Objective
To develop an observed-to-expected ratio (O/E) for adherence to National Comprehensive Cancer Network (NCCN) ovarian cancer treatment guidelines as a risk-adjusted hospital measure of quality care correlated with disease-specific survival.Methods
Consecutive patients with stages I-IV epithelial ovarian cancer were identified from the California Cancer Registry (1/1/96-12/31/06). Using a fit logistic regression model, O/E for guideline adherence was calculated for each hospital and distributed into quartiles stratified by hospital annual case volume: lowest O/E quartile or annual hospital case volume <5, middle two O/E quartiles and volume ≥5, and highest O/E quartile and volume ≥5. A multivariable logistic regression model was used to characterize the independent effect of hospital O/E on ovarian cancer-specific survival.Results
Overall, 18,491 patients were treated at 405 hospitals; 37.3% received guideline adherent care. Lowest O/E hospitals (n=285) treated 4661 patients (25.2%), mean O/E=0.77±0.55 and median survival 38.9months (95%CI=36.2-42.0months). Intermediate O/E hospitals (n=85) treated 8715 patients (47.1%), mean O/E=0.87±0.17 and median survival of 50.5months (95% CI=48.4-52.8months). Highest O/E hospitals (n=35) treated 5115 patients (27.7%), mean O/E=1.34±0.14 and median survival of 53.8months (95% CI=50.2-58.2months). After controlling for other variables, treatment at highest O/E hospitals was associated with independent and statistically significant improvement in ovarian cancer-specific survival compared to intermediate O/E (HR=1.06, 95% CI=1.01-1.11) and lowest O/E (1.16, 95% CI=1.10-1.23) hospitals.Conclusions
Calculation of hospital-specific O/E for NCCN treatment guideline adherence, combined with minimum case volume criterion, as a measure of ovarian cancer quality of care is feasible and is an independent predictor of survival.