REVIEW

Birth weight and adult bone mass: a systematicliterature review

M. M. Schlüssel & J. dos Santos Vaz & G. Kac

Received: 5 November 2009 /Accepted: 1 March 2010# International Osteoporosis Foundation and National Osteoporosis Foundation 2010

AbstractSummary This systematic literature review comprised 16studies. The association of birth weight with bone param-eters was much more evident for bone mineral content(BMC) rather than bone mineral density (BMD). This is animportant finding since a reduction in BMC is stronglyassociated with an increased risk of fractures.Introduction The purpose of this study was to conduct asystematic literature review of studies that have investigated theassociation between birth weight (BW) and adult bone mass.Methods The search included English language articles,indexed in MEDLINE, using the key words: (“birth size”OR “birth weight” OR birthweight) AND (osteoporosis OR“bone mass” OR “bone density” OR “bone mineraldensity” OR “bone mineral content” OR “bone area”). Amethodological quality appraisal of the reviewed studieswas performed.Results Sixteen articles were reviewed. Eleven of 13studies that measured BMC verified a positive effect ofBW on this parameter, and nine even after adjustment foradult body size. Among the ten studies that found anunadjusted association between BW and BMD, two reportedthat the significance remained after adjustment for current

body size. Interaction between prenatal and postnatalvariables on the determination of adult bone mass was onlytested by two studies. The results must be interpreted withcaution due to the existence of few papers on the issue, aswell as heterogeneous sample characteristics, investigatedbone sites, and implemented analysis procedures. The aspectsof methodological quality that frequently fail are as follows:representativeness of the planned and actually measuredsample as well as proper adjustment for confounding.Conclusion Based on the amount of accumulated evidence,it is probable that BW have a positive association with adultBMC rather than BMD, which is an important finding forclinical and public health policies since a reduction in BMCis strongly associated with an increased fracture risk.

Keywords Birth weight . Bone mass . Epidemiology . Fetalprogram . Nutrition . Prospective studies

Introduction

Scientific evidence derived from prospective studiesdemonstrates that malnourished individuals during intra-uterine life, and consequently with lower weight at birth,presents a higher risk for coronary heart disease, hyper-tension, type 2 diabetes, and hypercholesterolemia in laterlife [1, 2]. Such studies, relating low weight at birth withthe development of chronic diseases in adulthood, origi-nated the “Developmental Origins of Health and Diseases”(DOHaD) theory [3].

In brief, the biological basis for this theory postulatesthat certain types of stimuli in critical moments ofdevelopment, particularly nutritional deprivation duringfetal growth, “program” the individual organism, perma-nently changing his structure and function [4, 5]. In other

M. M. Schlüssel : J. dos Santos Vaz :G. KacInstitute of Nutrition Josué de Castro,Federal University of Rio de Janeiro,Rio de Janeiro, Brazil

G. Kac (*)Department of Social and Applied Nutrition, Institute of NutritionJosué de Castro, Federal University of Rio de Janeiro,Avenida Brigadeiro Trompowsky, s/nº CCS – Bloco J – 2º andar,sala 29 Cidade Universitária – Ilha do Fundão,21941-590 Rio de Janeiro, RJ, Brazile-mail: kacetal@gmail.come-mail: gkac@nutricao.ufrj.br

Osteoporos IntDOI 10.1007/s00198-010-1236-z

words, a genotype may create different phenotypes,according to the (adverse) conditions to which it wasexposed during intrauterine development [6]. Such nutri-tional insults would be reflected, therefore, in the reducedsize (weight) at birth.

Recently, lower weight at birth has been related withseveral diseases such as depression, schizophrenia, hyper-thyroidism, cancer, and osteoporosis [2, 3]. In relation toosteoporosis, evidence derived from longitudinal studies(birth cohorts) has demonstrated that its risk might bemodified by early environmental influences reflected inbirth weight, which is positively associated with bone massduring adulthood [7, 8]. Other studies corroborate thehypothesis that osteoporosis may be programmed in earlydevelopment. According to Cooper et al. [9], these studiesmay be classified in three groups: 1) those that explored therelationship between the physiological system of individu-als who may have been “programmed” and the rate of boneloss due to aging; 2) studies that investigated the influenceof body constitution, nutrition, and mothers' lifestyle onbone mass of their offspring; and 3) studies relating thegrowth rate of individuals during childhood and adoles-cence with the risk of fractures in adulthood.

From a clinical point of view, considering that bone lossis a natural and irreversible process that occurs with aging,one of the best strategies for the prevention of osteoporosisis to optimize the peak bone mass of an individual [10–13].From an epidemiological point of view, the earlier the startof a preventive procedure, the better is its chance to achievedesirable results. Therefore, the objective of the currentstudy was to conduct a systematic literature review (SLR)of studies on the association between birth weight and adultbone mass, aiming to elucidate whether improving fetalnutrition and adequate intrauterine growth might be astrategy for early prevention of future osteoporosis.

Methods

The study design comprised an SLR conducted accordingto the most recent guidelines [14]. The search wasconducted through September 2009 and updated in January2010 on MEDLINE, using the following keywords: (“birthsize” OR “birth weight” OR birthweight) AND (osteopo-rosis OR “bone mass” OR “bone density” OR “bonemineral density” OR “bone mineral content” OR “bonearea”). Selection criteria for the studies included (1)published in English; (2) abstract available on-line; (3)refer to original work (instead of literature reviews); (4)include among one of the main investigated outcomes theestimate of adult bone mass, measured by some of thefollowing parameters: bone area (BA), bone mineralcontent (BMC), or bone mineral density (BMD); (5)

include the weight at birth among one of the investigateddeterminants; (6) studied humans; and (7) include onlyadults and/or elderly individuals in the analysis.

