Where science meets policy: comparing longitudinal and cross-sectional designs to address diarrhoeal disease burden in the developing world

Background Longitudinal studies are considered preferable to cross-sectional studies for informing public health policy. However, when resources are limited, the trade-off between an accurate cross-section of the population and an understanding of the temporal variation should be optimized. When risk factors vary more across space at a fixed moment in time than at a fixed location across time, cross-sectional studies will tend to give more precise estimates of risk factor effects and thus may be a better source of data for policy judgments. Methods We conducted a diarrhoeal disease surveillance of 5616 individuals within 19 Ecuadorian villages. This data set was used to mimic cross-sectional and longitudinal studies by restricting focus to a single week and a single village, respectively. We compared the variability in risk factor effect estimates produced from each type of study. Results For household risk factors, the effect estimates produced by the longitudinal studies were more variable than their cross-sectional counterparts, which can be explained by greater spatial than temporal variability in the risk factor distribution. For example, the effect estimate of improved sanitation was almost twice as variable in longitudinal studies. Conclusions In our study, cross-sectional designs yielded more consistent evaluations of diarrhoea disease risk factors when those factors varied more between villages than over time. Cross-sectional studies can provide information that is representative across large geographic regions and therefore can provide insight for local, regional and national policy decisions. The value of the cross-sectional study should be reconsidered in the public health community.