Satoshi Katano, Toshiyuki Yano, Yuji Kono, Akinori Sawamura, Koshiro Kanaoka, Tetsufumi Motokawa, Yoshihiro Miyamoto, Yusuke Ohya, Shin-Ichiro Miura, Nagaharu Fukuma, Shigeru Makita, Yoshihiro Fukumoto, Hideo Izawa
CR was associated with a lower in-hospital mortality rate across all timing quartiles. The association pattern appeared to differ by outcome domain: Functional measures showed more favorable associations with earlier initiation (within 2-3 days), whereas a lower mortality rate was observed across a broader window. These findings warrant prospective evaluation.
BACKGROUND: Cardiac rehabilitation (CR) improves outcomes in heart failure, but optimal initiation timing during acute hospitalization remains undefined. We characterized the dose-response association between door-to-cardiac rehabilitation time and in-hospital outcomes, describing how this pattern appeared to differ between death and functional measures.
METHODS: Using the Japanese Registry of All Cardiac and Vascular Diseases-Diagnosis Procedure database, we identified 564 830 patients hospitalized for acute heart failure (2012-2023), categorized into non-CR (n=275 454) and CR groups by door-to-cardiac rehabilitation quartiles: quartile 1 (≤1 day), quartile 2 (2-3 days), quartile 3 (4-5 days), and quartile 4 (≥6 days). Generalized propensity score-based weighting adjusted for 32 confounders. Restricted cubic splines evaluated dose-response patterns among CR recipients. The primary outcome was in-hospital death; secondary outcomes were Barthel Index at discharge and Barthel Index improvement.
RESULTS: All door-to-cardiac rehabilitation quartiles showed lower in-hospital death versus non-CR (adjusted odds ratios, 0.39-0.49; all P<0.001). Barthel Index at discharge was higher in quartile 1 to quartile 3 (mean difference, +3.2 to +3.8 points), with a smaller difference in quartile 4. Spline analysis revealed nonlinear associations: The mortality nadir was observed at 6.1 days, whereas functional measures showed the most favorable associations at 3.3 days. Sensitivity analyses addressing immortal time bias yielded consistent mortality results.
CONCLUSIONS: CR was associated with a lower in-hospital mortality rate across all timing quartiles. The association pattern appeared to differ by outcome domain: Functional measures showed more favorable associations with earlier initiation (within 2-3 days), whereas a lower mortality rate was observed across a broader window. These findings warrant prospective evaluation.