1

Birth weight and lung function in adulthood:

A systematic review and meta-analysis

Published in: Ann Am Thorac Soc. 2017 (www.ncbi.nlm.nih.gov/pubmed/28362513 )

Authors:

Neil J. Saad1, Jaymini Patel1, Peter Burney1, Cosetta Minelli1

Affiliation:

1 National Heart and Lung Institute, Imperial College London, UK

Corresponding Author:

Neil J. Saad

National Heart and Lung Institute, Imperial College London

1b Manresa Road, SW3 6LR London, UK

E-mail: ns11@ic.ac.uk; Telephone number: + 44 (0) 2 7594 7952

Sources of support: NJS was supported by a National Heart and Lung Institute Foundation

studentship.

Keywords: Early life; Vital capacity; Airflow obstruction; Weight at one year; Weight gain

2

At a Glance Commentary

Scientific knowledge on the subject: Increasing evidence supports the influence of early life

factors on later lung function, which might prevent individuals from attaining maximum lung

function and predispose them to increased risk of lung disease in later life.

What this study adds to the field: This is the first systematic review and meta-analysis on the

association of birth weight, weight at one year and weight gain in the first year of life with lung

function measures of both restrictive and obstructive patterns in adulthood. We convincingly

show a positive association between birth weight and restrictive impairment (low FVC), with

findings consistent across studies. The meta-analysis for birth weight and airflow obstruction

(low FEV1/FVC) might suggest a positive association but the evidence is much weaker. We

identified gaps and limitations in the current evidence that should be addressed in future

research.

This article has an online data supplement, which is accessible from this issue’s table of

content online at www.atsjournals.org.

3

Abstract

Rationale: There is evidence suggesting that birth weight may influence lung function in

adulthood, but it is unclear whether it might differentially affect restrictive (FVC) and

obstructive (FEV1/FVC) patterns.

Objectives: To summarize evidence available on the association of birth weight, weight at one

year and weight gain in the first year of life with FVC and FEV1/FVC in adulthood.

Methods: We performed a systematic review of the literature by searching MEDLINE,

EMBASE and Web of Science through January 2015. Data were combined using inverse-

variance weighted meta-analysis with random effects models, and between-study heterogeneity

evaluated. We conducted a priori subgroup or sensitivity analyses by age, country wealth,

ethnicity, sex and smoking. Risk of bias was evaluated using the Newcastle-Ottawa Scale, and

reporting bias using funnel plots.

Measurements and Main Results: Eighteen articles were included in the review, and 13 in the

meta-analyses. Most studies were from high-income countries, and all showed low risk of bias.

We found strong evidence of association of birth weight with adult FVC, a 59.4mL higher FVC

in adulthood per kg increase in birth weight (95%CI: 43.3 to 75.5), with no evidence of

heterogeneity. Evidence of an association of birth weight with FEV1/FVC was weaker and

showed some inconsistency across studies. Only one study investigated weight at one year, and

another one weight gain in the first year.

Conclusions: Our meta-analyses show strong and consistent evidence of an association of birth

weight with adult FVC, a measure of restrictive impairment, with much weaker evidence for

airflow obstruction.

4

Introduction

Convincing evidence supports the notion that factors affecting intrauterine and early life can

influence lung function in adulthood (1–3). Thirty years ago, Barker and colleagues showed a

geographical relationship between high mortality from pneumonia and bronchitis in early life

and mortality from chronic lung disease half a century later (4). They subsequently showed

associations of birth weight and weight at one year with low lung function and mortality from

COPD (5). Barker and colleagues hypothesized that a poor prenatal or early postnatal

environment would predispose individuals to later chronic lung disease (6). They focused on

birth weight, as a marker of intrauterine development, and weight in early life, on the basis that

early infections associated with poverty might cause poor growth in the first 12 months of life

(4). These associations were mainly observed for forced vital capacity (FVC) rather than for

the ratio between forced expiratory volume in one second (FEV1) and FVC (FEV1/FVC),

suggesting an effect on restrictive impairment rather than airway obstruction. In the last two

decades, several studies have further investigated the relationship between birth weight and

lung function in adulthood.

Here we review all available epidemiological evidence on the association of birth weight, as

well as weight gain in the first year of life and weight at one year, with lung function in adults

from the general population. Our aim is to investigate the consistency of findings across

studies, provide a more precise estimate of these associations through meta-analysis, and

identify gaps and limitations in the current evidence that should be addressed in future research.

To distinguish between effects on restrictive and effects on obstructive patterns of low lung

5

function, we focus on both FVC and FEV1/FVC ratio (7). Some of the results of this study have

been previously reported in the form of a conference abstract (8).

Methods

All methods used were specified a priori in a detailed protocol, and the reporting followed the

guidelines from the MOOSE statement (9).

Inclusion criteria and search strategy

Studies of any design were eligible for inclusion if they assessed the association between birth

weight, weight at one year and/or weight gain in the first year of life, with FVC and/or

FEV1/FVC in adults (≥18 years) from the general population. We excluded studies where

participants were selected based on gestational age or presence of bronchopulmonary

dysplasia, as well as editorials, commentaries, case reports, conference abstracts and reviews.

We searched MEDLINE, EMBASE and Web of Science through January 2015, with no date

or language restrictions. The search was organized in two sections, one for birth weight, weight

at one year and weight gain in the first year, and the other for airflow obstruction, FVC, FEV1

and spirometry. In MEDLINE and EMBASE, we used both free text and controlled vocabulary,

whereas in Web of Science only free text searching was available (Table E1). One reviewer

(NJS) screened titles and abstracts for eligibility and a second reviewer (JP) independently

screened a random sample of 100 articles; there were no discrepancies between the two

reviewers. Identification of further studies through cross-checking of references from relevant

articles was performed in duplicate (NJS and JP).

