Association of childhood growth with adult lung function: A prospective urban birth cohort study from India

Abstract

Background and objectives

While some studies suggest that lower birth weight is associated with poorer adult lung function, findings regarding early childhood, adolescent, and adult growth variables remain inconsistent. We conducted this study to determine whether low lung function in Indians is related to lower birthweight and early growth.

Methods

The New Delhi Birth Cohort (NDBC) was used as the study population to examine patterns and periods of growth in infancy, childhood, and adolescence using a conditional growth model. Lung function data were collected using spirometry and the forced oscillation technique at approximately 50 yr of age. We further examined associations of conditional size measures with adult lung function outcomes using multiple linear regression models.

Results

Lung function was positively associated with weight till 5 yr of age. Further, an increase of 1 standard deviation (SD) in conditional weight between 5-10 yr resulted in a decrease in forced vital capacity (FVC) of 50 mL [95% confidence interval (CI) -0.10 to 0.00 (P= 0.08)]. Between 10-50 yr of age, a negative association was observed between weight and lung function. Positive associations of some of the measures of lung function were observed with length at birth, 2 yr, adolescent height, and final adult height. An increase of 1 SD in conditional height between birth and two years resulted in an increase in forced expiratory volume in 1 second (FEV1) of 130 mL (95% CI, 0.09 to 0.17).

Interpretation and conclusion

Birth weight and birth length, and greater than predicted linear growth throughout childhood, are independently associated with better adult lung function, in terms of spirometry and oscillometry. It is plausible that greater than predicted weight gain after 5 yr of age or during adolescence, or adulthood has an inverse association with lung function.

Birth weight and postnatal growth are important predictors of adult health.¹ The association of low birth weight with lower adult lung function is well established.² Adult body weight has an inverse relation with lung function.³ However, the association of childhood growth measures with lung function is not established.^4-6

The alveolar phase of lung development begins in utero and continues till approximately 8 yr of age.⁷ After this, alveolarisation is complete, and further growth is primarily of size. Differences in growth during these phases may not be fully captured by spirometry. Peak lung function, measured as forced expiratory volume in 1 second (FEV1) and forced vital capacity (FVC, total exhaled volume), is attained around 20 yr of age.⁸ Lung impedance, measured using oscillation technique, can be partitioned into parameters reflecting lung mechanics that may be more sensitive to differences in early lung growth than measures derived from spirometry,⁹ and lung oscillometry can be used to screen for early small airway changes, which progress to abnormal spirometry over a period of years.^10,11

Low lung function is partially related to socioeconomic factors and lifestyle, which are indirectly related to physical development.¹² Additionally, low and middle income countries (LMIC) like India are challenged by the dual burden of persistent undernutrition and emerging obesity, making information regarding the effects of linear growth faltering and relative weight gain in early life critical.^13,14

We are not aware of other studies from LMICs with rich growth data collected longitudinally from birth to adulthood with follow up of lung function assessment in middle age. The New Delhi Birth Cohort (NDBC) has tracked individuals born within a defined area of New Delhi, India, since their birth (1969-1973).¹⁵ Using data from this cohort, we evaluated the association of patterns of growth in infancy, childhood, and adolescence with adult lung function.

Methods

This prospective study was undertaken by the department of Genomics and Genome Biology, Institute of Genomics and Integrative Biology (IGIB), New Delhi, India after obtaining the ethical clearance from the Institutional Review Board. An informed written consent was also obtained from all participants.

The New Delhi birth cohort (NDBC)

Details of the NDBC have been published.¹⁵ Briefly, in the years 1969–1972, in a 12 km² area of South Delhi, eligible married couples were prospectively followed up for pregnancy, and 8181 singleton live births were recruited. The birth weight and length of the babies were recorded within 72 h of birth and thereafter 6 monthly until the age of 21 yr. Subsequently, 1526 men and women from the original cohort were retraced and recruited for a study of cardiometabolic risk factors between the years 2000 and 2003, at an average age of 30 yr. Data on socioeconomic status, attained education, family history of disease, tobacco and alcohol consumption, diet, and physical activity were obtained using standardised questionnaires and clinical assessments. These cohort participants have been followed prospectively since then.

