论文部分内容阅读
逻辑回归是研究协变量对二元结果影响的一种常用的统计方法。它已被广泛应用于临床试验和观察性研究。然而,单因素回归得到的结果和多元逻辑回归得到的结果往往是相互矛盾的。在多元回归中可能对结果会显示出非常强烈的影响的一个协变量在单因素回归中可能不会,反之亦然。这些事实在生物医学研究中并没有引起足够的重视。误用逻辑回归在医学出版物中非常普遍。在本文中,我们研究了单因素和多因素逻辑回归分析的不一致性,并在多元逻辑回归分析的模型部分中给出建议。
Logistic regression is a commonly used statistical method to study the influence of covariates on binary results. It has been widely used in clinical trials and observational studies. However, the results obtained by the univariate regression and the results obtained by the multiple logistic regression are often contradictory. A covariate that may have a very strong effect on the outcome in multivariate regression may not be in a one-way regression and vice versa. These facts did not attract enough attention in biomedical research. Misuse of logistic regression is very common in medical publications. In this paper, we study the inconsistencies in the logistic regression analysis of single and multiple factors and give recommendations in the model section of multivariate logistic regression analysis.