What is the clinical evidence from this study?

Study design: Systematic Review. Population: Patients with acute myocardial infarction (AMI) at risk for cardiac rupture (n=11603). Intervention: Risk prediction models for cardiac rupture after AMI vs. Not applicable. Primary outcome: Discriminative performance of cardiac rupture prediction models measured by C-statistic (AUC) (Pooled C-statistic 0.83 (95% CI 0.78–0.89), 95% CI 0.78–0.89).

What does this research mean for the field?

Risk prediction models for cardiac rupture after acute myocardial infarction showed good discriminatory ability with a pooled C-statistic of 0.83 (95% CI: 0.78–0.89), though they are undermined by substantial methodological limitations. Novelty: ClaimNovelty.SYNTHESIS. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

This review aims to evaluate and synthesize current risk prediction models for cardiac rupture following acute myocardial infarction.

February 14, 2026Open Access

Risk prediction models for cardiac rupture after acute myocardial infarction: a systematic review and meta-analysis

Key Result

Risk prediction models for cardiac rupture after acute myocardial infarction showed good discriminatory ability with a pooled C-statistic of 0.83 (95% CI: 0.78–0.89).

Key Points

This review aims to evaluate and synthesize current risk prediction models for cardiac rupture following acute myocardial infarction.
Systematic search across multiple databases for relevant studies from inception to August 2025.
Data extraction according to the CHARMS checklist and quality assessment using PROBAST.
Meta-analysis conducted to pool C-statistics and evaluate predictors of cardiac rupture.
Included ten studies with a total of 74–11,603 participants.
Pooled C-statistic for prediction models was 0.83, indicating good discrimination ability but high heterogeneity.
Identified robust predictors: advanced age, female sex, higher Killip grade, elevated heart rate, lower LVEF, and absence of emergency PCI.

Study Design

Type

Systematic Review (n=11,603)

Structured PICO

Population

10 studies pooling 54,991 patients with acute myocardial infarction (AMI) or ST-segment elevation myocardial infarction (STEMI) at risk of cardiac rupture.

Intervention

Risk prediction models for cardiac rupture post-AMI

Outcome

Model discrimination (C-statistic/AUC) for predicting cardiac rupture

Existing models demonstrate promising discriminatory ability (pooled C-statistic 0.83) for predicting cardiac rupture after AMI but are undermined by substantial methodological limitations.

Main Result

Effect estimate: Pooled C-statistic 0.83 (95% CI 0.78–0.89) (95% CI 0.78–0.89)

Limitations

Substantial methodological limitations including low events-per-variable ratios (1.0 to 9.2 vs recommended ≥10) indicating risk of overfitting.
High heterogeneity among included studies (I2 = 88%).
Most studies used univariate predictor selection, increasing risk of missing relevant factors or including spurious ones.
Inadequate handling or reporting of missing data in most studies.
Limited or absent external validation in several studies.
Lack of blinding in outcome assessment.
Insufficient reporting of timing between predictor assessment and cardiac rupture occurrence.

Abstract

Background Cardiac rupture (CR) is a catastrophic complication of acute myocardial infarction (AMI), accounting for 10%–20% of AMI-related deaths despite its low incidence. Several risk prediction models have been developed, but their methodological quality and clinical applicability remain uncertain. This study systematically reviewed and quantitatively synthesized existing prediction models for CR post-AMI. Methods We searched PubMed, Embase, Web of Science, Cochrane Library, CNKI, and Wanfang databases from inception to August 2025. Studies developing or validating risk prediction models for CR after AMI were eligible. Data extraction followed the CHARMS checklist, and methodological quality was assessed with PROBAST. A meta-analysis was performed to pool model discrimination (C-statistic) and evaluate predictors of CR. Subgroup analyses explored heterogeneity by publication period, study design, population, sample size, and validation approach. Results Ten studies (2017–2024) involving 74–11,603 patients were included. Among them, nine studies reported C-statistics (AUC) along with their confidence intervals (CIs), which were suitable for quantitative synthesis (Meta-analysis). The pooled C-statistic of CR prediction models was 0.83 (95% CI: 0.78–0.89), though with high heterogeneity ( I 2 = 88%). Consistently robust predictors included advanced age (OR = 2.26), female sex (OR = 2.43), higher Killip grade (OR = 3.58), elevated heart rate (OR = 2.29), lower LVEF (OR = 1.46), and absence of emergency PCI (OR = 0.37, protective). Most studies exhibited methodological flaws, including small events-per-variable ratios, univariate-based predictor selection, inadequate handling of missing data, and limited external validation. Conclusion Existing models demonstrate promising discriminatory ability for predicting CR after AMI but are undermined by substantial methodological limitations. Age, sex, Killip grade, LVEF, and PCI status represent robust predictors that should inform future consensus-based models. Large-scale, prospective, and externally validated studies are urgently needed to develop reliable tools for clinical risk stratification and targeted prevention of this lethal complication. Systematic Review Registration https://www.crd.york.ac.uk/PROSPERO/view/CRD420251105703 , PROSPERO CRD420251105703.

Mark Helpful

Bookmark

Relay

View Full Paper