Introduction Observational epidemiological studies and a systematic review have consistently shown the association between maternal exposure to biomass smoke and reduced birth weight.

Methods We analysed 47 139 most recent singleton births of 2005–2006 India DHS. Information on birth weight was obtained from health card or mothers' recall. “size of baby” at birth in a fivefold classification was re-coded as “smaller than average” and “average or larger” to represent low birth weight and normal birth weight respectively. The fuel type was classified as high pollution, and low pollution. Univariate and multivariate logistic regression models were developed using SURVEYLOGISTIC procedure in SAS system in which child factors, maternal factors and demographic factors were added step-by-step to the main exposure variable.

Results Babies not weighed at birth in households using high and low pollution fuels was 71.4% and 26.1% respectively. Missing information on size at birth was only 6.4%. The results for association between high pollution fuels use and low birth weight were as follows: univariate OR was 1.41 (95% CI 1.27 to 1.55). Adjusted ORs in models 1, 2, 3 were 1.41 (1.29, 1.57), 1.21 (1.06, 1.32), 1.07 (0.94, 1.22) respectively. Gender, birth order, mother's BMI, haemoglobin level and education were significant throughout. In model 3, there was no association between birth weight and fuel type but wealth index and religion were significant.

Conclusion Our results are consistent with previous observational studies. The effect size of association between size at birth and exposure to biomass smoke during pregnancy was small and was masked by wealth index and religion.