The selection process involved a search of references,initially based only on the key words, which retrieved 419articles. The list with the 419 articles was refined on-line,using as filters the following inclusion criteria: papers withavailable abstracts (403), published in English (382), andregarding original works (330).

The second stage of the selection process consisted of areading of the 330 remaining abstracts, based on which, theother inclusion criteria were verified. In the present work,16 studies were classified as appropriate for inclusion.Fig. 1, based on the proposal of the Preferred ReportingItems for Systematic reviews and Meta-Analyses (PRISMAGroup) [14], presents a flowchart with all stages of thestudy selection.

To conduct an appraisal of the studies' methodologicalquality, each of the selected articles was evaluated and

Note: Designed according to the proposal of the PRISMA Group14.

Fig. 1 Study search flow

Osteoporos Int

received a score, according to the criteria proposed byDowns and Black [15]. The original checklist was adapted,ruling out the criteria related exclusively to interventionstudies. Nineteen items were thus evaluated, allowing amaximum score of 20 points. Table 1 shows the itemsactually evaluated.

Each published paper was evaluated independently bytwo authors of the present SLR (MMS and JSV). Theagreement between evaluations was assessed using intra-class correlation coefficients (ICC). The result was classi-fied according to the scale proposed by Shrout [16] tomeasure agreement between different evaluators. This scaleconsists of five categories: virtually no agreement (<0.10),weak (0.11–0.40), reasonable (0.41–0.60), moderate (0.61–0.80), and substantial (0.81–1.00). To settle eventualdisagreements in assigned scores, a third author's opinion(GK) was consulted. No article was rejected due tomethodological limitations, but the findings of studies thatreceived higher scores were considered more consistent.

The studies were organized according to the followingfactors: year of publication, country where the study wasconducted, sample size, sex and subject age, instrumentused to measure bone mass, study design, study aims,parameters of bone mass evaluated and studied bone sites,main results, estimator used in the analysis, and themethodological quality score.

Finally, we grouped in a graph the studies that presentedsimilar results for a given parameter of bone mass, with theobjective of summarizing and facilitating understanding of

the review findings. For each investigated parameter (BA,BMC, and BMD), the possible associations with birthweight were attributed as follows: without association,unadjusted association (association only between birthweight and the specific parameter), or adjusted association(for body size in adulthood). Only studies that evaluated theindividuals' bone mass by DXA were considered.

Results

The majority of studies (13/16) were published after 2000and nine of them since 2005, which point to a recentgrowing trend of investigations on the issue. Most of thestudies originated from Europe (12/16) and, of these, sevenof 12 were conducted in England. The non-Europeanstudies were developed in the USA, Japan, New Zealand,and Brazil (Table 2).

Eleven of the reviewed studies included both men andwomen in their samples, and five studies included onlywomen. In relation to the individuals' ages, six studiesincluded young adults (18–27 years), three adults between30 and 36 years, and seven adults above 45 years. Of 16studies, 15 used DXA to estimate bone mass. Fourteenstudies referred to birth cohorts; one employed a cross-sectional design; and one, a randomized controlled trial(Table 2).

The more common source of information on birth weightwas the birth records, although in some cases this infor-

Table 1 Downs and Black [15] criteria used in the methodological quality evaluation of the reviewed studies

Clear description of study's hypothesis or objective(s)

Definition of the main outcomes in the introduction or methods section

Clear characteristics description of the individuals included in the study

Description of principal confounders

Clear description of the study's principal findings

Information on random data variability for the main outcomes

Characteristics description for the individuals lost to follow-up

Information on the real probability values for the main outcomes

Representativeness of the planned sample

Representativeness of the individuals actually included in the final sample

Clear description of results not based on a priori hypotheses (data dreading)

Adjustment of the analyses for different lengths of follow-up

Adequacy of the statistical tests used to evaluate the main outcomes

Accuracy of the main outcomes measures

Comparability of the individuals included in all comparison groups in relation to the population they were recruited from

Comparability of the individuals included in all comparison groups in relation to the period of time when they were recruited

Adequate adjustment for principal confounders in the analysis from which the main findings were drawn

Consideration of losses to follow-up in the analysis

Sufficient study power to detect an important effect, with a 5% significance level

All items scored 0 or 1, except “description of principal confounders”, which scored 0, 1, or 2

Osteoporos Int

mation has been obtained based on the individuals' mothers'memory (data not shown).

The sample size varied from 86 to 4,008 (Table 3). Itwas observed that birth weight presented an unadjustedpositive association with adult BMC. Among the 11 studiesthat found this result in at least one bone site, nine supportthat, although the effect has been reduced, the associationremains positive and significant even after adjustment forsome measure of adult body size (weight and/or height).However, only two studies reported a positive andsignificant association between low birth weight and adultBMD after adjustment for body size in adulthood. Only fivestudies had investigated BA as outcome (Table 3, Fig. 2).

The mean score for methodological quality among theselected studies was 15.7 with 10 being the minimum and19 the maximum. The comparison between scores assignedto the studies by the different evaluators (ICC = 0.845; 95%CI: 0.703–0.987) indicates a high level of concordance(substantial concordance) according to the qualificationscale used. The most recent studies (2005–2010) obtainedhigher scores, denoting greater attention to the dissemina-tion of results—or a higher rigorousness by journals inselecting the studies to be published. However, some issuessuch as “representativeness of the planned sample,”“representativeness of individuals effectively studied,” and“adjustment of the analyses for the main confounders”figured as the more frequent methodological flaws, beingobserved in almost all reviewed studies, independently ofyear of publication (Table 3).

