A population-level evaluation of emergency general surgery models of care and clinical outcomes

Importance — Outcomes in emergency general surgery vary between hospitals, and models with dedicated resources and personnel have been developed with the aim of improving care. Existing literature is limited in scope and often does not involve nonoperatively treated patients, reducing generalizability.

Objective — To use population-level data to determine whether treatment in an emergency general surgery model is associated with clinical outcomes.

Design, setting, and participants — This was a retrospective cohort study of adults in Ontario, Canada, diagnosed with 1 of 9 emergency general surgery conditions and hospitalized between April 1, 2002, and December 31, 2019, using linked administrative data housed at ICES. Data analysis was conducted from June 24, 2024, to October 24, 2025.

Exposure — Treatment at a hospital with an emergency general surgery model of care in comparison with a standard surgeon on-call model.

Main outcome and measures — Death in the hospital or within 30 days of discharge was the primary outcome. Secondary outcomes included death at 90 days and complications, failure to rescue, and readmission at 30 days. Generalized estimating equations were used with a negative binomial distribution for mortality outcomes and a binomial distribution for other secondary outcomes. Clustering at the hospital level was accounted for, and effect modification according to diagnosis risk category (low, medium, or high) was evaluated.

Results — A total of 494 609 patients were included (median [IQR] age, 56 [40-72] years; 263 267 [53.2%] female), with 88 889 (18.0%) treated in an emergency general surgery model hospital. A total of 3069 patients (3.4%) in an emergency general surgery model died within 30 days, compared with 15 013 (3.7%) in a surgeon on-call model. Adjusted analyses showed an association between an emergency general surgery model and decreased adjusted relative risk (aRR) of 30-day death for patients with high-risk conditions (aRR, 0.85; 95% CI, 0.77-0.95) but not for those with low- or medium-risk conditions. Death at 90 days was likewise lower in emergency general surgery models for high-risk conditions (aRR, 0.82; 95% CI, 0.74-0.92). The odds of complications in patients with high-risk conditions showed a similar association (adjusted odds ratio, 0.68; 95% CI, 0.53-0.87), but there was no association with failure to rescue or readmission.

Conclusions and relevance — This cohort study demonstrated that death and complications were lower for patients with high-risk conditions in an emergency general surgery model of care. These findings suggest that these patients would benefit from formal systems to prioritize management at emergency general surgery model centers but that patients with lower-risk conditions are less likely to benefit from this care.