Neurosurgical mortality rates: what variables affect mortality within a single institution and within a national database?

Abstract

Object: Mortality rate is a common outcome measure used by patients, families, physicians, insurers, and health care policy makers to evaluate and measure the quality of health care. The mortality index is a heavily used metric to measure survival, and is a key indicator in hospital report cards and national rankings. The significance of this metric is belied by the literature, which fails to accurately detail the overall mortality rate within the neurosurgical population. Given that there is no gold standard that can be used as a baseline, it is difficult to make durable interinstitutional comparisons concerning performance. In Part I of this paper, the authors examined an academic neurosurgical program's mortality rate and the effect of certain variables on this rate. In Part II, they assumed a broader perspective, examining a group of institutions, the University HealthSystem Consortium (UHC) Clinical Database/Resource Manager, and identifying factors that may be responsible for variability in the mortality index between hospitals.

Methods: Over a 36-month period, the authors' neurosurgical service performed 3650 procedures. Monthly "mortality and morbidity" conference logs were reviewed to collect information on the number of deaths. Deaths were classified according to elective or nonelective admission status. Additionally, the authors reviewed the UHC Clinical Database/Resource Manager for information regarding mortality rates in various other neurosurgical programs. These data reflected a 12-month period. Comparisons of hospital mortality indices were based on the percentage of transferred patients (both emergency department [ED] and inpatient), whether a hospital was a designated Level 1 trauma center, whether a hospital was designated a certified stroke center, and also based on the number of Medicaid patients treated.

Results: Sixty-two patients met the criteria to be considered neurosurgery-related deaths at the authors' institution (1.7% of all cases): 9 elective admissions (15%), 3 nonelective direct admissions (5%), 24 transfer patients (39%), and 26 ED admissions (42%). Causes of death included trauma (40%), stroke (33%), tumor (14%), spinal disease (8%), and infection (6%). Evaluation of the UHC data revealed that a mortality index of >or= 1.00 was seen in the following hospital types: trauma centers, hospitals with 11-20% Medicaid patients, and those with > 50,000 ED admissions. A nonstatistically significant trend toward increasing mortality rates was seen in hospitals with a lower percentage of elective neurosurgical cases, in Level 1 trauma centers, and in hospitals that were not certified stroke centers. Significance was seen in comparisons of hospitals with the highest and lowest mortality index quartiles in the following groups: trauma centers, hospitals with > 10% Medicaid patients, and hospitals with a high number of ED visits.

Conclusions: Many variables appear to impact the mortality rate within the neurosurgical population. The authors' observations have illuminated some of the reasons why: the data are elusive, documentation is variable, and the modes of statistical analysis are questionable. The first step in addressing this issue is to identify that there is a problem. The authors believe that this study has done so. Presently there is no definitive or reliable source for rating the quality of overall neurosurgical care, nor is there a good and complete source for understanding the quality of neurosurgical care in the US. It is important to view these results as the initial steps to a better understanding of patient outcomes, their measures, and their impact on neurosurgical practice.