Body composition and changes in health-related quality of life in older age : a 10-year follow-up of the Helsinki Birth Cohort Study

This is a sel -a chi ed e sion o an o iginal a icle. This e sion
may di e om he o iginal in pagina ion and ypog aphic de ails.
Au ho (s):
Ti le:
Yea :
Ve sion:
Copy igh :
Righ s:
Righ s u l:
Please ci e he o iginal e sion:
CC BY 4.0
h ps://c ea i ecommons.o g/licenses/by/4.0/
Body composi ion and changes in heal h- ela ed quali y o li e in olde age : a 10-yea
ollow-up o he Helsinki Bi h Coho S udy
© 2020 he Au ho s
Published e sion
Mikkola, Tuija M.; Kau iainen, Hannu; on Bonsdo , Mikaela B.; Salonen, Minna
K.; Wasenius, Niko; Kajan ie, Ee o; E iksson, Johan G.
Mikkola, T. M., Kau iainen, H., on Bonsdo , M. B., Salonen, M. K., Wasenius, N., Kajan ie, E., &
E iksson, J. G. (2020). Body composi ion and changes in heal h- ela ed quali y o li e in olde age
: a 10-yea ollow-up o he Helsinki Bi h Coho S udy. Quali y o Li e Resea ch, 29(8), 2039-
2050. h ps://doi.o g/10.1007/s11136-020-02453-1
2020
Vol.:(0123456789)
1 3
Quali y o Li e Resea ch
h ps://doi.o g/10.1007/s11136-020-02453-1
Body composi ion andchanges inheal h‑ ela ed quali y o li e inolde
age: a10‑yea ollow‑up o  heHelsinki Bi h Coho S udy
TuijaM.Mikkola1,2 · HannuKau iainen1,3· MikaelaB. onBonsdo 1,4· MinnaK.Salonen1,5· NikoWasenius1,6·
Ee oKajan ie5,7,8,9· JohanG.E iksson1,6,10,11
Accep ed: 17 Feb ua y 2020
© The Au ho (s) 2020
Abs ac
Pu pose Mos s udies examining he associa ions be ween body composi ion and heal h- ela ed quali y o li e (HRQoL) in
olde age ha e been c oss-sec ional and analyzed only a o lean mass. Hence, i is poo ly known whe he a and lean mass
a e independen ly associa ed wi h subsequen changes in HRQoL. We in es iga ed whe he baseline lean and a mass a e
associa ed wi h changes in HRQoL o e a 10-yea pe iod in olde adul s.
Me hods We s udied 1044 men and women om he Helsinki Bi h Coho S udy (age 57–70yea s a baseline). Bioelec i-
cal impedance analysis was used o de i e baseline a mass index (FMI, a mass/heigh 2) and lean mass index (lean mass/
heigh 2), dicho omized a sex-speci ic medians. HRQoL was assessed using RAND 36-i em Heal h Su ey a baseline and
ollow-up 10 yea s la e .
Resul s When con olled o lean mass and adjus ed o po en ial con ounde s, high baseline FMI was associa ed wi h a
g ea e decline in gene al heal h (s anda dized eg ession coe icien [β] = − 0.13, p = 0.001), physical unc ioning (β = − 0.11,
p = 0.002), ole physical (β = − 0.13, p = 0.003), i ali y (β = − 0.08, p = 0.027), ole emo ional (β = − 0.12, p = 0.007), and
physical componen sco e (β = − 0.14, p < 0.001). High baseline FMI was also associa ed wi h low HRQoL in all physical
domains a baseline (β: om − 0.38 o − 0.10). Lean mass was no s ongly associa ed wi h HRQoL a baseline o change
in HRQoL.
Conclusion In olde communi y-dwelling adul s, highe a mass is, independen o lean mass, associa ed wi h lowe physi-
cal HRQoL and g ea e decline in HRQoL. P e en ion o adiposi y may con ibu e o p ese a ion o a good quali y o li e
in olde age.
