Background: Randomized controlled trials (RCTs) are generally regarded as the gold standard for demonstrating causality because they effectively mitigate bias from both known and unknown confounders. However, conducting an RCT is not always feasible because of logistical and ethical considerations. This is especially true when evaluating surgical interventions, and non-randomized study designs must be utilized instead.

中文翻译：

---

高级统计：多元 Logistic 回归、Cox 比例风险和倾向得分

背景：随机对照试验 (RCT) 通常被认为是证明因果关系的金标准，因为它们有效地减轻了已知和未知混杂因素的偏倚。然而，出于后勤和伦理方面的考虑，进行 RCT 并不总是可行的。在评估手术干预时尤其如此，必须改用非随机研究设计。