评估公共卫生干预效果的中断时间序列方法及其应用  

An Introduction to Interrupted Time Series Design in Evaluating the Effect of Public Health Interventions and its Application

作　　者：徐笑寒 欧春泉 Xu Xiaohan;Ou Chunquan(Department of Biostatistics,Guangdong Provincial Key Laboratory of Tropical Disease Research,School of Public Health,Southern Medical University,510515,Guangzhou)

机构地区：[1]南方医科大学公共卫生学院生物统计学系,510515

出　　处：《中国卫生统计》2023年第1期41-44,共4页Chinese Journal of Health Statistics

基　　金：国家自然科学基金(81973140)。

摘　　要：目的中断时间序列(interrupted time series,ITS)方法是评估公共卫生干预效果的常用方法。本文详细介绍该方法的基本理论及统计分析中需重点关注的问题,并通过实例数据分析进一步阐述。方法探讨ITS分析中时间序列数据的过度离散、长期趋势和季节性的控制、非线性效应以及潜在的残差自相关等问题的解决方案。实例分析中,基于2002-2006年的西西里岛0~69岁人群的急性冠状动脉事件(acute coronary events,ACEs)数据,采用Quasi-Poisson回归评价2005年1月颁布的公共场所禁烟政策对ACEs住院率的影响,采用超额风险(excess risk,ER)和超额住院率(excess hospitalization rate,EHR)两个指标来反映干预的效应。结果基于不同的研究目的和数据特点,ITS统计建模时有不同的处理策略。西西里岛的禁烟政策使得ACE住院率下降12.28%(95%CI:7.40%~16.91%)。每年因此而避免发生的ACEs共有1440例,相当于每十万人口中避免396例(95%eCI:240~541)。结论ITS方法能够基于时间序列数据有效地估计公共卫生干预的效果,该方法也普遍适用于环境、医院管理和疫苗接种等领域的干预评估,但运用该方法时若相关统计问题处理不当可能导致结果偏差。Objective Interrupted Time Series(ITS)is a common method used in evaluating the effects of public health interventions.This study aimed to introduce the basic theory of ITS and the key issues in statistical analysis with an example.Methods To describe the solutions to the methodological problems in ITS,including over-dispersion of time series,control of long-term trend and seasonality,fitting of nonlinear intervention effects,and potential autocorrelation of residuals.In the example,the Quasi-Poisson regression was used to evaluate the effect of smoking ban in public places introduced in January 2005 on acute coronary events(ACEs)hospitalization based on the ACEs data of people aged 0-69 years in Sicily during 2002-2006.Two indicators of excess risk(ER)and excess hospitalization rate(EHR)were used to present the effect of intervention.Results The modelling strategies depend on research objectives and data characteristics.After the smoking ban policy,the risk of ACE hospitalization decreased by 12.28%(95%CI,7.40%~16.91%).A total of 1440 ACE cases were averted due to the smoking ban policy annually,correspondingly to 396(95%eCI,541~240)per 100000 population.Conclusion ITS method can effectively estimate the effects of public health interventions based on time-series data.This method is also generally suitable for interventions evaluation in the fields of environment,hospital management and vaccination.However,improper handling of relevant statistical problems would lead to biased results when using the ITS.

关 键 词：中断时间序列 公共卫生干预 POISSON回归 

分 类 号：R197.1[医药卫生—卫生事业管理]