Prognostic models for all‐cause and cardiovascular mortality in type 2 diabetes: Systematic review

AIMS: The elevated risk of all-cause and cardiovascular mortality in individuals with type 2 diabetes mellitus (T2DM) has led to growing efforts to develop prognostic models for early identification of high-risk individuals. This systematic review synthesised existing models to inform future model development, enhance predictive performance and guide targeted prevention strategies in diverse clinical and population health settings. METHODS: We systematically searched Ovid MEDLINE, Scopus and Web of Science for studies published between January 1, 2015, and June 11, 2024, reporting prognostic models developed and/or validated to predict all-cause or cardiovascular mortality in individuals T2DM. Data were extracted following the CHARMS checklist, and risk of bias assessed using the PROBAST tool. RESULTS: The search yielded 18,126 records; 10,921 were screened after deduplication. Of 147 full texts assessed, 26 cohort studies met inclusion criteria, with sample sizes (median [IQR]: 20,554 [1931-59,180]). Models were developed in diverse regions, with the highest number from Taiwan (n = 5) and the USA (n = 5). Most studies focused on all-cause mortality (n = 26); eight addressed cardiovascular mortality. Prediction horizons varied from 1 to 15 years, with 5-year risk being the most common (n = 10). Discrimination had a median C-statistic of 0.77 (IQR: 0.72-0.81). Calibration was reported in 20 studies, though methods varied. Cox regression was the most common statistical method (n = 16). CONCLUSIONS: Prediction models for mortality in T2DM show considerable heterogeneity in methodology, performance and validation. Limited external validation and inconsistent calibration reporting highlight the need for robust, generalisable and transparently reported models to improve clinical risk stratification in diabetes care.