Is poor hospital adherence to AMI process measures associated with increased patient mortality? An analysis of CMS clinical and administrative data

Popescu I, Werner R, Sarrazin MV, Cram P. Is poor hospital adherence to AMI process measures associated with increased patient mortality? An analysis of CMS clinical and administrative data. Paper presented at: Society of General Internal Medicine Annual Meeting; 2008 Apr 10; Pittsburgh, PA.

Related HSR&D Project(s)

• RCD 05-285 – The Role of Performance Measurement in Improving Healthcare Quality

BACKGROUND: Studies have documented that process measures at US hospitals have improved over recent years. At the same time, the gap between America´s best and worst hospitals appears to be widening and some hospitals have shown modest improvement. Little is known about the characteristics and outcomes of low-performing hospitals. METHODS: We used the Hospital Quality Alliance (HQA) database to identify all hospitals that reported AMI process measures to the Centers for Medicare and Medicaid Services (CMS) during 2004-2006. After excluding hospitals reporting less than 25 AMI cases during the three year study period, the remaining facilities were ranked and divided into deciles based on their compliance with AMI process measures. Compliance was measured as the ratio of the number of patients who received guideline-concordant AMI treatments divided by the number of patients eligible for these treatments. We defined low performing hospitals (N = 100) as all hospitals performing in the bottom AMI compliance decile for all three years. These hospitals were then compared to all other hospitals in the sample (N = 2,660). We then used 2005 MedPAR data to identify all patients aged 65 years or older admitted to the hospitals identified in the HQA database with AMI (N = 226,868). Risk-adjusted 30-day mortality was estimated for the low performing hospitals relative to all other hospitals using hierarchical models that adjusted for demographic and clinical patient risk factors, and included hospital random effects. The hospital random intercept was used as a measure of hospital risk-adjusted mortality. Individual hospitals were considered mortality outliers if the hospital random intercept was significantly different from the mean intercept (p < .05). RESULTS: Low-performing hospitals had significantly lower compliance with AMI process measures (average adherence 68% vs. 92%, p < .001), significantly fewer beds (102 vs. 236, p < .001), lower AMI volumes (29 vs. 111, p < .001), and were less likely to perform revascularization (7% vs 52%, p < .001). Medicare patients admitted to low performing hospitals tended to be older (mean age 81 vs 79 years, p < .001), were more likely to be female (56% vs. 50%, p < .001) and black (10% vs 7%, p < .001), and had more comorbid illnesses including CHF (51% vs. 43%, p < .001) and COPD (29% vs. 24%, p < .001). Low-performing hospitals had higher risk-adjusted mortality as compared to other hospitals (OR = 1.28, p < .001). When examining mortality among all hospitals , 41 hospitals had 30-day mortality rates that were significantly lower, and 10 hospitals had 30-day mortality rates that were significantly higher than the mean (p < .05). However, none of the low-performing hospitals as measured by process compliance was a high-mortality outlier. CONCLUSIONS: Hospitals with consistently low adherence to AMI process measures are significantly smaller and less likely to perform revascularization procedures compared to other hospitals; they are more likely to serve older minority patients, with more comorbidities. Risk-adjusted mortality appears to be higher for low-performing hospitals as a group. Viewed individually however, none of the low performing hospitals had significantly increased 30-day mortality rates. Poor performance in adherence to AMI standard of care is not necessarily correlated with significantly increased mortality rates for individual hospitals.