Keywo ds Heal h- ela ed quali y o li e· Aging· Body composi ion· Obesi y· Lean mass· Fa mass
* Tuija M. Mikkola
uija.mikkola@ olkhalsan. i
1 Folkhälsan Resea ch Cen e , Helsinki, Finland
2 Clinicum, Facul y o Medicine, Uni e si y o Helsinki,
Helsinki, Finland
3 P ima y Heal h Ca e Uni , Kuopio Uni e si y Hospi al,
Kuopio, Finland
4 Ge on ology Resea ch Cen e , Facul y o Spo andHeal h
Sciences, Uni e si y o Jy äskylä, Jy askyla, Finland
5 Public Heal h P omo ion Uni , Na ional Ins i u e o Heal h
andWel a e, Helsinki, Finland
6 Depa men o Gene al P ac ice andP ima y Heal h Ca e,
Uni e si y o Helsinki andHelsinki Uni e si y Hospi al,
Helsinki, Finland
7 PEDEGO Resea ch Uni , MRC Oulu, Oulu Uni e si y
Hospi al andUni e si y o Oulu, Oulu, Finland
8 Depa men o Clinical andMolecula Medicine, No wegian
Uni e si y o Science andTechnology, T ondheim, No way
9 Child en’s Hospi al, Helsinki Uni e si y Hospi al
andUni e si y o Helsinki, Helsinki, Finland
10 Singapo e Ins i u e o Clinical Sciences, Agency o Science,
Technology, andResea ch, Singapo e, Singapo e
11 Depa men o Obs e ics & Gynaecology, Yong Loo Lin
School o Medicine, Na ional Uni e si y o Singapo e,
Singapo e, Singapo e
Quali y o Li e Resea ch
1 3
In oduc ion
Obesi y is globally a g owing heal h conce n among
olde adul s, and i is associa ed wi h a numbe o physi-
cal heal h p oblems in olde age [1–3]. I inc eases he
isk o many non-communicable diseases, such as os eoa -
h i is, ca dio ascula disease (CVD), and ype 2 diabe es
[4], which may u he lead o educ ion in heal h- ela ed
quali y o li e (HRQoL). Excess a may con ibu e o he
de elopmen o mobili y limi a ions [5] subsequen ly lead-
ing o dec ease in physical HRQoL, since mobili y abili y
is an essen ial componen o physical HRQoL. Lean mass,
which is mos ly composed o skele al muscle issue, has
heal h implica ions al hough i s impac on heal h appea o
be less s iking han ha o a mass in he gene al popu-
la ion [6]. Howe e , he impo ance o lean mass is p o-
nounced in olde age. Muscle mass ends o dec ease wi h
aging [7] and may e en ually each a c i ical h eshold
whe e i s le el is oo low o managing daily ac i i ies
independen ly [8]. Lean mass can be used as a ma ke o
nu i ional s a us and i has been epo ed o be associa ed
wi h leng h o s ay in hospi al [9]. Sa copenia, ha is low
muscle mass co-occu ing wi h low muscle s eng h o
low physical pe o mance [10], has been epo ed o be
associa ed wi h wo se o e all HRQoL in olde adul s [11].
Howe e , no all s udies suppo his inding [12].
A numbe o s udies ha e examined he associa ions
be ween obesi y, based on BMI, and HRQoL among
olde adul s [13–16]. O e weigh and obesi y ha e been
epo ed o be associa ed wi h poo e o e all HRQoL [16]
and physical HRQoL [13–15] bu no wi h men al HRQoL
[14, 15]. Howe e , al hough BMI is a use ul measu e o
o e weigh and obesi y in he gene al adul popula ion,
p edic ing many heal h indica o s [4, 17], i is less appli-
cable o olde adul s [18, 19]. Body composi ion unde -
goes ma ked changes wi h aging; he e is a loss in muscle
mass, while he p opo ion o a issue inc eases and a is
edis ibu ed in he body [20–22] making body composi-
ion o an a e age olde adul e y di e en om ha o an
a e age younge adul . Fu he mo e, i espec i e o age,
he e is a la ge a ia ion in he ela i e p opo ions o a
and lean mass among pe sons wi h he same BMI. This
calls o use o mo e accu a e me hods o quan i y body
composi ion and body compa men s in s udies ocusing
upon olde adul s. Mos p e ious s udies in es iga ing he
associa ions be ween body compa men s and HRQoL in
olde adul s ha e examined ei he a o lean mass and
ha e no aken in o accoun hei mu ual e ec s [23–25].
This may cause con ounding bias as a and lean mass a e
closely co ela ed; hose wi h high a mass end o also
ha e high lean mass. To a oid con ounding in he analysis,
i is essen ial o include bo h a mass and lean mass in he
same analysis o yield hei independen e ec s and also
hei possible in e ac ions.
A as majo i y o p e ious s udies ha e analyzed he
associa ions o a and lean mass wi h HRQoL in c oss-sec-
ional se ings and only ew s udies ha e assessed whe he
body composi ion is associa ed wi h change in HRQoL in
olde age [26, 27]. Hence, he pu pose o his s udy was o
examine c oss-sec ional associa ions o body composi ion
wi h physical and men al HRQoL and o assess whe he
baseline body composi ion is associa ed wi h subsequen
change in HRQoL du ing a 10-yea pe iod among olde
adul s.
Me hods
Sample
The Helsinki Bi h Coho S udy (HBCS) consis s o 13,345
single ons, who we e bo n in Helsinki in 1934–1944 and
we e s ill ali e in 1971. A o al o 8760 indi iduals we e
bo n in he Helsinki Uni e si y Cen al Hospi al and o hese
people, 2902 indi iduals we e andomly selec ed o pa ici-
pa e in a clinical examina ion in 2001–2004 o each a a ge
o 2000 pa icipan s. In o al, 2003 indi iduals pa icipa ed
in he baseline clinical examina ion. The age ange o he
pa icipan s was 57–70yea s (mean 61yea s) a baseline.
Follow-up examina ions we e pe o med in 2011–2013
when he pa icipan s we e 67–79yea s old (mean 71yea s)
(n = 1094). O hese pa icipan s, body composi ion was no
ob ained om 39 pa icipan s a baseline, and hey we e
excluded om he analysis. Fu he , wo pa icipan s did
no e u n HRQoL ques ionnai e a baseline, and nine did no
e u n HRQoL ques ionnai e a ollow-up and we e, he e-
o e, excluded. Hence, 1044 pa icipan s wi h comple e da a
on body composi ion a baseline and HRQoL a bo h meas-
u emen poin s we e included in he p esen analysis. The
s udy design and assessmen s du ing he clinical isi s ha e
been desc ibed in de ail p e iously [28–30]. All measu e-
men s we e pe o med by ained s udy nu ses.