Study participants and setting

We used the 2000-2003 sample as our source population for tracing. Data collection was performed for 714 NDBC members from January 30, 2021 to December 30, 2021 in an office setting and between January 2, 2022 and July 30, 2022 in field camps ( Fig. 1).

Fig. 1.

Flowchart depicting the enrolment and attrition of New Delhi Birth Cohort members.

Lung function assessment

Spirometry was conducted according to the American Thoracic Society recommendations¹⁶ for subject manoeuvre, techniques, and quality control using a portable device (EasyOne Air, NDD, Zurich, Switzerland).¹⁷ Testing was performed in a quiet and comfortable environment by trained technicians in the participants’ preferred posture (seated erect or in standing position). A disposable in-line filter and mouthpiece were used for each participant and discarded thereafter. No pre-test or post-test bronchodilator was used for any participant. A disposable nose clip was used for some participants. The EasyOne Air spirometer is an ultrasound-based spirometer utilising a dual-beam Doppler approach to flow measurement and has a Windows-based software programme to collect, calculate, and store final spirometry data. EasyOne Air comes with lifetime calibration.

Airway oscillometry was performed using the forced oscillation technique (FOT) in a seated position prior to spirometry during tidal breathing (FOT device: Tremoflo C-100, Thorasys Labs, Montreal, Canada).¹⁸ Briefly, FOT permits passive measurement of lung mechanics from both smaller and larger airways. Sound waves transmitted sequentially are superimposed on normal tidal breathing, and disturbances in the flow and pressure of the external waves are used to calculate resistance (R), reactance (X), and reactance curve area (AX) at different frequencies. Oscillatory parameters were expressed as the mean value of three acceptable readings of at least 6 breathing cycles each.

Outcome variable definitions

FEV1 measurements were considered acceptable if there were no early procedural terminations, variable efforts, leaks, obstructed mouthpieces, or artifacts (e.g., cough, glottis closure); if the back-extrapolated volume was less than 150 mL or 5% of FVC.

Acceptability criteria for FOT included at least six breaths free from artefact due to occlusion, leak, or drift, or extreme (>5 SD of mean) or negative resistance (tremoFlo®software build 1.0.43.39). Calibration was performed every day using the calibrator load provided by Thorasys.¹⁹

Anthropometry and other measurements

Measurements included weight to the nearest 0.1 kg (SECA digital weighing scale Clara 803, Hammer Steindamm, Germany) and height to the nearest 1 mm, measured (SECA portable mechanical stadiometer 217, Hammer Steindamm, Germany).

A participant questionnaire, used in previous waves of data collection of the NDBC, was administered to obtain information on socio-economics status, tobacco use, and medical history. The WHO validated general physical activity questionnaire was used to estimate physical activity.²⁰ Smoking was recorded as never, former, and current smoker.

Sample size calculation

Prior to initiating field work, we estimated the available cohort to be 970, which would have provided 80% power to detect an effect size of 0.13. We collected lung function data for 714 individuals, but after quality assessment, 613 individuals had acceptable data. Our final sample size of 613 provides 80% power to detect an effect size of 0.16.²¹

Statistical analysis

Since only 8.7% of the original birth cohort participated in the present study, the subjects may well be unrepresentative of the cohort as a whole. To test the representativeness of the current participants and assess the risk of bias, we conducted a sensitivity analysis in which we compared early-life anthropometric variables (birth length, birth weight, and height and weight in infancy, childhood, and early adulthood) with members of the original cohort who did not participate in the current follow-up. There were minor differences in birth weight (full sample (original cohort in yr 1969)=2.82 kg, study sample (current phase in yr 2021)=2.88 kg) and length (48.4 cm and 48.7 cm, respectively) measurements between the final study sample and the full original cohort (Supplementary Table I).

