TY - JOUR T1 - Prevalence of and risk factors for sudden infant death syndrome in Canada JF - Canadian Medical Association Journal JO - CMAJ SP - 629 LP - 635 VL - 149 IS - 5 AU - W. J. Millar AU - G. B. Hill Y1 - 1993/09/01 UR - http://www.cmaj.ca/content/149/5/629.abstract N2 - OBJECTIVE: To analyse and describe the prevalence of sudden infant death syndrome (SIDS) in the Canadian population in relation to the distribution of known risk factors for the syndrome. To explain the observed interprovincial variation in SIDS rates. DESIGN: Retrospective population-based case-control study. SETTING: All the provinces and territories of Canada except Quebec. SUBJECTS: The birth and infant death records of singleton births for 1986-88 were linked. The linkage was successful in matching 904 (86%) of the 1053 deaths attributed to SIDS to the birth file for the infant. For each SIDS case three control babies who survived infancy were chosen at random, matched by province of birth. MAIN OUTCOME MEASURES: Infant death classified as a "sudden infant death." Independent variables included infant's sex, birth weight and gestational age, being small for gestational age, mother's age, marital status and parity, and father's age. RESULTS: The risk of SIDS was greater for boys (odds ratio [OR] 1.47, 95% confidence limits [CLs] 1.26 and 1.70) than for girls and was greater for the infants of unmarried women (OR 3.48, 95% CLs 2.94 and 4.11) than for those of married women. The risk of SIDS was inversely related to birth weight (p < 0.001), duration of pregnancy (p < 0.001) and mother's age (up to age 35) (p < 0.001) and was directly related to parity (up to four) (p < 0.001). The available information on birth and death registrations enabled about 30% of deaths from SIDS to be predicted. CONCLUSIONS: Logistic regression equations based on the risk factors available in vital statistics data have low power to predict provincial differences in rates of SIDS. Consequently, there may be additional factors that may explain provincial variation. There is a need for a well-designed case-control study that examines more variables than are available through vital registration systems. ER -