Discussion

The evidence based on the scientific literature concerningthe association between birth weight and adult bone masshas increased in the last decade. According to severalauthors [17–27], individuals with lower weight at birthbecome adults with lower bone mass. This way, birthweight would act as a risk marker for future osteoporosis,as this disease is characterized by low bone mass anddeterioration of bone tissue micro-architecture, withsubsequent increase in fragility and susceptibility tofractures [28]. The association between birth weight andadult bone mass showed to be much more marked forBMC rather than for BMD, both important determinantsof the individual's fracture risk. Eleven of 13 studies thatmeasured BMC verified the positive effect of birth weighton this parameter, and nine even after adjustment for adultbody size. In this regard, it seems that intrauterine growthrestriction (IUGR) is clearly an important determinant ofadult bone mass [20–22, 24, 25, 27]. On the other hand,prematurity seems also to be important [26], although therelevance of gestational age among term and pretermremains to be studied. Despite the above results, thepresence of birth weight effects on bone mass from earlyadulthood is still controversial [7, 23, 26, 27, 29, 30].

The systematization of the results from the reviewedstudies must take several issues into consideration. In anattempt to fully address these points, we organized thediscussion in three main parts as follows: (a) issues

Table 2 Year of publication; samples' origin, size and age; bone measures assessment instrument; and design of the selected studies, 1993–2010

Author Year Origin Individual's age (years) Instrument Design

Hamed et al. [29] 1993 – 20–23 DXA Cross-sectional

Cooper et al. [7] 1995 Bath, England 21 DXA Cohort

Cooper et al. [8] 1997 Hertfordshire, England 63–73 DXA Cohort

Yarbrough et al. [17] 2000 California, USA 47–89 DXA Cohort

Gale et al. [18] 2001 Sheffield, England 70–75 DXA Cohort

Antoniades et al. [19] 2003 England 47.5a DXA Cohort

te Velde et al. [20] 2004 Amsterdam, Netherlands 36 DXA Cohort

Dennison et al. [21] 2005 Hertfordshire, England 60–75 DXA Cohort

Laitinen et al. [22] 2005 Oulu and Lapland, Finland 31 DXA/QUS Cohort

Pearce et al. [48] 2005 Newcastle, England 49–51 DXA Cohort

Saito et al. [23] 2005 Niigata, Japan 18–21 DXA Cohort

Dalziel et al. [24] 2006 Auckland, New Zealand 31a DXA RCT

Oliver et al. [25] 2007 Hertfordshire, England 59–71 pQCT Cohort

Leunissen et al. [30] 2008 Netherlands 18–24 DXA Cohort

Hovi et al. [26] 2009 Helsinki, Finland 18–27 DXA Cohort

Schlüssel et al. [27] 2010 Ribeirão Preto, Brazil 23–24 DXA Cohort

DXA dual X-ray absorptiometry; pQCT peripheral quantitative computed tomography; QUS quantitative ultra-sound; RCT randomized controlledtriala Mean age

Osteoporos Int

Tab

le3

Sum

maryof

stud

iesinvestigatingtheassociationbetweenbirthweigh

tandadultbo

nemass

Author

Objectiv

eParam

eters(bonesites)

Sam

plesize

Mainresults

(birth

weightandadultbone

mass)

Estim

ators

Score

a

Ham

edet

al.[29]

ToinvestigateifLBW

orprem

aturity

areassociated

with

low

bone

massin

adulthood

BMD

(lum

barspineand

femoral

neck)

230wom

enThere

wereno

significantdifferencesbetween

BMD

means

(fem

oral

neck

p=0.60

andspine

p=0.64)betweenLBW

individuals(<2.5kg)

andindividualsborn

with

2.5kg

ormore.

Birth

weightwas

notassociated

with

BMD

atspine(r=0.08;95%

CI:−0

.05–

0.21)or

femoral

neck

(r=0.10;95%

CI:−0

.03–0.23)

Group

difference

correlationcoefficient

(linearregression)

10

Cooperet

al.[7]

Toinvestigatetheassociation

betweenearlygrow

thandadults'

bone

mass

BMCandBMD

(lum

bar

spineandfemoral

neck)

153wom

enBirth

weightwas

notassociated

with

BMCor

BMD

atspine(r=0.12

and0.05,p>0.05;

respectiv

ely)

orfemoral

neck

(r=0.14

and

0.05,p>0.05,respectiv

ely)

Correlatio

ncoefficient

(linearregression)

15

Cooperet

al.[8]

Toinvestigatetherelatio

nship

betweenweightgain

inearlylife

andbone

massin

theseventh

decade

oflife

BMCandBMD

(lum

bar

spineandfemoral

neck)

189wom

enand

224men

There

wereno

significantdifferencesbetween

BMCmeans

atspine(p

fortrend=0.14

for

men

and0.056forwom

en)or

femoral

neck

(pfortrend=0.36

formen

and0.21

for

wom

en)accordingbirthweighttertiles.The

samewas

observed

forBMD

means

atspine

(pfortrend=0.95

formen

and0.14

for

wom

en)or

femoralneck

(pfortrend=0.54

for

men

and0.43

forwom

en)

Group

difference

16

Yarbrough

etal.[17]

Todescribe

therelatio

nship

betweenbirthweightandbone

massof

wom

enin

thepost-

menopausalperiod

BMCandBMD

(lum

bar

spine,

hip,

forearm,and

wrist)

305wom

enAfter

adjustmentsforage,

BMI,alcohol

consum

ption,

smokinghabit,physical

activ

ity,osteoarthritisdiagnosis,estrogen

repositio

ntherapy,

calcium

supplementatio

n,andmedicationuse;

BMCmeanat

thespine

ofwom

enin

thethirdtertile

ofbirthweight

(53.57

g)was

significantly

higher

than

those

atsecond

(51.09

g)or

first(48.89

g)tertiles;

pfortrend=0.005.

