Abstract
Background:
Utilization of balloon aortic valvulopasty (BAV) has increased in recent years. We aimed to explore the relationship between hospital and operator experience on mortality.
Methods:
We queried Healthcare Cost and Utilization Project’s Nationwide Inpatient Sample (NIS) between 1998 and 2010 using the ICD9 procedure code of 35.96 for valvuloplasty. The NIS represents 20% of all hospitals in the US. Patients with age > 60years with aortic stenosis were included and those with concomitant mitral, tricuspid or pulmonic stenosis were excluded. Severity of co-morbid conditions was defined with Deyo modification of Charlson’s comorbidity index. Primary outcome was in-hospital mortality. Annual operator and hospital procedure volume were calculated based on the unique operator and hospital identification numbers and were split into tertiles of the total volume per year. Mixed effects logistic regression modeling incorporating hospital ID as random effects was constructed to find out the independent predictors of in-hospital mortality.
Results:
A total of 2,127 BAV (weighted n= 9,686) were identified for hospital volume analysis and 1,030 BAV (weighted n = 5,125) were available for both operator and hospital volume analysis. In models incorporating either operator or hospital volume alone (hazard ratio (HR), 95% CI, p value) both operator volume (HR 0.5 per every unit increase in operator volume, 0.3-0.9, p<0.05) and hospital volume (0.8 per every 10 unit increase in hospital volume, 0.8-0.9, p <0.05) were independently associated with decreased mortality. However in a model incorporating both operator and hospital volume, only increasing operator volume was associated with decreased mortality (0.85, 0.73-0.98, p=0.03) and influence of hospital volume was no longer statistically significant. Relative risk reduction of mortality was 51% when BAV was performed in the highest tertile of both hospital and operator volume (p=0.05).
Conclusion:
We demonstrated that both annual operator and hospital procedure volume significantly impact in-hospital mortality for BAV, with operator volume playing a more significant role.