Leung K M, Elashoff R M, Rees K S, Hasan M M, Legorreta A P
Quality Initiatives Division, Health Net, Woodland Hills, Calif. 91367, USA.
Am J Public Health. 1998 Mar;88(3):377-81. doi: 10.2105/ajph.88.3.377.
The purpose of this study was to identify factors related to pregnancy and childbirth that might be predictive of a patient's length of stay after delivery and to model variations in length of stay.
California hospital discharge data on maternity patients (n = 499,912) were analyzed. Hierarchical linear modeling was used to adjust for patient case mix and hospital characteristics and to account for the dependence of outcome variables within hospitals.
Substantial variation in length of stay among patients was observed. The variation was mainly attributed to delivery type (vaginal or cesarean section), the patient's clinical risk factors, and severity of complications (if any). Furthermore, hospitals differed significantly in maternity lengths of stay even after adjustment for patient case mix.
Developing risk-adjusted models for length of stay is a complex process but is essential for understanding variation. The hierarchical linear model approach described here represents a more efficient and appropriate way of studying interhospital variations than the traditional regression approach.
本研究旨在确定与妊娠和分娩相关的因素，这些因素可能预测患者产后住院时间，并对住院时间的差异进行建模。
分析了加利福尼亚州产妇患者（n = 499,912）的医院出院数据。采用分层线性模型来调整患者病例组合和医院特征，并考虑医院内部结果变量的依赖性。
观察到患者住院时间存在显著差异。这种差异主要归因于分娩类型（阴道分娩或剖宫产）、患者的临床风险因素以及并发症的严重程度（如有）。此外，即使在调整患者病例组合后，各医院的产妇住院时间仍存在显著差异。
开发住院时间的风险调整模型是一个复杂的过程，但对于理解差异至关重要。本文所述的分层线性模型方法比传统回归方法更有效、更适用于研究医院间的差异。
