Ten clinically important subgroups of myocardial infarction survivors: Identification by latent class analysis

BACKGROUND: Given the substantial heterogeneity among patients with myocardial infarction (MI), stratification of this population may support improvements in clinical management and patient outcomes. Therefore, we aimed to identify homogenous subgroups among MI survivors using latent class analysis (LCA) and, secondarily, to demonstrate the clinical relevance of the subgroups by comparing their prognoses measured by risk of death and recurrent MI (RMI).

METHODS: Patients surviving a hospitalization with a first-time MI from 2016 to 2018 were identified from nationwide Danish registries. We included 30 patient characteristics in the LCA and selected the best model based on statistical fit, clinical relevance, and reproducibility across data samples. The five-year mortality and risk of RMI were quantified by cumulative incidence functions and proportional hazard models.

RESULTS: In total, 17,018 patients were included, and 10 latent classes were identified. The classes expressed distinct patterns of characteristics, but clusters with similarities were identified: three classes were characterized by high probabilities of ST-elevation MI, two classes by other MI-specific characteristics, and five classes by comorbidity and/or healthcare utilization patterns. The five-year cumulative incidences ranged from 3.2 % to 60.5 % (death) and 4.4 % to 13.5 % (RMI). Using the class with the best prognosis as reference group, the age- and sex-adjusted hazard ratios ranged from 1.67 to 12.94 (death) and 1.28 to 4.39 (RMI).

CONCLUSION: Ten clinically distinct subgroups were identified in a population of first-time MI survivors by using LCA. The subgroups showed markedly differing prognoses in terms of five-year mortality and risk of RMI.