To the Editor—Propensity score methods are increasingly being used to control for confounding by indication in observational studies when random assignment may not be feasible. In this context, adjustment for baseline differences between study groups by propensity score methods are often more practical and statistically more efficient compared to techniques such as multivariable regression or stratification [1]. Unlike conventional adjustment methods that use a limited number of covariate factors, propensity score methods may allow the integration of a larger number of covariables and can reduce confounding effects to make internally valid estimates of the relationship between an exposure (eg, treatment duration) and an outcome (eg, mortality) [1].

中文翻译：

---

要求标准化报告和倾向评分方法的合理性

致编辑-在随机分配可能不可行的观察研究中，越来越多地采用倾向得分方法来控制混淆。在这种情况下，与多变量回归或分层技术相比，通过倾向评分法对研究组之间的基线差异进行调整通常更为实用，并且在统计上更为有效[1]。与使用有限数量的协变量因子的常规调整方法不同，倾向评分方法可能允许集成大量协变量，并且可以减少混淆影响，从而对暴露（例如治疗持续时间）与治疗风险之间的关系做出内部有效的估计。结果（例如死亡率）[1]。