|Home | About | Journals | Submit | Contact Us | Français|
In a recent review in the Archives of Disease in Childhood, Fleming et al sound a note of caution concerning recent recommendations by the American Academy of Pediatrics (AAP) and others on the avoidance of bed‐sharing and the use of dummies (pacifiers) in reducing the risk of sudden infant death syndrome (SIDS).1 It is certainly appropriate that changes in such advice should only be made after proper consideration of the evidence and, as far as possible, with due account taken of any possible unintended adverse consequences. However, there are also risks in being insufficiently critical of the scientific basis of the objections to, as well as to the arguments in favour of, revision of advice.
The defence usually mounted in favour of bed‐sharing rests mainly on three arguments, which may be called the anthropological argument, promotion of breast feeding and the claimed benefits to mother and child of the increased close body contact that bed‐sharing entails. The anthropological argument can be summarised thus: bed‐sharing is widely practised among many cultures and communities, and has “evolved” over a long period of time and therefore must have advantages for infant, and therefore species, survival. It is at least questionable whether practices that were and are widespread in pre‐industrial, pre‐literate and tribal societies should be accepted uncritically as the best guidelines for child care in modern, developed countries.
The breast‐feeding argument rests on the assumption that the association found in a number of studies between bed‐sharing and breast feeding implies causality in the sense that bed‐sharing promotes breast feeding. This has not been clearly demonstrated in any study designed to answer this particular question, and it is at least possible that the causality is in the other direction: when mothers stop breast feeding, for whatever reason, they are less likely to take the infant into the parental bed. One recent study from New Jersey, USA, found a trend toward less breast feeding in bed‐sharing cases than in non‐bed‐sharing controls (22% vs 35%).2 Finally, the benefits for infant sleep that are claimed to be the consequence of bed‐sharing have been seriously challenged by one recent, careful study.3 In addition to those cited by Fleming et al, one published4 and one unpublished study have found bed‐sharing to be a risk factor for SIDS. The unpublished report is an update on the work of McGarvey et al, presented at the International SIDS Meeting in Yokohama, 2006 (in the most recent published report from this group5 the trend towards bed‐sharing as a risk factor when the mother was a non‐smoker was not statistically significant, but only just). Interpretation of the observation by Chen and Rogan6 (cited by Fleming et al) that postneonatal deaths from all causes were 26% higher for bottle fed than breast fed infants is complicated by the fact that when the deaths were divided into four subgroups (infections, injuries, SIDS and others) the effect was only significant for those who died of injuries. This finding is not easily explained, unless it is due to confounding social factors associated with breast feeding that were not included in the analysis.
As for dummy use, it may be the case, as Fleming et al argue, that the evidence base for it being a protective factor does not meet all the criteria for the evidence based medicine of Sackett et al,7 but one would have to apply this criticism to most and probably all of the current recommendations. In particular, the fact that the mechanism of the putative protective effect of dummy use is unknown applies equally to the recommendation to put infants to sleep on their backs, which few advisors would now disagree with. Since the publication of the AAP recommendations, there has been at least one additional study8 showing a statistically significant and extremely strong protective effect of dummy use, not included in the original meta‐analysis on which the AAP recommendations were based. Addition of these data to the model used by Hauck et al9 would increase the magnitude of the apparent protective effect over the odds ratio of 0.39 (95% CI 0.31 to 0.50) calculated by multivariate analysis in the meta‐analysis. As with bed‐sharing, any negative association between dummy use and breast feeding may well be due to multiple confounders and not cause and effect. In another recent publication10 (which shares one author with the current Fleming et al review), the recommendation is more cautious and probably more balanced: “It seems appropriate to stop discouraging the use of pacifiers. Whether it is appropriate to recommend pacifier use in infants is open to debate”.
Competing interests: None declared.