Data published in BMC Infectious Diseases indicated that the diagnostic criteria for community-acquired pneumonia (CAP) within randomized controlled trials (RCTs) are highly heterogeneous, a fact that may have consequences on the validity of trial results.
The study investigators analyzed 47 RCTs related to CAP recorded on ClincalTrials.gov, and as a result of high heterogeneity, they further divided similar CAP inclusion criteria into 8 patterns, based on 42 different CAP inclusion criteria combinations. The diagnostic performances of these CAP definition patterns were then applied to a reference population of 319 suspected cases. The diagnosis was confirmed in 163 cases and excluded in 156 cases by an adjudication committee after a systematic thoracic computed tomography (CT)-scan and a 28-day follow-up period.
The heterogeneity of the inclusion criteria could not be explained by methodology or study objectives. The range of diagnostic performance of the 8 patterns were 9.8–73.0% for sensitivities, 56.4% to 97.4% for specificities, 63.6% to 83.6% for positive predictive values, and 50.8% to 66.7% for negative predictive values. No CAP definition had either sensitivity or specificity higher than 65%. The rate of false positives ranged from 1% to 21%, depending on the CAP definition.
The study investigators concluded that the high level of heterogeneity found in the inclusion criteria for CAP in RCTs and their considerable variation in diagnostic performance could mean RCTs are including false positives, as well as missing patients. This may subsequently “[affect] the interpretation of the results of randomized trials, which remain the basis of evidence-based guidelines.” They further suggested “a ‘two-step strategy’, associating a sensitive combination of inclusion criteria identified in our study, with a systematic per-protocol analysis on the sub-population of patients whose CAP diagnosis is established a posteriori by an adjudication committee, including if possible a thoracic CT-scan.”