Trends in clinical trials reports in common cancers between 1989 and 2000

Purpose. Cancer clinical trials can be dichotomized into pilot (phase I and phase 11) and randomized controlled clinical trials (RCTs). The best data source for evidence-based medicine is from RCTs. However, many patients prefer to enroll onto pilot trials, and many investigators prefer to conduct or refer their patients to pilot trials. This exploratory study sought to describe the epidemiologic patterns of clinical trial reports in common cancers. Methods: A structured review was conducted of MEDLINE citations of all English clinical trials reports published between 1989 and 2000 in breast, lung, colorectal, prostate, and female genital cancers, plus leukemias and lymphomas. Each report was classified by design (RCT, pilot, or other) and country. The abstracts of RCTs were reviewed for sample size. Reports addressing screening or prevention were excluded. Results: A total of 12,035 reports, of which 3,560 were from RCTs, were found. The annual growth in RCT reports in breast, colorectal, and prostate cancer was significant (range, 4.8% to 8.5% per year) but was insignificant in leukemias, lymphomos, and female genital and lung cancers (range, 0.1% to 4.3% per year). Within each cancer, the average sample size per report did not change during the 12 years. Nonrandomized trial reports increased on average 15.1% per year (range, 10.1% to 23.2%). The United States accounted for 30% of all RCT reports and 45% of pilot trial reports. Conclusion: The faster growth in nonrandomized compared with RCT reports may reflect rapid advances in cancer biology or different structural, commercial, and financial incentives, especially in the United States compared with Europe. Unless additional studies show evidence of an increase in their quality, the modest growth in RCT reports may limit future evidence-based cancer care. (C) 2003 by American Society of Clinical Oncology.