Sudden infant death syndrome: after the “back to sleep” campaign

Further declines may come from reducing maternal smoking

The incidence of sudden infant death syndrome in England and Wales fell by two thirds between 1989 and 1993. The decline in deaths followed the “back to sleep” health education campaign (started in 1991), which advised parents to place babies on their back or side to sleep, to avoid overheating and smoky environments, and to contact a doctor if their baby was unwell.1 Similar campaigns were conducted in Australia, New Zealand, and several other countries, with a decrease in the incidence of sudden infant death syndrome of 50% or more.2 In Tasmania a within-cohort analysis of more than 30 factors that might have contributed to the decline in incidence found that 70% of the decline could be attributed to the reduction in the proportion of infants who usually slept prone.3 Much of the research effort is now focused on understanding the mechanisms through which the prone position exerts its effect and on identifying the causal pathway for sudden infant death in infants who do not sleep in the prone position.

Two papers in this week's BMJ present the findings from a two year case-control study conducted in England after the “back to sleep” campaign was launched.4 5 The study used a similar design to that used in the 1990 Avon and Somerset study, which identified the prone sleeping position and overheating as risk factors for sudden infant death.6 Its strengths include a standardised and detailed system for case ascertainment over a large population, a short interval between death and interview, high response rates, and an emphasis on data collection for potentially amenable risk factors. With the rapid decline in the incidence of sudden infant death syndrome in most countries, this will be one of the largest studies of sudden infant death in a post-intervention population. However, recall bias must be considered in a case-control study on a population that was well informed about the risk factors for sudden infant death syndrome.

The first paper by Fleming et al (p 191) makes the important observation that the risk of sudden infant death syndrome was higher in infants placed to sleep on their side than those placed on their back.4 The highest risk was for infants reportedly placed on their side and found prone. The New Zealand case-control study also found the risk to be higher in babies placed on their side rather than on their back.7 While neither study was able to demonstrate conclusively whether this was due to differences in reporting or in infant care practices, the likelihood that infants sleeping on their side are at greater risk is supported by evidence that the side position is unstable, with a proportion of infants changing to the prone position during sleep.8 As the effect of sleeping position on the risk of sudden infant death syndrome depends on features of the infant and the infant's environment such as swaddling,9 Fleming et al could usefully analyse their data on sleeping position with regard to effect modification as well as confounding.

Bed sharing has been a controversial area, particularly since the New Zealand case-control study found that bed sharing was a risk factor among infants whose mothers smoked.10 Infants who shared their parents' bed all night during the last sleep were four times more likely to suffer sudden infant death after adjusting for confounding factors, and the risk was potentiated if the mother smoked.4 Even among the infants of non-smoking mothers, bed sharing doubled the risk of sudden infant death syndrome. This increase in risk was not significant, but this may have been due to a type II statistical error in view of the small numbers in the subgroup.

With only a small proportion of infants sleeping prone after the “back to sleep” campaign, Blair et al in the second paper (p 195) found parental smoking to be the major contributing factor to sudden infant death syndrome, accounting for 62% of cases.5 This estimate is based on an unusual composite measure which records a positive exposure if the mother smoked during pregnancy or the father smokes in the household. Previous work supports the view that prenatal and postnatal smoking by the mother and probably postnatal smoking by others increase the risk of sudden infant death syndrome.11 12 13 The findings of relative risks greater than three and a dose-response relation in these studies support the inference that this is causal. However, lingering doubts remain that the association may be due, in part, to the association of parental smoking with other infant care practices, which might vary by socioeconomic status.

Blair et al have not helped resolve to this issue, and the separate contributions to risk of prenatal and postnatal exposure to tobacco smoke from various sources are difficult to ascertain from their analysis. Importantly, the paper does not report estimates of risk due to maternal postnatal smoking after adjustment for maternal smoking during pregnancy. The dose-response patterns for postnatal exposure (table 4) are reported without adjustment for socioeconomic status, and the multivariate models used are not described in detail. Further analysis could provide an estimate of the population risk attributable to smoking, which could be more readily evaluated and may contribute more to the already strong case for avoiding cigarette smoking during pregnancy or in the vicinity of babies.