Problems of studying infections in the compromised host.

Although large numbers of patients have been enrolled in studies of the efficacy and safety of new antimicrobial agents in immunocompromised hosts, many variables other than number affect the outcome and interpretation of therapeutic and prophylactic trials. Most therapeutic studies, even of a comparative nature, show relatively few differences in the outcome of treatment of staphylococcal and gram-negative bacillary infections. A major problem has been the accrual of large numbers of cases of proven infection. In the future, clinical studies should concentrate on proven infections due to the more problematic organisms, such as Pseudomonas, Serratia, and Enterobacter species, which are often associated with clinical failure and emergence of high-level resistance. Comparisons of cost-effectiveness of regimens and duration of survival of the cohorts treated in randomized studies should be made. The desirability of large or multicenter therapeutic studies seems obvious, but there is room for improvement in the design and execution of such studies.