Comparative analysis of methods for identifying multimorbidity patterns: a study of 'real-world' data

Area of Science:

• Epidemiology
• Health Services Research
• Biostatistics

Background:

• Multimorbidity, the presence of multiple chronic conditions, is a growing public health concern.
• Identifying patterns of co-occurring diseases is crucial for effective clinical management and resource allocation.
• Existing methods for pattern identification, such as hierarchical cluster analysis (HCA) and exploratory factor analysis (EFA), have not been directly compared in large primary care datasets.

Purpose of the Study:

• To compare multimorbidity patterns identified by HCA and EFA in a large primary care database.
• To determine if the choice of analytical method influences the composition of identified multimorbidity patterns.
• To explore the potential clinical applications of HCA and EFA in understanding disease co-occurrence.

Main Methods:

• A cross-sectional study utilizing electronic health records from 408,994 patients aged 45-64 with multimorbidity.
• Diagnoses were coded using the International Classification of Diseases, version 10.
• Multimorbidity patterns were identified using both HCA and EFA, with analyses stratified by sex.

Main Results:

• HCA identified a larger number of clusters (53 for women, 15 for men) compared to EFA (9 factors for women, 10 for men).
• Significant differences in pattern composition were observed between HCA and EFA, and also by sex.
• Consistent patterns across methods and sexes included hypertension with obesity, spondylopathies with deforming dorsopathies, and dermatitis eczema with mycosis.

Conclusions:

• The method used significantly impacts the identification of multimorbidity patterns, highlighting limitations in comparing findings across studies.
• EFA is valuable for characterizing comorbidity relationships, whereas HCA is better suited for detailed multimorbidity investigations.
• Findings provide evidence for the distinct utility of HCA and EFA in clinical and research settings and inform future study standardization.