For

BMD,thistrendwas

only

observed

intheunadjusted

analysis

Group

difference

(linear

regression)

17

Galeet

al.[18]

Toinvestigatetherelatio

nship

betweenanthropometricmeasures

atbirthandadults'b

ody

compositio

n

BA,BMC,andBMD

(total

body,lumbarspine,

and

femoral

neck)

41wom

enand

102men

After

adjustmentsforsex,

age,andadulth

eight,

BMCmeans

ofindividualsin

thethirdtertile

ofbirthweightwerehigher

fortotalbody

(23.21

kg),femoralneck

(4.34g),and

lumbar

spine(69.44

g)than

thoseat

second

(22.71

kg;4

.28gand68.36g,

respectiv

ely)

orfirsttertile

(22.25

kg;4.11

g;65.25g,

respectiv

ely);pfortrend<0.05

forallbone

sites.How

ever,allassociations

losteffect

afterinclusionof

adultweightin

themodels.

For

BA

andBMD,thistrendwas

only

observed

intheunadjusted

analysis

Group

difference

(linear

regression)

16

Antoniadeset

al.[19]

Toinvestigatetheassociation

betweenbirthweightand

osteoporosis/osteoarthritis

inadult

wom

en

BMCandBMD

(lum

bar

spine,

hip,

femoral

neck,

andforearm)

4,008wom

enb

After

adjustmentsforadultweightandheight,

twinsintra-pairbirthweightdifferences

remainedpositiv

elyassociated

with

twins

intra-pairBMC

differencesat

spine

(β=1.681g;

95%

CI=

0.198–

3.164;

p<0.05)

femoral

neck

(β=0.100g;

95%

CI=

0.004–0.196;

p<0.05)andforearm

(β=0.261

g;95%

CI=

0.018–

0.504;

p<0.05).

β(linearregression)

13

Osteoporos Int

Tab

le3

(con

tinued)

Author

Objectiv

eParam

eters(bonesites)

Sam

plesize

Mainresults

(birth

weightandadultbone

mass)

Estim

ators

Score

a

Birth

weighthasshow

nassociationwith

BMD

only

intheunadjusted

analysis

teVelde

etal.[20]

Toinvestigatetheassociation

betweenbirthweightandadults'

musculo-skeletalhealth

BMCandBMD(totalbody,

lumbarspine,

andhip)

151wom

enand117men

Birth

weightwas

positiv

elyassociated

with

BMCof

totalbody

(β=189.1g;

95%

CI=

18.9–3

59.2;p≤0

.05)

andhip(β=2.24

g;95%

CI=

0.39–4

.08;

p<0.05).How

ever,these

associations

lostsignificance

afteradjustment

foradultweight.Birth

weightshow

edno

associationwith

BMD

β(linearregression)

15

Dennisonet

al.[21]

Toevaluate

theinfluenceof

intrauterine

andpostnatalgrow

thon

bone

massof

elderly

individuals

BA,BMCandBMD

(lum

barspineand

proxim

alfemur)

498wom

enand468men

Amongmen,birthweightwas

associated

with

BA

ofspine(r=0.12;p=0.01)andproxim

alfemur

(r=0.17;p=0.0002);BMCof

spine

(r=0.10;p=0.03)andproxim

alfemur

(r=0.16;p=0.0003);andBMD

ofproxim

alfemur

(r=0.10;p=0.03).Amongwom

en,

birthweightwas

associated

with

BA

ofspine

(r=0.18;p=0.0001)andproxim

alfemur

(r=0.26;p<0.0001)andBMCof

spine

(r=0.11;p=0.03)andproxim

alfemur

(r=0.16;p=0.0008)

Correlatio

ncoefficient

(linearregression)

18

Laitin

enet

al.[22]

Toinvestigatetheassociation

betweenbody

size,in

different

periodsof

grow

th,andmeasures

ofbone

massat

adulthood

BMCandBMD

(distaland

ultradistal

radius)

534wom

enand561men

After

adjustmentsforadultBMIandcalcium

intake,SGA

individualsshow

edahigher

chance

tobe

classified

aslow

BMC

(under

the10th

centile

ofsampledistributio

n)in

both

regions(distalandultradistal)of

radius

(OR=

2.50;95%

CI=

1.50

–4.20,

forboth

sites)

but

notas

low

BMD.IU

GRindividualsshow

eda

higher

chance

tobe

classified

aslow

BMCin

thedistal(O

R=2.61;95%

CI=

1.42

–4.80)

and

ultradistalregion

(OR=2.54;95%

CI=

1.40

–4.59)of

radius;andlow

BMD

inthedistal

region

(OR=1.92;95%

CI=

1.01

–3.66)

OR(logistic

regression)

14

Pearceet

al.[48]

Toinvestigatethedeterm

inantsof

someindicators

ofadultbone

health

BA

andBMD

(lum

bar

spine+hip)

218wom

enand171men

Birth

weightwas

positiv

elyassociated

with

BA

(lum

barspine+hip)

ofmen

(β=2.57

cm2;9

5%CI=

1.15–3

.98)

andwom

en(β=1.64

cm2;

95%

CI=

0.65

–2.63).After

adjustmentsfor

adultweightandheight,thisassociation

remainedsignificantonly

formen

(β=

1.11

cm2;95%

CI=

0.04–2

.18).Birth

weight

show

edno

associationwith

BMD

β(linearregression)

19

Saito

etal.[23]

Toinvestigatetheassociation

betweengrow

thandadult

wom

en'sbone

mass

BMCandBMD

(lum

bar

spine,

hip,

andfemoral

neck)

86wom

enAfter

adjustmentsforweightgain

between

0–1,5;

6–9and9–

12years;adultweight;

calcium

intake;andphysical

activ

ity,birth

weightremainedpositiv

elyassociated

with

BMCof

spine(β=3.48

g;SE=1.72;R2=

0.042;

p=0.047)

andhip(β=2.25

g;SE=

1.05;R2=0.039;

p=0.035).Birth

weighthas

show

nassociationwith

BMD

only

inthe

unadjusted

analysis

βandR2(linearregression)

