Predicting Heart Failure Outcomes Using Patient-Reported Health Status

Abstract

Background

The Kansas City Cardiomyopathy Questionnaire-12 (KCCQ-12), a patient-reported outcome measure for adults with heart failure, is associated with hospitalizations and mortality in clinical trials. Curated data sets from controlled trials differ substantially from pragmatic data collected from real-world settings, however, and few data exist on the KCCQ-12's predictive utility in clinical practice.

Objectives

This study sought to evaluate the predictive utility of the KCCQ-12 for hospitalizations and mortality when administered during outpatient heart failure care.

Methods

We conducted a cohort study of patients assigned the KCCQ-12 in heart failure clinics from July 2019 through March 2024. The primary exposure was KCCQ-12 Overall Summary (KCCQ-OS) score. The primary outcomes were 90-day hospitalization and cumulative mortality. Multivariable-adjusted associations were assessed using logistic regression and Cox proportional hazards models. Gradient boosting (XGBoost) and random survival forest machine learning models were used to evaluate KCCQ-OS feature importance in predicting 90-day hospitalizations and cumulative mortality, respectively.

Results

Among 4,406 patients assigned the KCCQ-12, 2,888 (66%) completed at least 1 questionnaire. The median KCCQ-OS score was 59.4 (Q1-Q3: 35.4-81.8). Patients with KCCQ-OS scores <25 had higher adjusted risks of 90-day hospitalization (OR: 3.49; 95% CI: 2.50-4.90) and cumulative mortality (HR: 3.09; 95% CI: 2.29-4.17) compared with those with scores ≥75. The KCCQ-OS score was the most important feature for predicting 90-day hospitalizations in the XGBoost model (area under the receiver-operating characteristic curve: 0.760; 95% CI: 0.706-0.811) and the most important feature for predicting cumulative mortality in the random survival forest model (C-index 0.783; 95% CI: 0.742-0.824) compared with other clinical, demographic, and laboratory variables. KCCQ-12 noncompletion was independently associated with increased 90-day hospitalization (OR: 1.72; 95% CI: 1.46-2.02) and 1-year mortality (HR: 1.52; 95% CI: 1.25-1.84) after adjusting for all variables in the primary analysis.

Conclusions

In outpatient heart failure care, lower KCCQ-OS scores were strongly associated with increased hospitalizations and mortality, with the greatest risk among patients with scores <25. Noncompletion of the KCCQ-12 was itself associated with worse outcomes. The KCCQ-OS score was the dominant predictor of 90-day hospitalizations and cumulative mortality in machine learning models, supporting the KCCQ-12 as a prognostic tool in routine clinical practice.