Body composi ion andan h opome y
Body composi ion was assessed a baseline by bioelec ical
impedance analysis using he InBody 3.0 eigh -pola ac ile
elec ode sys em (Biospace Co, L d, Seoul, Ko ea) [31]. The
ins umen es ima es lean body mass and body a mass by
segmen al mul i- equency (5, 50, 250, and 500kHz) analy-
sis. The measu emen s we e made wi h he subjec s and-
ing in ligh indoo clo hing on he ou oo elec odes on
he pla o m o he analyze and g ipping he wo palm and
humb elec odes. Be ween-day p ecision o InBody 3.0 has
been epo ed o be 2.7% [31]. When compa ed o DXA
Quali y o Li e Resea ch
1 3
a - ee mass, pe cen oo mean squa e e o o InBody 3.0
was 6% [31]. Fa and lean mass indices we e calcula ed as
ollows: a mass index (FMI, kg/m2) = a mass/heigh 2 and
lean mass index (LMI, kg/m2) = lean mass/heigh 2.
Heigh and weigh we e measu ed in ligh indoo clo h-
ing and wi hou shoes. Heigh was measu ed o he nea es
0.1cm using a Kawi s adiome e and weigh o he nea -
es 0.1kg using Seca Alpha 770 scales. Body mass index
(BMI) was calcula ed as weigh in kilog ams di ided by he
squa e o heigh in me e s. Wais ci cum e ence was meas-
u ed midway be ween he lowes ib and he iliac c es wice
using a so ape. The a e age o he wo measu emen s was
calcula ed.
Heal h‑ ela ed quali y o li e
Heal h- ela ed quali y o li e was assessed using he Finn-
ish alida ed e sion o he RAND 36-I em Hea h Su ey
e sion 1.0 ques ionnai e [32]. The RAND-36 is composed
o eigh domains: physical unc ioning (10 i ems), ole
limi a ions caused by physical heal h p oblems (4 i ems),
bodily pain (2 i ems), gene al heal h (5 i ems), ole limi a-
ions caused by emo ional p oblems (3 i ems), i ali y (4
i ems), men al heal h (5 i ems) and social unc ioning (2
i ems). The single i ems in he ques ionnai e we e coded
o ange be ween 0 and 100, wi h 100 ep esen ing he bes
le el o unc ioning o well-being. Domain sco es we e he
a e ages o he i ems. Domain sco e was se as missing i
mo e han hal o he i ems in a gi en domain was missing.
Physical (PCS) and men al heal h summa y sco es (MCS)
we e agg ega ed om he eigh domain sco es. Fi s , hese
domains we e s anda dized using he means and s anda d
de ia ions o he US e e ence popula ion (1990) [33]. Nex ,
he domains we e weigh ed using ac o sco e coe icien s
ob ained om he same e e ence popula ion and summed
o yield PCS and MCS. Finally, PCS and MCS we e s and-
a dized using a mean o 50 and a s anda d de ia ion o 10
[34]. The sub-scales o RAND-36 ha e high unidimension-
ali y and su icien ly high in e nal consis ency among olde
adul s, which makes RAND-36 a use ul measu e in s udies
o olde adul s [35].
Socioeconomic a iables
Da e o bi h was e ie ed om hospi al bi h eco ds and
was used o calcula e subjec s’ age a he s a o he ollow-
up. Yea s o educa ion we e de i ed om in o ma ion on he
le el o educa ion ob ained om he na ion-wide egis e o
S a is ics Finland. Income (co esponding o alue in eu os
in 2018) pe consump ion uni in 2000 was calcula ed as ol-
lows: household axable income di ided by he squa e oo
o he numbe o people in he household [36]. These da a
we e ob ained om S a is ics Finland and we e linked using
unique pe sonal iden i ica ion codes.
Long‑ e m illness
In o ma ion on se e e long- e m diseases was ex ac ed om
he na ional Ca e Regis e o Heal h Ca e ( o me Hospi al
Discha ge Regis e ) o he ime pe iod be ween Jan 1s 1971
and June 30 h 2000 and was linked using unique pe sonal
iden i ica ion codes. The egis e con ains in o ma ion on
all specialized ou pa ien isi s ( om 1998 onwa ds) and
hospi al inpa ien discha ges in Finland. F om he egis e ,
we aimed o iden i y se e e long- e m illnesses, which se i-
ously comp omise physical o men al unc ioning o he sub-
jec . Se e e long- e m illnesses om he ollowing disease
ca ego ies we e included: cance , ca dio ascula disease,
pulmona y, demen ia, psychia ic, neu ological, musculo-
skele al, kidney, gas oin es inal, and diabe es. Any se e e
long- e m illness a iable was coded as 1 (se e e long- e m
illness) and 0 (no se e e long- e m illness).