16

Osteoporos Int

Tab

le3

(con

tinued)

Author

Objectiv

eParam

eters(bonesites)

Sam

plesize

Mainresults

(birth

weightandadultbone

mass)

Estim

ators

Score

a

Dalziel

etal.[24]

Toinvestigatetheeffect

ofintrauterine

exposure

toglucocorticoids,low

birthweight,

andprem

aturity

onpeak

bone

massandfemoral

geom

etry

ofadults

BA,BMC,andBMD

(total

body,lumbarspine,

proxim

alfemur,and

femoral

neck)

86wom

enand88

men

After

adjustmentsforsex,

gestationalage,PBT,

andadultleglength

Z-score,b

irth

weightwas

positiv

elyassociated

with

BA

(β=0.56

cm2;

SE=0.22;p=0.013)

ofproxim

alfemur.Birth

weightZ-score

was

positiv

elyassociated

with

BMC(β=64

g;SE=29;p=0.029)

andBMD

(β=0.02

g/cm

2;SE=0.01;p=0.033)

oftotal

body

afteradjustmentsforsex,

gestational

age,PBT,

andadultheight

Z-score;andBMC

(β=1.24

g;SE=0.51;p=0.016)

andBMD

(β=0,03

g/cm

2;SE=0,01;p=0,031)

ofprox-

imal

femur,afteradjustmentsforsex,

gesta-

tionalage,

PBT,

andadultleglength

Z-score.

Allassociations

wereabolishedafteradjust-

mentsforadultsize

β(linearregression)

16

Oliv

eret

al.[25]

Toinvestigatetherelatio

nship

betweengrow

thandbone

strength

BA

andBMD

(forearm

)318wom

enand313men

After

adjustmentsforage,BMI,sm

okinghabit,

alcoholconsum

ption,

physical

activ

ity,

calcium

intake,social

status,HRT,

and

numberof

yearsaftermenopause;birth

weightwas

associated

with

BA

oftotalbody

(r=0.18;p

<0.01)andtib

ia(r=0.29;p

<0.001)

andtrabecular

BMD

ofradius

(r=−0

.15;

p<0.05),am

ongwom

en.Amongmen,birth

weightwas

associated

with

BA

oftotalbody

(r=0.13;p<0.05)andtib

ia(r=0.13;p<0.05)

andtrabecular

BMD

oftib

ia(r=−0

.13;

p<0.05)

Pearson

correlation

coefficient

15

Leunissen

etal.[30]

Toinvestigatetheassociation

betweenbirthsize

(weightand

length)andadultbone

mass

BMCandBMD

(total

body

andlumbarspine)

121wom

enand191men

After

correctio

nforadultheight,individuals

born

SGAwith

(−1.29

SDS)andwith

out

catch-up

grow

th(−0.38

SDS)presentedlower

totalbody

BMCthan

AGA

individualswith

adequate

postnatalgrow

th(p<0.02).Birth

weightZ

-score

was

negativ

elyassociated

with

BMD

atthespine,

afteradjustmentforadult

height

(β=−1

.80;

p=0.026)

andweight

(β=−1

.65;

p=0.037).How

ever,thisassocia-

tionlostsignificance

afteradjustmentsfor

adultheight

andbirthlength

Z-score

differ-

ence,as

wellas

foradultweightandbirth

weightZ-score

difference

Group

difference

β(linearregression)

15

Hoviet

al.[26]

Toevaluate

whether

individuals

born

with

VLBW

becameadults

with

reducedbone

mass

BMCandBMD(totalbody,

lumbarspine,andfemoral

neck)

168wom

enand115men

After

adjustmentsforsex,

currentheight,and

exercise

intensity,VLBW

individuals

presentedlower

BMD

atlumbarspine

(β=−0

.26Z-score

units;95%

CI=

−0.51

to−0

.01;

p=0.04)andfemoral

neck

(β=−0

.40

Z-score

units;95%

CI=

−0.64to

−0.17;

p<

0.001)

andtotalbody

BMC(β=−6

9g;

95%

CI=

−138

to−1

;p=0.05)than

theirAGApeers

Group

Differenceβ

(linearregression)

17

Schlüssel

etal.[27]

Toexploretheassociationbetween

birthweight,postnatallin

ear

grow

th,andbone

massin

young

adults

BA,BMC,andBMD

(lum

barspine,

proxim

alfemur,andfemoral

neck)

250wom

enand246men

After

adjustmentsforsex,

skin

color,

gestationalage,

physical

activ

ity,sm

oking

habit,alcohol,calcium

andproteinintake,

adultweight,andheight

Z-score,birthweight

Z-score

remainedpositiv

elyassociated

with

BA

(β=0.62

cm2,SE=0.2;

p=0.002)

and

β(linearregression)

19

Osteoporos Int

regarding study variability in relation to sex, age, bonesites, parameters of bone mass, and data analysis proce-dures employed; (b) issues concerning differences in theassociation between birth weight, BMC and BMD, beforeand after adjustments for measures of adult body size; and(c) the importance of using adequate models, includinginteraction tests between prenatal and postnatal variables, toanalyze life course data.

The reviewed studies are quite different with regard toseveral aspects that impede the conduction of a formalmeta-analysis, with the wide variability in study samplesize and characteristics being the most important. Elevenstudies included both men and women, while only sixpresented results stratified by sex. None presented analysisof interaction between birth weight and sex on theassociation with bone mass. Further, it is worth noting thefact that in general, the studies included a wide range of agestrata in their sample. Thus, it is important to consider thisvariability before drawing any conclusions because bothsex and age are important determinants of bone metabo-lism. In this regard, these variables directly influence theindividual's peak bone mass (PBM) and the subsequentbone loss that naturally occurs with aging [11].