6

Data extraction

Data were extracted in duplicate (NJS and JP) using a pre-piloted data extraction form (Table

E2), which covered: study design; characteristics of the study population; birth weight, weight

at one year and weight gain in the first year; FVC and FEV1/FVC; information on main and

secondary statistical analyses, including adjustment for covariates, adjustment for multiple

comparisons, sample size and power calculations; study findings. Disagreements were resolved

through consensus and, when necessary, through discussion with a third reviewer (CM).

Risk of bias

One reviewer (NJS) assessed the risk of bias of each study using the Newcastle-Ottawa Quality

Assessment Scale for cohort studies (NOS) (10). We modified the original NOS to fit our

research questions (Table E3) and high, moderate and low risk of bias were defined as NOS<4,

between 4 and 6, and >6, respectively, as in the original NOS (11–13).

Statistical analysis

Whenever possible, study results were translated into differences in mL of FVC and %

FEV1/FVC per kg increase in birth weight, weight at one year or weight gain in the first year.

The standardized mean difference (SMD), a unitless measure representing the difference in

outcome between the groups divided by the standard deviation of the outcome, was calculated

for each study and used for the meta-analysis whenever the outcome of any of the studies could

not be translated into FVC mL or % FEV1/FVC.

For studies that had analyzed the exposure (e.g., birth weight) as a continuous variable and

used linear regression, we extracted regression coefficients and their standard errors for the

meta-analysis. We used results from a model adjusted for age, height and sex if available, or

7

from a model judged as similar as possible to this. For studies that used a categorical exposure

(e.g., low vs. normal birth weight), we extracted (or calculated) the mean difference in lung

function between the two groups and its standard error. In the presence of more than two

groups, we dichotomized the exposure to make it consistent across studies.

We used inverse-variance weighted random effects meta-analysis. The choice of a random

effects model was made a priori to allow for between-study heterogeneity of results. Between-

study heterogeneity was tested using the Q test and its magnitude estimated using the I2

statistic, which represents the percentage of variation in effect size between studies that can be

attributed to genuine variability rather than chance (14). In our main analyses we did not

combine study estimates if I2 was higher than 50%, but we do present these results in secondary

analyses.

We performed a priori subgroup analyses by age, sex, country wealth (high- versus low- to

middle-income countries, based on the World Bank classification (15)), and country

geographic region (defined by the United Nations statistics division (16)) as a proxy for

ethnicity. We also investigated the importance of smoking in a sensitivity analysis by

combining effect estimates that were adjusted for smoking. Finally, the presence of reporting

bias was assessed using funnel plots. All analyses were performed using Stata 13.1 (StataCorp

LP, College Station, TX, USA). We defined statistical significance at a p-value threshold of

<0.05.

Results

Inclusion and exclusion of studies

8

After title and abstract screening of the 718 unique records identified (Figure 1), we assessed

21 full-text articles. Eighteen articles were included in the systematic review, of which 13 in

the meta-analyses. Reasons for exclusions are reported in Figure 1. All studies were in English

and all 18 articles reported associations for FVC, while only 10 also reported FEV1/FVC. One

article reported results from two different study populations (17) and five reported results for

males and females separately (18–22), so the total number of comparisons was 24 for FVC and

14 for FEV1/FVC (Figure E1). Birth weight was investigated in all 18 studies, whereas there

was only one for weight at one year and another one for weight gain in the first year of life.

Characteristics of the studies included

Table 1 reports the characteristics of the studies included. Seventeen of the 18 studies were

either prospective cohort studies, where birth weight was obtained in real time, or retrospective

cohort studies, where birth weight was obtained from hospital records. Only Lawlor et al. (23)

concurrently measured self-reported birth weight and lung function. The average birth weight

was above 3.0 kg in all studies and varied between 3.1 and 3.6 kg, except for a study from

South India where the average was 2.8 kg (18). The average age at the time lung function was

measured was between 18 and 58 years (<40 in 9 studies and ≥40 in the other 9). The average

FVC varied from 2.7 to 5.0 L and FEV1/FVC from 73.0 to 88.6%. Studies with older

participants had lower FVC, but not always lower FEV1/FVC, compared to studies with

younger individuals. Most studies were conducted in high-income countries, with only four

studies from low- to middle-income countries (China, Brazil and India). Gestational age was

not commonly reported, with only 8/18 studies providing any information; in these studies,

only a small number of preterm births were reported.

9

Birth weight was evaluated differently across studies, either as a continuous or binary exposure;

as a binary exposure, the threshold of 2.5 kg was mostly used, although one study employed

another cut-off (low birth weight as ≤2.5kg; normal birth weight as >3.0kg) (Table E4). FVC

was analysed in litres (or ml) or percent predicted, while FEV1/FVC was predominantly

expressed in percent, though one study used percent predicted (Table E4).

For each of the studies, Table E5 reports all confounding variables considered for adjustment

in any of their analyses, regardless of whether adjustments were made simultaneously or one

by one. Most studies adjusted for age, height and sex, with several also adjusting for smoking

and socio-economic or educational status. Adjustment for other variables, including prenatal

or childhood factors, was rare. The adjustments for the effect estimates used in our main meta-

analyses are reported in Tables E6 and E7, for FVC and FEV1/FVC respectively. All studies

adjusted for age, sex and height, except for two that did not adjust for height. In particular, one

study (24) did not adjust the association of FVC for height. Tables E8 and E9 show the effect

estimates for FVC and FEV1/FVC, respectively, used in the sensitivity analyses that included

adjustment for smoking.

