1 Department of Maternal and Child Health, University of Alabama at Birmingham, USA.
2 Department of Social Medicine, University of Copenhagen, Denmark.
3 Department of Biostatistics, University of Copenhagen, Denmark.
4 Department of Organismic & Evolutionary Biology, Harvard University, Boston, Massachusetts, USA.
5 Department of Pediatrics, Aarhus University, Denmark.
6 Department of Epidemiology & Preventive Medicine, University of California at Davis, USA.
7 Section for Medical Statistics, University of Bergen, Norway.
8 Danish Epidemiology Science Centre, Aarhus, Denmark.
9 Section for Medical Statistics, University of Bergen, Norway.
10 Epidemiology Branch, National Institute of Environmental, Health Sciences, Durham, North Carolina, USA.
Jørn Olsen, The Danish Epidemiology Science Centre, Aarhus University, Vennelyst Boulevard, 6 DK-8000 Aarhus C Denmark. E-mail: jo{at}soci.au.dk
SirsBirthweight 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 morbiditycorrelations 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 89 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.
Per cent low birthweight is a poor index of a populations perinatal health
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.
In population studies, per cent low birthweight is a poor research tool for detecting factors or conditions that damage perinatal health
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.
Adjustment for absolute birthweight is rarely justifiable in looking for the effects of specific exposures on infant or perinatal mortality
Such adjustments, including multivariate adjustments for birthweight, may lead to incorrect results.
Some exposures or conditions may compromise fetuses without causing preterm delivery or impairing fetal growth
The only sure strategies for fully assessing perinatal health are those that measure a range of endpoints related to health and development.