Three of the reviewed studies were conducted based onthe Hertfordshire birth cohort. These studies probably hadsome subjects in common. However, we decided tomaintain all three studies in the review because they havea different number of subjects in their samples, differentbone sites measured, as well as different techniques of bonemass assessment.

Another source of variability relates to the great numberof bone sites that were studied. We included in the reviewstudies that investigated the effect of birth weight on thebone mass of peripheral sites, such as the wrist and theforearm, and studies that investigated the association inmore central sites, such as the hip, the lumbar spine, oreven the whole body. Homogeneity was not observed in theselection of the evaluated bone mass parameters. The mostcommonly studied parameters were BMC and BMD. Fewstudies included BA among the outcomes.

Several different approaches to data analysis were chosenby the authors to evaluate the association between birthweight and adult bone mass. Both the explanatory variable(birth weight) and the outcomes (BA, BMC and BMD) wereappraised in different ways. Some studies investigated theoutcome as categorical, others as ordinal, or as continuous.Further, the adjustments employed in the analyses were nothomogeneous among studies. Physical activity, smokinghabit, alcohol consumption, and calcium intake were themost common co-variables. Therefore, it became quitedifficult to compare the results of the reviewed studies—aswell as provide an estimation of effect size—that otherwiseallowed us to provide an in-depth description of the findings.T

able

3(con

tinued)

Author

Objectiv

eParam

eters(bonesites)

Sam

plesize

Mainresults

(birth

weightandadultbone

mass)

Estim

ators

Score

a

BMC(β=0.90

g,SE=0.41;p=0.029)

ofproxim

alfemur.BMD

meanat

thespineof

men

inthethirdtertile

ofbirthweight

(1.09g/cm

2)was

significantly

higher

than

thoseat

second

(1.06g/cm

2)or

first

(1.04g/cm

2)tertiles;pfortrend=0.015.

How

ever,thisassociationlostsignificance

afteradjustmentsforadultheight

andweight

AGAadequate

forgestationalage;

BAbo

nearea;BMCbo

nemineral

content;BMD

bone

mineral

density

;BMIbo

dymassindex;

CIconfidence

interval;HRTho

rmon

erepo

sitio

ntherapy;

IUGR

intrauterine

grow

threstriction;

LBW

low

birthweigh

t;ORod

dsratio

;SD

Sstandard

deviationscores;SE

standard

error;SG

Asm

allforgestationalage;VLBW

very

low

birthweigh

t;PBTprenatal

betamethasone

treatm

ent

aMethodologicalquality

scoreof

thestudy,

accordingto

thecriteriaproposed

byDow

nsandBlack

[15]

b2,004twin

pairs

Osteoporos Int

A second aspect that we shall discuss concerns thestrength of the association between birth weight and bonemass parameters. The influence of birth weight was muchmore evident on BMC rather than on BMD. In almost allstudies that found an unadjusted association between birthweight and BMD [17–19, 22–24], it lost the significanceafter the inclusion of some measure of adult body size inthe models. A potential explanation consists of the fact thatprenatal growth sets the growth trajectory and hencepredicts bone size (in terms of length and width), which ismore closely related to BMC and seems to be establishedearlier. In contrast, BMD continues to increase for someyears and may be influenced by other locally acting factorssuch as mechanical load due to physical activity andhormonal status [7, 8, 17]. Nevertheless, this is a veryimportant finding and has clear implications for clinical andpublic health policies, considering the known strongassociation between a reduction in BMC and an increasein the adult's fracture risk [31–36].

Another important fact that needs to be mentioned is thatone of two studies that found a significant associationbetween birth weight and BMD, even after adjustments forcurrent body size, included subjects with very low birthweight (VLBW). This is challenging considering thatVLBW newborns probably have some postnatal complica-tions that are likely to influence bone mass. Further, VLBWindividuals are suggested to attain shorter body height inadulthood [37–39]. The authors argue that VLBW individ-uals are prone to reach only submaximal peak bone mass,although this may partially be attributable to their smaller

body size. Nonetheless, these findings suggest that VLBWindividuals are programmed not only to develop bones thatare compromised in size but also in its density.

The third aspect that needs to be discussed refers to thecorrect analysis of life course data. Lucas et al. [40] suggestedseveral years ago that a crucial issue in reporting the resultsof life course studies is the employment of adequate analysisprocedures and interpretations. These authors suggested alogical sequence of four statistical models to be fit beforereaching any conclusion: 1) an early model, relating earlysize (birth weight) with the outcome; 2) a combined model,including both early (birth weight) and later size (weight orheight); 3) an interaction model, obtained by adding theinteraction term between early and later size to the combinedmodel; and 4) a late model, in which later size alone isrelated to the outcome. These authors were aware that thecorrect analysis and interpretation of data require theemployment of four models. Most studies reached the sec-ond level, but unfortunately, it seems that this suggestion hasbeen overlooked. Only two of the 16 reviewed studies havereported an interaction analysis between prenatal andpostnatal variables but observed no statistical significanceon the determination of adult bone mass.

Studies conducted by Cooper and colleagues [7, 8, 21, 25,41] have proposed that low weight at 1 year would be astronger marker of some “program” for lower bone mass, bothin young adulthood [7] and in the seventh decade of life[8, 21, 25], than would low birth weight. It is also proposedthat the bone mass “track” is defined very early in life,meaning that individuals in the high end of the population

0

5

2 2

11

9

5

10

2

0

1

2

3

4

5

6

7

8

9

10

Nu

mb

er o

f st

ud

ies

Bone Area Bone Mineral Content Bone Mineral Density

Without Association

Unadjusted Association

Adjusted Association

Notes: A study might figure in the unadjusted and the adjusted association column. Only studies that have employed DXA for bone mass evaluation were included. In addition to height and weight, adjustments were made for physical activity, smoking habit,alcohol consumption and calcium intake.