Association of birth weight with lung function

Birth weight and FVC

Five of the 18 studies could not be used for meta-analysis for reasons reported in Figure E1,

which resulted in 18 comparisons from 13 studies included in the meta-analyses. The excluded

studies were all from high-income countries, with sample size between 127 and 1,494, and had

similar average birth weight and lung function as the included studies.

10

In the meta-analysis of birth weight as a continuous exposure, the combined estimate for the

association between FVC and birth weight was 59.4 mL/kg (95% confidence interval (CI):

43.3 to 75.5; p<0.001), with no evidence of between-study heterogeneity (I2: 0%, p_het: 0.5)

(Figure 2A). We also repeated the analyses by adding the recent follow-up results from the

study by Orfei et al. (17), published after our literature search by Cai et al. (25). The result of

this secondary analysis for the association of birth weight (continuous) with FVC did not

substantially change compared with the main analysis (49.9 vs. 59.4 mL higher FVC in

adulthood per kg increase in birth weight; p=0.48) (details in Online Data Supplement).

In the meta-analysis of birth weight as a binary exposure, the standardized mean difference

between normal and low birth weight was 0.13 (95% CI: 0.001 to 0.26; p=0.048), with no

between-study heterogeneity (I2: 3.8%, p_het: 0.4) (Figure 2B). This would correspond to a

difference between normal and low birth weight of 77 mL (95% CI: 0.6 to 154) if we assumed

an FVC standard deviation of 592 mL, as reported in the study (26) with the largest weight in

the meta-analysis. Excluding the only study (24) that did not adjust for height did not alter the

results (p=0.71).

Subgroup analysis by age (≥40 vs. <40 years at time FVC was measured) suggested a smaller

effect estimate for older individuals for the effect of birth weight as a continuous exposure,

p=0.037 (Figure E3A). For birth weight as a binary exposure, however, age seemed to modify

the effect in the opposite direction, with no apparent effect on FVC for younger individuals

(based on three studies, two including only 18-19 year olds), p=0.038 (Figure E3B). Subgroup

analyses by country-level income, country region and sex did not show evidence of

modification of the association of birth weight with FVC by any of these factors. The results

11

also remained unchanged when further adjusting the main analysis for smoking (details in

Online Data Supplement).

Birth weight and FEV1/FVC

Three of the 10 studies for FEV1/FVC did not include a standard error or standard deviation

for the regression coefficient of the association or only reported the association in text without

providing any data. Therefore, these studies could not be included in the meta-analysis,

resulting in 10 comparisons (Figure E1). The excluded studies, all conducted in Western

Europe and Scandinavia, were similar in study characteristics to the included studies.

Five comparisons used birth weight as a continuous exposure and five as a binary exposure

(Figure E1). For birth weight as a continuous exposure, between-study heterogeneity was

higher than 50% (I2: 55%, p_het: 0.064) and therefore results were not synthesized in the main

analysis (Figure 3A). A meta-analysis performed as a secondary analysis showed no

association of birth weight with FEV1/FVC (0.1% higher FEV1/FVC in adulthood per kg

increase in birth weight; 95% CI: -0.4% to 0.5%; p=0.7).

There was, however, no between-study heterogeneity (I2: 0%, p_het: 0.65) for birth weight as

a binary exposure (Figure 3B) and meta-analysis showed a statistically significant positive

association, with an SMD of 0.21 (95% CI: 0.06 to 0.37; p=0.007) which corresponds to an

higher FEV1/FVC of 1.9 % (95% CI: 0.5 to 3.2) for normal vs. low birth weight if we assume

a standard deviation of 8.87% for FEV1/FVC%, as reported in Stein et al. (18). Subgroup

analyses by age, country-income level, country region or sex did not show a difference in the

association between birth weight and FEV1/FVC by any of these factors. Moreover, the

12

combined effect estimate was not markedly altered when further adjusting the main analysis

for smoking (details in Online Data Supplement).

Association of weight at one year and weight gain in the first year with lung function

We identified only one study on weight at one year and one study on weight gain in the first

year of life, both in relation to FVC. Barker et al. (5) reported a positive association, with a

higher FVC in adulthood of 44 mL (95% CI: 0 to 66) per kg increase in weight at one year,

while Canoy et al. (20) showed a higher FVC in adulthood of 41.8mL (95% CI: 16 to 67) and

23.5 mL (95% CI: 4.9 to 42) per kg increase in weight gain in the first year of life, for men and

women respectively.

Risk of bias and investigation of reporting bias

Figure 4 shows the percentage of studies that fulfilled each of the eight NOS criteria, with

study-specific results available in Table E10. There were no studies with high risk of bias;

16/18 studies selected participants representative of the target populations, and 12/18 studies

achieved a follow-up greater than 70% (Table E10). Most studies (78%) ascertained birth

weight from birth or hospital records, although 3/18 studies (27–29) provided no description,

and 1/18 (23) used self-reported birth weight. Assessment of lung function did not always

adhere to ATS/ERS criteria in use at the time of the study, with 10/18 studies stating that they

had used other criteria (Table E10).

We could only draw a funnel plot for FVC and birth weight as a continuous exposure (9 studies)

to investigate reporting bias, given the small number of studies in the other meta-analyses (3

to 5). We did not see evidence of asymmetry (Figure 5).

