Sklar D P, Tandberg D
Department of Emergency Medicine, University of New Mexico School of Medicine, Albuquerque 87131-5246, USA.
Am J Emerg Med. 1996 Mar;14(2):170-2. doi: 10.1016/S0735-6757(96)90126-6.
A nonconcurrent prospective cohort study was conducted to evaluate if National Resident Matching Program (NRMP) rank developed using multivariate regression followed by consensus group activity is associated with perceived general performance during emergency medicine residency. All residents graduating from a university hospital-based residency program between 1990 to 1993 were ranked by university faculty, private attendings, charge nurses, and a clerk. Each evaluator was asked to order (from the strongest to the weakest) a deck of cards that contained only each graduate's name and picture. NRMP ranks were scaled to adjust for differences in each year's match list length. Evaluators were unaware of the graduates NRMP ranks. Agreement among raters was analyzed using Kendall's coefficient of concordance. Perceived ranks were compared with actual NRMP ranks using the Spearman correlation procedure. Twenty graduates were evaluated by eight different individuals, yielding 160 pairs of ranks. There was moderately strong agreement among evaluators about the relative strength of the 20 residents (W = 0.67, P < .001). The association of perceived rank with NRMP rank was much greater than that expected by chance (r(s) = .35, P < .0001). Applicants with better NRMP ranks were perceived as stronger residents, which supports the strategy of using formal statistical modelling followed by consensus group activity as a method of generating NRMP rank lists.
开展了一项非同期前瞻性队列研究，以评估使用多变量回归随后进行共识小组活动得出的全国住院医师匹配计划（NRMP）排名，是否与急诊医学住院医师培训期间的总体表现认知相关。1990年至1993年间从一所大学附属医院住院医师培训项目毕业的所有住院医师，由大学教员、私人主治医师、护士长和一名办事员进行排名。要求每位评估者将一副仅包含每位毕业生姓名和照片的卡片按顺序排列（从最强到最弱）。NRMP排名进行了调整，以适应每年匹配名单长度的差异。评估者不知道毕业生的NRMP排名。使用肯德尔和谐系数分析评估者之间的一致性。使用斯皮尔曼相关程序将认知排名与实际NRMP排名进行比较。20名毕业生由8名不同的人员进行评估，得出160对排名。评估者对20名住院医师的相对实力有中度较强的一致性（W = 0.67，P < .001）。认知排名与NRMP排名的关联远大于偶然预期（r(s) = .35，P < .0001）。NRMP排名较好的申请者被认为是更强的住院医师，这支持了使用正式统计建模随后进行共识小组活动作为生成NRMP排名列表的一种方法的策略。