Assessment of growth

Measurements of weight and height were repeated in the same individual throughout their lifespan and are expressed as z-scores reflecting differences from the reference population. For the present analyses, we used measures of length/height and weight at birth and at 2, 5, 10, 15, and 50 yr. Growth parameters are highly correlated and thus unsuitable for regression analysis due to collinearity. As described elsewhere, we derived height- and weight-specific conditional growth measures which express the participant’s growth trajectory as a series of uncorrelated variables.²² Conditional variables represent a child’s deviation from their expected size based on their own previous measures and the growth of the other children in the cohort. These conditional measures are used as exposures in our analyses.

A thorough literature search and previous physiological knowledge were used to select potential confounders. We developed multiple linear regression models with the conditional growth variables derived from birth to young adulthood as exposures and the lung function parameters as outcomes. Next, we tested for confounding for candidate variables, i.e., sex, wealth index, physical activity, and smoking, using a likelihood ratio test. For each potential confounder, a likelihood ratio P>0.5 and a difference in regression coefficients between the full and reduced model was <10% resulted in the removal of the variable unless it was known from published literature to be a confounder (Supplementary Table II). Each time a confounder was removed, another full and reduced model was generated. A separate regression model was run for each outcome. Following this process, sex and smoking status were identified as confounders and were retained in all regression models. R (version 4.2.2) was used for data analysis and visualisation.²³

Results

The characteristics of the 714 study participants are summarised in Table I and Supplementary Table III. 665 individuals performed spirometry tests of acceptable quality, and 640 participants performed FOT tests of acceptable quality ( Table II). Total of 107 (16%) participants had obstructive lung disease.

Table I.

Summary characteristics of participants (N=714) stratified by sex

	Parameter	Men (N=463)			Women (N=251)
		n	Mean or % or median	SD or IQR	n	Mean or % or median	SD or IQR
1.	Age at participation in current study (yr)		50.1	0.97	251	50.1	0.95
2.	Schooling attainment (%)
	Middle school or less	47	10.2	-	9	3.6	-
	High school or diploma	161	34.8	-	65	25.89	-
	Graduate	200	43.2	-	143	56.97	-
	Postgraduate or Professional	55	11.9	-	34	13.54	-
3.	Occupation of participant (%)
	Homemaker	13	2.8	-	160	63.7	-
	Unemployed	9	1.9	-	0	0	-
	Unskilled or semi-skilled manual	53	11.5	-	2	0.8	-
	Skilled manual	256	55.3	-	46	18.3	-
	Non-manual, business or professional	132	28.5	-	43	17.1	-
4.	No. of people living together in the house		4.5	2	251	4.2	1
5.	Smoking (%)
	Never	346	75	-	246	98	-
	Former	11	2	-	0	0	-
	Current	106	23	-	5	2	-

Table II.

Lung function measures among cohort participants

Lung function parameters	n	Mean (SD)
Forced vital capacity (FVC) (in L)	665	2.9 (0.8)
Forced expiratory volume in the 1st second (FEV1) (in L)	665	2.2 (0.6)
Resistance at 5Hz (R5) (in cmH2O*L-1*s)	640	4.4 (1.3)
Resistance at 31Hz (R31) (in cmH2O*L-1*s)	640	3.3 (0.8)
Difference of resistance at 5Hz and that at 20 Hz (R5-R20) (in cmH2O*L-1*s)	640	1.3 (0.8)
Reactance at 5 Hz (X5) (in cmH2O*L-1*s)	640	-2.1 (1.1)
Reactance at 31 Hz (X31) (in cmH2O*L-1*s)	640	0.6 (0.4)
Area under the curve (AX) (in cmH2O*L-1)	640	18.4 (12.3)
Resonant frequency (frees) (in Hz)	640	21.4 (4.7)

Regression models were developed using data from 613 participants with complete anthropometric data and acceptable lung function tests. Of the 111 participants who reported being current smokers, 18 (all males) reported more than 40 pack years.

