Effects of type and level of training on variation in physician knowledge in the use and acquisition of blood cultures: A cross sectional survey

Background: Blood culture (BCX) use is often sub-optimal, and is a user-dependent diagnostic test. Little is known about physician training and BCX-related knowledge. We sought to assess variations in caregiver BCX-related knowledge, and their relation to medical training. Methods: We developed and piloted a self-administered BCX-related knowledge survey instrument. Expert opinion, literature review, focus groups, and mini-pilots reduced > 100 questions in multiple formats to a final questionnaire with 15 scored content items and 4 covariate identifiers. This questionnaire was used in a cross-sectional survey of physicians, fellows, residents and medical students at a large urban public teaching hospital. The responses were stratified by years/level of training, type of specialty training, self-reported practical and theoretical BCX-related instruction. Summary scores were derived from participant responses compared to a 95% consensus opinion of infectious diseases specialists that matched an evidence based reference standard. Results: There were 291 respondents (Attendings=72, Post-Graduate Year (PGY)≥3=84, PGY2=42, PGY1=41, medical students=52). Mean scores differed by training level (Attending=85.0, PGY3=81.1, PGY2=78.4, PGY1=75.4, students=67.7)[p≤0.001], and training type (Infectious Diseases=96.1, Medicine=81.7, Emergency Medicine=79.6, Surgery=78.5, Family Practice=76.5, Obstetrics-Gynecology=74.4, Pediatrics=74.0)[p≤0.001]. Higher summary scores were associated with self-reported theoretical [p≤0.001] and practical [p=0.001] BCX-related training. Linear regression showed level and type of training accounted for most of the score variation. Conclusions: Higher mean scores were associated with advancing level of training and greater subject-related training. Notably, house staff and medical students, who are most likely to order and/or obtain BCXs, lack key BCX-related knowledge. Targeted education may improve utilization of this important diagnostic tool.