Discrepancy between results of randomized control studies and retrospective analysis. The case of statin therapy effect on one-year mortality in patients with decompensated heart failure

Background: In view of the recent reports on a lack of protective effect of statin therapy on clinical outcomes in patients with heart failure, the present study investigated the impact of statin therapy before admission on one-year survival of patients hospitalized due to decompensated heart failure in two groups of patients; with and without ischemic heart disease. Methods: We performed a retrospective cohort analysis of 887 consecutive patients older than 18 years hospitalized with decompensated heart failure between 11/December 2001 and 06/June 2005. Two groups of patients were compared: those who received statins within 3 months before the admission (S) and those who did not (NS). The primary outcome was one-year all-cause mortality. To adjust for a potential misbalance between S and NS groups in baseline characteristics, a propensity score for statin therapy was incorporated into the survival model. Results: Two hundred eighty-one patients (31.7%) received statins prior to admission. Patients with ischemic heart disease (IHD) (656/887 subjects, 74%) had higher rate of S therapy as compared to the rest 36.3% vs. 18.6%, p < 0.001. Overall one-year mortality rate in the S group was 21% vs. 31.8% in the NS group, p < 0.001. In the subgroup of patients with IHD, statins were protective after adjustment for comorbidities and propensity score (hazard ratio [HR], 0.66; 95% CI 0.4-0.97), but in patients with non-ischemic HF statin therapy was not associated with a protective effect (HR 0.77; 95% CI 0.38-1.6). Conclusions: In our study statins' protective effect on one-year survival in HF patients is restricted to patients with IHD. This stands in disagreement with the results of the randomized trials showing no effect of statin therapy. Statin use may be a marker of better health care and despite an extensive adjustment for baseline characteristics, unaccounted bias inherent to retrospective studies may explain the discrepancy in the results.