alcohol use and smoking as ime- a ying co a ia es and ma i al s a us, body mass index, numbe o ch onic diseases and job s ain be o e e i emen
S enholm e al. In e na ional Jou nal o Beha io al Nu i ion and Physical Ac i i y (2017) 14:85 Page 6 o 8
changes in he dis ibu ion o isk ac o s may ela e
o conside able changes in bene i s and ha ms in ab-
solu e e ms as he majo i y o inciden cases o dis-
ease occu in people wi h a medium isk [34, 35].
We also sough o examine he ole o unde lying
ac o s o weigh change by including ime- a ian
leisu e- ime physical ac i i y, alcohol consump ion
and smoking in ou models. Howe e , none o hese
ac o s explained he obse ed changes du ing e i e-
men ansi ion. A u he impo an aspec in ene gy
balance is nu i ion and ea ing habi s, which we e no
assessed in his s udy. In a p e ious smalle s udy
om he Ne he lands, ea ing habi s pa ially explained
changes in men’s weigh du ing e i emen ansi ion
[16]. Fu he esea ch is needed o examine whe he
e i emen is associa ed wi h changes in nu i ion and
ea ing habi s and whe he his could explain he ob-
se ed changes in body weigh .
S eng hs o ou s udy include he epea ed measu e-
men o body weigh a ound an objec i ely de e mined
e i emen ansi ion which enabled us o es ima e body
weigh de elopmen be o e, du ing and a e e i emen .
All pa icipan s e i ed om ull- ime wo k o ull- ime
e i emen , hus emo al o wo k- ela ed exposu es and
inc ease in leisu e- ime was simila o all. The e a e
some limi a ions o ou s udy. This s udy elied on
sel - epo ed body weigh which is subjec o ecall
and in o ma ion bias, possibly esul ing in unde -
epo ing o body weigh [36, 37]. Howe e , he e is
no eason o assume ha hese biases would be di -
e en o he di e en e i emen pe iods. The
gene alizabili y o he indings may be limi ed as he
coho consis ed o ela i ely heal hy public sec o
employees o Eu opean o igin in a Scandina ian wel-
a e s a e wi h a ela i ely gene eous e i emen
scheme.Inaddi ion,since he ocuso ou s udywas
on s a u o y e i emen , he esul s canno be gene al-
ized o o he e i emen ypes, such as ea ly e i e-
men due o poo heal h. Da a om he US shows
ha obese indi iduals a e mo e likely o seek ea ly e-
i emen bene i s han non-obese [38] and i is pos-
sible ha o hem changes in body weigh du ing
e i emen ansi ion may di e om hose en e ing
o s a u o y e i emen .
Conclusions
This s udy sugges s ha s a u o y e i emen is associ-
a ed wi h sligh weigh loss in men e i ing om seden-
a y jobs and a sligh weigh gain in women e i ing
om di e se and physically hea y jobs. The magni ude
o change in BMI sugges s ha e i emen does no ha e
clinically ele an in luence on weigh de elopmen
among olde adul s.
Addi ional ile
Addi ional ile 1: Table S1. S udy design. Su ey yea s, hei ela ion o
he s udy wa es a ound e i emen and he cons uc ion o he p e-
e i emen , e i emen ansi ion and pos - e i emen pe iods. Table S2.
Risk o obesi y du ing p e- e i emen , e i emen ansi ion and pos -
e i emen pe iods in men and women. Table S3. Change in body mass
index be o e, du ing and a e e i emen ansi ion in men and women.
Only pa icipan s wi h ou obse a ions a e included (n= 3949). (DOCX
20 kb)
Acknowledgemen s
No applicable.
Funding
This s udy was suppo ed by Academy o Finland (g an numbe 286294 and
294154 o SS , g an numbe 129364 o EVJ, g an numbe 311492 o MK);
Finnish Minis y o Educa ion and Cul u e ( o SS ); The Finnish Wo k
En i onmen Fund (g an numbe 109364 o SSo); Juho Vainio Founda ion
( o SS ); Medical Resea ch Council (g an numbe K013351 o MK); and
No dFo sk, he No dic P og amme o Heal h and Wel a e (g an numbe
75021 o MK, g an numbe 76679 o SSo).
