RESUMEN
OBJECTIVES: The relationship between outcomes, patient safety indicators and volume has been well established in patient's undergoing craniotomy for brain tumor. However, the determination of "high" and "low" volume centers have been subjectively derived. We present a paper with a novel method of objectively determining "high" volume centers for craniotomy for brain tumor. METHODS: Patients from 2002 to 2011 were identified in the Nationwide Inpatient Sample database using ICD-9 codes related to craniotomy for brain tumor. Primary endpoints of interest were hospital PSI event rate, in-hospital mortality rate, observed-to-expected PSI event ratio, and O/E in-hospital mortality ratio. Using a zero-inflated gamma model analysis and a cutpoint analysis we determined the volume threshold between and "high" and "low" volume hospitals. We then completed an analysis using this determined threshold to look at PSI events and mortality as they relate to "high" volume and "low" volume hospitals. RESULTS: 12.4 % of hospitals were categorized as good performers using O/E ratios. Regarding in-hospital mortality, 16.8 % were good performers. Using the above statistical analysis the threshold to define high vs. low volume centers was determined to be 27 craniotomies. High volume centers had significantly lower O/E ratios for both PSI and mortality events. The PSI O/E ratio was reduced 55 % and mortality O/E ratio reduced 73 % at high volume centers as defined by our analysis. CONCLUSIONS: Patients treated at institutions performing >27 craniotomies per year for brain tumors have a lower likelihood of PSI events and decreased in-hospital morbidity and mortality.