The association between polypharmacy and mortality in patients with heart failure: Results from the PULSE dataset.

Aims: Mortality remains high in heart failure despite advances in heart failure therapy. Heart failure patients are generally older, with multiple long-term conditions, and polypharmacy is common. This study explores the association between polypharmacy and mortality.

Methods: This retrospective longitudinal observational cohort study collected medication data on admission and discharge from the first heart failure hospitalisation. Association with mortality was explored using Cox proportional hazard models and inverse probability weighting regression analysis.

Results: A total of 660 patients were included, 367 (56%) male, mean age 76.1 (SD ±12.3) and almost 60% (338/660) had died at study end. Median follow-up time was 2.9 years (25th and 75th quartiles 1.6 and 4.5). It was rare to be discharged from hospital with no polypharmacy (5%, n = 31). Heart failure with preserved ejection fraction (HFpEF) was associated with a 32% (HR 1.32, CI 1.08-1.61, P = 0.007) higher mortality compared to HFrEF. In those with heart failure with reduced ejection fraction (HFrEF), univariable analysis showed hyperpolypharmacy was associated with twice the mortality compared to polypharmacy (HR 1.95, CI 1.36-2.82, P < 0.001). In multivariable analysis, the association between polypharmacy and mortality was lost. The average treatment effect for hyperpolypharmacy was associated with 26% (Coeff. -0.26, CI -0.43 to -0.09, P = 0.003) higher mortality than polypharmacy. The chance of survival to the end of follow-up was 80% (Coeff. 0.80, CI 0.64-0.95, P < 0.01) for those with polypharmacy, and 54% (Coeff. 0.54, CI 0.46-0.61, P < 0.01) for those with hyperpolypharmacy. In HFpEF, hyperpolypharmacy, univariable analysis was not associated with mortality (HR 0.93, CI 0.70-1.24, P = 0.63). Average treatment effect also showed that hyperpolypharmacy was not associated with mortality (Coeff. -0.03, CI -0.15 to 0.08, P = 0.55). The chance of survival to the end of follow-up was 67% (Coeff. 0.67, CI 0.58-0.77, P < 0.01) with polypharmacy and 64% (Coeff. 0.64, CI 0.57-0.71, P < 0.01) with hyperpolypharmacy.

Conclusions: Age, sex, CCI, and CFS are strong mortality predictors for HF irrespective of HF subgroup. Rigorous confounding adjustment suggests polypharmacy is associated with mortality following hospitalisation for HFrEF but not HFpEF. Further studies are needed to address the complex interplay between polypharmacy, age, comorbidity, and frailty.