Bias in reporting of endpoints of efficacy and toxicity in randomized clinical trials (RCTs) for women with breast cancer (BC).

Abstract

6043 Background: Phase III RCTs are designed to assess clinically important differences in endpoints that reflect benefit to patients. Accurate and unbiased reporting is essential to guide rational therapy. Here we evaluate the quality of reporting of the primary endpoint (PE) and of toxicity in RCTs for BC. Methods: PUBMED was searched from 1995-2011 to identify RCTs for BC. Scales for assessing bias in reporting of the PE and of toxicity were developed. For the PE, scales assessed whether the concluding statement of the abstract (CSAbs) was (a) based on the PE, (b) described appropriately a statistically positive or negative result for the PE, or (c) was based on secondary endpoints. Bias and completeness of reporting of toxicity was assessed using a hierarchy scale of whether reporting occurred in the CSAbs, elsewhere in the abstract, in the discussion of the article or only in the results. Association of bias with Journal Impact Factor (JIF); changes in the PE compared to protocol information in clinicaltrials.gov and funding source was also evaluated. Results: 164 trials were evaluated; 33% showed bias in reporting of the PE. The PE was more likely to be reported in the CSAbs if statistically significant [OR 5.2, 95% CI 1.9-14.3, p=0.001]. A statistically negative PE was not reported in the CSAbs in 27% of trials. Of the 30 trials where protocol information was available in clinicaltrials.gov, there were non-significant associations for the PE to show a positive result if had been changed, and for greater bias in reporting the PE if it was unchanged. 67% of studies showed bias in reporting of toxicity. Only 14% mentioned toxicity in CSAbs. When the PE was positive, bias in reporting of toxicity was more common [OR 2.0, 95% CI 1.0-3.9, p=0.04]. There was no apparent association between bias in reporting of either the PE or toxicity and JIF or funding source, but a non-significant association between change in the PE and industry funding. Conclusions: Bias in reporting of the PE is common especially for studies with a negative PE. Reporting of toxicity is poor especially for studies with a positive PE. Changing of the PE appears to be a strategy to increase the likelihood of observing a statistically significant result.