The association between birthweight and adult lung function: insights from the UK Biobank

Abstract

Background

The potential association between birthweight and adult lung function has been postulated, yet the precise nature of this association in later life remains inconclusive.

Methods

The analysis included 201,615 individuals from the UK Biobank dataset. To identify birthweight subgroups differences in lung function, propensity score matching and a pairwise t-test were conducted. The association between birthweight and lung function was assessed using a linear regression, gradually adjusting for variables. Subgroup analyses were conducted to investigate whether the relationship between birthweight and lung function was modified by variations in age-related lung function changes.

Results

The low birthweight group demonstrated significantly lower forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC) in comparison to those in the adequate birthweight group (P < 0.001). Conversely, no statistically significant distinction was observed in macrosomia. A positive linear correlation between birthweight and lung function was observed within each interval. In the adequate birthweight group, every 1 Kg increase in birthweight was found to be significantly associated with a mean increase in FEV1 of 70.9 mL in males and 56.0 mL in females. Additionally, there was a significant increase in FVC of 86.7 mL in males and 62.2 mL in females. Macrosomia group demonstrated a more pronounced decrease in lung function, in contrast to the adequate birthweight group which exhibited a similar decline pattern to the low birthweight group.

Conclusion

Our study suggests an association between optimal birthweight and later-life respiratory health, as observed through stringent statistical analysis.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12890-026-04174-6.

Background

Lung function serves as a pivotal indicator of overall health within the general population and is closely linked to the occurrence of early comorbidities and all-cause mortality [1, 2]. Throughout the stages of childhood and adolescence, lung function generally exhibits a steady increase, reaching its peak during the mid-20s, followed by a gradual decline as individuals age [3]. This trajectory is influenced by genetic mechanisms, antenatal factors, and a multitude of early-life exposures [4]. It is approximated that low birthweight affects 15–20% of all births worldwide, corresponding to approximately 20 million infants each year [5]. As a result of notable progress in neonatal intensive care technology, there has been a substantial enhancement in the survival rate of infants with extremly low birthweight [6]. Consequently, this development facilitates a more comprehensive exploration of the association between birthweight and lung function across a broader spectrum.

Birthweight offers a comprehensive understanding of the intrauterine environment’s quality, fetal growth and development, and maternal nutritional status [7]. A considerable body of evidence indicates that low birthweight is associated with pathophysiologic aspects of neonatal lung development, and with an increased likelihood of developing lung disease in the future [8]. Measures of lung function, including forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC), are standardized parameters used for evaluating respiratory function. These measures are extensively employed in population health screening to ascertain the overall health condition and identify diseases at an early stage. Examining the association between birthweight and lung function in later years can help identify subpopulation at risk for reduced respiratory capacity and may inform future research into the shared factors linking reduced lung function with nonpulmonary comorbidities [9].

The results obtained from various studies exhibit incongruity, as certain studies indicate a correlation between birthweight and lung function [10, 11], others report no such association [12, 13]. These studies assessed lung function at a singular time point and were conducted at varying stages of life. Consequently, the longitudinal connection between birthweight and adult lung function remains ambiguous. A meta-analysis revealed that a 1 Kg increase in birthweight corresponds to a 59 mL (95% CI 43–76 mL) increase in FEV1 [14]. It is crucial to acknowledge that employing a meta-analysis methodology enables a larger sample size and enhanced statistical potency. However, the influence of the heterogeneity of the original study populations and research designs cannot be disregarded. Furthermore, recent findings indicate that a straightforward linear correlation between birthweight and adult lung function does not exist [15].

In order to address these identified gaps, we have utilized the extensive resources of the UK Biobank study, which is a substantial population-based cohort. The objective of this study is to investigate a quantitative relationship between birthweight and adult lung function.

Methods

Study population

The UK Biobank consists of over 500,000 individuals who were recruited between the ages of 33 and 73 from 2006 to 2010 [16]. The research protocol is accessible to the public. At the outset, participants were requested to provide digitally signed consent, respond to touch-screen questionnaires, and undergo physical and anthropometric assessments. Through health record linkage, the study enables the ongoing monitoring of health events for all participants. Algorithmically determined outcomes are generated for major health conditions. The UK Biobank has obtained ethical approvals from the UK Biobank Research Ethics Committee and Human Tissue Authority.