13

Discussion

Through meta-analysis, we show strong evidence of a positive association of birth weight with

adult FVC, with highly consistent findings across studies. Our findings also suggest a positive

association of birth weight with the adult FEV1/FVC ratio, but such evidence is much weaker

than for the association between birth weight and FVC, and we observed more heterogeneity

across studies.

The association that we observed between birth weight and adult FVC supports Barker’s

hypothesis that early life exposures may be important to future outcomes in an individual (30),

but it is also consistent with alternative explanations such as confounding. The effect of birth

weight on adult FVC might be mediated by adverse early life factors affecting birth weight,

such as maternal smoking (31–33) or poor maternal nutrition in pregnancy (1, 34, 35).

Gestational age is also a possible confounder, since prematurity is a strong predictor of birth

weight and has been associated with lower lung function in adulthood (36–38). Disentangling

the effects of birth weight and gestational age, however, is complex, as gestational age is highly

correlated with birth weight and is also difficult to estimate reliably; the latter is particularly

an issue for studies performed when early fetal ultrasound scans were less common. There is a

further difficulty in disentangling the effects of prematurity from the potential side effects of

treatments used to manage the condition. However, this is an important aspect that requires

further investigation. In our systematic review we could not fully investigate whether prenatal

factors and gestational age could explain the observed associations, since these were seldom

considered in the analyses reported by the studies included. In those few studies that did adjust

for gestational age, in addition to adjusting for height and sex, the effect estimate of the

association between birth weight and FVC did not markedly differ from those when adjusting

14

only for height and sex (26, 28).

Subgroup analyses by age suggested a weaker association between birth weight and FVC in

older compared to younger adults. This might reflect an increasing lifetime impact of

environmental factors such as smoking, diet or air pollution (25). Subgroup analysis by

country-level income was more problematic, as those with birth weight as a binary exposure

were all, except one, from low- to middle-income countries. However, both meta-analyses of

studies with binary or continuous exposure showed statistically significant associations in the

same direction, suggesting that country wealth does not substantially influence the association.

We used country regions as a proxy for ethnicity, as the latter was never reported, and we found

no marked differences between regions, nor were the effect estimates influenced by smoking

or sex.

Our findings suggest a positive association between birth weight and adult FEV1/FVC, but this

evidence is much weaker than the association with adult FVC. Moreover, we observed

inconsistencies of findings across studies for FEV1/FVC in the meta-analysis of birth weight

as a continuous exposure (I2=55%), which we could not fully investigate due to limited

information for subgroup analysis. However, no heterogeneity was found in the analysis of the

five studies that expressed birth weight as a binary exposure.

Studies that selected participants based on gestational age, or presence of bronchopulmonary

dysplasia, were excluded to ensure that the findings were applicable to the general population.

None of the individual studies included appeared to be at high risk of bias, as defined by the

NOS scale, suggesting that bias in the individual study methodology is not likely to explain the

observed positive associations or variability in the results.

15

We performed our systematic review following a strict methodology to minimise possible bias,

including selection bias (e.g. we performed a comprehensive search with no language

restrictions). Moreover, the small number of studies included in each meta-analysis limited our

investigation of reporting bias; the only funnel plot drawn for the association of birth weight

as a continuous exposure with FVC (9 studies) did not show asymmetry. Reporting bias might

have occurred for the association of birth weight with FEV1/FVC. Eight of the 18 studies

included did not report on FEV1/FVC despite reporting results for both FVC and FEV1

separately. We cannot exclude the possibility that null results for FEV1/FVC had been omitted

for this reason. However, not reporting FEV1/FVC might have been due to lack of interest in

this measure, given that many studies defined “airflow obstruction” based on FEV1 alone. A

meta-analysis on birth weight and lung function published ten years ago and including eight

studies, analysed only FEV1 because too few studies reported results for other outcomes (23).

This meta-analysis showed a higher FEV1 in adulthood of 48 mL (26 to 70 mL) per kg increase

in birth weight. The 1946 and the 1958 British national birth cohorts subsequently found

similar estimates of 41.3 and 66.6 mL/kg for the association between FEV1 and birth weight,

respectively (17). Although not uncommon, focusing on FEV1 does not distinguish between

restrictive and obstructive patterns of low lung function (7).

There is insufficient evidence on the effect of weight at one year and weight gain in the first

year of life, with only two studies identified for FVC and none for the FEV1/FVC ratio. The

limited evidence, however, suggests a positive relationship with FVC, as for birth weight. This

systematic review focused on weight at one year and weight gain in the first year of life, but

other studies not included here have considered weight at different times in early childhood.

Two studies found no evidence of association of FVC or FEV1/FVC with weight gain in the

16

first three years of life (28, 39), while one other study reported a positive association of FVC

with weight gain between birth and age five (40). More research is required to provided

evidence of whether weight gain in early life is a predictor of lung function in later life.

Our review focused on two measures, FVC and FEV1/FVC, which we chose as together they

characterize impaired lung function. However, the relation between birth weight and other

measures of lung function has also been previously studied and these findings are in line with

our results. A study among 884 young adults in New Zealand assessed the relation between

birth weight and total lung capacity (TLC), for which FVC generally acts as a proxy in

epidemiological studies. They reported a positive, although not statistically significant,

association (effect estimate: 82 mL higher TLC in adulthood per kg increase in birth weight;

95% CI: -11 to 174) (28), in line with our findings for the relation of birth weight and FVC.

Two other studies, both in young adults, investigated the association of birth weight with mid-

expiratory flow (FEF25-75), a measure of small airways obstruction (21, 23). They found no

evidence of association, which is consistent with our findings for birth weight and FEV1/FVC.

