Objective: To determine which risk factors are specific for the sudden infant death syndrome (SIDS) rather than characteristic of postneonatal deaths in general.
Study population: The live births and infant death cohorts of the 1988 National Maternal and Infant Health Survey.
Methods: Information on live births, deaths from SIDS, and postneonatal deaths from other causes was abstracted from the National Maternal and Infant Health Survey. To account for oversampling of certain populations, the data were weighted to reflect national counts. Risk factors were defined as black race, birth weight less than 1500 gm, birth weight less than 2500 gm, gestational age at birth less than 37 weeks, 5-minute Apgar score less than 7, male gender, more than two previous pregnancies, maternal age less than 20 years, maternal education level less than 12 years, multiple births, and maternal smoking during pregnancy. Odds ratios (ORs) and 95% confidence intervals (Cls) were calculated to compare the SIDS with the live births cohort, infants who died of other causes with the live births cohort, and SIDS with non-SIDS deaths. The population-attributable risk percentage was computed for risk factors independently associated with SIDS when compared with other postneonatal deaths.
Results: For all characteristics other than a 5-minute Apgar score less than 7, the ORs comparing infants who died of SIDS with the live births cohort were significantly greater than 1.0. Similarly, ORs comparing infants who died of other causes with the live births cohort were also greater than 1.0, except for male gender. When the two infant death cohorts were compared, only maternal smoking during pregnancy and low maternal education level were significantly more common among the SIDS group. After we controlled for cigarette smoking, the adjusted OR for low maternal education level was not significantly greater than 1.0. However, even after control for low maternal education level, prenatal exposure to tobacco was significantly more common among the SIDS group than in infants dying of other causes (OR = 1.97; 95% Cl, 1.59 to 2.45). On the basis of an adjusted OR of 2.92 when the SIDS group was compared with the live births cohort, the population-attributable risk percentage for maternal smoking as a risk factor for SIDS was 30%.
Conclusion: Among characteristics generally thought to be risk factors, only maternal smoking during pregnancy was independently associated with SIDS. Data from this nationally representative sample indicate that if women refrained from smoking while pregnant, up to 30% of SIDS might be prevented.