Statistical considerations in clinical trials testing empiric antibiotic regimens in patients with febrile neutropenia.

The Chalmers scoring system for assessing the quality of randomized controlled trials was applied to 19 papers (published in 1976-1984) reporting the results of trials of antibiotics in febrile neutropenic patients, it was found at the time of this survey (publication 1986) that statistical issues were inadequately addressed in the papers and that the scoring system was a measure of the quality of reporting in the papers rather than the quality of the actual trials. Some new items have been added to the system to improve the validity of results, and 15 more recent papers have been reviewed with reference to the updated scoring procedure, to find whether the quality of trials and of their reporting has improved since 1986. Further discussion of statistical issues in clinical trials is recommended.