P. Kandagatla1, A. Worden1, A. Woodward1, A. H. Gupta1 1Henry Ford Health System/Wayne State University,Department Of Surgery,Detroit, MICHIGAN, USA
Introduction: Previous studies have shown a lack of required biostatistical knowledge among residents. There are few data on the departmental variation of biostatistical knowledge among residents. Furthermore, there is a paucity of data comparing residents and medical students. We hypothesized that residents’ biostatistical knowledge varies across departments and there are individual characteristics that may serve as predictors for increased knowledge.
Methods: We surveyed residents and medical students rotating at our institution. The survey tool included questions on demographics, prior biostatistics exposure, educational history, and confidence in biostatistics. The remainder of the survey included a previously validated 17-question biostatistics quiz. Descriptive statistics are used to summarize the responses. Univariate and bivariate analyses were done to compare means and calculate correlation, respectively. Multivariate analysis was performed to derive independent predictors of increased knowledge.
Results: A total of 218 participants completed the survey. Twenty-five (11.5%) were medical students and 193 (88.5%) were residents. The overall mean (SD) percent correct on the quiz was 48.3% (14.5%), and average correct was not different between students and residents(50.4% SD 13.2% vs 48.2% SD 14.6%, p = 0.45). There was a significant difference in the mean scores across all departments (range: 39.2-58.1%, p<0.01), which included Anesthesia, Emergency Medicine, Family Medicine, General Surgery, Internal Medicine, Neurosurgery, Ob/Gyn, Orthopedic Surgery, Radiology, and Urology. Ninety-three (42.7%) participants had a prior epidemiology course, 133 (61.0%) had a biostatistics course, and 117 (53.7%) had an evidence-based medicine course. There was no significant difference in the mean scores between those that had each type of course compared to those that did not. There was no significant correlation between average journal articles read per week (r = 0.09, p = 0.2), previous research publications (r = 0.12, p = 0.08), number of biostatistics didactics (r = -0.07, p = 0.33), and level of training (r = -0.01, p=0.84) with percent correct. Journal club attendance, however, was correlated with performance (r = 0.22, p<0.01). Confidence in participants’ understanding of statistical terms (r = 0.11, p = 0.11) was not correlated with performance. Linear regression revealed journal club attendance (b = 1.5, 95% CI 0.11-2.88, p = 0.03) and number of articles read per week (b = 2.8, 95% CI 0.11-5.45, p = 0.04) to be the only significant independent predictors for increased performance.
Conclusion: There was a significant variation in biostatistical knowledge across residency departments. Future educational interventions attempting to increase knowledge of biostatistics should perhaps focus on increasing journal club attendance and regular reading of medical literature.