We observed a positive association of all six lung function parameters with birth weight. The association with conditional weight weakened through childhood and became inverse by age 10; the inverse association strengthened thereafter. For example, birth weight was positively associated with FVC [coefficient 0.09; confidence intervals (CI) 0.04, 0.14], and FEV1 (0.05 [0.01, 0.09]) ( Fig. 2).

Fig. 2.

Relationship between conditional weight at different ages and lung function. The coefficients plotted for each age along with the error bars depicting 95% confidence interval have been derived using multiple linear regression model. Model adjusted for sex and smoking.

Birth length was positively associated with FEV1, FVC, and X5 at 50 yr ( Tables III and IV). Conditional measures of length/height at ages 2, 10, and 20 yr were positively associated with these same three measures ( Fig. 3). Conversely, birth length and conditional measures of linear growth were inversely associated with R5, R5-20, and AX at 50 yr ( Table IV).

Table III.

Coefficients for the linear regression model for spirometry outcome measures. Model has been adjusted for sex and smoking status (N=613)

Size measures	Forced vital capacity (in L) (95% confidence interval)	P value	Forced expiratory volume in 1st second (in L) (95% Confidence interval)	P value
Length at birth	0.06 (0.01, 0.10)	0.02	0.07 (0.03, 0.10)	<0.001
Conditional height at 2 yr	0.16 (0.11, 0.21)	<0.001	0.13 (0.09, 0.17)	<0.001
Conditional height at 5 yr	0.04 (-0.01, 0.08)	0.11	0.02 (-0.01, 0.06)	0.20
Conditional height at 10 yr	0.03 (-0.02, 0.07)	0.24	0.003 (-0.04, 0.04)	0.86
Conditional height at 15 yr	0.06 (0.01, 0.11)	0.02	0.04 (0.00, 0.08)	0.06
Conditional height at 50 yr	0.19 (0.15, 0.24)	<0.001	0.12 (0.09, 0.16)	<0.001
Conditional weight at birth	0.09 (0.04, 0.14)	<0.001	0.05 (0.01, 0.09)	0.01
Conditional weight at 2 yr	0.02 (-0.03, 0.06)	0.53	0.007 (-0.03, 0.05)	0.73
Conditional weight at 5 yr	0.03 (-0.02, 0.08)	0.37	0.02 (-0.02, 0.06)	0.29
Conditional weight at 10 yr	-0.05 (-0.10, 0.00)	0.08	-0.04 (-0.08, 0.00)	0.08
Conditional weight at 15 yr	-0.08 (-0.13, -0.03)	0.002	-0.07 (-0.11, -0.03)	0.002
Conditional weight at 50 yr	-0.07 (-0.12, -0.02)	0.003	-0.04 (-0.08, 0.00)	0.03

Table IV.

Coefficients for linear regression model for oscillometry outcome measures. Model has been adjusted for sex and smoking status (N=613)

