Back

HALP and mHALP as Effective Tools for 90-Day Mortality Prediction in Heart Failure

Hourani, A. Z. O.; Sürmeli, A. D.; Devarapalli, S. K.; Oreziak, M.

2025-05-22 cardiovascular medicine
10.1101/2025.05.22.25328139 medRxiv
Show abstract

BackgroundCongestive heart failure (CHF) is a chronic syndrome with high mortality rates, necessitating risk stratification tools. Existing models are complex and may not incorporate crucial inflammatory and nutritional biomarkers. The Hemoglobin, Albumin, Lymphocyte, and Platelet (HALP) score and modified HALP (mHALP) score, which reflect nutritional status and systemic inflammation, emerged as potential prognostic tools. This study aimed to evaluate the predictive performance of HALP and mHALP scores for 90-day mortality in CHF patients. MethodsThis retrospective study utilized the MIMIC-IV database for 1000 adult CHF patients. HALP was calculated as [Hemoglobin (g/L) * Albumin (g/L) * Lymphocytes (/L)] / Platelets (/L), and mHALP by multiplying these components. The primary outcome was 90-day all-cause mortality. Univariate and multivariate logistic regression analyses were performed, and models were assessed using Receiver Operating Characteristic (ROC) analysis and Youdens Index. ResultsNon-survivors (13.2%) had significantly lower median HALP (19.69 vs. 27.18) and mHALP (583,920.75 vs. 1,151,898.94) scores compared to survivors. In univariate analysis, HALP had an AUC of 0.63 and an OR of 0.25 for 90-day mortality, while mHALP showed a higher AUC of 0.68 and an OR of 0.22. After confounder adjusting, both log-transformed HALP (OR: 0.44) and mHALP (OR: 0.57) remained significant predictors of mortality. mHALP demonstrated better overall predictive performance. ConclusionBoth HALP and mHALP scores are significantly associated with 90-day mortality in CHF patients, with mHALP exhibiting superior prediction. Easily calculable scores, derived from routine laboratory tests, offer a cost-effective, early risk stratification in CHF, potentially aiding in tailoring interventions.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.