Background: Heart failure presents a growing clinical and economic burden in the USA. Robust cost data on the burden of illness are critical to inform economic evaluations of new therapeutic interventions.
Objectives: This systematic literature review of heart failure-related costs in the USA aimed to assess the quality of the published evidence and provide a narrative synthesis of current data.
Methods: Four electronic databases (MEDLINE, EMBASE, EconLit, and the Centre for Reviews and Dissemination York Database, including the NHS Economic Evaluation Database and Health Technology Assessment Database) were searched for journal articles published between January 2014 and March 2020. The review, registered with PROSPERO (CRD42019134201), was restricted to cost-of-illness studies in adults with heart failure events in the USA.
Results: Eighty-seven studies were included, 41 of which allowed a comparison of cost estimates across studies. The annual median total medical costs for heart failure care were estimated at $24,383 per patient, with heart failure-specific hospitalizations driving costs (median $15,879 per patient). Analyses of subgroups revealed that heart failure-related costs are highly sensitive to individual patient characteristics (such as the presence of comorbidities and age) with large variations even within a subgroup. Additionally, differences in study design and a lack of standardized reporting limited the ability to compare cost estimates. The finding that costs are higher for patients with heart failure with reduced ejection fraction compared with patients with preserved ejection fraction highlights the need for differentiating among different heart failure types.
Conclusions: The review underpins the conclusion drawn in earlier reviews, namely that hospitalization costs are the key driver of heart failure-related costs. Analyses of subgroups provide a clearer understanding of sources of heterogeneity in cost data. While current cost estimates provide useful indications of economic burden, understanding the nuances of the data is critical to support its application.