A ailabili y o da a and ma e ials
Da a o esea ch pu poses a e a ailable upon eques .
Au ho s’con ibu ions
SS and JV concep ualized he s udy. VA compiled he da a. SS led he
s a is ical analyses and d a ed he manusc ip . All au ho s con ibu ed
meaning ully o he in e p e a ion o analyses and e isions o he
manusc ip . All au ho s ha e ead and app o ed he inal manusc ip .
E hics app o al and consen o pa icipa e
The FPS s udy was app o ed by he E hics Commi ee o he Hospi al
Dis ic o Helsinki and Uusimaa. The pa icipan s ga e hei in o med
consen o ake pa when esponding o he ques ionnai es.
Consen o publica ion
No applicable.
Compe ing in e es s
The au ho s decla e ha hey ha e no compe ing in e es s.
Publishe ’sNo e
Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in
published maps and ins i u ional a ilia ions.
Au ho de ails
1
Depa men o Public Heal h, Uni e si y o Tu ku and Tu ku Uni e si y
Hospi al, Tu ku, Finland.
2
Facul y o Social Sciences (Heal h Science),
Uni e si y o Tampe e, Tampe e, Finland.
3
Finnish Ins i u e o Occupa ional
Heal h, Helsinki, Finland.
4
Depa men o Epidemiology and Public Heal h,
Uni e si y College London Medical School, London, UK.
5
Clinicum, Facul y o
Medicine, Uni e si y o Helsinki, Helsinki, Finland.
Recei ed: 14 Decembe 2016 Accep ed: 14 June 2017
Re e ences
1. Eke d DJ. F on ie s o esea ch on wo k and e i emen . J Ge on ol B
Psychol Sci Soc Sci. 2010;65B(1):69–80.
2. S enholm S, Pulakka A, Kawachi I, Oksanen T, Halonen JI, Aal o V, e al.
Changes in physical ac i i y du ing ansi ion o e i emen : a coho s udy.
In J Beha Nu Phys Ac . 2016;13(1):51.
3. Sjös en N, Ki imäki M, Singh-Manoux A, Fe ie JE, Goldbe g M, Zins M, e al.
Change in physical ac i i y and weigh in ela ion o e i emen : he F ench
GAZEL Coho S udy. BMJ Open. 2012;2:e000522–2011. P in 2012
4. Halonen JI, S enholm S, Pulakka A, Kawachi I, Aal o V, Pen i J, Lallukka T,
Vi anen M, Vah e a J, Ki imäki M. T ajec o ies o isky d inking a ound he
S enholm e al. In e na ional Jou nal o Beha io al Nu i ion and Physical Ac i i y (2017) 14:85 Page 7 o 8
ime o s a u o y e i emen : a longi udinal la en class analysis. Addic ion.
2017;112(7):1163–70.
5. Zins M, Gueguen A, Ki imaki M, Singh-Manoux A, Lecle c A, Vah e a J, e al.
E ec o e i emen on alcohol consump ion: longi udinal e idence om
he F ench Gazel coho s udy. PLoS One. 2011;6(10):e26531.
6. Eng SM, Gammon MD, Te y MB, Kushi LH, Tei elbaum SL, B i on JA, e al.
Body size changes in ela ion o pos menopausal b eas cance among
women on Long Island, New Yo k. Am J Epidemiol. 2005;162(3):229–37.
7. Fine JT, Coldi z GA, Coakley EH, Moseley G, Manson JE, Wille WC, e al.
A p ospec i e s udy o weigh change and heal h- ela ed quali y o li e in
women. J Am Med Assoc. 1999;282(22):2136–42.
8. Wanname hee SG, Shape AG, Walke M. O e weigh and obesi y and
weigh change in middle aged men: impac on ca dio ascula disease and
diabe es. J Epidemiol Communi y Heal h. 2005;59(2):134–9.
