Section: 
Original articles
Original articlesThe assessment of disease symptom dynamics using survival analysis methods
A. A. Korneenkov (1), I. V. Fanta (1), Е. E. Vyazemskaya (1)
(1) Saint Petersburg Research Institute of Ear, Throat, Nose and Speech Ministry of Healthcare of Russia, Saint Petersburg, 190013, Russia
UDK: 519.237.5:303.7: 004.43:616.24-008.41
DOI: https://doi.org/10.18692/1810-4800-2019-4-8-14
ABSTRACT
The article describes an approach to the analysis of disease symptom dynamics from the viewpoint of survival analysis methods. It is justified in the events where the observation time is insufficient to record the disappearance or critical reduction of the severity of the symptom. Using the example of analysis of the cough disappearance time in ARVI, the authors demonstrated the solution of several typical tasks: the assessment of dependence of cough symptom disappearance on time, the comparison of antitussive preparations effect, building a model of probability of cough disappearance within the observation period with consideration of additional signs – covariates. All calculations were made in the statistical software R. To evaluate the probability of a symptom disappearance by a particular observation day, the authors used a non-parametric analysis method – the Kaplan–Meier method; to compare the effects of drugs on cough preservation, the low-rank test was used; the Cox semi-parametric proportional risk method was used to build the symptom disappearance, probability model. It has been shown that the presented method provides an efficient prediction of the likelihood of the symptom disappearance within the observation period, patients stratification according to the risk of preservation of the symptom in question, adequately changing the diagnostic and treatment approaches and considering the factors that are important for the outcome in question; besides, it provides the assessment of their contribution to the probability of disappearance of clinically significant symptom. The text of the article contains the program codes of R language with explanations.
Publication date:
06.08.2019
Keywords:
symptom dynamics assessment, survival analysis, survival curve, censored data, Kaplan-Meier method, low-rank test, Cox risk regression model, R-software, R-language. For citation:
Korneenkov A. A., Fanta I. V., Vyazemskaya E. E. The assessment of disease symptom dynamics using survival analysis methods. Rossiiskaya otorinolaringologiya. 2019;18(4):8–14. https://doi.org/10.18692/1810-4800-2019-4-8-14