Patterns of multimorbidity, their outcomes and analytical methods for their determination: a systematic review

BACKGROUND: Multimorbidity is becoming a pressing public health problem, as more than 50% of the world's adult population over the age of 60 is multimorbid. In order to better stratify risk in patients pattern identification is a promising approach. We conducted this systematic review to inform future research on multimorbidity patterns, their related outcomes and analytical methods used to derive them. METHODS: This systematic review followed PRISMA 2020 guidelines and was registered in PROSPERO registry (CRD420261290771). Bibliographic databases PubMed, Ovid MEDLINE(R), and Web of Science were systematically searched. Original observational studies examining multimorbidity patterns and associated health outcomes were included. Data was independently extracted by two reviewers and synthesised qualitatively. Methodological quality was assessed using a modified Newcastle-Ottawa Scale. RESULTS: We examined a total of 7897 articles based on title and abstract. This systematic review includes an analysis of 30 studies and provides an update on the current state of multimorbidity patterns identification, their outcomes and the analytical methods used to derive them. We found that across all included studies the most commonly identified and stable patterns are those which include a cardiovascular, metabolic, respiratory or neurological/neuropsychiatric component. The cardiometabolic pattern showed the least heterogeneity in its composition. The most commonly measured outcome between patterns and outcomes was mortality. Higher all-cause mortality was observed in the majority of studies in patterns including a cardiovascular, metabolic or neurological/neuropsychiatric component. We found that identifying multimorbidity patterns using electronic health records and questionnaires are the most prevalent methods of data gathering. The most commonly used methodology in this field was latent class analysis followed by other methods such as, fuzzy c-means, k-means clustering, and factor analysis. CONCLUSION: Across the included studies, the most consistently identified and stable multimorbidity patterns were those comprising cardiovascular, metabolic, respiratory, and neurological/neuropsychiatric components. Patterns that included cardiovascular, metabolic, or neuropsychiatric conditions were most frequently associated with increased all-cause mortality. These findings support the clinical relevance of pattern-based approaches to multimorbidity. Future research should prioritise methodological standardisation, transparent reporting, and systematic validation of identified patterns to improve comparability and facilitate their translation into clinical practice and public health planning.