Quality of Reporting of Harms in the NCI Sponsored Phase III Hematological Malignancies Trials.

Background: To make informed decisions, patients and physicians need accurate information both on treatment benefits and harms. However, reports of clinical trials typically focus on description of benefits only. Research supported by the NCI account almost for 100% of all publicly sponsored phase III trials in the US, and is widely considered as the most important for advancement of treatment against cancer. Therefore, it is important to assess the quality of reporting of treatment-related mortality (TRM) and morbidity (TRMorb) in this setting.

Methods: We evaluated all randomized controlled trials (RCTs) that have been conducted by six NCI sponsored cooperative oncology groups (COGs). We included all trials that were completed by 2002. Here we focus on hematological malignancy trials. Data related to TRM and TRMorb were extracted from publications that reported benefits (i.e. survival or event-free survival). Data were pooled using meta-analytic techniques to assess if the harms were worse in experimental vs. standard treatments.

Results: We evaluated,114 hematological-malignancy trials. Seventy-nine trials studied leukemias, 32 lymphomas and 3 either multiple myeloma or myelodysplastic syndrome. Quality of reporting for the most important methodological domains (such as randomization methodology, power calculation, drop-out rates, use of intention-to-treat analysis) was high. Data related to TRM were extractable from 41 trials (36%) only. Overall, TRM accounted for less than 5% of all deaths and was shown to be significantly higher in patients allocated to innovative treatments [Odds ratio 1.93: 99%CI (1.11–1.76)]. TRM was similar according to the type of hematological malignancy or COG. Treatment-related morbidity data were reported in different non-standardized formats; therefore, it was not possible to perform a meta-analysis to give an accurate assessment of TRMorb in experimental and conventional treatment arms.

Conclusion: Treatment-related harms have been inadequately reported even in the currently best available research. Given the fact, that the users of research evidence can only rely on the available published reports, the NCI should urgently improve the requirment for the standards of reporting of clinical trials it funds.