Popovich Irina Grigorevna
To determine the feasibility of using cluster analysis to manage treatment costs of systemic autoimmune rheumatic diseases (SARDs). MATERIALS AND METHODS. The object of the study was the case histories of patients hospitalized in 2020. A total of 954 case histories of patients with SARDs were analyzed, among them systemic scleroderma – 411 patients (43.1%), systemic lupus erythematosus – 263 (27.5%), rheumatoid arthritis seropositive and seronegative – 103 (10.8%), systemic vasculitis associated with anti-neutrophil cytoplasmic antibodies – 98 (10.3%), idiopathic inflam- matory myopathies (polymyositis, dermatomyositis) – 57 (6%), Sjögren’s disease – 22 (2.3%). Hierarchical cluster analysis by weighted pairwise mean using Euclidean distance and Chebyshev distance, and non-hierarchical clus- tering by k-means were performed for the sample of case histories in the STA- TISTICA 13.3. Multiple regression model with a dependent factor of treatment costs was built in MS Excel. RESULTS. Hierarchical clustering resulted in 2 dendrograms, which yielded the same number of clusters – 4. The cluster analysis identified 4 clusters with significant differences (p<0.05) in treatment costs with formative indicators: gender, age of patients, duration of disease, number of bed-days, number of hospitalizations, disease activity level, number of comorbidities and compli- cations, previous use of biologic agents, total number of medications pre- scribed, use of Acellbia, Benlista, other biologic agents, and immunoglobulins during hospitalization. General multiple regression model for all patients with SARDs and separate multiple regression models for each cluster were con- structed using statistically significant factors. CONCLUSIONS. Factors affecting the treatment costs for patients with SARDs include the number of days of hospitalization, the degree of disease activity, the number of medications prescribed, and the use of biologic agents and immunoglobulins during hospitalization.
To determine the feasibility of using cluster analysis to manage treatment costs of systemic autoimmune rheumatic diseases (SARDs). MATERIALS AND METHODS. The object of the study was the case histories of patients hospitalized in 2020. A total of 954 case histories of patients with SARDs were analyzed, among them systemic scleroderma – 411 patients (43.1%), systemic lupus erythematosus – 263 (27.5%), rheumatoid arthritis seropositive and seronegative – 103 (10.8%), systemic vasculitis associated with anti-neutrophil cytoplasmic antibodies – 98 (10.3%), idiopathic inflam- matory myopathies (polymyositis, dermatomyositis) – 57 (6%), Sjögren’s disease – 22 (2.3%). Hierarchical cluster analysis by weighted pairwise mean using Euclidean distance and Chebyshev distance, and non-hierarchical clus- tering by k-means were performed for the sample of case histories in the STA- TISTICA 13.3. Multiple regression model with a dependent factor of treatment costs was built in MS Excel. RESULTS. Hierarchical clustering resulted in 2 dendrograms, which yielded the same number of clusters – 4. The cluster analysis identified 4 clusters with significant differences (p<0.05) in treatment costs with formative indicators: gender, age of patients, duration of disease, number of bed-days, number of hospitalizations, disease activity level, number of comorbidities and compli- cations, previous use of biologic agents, total number of medications pre- scribed, use of Acellbia, Benlista, other biologic agents, and immunoglobulins during hospitalization. General multiple regression model for all patients with SARDs and separate multiple regression models for each cluster were con- structed using statistically significant factors. CONCLUSIONS. Factors affecting the treatment costs for patients with SARDs include the number of days of hospitalization, the degree of disease activity, the number of medications prescribed, and the use of biologic agents and immunoglobulins during hospitalization.
