Learning from disaster.
Learning from disaster.
复制标题
DOI：
10.1136/bmj.302.6773.414-c
复制
发表时间：
1991-02
影响因子：
--
通讯作者：
D. Dodwell
中科院分区：
文献类型：
--
作者：
D. Dodwell
先生，--马丁·桑德勒博士和彼得·梅尔博士对我们论文的评论中有几处错误。[24]它建议将计划的再入院纳入再入院率的分子中。他们没有被包括在内。它还建议将日间病例包括在分子中。日间病例入院总是有计划的，因此也被排除在外。这封信暗示死亡被意外地排除在分母之外。事实上，死亡被故意排除在外。已经死亡的人不再有被重新接纳的风险，因此不能列入分子。标准流行病学实践表明，他或她也应该被排除在分母之外。信中指出，很大一部分被选中进行个别审计的再次入院患者的病例记录无法获得。事实上，93名再次入院的患者中有74名（79%）有病例记录：这是一个相当可观的“应答率”。“然而，我们同意，一些选择偏见的因素可能在起作用，尽管我们认为推测病例记录不可用的患者是否更有可能属于可避免或不可避免的再入院类别是不明智的。我们完全不同意一个有用的结果指标总是受到不正当的激励的影响。再入院率（无论如何衡量），如果被用作结果的指标，特别容易受到不正当的激励，因为再入院率是一个特别糟糕和无法解释的结果代理。很难想象一个有效的结果指标--一个真正衡量健康状况改善的指标--怎么会导致不正当的激励。艾伦·麦克
SIR,-There are several errors in the letter by Drs Martin Sandler and Peter Mayer' commenting on our papers.`24 It suggested that planned readmissions were included in the numerator of a readmission rate. They were not included. It also suggested that day cases were included in the numerator. Day case admissions are invariably planned and would therefore also be excluded. The letter suggested that deaths were accidentally excluded from the denominator. In fact, deaths were purposely excluded. A person who has died is no longer at risk of being readmitted and therefore cannot be included in the numerator. Standard epidemiological practice suggests that he or she should also be excluded from the denominator. The letter suggested that the case notes of a substantial proportion of readmitted patients selected for individual audit were unavailable. In fact, case notes were available for 74 of the 93 patients readmitted (79%): a respectable "response rate." We agree, however, that some element of selection bias might be in play, although we think it unwise to speculate whether patients whose case notes were unavailable would be more likely to fall into the category of avoidable or unavoidable readmissions. We entirely disagree that a useful outcome indicator will always be subject to perverse incentives. Readmission rate (however measured) if used as an indicator of outcome is particularly subject to perverse incentives because a readmission rate is a particularly bad and uninterpretable proxy for outcome. It is hard to see how a valid outcome indicator-a true measurement of improvement in health status-could cause perverse incentives. AILEEN CLARKE