Income level or bed sharing would confound any effect of previous use of mattress

Editor—Hazel Brooke and colleagues have drawn attention to the possible risk of the sudden infant death syndrome associated with “old” mattresses.1 In their study they did not ask about mattress age or the number of previous users, only whether the mattress had been used by another infant or adult. The use of the word “old” is therefore misleading. By this definition, all adult mattresses would be considered old, as would recently bought mattresses used by one previous infant. In our study, mattress age as a continuous variable was not significant, whether mattresses totally covered with polyvinyl chloride were included or not.2 3 When mattress age was taken as a dichotomous variable, a small but significant proportion of index mattresses were ≥4 years old; this factor became non-significant, however, when parity or family income was added in a bivariate analysis. We also asked about previous use of infant mattresses. This was not a significant factor in the univariate analysis, but, when we included infants who routinely bed shared, the risk associated with previous use became significant; the significance increased if we included instead infants who bed shared during their last sleep. Thus, distinguishing between infant and parental mattresses and what practice was adopted on the last night seems important.

Given this, one would expect that, in Brooke and colleagues' multivariate model, the factors representing income or bed sharing would confound any effect of previous use of the mattress. However, income was not measured directly and the study focused on routine bed sharing practice rather than bed sharing around the time of death. Both our study and a similar large study in New Zealand4 showed considerable differences in the risk associated with routine practices and the practice around death. The risk associated with routine bed sharing rises twofold when …