9. Jenkins KR. Body-weigh change and physical unc ioning among young
old adul s. J Aging Heal h. 2004;16(2):248–66.
10. Wijnho en HA, an Zon SK, Twisk J, Visse M. A ibu ion o causes o
weigh loss and weigh gain o 3-yea mo ali y in olde adul s: esul s om
he Longi udinal Aging S udy Ams e dam. J Ge on ol A Biol Sci Med Sci.
2014;69(10):1236–43.
11. Rejeski WJ, Ma sh AP, Chmelo E, Rejeski JJ. Obesi y, in en ional weigh loss
and physical disabili y in olde adul s. Obes Re . 2010;11(9):671–85.
12. Goldman D, Lakdawalla D, Zheng Y. Re i emen and weigh mimeo 2008.
13. Gueo guie a R, Sindela JL, Wu R, Gallo WT. Di e en ial changes in body
mass index a e e i emen by occupa ion: hie a chical models. In J Public
Heal h. 2011;56(1):111–6.
14. Chung S, Domino ME, S ea ns SC. The e ec o e i emen on weigh .
J Ge on ol B Psychol Sci Soc Sci. 2009;64(5):656–65.
15. Fo man-Ho man VL, Richa dson KK, Yankey JW, Hillis SL, Wallace RB, Wolinsky
FD. Re i emen and weigh changes among men and women in he heal h
and e i emen s udy. J Ge on ol B Psychol Sci Soc Sci. 2008;63(3):S146–53.
16. Nooyens AC, Vissche TL, Schui AJ, an Rossum CT, Ve schu en WM, an
Mechelen W, e al. E ec s o e i emen on li es yle in ela ion o changes in
weigh and wais ci cum e ence in Du ch men: a p ospec i e s udy. Public
Heal h Nu . 2005;8(8):1266–74.
17. Goda d M. Gaining weigh h ough e i emen ? Resul s om he SHARE
su ey. J Heal h Econ. 2016;45:27–46.
18. Dug a o A, Sabia S, S inghini S, Ki imäki M, Wes e lund H, Vah e a J, e al.
Do socioeconomic ac o s shape weigh and obesi y ajec o ies o e he
ansi ion om midli e o old age? Resul s om he F ench GAZEL coho
s udy. Am J Clin Nu . 2010;92(1):16–23.
19. T ansame ica Cen e o Re i emen S udies. 17 hAnnual T ansame ica
Re i emen Su ey. In: A Compendium o Findings Abou Ame ican Wo ke s;
2016.
20. Ki imäki M, Hame M, Ba y GD, Geddes JR, Tabak AG, Pen i J, e al.
An idep essan medica ion use, weigh gain, and isk o ype 2 diabe es: a
popula ion-based s udy. Diabe es Ca e. 2010;33(12):2611–6.
21. Suoy jö H, Oksanen T, Hinkka K, Ki imäki M, Klaukka T, Pen i J, e al. The
e ec i eness o oca ionally o ien ed mul idisciplina y in e en ion on
sickness absence and ea ly e i emen among employees a isk: an
obse a ional s udy. Occup En i on Med. 2009;66(4):235–42.
22. Wo ld Heal h O ganiza ion. Obesi y: P e en ing and managing he global
epidemic. Repo o a WHO Consul a ion. Gene a: WHO Technical Repo
Se ies 894, 2000 Con ac No.: Repo .
23. S a is ics Finland. Classi ica ion o occupa ions 2001 2001 [ci ed A ailable a :
h p://www.s a . i/me a/luoki ukse /amma i/001-2001/index_en.h ml.
Accessed 17 Oc 2016.
24. Solo ie a S, Pehkonen I, Kaus o J, Mi anda H, Shi i R, Kauppinen T, e al.
De elopmen and alida ion o a job exposu e ma ix o physical isk
ac o s in low back pain. PLoS One. 2012;7(11):e48680.