Li es yle a iables, blood p essu e, andblood lipids
In o ma ion on li es yle was ob ained using a ques ionnai e
adminis e ed a baseline (2001–2004). The pa icipan s we e
asked abou hei ma i al s a us, smoking, and alcohol con-
sump ion. Ma i al s a us and smoking we e dicho omized
(cohabi ing yes/no, cu en smoke yes/no). F equency
o alcohol consump ion was ecoded in o ou ca ego ies
(0/1–2 imes pe mon h/1–2 imes pe week/ > 2 imes pe
week). The pa icipan s also comple ed a alida ed Kuopio
Ischemic Hea Disease Risk Fac o S udy (KIHD) ques-
ionnai e o assess he du a ion, equency and in ensi y o
leisu e- ime physical ac i i y du ing he pas 12mon hs [37].
Fo each in ensi y g ade, ac i i y-speci ic me abolic equi a-
len (MET) alues we e used. To al leisu e- ime physical
ac i i y, including bo h non-condi ioning (e.g., housewo k)
and condi ioning (e.g., esis ance aining) physical ac i i y,
was compu ed and is exp essed in METhou s pe day.
Blood p essu e was measu ed om he igh a m while
he subjec was in a si ing posi ion and was eco ded as he
mean o wo successi e eadings om a me cu y sphyg-
momanome e . Blood samples we e d awn o he assess-
men o glucose and lipids. Se um choles e ol and iglyc-
e ide concen a ions we e measu ed wi h he use o s anda d
enzyma ic me hods. LDL choles e ol concen a ions we e
calcula ed using he F iedewald o mula [38].
S a is ical analysis
Fa and lean mass indices we e di ided acco ding o median
alues, sepa a ely o men and women, a e which ou body
composi ion ca ego ies we e c ea ed: (1) low FMI and low
Quali y o Li e Resea ch
1 3
LMI (LFLL), (2) low FMI and high LMI (LFHL), (3) high
FMI and low LMI (HFLL), (4) high FMI and highLMI
(HFHL). By di iding pa icipan s in o hese ca ego ies, we
aimed a educing con ounding be ween a and lean mass
indices. Median alues o FMI we e 17.8kg/m2 in women
and 6.0kg/m2 in men and o LMI 8.84kg/m2 in women
and 20.8kg/m2 in men, espec i ely. The associa ions o
body composi ion ca ego ies wi h con inuous backg ound
cha ac e is ics and HRQoL measu es we e analyzed using
gene al linea models. The associa ions o body composi ion
ca ego ies wi h ca ego ical backg ound cha ac e is ics we e
analyzed using (o dinal) logis ic eg ession analysis. In he
c oss-sec ional linea eg ession models wi h HRQoL meas-
u es as dependen a iables, con inuous FMI and LMI and
hei in e ac ion we e en e ed in o he models as independen
a iables. In he adjus ed models, sex, age, educa ion, smok-
ing, alcohol consump ion, physical ac i i y, and any se e e
long- e m illness we e se as co a ia es because hey po en-
ially a ec bo h body composi ion and HRQoL. Changes
in HRQoL measu es du ing he ollow-up we e calcula ed
as he absolu e di e ences be ween he ollow-up alues and
co esponding baseline alues. In he longi udinal eg ession
models, changes in HRQoL measu es we e i s eg essed
on con inuous baseline FMI and LMI and hei in e ac ion.
These models we e adjus ed o he same co a ia es as he
c oss-sec ional analyses and addi ionally o he co espond-
ing baseline HRQoL domain sco es. These c oss-sec ional
and longi udinal linea eg ession analyses we e epea ed
wi h dicho omized baseline FMI and LMI and hei in e ac-
ion as independen a iables. P edic i e ma gins we e com-
pu ed o he in e ac ion e m o yield adjus ed p edic ions
o HRQoL means o he ou body composi ion ca ego ies.
Boo s apping wi h 5000 epe i ions was used o calcula e
95% con idence in e als. S anda dized eg ession coe i-
cien s om models using con inuous FMI and LMI as inde-
penden a iables we e used as e ec size indica o s. The
alues o he s anda dized eg ession coe icien s abo e 0.10,
0.30, and 0.50 ep esen small, mode a e and la ge e ec
sizes, espec i ely [39]. The da a we e analyzed using S a a
15.1 (S a aCo p, College S a ion, TX, USA).
Resul s
The ela ionships be ween lean mass index and a mass
index and he median-spli ca ego ies o body composi ion
in women and men a e shown in Fig.1.
Compa ed o hose included in he analysis, he subjec s
who ook pa only in he baseline measu emen s, and hence
we e excluded om he p esen s udy, we e olde (mean
61.4 s. 60.7yea s, p < 0.001) and less educa ed (11.0 s.
