Kassie Gizat M, Nguyen Tuan A, Kalisch Ellett Lisa M, Pratt Nicole L, Roughead Elizabeth E
Quality Use of Medicines and Pharmacy Research Centre, School of Pharmacy and Medical Sciences, Sansom Institute for Health Research, University of South Australia, GPO Box 2471, Adelaide, SA, 5001, Australia.
University of South Australia, GPO Box 2471 (CEA-18), Adelaide, SA, 5001, Australia.
Drugs Aging. 2018 Mar;35(3):213-222. doi: 10.1007/s40266-018-0526-6.
Medicines are potentially modifiable risk factors for postoperative delirium. However, the extent to which preoperative medicines are included in risk prediction models (RPMs) is unknown.
This systematic review aimed to assess the extent of inclusion of preoperative medications in RPMs for postoperative delirium.
Articles were systematically searched from MEDLINE, EMBASE and CINAHL using Medical Subject Headings (MeSH) where possible and keywords for postoperative delirium and prediction model. Studies published until May 2017 with a primary outcome of postoperative delirium that developed an RPM containing preoperative patient information were considered. Where a study had two cohorts, a derivation and a validation cohort, findings from the derivation cohort were extracted and reported.
Eighteen prospective and one retrospective cohort studies were included for review. Of the 19 studies, only nine considered preoperative medication data, with medications appearing as predictor variables in five models. There was wide variability in the factors included in the final models, with the most frequent predictors being age and cognitive impairment, appearing in 13 (68%) and 11 (58%) RPMs, respectively.
While medications are commonly cited risk factors for delirium, they are not adequately considered when developing RPMs. Future studies aiming to develop an RPM for postoperative delirium should include preoperative medication data as a potential predictor variable because of the modifiable nature of medication use and its impact on other factors commonly in models, such as cognition.
药物是术后谵妄潜在的可改变风险因素。然而，术前药物在风险预测模型（RPMs）中的纳入程度尚不清楚。
本系统评价旨在评估术前用药在术后谵妄RPMs中的纳入程度。
尽可能使用医学主题词（MeSH）以及术后谵妄和预测模型的关键词，从MEDLINE、EMBASE和CINAHL中系统检索文章。纳入2017年5月之前发表的、以术后谵妄为主要结局且建立了包含术前患者信息的RPMs的研究。若一项研究有两个队列，即推导队列和验证队列，则提取并报告推导队列的结果。
纳入18项前瞻性队列研究和1项回顾性队列研究进行综述。在这19项研究中，只有9项考虑了术前用药数据，其中有5个模型将药物作为预测变量。最终模型中纳入的因素差异很大，最常见的预测因素是年龄和认知障碍，分别出现在13个（68%）和11个（58%）的RPMs中。
虽然药物是谵妄常见的风险因素，但在开发RPMs时未得到充分考虑。由于药物使用具有可改变性及其对模型中常见的其他因素（如认知）的影响，未来旨在开发术后谵妄RPMs的研究应将术前用药数据作为潜在的预测变量纳入。
