Bristol Eye Hospital, Bristol, UK.
Eye (Lond). 2011 Aug;25(8):1010-5. doi: 10.1038/eye.2011.103. Epub 2011 May 6.
To develop a methodology for case-mix adjustment of surgical outcomes for individual cataract surgeons using electronically collected multi-centre data conforming to the cataract national data set (CND).
Routinely collected anonymised data were remotely extracted from electronic patient record (EPR) systems in 12 participating NHS Trusts undertaking cataract surgery. Following data checks and cleaning, analyses were carried out to risk adjust outcomes for posterior capsule rupture rates for individual surgeons, with stratification by surgical grade.
A total of 406 surgeons from 12 NHS Trusts submitted data on 55,567 cataract operations between November 2001 and July 2006 (86% from January 2004). In all, 283 surgeons contributed data on >25 cases, providing 54,319 operations suitable for detailed analysis. Case-mix adjusted results of individual surgeons are presented as funnel plots for all surgeons together, and separately for three different grades of surgeon. Plots include 95 and 99.8% confidence limits around the case-mix adjusted outcomes for detection of surgical outliers.
Routinely collected electronic data conforming to the CND provides sufficient detail for case-mix adjustment of cataract surgical outcomes. The validation of these risk indicators should be carried out using fresh data to confirm the validity of the risk model. Once validated this model should provide an equitable approach for peer-to-peer comparisons in the context of revalidation.
利用符合白内障国家数据集(CND)的电子收集多中心数据,为个体白内障外科医生开发一种手术结果病例组合调整方法。
从 12 家参与 NHS 信托机构的电子患者记录(EPR)系统中远程提取常规收集的匿名数据,这些机构正在进行白内障手术。在进行数据检查和清理后,针对个别外科医生的后囊破裂率进行风险调整结果分析,并按手术级别进行分层。
共有 12 家 NHS 信托机构的 406 名外科医生提交了 2001 年 11 月至 2006 年 7 月期间 55567 例白内障手术的数据(2004 年 1 月以来的占比为 86%)。共有 283 名外科医生提供了>25 例手术的数据,为详细分析提供了 54319 例手术。为所有外科医生以及三个不同等级的外科医生分别呈现了个体外科医生病例组合调整结果的漏斗图。图中包括 95%和 99.8%置信限,以检测手术离群值的病例组合调整结果。
符合 CND 的常规收集电子数据为白内障手术结果的病例组合调整提供了足够的细节。应使用新数据验证这些风险指标的有效性,以确认风险模型的有效性。一旦验证,该模型应在重新验证的背景下为同行比较提供公平的方法。