A total of 502,371 participants were assessed during the survey cycles from 2006 to 2010, with publicly available data being collected. After excluding participants with missing data on interviews or exams, as well as those with invalid spirometry measurements, a total of 202,275 participants were deemed suitable. Further exclusion of participants whose age did not fall within the range of 40 to 70 resulted in a final analysis cohort consisting of 201,615 participants (Fig. 1).

Fig. 1.

Flow chart of participants used in analysis

Birthweight

Birthweight was self-reported at baseline assessment. In order to account for the influence of racial heterogeneity on birth weight, we examined the racial composition of the final sample and determined that all included individuals were Caucasian. Birthweight was categorized using the criteria established by the World Health Organization, which includes classifications such as low birthweight (< 2.5 kg), adequate birthweight (≥ 2.5 kg - <4.0 kg), and macrosomia (≥ 4 kg) [17].

Lung function measurements

The lung function variables analysed in this study were FEV1 and FVC. These variables underwent quality control procedures as outlined in a previous publication [18]. To ensure accuracy, at least two acceptable spirograms were obtained from each participant, with acceptable spirograms defined as having a difference of no more than 250 ml between them. Multiple forced expirations were performed, and the best reading was selected. The Vitalograph Pneumotac 6800 spirometer (Maids Moreton, UK), operated by trained personnel, was used to conduct the tests.

Covariates

This study gathered additional variables that could potentially influence birth weight and lung function, including age, gender, height, weight, smoking status, alcohol consumption, educational level, household income, Townsend deprivation index, and place of birth. smoking and drinking status were classified into three distinct categories, namely never, former, and current usage. Educational levels were categorized as high, encompassing college, university, National Vocational Qualification (NVQ), or other professional qualifications, and low, including A-level, O-level, or CSEs. Household income was dichotomized as less than GBP 31,000 and greater than or equal to GBP 31,000. The place of birth was categorized as either England or elsewhere.

Statistical analysis

The statistical analyses were performed in R statistical software (version 4.3.0). Figures were prepared using GraphPad Prism v9.0.2 and Adobe Illustrator CC v22.0.1. The baseline characteristics of both included and excluded participants were compared using Cohen’s w and Cohen’s d effect size. For categorical variables, a value of < 0.1 indicates a small difference between groups, while for continuous variables, a value of < 0.2 indicates a small difference between groups [19]. The between-group differences of continuous and categorical variables at baseline were assessed using analysis of variance and chi-square test. To improve comparability across birthweight groups by balancing observed covariates, propensity score matching (PSM) was conducted using the TriMatch package of the R statistical software. Participantswere matched at a 1:1:1 ratio, using nearest neighbor matching with a caliper distance of 0.2. exact matching was applied for age, gender, and height. Then, a pairwise t-test for group difference was conducted.

The linear and restricted cubic spline specifications were utilized to examine the shape of the association between birthweight and lung function, employing both statistical and graphical analyses. In order to assess the association of birthweight and lung function, we employed a linear regression within the range of variation trends. These regressions were sequentially adjusted for variables that may be linked to both lung function and birthweight: age, height, weight, smoking status, alcohol consumption, educational level, household income, Townsend deprivation index, and place of birth. Furthermore, subgroup analyses were conducted to investigate whether the relationship between birthweight and lung function was modified by variations in age-related lung function changes. The modifying effect was examined by adding a multiplicative interaction term. Differences were considered statistically significant for P value ≤ 0.05.

Results

From the total UK biobank population, 201,615 participants aged 40–70 years (mean age 55.6 ± 8.00 years) were included in the final analyses (Fig. 1), of whom 121,503 (60.2%) were female. The baseline comparison between included and excluded participants is given in Table S1. In general, the excluded individuals exhibited comparable birthweight and lung function. The demographic and socioeconomic attributes displayed a nearly random distribution, with the exception of the excluded group having a higher average age and a larger proportion of households with income below GBP 3100.

