CLUSTER ANALYSIS AS A METHOD FOR TESTING THE HYPOTHESIS ABOUT THE INFLUENCE OF GLUCOSE METABOLISM DISORDERS ON THE COURSE OF ACUTE OTITIS MEDIA IN ADULTS

Abstract

Relevance. Relevance of the research is caused by the clinical necessity to increase efficiency of diagnostics and treatment of acute otitis media (AOM) from the modern positions personalized medicine and the feasibility of statistical verification of the validity of the hypothesis about the influence of glucose metabolism disorders on the formation of clinical and pathogenetic features of acute otitis media in adults.
 Objective. To carry out statistical verification of the hypothesis about the influence of carbohydrate metabolism disorders on the complicated course of AOM on the basis of cluster analysis and to determine its informativity.
 Materials and methods. The cluster analysis included 73 with AOM. The 1 group (n=17) - patients with AOM; 2 group (n=15) - patients with AOM with presence of the sinus pathology; 3 group (n=20) - patients with AOM due to type 2 diabetes mellitus (DM2); 4 group (n=21) - patients with AOM due to transient glucose metabolism disorders. Apart from the clinical examination, all patients were tested for HbAlc, parathormone, osteocalcin, total calcium, IL-4, IL-6, FNP-α, and low-density lipoproteins (LDL). Cluster analysis on the basis of laboratory indicators was performed by the hierarchical method using the iterative algorithm of k-means and prior standardization of the data.
 Result. The hypothesis was tested at the level of stratification of patients into 4 and 2 clusters. In the first case, 28.8 %, 31.5 %, 15.6 %, and 24.6 % of the total number of patients were included in clusters I, II, III, and IV. Patients with AOM due to impaired glucose metabolism were predominantly in clusters I and II, comprising 51.2 % and 69.5 % of the cluster contingent respectively, while DM2 prevailed in cluster I and transient glucose metabolism disorders in cluster II (61.9 % and 62.5 % accordingly), 72.7 % and 94.4 % of patients with AOM without impaired glucose metabolism were stratified in clusters III and IV.
 The analysis of laboratory indicators using the k-means algorithm showed a general tendency for the increase in glucose, HbAlc, parathormone, TNF-α and LDL in patients with AOM in clusters I and II with increased levels of osteocalcin, calcium and IL-4 in patients in clusters III and IV with the presence of a significant difference (p˂0.05).
 At the level of patients' grouping into two clusters, cluster I included 41.9 % of the total number of patients due to the presence of all glucose metabolism disorders (of these, DM2 occurred in 57.6 % and transient glucose metabolism disorders in 43.3 %). The II cluster (58.9 % of patients) included all patients with AOM without impaired glucose metabolism, accounting for 74.4 % of its structure, and 25.6 % of patients with impaired glucose metabolism (6.97 % for DM2 and 18.6 % for transient impairments).
 The k-means algorithm revealed a clear and reliable distribution of clusters for all 9 studied indicators, which resulted in an increase in glucose content, HbAlc, levels of parathormone, IL-6, TNF-α and LDL and decreased osteocalcin, calcium and IL-4 in I cluster compared to II. Differences in the indicators at the group-group levels, group-cluster and cluster-cluster revealed a sufficient difference in the first and third cases (with the exception of osteocalcin) and their absence at the group-cluster level (with the exception of parathormone for cluster I, p˂0.01 and glucose and HbAlc for cluster II, p˂0.05).
 Conclusion. The cluster analysis was informative and confirmed the hypothesis about the influence of glucose metabolism disorders on peculiarities of AOM pathogenesis in adults according to laboratory indices. The principal importance of the presence or absence of impaired glucose metabolism in patients with AOM is confirmed by stratification of patients into two clusters. Thus, all patients without glucose metabolism disorders were included in one (II) cluster. The contingent of patients with impaired glucose metabolism was not homogeneous, with cluster I comprising only patients with impaired glucose metabolism, but 15 % of patients with DM2 and 38.1 % of patients with transient disorders were in cluster II. Therefore, to improve the quality of diagnostics of complicated course of acute otitis media on the basis of glucose metabolism disorders it is necessary to supplement the laboratory indicators with informative clinical criteria. Parathormone level, as well as the combination of IL-6, TNF-α and LDL levels can be used as a marker of impaired glucose metabolism.