Our systematic review identified methodological limitations and gaps in knowledge that should

be addressed in future research. Prenatal influences and gestational age have been rarely

considered in published studies and this limits our understanding of what could mediate the

observed associations of birth weight with lung function in later life. The different units of

measurement used for FVC and FEV1/FVC and in particular the different ways of evaluating

birth weight (continuous vs. binary) across studies make comparisons difficult. Evaluating

birth weight as a binary exposure should be avoided unless there are specific reasons such as a

belief that birth weight only affects adult lung function below a certain threshold, as this leads

to loss in statistical power (41). This is particularly problematic in those relatively small

17

studies. Moreover, unmeasured factors influencing birth weight and weight gain in early life

as well as later lung function could potentially confound the observed associations. One such

factor that was never taken into account is ethnicity, which may cause confounding in

populations of mixed ethnicity as it is associated with both lower FVC (42, 43) and lower birth

weight (44). Finally, the included studies that used birth weight as a continuous exposure

always assumed a linear relation between weight in early life and lung function in adulthood.

A departure from this assumption, for example if both low and very high birth weight had a

detrimental effect on later lung function, could have an impact on the association estimates,

most likely biasing the estimates towards no effect.

If the association between birth weight and adult lung function is causal, early life public health

interventions to improve the average birth weight could lead to improvements in population

respiratory health. In particular, higher FVC has been associated with reduced mortality and

less respiratory symptoms, such as dyspnoea (45, 46). Our findings suggest a relatively small

effect of birth weight on adult FVC, a 60mL higher FVC in adulthood per kg increase in birth

weight, thus also suggesting that higher birth weight is unlikely to markedly benefit a single

individual. Shifting the distribution of birth weight in the population towards higher values,

however, might reduce the population burden of respiratory disease. This exemplifies the

prevention paradox proposed by Geoffrey Rose; “a preventive measure which brings much

benefit to the population offers little to each participating individual” (47, 48).

In conclusion, there is strong consistent evidence that low birth weight is associated with lower

FVC in adults. Most of the evidence is from studies performed in high-income countries, with

more limited data from low- to middle-income countries. There is more ambivalent evidence

for an association between birth weight and airflow obstruction in adult life.

18

Acknowledgments

We would like to thank Kay Dickersin for fruitful discussions and comments.

19

References

1. Checkley W, West KP, Wise RA, Baldwin MR, Wu L, LeClerq SC, Christian P, Katz

J, Tielsch JM, Khatry S, Sommer A. Maternal vitamin A supplementation and lung

function in offspring. N Engl J Med 2010;362:1784–94.

2. Bartley M, Kelly Y, Sacker A. Early life financial adversity and respiratory function in

midlife: A prospective birth cohort study. Am J Epidemiol 2012;175:33–42.

3. Morales E, Garcia-Esteban R, Asensio de la Cruz O, Basterrechea M, Lertxundi A,

Martinez López de Dicastillo MD, Zabaleta C, Sunyer J. Intrauterine and early

postnatal exposure to outdoor air pollution and lung function at preschool age. Thorax

2015;70:64–73.

4. Barker D, Osmond C. Childhood respiratory infection and adult chronic bronchitis in

England and Wales. Br Med J 1986;293:1271–5.

5. Barker DJP, Godfrey K, Fal C, Osmond C, Winter P, Shaheen S. Relation of birth

weight and childhood respiratory infection to adult lung function and death from

chronic obstructive airways disease. Br Med J 1991;303:671–675.

6. Barker D. Developmental origins of chronic disease. Public Health 2012;126:185–9.

7. Burney P. Coming off the GOLD standard. Lancet Respir Med 2014;2:174–6.

8. Saad NJ, Patel J, Minelli C, Dickersin K, Burney PGJ. Birth weight and lung function

in later life: a systematic review & meta-analysis. Eur Respir Soc 2015 Int Congr

Amsterdam: 2015.

9. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, Moher D,

Becker BJ, Sipe TA, Thacker SB. Meta-analysis of observational studies in

epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in

Epidemiology (MOOSE) group. JAMA 2000;283:2008–12.

10. Ottowa Hospital Research Institute. The Newcastle-Ottawa Scale (NOS) for assessing

the quality of nonrandomised studies in meta-analyses. 2010; at

http://www.ohri.ca/programs/clinical_epidemiology/ oxford.asp.

11. Suthar AB, Granich R, Mermin J, Van Rie A. Effect of cotrimoxazole on mortality in

HIV-infected adults on antiretroviral therapy: a systematic review and meta-analysis.

Bull World Health Organ 2012;90:128–138C.

12. Dou H, Ma E, Yin L, Jin Y, Wang H. The Association between Gene Polymorphism

of TCF7L2 and Type 2 Diabetes in Chinese Han Population: A Meta-Analysis. PLoS

One 2013;8:e59495.

13. Dersch R, Freitag MH, Schmidt S, Sommer H, Rücker G, Rauer S, Meerpohl JJ.

Efficacy and safety of pharmacological treatments for neuroborreliosis—protocol for a

systematic review. Syst Rev 2014;3:117.

14. Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-

analyses. Br Med J 2003;327:557–560.

15. The World Bank. Country and Lending Group data. 2016;at

<http://data.worldbank.org/about/country-and-lending-groups>.

16. United Nations Statistics Division. Standard Country or Area Codes for Statistical Use.

2013;at <http://millenniumindicators.un.org/unsd/methods/m49/m49.htm>.

