Objective To assess the association between childbearing age and risk of offspring mortality in three birth cohort studies.

Setting Pelotas, Southern Brazil.

Methods All hospital births which occurred in 1982 (n=6011), 1993 (n=5304) and 2004 (n=4287) were identified and these infants were followed-up. Deaths were monitored through vital registration, visits to hospitals and cemeteries. The analyses were restricted to women younger than 30 years who delivered singletons (72%, 70% and 67% of the original cohorts, respectively). Maternal age was categorised into three groups (<16, 16–19, and 20–29 years). Additional analyses compared mothers aged 12–19 and 20–29 years. The outcome variables included fetal, perinatal, neonatal, post-neonatal and infant mortality. Potential confounders were family income, maternal education, maternal colour/race, marital status, parity, and pre-pregnancy body mass index. Crude and adjusted ORs were estimated with logistic regression models.

Results There were no interactions between maternal age and cohort year. After adjustment for confounding, the pooled ORs for infant death were 1.08 (95% CI 0.42 to 2.78) for mothers younger than 16 years, 1.48 (1.03 to 2.12) for 16–19-year-olds, and 1.45 (1.01 to 2.07) for those aged 12–19, compared to a 20–29-year-old mothers. The excess risk was due to post-neonatal deaths.

Conclusion The slightly increased risk of post-neonatal mortality among the offspring of adolescent mothers suggests that social and environmental factors may be more important than maternal biologic immaturity. The possibility of residual confounding cannot be ruled out.