Risk prediction for heart failure incidence within 1-year using clinical and laboratory factors

Area of Science:

• Cardiology
• Medical Informatics
• Biostatistics

Background:

• Validated risk scores for short-term heart failure (HF) incidence are limited.
• Accurate prediction models are crucial for timely intervention and patient management.

Purpose of the Study:

• To develop and validate a 1-year HF incidence risk prediction model.
• To assess the added value of laboratory variables to clinical risk factors for HF prediction.

Main Methods:

• Utilized the MIMIC II clinical database.
• Developed two multivariable Cox models: one with clinical factors, another including laboratory parameters (serum creatinine, BUN, glucose, PT, APTT, TBIL).
• Internally validated model performance using bootstrapping.

Main Results:

• Identified pulmonary circulation diseases, peripheral vascular disease, chronic pulmonary disease, hypothyroidism, electrolyte/fluid disorders, BUN, and APTT as independent predictors of HF incidence.
• Found hypertension has an inverse association with short-term HF risk.
• The combined model achieved a C-statistic of 0.712 with internal validation, demonstrating effectiveness.

Conclusions:

• A prediction model incorporating both clinical and laboratory factors enhances 1-year HF risk assessment.
• Several clinical and laboratory indices, beyond traditional factors, are significant predictors of short-term HF.
• Hypertension's complex role in short-term HF risk warrants further investigation.