Mortality trends for 23,505 Medicare patients hospitalized with heart failure in Northeast Ohio, 1991 to 1997

Background Clinical trials have identified major therapeutic advances for heart failure (HF), but the degree to which survival has improved among the general population of patients with HF is not known. This study analyzed mortality trends from 1991 to 1997 for 23,505 Medicare patients hospitalized with a first admission for HF at 29 Northeast Ohio hospitals. Methods We linked databases from the Cleveland Health Quality Choice (CHQC) program and Medicare to allow identification of first admissions,for HF and death, date. We adjusted for changes in admission illness severity using chart data from CHQC (eg, vital signs, do-not-resuscitate status, comorbid conditions, and laboratory results). Logistic regression was used to analyze trends in risk-adjusted mortality. Results At baseline (1991), crude inhospital, 30-day and 1-year mortality rates were 6.4%, 8.6% and 36.5%, respectively. Between 1991 and 1997, mean length of stay declined steeply from 9.2 days to 6.6 days (P <.001 for trend). Risk-adjusted inhospital mortality also declined markedly (absolute-decline -3.7%, 95% Cl -4.3 to -3.0), a 52.8% relative decrease. However, the decline in 30-day mortality was only - 1.4% (95% Cl -2.5 to -0.1, P <.05), a 15.3% relative decrease. The 1-year mortality declined -5.3% (95% Cl -3.2 to -7.4, P <.001), a 14.6% relative decrease. Conclusions Long-term mortality for patients hospitalized with HF improved from 1991 to 1997, although mortality remains very high. The 30-day mortality declined far less than inhospital mortality, indicating that mortality shortly after discharge increased. This raises concerns that the marked reduction in length of stay is causing adverse consequences.