An article was published in the Journal of the American Academy of Pediatrics in May of this year that addressed the potential for a relationship between swaddling and sudden infant death syndrome. The article caused a bit of a stir and, unfortunately, the data was presented in some scary ways in the press. So I thought it worthwhile to review what the study actually tells us and what it does not.
The design of this study, a meta-analysis, is meant to pool data from several smaller studies in order collect enough power to look at a given outcome. In this case, the authors found four collections of data looking at sudden infant death syndrome that included information about whether the infants were or were not swaddled, the ages of the infants, and their sleeping positions. The collective data spanned the years between 1988 and 2006 and included subjects in the US, the UK and Australia.
What the authors learned about their data during the analysis is very important. They found that there was significant heterogeneity among the studies. This means that the original study protocols were quite different from one another which complicates the idea of pooling together the data and treating the group of individual studies like one big study. To account for this complication, the researchers needed to apply a statistical calculation to make sure that they were not drawing conclusions that could be attributed to chance.
So, when the researchers compared the swaddled to the not swaddled group BEFORE accounting for potentially problematic differences among the studies, they found that swaddled babies were 1.53 more likely than babies who were not swaddled to die of SIDS. However, when they did account for the differences in the original studies, this increase in risk disappeared. Of note, the study authors call this set of data “of borderline significance” but it is actually statistically insignificant. They then drop one of the studies from the calculation and recalculate. The risk continues to be, statistically speaking, insignificant. However, they apply a substandard definition of statistical significance here again.
Where the data does seem to hold up is in the authors’ analysis of risks of swaddling older babies and risks associated with swaddled babies sleeping in any position but on their backs. The data revealed a trend toward an increased risk of SIDS as age increased with the highest and most statistically significant risk in infants older than six months. Babies who were found swaddled and on their bellies were rare in the study, but were 19 times more likely than unswaddled babies to die of SIDS. In a subset of the data that included the position in which the babies were last placed as well as found, it became evident that the practice of placing swaddled infants on their bellies to sleep was very rare and so it seemed that the babies who were found on their bellies had rolled over.
In their discussion, the authors offer the insight that in the Netherlands, a country with low rates of SIDS, swaddling is encouraged but parents are advised not to initiate swaddling after the fourth month and to discontinue swaddling any infant who is signaling that they are interested in rolling over and all infants at six months when they are very likely to be able to roll over even if it has not yet been witnessed. These seem to be very reasonable recommendations given the information that is available to us including how useful swaddling can be in soothing young babies. Perhaps the most useful takeaway from this study is the need for improvements in messaging to parents considering swaddling from the medical community here in the US.
The article can be found in its entirety on the AAP website here.