Risk prediction models for mortality in patients with multimorbidity: a systematic review and meta-analysis

Area of Science:

• Gerontology
• Public Health
• Epidemiology

Background:

• Multimorbidity presents a substantial global aging and public health challenge.
• Existing risk prediction models for mortality in multimorbidity patients vary in quality and clinical applicability.
• Uncertainty surrounds the reliability of current models for practice and future research.

Purpose of the Study:

• To systematically review and assess the quality of published risk prediction models for mortality in patients with multimorbidity.
• To evaluate the risk of bias and applicability of these models using a standardized tool.

Main Methods:

• A comprehensive search of multiple databases (PubMed, Embase, Web of Science, etc.) was conducted up to May 30, 2024.
• Two independent reviewers performed study selection, data extraction, and quality assessment using the Prediction Model Risk of Bias Assessment Tool (PROBAST).

Main Results:

• Eighteen studies comprising 21 prediction models were included.
• Logistic regression was the most common modeling technique; age and BMI were frequent predictors.
• While pooled AUC was 0.81 (fair discrimination), three studies had low risk of bias, and 11 had high risk, mainly due to reporting issues.

Conclusions:

• Included models demonstrated some discriminatory ability for mortality prediction in multimorbidity.
• However, all models exhibited significant risks of bias, impacting their reliability.
• Future research must prioritize rigorous designs and multicenter external validation to enhance prediction precision and inform global health strategies.