Size measures	Resistance at 5 Hz (R5) (in cmH20.s/L) (95% confidence interval)	P value	Resistance at 5 Hz – resistance at 20 Hz (R5-20) (in cmH20.s/L) (95% confidence interval)	P value	Reactance at 5 Hz (X5) (in cmH20.s/L) (95% confidence interval)	P value	Area under the curve (Ax) (in cmH20/L) (95% confidence interval)	P value
Length at birth	-0.16 (-0.25, -0.07)	0.001	-0.08 (-0.14, -0.01)	0.02	0.15 (0.08, 0.23)	<0.001	-1.46 (-2.30, -0.60)	<0.001
Conditional height at 2 yr	-0.15 [-0.24, -0.06]	0.001	-0.04 [-0.10, 0.02]	0.17	0.15 [0.08, 0.23]	<0.001	-1.60 [-2.46, -0.75]	<0.001
Conditional height at 5 yr	-0.02 [-0.11, 0.07]	0.59	-0.02 [- 0.08, 0.05]	0.62	-0.01 [-0.09, 0.06]	0.74	0.30 [-0.55, 1.16]	0.48
Conditional height at 10 yr	0.01 [-0.08, 0.10]	0.85	0.07 [0.01, 0.13]	0.02	0.001 [-0.08, 0.08]	0.98	0.28 [-0.58, 1.14]	0.52
Conditional height at 15 yr	-0.08 [-0.18, 0.01]	0.07	-0.06 [-0.13, 0.00]	0.04	0.01 [-0.06, 0.09]	0.78	-0.44 [-1.30, 0.42]	0.33
Conditional height at 50 yr	-0.17[-0.26,-0.08]	<0.001	-0.11 [-0.18, -0.05]	<0.001	0.18 [0.10, 0.25]	<0.001	-1.80 [-2.69, -0.96]	<0.001
Conditional weight at birth	-0.02 [- 0.11, 0.07]	0.66	-0.04 [- 0.10, 0.02]	0.23	0.05 [- 0.03, 0.12]	0.22	-0.70 [-1.56, 0.14]	0.10
Conditional weight at 2 yr	0.07 [-0.02, 0.16]	0.13	0.04 [-0.02, 0.10]	0.24	0.001 [-0.08, 0.08]	0.98	-0.18 [-1.04, 0.67]	0.69
Conditional weight at 5 yr	0.02 [-0.08, 0.11]	0.71	0.02 [-0.05, 0.08]	0.54	-0.04 [-0.11, 0.04]	0.37	0.21 [-0.65, 1.08]	0.63
Conditional weight at 10 yr	0.16 [0.07, 0.25]	<0.001	0.08 [0.02, 0.14]	0.01	-0.08 [-0.16, -0.01]	0.04	1.30 [0.43, 2.17]	0.01
Conditional weight at 15 yr	0.11 [0.02, 0.21]	0.40	0.03 [-0.03, 0.09]	0.37	-0.08[-0.16, 0.00]	0.04	0.66 [-0.21, 1.52]	0.14
Conditional weight at 50 yr	0.29[0.19, 0.38]	<0.001	0.11 [0.05, 0.17]	<0.001	-0.18 [-0.25, -0.10]	<0.001	[1.02, 2.77]	<0.001

Fig. 3.

Relationship between conditional height at different ages and lung function. The coefficients plotted for each age along with the error bars depicting 95% confidence interval have been derived using multiple linear regression model. Model adjusted for sex and smoking.

Discussion

Our study suggests that greater birth weight and birth length, and greater than predicted linear growth throughout childhood, are associated with better adult lung function, but greater than predicted weight gain after 5 yr of age or during adolescence or adulthood has an inverse association with lung function. It is important to note that increased weight in this context is relative to the average for the population and doesn’t necessarily imply that individuals are overweight. Weight gain is a result of linear growth and soft tissue gain (fat-free mass and fat mass); our conditional relative weight variables represent weight change that is separated from change in height.

In addition to spirometry, we also obtained oscillometry data, which is a first for a birth cohort. R5, R5-R20, and Ax were inversely correlated with FEV1 and FVC, except X5, which becomes more negative with low spirometric lung function. The same trends were observed for oscillometry as for spirometry, with higher length at birth, higher linear growth between birth and 2 yr, and higher final attained adult height associated with lower values for R5, R5-R20, and Ax, and less negative values for X5. Weight gain after 5 yr changed the direction of the association, becoming statistically significant after 10 yr. This fits well with previous knowledge of low birth weight and low adult spirometry values. To this, we can perhaps now add, increased weight gain in childhood after 5 yr.

