ABSTRACT Continuous monitoring of surgical quality reflected in time‐to‐event (TTE) outcomes is crucial for protecting patients from unnecessary health risks. The control chart scheme has gradually become one of the most common systems for assessing surgical quality levels and have made outstanding contributions. However, they overlook an inherent imperfection in surgical TTE outcomes: a substantial proportion of patients are cured postoperatively, and their cure status is generally unobservable. This limitation causes existing methods to underestimate the actual hazard, leading to unreliable monitoring results. Meanwhile, most of them lack a mechanism for monitoring the stability of surgical quality. In this paper, we propose a risk‐adjusted EWMA (Exponentially Weighted Moving Average) scheme based on a semi‐parametric mixture cure model (named RAC‐EWMA) and establish its monitoring scheme by explicitly deriving the corresponding weighted score test statistics. The former enables RAC‐EWMA to specifically monitor both the cure rate and hazard, while the latter provides it with the ability to simultaneously assess changes in average level and stability. In simulations and case studies, the average run lengths (ARLs) required for RAC‐EWMA to detect shifts are reduced by 24.23% and 8.95% compared to the best‐performing existing methods, respectively. Furthermore, it provides more detailed information about shift sources to support downstream analysis. In summary, RAC‐EWMA not only effectively monitors changes in both fixed and random effects for cure‐presenting survival data but also demonstrates superior performance over existing methods.
Liu et al. (Thu,) studied this question.