Approaches to forecasting of time dynamics of the processes and use of Farr’s epidemic evolution law on example of dynamics of new cases of HIV-infection in Russian Federation

Abstract

Introduction. Forecasting of different processes in health, including epidemics, are important area of public health. There exists an idea that in some cases simple models can give adequate forecasts.Goal of this study was to evaluate possible use and results of forecasting of registration of new cases of HIV infection in Russian Federation based on well-known Farr’s law.Materials and methods. The official statistical data on new HIV cases in 1999–2020 in Russian Federation were used. Parameters for forecasting new cases until 2027 were calculated according to Bregman and Langmuir. For evaluation of forecasting the calculations were done for 2015–2020 after fitting model with data until 2014. Normal components of the empirical epidemic curve were estimated and more appropriately fitted distributions were found for the data described by those components.Results. Estimations according to the Farr’s law somewhat undercount number of the new cases of HIV infection (it forecast 99% (95% CI 92–106%) cases when smoothing was used and 97% (95% CI 89–106%) when raw data were used). In general, especially when smoothing was used, fit was satisfactory. Forecast until 2027 show that total number of HIV cases in 1999–2027 will be 1.7–2.0 mln people. Analysis of most probable distribution of the second peak of epidemic curve show that it is lognormal, which allow for much larger number of infected in medium- and long-term perspective.Conclusion. Though Farr’s law could be used for short-term forecast it is not recommended to weaken preventive programs due to possibility of large increase in number of HIV-infected in comparison with Farr’s law forecast.