Health status trajectories using Markov modeling in patients with heart failure with preserved ejection fraction

Health-related quality of life in patients with heart failure (HF) is dynamic. This post hoc study aimed to examine the predictive value of serial Kansas City Cardiomyopathy Questionnaire (KCCQ) assessments for forecasting clinical outcomes in patients with HF with preserved ejection fraction (HFpEF) using a multistate Markov model in the PARAGON-HF study.

A total of 4707 patients were categorized into 5 health states based on KCCQ scores: excellent (75-100), good (50-74), fair (25-49), poor (0-24), and death. Endpoints included health state duration (sojourn time), transition probabilities, associations with mortality, and the impact of HF treatments on health state transitions and sojourn times.

At baseline, 2384 patients were in the excellent state, 1687 in good, 568 in fair, and 68 in poor. Bidirectional transitions were observed, with the longest sojourn time in the excellent state (1.57 years; 95%CI, 1.51-1.63), compared with good (0.70 years; 95%CI, 0.68-0.72), fair (0.72 years; 95%CI, 0.69-0.74), and poor (0.80 years; 95%CI, 0.73-0.87) states. Annual mortality increased progressively with worsening health states: excellent (3.1%), good (4.3%), fair (9.4%), and poor (15.6%). Overall sojourn times did not differ between sacubitril/valsartan and valsartan. However, among patients in the poor health state, sacubitril/valsartan was associated with a higher likelihood of improvement to the fair state (93.1% vs 86.9%) and a lower risk of death (6.9% vs 13.1%).

Health states in HFpEF exhibit bidirectional changes with varying sojourn times. Mortality was higher in poorer health states, and sacubitril/valsartan was associated with more favorable outcomes in these patients.