Generalizability and Longitudinal Outcomes of a National Heart Failure Clinical Registry: Comparison of Acute Decompensated Heart Failure National Registry (ADHERE) and Non-ADHERE Medicare Beneficiaries

Study Questions:

Can clinical registry data be generalized to the general population of interest?


Elderly patients enrolled into the Acute Decompensated Heart Failure National Registry (ADHERE) (patients ages ≥65 years with discharge diagnosis of heart failure between 2001-2006) were matched to records to fee-for-service Medicare claims data, forming two groups of patients: those ADHERE patients linked to Medicare claims and those not linked. ADHERE patients who were Medicare linked were also compared to a group of Medicare patients not enrolled into ADHERE. The primary outcome of interest was the cumulative incidence of mortality, and readmission within each group was assessed via Kaplan-Meier analysis.


There were 104,808 admissions (82,074 elderly patients) in the ADHERE registry matched to Medicare claims. Mortality in ADHERE patients linked (4.5%, n = 4,704) and not linked (4.5%, n = 1,389) to Medicare was similar (p = 0.92). In the Medicare groups, patients enrolled into ADHERE had statistically lower unadjusted mortality (4.4% in-hospital and 36% 1-year) than those not enrolled (4.9% in-hospital and 38% 1-year, p < 0.001). Likewise, readmissions were statistically fewer in the ADHERE Medicare patients (42.6% at 1 year vs. 43.4%, p < 0.001). ADHERE hospitals were more likely to be teaching hospitals with higher heart failure volumes.


The authors concluded that elderly patients in the ADHERE registry are similar to Medicare beneficiaries hospitalized with heart failure.


Extrapolating the results from this registry analysis to that of others, this study suggests that large patient registries in heart failure can accurately reflect those patients cared for in the general population. This appears to be true despite differences in the types of hospitals caring for patients in the ADHERE versus non-ADHERE groups. The large patient numbers yielded highly statistically significant p values when assessing mortality and readmissions, but these are unlikely to be clinically relevant. Overall, this was a relatively ill group, with a mortality of 43% at 1 year.

< Back to Listings