Sostrup statement on low birthweight

Sirs—Birthweight data are available in most countries and have been used extensively in perinatal monitoring and in research. The per cent of babies born low weight (<2500 g) has been widely accepted as a population-level marker for perinatal morbidity and mortality. This has been justified by the correlation between low birthweight (LBW) and high perinatal mortality and morbidity—correlations that extend even into adult life. However, the relation of birthweight to these health endpoints is more complex (and perhaps less causal) than simple correlations imply. We offer the following statement on birthweight as a consensus developed at a retreat held 8–9 June 2002 at Sostrup Slot, and presented at the Second Nordic Epidemiologic Conference in Aarhus, Denmark. It is our hope that this statement might provoke constructive discussion of the limits of LBW as an epidemiological measure, and an exploration of possible alternatives. The per cent LBW is a simple and time-honoured tool for monitoring aspects of perinatal health around the globe. However, epidemiological research has shown that per cent LBW can be misleading. For example, the per cent LBW can be affected by a change of mean birthweight. Changes in mean weight can reflect physiological changes that do not necessarily affect health. For purposes of surveillance, per cent LBW should be de-emphasized in favour of other population measures. In principle, these alternatives include rates of fetal and infant mortality, and direct indicators of infant health and development. Preterm delivery (even a rough surrogate such as the residual distribution based on the complete birthweight distribution) is a better health indicator than LBW. We recognize that any change in the monitoring of health in less-developed countries requires redirection of scarce resources. In practice, therefore, research is needed to determine which of the possible measures of perinatal health are the most useful and feasible. LBW babies comprise a mix of preterm delivery, decreased fetal growth, and genetically small body size (a physiologically normal feature). The first of these conditions (preterm delivery) is a measure of poor infant health. Babies delivered preterm are typically at a disadvantage either because of early delivery itself or because of the forces causing the baby to be born early. However, effects on fetal growth do not necessarily affect perinatal health. As a research tool, per cent LBW should be replaced by measures that distinguish between preterm delivery and decreased fetal growth. Such adjustments, including multivariate adjustments for birthweight, may lead to incorrect results. The only sure strategies for fully assessing perinatal health are those that measure a range of endpoints related to health and development.