Development of methodological approaches to the formation of a risk-based model to minimize the prevalence of adverse reactions in drug application in medical organizations of Moscow

Objective: development of approaches to predict the likelihood of adverse reactions (ARs) when using drugs based on a comprehensive assessment of risk factors.Material and methods. We used a database containing 1,450 drug-related ARs reports from January through December 2021. A list of antibacterial drugs by international nonproprietary name (INN) with 4 or more ARs reports was selected as a reference group to perform various types of statistical analysis. A cumulative multivariate regression analysis was carried out on a database of 187 ARs notifications for 13 INN of antibacterial drugs. The study was performed in two stages. In the first stage, a statistical method was used (classical multiple regression, linear discriminant analysis, factor analysis, principal component regression, partial least squares regression, estimation of variance accuracy); at the second stage a modeling method was used. As part of the modeling stage, the integral score of the risk of ARs was presented as a sum of values for individual risk factors. Two groups of risks were proposed to be assessed: 1) intrinsic risk value for each factor (attribute), which was equal to the sum of risks of all factors (conditions) in which the drug had been used; 2) intrinsic risk value for antibacterial drugs by each INN. The total risk value was defined as the sum of the risk of the drug and all factors (conditions) in which this drug had been used.Results. The results were visualized in the form of a two-level risk-based model matrix, with a “heat map” of the risk level overlaid on it. The maximum total risk of ARs was obtained for ceftriaxone – 404.96 points, depending on patient’s gender. The minimum total risk was calculated for azithromycin and cefotaxime depending on the International Classification of Diseases (10th revision) code – 88.46 points. The proposed methodological approach also allows combining all possible combinations of drugs and conditions of their use. For example, for the use of vancomycin in hospital conditions by intravenous administration: intrinsic risk of use – 42.93 points; risk of use in hospital conditions – 183.68 points; risk when administered intravenously – 209.95 points; the total risk value in the designated situation – 436.56 points.Conclusion. The proposed approach can allow medical organizations to reduce significantly the number of ARs when using drugs by categorizing and preventing risks before they occur. It also has significant prospects of application at the federal level, given its modification on a large volume of data.