Background:
Although the distribution of cardiologists and mortality for cardiovascular conditions are both known to vary across regions of the United States, no study has examined the relationship between regional cardiologist density and patient mortality for acute myocardial infarction (AMI) or heart failure (HF).
Methods:
We used 2010 Medicare administrative claims data for AMI and HF. Pneumonia (PN) was used as a control condition. Primary outcomes were death at 30 days and 1 year from admission. For each Hospital Referral Region (HRR), we used the 2010 Bureau of Health Professionals’ Area Resource File to define cardiologist density (number of cardiologists divided by population aged 65+) and 4 HRR characteristics: primary care physician density, total physician density, unemployment rate, and percent white race. We used 2-level hierarchical logistic regression models to examine the association between cardiologist density by tertile and mortality for each condition adjusting for (Model A) patient age, sex, and condition-specific comorbidities, and (Model B) patient and HRR characteristics.
Results:
Median (interquartile range) cardiologist density per 100,000 in the low, middle, and high tertiles of HRRs was 26.3 (22.9-29.9), 38.6 (36.5-43.1), and 64.5 (54.4-85.3), respectively. There were 171,126 admissions for AMI, 352,853 for HF, and 343,053 for PN. The 30-day mortality rates were 15.3% (26,290), 11.7% (41,121), and 11.9% (40,906), and 1-year mortality rates were 32.1% (55,292), 40.4% (142,612), and 35.2% (120,666), respectively (Table). For 30-day mortality, while model A showed lower mortality with higher cardiologist density for all conditions (odds ratios (ORs): 0.84-0.95), model B showed no associations. For 1-year mortality, while model A showed lower mortality in the high cardiologist density tertile for AMI (OR=0.93) and HF (OR=0.91) and no associations for PN, model B showed no associations for AMI or HF and higher mortality with higher cardiologist density for PN (ORs=1.04-1.06).
Conclusion:
After adjusting for patient and HRR characteristics, regional cardiologist density was not associated with 30-day or 1-year mortality for AMI or HF, suggesting that the uneven regional distribution of cardiologists across the United States does not affect patient outcomes.
Association Between Medicare Expenditures and Adverse Events for Patients With Acute Myocardial Infarction, Heart Failure, or Pneumonia in the United States
