Seaman and colleagues [1] recently found that increasing age, long duration of illness, decreasing hemoglobin level, and decreasing body mass index were independently associated with death from (or with) visceral leishmaniasis in adults in Sudan. The large sample size in their study makes it unlikely that random error had any effect on the risk ratio estimates, but measurement errors resulting from the limited resources that were available to the investigators may have affected the validity of the results. Age was particularly likely to be systematically overestimated in patients whose disease was more severe and had lasted longer (and who therefore were at higher risk for death) because patients with severe malnutrition tend to look older than they really are. If this differential misclassification bias occurred, the effect of age as a predictor of death was overestimated.

In Brazil, I also found that patients with visceral leishmaniasis, even those who are severely debilitated, respond to treatment with a pentavalent antimonial, with the exception of patients with severe neutropenia who develop bacterial infections. Visceral leishmaniasis leads to pancytopenia; although anemia can be interpreted as a risk factor by itself, perhaps it is better seen as a surrogate for neutropenia, which was not measured in the study by Seaman and colleagues [1]. The immediate cause of death for many patients with visceral leishmaniasis is pneumonia or another bacterial infection [2]; such infections are better explained by neutropenia than by anemia.

The distinction between neutropenia and anemia as risk factors for death in patients with visceral leishmaniasis is important. Measures to correct anemia, such as blood transfusions, are expensive and logistically difficult to implement in such conditions as those prevailing in southern Sudan. These measures would probably not have a major role in the reduction of visceral leishmaniasis-associated mortality rates.