11.7yea s, p < 0.001). The excluded coho membe s we e
mo e equen ly men (49% s. 43%, p = 0.008) and mo e
likely o ha e a high a mass index (55% s. 46%, p < 0.001)
compa ed o coho membe s who we e included in he
p esen s udy. Howe e , he e was an equal p opo ion o
Lean Mass Index (kg/m
2
)
10 15 20 25 30
m/gk(xednIssaM aF 2)
0
5
10
15
20
25
30
AB
CD
Women
= 0.71 (95% CI:0.67 o 0.75)
Lean Mass Index (kg/m
2
)
10 15 20 25 30
0
5
10
15
20
25
30
AB
CD
Men
= 0.56 (95% CI:0.49 o 0.62)
Fig. 1 Rela ionships be ween a mass index and lean mass index in
women (le panel) and men ( igh panel). Fa and lean mass indices
we e each di ided acco ding o sex-speci ic median alues in o wo
ca ego ies and based on hem, ou ca ego ies we e c ea ed: a lean
low a low, b lean high a low, c lean low a high, d lean high a
high

Quali y o Li e Resea ch
1 3
hose wi h a high lean mass index among he included and
excluded subjec s (50% s. 50%, p = 0.91).
Backg ound cha ac e is ics
Table1 shows he backg ound cha ac e is ics o he sub-
jec s g ouped acco ding o he ou body composi ion
ca ego ies de i ed based on dicho omized a mass index
and lean mass index. The ou body composi ion ca ego-
ies (LFLL, LFHL, HFLL, HFHL) did no di e wi h
espec o sex, income, ma i al s a us, smoking, physical
ac i i y, o al choles e ol concen a ion, o p e alence o
se e e long- e m illnesses o he han diabe es. Age was
somewha lowe in he LFHL ca ego y han in he o he
ca ego ies (p = 0.054). LDL choles e ol concen a ion
was lowes in he LFLL ca ego y and highes in he LFHL
ca ego y (p = 0.063). Wais ci cum e ence, BMI, blood
p essu e, and iglyce ides inc eased wi h inc easing a
and lean mass index, while HDL choles e ol concen a ion
dec eased wi h inc easing a and lean mass index. The
Table 1 Backg ound cha ac e is ics o he Helsinki Bi h Coho S udy pa icipan s in 2001–2004
BMI body mass index, MET me abolic equi alen , LDL low-densi y lipop o ein, HDL high-densi y lipop o ein
a Hommel’s p < 0.05
Fa mass index low Fa mass index high p- aluea
Lean mass index low
N = 377
Lean mass index high
N = 144
Lean mass index low
N = 144
Lean mass index high
N = 379
Women, n (%) 222 (59) 73 (51) 73 (51) 223 (59) 0.13
Age, yea s mean (SD) 60.6 (2.7) 60.3 (2.2) 61.1 (3.0) 60.9 (2.9) 0.054
Educa ion yea s, mean (SD) 12.0 (3.7) 12.3 (3.1) 11.6 (3.6) 11.1 (3.3) < 0.001
Income in 1000 eu os, mean
(SD)
60.5 (48.8) 56.1 (34.1) 55.2 (66.4) 53.3 (47.8) 0.25
Cohabi ing, n (%) 290 (77) 112 (78) 105 (73) 301 (79) 0.46
Cu en smoke , n (%) 69 (18) 35 (24) 31 (22) 64 (17) 0.22
Alcohol consump ion, n (%) < 0.001
No a all 24 (6) 1 (1) 10 (7) 21 (6)
1–2 imes/mon h 139 (37) 54 (38) 48 (34) 179 (48)
1–2 imes/week 151 (40) 52 (37) 53 (37) 130 (35)
> 2 imes/week 62 (16) 35 (25) 32 (22) 46 (12)
Daily physical ac i i y in
METhou s, mean (SD)
5.6 (3.6) 5.9 (4.1) 4.8 (3.2) 5.3 (3.8) 0.085
Wais ci cum e ence in cm,
mean (SD)
85 (9) 91 (7) 96 (8) 104 (11) < 0.001
BMI in kg/m2, mean (SD) 23.4 (1.8) 26.1 (1.4) 27.4 (1.6) 31.4 (3.5) < 0.001
Blood p essu e in mmHg
Sys olic, mean (SD) 139 (19) 141 (19) 145 (22) 149 (19) < 0.001
Dias olic, mean (SD) 85 (10) 87 (10) 90 (9) 91 (10) < 0.001
To al choles e ol in mmol/l,
mean (SD)
5.82 (0.94) 5.94 (0.92) 5.93 (1.15) 5.91 (1.11) 0.48
LDL in mmol/l, mean (SD) 3.53 (0.80) 3.73 (0.83) 3.66 (0.99) 3.63 (0.89) 0.063
HDL in mmol/l, mean (SD) 1.76 (0.45) 1.66 (0.44) 1.60 (0.41) 1.51 (0.40) < 0.001
T iglyce ides in mmol/l, mean
(SD)
1.20 (0.57) 1.22 (0.52) 1.47 (0.60) 1.70 (0.86) < 0.001
Se e e long- e m illnesses, n
(%)
Ca dio ascula 28 (7) 7 (5) 8 (6) 28 (7) 0.65
Pulmona y 5 (1) 2 (1) 1 (1) 5 (1) 0.96
Psychia ic 7 (2) 0 (0) 1 (1) 2 (1) 0.18
Neu ological 2 (1) 0 (0) 1 (1) 1 (0) 0.74
Musculoskele al 15 (4) 8 (6) 4 (3) 22 (6) 0.42
Diabe es 9 (2) 4 (3) 8 (6) 53 (14) < 0.001
Any illness 57 (15) 19 (13) 21 (15) 85 (22) 0.016
Quali y o Li e Resea ch
1 3
p opo ion o indi iduals wi h diabe es was highes in he
HFHL ca ego y.
