Short-term survival analysis of a risk-adjusted model for ovarian cancer care

To quantify the impact on short-term ovarian cancer survival associated with treatment at high-performing hospitals using the observed-to-expected ratio (O/E) for adherence to ovarian cancer treatment guidelines as a risk-adjusted measure of hospital quality care.

This was a retrospective population-based study of stage I-IV invasive epithelial ovarian cancer reported to California Cancer Registry 1996-2017. A fit logistic regression model, risk-adjusted for patient and disease characteristics, was used to calculate O/E for each hospital stratified by hospital annual case volume. Cox proportional hazards model was used for survival analyses at 3, 6, 12, 24 months and stratified according to sociodemographic characteristics.

The study population included 35,725 subjects treated at 443 hospitals: Low-O/E - 26.4% of cases; Intermediate-O/E - 55.5% of cases; and High-O/E - 18.1% of cases. Overall median survival by hospital category was: High-O/E = 72.5 months (95% CI = 68.6-78.6 months), Intermediate-O/E = 68.6 months (95% CI = 65.9-71.6 months), Low-O/E = 47.0 months (95% CI = 44.2-49.2 months). Initial treatment at a High-O/E hospital (HR = 1.00) was a statistically significant and independent predictor of improved short-term survival compared to Low-O/E hospitals at 3 months (HR = 1.46, 95% CI = 1.29-1.65), 6 months (HR = 1.35, 95% CI = 1.22-1.50), 12 months (HR = 1.27, 95% CI = 1.17-1.38), and 24 months (HR = 1.19, 95% CI = 1.11-1.27). Significant and independent associations between improved sort-term survival and High/O/E care were observed for Whites, Hispanics, Asian/Pacific Islanders (A/PI), across SES strata, and among all payer categories.

Ovarian cancer care at a High-O/E hospital is an independent predictor of improved outcome and the survival advantage is disproportionately weighted toward the short-term time horizon following diagnosis.