The Surgical Assessment and Healthcare (SAH) Index: A Risk-Adjusted Framework for Surgeon-Level Quality Audit in Gastric Cancer
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Gastric cancer outcomes vary, yet surgeon-level variation is rarely quantified after risk adjustment. We tested whether surgeon identity independently predicts survival and developed and internally validated a risk-adjusted, surgeon-level framework for auditing surgical treatment quality, the Surgical Assessment and Healthcare (SAH) Index.
Study Design
Single-institution retrospective cohort study (Ruijin Hospital, Shanghai Jiao Tong University; NCT07180966 ) of 690 patients undergoing curative-intent resection for pStage I–III gastric adenocarcinoma in 2019 by eight consultant surgeons. Overall survival was modeled by multivariable Cox regression (199 events), with complete-case, stage-stratified, and upfront-surgery sensitivity models. The SAH Index expressed surgeon × stage observed-to-expected ratios for 5-year mortality and major morbidity (Clavien–Dindo ≥ IIIa), with flags from Poisson intervals.
Results
Independent predictors were stage (HR 2.969/step), age (HR 1.031/year), and non-distal gastrectomy (HR 1.480; all p ≤ .008). Surgeon identity was marginally associated with survival (Wald p = .047, likelihood-ratio p = .074): S6 (HR 2.167, 95% CI 1.288–3.647) and S8 (HR 1.968, 1.033–3.749) carried roughly double the reference hazard; S6 persisted across sensitivity, bootstrap, and propensity-score analyses. S8 attenuated on resampling. Stage-stratified models localized these (S6, Stage II; S8, Stage III). Both risk models were well calibrated. Benchmarking flagged one survival-outlier cell (S6, Stage II); no morbidity outlier was detectable, given few complication events and a weakly discriminating morbidity model.
Conclusions
A single surgeon showed a persistent risk-adjusted survival signal, localized to one stage and without a morbidity excess; the marginal overall association is best read as this one localized outlier rather than broad between-surgeon variation. The SAH Index is presented as a development-stage, internally validated framework for surgeon-level audit; prospective, multi-center validation is the planned next phase.