Fig. 2 Number of studies according to the type of reported association between birth weight and parameters of adult bone mass, in at least onebone site

Osteoporos Int

distribution are likely to remain there over time [42–45].Other authors have also shown the positive effect of earlypostnatal growth on adult bone mass [23, 46, 47]. Therefore,if both prenatal and early postnatal growth play a role inbone mass improvement, it is important to determine moreprecisely which period of postnatal development marks thecritical point that programs the achievement of adult bonemass.

No study reached the maximum score in the methodo-logical quality evaluation. Although displaying someimportant limitations, several studies received high scoresfor their methodological quality. The internal consistency ofthe scores assigned to the studies gave greater robustness tothis evaluation. Among the criteria considered here, thefinal sample's representativeness was defective in allreviewed studies. Birth cohorts, particularly those withlarge follow-up periods, have final samples that were rarelyrepresentative of the sample's initial drift. Furthermore,since this review was conducted in a database with rigorousindexing criteria and incorporated only articles published inEnglish, publication and language bias cannot be ruled out.Therefore, it is under the light of these considerations thatthe results described here should be considered.

In conclusion, based on the amount of accumulatedevidence, it is probable that birth weight has a positiveassociation with adult bone mass. The association is muchmore evident for BMC rather than BMD. As stated above,this represents an important finding for clinical and publichealth policies since a reduction in BMC is stronglyassociated with an increased fracture risk. However, interac-tion between prenatal and postnatal growth must be ade-quately investigated as well as reported more often, to identifywhich stage (or stages) of growth are the real critical windowfor bone mass accrual. Further, despite adequate adjustments,the results of studies investigating the association betweenbirth weight and bone mass in early adulthood are scanty aswell as still controversial. These issues remain difficult, andmore studies aiming to answer these questions are needed.

Funding Gilberto Kac is a research fellow from the “NationalCouncil for Scientific and Technological Development” (CNPq).Michael Maia Shlüssel was a recipient of a scholarship from the“Brazilian Coordination Body for the Training of University LevelPersonnel” (CAPES).

Conflicts of interest None.

References

1. Barker DJP (2007) The origins of the developmental originstheory. J Intern Med 261:412–417

2. Kajantie E (2008) Early-life events. Effects on aging. Hormones7:101–113

3. DeBoo HA, Harding JE (2006) The developmental origins ofadult disease (Barker) hypothesis. Aust NZ J Obstet Gynaecol46:4–14

4. Bateson P (2001) Fetal experience and good adult design. Int JEpidemiol 30:928–934

5. Harding JE (2001) The nutritional basis of the fetal origins ofadult disease. Int J Epidemiol 30:15–23

6. Bateson P, Barker D, Clutton-Brock T, Deb D, D’Udine B, FoleyRA, Gluckman P et al (2004) Developmental plasticity and humanhealth. Nature 430:419–421

7. Cooper C, Cawlay M, Bahalla A, Egger P, Ring F, Mortn L,Barker D (1995) Childhood growth, physical activity, and peakbone mass in women. J Bone Miner Res 10:940–947

8. Cooper C, Fall C, Egger P, Hobbs R, Eastell R, Barker D (1997)Growth in infancy and bone mass in later life. Ann Rheum Dis56:17–21

9. Cooper C, Westlake S, Harvey N, Javaid K, Dennison E, HansonM (2006) Developmental origins of osteoporotic fracture. Osteo-poros Int 17:337–347

10. Heaney RP, Abrams S, Dawson-Hughes B, Looker A, Marcus R,Matkovic V, Weaver C (2000) Peak bone mass. Osteoporos Int11:985–1009

11. Mora S, Gilsanz V (2003) Establishment of peak bone mass.Endocrinol Metab Clin N Am 32:39–63

12. Hernandez CJ, Beaupré GS, Carter DR (2003) A theoreticalanalysis of the relative influences of peak BMD, age-related boneloss and menopause on the development of osteoporosis.Osteoporos Int 14:843–847

13. Bonjour JP, Chevalley T, Ferrari S, Rizzoli R (2009) Theimportance and relevance of peak bone mass in the prevalenceof osteoporosis. Salud Publica Mex 51(S1):S5–S17

14. Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group(2009) Preferred reporting items for systematic reviews and meta-analyses: the PRISMA Statement. PloS Med 6(7):e1000097, Epub2009 Jul 21

15. Downs SH, Black N (1998) The feasibility of creating a checklistfor the assessment of the methodological quality both ofrandomized and non randomized studies of health care interven-tions. J Epidemiol Community Health 52:377–384

16. Shrout PE (1998) Measurement reliability and agreement inpsychiatry. Stat Methods Med Res 7:301–317

17. Yarbrough DE, Barrett-Connor E, Morton DJ (2000) Birth weightas a predictor of adult bone mass in postmenopausal women: theRancho Bernardo Study. Osteoporos Int 11:626–630

18. Gale CR, Martyn CN, Kellingray S, Eastell R, Cooper C (2001)Intrauterine programming of adult body composition. J ClinEndocrinol Metab 86:267–272

19. Antoniades L, MacGregor AJ, Andrew T, Spector TD (2003)Association of birth weight with osteoporosis and osteoarthritis inadult twins. Rheumatology 42:791–796

20. te Velde SJ, Twisk JWR, van Mechelen W, Kemper HCG (2004)Birth weight and musculoskeletal health in 36-years-old men andwomen: results from the Amsterdam Growth and Health Longi-tudinal Study. Osteoporos Int 15:383–388

21. Dennison EM, Syddall HE, Sayer AA, Gilbody HJ, Cooper C(2005) Birth weight and weight at 1 year are independentdeterminants of bone mass in the seventh decade: the Hertford-shire cohort study. Pediatr Res 57:582–586

