Abstract
In patients with chronic heart failure (CHF), exercise therapy is generally recommended in addition to pharmacotherapy such as ACE inhibitors and beta-blockers. However, our previous model-based meta-analysis showed that the reduction in mortality with pharmacotherapy requires a reduction in cardiac load prior to an improvement in cardiac function; thus increased workload caused by exercise therapy may be conflicting depending on patients. In the present study, using individual patient data from the HF-ACTION trial, we explored the influence of pharmacotherapy on the effect of exercise therapy with a Cox proportional hazards model focusing on interactions. The results showed that the exercise effect on all-cause mortality varied greatly depending on the pharmacotherapy at randomization, with exercise therapy significantly decreasing the number of deaths only in patients on a beta-blocker. In addition, various other factors were also identified as influencing the exercise effect, and clinical scores generated using these factors were useful in discriminating patients whose prognosis was improved or worsened by exercise. These results could compel a review of the original primary analysis of the HF-ACTION study, which found that exercise therapy had little effect on the prognosis of patients with CHF.
