1. Associations between lower birthweight and higher blood pressure, increased risk of type 2 diabetes and coronary heart disease (CHD) have been observed in a number of different populations worldwide.
2. The reason for this is still debated. Some believe that the observed associations can be explained on the basis of differences in postnatal growth, socioeconomic confounding or genetic factors. Two published studies of birthweight and CHD, with information on later size, suggest that both gestational and postnatal exposures are important. Associations between birthweight and blood pressure, seen in cohorts of twins treated as individuals, have generally remained when data are analysed within twin pairs. Furthermore, similar associations are seen in studies of animals with relative genetic homogeneity kept in standard conditions. These findings suggest that neither socioeconomic nor genetic factors are wholly responsible for the observed associations.
3. If then, there is an underlying causal association, two issues are of fundamental importance. First, is fetal growth (for which birthweight is a summary measure) involved in the causal pathway or is the causal factor a fetal exposure independently associated with fetal growth and increased risk of adult cardiovascular disease? The answer is important in terms of our understanding, the potential for intervention and estimation of the public health implications. Second, are the classic risk factors for CHD in the causal chain between fetal exposures or growth and adult CHD? Most prospective studies measure these factors, but their role as intermediates is unproven.
4. Intervention studies are the best way to test causal hypotheses, but our level of understanding is insufficient to justify such studies in humans, so we rely on animal studies to formally test causal hypotheses. In the present paper, we discuss design and statistical issues in relation to animal studies. The challenge in this field is to devise ways to identify and test potential causal hypotheses in humans.