22. Laitinen J, Kuikaanniemi K, Heikkien J, Koiranen M,Nieminen P, Sovio U, Keinanen-Kiukaanniemi S, JärvelinMS (2005) Body size from birth to adulthood and bonemineral content and density at 31 years of age: results fromthe northern Finland 1966 birth cohort study. Osteoporos Int16:1417–1424

Osteoporos Int

23. Saito T, Nakamura K, Okuda Y, Nashimoto M, Yamamoto N,Yamamoto M (2005) Weight gain in childhood and bone mass infemale college students. J Bone Miner Metab 23:69–75

24. Dalziel SR, Fenwick S, Cundy T, Parag V, Beck TJ, Rodgers A,Harding JE (2006) Peak bone mass after exposure to antenatalbetamethasone and prematurity: follow-up of a randomizedcontrolled trial. J Bone Miner Res 21:1175–1186

25. Oliver H, Jameson KA, Sayer AA, Cooper C, Dennison EM, TheHertfordshire Cohort Study Group (2007) Growth in early lifepredicts bone strength in late adulthood: the Hertfordshire CohortStudy. Bone 41:400–405

26. Hovi P, Andersson S, Järvenpää A, Eriksson JG, Strang-KarlssonS, Kajantie E, Mäkitie O (2009) Decreased bone mineral densityin adults born with very low birth weight: a cohort study. PLoSMed 6(8):e1000135. doi:10.1371/journal.pmed.1000135

27. Schlüssel MM, Castro JAS, Kac G, Silva AAM, Cardoso VC,Bettiol H et al (2010) Birth weight and bone mass in young adultsfrom Brazil. Bone 46:957–963

28. Genant HK, Cooper C, Poor G, Reid I, Ehrlich G, Kanis J et al (1999)Interim report and recommendations of the World Health Organiza-tion Task-Force for Osteoporosis. Osteoporos Int 10:259–264

29. Hamed HM, Purdie DW, Ramsden CS, Carmichael B, Steel SA,Howey S (1993) Influence of birth weight on adult mineraldensity. Osteoporos Int 3:1–2

30. Leunissen RWJ, Stijnen T, Boot AM, Hokken-Koelega ACS(2008) Influence of birth size and body composition on bonemineral density in early adulthood: the PROGRAM study. ClinEndocrinol (Oxf) 69:386–392

31. Wasnich RD, Ross PD, Heilbrun LK, Vogel JM (1985) Predictionof postmenopausal fracture risk with use of bone mineralmeasurements. Am J Obstet Gynecol 1(153):745–751

32. Wasnich RD, Ross PD, Heilbrun LK, Vogel JM (1987) Selectionof the optimal skeletal site for fracture risk prediction. Clin OrthopRelat Res 216:262–269

33. Gärdsell P, Johnell O, Nilsson BE (1989) Predicting fractures inwomen by using forearm bone densitometry. Calcif Tissue Int44:235–242

34. Wasnich RD, Ross PD, Davis JW, Vogel JM (1989) A comparisonof single and multi-site BMC measurements for assessment ofspine fracture probability. J Nucl Med 30:1166–1171

35. Gärdsell P, Johnell O, Nilsson BE, Gullberg B (1993) Predictingvarious fragility fractures in women by forearm bone densitom-etry: a follow-up study. Calcif Tissue Int 52:348–353

36. Melton LJ 3rd, Crowson CS, O’Fallon WM, Wahner HW, RiggsBL (2003) Relative contributions of bone density, bone turnover,and clinical risk factors to long-term fracture prediction. J BoneMiner Res 18:312–318

37. Ericson A, Kallen B (1998) Very low birthweight boys at the ageof 19. Arch Dis Child Fetal Neonatal Ed 78:F171–F174

38. Hack M, Schluchter M, Cartar L, Rahman M, Cuttler L, BorawskiE (2003) Growth of very low birth weight infants to age 20 years.Pediatrics 112:e30–e38

39. Cooke RW (2004) Health, lifestyle, and quality of life for youngadults born very preterm. Arch Dis Child 89:201–206

40. Lucas A, Fewtrell MS, Cole TJ (1999) Fetal origins of adultdisease—the hypothesis revisited. Br Med J 319:245–249

41. Javaid MK, Lekamwasam S, Clark J, Dennison EM, SyddallHE, Loveridge N, Reeve J, Beck TJ, Cooper C, TheHertfordshire Cohort Study Group (2006) Infant growthinfluences proximal femoral geometry in adulthood. J BoneMiner Res 21:508–512

42. Ferrari S, Rizzoli R, Slosman D, Bonjour JP (1998) Familialresemblance for bone mineral mass is expressed before puberty. JClin Endocrinol Metab 83:358–361

43. Loro ML, Sayre J, Roe TF, Goran MI, Kaufman FR, Gilsanz V(2000) Early identification of children predisposed to low peakbone mass and osteoporosis later in life. J Clin Endocrinol Metab85:3908–3918

44. Ruff C (2005) Growth tracking of femoral and humeral strengthfrom infancy through late adolescence. Acta Paediatr 94:1030–1037

45. Foley S, Quinn S, Jones G (2009) Tracking of bone mass fromchildhood to adolescence and factors that predict deviation fromtracking. Bone 44:752–757

46. Weiler HA, Yuen CK, Seshia MM (2002) Growth and bonemineralization of young adults weighing less than 1500 g at birth.Early Hum Dev 67:101–112

47. Jensen RB, Vielwerth S, Frystyk J, Veldhuis J, Larsen T,Mølgaard C, Greisen G, Juul A (2008) Fetal growth velocity,size in early life and adolescence, and prediction of bonemass: association to the GH–IGF axis. J Bone Miner Res23:439–446

48. Pearce MS, Birrell FN, Francis RM, Rawlings DJ, Tuck SP,Parker L (2005) Lifecourse study of bone health at age 49–51years: the Newcastle thousand families cohort study. J EpidemiolCommunity Health 59:475–480

Osteoporos Int