17. Orfei L, Strachan DP, Rudnicka a R, Wadsworth MEJ. Early influences on adult lung

function in two national British cohorts. Arch Dis Child 2008;93:570–4.

18. Stein C, Kumaran K, Fall C, Shaheen S, Osmond C, Barker D. Relation of fetal growth

to adult lung function in south India. Thorax 1997;52:895–9.

19. Edwards CA, Osman LM, Godden DJ, Campbell DM, Douglas JG. Relationship

20

between birth weight and adult lung function: Controlling for maternal factors. Thorax

2003;58:1061–1066.

20. Canoy D, Pekkanen J, Elliott P, Pouta A, Laitinen J, Hartikainen A-L, Zitting P, Patel

S, Little MP, Järvelin M-R. Early growth and adult respiratory function in men and

women followed from the fetal period to adulthood. Thorax 2007;62:396–402.

21. Ubilla C, Bustos P, Amigo H, Oyarzun M, Rona RJ. Nutritional status, especially body

mass index, from birth to adulthood and lung function in young adulthood. Ann Hum

Biol 2008;35:322–33.

22. Suresh S, Mamun A a, O’Callaghan M, Sly PD. The impact of birth weight on peak

lung function in young adults. Chest 2012;142:1603–10.

23. Lawlor D, Ebrahim S, Davey Smith G. Association of birth weight with adult lung

function: findings from the British Women’s Heart and Health Study and a meta-

analysis. Thorax 2005;60:851–858.

24. Pei L, Chen G, Mi J, Zhang T, Song X, Chen J, Ji Y, Li C, Zheng X. Low birth weight

and lung function in adulthood: retrospective cohort study in China, 1948-1996.

Pediatrics 2010;125:e899-905.

25. Cai Y, Shaheen SO, Hardy R, Kuh D, Hansell AL. Birth weight, early childhood

growth and lung function in middle to early old age: 1946 British birth cohort. Thorax

2015;1–7.doi:10.1136/thoraxjnl-2014-206457.

26. Lima RDC, Victora CG, Menezes AMB, Barros FC. Respiratory function in

adolescence in relation to low birth weight, preterm delivery, and intrauterine growth

restriction. Chest 2005;128:2400–7.

27. Cheung YB, Karlberg JPE, Low L. Birth weigth and adult lung function in China.

Thorax 2001;56:85.

28. Hancox RJ, Poulton R, Greene JM, McLachlan CR, Pearce MS, Sears MR.

Associations between birth weight, early childhood weight gain and adult lung

function. Thorax 2009;64:228–32.

29. Odberg MD, Sommerfelt K, Markestad T, Elgen IB. Growth and somatic health until

adulthood of low birthweight children. Arch Dis Child 2010;95:F201–F205.

30. Barker D. The fetal and infant origins of adult disease The womb may be more

important than the home. Br Med J 1990;1990.

31. Gilliland FD, Berhane K, McConnell R, Gauderman WJ, Vora H, Rappaport EB, Avol

E, Peters JM. Maternal smoking during pregnancy, environmental tobacco smoke

exposure and childhood lung function. Thorax 2000;55:271–276.

32. Moshammer H, Hoek G, Luttmann-Gibson H, Neuberger M a., Antova T, Gehring U,

Hruba F, Pattenden S, Rudnai P, Slachtova H, Zlotkowska R, Fletcher T. Parental

smoking and lung function in children: An international study. Am J Respir Crit Care

Med 2006;173:1255–1263.

33. Hayatbakhsh MR, Sadasivam S, Mamun a a, Najman JM, Williams GM, O’Callaghan

MJ. Maternal smoking during and after pregnancy and lung function in early

adulthood: a prospective study. Thorax 2009;64:810–4.

34. Christian P, Lee SE, Angel MD, Adair LS, Arifeen SE, Ashorn P, Barros FC, Fall

CHD, Fawzi WW, Hao W, Hu G, Humphrey JH, Huybregts L, Joglekar C V., Kariuki

SK, Kolsteren P, Krishnaveni G V., Liu E, Martorell R, Osrin D, Persson LA,

Ramakrishnan U, Richter L, Roberfroid D, Sania A, Kuile FO Ter, Tielsch J, Victora

CG, Yajnik CS, et al. Risk of childhood undernutrition related to small-for-gestational

age and preterm birth in low- and middle-income countries. Int J Epidemiol

2013;42:1340–1355.

35. McEvoy CT, Schilling D, Clay N, Jackson K, Go MD, Spitale P, Bunten C, Leiva M,

21

Gonzales D, Hollister-Smith J, Durand M, Frei B, Buist a S, Peters D, Morris CD,

Spindel ER. Vitamin C supplementation for pregnant smoking women and pulmonary

function in their newborn infants: a randomized clinical trial. JAMA 2014;311:2074–

82.

36. Saarenpaa H-K, Tikanmaki M, Sipola-Leppanen M, Hovi P, Wehkalampi K, Siltanen

M, Vaarasmaki M, Jarvenpaa A-L, Eriksson JG, Andersson S, Kajantie E. Lung

Function in Very Low Birth Weight Adults. Pediatrics 2015;136:642–650.

37. Vollsæter M, Røksund OD, Eide GE, Markestad T, Halvorsen T. Lung function after

preterm birth: development from mid-childhood to adulthood. Thorax 2013;1–

10.doi:10.1136/thoraxjnl-2012-202980.

38. Vollsæter M, Clemm HH, Satrell E, Eide GE, Røksund OD, Markestad T, Halvorsen

T. Adult Respiratory Outcomes of Extreme Preterm Birth. A Regional Cohort Study.

Ann Am Thorac Soc 2015;12:313–322.

