Birth weight-specific rates as a bias in the effects of smoking and other perinatal hazards.

Perinatal mortality was examined in the Ontario Perinatal Mortality Study (1960 to 1961) subset of 31,788 births and perinatal deaths among English-speaking Canadian-born women. Overall perinatal mortality rose from 22.0 to 28.2 per 1000 if the mother smoked, a 27% increase. When perinatal mortality was computed in birth weight-specific groups (less than 2500 g and 2500 g or more), the rates for smokers were lower than those for nonsmokers in both subgroups. These results occurred because maternal smoking had an even stronger effect on birth weight than on perinatal mortality, increasing the proportion of births of infants weighing less than 2500 g by 106%. These data confirm the hazard of maternal smoking for the unborn, and pinpoint how some studies that have seemingly inconsistent results have erred in analysis.