Cluster analysis identifies long COVID subtypes in Belgian patients

Abstract

Severe acute respiratory syndrome coronavirus infection presents complications known as long COVID, a multisystemic organ disease which allows multidimensional analysis. This study aims to uncover clusters of long COVID cases and establish their correlation with the clinical classification developed at the Clinical Research Unit of Brugmann University Hospital, Brussels. Such an endeavour is instrumental in customizing patient management strategies tailored to the unique needs of each distinct group. A two-stage multidimensional exploratory analysis was performed on a retrospective cohort of 205 long COVID patients, involving a factorial analysis of mixed data, and then hierarchical clustering post component analysis. The study's sample comprised 76% women, with an average age of 44.5 years. Three clinical forms were identified: long, persistent, and post-viral syndrome. Multidimensional analysis using demographic, clinical, and biological variables identified three clusters of patients. Biological data did not provide sufficient differentiation between clusters. This emphasizes the importance of identifying or classifying long COVID patients according to their predominant clinical syndrome. Long COVID phenotypes, as well as clinical forms, appear to be associated with distinct pathophysiological mechanisms or genetic predispositions. This underscores the need for further research.

Keywords: SARS CoV-2; clinical forms; clusters; long COVID; phenotypes.

Figure 1.

Patient file selection flow chart.

Figure 2.

Optimal principal components for dimension reduction. y-axis: the eigenvalues of the factors or principal components. x-axis: the corresponding factor or component number.

Figure 3.

Factor map representing clusters of individuals.