观察性研究方法与医院资料的统计分析

Analysis of Hospital Data by Statistical Methods for Observational Studies

目的通过个案分析,说明如何采用观察性研究的方法对医院资料进行统计分析。方法正确的研究设计与搜集客观准确的基础数据,明确观察对象所能推论的总体,设立对比组,通过分层分析及率的标准化等方法进行组间差别比较。结果 医院资料不同于试验数据。由于医院资料通常不能随机分组,并存在混杂,不能简单地用假设检验方法推论因果关系。结论在对医院资料进行统计分析时,应遵循观察性研究的原则和方法,以避免可能存在的信息偏倚、选择偏倚和混杂偏倚。

Objective To demonstrate, by case study, how to analyze the hospital data through the statistical methods in observational studies. Methods Based on correct research design and reliable basic data, clearly understanding the reference population of the subjects, to compare the differences by setting up contrast group and grouping the subjects into strata. Results The hospital data is totally different with the experiment data. The causation can not be directly tested by the statistical inference from the differences among groups because the observations in the hospital data can not be randomly allocated, and the confounding exist. Conclusion The principles and methods for observational studies should be followed to reduce information bias, selection bias and confounding bias in statistical analysis of hospital data.