There were only small differences between the estimates provided by models with and without adjustment for smoking status. There can be a few explanations for this. The air quality in Delhi is notoriously bad (and has been getting worse in recent decades).²⁴ This baseline exposure might overwhelm other differences between smokers and nonsmokers, especially as only a small proportion of smokers in the cohort were characterised as heavy. There is consistent evidence from other studies that overweight, obesity, and weight gain in adulthood are detrimental to lung function independent of age and smoking status.³

Certain limitations to these data should be considered, attrition being the most important. As described previously,²⁵ municipal redevelopment of the Lajpat Nagar area led to closing unauthorised housing and loss to follow up of displaced members. Mortality (childhood and COVID-related) and outmigration (nationally and internationally), have also contributed to attrition. However, even after substantial attrition, the present cohort in follow-up is similar to the original cohort with respect to several parameters of growth in childhood. Since our analysis was based on internal comparisons within the study sample, bias would only exist if the association between early growth and lung function differed between those who were included in the current study and those who were not. There is, however, a possibility of dilution of true association if those surviving to 50 yr and being willing to participate are healthier than those who dropped out. Second, almost two-thirds of the participants are male. This is because married women shift residence to their spouse’s home, which may be far away. Also, many women are not comfortable sharing the fact that they enrolled in a health study in childhood and often opt out. Thirdly, we do not have data on lung function at age 20 yr, and thus, associations of childhood growth with peak lung function are unavailable. Additionally, the absence of relevant data precluded adjustment for potentially important confounders, including occupational and environmental exposures, early life respiratory history, and comorbidities that could influence adult pulmonary function. Moreover, NDBC is a cohort recruited from an exclusively urban neighbourhood, and the patterns of growth and of lung function in this setting may differ from those born and living in rural areas or in more recent birth cohorts, in which mean birth weights are higher and who are exposed to the higher pollution levels of Delhi from earlier ages. Finally, our data are observational, and the causality of associations between growth and lung function cannot be inferred.

Our study is the first to study the association of linear growth and of relative weight throughout childhood on lung function parameters in middle age. Strengths of our study include the use of prospectively recorded growth data and the use of conditional measures of size, which enable us to eliminate the inherent correlation between repeated weight and height measurements obtained at different ages. However, while conditional relative weight does not differentiate between lean and fat mass, we note that measures of length, which are strongly correlated with lean mass, are included in our models.

These data are potentially useful to paediatricians and public health practitioners in motivating efforts aimed at increasing birth size and preventing excessive weight gain after the age of five years.

शोध-संदेश

शोध से पता चला कि जिन बच्चों का जन्म के समय वजन और लंबाई अच्छी होती है तथा जिनकी बचपन में लंबाई सही तरीके से बढ़ती है, उनके फेफड़े वयस्क होने पर बेहतर तरीके से काम करते हैं। वहीं, पाँच वर्ष की आयु के बाद जरूरत से अधिक वजन बढ़ना फेफड़ों के स्वास्थ्य पर नकारात्मक प्रभाव डाल सकता है।

Author contributions

AB: Project administration, resources, conceptualisation, supervision, manuscript writing, investigation and methodology; ADS: Supervision, manuscript writing; SKB: Conceptualisation, resources; SS: Resources; AA: Resources, conceptualisation, supervision, manuscript writing; HSS: Conceptualisation, supervision, resources, manuscript writing. All authors have read and approved the final printed version of the manuscript.

Financial support and sponsorship

The study received the funding support from the Wellcome Trust/DBT India Alliance Early Career Clinical and Public Health Research Fellowship (IA/CPHE/19/1/504602), which was awarded to Dr. Anubhuti Bansal.

Conflicts of Interest

None.

Use of Artificial Intelligence (AI)-Assisted Technology for manuscript preparation

The authors confirm that there was no use of AI-assisted technology for assisting in the writing of the manuscript and no images were manipulated using AI.