39. Sherrill DL, Guerra S, Wright AL, Morgan WJ, Martinez FD. Relation of early

childhood growth and wheezing phenotypes to adult lung function. Pediatr Pulmonol

2011;46:956–63.

40. Suresh S, O’Callaghan M, Sly PD, Mamun A a. Impact of Childhood Anthropometry

Trends on Adult Lung Function. CHEST J 2015;147:1118.

41. Altman DG, Royston P. Statistics Notes 52: The cost of dichotomising continuous

variables. Br Med J 2006;332:1080.

42. Burney PGJ, Hooper RL. The use of ethnically specific norms for ventilatory function

in African-American and white populations. Int J Epidemiol 2012;41:782–90.

43. Hooper R, Burney P. Cross-sectional relation of ethnicity to ventilatory function in a

West London population. Int J Tuberc Lung Dis 2013;17:400–5.

44. Kelly Y, Panico L, Bartley M, Marmot M, Nazroo J, Sacker A. Why does birthweight

vary among ethnic groups in the UK? Findings from the Millennium Cohort Study. J

Public Health (Bangkok) 2009;31:131–7.

45. Grnøseth R, Vollmer WM, Hardie JA, Ólafsdóttir IS, Lamprecht B, Buist AS, Gnatiuc

L, Gulsvik A, Johannessen A, Enright P. Predictors of dyspnoea prevalence: Results

from the BOLD study. Eur Respir J 2014;43:1610–1620.

46. Burney PGJ, Hooper R. Forced vital capacity, airway obstruction and survival in a

general population sample from the USA. Thorax 2011;66:49–54.

47. Rose G. Strategy of prevention: lessons from cardiovascular disease. Br Med J (Clin

Res Ed) 1981;282:1847–1851.

48. Rose G. Sick Individuals and Sick Populations. Int J Epidemiol 1985;14:32–38.

49. Lærum B., Svanes C, Gulsvik A, Iversen M, Thorarinsdottir H., Gislason T, Jögi R,

Norrman E, Gunnbjörnsdottir M, Wentzel-Larsen T, Janson C, Omenaas E. Is birth

weight related to lung function and asthma symptoms in Nordic-Baltic adults? Respir

Med 2004;98:611–618.

50. Shaheen SO, Sterne J a, Tucker JS, Florey CD. Birth weight, childhood lower

respiratory tract infection, and adult lung function. Thorax 1998;53:549–53.

51. Lopuhaä CE, Roseboom TJ, Osmond C, Barker DJ, Ravelli a C, Bleker OP, van der

Zee JS, van der Meulen JH. Atopy, lung function, and obstructive airways disease after

prenatal exposure to famine. Thorax 2000;55:555–61.

52. Boezen HM, Vonk JM, van Aalderen WMC, Brand PLP, Gerritsen J, Schouten JP,

Boersma ER. Perinatal predictors of respiratory symptoms and lung function at a

young adult age. Eur Respir J 2002;20:383–390.

53. Roca J, Burgos F, Sunyer J, Saez M, Chinn S, Ant?? JM, Rodr??guez-Roisin R,

Quanjer PH, Nowak D, Burney P. References values for forced spirometry. Eur Respir

22

J 1998;11:1354–1362.

54. Quanjer PH, Borsboom GJ, Brunekreef B, Zach M, Forche G, Cotes JE, Sanchis J,

Paoletti P. Spirometric reference values for white European children and adolescents:

Polgar revisited. Pediatr Pulmonol 1995;19:135–42.

55. Hancox RJ, Poulton R, Greene JM, McLachlan CR, Pearce MS, Sears MR.

Associations between birth weight, early childhood weight gain and adult lung

function. Thorax 2009;64:228–232.

23

Tables & Figures

24

Table 1: Characteristics of the 18 studies included. Age, birth weight, gestational age, FVC and FEV1/FVC are shown as mean and standard deviation (SD);

when these were unavailable, median, absolute range (Range) or interquartile range (IQR) were reported instead, as specified in the table; BW: birth weight

Author, year Country Study population Sample

size

Age

(yrs)

Sex

(%

male)

Birth

weight

(kg)

Gestational

age (weeks)

FVC

(L)

FEV1/FVC

(%)

Barker, 1991 (5) England Men born in Hertfordshire, England in 1911-30 639 Range:

59-67 100 - - 3.0 (0.7)1 -

Stein, 1997 (18) India Subjects born in hospital in 1934-53, identified via house-to-house

visits in a two-mile radius 286 47 (4.6) 66 2.8 (0.4) - 2.7 (0.4) 1 79.5 (8.5) 1

Shaheen, 1998 (50) Scotland Follow-up study in 1985-86 of children who lived in St Andrews or

the Fife or Tayside regions 239 58 (4.3) 53 3.4 (0.6) - 3.7 (1.0)3 -

Lopuhaa, 2000 (51) Netherlands Born in hospital in 1943-47; traced and invited for spirometry 733 29 (6.4) 2 462 3.3 (0.5) 2 - 4.3 (0.7)1 73.0 (11)1

Cheung, 2001 (27) China Birth cohort of born in two hospitals in Hong Kong in 1967 120 30 49

3.1

Range:

