Mattresses, microenvironments, and multivariate analyses

BMJ 2002; 325 doi: http://dx.doi.org/10.1136/bmj.325.7371.981 (Published 02 November 2002) Cite this as: BMJ 2002;325:981

No reason to change current practices for reducing risk of sudden infant death

  1. Peter J Fleming, professor of infant health and developmental physiology,
  2. Peter S Blair, medical statistician
  1. Division of Child Health, University of Bristol, Bristol BS1 2LY
  2. Department of Paediatrics, University of Auckland, Private Bag 92019, Auckland, New Zealand

    Papers p 1007

    Despite the success of “Back to Sleep” campaigns in many countries, sudden infant death syndrome remains responsible for the largest group of deaths in infants between one month and one year of age.1 The importance of sleeping in the prone position as a contributory factor has led to studies of the pathophysiological effects of the prone position on the infant and to studies of microenvironmental factors that might contribute to this risk. 2 3 The carefully conducted study by Tappin and colleagues in this issue (p 1007) is set in Scotland and emphasises the potential importance of the infant's microenvironment during sleep as a contributory factor to the risk of sudden infant death syndrome, but as emphasised by the authors, caution must be exercised in the interpretation of these results.4

    Tappin and colleagues have shown an apparently increased risk of sudden infant death syndrome for infants sleeping on a mattress previously used by another infant (54% cases, 28% controls), confirming the observation by the same group in an earlier study.5 The previous Scottish study was criticised because infants sharing beds with adults were included with infants sleeping on mattresses used by another infant.6 In the present study this criticism has been addressed. The earlier study was, however, also criticised for a lack of adjustment of potential confounders related to the use of mattresses, in particular the socioeconomic status and the number of siblings in each family. The only other study that has addressed this question, the Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI SUDI), conducted in England around the same time, reported similar differences in the reuse of mattresses, but the difference was less marked (70% cases, 53% controls) and not statistically significant after adjusting for either socioeconomic status or size of family, the risk of previously used mattresses approaching unity when adjusted for both factors. 1 7 8

    The study by Tappin and colleagues has attempted to adjust for socioeconomic status by using the postcode dependent tool DEPCAT (deprivation category). This is a relatively insensitive tool,9 and, given the low response rate (71%) among “control” families, some of the difference in the use of mattresses may be explained by the higher prevalence of deprivation among “sudden infant death syndrome” families. A careful analysis of mattress reuse, stratified by families of different sizes, would clarify the possible importance of parity, which is obscured by the multivariate model that has been used.

    Re-examination of the data from the Confidential Enquiry into Stillbirths and Deaths in Infancy (PS Blair, unpublished data) shows that use of a previously used mattress rose with increasing numbers of siblings—in families with four or more children. Seventy three per cent of both cases of sudden infant death syndrome and control infants slept on previously used mattresses. In a conditional logistic regression model (taking account of the age matching), high parity (more than two children in the family) remained significantly associated with an increased risk of sudden infant death syndrome (odds ratio 2.67, 95% confidence interval 1.81 to 3.92, P<0.0001), but the risk associated with reused mattresses became non-significant (1.05, 0.72 to 1.51), P=0.80) among smaller families. Careful subgroup analysis may help explain the strong univariate effect of parity in the data by Tappin and colleagues but subsequent non-significance in the multivariate model.

    How might a previously used mattress contribute to the risk of sudden infant death syndrome? Several studies and two multidisciplinary expert groups have not found any evidence to support the hypothesis that toxic gases were produced by fungal activity on the polyvinyl chloride coverings of mattresses—the use of mattresses with integral polyvinyl chloride coverings was shown to be associated with a lower risk of sudden infant death. 1 4 7 8 10 Tappin and colleagues also explicitly reject this hypothesis.

    One postulated effect of two or more infants successively using the same mattress, particularly without an impermeable cover, would be that saliva, urine, or milk soaking into the mattress would produce a change in the microbial flora, with increased growth of potentially toxigenic bacteria (for example, Staphylococcus aureus) that might in turn contribute to illness or death of the infant, particularly at the time of an acute viral infection.1113 Thus reuse of mattresses would lead to the risk occurring at a younger age for the second or subsequent infant using the mattress. Reanalysis of the data from the Confidential Enquiry into Stillbirths and Deaths in Infancy (PS Blair, unpublished data) shows, however, that infants who died with the sudden infant death syndrome who died on previously used mattresses were on average two weeks older than infants who died of the sudden infant death syndrome on new mattresses (median 100 days v 85 days, respectively). Unfortunately information on the ages of the infants is not given by Tappin.

    The similarities in observations, but marked differences in the results of multivariate analyses between the Scottish studies and the English Confidential Enquiry into Stillbirths and Deaths in Infancy study may represent the effects of real (but possibly unmeasured) factors, differences in child care practice north and south of the border, or a limitation of the complex multivariate analysis. Certainly more studies are needed to compare the precise microenvironment of sleep for victims of the sudden infant death syndrome and healthy infants from similar socioeconomic backgrounds.

    In the meantime, the advice to parents should remain as internationally agreed:

    • Place your baby to sleep on his or her back

    • Create a smoke free environment for your baby during pregnancy and after birth.


    • Competing interests Peter Fleming was a member of the expert group set up to advise the chief medical officer in 1994 on cot death theories—with particular reference to “toxic gas” hypotheses implicating mattresses in cot deaths (the Limerick Committee). Both Peter Fleming and Peter Blair were grant holders and principal investigators in the CESDI SUDI study on mattresses and SIDS.