C oss‑sec ional associa ions be weenbaseline body
composi ion andbaseline HRQoL
In he c oss-sec ional analyses, baseline FMI was nega i ely
associa ed wi h se e al domains o HRQoL (adjus ed o
sex, age, educa ion, smoking, alcohol consump ion, physical
ac i i y, any se e e long- e m illness), i.e., gene al heal h
(β = − 0.27, 95% con idence in e al [95% CI] − 0.35 o
− 0.18), physical unc ioning (β = − 0.38, 95% CI − 0.45 o
− 0.30), ole limi a ions caused by physical heal h p oblems
(β = − 0.18, 95% CI − 0.27 o − 0.09), i ali y (β = − 0.12,
95% CI − 0.21 o − 0.03), physical heal h componen sco e
(β = − 0.32, 95% CI − 0.40 o − 0.23), and bodily pain (β
− 0.10, 95% CI − 0.19 o − 0.01) (Table2, means o he ou
body composi ion ca ego ies in Table3). C ude and adjus ed
model es ima es di e ed only ma ginally, and, he e o e,
only adjus ed es ima es a e shown. Highe lean mass index
Table 2 S anda dized eg ession coe icien s o he adjus ed linea eg ession analyses wi h con inuous baseline a mass index (FMI), lean mass
index (LMI), and hei in e ac ion explaining heal h- ela ed quali y o li e (HRQoL, RAND-36) domains a baseline
Independen a iables included in he eg ession models: FMI, LMI, FMI × LMI in e ac ion, sex, age, educa ion, smoking, alcoholuse, physical
ac i i y and se e e long- e m illness
The alues o he s anda dized eg ession coe icien s abo e 0.10, 0.30, and 0.50 ep esen small, mode a e and la ge e ec sizes, espec i ely
HRQoL baseline Fa mass index Lean mass index FMI × LMI in e ac ion
β (95% CI) p β (95% CI) p β (95% CI) p
Gene al heal h − 0.27 (− 0.35 o − 0.18) < 0.001 0.13 (0.03 o 0.23) 0.009 − 0.03 (− 0.10 o 0.03) 0.29
Physical unc ioning − 0.38 (− 0.45 o − 0.30) < 0.001 0.07 (− 0.02 o 0.17) 0.12 0.01 (− 0.06 o 0.06) 0.98
Role physical − 0.18 (− 0.27 o − 0.09) < 0.001 0.07 (− 0.04 o 0.17) 0.21 0.03 (− 0.03 o 0.10) 0.30
Bodily pain − 0.10 (− 0.19 o − 0.01) 0.028 0.05 (− 0.05 o 0.16) 0.33 0.01 (− 0.05 o 0.08) 0.73
Men al heal h 0.01 (− 0.09 o 0.10) 0.90 0.01 (− 0.10 o 0.11) 0.88 − 0.04 (− 0.10 o 0.03) 0.25
Vi ali y − 0.12 (− 0.21 o − 0.03) 0.007 0.08 (− 0.02 o 0.19) 0.12 − 0.06 (− 0.12 o 0.01) 0.089
Role emo ional 0.01 (− 0.08 o 0.10) 0.81 − 0.08 (− 0.19 o 0.02) 0.12 − 0.04 (− 0.11 o 0.02) 0.20
Social unc ioning − 0.06 (− 0.16 o 0.03) 0.18 0.05 (− 0.05 o 0.16) 0.32 0.03 (− 0.03 o 0.10) 0.31
Physical componen sco e − 0.32 (− 0.40 o − 0.23) < 0.001 0.13 (0.03 o 0.22) 0.013 0.02 (− 0.04 o 0.08) 0.46
Men al componen sco e 0.07 (− 0.03 o 0.16) 0.16 − 0.03 (− 0.13 o 0.08) 0.63 − 0.05 (− 0.11 o 0.02) 0.17
Table 3 Means (95% con idence in e als) o heal h- ela ed quali y o li e (RAND-36) domains acco ding o body composi ion ca ego ies
Models a e adjus ed o sex, age, educa ion, smoking, alcohol use, physical ac i i y, and se e e long- e m illness
p- alues o he e ec o a mass index, lean mass index and hei in e ac ion
Fa mass index low Fa mass index high p a mass index p lean mass
index
p in e ac ion
Lean mass index
low
N = 377
Lean mass index
high
N = 144
Lean mass index
low
N = 377
Lean mass index
high
N = 144
Gene al heal h 69 (67–71) 68 (65–70) 64 (62–67) 65 (63–67) < 0.001 0.61 0.31
Physical unc-
ioning
89 (88–90) 89 (87–91) 86 (84–88) 82 (81–84) < 0.001 0.050 0.030
Role physical 88 (85–90) 88 (84–92) 84 (79–89) 81 (78–85) 0.009 0.51 0.49
Bodily pain 80 (78–83) 78 (75–82) 79 (75–83) 79 (77–81) 0.74 0.43 0.52
Men al heal h 82 (80–83) 82 (80–84) 82 (79–84) 82 (80–83) 0.97 0.96 0.94
Vi ali y 72 (70–74) 74 (71–77) 72 (69–75) 71 (69–73) 0.15 0.75 0.23
Role emo ional 87 (84–90) 85 (81–90) 88 (83–92) 86 (83–88) 0.73 0.35 0.92
Social unc ion-
ing
91 (89–93) 92 (89–94) 91 (89–94) 90 (89–92) 0.67 0.77 0.37
Physical compo-
nen sco e
51 (50–52) 51 (49–52) 49 (48–50) 48 (47–50) < 0.001 0.29 0.58
Men al compo-
nen sco e
54 (53–55) 54 (53–55) 55 (53–56) 54 (54–55) 0.29 0.94 0.73
Quali y o Li e Resea ch
1 3
was associa ed wi h be e gene al heal h (β = 0.13, 95%
CI 0.03 o 0.23) and a highe physical componen sco e
(β = 0.13, 95% CI 0.03 o 0.22) a baseline. The e we e no
clea indica ions ha baseline body composi ion would be
associa ed wi h baseline men al heal h, ole limi a ions
caused by men al heal h p oblems, social unc ioning o
men al componen sco e.
