Perinatal mortality rates: adjusting for risk factor profile is essential.

Perinatal mortality has long been used as a comparative measure of health and health care across regions, countries and over time. Recently, the validity of the measure has been questioned. Using data from a population-based survey of late fetal losses, stillbirths and infant deaths, the Northern Perinatal Mortality Survey, we demonstrate the potential for inaccuracy of crude measures of perinatal mortality. Such measures are generally not adjusted for characteristics of the population (e.g. birthweight, maternal age, plurality, gender) which are known to affect risk of adverse pregnancy outcome when comparing temporal or geographical trends. We also show the effect of standardising for these factors on the most frequent causes of perinatal death. We recommend the construction of a 'standard birth population' for calculating standardised perinatal mortality rates that would improve direct comparisons between populations.