Yasuyuki Shiraishi1, Shun Kohsaka2, Toshiyuki Nagai3, Ayumi Goda4, Atsushi Mizuno5, Yuji Nagatomo6, Yasumori Sujino7, Ryoma Fukuoka1, Mitsuaki Sawano1, Takashi Kohno1, Keiichi Fukuda1, Toshihisa Anzai3, Ramin Shadman8, Todd Dardas9, Wayne C Levy9, Tsutomu Yoshikawa6. 1. Department of Cardiology, Keio University School of Medicine, Tokyo, Japan. 2. Department of Cardiology, Keio University School of Medicine, Tokyo, Japan. Electronic address: sk@keio.jp. 3. Department of Cardiovascular Medicine, National Cerebral and Cardiovascular Center, Osaka, Japan; Department of Cardiovascular Medicine, Hokkaido University Graduate School of Medicine, Hokkaido, Japan. 4. Division of Cardiology, Kyorin University School of Medicine, Tokyo, Japan. 5. Department of Cardiology, St. Luke's International Hospital, Tokyo, Japan. 6. Department of Cardiology, Sakakibara Heart Institute, Tokyo, Japan. 7. Department of Cardiology, Saitama Medical University International Medical Center, Saitama, Japan. 8. Southern California Permanente Medical Group, Los Angeles, California. 9. Division of Cardiology, Department of Internal Medicine, University of Washington, Seattle, Washington.
Abstract
BACKGROUND: Precise risk stratification in heart failure (HF) patients enables clinicians to tailor the intensity of their management. The Seattle Heart Failure Model (SHFM), which uses conventional clinical variables for its prediction, is widely used. We aimed to externally validate SHFM in Japanese HF patients with a recent episode of acute decompensation requiring hospital admission. METHODS AND RESULTS: SHFM was applied to 2470 HF patients registered in the West Tokyo Heart Failure and National Cerebral And Cardiovascular Center Acute Decompensated Heart Failure databases from 2006 to 2016. Discrimination and calibration were assessed with the use of the c-statistic and calibration plots, respectively, in HF patients with reduced ejection fraction (HFrEF; <40%) and preserved ejection fraction (HFpEF; ≥40%). In a perfectly calibrated model, the slope and intercept would be 1.0 and 0.0, respectively. The method of intercept recalibration was used to update the model. The registered patients (mean age 74 ± 13 y) were predominantly men (62%). Overall, 572 patients (23.2%) died during a mean follow-up of 2.1 years. Among HFrEF patients, SHFM showed good discrimination (c-statistic = 0.75) but miscalibration, tending to overestimate 1-year survival (slope = 0.78; intercept = -0.22). Among HFpEF patients, SHFM showed modest discrimination (c-statistic = 0.69) and calibration, tending to underestimate 1-year survival (slope = 1.18; intercept = 0.16). Intercept recalibration (replacing the baseline survival function) successfully updated the model for HFrEF (slope = 1.03; intercept = -0.04) but not for HFpEF patients. CONCLUSIONS: In Japanese acute HF patients, SHFM showed adequate performance after recalibration among HFrEF patients. Using prediction models to tailor the care for HF patients may improve the allocation of medical resources.
BACKGROUND: Precise risk stratification in heart failure (HF) patients enables clinicians to tailor the intensity of their management. The Seattle Heart Failure Model (SHFM), which uses conventional clinical variables for its prediction, is widely used. We aimed to externally validate SHFM in Japanese HF patients with a recent episode of acute decompensation requiring hospital admission. METHODS AND RESULTS: SHFM was applied to 2470 HF patients registered in the West Tokyo Heart Failure and National Cerebral And Cardiovascular Center Acute Decompensated Heart Failure databases from 2006 to 2016. Discrimination and calibration were assessed with the use of the c-statistic and calibration plots, respectively, in HF patients with reduced ejection fraction (HFrEF; <40%) and preserved ejection fraction (HFpEF; ≥40%). In a perfectly calibrated model, the slope and intercept would be 1.0 and 0.0, respectively. The method of intercept recalibration was used to update the model. The registered patients (mean age 74 ± 13 y) were predominantly men (62%). Overall, 572 patients (23.2%) died during a mean follow-up of 2.1 years. Among HFrEF patients, SHFM showed good discrimination (c-statistic = 0.75) but miscalibration, tending to overestimate 1-year survival (slope = 0.78; intercept = -0.22). Among HFpEF patients, SHFM showed modest discrimination (c-statistic = 0.69) and calibration, tending to underestimate 1-year survival (slope = 1.18; intercept = 0.16). Intercept recalibration (replacing the baseline survival function) successfully updated the model for HFrEF (slope = 1.03; intercept = -0.04) but not for HFpEF patients. CONCLUSIONS: In Japanese acute HF patients, SHFM showed adequate performance after recalibration among HFrEF patients. Using prediction models to tailor the care for HF patients may improve the allocation of medical resources.
Authors: Scott C Silvestry; Claudius Mahr; Mark S Slaughter; Wayne C Levy; Richard K Cheng; Damian M May; Eleni Ismyrloglou; Stelios I Tsintzos; Edward Tuttle; Keziah Cook; Erica Birk; Aparna Gomes; Sophia Graham; William G Cotts Journal: ASAIO J Date: 2020-08 Impact factor: 2.872