Statistical competencies for medical research learners: What is fundamental?

Abstract

IntroductionIt is increasingly essential for medical researchers to be literate in statistics, but the requisite degree of literacy is not the same for every statistical competency in translational research. Statistical competency can range from ‘fundamental’ (necessary for all) to ‘specialized’ (necessary for only some). In this study, we determine the degree to which each competency is fundamental or specialized.MethodsWe surveyed members of 4 professional organizations, targeting doctorally trained biostatisticians and epidemiologists who taught statistics to medical research learners in the past 5 years. Respondents rated 24 educational competencies on a 5-point Likert scale anchored by ‘fundamental’ and ‘specialized.’ResultsThere were 112 responses. Nineteen of 24 competencies were fundamental. The competencies considered most fundamental were assessing sources of bias and variation (95%), recognizing one’s own limits with regard to statistics (93%), identifying the strengths, and limitations of study designs (93%). The least endorsed items were meta-analysis (34%) and stopping rules (18%).ConclusionWe have identified the statistical competencies needed by all medical researchers. These competencies should be considered when designing statistical curricula for medical researchers and should inform which topics are taught in graduate programs and evidence-based medicine courses where learners need to read and understand the medical research literature.

Learning gaps among statistical competencies for clinical and translational science learners

“…These competencies include 21 (data safety and monitoring), 23 (meta-analysis), and 24 (early stopping rules). In our previous work [4], we specifically noted that these competencies are not fundamental for all learners although they are important for some, and we suggested that these competencies might be considered specialized statistical topics more appropriate for targeted training rather than for general graduate programs in CTS. We note that more than 20% of the statistical educators reported that instruction on data safety and monitoring and meta-analysis is not available in any master's coursework and that nearly 10% of the statistical educators reported that instruction on early stopping rules (in clinical trials) is not available in any doctoral coursework.…”

“…We suggest that greater attention should be paid to these competencies (2,7,10,11,12,13,16,17) and that graduate programs for CTS learners might need to fill the learning gaps by including these competencies in required coursework. Competencies seen as less fundamental in Enders et al [4] were less frequently included in the training of most CTS learners. These competencies include 21 (data safety and monitoring), 23 (meta-analysis), and 24 (early stopping rules).…”

“…To contextualize coverage of the competencies, we crossreferenced our data with a previous survey of 112 biostatisticians who graded each competency on whether or not they consider it to be fundamental to general training for CTS learners [4]. We define "percent fundamental" as the proportion of prior respondents who rated the competency as fundamental in the prior survey [4]. We evaluated the association between the percent fundamental from our prior work [4] and the percent each competency was covered in our current survey.…”

“…Committee of the CTSA National Consortium [12] and the Association of Schools of Public Health [13] and from established guidelines for the reporting of clinical and observational studies [14][15][16][17], we previously developed a comprehensive set of 24 statistical competencies for graduate coursework in the health sciences [4,5,12,18]. These competencies help define statistical topics that should be taught to health science learners of different types, including CTS learners.…”

What makes a biostatistician?

“…Biostatisticians need to be experts in their field (not in all of the mentioned above), but it is not enough to perform a methodologically correct and comprehensive analysis. The biostatistician must also be able to interpret the output and communicate the results to collaborators . Therefore, there are competencies required in addition to the biostatistics expertise.…”