Multimorbidity and comorbidity of chronic diseases among the senior Australians: prevalence and patterns

Abstract

Understanding patterns and identifying common clusters of chronic diseases may help policymakers, researchers, and clinicians to understand the needs of the care process better and potentially save both provider and patient time and cost. However, only limited research has been conducted in this area, and ambiguity remains as those limited previous studies used different approaches to identify common clusters and findings may vary with approaches. This study estimates the prevalence of common chronic diseases and examines co-occurrence of diseases using four approaches: (i) identification of the most occurring pairs and triplets of comorbid diseases; performing (ii) cluster analysis of diseases, (iii) principal component analysis, and (iv) latent class analysis. Data were collected using a questionnaire mailed to a cross-sectional sample of senior Australians, with 4574 responses. Eighty-two percent of respondents reported having at least one chronic disease and over 52% reported having at least two chronic diseases. Respondents suffering from any chronic diseases had an average of 2.4 comorbid diseases. Three defined groups of chronic diseases were identified: (i) asthma, bronchitis, arthritis, osteoporosis and depression; (ii) high blood pressure and diabetes; and (iii) cancer, with heart disease and stroke either making a separate group or "attaching" themselves to different groups in different analyses. The groups were largely consistent across the approaches. Stability and sensitivity analyses also supported the consistency of the groups. The consistency of the findings suggests there is co-occurrence of diseases beyond chance, and patterns of co-occurrence are important for clinicians, patients, policymakers and researchers. Further studies are needed to provide a strong evidence base to identify comorbid groups which would benefit from appropriate guidelines for the care and management of patients with particular disease clusters.

Figure 1. Dendrogram of likely clusters using variable clustering of 10 chronic diseases.

Figure 2. Scree test with Eigenvalues for range of factors.