Clustering Diagnoses from 58M Patient Visits in Finland 2015–2018 (Preprint)
BACKGROUND Patients with multiple chronic diseases cause a major burden to the health service system. Currently, diseases are mostly treated separately without paying enough attention to their relationships, which results in a fragmentation of the care process. Better integration of services can lead to more effective organization of the overall health care system. OBJECTIVE To analyze the connections between diseases based on their co-occurrences in order to support decision-makers in better organizing health care services. METHODS We performed cluster analysis of diagnosis using data from the Finnish Health Care Registers for primary and specialized health care visits and inpatient care. The target population of this study comprised all individuals aged 18 years or older who used health care services during the years 2015–2018. Clustering was performed based on the co-occurrence of diagnoses. The more the same pair of diagnoses appears in the records of same patients, the more the diagnoses correlate. Based on the co-occurrences, we calculated the relative risk of each pair of diagnoses and clustered the data using a graph-based clustering algorithm called M-algorithm, a variant of k-means. RESULTS The results reveal multimorbidity clusters, of which some are expected, for example one representing hypertensive and cardiovascular diseases. Other clusters are more unexpected, such as a cluster containing lower respiratory tract diseases and systemic connective tissue disorders. We also report the estimated cost effect of each cluster to society. CONCLUSIONS The method and achieved results provide new insight to identify key multimorbidity groups, especially ones resulting in burden and costs in health care services.