# Systematic review and meta - analysis of risk prediction models for heart failure after PCI in patients with acute myocardial infarction

## TL;DR

This study reviews and evaluates heart failure risk prediction models for patients with heart attacks after a common heart procedure, finding strong predictive power but high bias risks.

## Contribution

The paper provides a systematic review and meta-analysis of heart failure risk models post-PCI in AMI patients, identifying key predictors and methodological shortcomings.

## Key findings

- 14 studies with 14 models were analyzed, showing strong predictive ability with AUCs between 0.847 and 0.966.
- 19 effective predictors of heart failure were identified, including age, biomarkers, and clinical indicators.
- All studies had high bias risk due to small sample sizes and methodological issues like overreliance on univariate analysis.

## Abstract

The incidence of heart failure (HF) following percutaneous coronary intervention (PCI) in patients with acute myocardial infarction (AMI) remains relatively high, severely impairing long-term prognosis and quality of life. In recent years, advances in biomarker identification and imaging technologies have driven growing research into developing HF risk prediction models for AMI patients post-PCI. However, significant heterogeneity exists across current studies in terms of model construction methods, variable selection, and validation strategies, and the predictive performance and clinical utility of these models have not been systematically evaluated.

To systematically review published studies on risk prediction models for HF after PCI in AMI patients.

Systematic review and meta-analysis.

Databases including PubMed, Web of Science, Embase, Cochrane Library, CNKI, VIP, Wanfang Data Knowledge Service Platform, Chinese Journal Full-text Database, and Chinese Biomedical Literature Database were searched from inception to December 31, 2024. Eligibility criteria included: study participants aged ≥ 18 years with AMI post-PCI; studies focused on HF risk prediction model development; and publication in Chinese or English. The Prediction Model Risk of Bias Assessment Tool (PROBAST) was used to evaluate bias risk in model construction and validation, and RevMan 5.4 software was employed for meta-analysis.

A total of 14 studies involving 14 HF risk prediction models for AMI patients post-PCI were included. All studies adopted a retrospective design, with HF incidence ranging from 3.0% to 37.5%. The area under the curve (AUC) for model discriminative ability ranged from 0.847 to 0.966, indicating strong predictive power. Meta-analysis identified 19 effective predictors of post-PCI HF in AMI patients (all P < 0.05): Killip classification, age, Gensini score, D-dimer level, N-terminal pro-brain natriuretic peptide (NT-proBNP), neutrophil-to-high-density lipoprotein cholesterol ratio (NHR), neutrophil-to-lymphocyte ratio (NLR), modified shock index (MSI), troponin I, cardiac troponin T, serum creatinine, high-sensitivity C-reactive protein (hs-CRP), ventricular wall motion amplitude, hypertension, time from symptom onset to hospital admission, number of diseased coronary vessels, diabetes mellitus, arrhythmia, and cardiac structural changes. However, all studies were rated as high overall bias risk via PROBAST, primarily due to insufficient sample size, improper handling of continuous variables, and overreliance on univariate analysis for variable selection.

While existing HF risk prediction models for AMI patients post-PCI demonstrate favorable discriminative ability, they suffer from high overall bias risk. Future studies should adhere strictly to PROBAST guidelines, optimize study design, improve methodological quality, and conduct multicenter prospective studies to validate model stability and generalizability. This will facilitate the development of clinically robust, applicable, and evidence-based prediction tools for clinical practice.

CRD42025639512

The online version contains supplementary material available at 10.1186/s12872-025-05406-z.