Spectrum of microbial etiology of community-acquired pneumonia in hospitalized patients: implications for selection of the population for enrollment in clinical trials.

The title of this article implies that knowledge of the etiological pathogen may be useful in selection of patients for clinical trials for community-acquired pneumonia (CAP). However, this remains to be seen. The clinical course of a patient with CAP admitted to the hospital but not to the intensive care unit depends on a number of variables, including the patient, the pathogen, and the hospital itself. The site-of-care decision can be based on 1 of 2 prediction rules. Neither of these rules, however, correlates with the etiology of CAP, and it is not clear whether they can be used to stratify patients according to prognostic factors. A pathogen may be found in only approximately one-third of hospitalized patients with CAP overall. An etiological diagnosis is more likely to be made for patients with CAP who are hospitalized in the intensive care unit (39%) than for those hospitalized in other wards (20%). The issue of randomization to treatment regimens and possible approaches to randomization are discussed. It seems clear, however, that randomization would have to take place immediately after entry of the patient into the study. The possibility of using risks for specific pathogens or risks for antimicrobial resistance is also addressed. However, there are no data to support the use of such risks as prognostic factors in CAP. The best approach for noninferiority trials involving hospitalized patients with CAP is to randomize patients who meet the inclusion criteria and to stratify them by hospital site, with block randomization within each site. Stratification by site takes into account local epidemiology and can balance differences in unmeasured confounders among sites.