Based on the birthweight, participants were divided into three groups: low birthweight (< 2.5 kg), adequate birthweight (≥ 2.5 kg - <4.0 kg), and macrosomia (≥ 4 kg). We observed statistically significant differences in either FEV1 or FVC among groups (Table 1). It is important to highlight that there were significant differences among groups for all baseline data (P < 0.001). In order to address the uneven distribution of baseline features among groups, we employed PSM for subsequent analysis. The implementation of PSM led to a reduction in the standardized mean differences of all variables compared to pre-matching values, demonstrating improved balance of covariates​ (Fig. S1). Subsequently, we conducted pairwise t-tests on the 10,011 matched triplets resulting from PSM (Table 2). Our study revealed that the low birthweight group exhibited significantly lower FEV1 and FVC compared to the adequate birthweight group (P < 0.001). However, no significant difference was observed in macrosomia (P = 0.374 for FEV1; P = 0.911 for FVC).

Table 1.

Association between birthweight and lung function

Variables	Birthweight
	< 2.5 Kg (n = 19751)	2.5–4 Kg (n = 154679)	≥ 4 Kg (n = 27185)	P
Age, mean ± SD	56.80 ± 7.85	55.28 ± 7.98	56.43 ± 8.06	< 0.001
Male, n (%)	5732 (29.02)	59,837 (38.68)	14,543 (53.50)	< 0.001
Height, mean ± SD	164.12 ± 8.67	168.25 ± 8.98	172.51 ± 9.37	< 0.001
Weight, mean ± SD	74.38 ± 15.62	77.05 ± 15.70	83.54 ± 16.79	< 0.001
Smoking status, n (%)				< 0.001
Never	11,643 (58.95)	87,247 (56.41)	13,697 (50.38)
Former	6171 (31.24)	52,516 (33.95)	10,599 (38.99)
Current	1937 (9.81)	14,916 (9.64)	2889 (10.63)
Drinking status, n (%)				< 0.001
Never	783 (3.96)	4354 (2.81)	707 (2.60)
Former	685 (3.47)	4664 (3.02)	922 (3.39)
Current	18,283 (92.57)	145,661 (94.17)	25,556 (94.01)
Household income, n (%)				< 0.001
< GBP 31,000	10,206 (51.67)	64,294 (41.57)	12,016 (44.20)
≥GBP 31,000	9545 (48.33)	90,385 (58.43)	15,169 (55.80)
Education, n (%)				< 0.001
A-level, O-level or CSEs	9874 (49.99)	80,827 (52.25)	14,511 (53.38)
College, university, NVQ or other professional qualifications	9877 (50.01)	73,852 (47.75)	12,674 (46.62)
Townsend deprivation index, mean ± SD	−1.39 ± 3.00	−1.62 ± 2.89	−1.57 ± 2.93	< 0.001
Birth place, n (%)				0.001
England	16,135 (81.69)	126,958 (82.08)	22,062 (81.16)
Elsewhere	3616 (18.31)	27,721 (17.92)	5123 (18.84)
FEV1, mean ± SD	2.56 ± 0.71	2.86 ± 0.76	3.06 ± 0.83	< 0.001
FVC, mean ± SD	3.39 ± 0.89	3.76 ± 0.96	4.06 ± 1.04	< 0.001

FEV1 forced expiratory volume in 1 s, FVC forced vital capacity, BMI body mass index, GBP Great Britain Pound, NVQ National Vocational Qualification

Table 2.

