Assessing pharmacy residents' knowledge of biostatistics and research study design

Abstract

Background: Historically, clinicians have demonstrated a lack of confidence and poor aptitude for biostatistics as a tool for medical literature interpretation. Evaluation of pharmacy residents' ability to interpret biostatistics commonly used in peer-reviewed literature has not been previously conducted.

Objective: To evaluate the level of understanding and perception of biostatistics concepts among pharmacy residents.

Methods: A survey of postgraduate year 1 (PGY1) residents in American Society of Health-System Pharmacists-accredited residency programs was conducted in May 2009. The survey instrument consisted of 27 items, including 10 knowledge-based questions, and was distributed to residency programs for anonymous reporting via SurveyMonkey. The primary outcome of interest was biostatistics knowledge, defined as the percent total score of correct knowledge items. Statistical attitude and confidence questions were rated on a 5-point Likert-type scale (1 = strongly disagree, 5 = strongly agree). The t-test or 1-way analysis of variance was conducted, as appropriate, to assess for differences in mean biostatistics knowledge scores by respondent characteristics. Forward stepwise regression was used to identify which characteristics were independently associated with biostatistics knowledge.

Results: A total of 214 PGY1 residents responded to the online survey assessment, and a subset of respondents (n = 166) answered 1 or more of the biostatistics knowledge questions. Of those who responded to at least 1 knowledge assessment, the overall mean (SD) biostatistics knowledge score was 47.3% (18.50%; range 0-90). Overall, respondents were predominantly female (74%) and younger than 30 years (81%). Residents scored highest in the recognition of the purpose of a double-blind study (92.6%; 95% CI 88.52 to 96.67), interpretation of relative risk (75.8%; 95% CI 69.02 to 82.57), and identification of the appropriate analytic method for a nominal variable (69.4%; 95% CI 62.16 to 76.59). Bivariate analyses showed that there were statistically significant mean differences in knowledge scores by attitude (p = 0.001) and confidence (p < 0.001). The multivariate model showed that above-average confidence ratings were associated with an absolute increase of 7.6% in biostatistics knowledge score (p < 0.019) compared to those whose confidence rating was at or below average.

Conclusions: Overall, pharmacy residents' perception and understanding of biostatistics were poor in this assessment, which correlates with previous reports. Enhanced training in biostatistics and literature evaluation of both mentors and trainees should be incorporated in PharmD programs and residency training sites.