25. A omaa A, Koskinen S. Heal h and unc ional capaci y in Finland. In: Baseline
esul s o he Heal h 2000 heal h examina ion su ey. Helsinki: Publica ions
o he Na ional Public Heal h Ins i u e B12/2004; 2004. p. 175. A ailable a
h ps://www.julka i. i/handle/10024/78534.
26. Kujala UM, Kap io J, Sa na S, Kosken uo M. Rela ionship o leisu e- ime
physical ac i i y and mo ali y: he Finnish win coho . J Am Med Assoc.
1998;279(6):440–4.
27. Ainswo h BE, Haskell WL, He mann SD, Meckes N, Basse DR J , Tudo -
Locke C, e al. 2011 Compendium o Physical Ac i i ies: a second upda e o
codes and MET alues. Med Sci Spo s Exe c. 2011;43(8):1575–81.
28. Halonen JI, Ki imaki M, Pen i J, Vi anen M, Sub amanian SV, Kawachi I,
e al. Associa ion o he a ailabili y o bee , wine, and liquo ou le s wi h
be e age-speci ic alcohol consump ion: a coho s udy. Alcohol Clin Exp
Res. 2014;38(4):1086–93.
29. Ka asek R, B isson C, Kawakami N, Hou man I, Bonge s P, Amick B. The Job
Con en Ques ionnai e (JCQ): an ins umen o in e na ionally compa a i e
assessmen s o psychosocial job cha ac e is ics. J Occup Heal h Psychol.
1998;3(4):322–55.
30. F ansson EI, Nybe g ST, Heikkila K, Al edsson L, Bacque de D, Ba y GD,
e al. Compa ison o al e na i e e sions o he job demand-con ol scales
in 17 Eu opean coho s udies: he IPD-Wo k conso ium. BMC Public
Heal h. 2012;12:62.
31. Zege SL, Liang KY. Longi udinal da a analysis o disc e e and con inuous
ou comes. Biome ics. 1986;42:121–30.
32. Diggle PJ, Liang KY, Zege SL. Analysis o Longi udinal Da a. Ox o d, U.K:
Ox o d Uni e si y P ess; 1994.
33. McMinn J, S eel C, Bowman A. In es iga ion and managemen o
unin en ional weigh loss in olde adul s. BMJ. 2011;342:d1732.
34. Rose G. Sick indi iduals and sick popula ions. In J Epidemiol. 1985;14(1):32–8.
35. Rose G. Ancel Keys Lec u e. Ci cula ion. 1991;84(3):1405–9.
36. Conno Go be S, T emblay M, Mohe D, Go be B. A compa ison o di ec
s. sel - epo measu es o assessing heigh , weigh and body mass index:
a sys ema ic e iew. Obes Re . 2007;8(4):307–26.
37. Kuczma ski MF, Kuczma ski RJ, Najja M. E ec s o age on alidi y o sel -
epo ed heigh , weigh , and body mass index: indings om he Thi d
Na ional Heal h and Nu i ion Examina ion Su ey, 1988-1994. J Am Die
Assoc. 2001;101(1):28–34. quiz 5-6
38. Bu khause R, Cawley J. The Impo ance o Objec i e Heal h Measu es in
P edic ing Ea ly Receip o Social Secu i y Bene i s: The Case o Fa ness. Ann
A bo : Wo king Pape s wp148, Uni e si y o Michigan, Michigan Re i emen
Resea ch Cen e ; 2006.
• We accep p e-submission inqui ies
• Ou selec o ool helps you o ind he mos ele an jou nal
• We p o ide ound he clock cus ome suppo
• Con enien online submission
• Tho ough pee e iew
• Inclusion in PubMed and all majo indexing se ices
• Maximum isibili y o you esea ch
Submi you manusc ip a
www.biomedcen al.com/submi
Submi you nex manusc ip o BioMed Cen al
and we will help you a e e y s ep:
S enholm e al. In e na ional Jou nal o Beha io al Nu i ion and Physical Ac i i y (2017) 14:85 Page 8 o 8