Reliability of a 25-Item Low-Stakes Multiple-Choice Assessment of Bronchoscopic Knowledge

Abstract

A need for improved patient safety, quality of care, and accountability has prompted the development of competency-based educational processes. Assessment tools related to bronchoscopy training, however, have not yet been developed or validated. To determine whether 25 multiple-choice questions (MCQs) extracted from the free, Web-based Essential Bronchoscopist (EB) learning guide qualify in their original form as a preliminary pool of questions for a low-stakes assessment of bronchoscopic knowledge. Twenty-five randomly selected MCQs from among the top 70 question-answer sets of the EB were administered to 40 self-declared novice bronchoscopists (n = 13), experienced bronchoscopists (n = 21), and expert bronchoscopists (n = 6). A difficulty index and a discrimination index (DI) were calculated for each item. Internal consistency reliability was calculated using item-total correlation and Cronbach alpha. Content validity was determined by five independent experts. Ideal test items based on a difficulty index and item-total correlation were administered to a different group of 24 bronchoscopists to prospectively reassess internal consistency reliability. The mean (+/- SD) score for the 40 participants was 16.47 +/- 3.72 (median score, 17; score range, 7 to 22). The mean difficulty index was 0.65 +/- 0.22, and the mean DI was 0.52 +/- 0.28. Item total-correlations ranged from - 0.01 to + 0.71. Test content was unanimously validated. The Cronbach alpha was 0.69. There was no significant correlation between scores and the number of bronchoscopies performed or self-declared expertise. Eleven ideal test MCQs were identified. The internal consistency of these items remained satisfactory (Cronbach alpha = 0.75) when assessed prospectively in a different cohort. Reliable and valid MCQs were identified to initiate a preliminary pool of questions for a low-stakes assessment of bronchoscopic knowledge.