Objective To explore the difference between continuous variables method and median split method in fitting logistic regression model,when independent variables and logitr meet linear relationship.Methods The simulation data were generated on the basis of two group divisible design case-control study.Justify the monotonerelationship between covariate and logitr by the “SmeiPar” software package of R software.Using continuous variables and median split method to discretize raw data,and then fit the logistic regression models.Results Compared with the median split method,the continuous variables method did better in goodness-of-fit test.Conclusion It is recommended to use two cut-off points maximum OR values method to discrete continuous variables if the relationship between these variables and logitπ-is monotonic.%目的 探讨当logitr与连续型自变量之间呈线性关系时,采用原始值法和中位数截断法拟合logistic回归模型的差异.方法 以成组设计病例对照研究为背景产生模拟数据,应用R软件“SmeiPar”软件包对logitπr与连续型自变量间函数关系进行判断,进而采用原始值法和中位数截断法分别拟合logistic回归模型.结果 原始值法较中位数截断法能够更好地拟合logistic回归模型.结论 当logitπ与连续型自变量间呈线性关系时,建议采用原始值法.
展开▼