Longi udinal associa ions be weenbaseline body
composi ion andchange inHRQoL
The e we e no in e ac ions o FMI and LMI on HRQoL
a iables. Adjus ed o he baseline co a ia es (sex, age,
educa ion, smoking, alcohol consump ion, physical ac i -
i y, any se e e long- e m illness, and co esponding baseline
HRQoL a iable), highe FMI was associa ed wi h g ea e
decline in gene al heal h (β = − 0.13, 95% CI − 0.20 o
− 0.06), physical unc ioning (β = − 0.11, 95% CI − 0.18 o
− 0.04), bodily pain (β = − 0.08, 95% CI − 0.16 o 0.01), ole
limi a ions caused by physical heal h p oblems (β = − 0.13,
95% CI − 0.22 o − 0.05), i ali y (β = − 0.08, 95% CI − 0.16
o − 0.01), ole limi a ions caused by emo ional p oblems
(β = − 0.12, 95% CI − 0.20 o − 0.03), and physical compo-
nen sco e (β = − 0.14, 95% CI − 0.22 o − 0.07) (Table4,
means o he ou body composi ion ca ego ies in Figs.2
and 3). The changes in physical componen sco e in he
ou body composi ion ca ego ies we e he ollowing: − 1.3
(LFLL), − 1.8 (LFHL), − 2.4 (HFLL), and − 3.7 (HFHL)
poin s, espec i ely (Fig.3). FMI was no associa ed wi h
changes in he es o he HRQoL measu es. Lean mass
index was no associa ed wi h changes in HRQoL measu es.
Discussion
We examined he associa ions o body composi ion wi h
HRQoL and i s change du ing a 10 yea pe iod in olde men
and women simul aneously adjus ing o a and lean mass.
Compa ed o hose wi h low a mass index, hose wi h high
a mass index had lowe sco es in se e al physical heal h
domains a baseline. They also showed la ge declines in all
physical heal h domains, and in he i ali y and emo ional
ole domains. Highe lean mass index was only associa ed
wi h be e gene al heal h and a highe physical componen
sco e a baseline.
Al hough se e al c oss-sec ional s udies ha e examined
he ela ionships be ween body composi ion and HRQoL in
olde adul s [23–25, 40], only ew ha e u ilized a longi udi-
nal design. In he p esen s udy, se e al domains o HRQoL
declined in all ou body composi ion ca ego ies o e he
10-yea ollow-up. These we e physical unc ioning, ole
limi a ions caused by physical heal h p oblems, and bodily
pain. We obse ed mean declines o 1.3–3.7 poin s in he
physical composi e sco e (PCS) wi hin he ou body com-
posi ion ca ego ies in he p esen s udy. A p e ious s udy
epo ed a mean decline o abou 4 poin s in PCS om he
age 60 o 70yea s, an age pe iod compa able o ha in he
p esen s udy [41]. The la ges declines we e obse ed in
ole limi a ions caused by physical heal h p oblems bu also
he a iance o change was la ges in his domain.
