Hospitals vary significantly in cirrhosis mortality. We aimed to determine whether hospital structural characteristics mediate this effect.

Methods: We used 2009 data from the HCUP Nationwide Inpatient Sample and the American Hospital Association to create a multi-level model to evaluate the effect of patient factors and specific hospital characteristics on risk of in-patient mortality from cirrhosis. Structural factors were cumulatively added to the model to determine how much a specific factor affected inter-hospital variation in cirrhosis mortality.

Results: 13,029 cirrhosis admissions were sampled in 477 hospitals. Patient characteristics including age, sex, race, payer status, and APR-DRG morbidity varied significantly by hospital cirrhosis volume (p < 0.05). Structural characteristics including presence of liver transplant program, bed capacity, teaching status, payer mix, staffing patterns also varied across hospitals (p< 0.01). Patient factors were significantly predictive of mortality including hepatitis C, alcoholic liver disease, portal hypertension, hepatorenal syndrome and co-morbidities (all, p<0.04). After adjusting for patient factors, there was nearly a 12-fold difference in mortality across hospitals (1.2% – 14.2%, p = 0.05). Hospital cirrhosis volume explained 10.4% of risk-adjusted variation in mortality. Liver transplant program status accounted for 13.9% of variation, however, payer mix, teaching status, ICU capacity, and staffing explained an additional 40.2% of variation in mortality between hospitals (Table 1).

Conclusions: There are tremendous differences between hospitals admitting cirrhotic patients and their outcomes. After accounting for patient differences, hospital structural characteristics account for 64.5% of the overall variation between hospitals. Cirrhosis mortality variation may be ameliorated by structural, programmatic, and process improvement.

« Back to 2015 American Transplant Congress