Is prevalence of diarrhea a better predictor of subsequent mortality and weight gain than diarrhea incidence?

A number of different outcome measures have been proposed for use in prospective studies of morbidity associated with childhood diarrhea. These include the number of episodes experienced by each child over a defined period (a measure of incidence) and the number of days of diarrhea divided by the total number of days of observation for each child (a measure denoted "longitudinal prevalence"). The authors examined data from Ghana to determine which of these measures is more strongly associated with weight gain over a 4-month period and subsequent mortality. Both diarrhea incidence and longitudinal prevalence were associated with weight gain in children aged 6-23 months, but a statistically stronger association was observed with longitudinal prevalence (likelihood ratio statistic 28.95 on 1 degree of freedom against 19.70 for incidence). Neither measure was associated with weight gain in younger or older children. Longitudinal prevalence, but not incidence, was strongly associated with subsequent mortality (p = 0.002 for longitudinal prevalence; p = 0.557 for incidence). Although many epidemiologic studies of diarrhea focus on incidence, these data suggest that longitudinal prevalence is more strongly predictive of long-term health outcome. The authors conclude that longitudinal prevalence merits greater attention as a measure of outcome in diarrhea studies.