Integration of Patient Reported Quality-of-life Data into Risk Assessment in Heart Failure.

Area of Science:

• Cardiology
• Health Informatics
• Clinical Prediction Modeling

Background:

• Optimal management of heart failure (HF) necessitates accurate, serial risk assessment for adverse outcomes.
• Patient-reported outcomes (PROs) are increasingly integrated into clinical practice.
• The utility of PROs in refining risk prediction models for ambulatory HF patients remains an area of investigation.

Purpose of the Study:

• To evaluate whether incorporating PROs into risk prediction models improves accuracy for ambulatory patients with HF.
• To assess the impact of PROs on predicting HF hospitalization and mortality.

Main Methods:

• Utilized Cox regression with LASSO regularization and gradient boosting machine analyses.
• Included consecutive patients with HF with reduced ejection fraction (HFrEF) and HF with preserved ejection fraction (HFpEF) from a HF clinic (2015-2019).
• Evaluated model performance using the time-dependent concordance index (Cτ).

Main Results:

• Gradient boosting models incorporating PROs demonstrated superior prediction performance (Cτ = 0.73 for HFrEF, 0.74 for HFpEF).
• Models showed excellent risk stratification across patient quintiles.
• Key PROs influencing prediction included the Kansas City Cardiomyopathy Questionnaire, visual analogue scale, and Patient Reported Outcomes Measurement Information System dimensions.

Conclusions:

• PROs significantly enhance risk prediction for both HFrEF and HFpEF patients, independent of traditional clinical factors.
• Routine assessment of PROs, alongside electronic health record data, can lead to more accurate risk stratification.
• Improved risk assessment supports timely and intensified treatment strategies for patients with HF.