GEOGRAPHIC-VARIATION OF PROCEDURE UTILIZATION - A HIERARCHICAL MODEL APPROACH
GEOGRAPHIC-VARIATION OF PROCEDURE UTILIZATION - A HIERARCHICAL MODEL APPROACH
复制标题
DOI：
10.1097/00005650-199305001-00008
复制
发表时间：
1993-05-01
期刊：
影响因子：
3
通讯作者：
MORRIS, C
中科院分区：
文献类型：
--
作者：
GATSONIS, C;NORMAND, SL;MORRIS, C
在本研究中，简要介绍了用于量化和解释医疗保健利用变化的分层统计模型。说明性示例源自对最近患有急性心肌梗死的医疗保险患者冠状动脉造影利用率的州际差异分析。分层模型区分州内和州间变化：前者通过每个州的单独逻辑回归建模，以年龄和性别作为自变量，而后者通过州特定逻辑模型系数的多元正态分布建模。比较了替代计算方法并评估了模型拟合度。获得了平均患者和按性别分层的血管造影状态率分布的估计值。结果显示，各州之间血管造影的使用情况存在显着差异，但年龄和性别对是否进行血管造影的决定的影响仅存在中等程度的州际差异。这种分析方法比标准化方法可以获得更详细的结果，并考虑了聚合单位之间样本量的差异。分析的下一个主要步骤是通过汇集各州的数据来得出各个州逻辑模型的平滑估计。该分析还可以扩展到纳入其他患者特征，例如种族和合并症，以及州特征，例如地理位置和手术的可用性。
In this study, an abbreviated introduction to hierarchical statistical models for quantifying and explaining variations in the utilization of medical care is presented. The illustrative example was derived from an analysis of interstate variation in coronary angiography utilization for Medicare patients with a recent acute myocardial infarction. The hierarchical model distinguished within-from between-states variation: the former was modeled via a separate logistic regression for each state, with age and sex as the independent variables, while the latter was modeled via a multivariate normal distribution for the coefficients of the state-specific logistic models. Alternative computation approaches were compared and model fit was assessed. Estimates of the distribution of state rates of angiography for an average patient and for age-by-sex strata were obtained. The results showed substantial interstate variation in angiography utilization, but only moderate interstate variation in the effects of age and sex on the decision to perform angiography. This analytic approach allows substantially more detailed results than those by standardization, and accounts for sample size differences between units of aggregation. The next major step in the analysis would be to derive smoothed estimates of the individual state logistic models by pooling data across states. The analysis can also be extended to incorporate other patient characteristics, such as race and comorbidity, and state characteristics, such as geographic location and availability of the procedure.