2.3-4.1

- 3.8 (0.5)1,3 88.6 (6.2)1,3

Boezen, 2002 (52) Netherlands Random sample of newborns in 1975-78 invited for lung function

testing 590 20 (0.9)2 45 3.3 (0.6) 2

Median: 40

Range: 28-44 - -

Edwards, 2003 (19) Scotland Follow-up of random community sample of 1 in 5 children who

attended school in Aberdeen in 1964 323 48 (1.5) 51 3.3 (0.5) 40.3 (1.7) - -

Laerum, 2004 (49)

Denmark,

Norway, Iceland,

Sweden, Estonia

Random subsample of subjects who had competed a questionnaire on

respiratory health invited for lung function 1,494

Range:

20-472 536 3.52 3.2%5 104.7 (13.4)4 -

Lawlor, 2005 (23) United Kingdom Women randomly selected from 23 British towns between 1999 and

2001 2,257

Range:

60-79 0 3.3 (0.8) - - -

Lima, 2005 (26) Brazil All newborns born in 1982 in Pelotas; follow-up in 2000 on all <2.5kg

and on two random controls (BW ≥2.5kg) 354 18 (0.3) 100 3.0 (0.7)3 38.7 (3.3)3 4.9 (0.1)3

-

Canoy, 2007 (20) Finland Birth cohort of individuals born in Northern Finland in 1966 5,390 31 50 3.5 (0.5) 40.3 (1.5) 4.7 (0.7) 84.1 (6.5)

Orfei-1, 2008 (17) Great Britain All British children born 1 week in March 1946. At first follow-up a

stratified subsample by socio-economic status selected 2,167 43 - 3.4 (0.5)3 - 3.7 (0.9)3 -

Orfei-2, 2008 (17) Great Britain All British children born during 1 week in March 1958 and

complemented with immigrants born during the same week. 5,947

Range:

44-45 - 3.4 (0.5)3 - 4.2 (1.0)3 -

Ubilla, 2008 (21) Chile Participants randomly selected from those born in the hospital in

Limache in 1974-78; follow-up conducted in 2001-03 1,221

Median:

25 45 3.2 (0.5) - 4.0 (0.5) 87.0 (4.6)

Hancox, 2009 (28) New Zealand Birth cohort, born in 1972-73 in a hospital and those living Otago at

age 3 invited for follow-up 921 32 53 3.4 (0.5)3 4.0%5 5.0 (0.7)3 78.2 (6.5)3

Odberg, 2010 (29) Norway All non-handicapped low BW children (<2.0kg) born in 1986-88 and

controls (BW>3.0kg) 269 19 54 - 32.2 (3.3)5 115.6 (14.2) 3,4 83.5 (12.3)3

Pei, 2010 (24) China Born in Hospital in 1948-54, traced if living in Bejing to attend a

physical examination in 1995-96 627 45 (1.4)5, 49 3.1 (0.4)3 40 (2.0)2,3 3.5 (0.7)3 -

Sherrill, 2011 (39) USA Healthy newborns enrolled between 1980-84 if parents used a health

organisation in Tucson with follow-up 127 22 55 3.6 (0.5) - - -

Suresh, 2012 (22) Australia Pregnant women in 1981-83 and their offspring; both followed after

up 21 years 2,368 21 50 3.4 (0.5) - 4.6 (0.7) -

25

1 FVC or FEV1/FVC adjusted for height and age (Barker et al. and Stein et al.)(5, 18), for sex, height and smoking (Cheung et al.)(27) or for sex height and age (Lopuhaa et al.)(51).

2 The value refers to the original cohort rather than the subset that performed lung function testing, for which the association with weight in early life was assessed. Except for Pei et al.(24), where the

information was only available for 611 participants. 3 Value not reported in the paper but estimates obtained by combing subgroups. 4 Value reported as percent predicted. Reference equation by Roca et al.(53) used for Laerum et al.(49) and reference equation by Quanjer et al.(54) used for Odberg et al.(29). 5 For Hancox et al. (28) and Laerum et al.(49), only the percentage of preterm births reported with preterm birth defined as gestational age<37 weeks for Hancox et al. (28) and gestational age<36 weeks

for Laerum et al.. For Odberg et al.(29), the mean and standard deviation were only reported for low birth weight participants.

26

Figure 1: Flow chart of inclusion of studies in the systematic review. Five studies could not be

included in meta-analysis for reasons reported in Figure E1.

27

Figure 2: Forests plots for FVC. A) Birth weight as a continuous exposure (9 studies, 12

comparisons); B) Birth weight as a binary exposure (above/below 2.5kg) (5 studies, 6 comparisons).

Orfei et al. (17) included two different study populations, the British 1946 national cohort and the

British 1958 national cohort and were therefore counted as two separate studies. Study estimates were

ordered by increasing age. Some associations were reported by sex and Odberg et al. (29) defined birth

weight as below 2.5 kg and above 3.0 kg.

28

Figure 3: Forest plots for FEV1/FVC. A) Birth weight as a continuous exposure (3 studies, 5

comparisons); B) Birth weight as a binary exposure (above/below 2.5kg) (4 studies, 5 comparisons).

Study estimates were ordered by increasing age. Some associations were reported by sex. Odberg et

al. (29) defined birth weight as below 2.5 kg and above 3.0 kg.

29

Figure 4: Percentage of 18 studies that fulfilled each of eight NOS criteria. Regarding

adjustment for potential confounders (last two items), 3/18 studies did not control the association

with FEV1/FVC for sex, age or current smoking (19, 29, 49).

Figure 5: Funnel plot for FVC. Birth weight as a continuous exposure (9 studies). Beta is the effect

estimate from a linear regression of FVC on birth weight, while s.e. of beta is the standard error of the

effect estimate. The dashed lines represent the pseudo 95% confidence limits.