Two-stage piecewise linear model for investigating doseresponse relationship in meta-analysis:methodology,examples,and applicationShort title:Two-stage piecewise linear model for dose-response meta-analysis
作者单位：Chinese Evidence-Based Medicine Center and Chinese Cochrane CenterWest China HospitalSichuan University Department of Clinical Epidemiology and BiostatisticsMcMaster University Center for Evidence-Based Medicine and Clinical ResearchTaihe HospitalHubei University of Medicine Department of UrologyZhongnan Hospital of Wuhan UniversityWuhan University
会议名称：《2017年中国卫生统计学学术年会》
会议日期：2017年
学科分类：1007[医学-药学(可授医学、理学学位)] 1006[医学-中西医结合] 100706[医学-药理学] 100602[医学-中西医结合临床] 10[医学]
关 键 词：piecewise linear function dose-response meta-analysis discrete exposure two-stage approach
摘 要：Background:Dose-response meta-analysis（DRMA）is widely employed to establishing the potential dose-response relationship between continuous exposures and disease ***,no method is readily available for exploring the relation between a discrete exposure and a binary or continuous ***:We proposed a two-stage piecewise linear（PL）DRMA model as a solution to this *** used sleep（a continuous exposure）and parity（a discrete exposure）data to illustrate how to apply PL model in DRMA using the Stata code *** also empirically compared the slopes of PL model with restricted cubic spline（RCS）model by sleep ***:PL DRMA model fitted well in our two *** sleep duration,our meta-analysis showed that the relative risk（RR）of all-cause mortality was 1.02（95%CI:1.01,1.04）for every 1-hour reduction of sleep duration among people who slept less than 7 hours,and 1.15（95%CI:1.06,1.25）for every 1-hour increase of sleep duration among people who slept more than 7 *** parity and risk of rheumatoid arthritis:among women with 3 or less births,the RR was 0.88（95%CI:0.77,1.00）for every 1-birth increment;otherwise（3 or more births）,the RR was 1.10（95%CI:0.99,1.23）for every 1-birth *** slopes of PL model fitted well to RCS curve based on sleep ***:Piecewise linear function is a simple and valid method for DRMA and can be used for discrete *** also represents a superior model to linear model in DRMA and may be an alternative model to non-linear model.