Von Korff M, Wagner E H, Saunders K
Center for Health Studies, Group Health Cooperative of Puget Sound, Seattle, WA 98101.
J Clin Epidemiol. 1992 Feb;45(2):197-203. doi: 10.1016/0895-4356(92)90016-g.
Using population-based automated pharmacy data, patterns of use of selected prescription medications during a 1 year time period identified by a consensus judgement process were used to construct a measure of chronic disease status (Chronic Disease Score). This score was evaluated in terms of its stability over time and its association with other health status measures. In a pilot test sample of high utilizers of ambulatory health care well known to their physicians (n = 219), Chronic Disease Score (CDS) was correlated with physician ratings of physical disease severity (r = 0.57). In a second random sample of patients (n = 722), its correlation with physician-rated disease severity was 0.46. In a total population analysis (n = 122,911), it was found to predict hospitalization and mortality in the following year after controlling for age, gender and health care visits. In a population sample (n = 790), CDS showed high year to year stability (r = 0.74). Based on health survey data, CDS showed a moderate association with self rated health status and self reported disability. Unlike self-rated health status and health care utilization, CDS was not associated with depression or anxiety. We conclude that scoring automated pharmacy data can provide a stable measure of chronic disease status that, after controlling for health care utilization, is associated with physician-rated disease severity, patient-rated health status, and predicts subsequent mortality and hospitalization rates. Specific methods of scoring automated pharmacy data to measure global chronic disease status may require adaptation to local prescribing practices. Scoring might be improved by empirical estimation of weighting factors to optimize prediction of mortality and other health status measures.
利用基于人群的自动化药房数据，通过共识判断过程确定的1年时间段内选定处方药的使用模式被用于构建慢性病状态指标（慢性病评分）。该评分从其随时间的稳定性及其与其他健康状况指标的关联方面进行了评估。在医生熟知的门诊医疗高利用率者的试点测试样本（n = 219）中，慢性病评分（CDS）与医生对身体疾病严重程度的评分相关（r = 0.57）。在第二个患者随机样本（n = 722）中，其与医生评定的疾病严重程度的相关性为0.46。在总体人群分析（n = 122,911）中，发现它在控制年龄、性别和医疗就诊次数后能够预测次年的住院率和死亡率。在一个人群样本（n = 790）中，CDS显示出较高的年度稳定性（r = 0.74）。基于健康调查数据，CDS与自我评定的健康状况和自我报告的残疾有中度关联。与自我评定的健康状况和医疗利用率不同，CDS与抑郁或焦虑无关。我们得出结论，对自动化药房数据进行评分可以提供一种稳定的慢性病状态指标，在控制医疗利用率后，该指标与医生评定的疾病严重程度、患者评定的健康状况相关，并能预测随后的死亡率和住院率。对自动化药房数据进行评分以衡量全球慢性病状态的具体方法可能需要根据当地的处方习惯进行调整。通过对加权因子进行实证估计以优化对死亡率和其他健康状况指标的预测，评分可能会得到改善。
