Bioethical Issues in Biostatistical Consulting Study: Additional Findings and Concerns.

Abstract

To determine the frequency and severity of 17 requests for inappropriate analysis and reporting of data that biostatisticians received from researchers in statistical consultations, as indirectly experienced (i.e., either heard about or observed being asked of other biostatisticians). A randomly drawn sample of 522 members of the American Statistical Association who self-identified as consulting biostatisticians were selected to participate in an online survey with the Bioethical Issues in Biostatistical Consulting Questionnaire, specifically developed for this study. Of the 522 consulting biostatisticians contacted, 390 (74.7%) completed the survey. The top 4 most frequently reported indirectly experienced inappropriate requests rated as "high severity" by at least 50% of the respondents were as follows: 1) proposing a study with a flawed design, including insufficient power; 2) setting aside values when the outcome turns on a few outliers; 3) reporting results of data analysis from only subsets of the data; and 4) overstating the statistical findings well beyond what the data support so that readers are misled. This article is a follow-up report to our recently published article on 18 directly experienced inappropriate and/or unethical requests by biomedical researchers of their consulting biostatisticians. These additional survey findings from the Bioethical Issues in Biostatistical Consulting Questionnaire on 17 indirectly experienced inappropriate requests show the following: 1) the widespread nature of inappropriate, if not, frankly unethical requests made by biomedical researchers of their consulting biostatisticians and 2) the urgent need to develop and offer educational programs for biomedical researchers in training to correct these behaviors-be they intentional or accidental. This report speaks to the urgency for developing training programs for new and existing researchers to reduce the frequency of these inappropriate bioethical requests during biostatistical consultations. The findings from this U.S. national survey of biostatisticians on inappropriate requests for data analyses by biomedical researchers strongly suggest a need for remedial actions, including 1) new educational modules for in-development and currently employed biomedical researchers, 2) improved institutional environments by research universities and companies regarding job and publication pressures, and 3) inclusion of a collaborating biostatistician as a working research team member from the early planning stages of all biomedical studies.