Differences in FEV1 and FVC in the groups after propensity score matching

Birthweight	FEV1			FVC
	Mean, L	SD	p	Mean, L	SD	P
2.5–4 Kg	2.91	0.82	Reference	3.86	1.05	Reference
< 2.5 Kg	2.85	0.83	< 0.001	3.79	1.06	< 0.001
≥ 4 Kg	2.92	0.83	0.374	3.85	1.05	0.911

FEV1 forced expiratory volume in 1 s, FVC forced vital capacity, SD standard deviation

In order to describe a more precise association between birthweight and lung function, the Pearson correlation test was utilized to estimate correlation coefficients (r) and p-values (Fig. S2). The findings indicate a relatively weak correlation between birthweight and FEV1 (R = 0.15, P < 0.001 for females; R = 0.095, P < 0.001 for males) as well as FVC (R = 0.15, P < 0.001 for females; R = 0.12, P < 0.001 for males). Consequently, the shape of these associations was further examined using restricted cubic spline models (Fig. 2). Notably, significant nonlinear associations between birthweight and both FEV1 and FVC were observed for both genders (P < 0.001). Visual analysis revealed that individuals with higher birthweight exhibited higher levels of FEV1 and FVC. It is important to highlight that FEV1 and FVC demonstrated the most pronounced increase in the adequate birthweight range (≥ 2.5 kg - <4.0 kg), particularly among males.

Fig. 2.

Nonlinear relationship between birthweight and lung function. Relationship between birthweight and FEV1 in male (A); FEV1 in female (B); FVC in male (C); FVC in female (D). Analyses were adjusted for age, height, weight, smoking status, drinking status, educational level, household income, Townsend deprivation index, and place of birth. Solid blue line with Shaded areas delimit the mean ± standard deviation. FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity

As FEV1 and FVC exhibited distinct variation trends in different birthweight ranges, we constructed a linear regression model within each region separately (Fig. 3). In detail, either FEV1 or FVC showed significant increases with higher birthweight in all the birthweight subgroups without adjusting for other variables (P < 0.001). After adjusting for various factors including age, height, weight, smoking and drinking habits, educational level, household income, Townsend deprivation index, and place of birth, the subgroup of individuals with a birthweight below 2.5 Kg exhibited an increase in FEV1 of 42.7 (95% CI 25.8 to 59.6; P = 0.012) mL/Kg birthweight in males and 37.8 (95% CI 28.8 to 46.8, P < 0.001) mL/Kg birthweight in females. Additionally, the FVC increased by 28.4 (95% CI 17.8 to 39.0; P = 0.007) mL/Kg birthweight in females, while no statistically significant difference was observed in FVC among males (P = 0.065). In the subgroup of individuals with a birthweight between 2.5 and 4 Kg, a 1 Kg increase in birthweight was associated with a significant increase in FEV1 of 70.9 mL (95% CI 63.7 to 78.1, P < 0.001) in males and 56.0 mL (95% CI 52.1 to 59.9, P < 0.001) in females, after adjusting for various factors. Similarly, there was a significant increase in FVC of 86.7 mL (95% CI 78.3 to 95.1, P < 0.001) in males and 62.2 mL (95% CI 57.5 to 66.9, P < 0.001) in females.

Fig. 3.

Association between birthweight and FEV1 (A) and FVC (B) in linear regression models. Model 1: adjusted by age; Model 2: adjusted by age, height, weight; Model 3: adjusted by age, height, weight, smoking status, drinking status; Model 4: adjusted by age, height, weight, smoking status, drinking status, educational level, household income, Townsend deprivation index, and place of birth. FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity

However, among individuals with a birthweight exceeding 4 Kg, our findings indicated that the relationship between birthweight and both FEV1 and FVC did not exhibit a positive correlation, with the exception of FVC in males (mean 31.9, 95% CI 17.3 to 46.5, P = 0.029).

A gradual decline in FEV1 and FVC was observed across birthweight subgroups, as showed in Fig. 4. In both males and females, the highest values for both FEV1 and FVC were consistently observed in the subgroup with a birthweight of ≥ 4 Kg, while the lowest values were consistently observed in the subgroup with a birthweight of < 2.5 Kg, spanning from ages 40 to 70. However, the rate of decline differed between these subgroups. Specifically, the macrosomia group exhibited a more rapid decline (between-group P < 0.05), whereas the low birthweight group displayed a nearly parallel descent pattern.

Fig. 4.

