Standardized quality control in esophageal squamous cell carcinoma surgery: associations with favorable perioperative outcomes and long-term survival

Background Real-world evaluations associating surgical quality indicators with outcomes in esophageal squamous cell carcinoma (ESCC) remain limited. This study assessed processes of care, perioperative outcomes, and long-term survival within a high-volume ESCC surgical program under a standardized quality framework. Methods We analyzed 2952 patients undergoing esophagectomy (June 2018-September 2023) under standardized clinical pathways. The primary endpoint was overall survival (OS); secondary endpoints included preoperative positron emission tomography-computed tomography (PET-CT) use, neoadjuvant therapy adherence, lymph node yield, R0 resection rate, complications, length of hospitalization stay, and 30- and 90-day mortality. OS was estimated by the Kaplan-Meier method; multivariable Cox regression assessed associations between selected quality indicators and OS, adjusted for clinical confounders and calendar period. Results The 1-, 3-, and 5-year OS rates were 95.7%, 85.2%, and 76.4%. PET-CT was performed in 83.1%; 76.0% of indicated patients received neoadjuvant therapy. Median lymph node yield was 34; R0 resection rate was 95.6%. Complications declined from 18.9% to 15.6%; median length of hospitalization stay was 8 days; 30- and 90-day mortality were 0.3% and 0.6%. Higher lymph node yield (per 10-node increase: hazard ratio [HR] 0.88 [95% CI: 0.81, 0.96]), R0 resection (HR, 0.40 [95% CI: 0.29, 0.55]), and neoadjuvant therapy adherence (HR, 0.67 [95% CI: 0.50, 0.88]) were associated with improved OS after adjustment. Conclusions Within a unified quality framework, this ESCC cohort shows favorable performance on quality indicators, low perioperative risk, and durable survival. Higher lymph node yield, R0 resection, and neoadjuvant therapy adherence are associated with improved survival after adjustment. Together, the predefined quality indicators may inform multicenter quality benchmarking, though causal inference remains limited.