Malnutrition zinc deficiency and malaria in Africa [letter]

Sunil Sazawal and colleagues in their large trial in Pemba Zanzibar found that zinc supplementation led to a less-than-expected reduction of mortality (7%; 95% CI -6 to 19). The study was designed with a 90% probability to detect a 20% reduction in mortality. It is not helpful that Sazawal and colleagues take this power into consideration when interpreting the results. First the results indicate that power calculations were based on assumptions that were not met during trial implementation. A study with an expected upper limit equal to a particular value will have only 50% probability of yielding an upper limit below that value. Thus under the conditions prevailing in the Pemba trial the probability of detecting a mortality reduction of 19% is only 50%. (excerpt)