Tobacco Documents Online

2 including nonsmoking mothers, and the text of the abstract is therefore seriously misleading. (iv) Although data were collected on quite a number of potential confounding variables (see Table 1 of the paper), the final estimate was only adjusted for maternal age. This was because variables were only included in the final model "if they changed the measure of association between smoking and SIDS by more than 10%." This, to say the least, seems an unwise procedure. Few variables would be expected on their own to have such a large confounding effect, and it is eminently possible that the joint effect of adjustment for all the confounding variables they studied could have much more effect on the relative risk estimate than the single variable maternal age on its own. (v) As seen in Table 1, there was quite a substantial amount of missing data for some of the lifestyle and sociodemographic factors studied, notably years of education, occupational status during pregnancy and caffeine intake during pregnancy, in each case over 25% of the total population. This would reduce the ability to control adequately for these variables. (vi) No data were available on a considerable number of other variables that have been linked to SIDS. These include sleeping position, heating, recent illnesses, breast or bottle feeding, bed sharing, drug use by the mother and family size. Other studies have found that adjusting simultaneously for a wide range of other factors dramatically reduced the association of SIDS with maternal smoking. Overall, the study adds little to the literature on maternal smoking and SIDS. P N Lee 22.1.2001

1 REVIEW 1141 CONFIDENTIAL Subject ref 9 "A prospective study of smoking during pregnancy and SIDS" K Wisborg et al Archives of Diseases in Childhood (2000), 83, 203-206 In this prospective study in 1989-96 around 25000 pregnant women in Aarhus, Denmark completed questionnaires during pregnancy concerning smoking habits before and during pregnancy and a range of other potential confounding variables. Subsequently information about deaths during the first year of life was obtained, 20 deaths being identified as being from sudden infant death syndrome (SIDS). After adjustment for maternal age, the results reported in Table 2 show that the risk of SIDS was found to be 3.5 times higher (95% CI 1.4-8.7) if the mother smoked during pregnancy. The estimates were similar, 3.4 and 3.7, whether the mother smoked 1-9 or 10+ cigarettes per day. The authors note that "Our study adds to earlier evidence for an association between smoking during pregnancy and SIDS. The strengths of the study are the possibility to adjust for a number of potential confounders and the fact that information about smoking habits during pregnancy was prospectively collected." In commenting on these findings a number of points should be noted: (i) The total number of SIDS cases is small, only 20. (ii) The results are inconsistently reported - in the abstract it is stated that the unadjusted relative risk was 3.5 (1.4-8.7) and that adjustment for maternal age reduced the estimate to 3.0 (1.2-7.3). Table 2, however, states that the relative risk adjusted for maternal age was 3.5 (1.4-8.7). (iii) The abstract also states that the risk of SIDS increases with the number of cigarettes smoked per day (p<0.05). However, the actual data show that the estimated risk (adjusted or unadjusted) for smoking of 1-9 and 10+ cigarettes per day is much the same, and clearly not statistically significantly different. The p<0.05 relates to a trend analysis