Mother-infant co-sleeping or bed sharing is discouraged by health organisations due to evidence that it is associated with unexplained sudden infant death. On the other hand, there is evidence that it should theoretically be beneficial for infants. One line of this evidence concerns breathing regulation, which at night is influenced by the rocking movement of the mother's chest as she breathes. Here, the hypothesis that mother-infant co-sleeping will be associated with a lower probability of infant breathing distress is tested in the UK Millennium Cohort Study (n = 18,552 infants). Maternal, infant, family, and socio-economic covariates were included in logistic regression analysis, and in a machine learning algorithm (Random Forest) to make full use of the number of variables available in the birth cohort study data. Results from logistic regression analysis showed that co-sleeping was associated with a reduced risk of breathing difficulties (OR = 0.69, p = 0.027). The Random Forest algorithm placed high importance on socio-economic aspects of infant environment, and indicated that a number of maternal, child, and environmental variables predicted breathing distress. Co-sleeping by itself was not high in the Random Forest variable importance ranking. Together, the results suggest that co-sleeping may be associated with a modest reduction in risk of infant breathing difficulties.
Keywords: Neonatal health; SIDS; SUDI; cohort studies; infant sleep; machine learning; safe sleeping.