Perinatal mortality: standardizing for birthweight is biased.

Standardization is often used to compare perinatal mortality between populations with different birthweight distributions. Unfortunately, the relationship between perinatal mortality and birthweight seldom satisfies one prerequisite for standardization. Furthermore, standardizing for birthweight is biased against populations with heavier birthweights. For example, the standardized perinatal mortality rate of the heavier of two populations is biased upwards by 15 to 25% when the difference in mean birthweight between those populations is 150 g. As a result, the use of standardization may confuse the interpretation of differences in perinatal mortality.