Ou indings a e in ag eemen wi h a s udy epo ing ha
a highe a mass (assessed using dual-ene gy x- ay abso p-
iome y, DXA) was associa ed wi h a g ea e decline in
SF-36 physical unc ioning domain o e a 3-yea ollow-up
among olde women [26]. Tha s udy did no epo o he
Table 4 S anda dized eg ession coe icien s o adjus ed linea eg ession analyses wi h con inuous baseline a mass index (FMI), lean mass
index (LMI), and hei in e ac ion explaining changes in heal h- ela ed quali y o li e (HRQoL, RAND-36) domains
Independen a iables included in he eg ession models: FMI, LMI, FMI × LMI in e ac ion, sex, age, educa ion, smoking, alcoholuse, physical
ac i i y, se e e long- e m illness, and co esponding HRQoL domain a baseline
The alues o he s anda dized eg ession coe icien s abo e 0.10, 0.30, and 0.50 ep esen small, mode a e and la ge e ec sizes, espec i ely
HRQoL change Fa mass index Lean mass index FMI × LMI in e ac ion
β (95% CI) p β (95% CI) p β (95% CI) p
Gene al heal h − 0.13 (− 0.20 o − 0.06) 0.001 − 0.01 (− 0.09 o 0.08) 0.90 − 0.01 (− 0.06 o 0.05) 0.84
Physical unc ioning − 0.11 (− 0.18 o − 0.04) 0.002 − 0.01 (− 0.09 o 0.07) 0.80 − 0.01 (− 0.05 o 0.05) 0.94
Role physical − 0.13 (− 0.22 o − 0.05) 0.003 0.01 (− 0.09 o 0.11) 0.82 0.01 (− 0.05 o 0.07) 0.67
Bodily pain − 0.08 (− 0.16 o 0.01) 0.058 − 0.07 (− 0.17 o 0.02) 0.12 0.05 (− 0.01 o 0.11) 0.075
Men al heal h − 0.02 (− 0.10 o 0.06) 0.57 − 0.07 (− 0.16 o 0.02) 0.15 − 0.03 (− 0.08 o 0.01) 0.33
Vi ali y − 0.08 (− 0.16 o − 0.01) 0.027 − 0.04 (− 0.12 o 0.05) 0.37 0.02 (− 0.03 o 0.08) 0.41
Role emo ional − 0.12 (− 0.20 o − 0.03) 0.007 0.04 (− .06 o 0.14) 0.40 − 0.01 (− 0.06 o 0.06) 0.94
Social unc ioning − 0.06 (− 0.14 o 0.03) 0.19 − 0.03 (− 0.13 o 0.07) 0.51 − 0.04 (− 0.10 o 0.02) 0.24
Physical componen sco e − 0.14 (− 0.22 o − 0.07) < 0.001 − 0.02 (− 0.11 o 0.06) 0.58 0.02 (− 0.03 o 0.08) 0.43
Men al componen sco e − 0.03 (− 0.11 o 0.05) 0.49 − 0.02 (− 0.12 o 0.08) 0.68 − 0.03 (− 0.08 o 0.03) 0.40
Quali y o Li e Resea ch
1 3
SF-36 domains. Fu he , hey epo ed a U-shaped asso-
cia ion be ween lean mass and change in physical unc ion-
ing; hose in he lowes and highes qua ile o lean mass
showed he g ea es decline in physical unc ioning. When
hese analyses we e adjus ed o a mass, he associa ion
o lean mass wi h physical unc ioning a enua ed sugges -
ing con ounding by a mass. Howe e , mos s udies ha e
igno ed a mass when s udying lean mass and ice e sa. A
s udy including 65yea s and olde indi iduals epo ed no
associa ions be ween change in a EQ-5D index o e 3 yea s
and a ange o obesi y measu es, such as BMI, wais ci -
cum e ence, and a pe cen age, al hough in c oss-sec ional
analysis almos all measu es we e associa ed wi h HRQoL
[27]. Ano he s udy explo ed HRQoL ajec o ies among
o e 4000 women, who we e 60–79yea s a baseline. The
esul s sugges ed ha obese women we e mo e likely han
o he s o be in a ajec o y ha had pe sis en ly low HRQoL
index o e he 7-yea ollow-up [42]. Acco ding o Cohen’s
Low High
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
2
4
6PF
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
2
4
6GH
LMI low
LMI high
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
2
4
6VT
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
2
4
6MH
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
2
4
6RP
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
2
4
6RE
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
2
4
6SF
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
2
4
6BP
Low High
FMI
LowHigh
FMI
High
FMI
Low
FMI
Low High
FMI
Low High
FMI
Low High
FMI
LowHigh
FMI
(a)(b) (c) (d)
(e)( ) (g)(h)
FMI p=0.002
LMI p=0.35
In e ac ionp=0.72
FMI p=0.010
LMI p=0.20
In e ac ionp=0.57
FMI p=0.016
LMI p=0.36
In e ac ionp=0.61
FMI p=0.050
LMI p=0.055
In e ac ionp=0.19
FMI p=0.081
LMI p=0.29
In e ac ionp=0.38
FMI p=0.001
LMI p=0.57
In e ac ionp=0.93
FMI p=0.054
LMI p=0.83
In e ac ion p=0.45
FMI p=0.086
LMI p=0.33
In e ac ionp=0.95
Fig. 2 Adjus ed changes in RAND-36 domains ac oss body com-
posi ion ca ego ies. ‘LMI low’ e e s o lean mass index (lean body
mass/heigh 2) equal o below he sex-speci ic median and ‘LMI high’
o lean mass index abo e he sex-speci ic median. ‘FMI low’ and
‘FMI high’ a e de ined co espondingly. Models we e adjus ed o
sex, age, educa ion, smoking, alcohol consump ion, physical ac i i y,
any se e e long- e m illness, and co esponding baseline RAND-36
domain. GH gene al heal h, PF physical unc ioning, RP ole limi-
a ions caused by physical p oblems, BP bodily pain, MH men al
heal h, VT i ali y, RE ole limi a ions caused by emo ional p oblems,
SF social unc ioning