Lung function measures across age groups by birthweight category. The decrement of FEV1 (A) and FVC (B) in male with advancing age. The decrement of FEV1 (C) and FVC (D) in female with advancing age. P values were calculated by the interaction between groups using fitting interaction terms within regression models. All analyses were adjusted by height, weight, smoking status, alcohol consumption, educational level, household income, Townsend deprivation index, and place of birth

Discussion

Based on data collected from a sample of 201,615 individuals from the UK Biobank, our study observed significant associations between birthweight and lung function measures. We found that both FEV1 and FVC were lower in individuals with a birthweight < 2.5 Kg compared to those with adequate birthweight. Furthermore, a positive correlation between birthweight and lung function was observed within the range of 2.5 to 4 kg, characterized by a linear relationship across intervals —though this should be interpreted as descriptive rather than demonstrative of a strict dose-effect relationship. Beyond these associations, we also identified differential patterns in the decline​ of lung function with age across birthweight subgroups. The observed associations between birthweight and variations in lung function may inform future refinements of reference equations, potentially contributing to more comprehensive evaluations of lung function.

The consideration of the biological and physiological factors of lung development in relation to birthweight is crucial in understanding the implications of our findings. Extensive research has suggested that low birthweight may serve as an indicator of suboptimal intrauterine conditions, which has been associated with variations in lung development [20, 21]. Moreover, the development of the respiratory system is intricately associated with both gestational age and birth weight, as evidenced by the delayed lung growth and development often observed in low-birthweight infants [22]. A comprehensive review of existing literature substantiates that individuals with low birthweight consistently exhibit lower lung function, which aligns with our findings [23].This suggests that the association between birthweight and lung development may not be temporary but rather persist​ into later stages of life, with potential implications for understanding patterns of lung function change between ages 40 and 70.

The association between birth weight and lung function is apparent, with the most significant increase in lung function occurring within the weight range of 2.5 to 4 kg. It is probable that the fetal environment during this range facilitates optimal lung development [23, 24]. The presence of a nonlinear association beyond this range may reflect distinct physiological patterns, such as correlations with obesity and related comorbidities, which could be associated with the relationship between higher birthweight and lung function measures [15, 25]. This exploratory analysis revealed nonlinear patterns in the association between birthweight and lung function, characterized by varying slopes across birthweight ranges [25]. However, these patterns should be interpreted as descriptive rather than evidence of a statistically tested threshold effect. To contextualize the steeper decline observed in the macrosomia group,​ several factors must be considered. First, higher baseline lung function in macrosomia individuals may affect perceived decline rates due to mathematical coupling. Second, unmeasured factors like cardiometabolic conditions or obesity, more common in high birthweight individuals, could confound results [26]. Third, these patterns are descriptive associations, not proof of causation, as our study design cannot determine whether birthweight has a direct effect on lung aging. Speculatively, macrosomia may be linked to metabolic dysregulation (e.g., fetal hyperinsulinemia and insulin resistance) and adult obesity traits, which could contribute to accelerated lung function decline through inflammatory or mechanical pathways [27–31]. However, this remains speculative, and future studies with detailed longitudinal data on metabolic health are needed to verify this hypothesis.

The associations observed in our study may have implications for understanding respiratory health patterns. The correlation between birthweight and lung function suggests that birthweight could be considered as one of the multiple factors in respiratory health assessment [32]. These findings contribute to the body of knowledge that might eventually inform strategies for identifying individuals with different lung function trajectories, though further research is needed before considering clinical applications. Importantly, the nonlinear associations observed via restricted cubic splines are hypothesis-generating and require validation through formal threshold detection methods (e.g., piecewise regression) in future studies.

The observed effect sizes between birthweight and lung function, though statistically significant, are not clinically meaningful in mid-to-late adulthood. Our study found FEV1 differences of 37.8 to 86.7 mL per kg birthweight, which fall below the clinically important difference of 100 to 140 mL for chronic respiratory diseases [33]. While birthweight is associated with lung function at a population level, these findings are more relevant for understanding population-level lung health patterns over a lifetime rather than for individual clinical decisions. However, the persistence of these associations into later adulthood highlights a potential long-term link between early-life factors and respiratory aging, even if the differences remain subclinical.

