High-volume facilities are not always low risk: comparing risk-standardized mortality rates versus facility volume as quality measures in surgical neuro-oncology

Abstract

Objective: Risk-standardized mortality rates (RSMRs) have recently been shown to outperform facility case volume as a proxy for surgical quality in lung and gastrointestinal cancer. The aim of this study was to investigate RSMR as a surgical quality metric in primary CNS cancer.

Methods: This retrospective observational cohort study used data from the National Cancer Database, a population-based oncology outcomes database sourced from more than 1500 institutions in the United States, and included adult patients 18 years of age and older who were diagnosed with glioblastoma, pituitary adenoma, or meningioma and were treated with surgery. For each group, RSMR quintiles and annual volume were calculated in a training set (2009-2013) and these thresholds were applied to the validation set (2014-2018). In this paper, the authors compared the effectiveness and efficiency of facility volume-based versus RSMR-based hospital centralization models and evaluated the overlap between the two systems. A patterns-of-care analysis was also performed to explore socioeconomic predictors of being treated at better-performing treating facilities.

Results: A total of 37,838 meningioma, 21,189 pituitary adenoma, and 30,788 glioblastoma patients were surgically treated from 2014 to 2018. There were substantial differences between RSMR and facility volume classification schemes among all tumor types. In an RSMR-based centralization model, an average of 36 patients undergoing glioblastoma surgery would need to relocate to a low-mortality hospital to prevent one 30-day mortality following surgery, whereas 46 would need to relocate to a high-volume hospital. For pituitary adenoma and meningioma, both metrics were inefficient in centralizing care to reduce surgical mortality. Additionally, overall survival for glioblastoma patients was better modeled in an RSMR classification scheme. Analyses to investigate the impact of care disparities found that Black and Hispanic patients, patients earning less than $38,000, and uninsured patients were more likely to be treated at high-mortality hospitals.

Conclusions: RSMR is more effective and efficient than a traditional volume-based approach for preventing early postoperative death in glioblastoma surgery. These data have important implications for future quality-related studies in neurosurgical oncology and may be relevant for healthcare/insurance payments, hospital evaluation assessments, healthcare disparities, and the standardization of care across hospitals.

Keywords: facility volume; glioblastoma; meningioma; neuro-oncology; oncology; pituitary adenoma; risk-standardized mortality rates.