We used PSM with nearest-neighbor matching with a 0.2 caliper to balance baseline covariates like age, gender, and height, which differed significantly across birthweight categories and are key determinants of lung function. This approach aimed to improve covariate balance, as shown by standardized mean differences before and after matching (Fig. S1). However, PSM has limitations: it cannot address unmeasured confounders like genetic factors or exact gestational age, and matching on adult height may introduce overadjustment bias, which could influence the associations we aimed to estimate. Importantly, the results from PSM were consistent with those from multivariable regression (Fig. 3), elucidating that​ PSM only supports associational, not causal, interpretations.

To the best of our knowledge, the current study represents the most extensive sample size to date in examining the correlation between birthweight and adult lung function within a population-based investigation. This advantage allowed us to perform PSM analysis, thereby supporting the robustness of our observed associations within the matched sample. Furthermore, it allows for the evaluation of a linear association between birthweight and lung function across a continuous age range. Despite the potential for robust statistical power achieved through previous meta-analyses by integrating data from multiple studies [14], the presence of heterogeneity in terms of ethnicity, methodologies, and various confounding factors is inevitable, which may compromise the reliability of the findings [34, 35]. First, we recognize that birthweight data were self-reported, potentially introducing recall bias. Nonetheless, prior studies support the idea that self-reported birthweights, despite their imperfections, are adequately accurate for research purposes, especially when precise measurement is not essential and the sample size is sufficiently large [36]. Any misclassification is likely to be non-differential, which could attenuate the observed associations. Second, our study exclusively involved individuals of Caucasian ancestry from the UK Biobank, thereby constraining the generalizability of our findings to other ethnic populations. Future research involving multi-ethnic cohorts is essential to determine whether these associations differ across diverse racial and ethnic groups, especially considering the well-documented variations in lung function reference values and birthweight distributions among different populations [37]. Third, the lack of detailed gestational age data prevents us from distinguishing between preterm birth and small-for-gestational-age (SGA) status, which are differentially associated with lung development. Preterm birth affects alveolarization and surfactant production, while SGA is linked to placental insufficiency and altered growth [38]. This limitation may obscure specific associations between birthweight and lung function, as SGA infants with appropriate gestational age may have different lung outcomes than preterm infants of similar birthweight. Fourth, despite adjusting for numerous confounders, we acknowledge the potential influence of unmeasured early-life factors on our findings. Our study lacked data on crucial determinants of lung development, such as early-life respiratory infections, maternal smoking during pregnancy, detailed prenatal nutrition indicators, and childhood socioeconomic and environmental conditions [39, 40]. These factors may partially account for the observed association between birthweight and adult lung function, as they are associated with both fetal growth and long-term respiratory health. Although we employed PSM and multivariate adjustment to mitigate confounding, the absence of these critical early-life variables represents a significant limitation.

Conclusion

Our analysis of the UK Biobank cohort suggests that birthweight is associated with lung function in middle to late adulthood. We observed that both low (< 2.5 kg) and high (> 4 kg) birthweights are associated with differences in lung function parameters compared to the adequate birthweight range. These findings underscore birthweight as an early-life factor associated with variations in respiratory health measures later in life.

Authors’ contributions

Y.W., and Y.Z., contributed to the conception and design of the study, data analysis and draft the manuscript. Y.B., involved in the data collection and interpretation of the data. P.L., T.J., and Y.L., conducted the statistical analyses and graph visualization. Z.C., and Y.Z., (Yongjian Zhu) were responsible for the conception, interpretation of the data, review and edit the manuscript. All authors participated in revising and editing the manuscript and have read and approved the final manuscript.

Funding

This study was supported by National Natural Science Foundation of China (82170037 and 82000015). Key scientific research projects, Science and Technology Department of Henan Province (231111310800).

Data availability

UK Biobank data are publicly available upon request to UK Biobank: (https://www.ukbiobank.ac.uk).

Declarations

Ethics approval and consent to participate

The analyses were conducted using the UK Biobank Resource under application number 93398. The UK Biobank received ethical approval from the Research Ethics Committee (REC reference for the UK Biobank is 11/NW/0382).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.