A Statistical Analysis of the Factors with Nonrandom Impact on the Survival Rate of Women with Breast Cancer, by Type of Medical Treatment

Abstract

A statistical analysis of the factors with nonrandom impact of the survival rate of women with breast cancer, by type of medical treatment in Ukraine, was made using data from the National Cancer Register. The complexity and ambiguity of the problem related with the choice of a special treatment type, i. e. surgical treatment or combined surgical treatment, was emphasize. The combined surgical treatment is available in two options: as a combination of surgical treatment and radiation therapy and a combination of surgical treatment, radiation therapy and chemical therapy. The data on female patients who had medical treatment in the National Institute of Cancer used for a statistical analysis to reveal positive and negative results from each type of special medical treatment. The need for the assumption on proportionality was substantiated, with its testing based on a graphic analysis by grouping factors. The obtained results led to the conclusion that the model could be extended by the categorical factors: type of treatment (grouping factor), type of surgical operation, phase of decease. The assessment was made based on analyzing the factors’ impact by type of medical treatment. It was demonstrated that the model could be extended by interaction effects that would enable for estimating the relative risk depending on a combination of the treatment group and the respective factor. None of the effects included in the model could prove to be statistically significant. It means that no conclusions could be made about the factors’ interactions by type of the medical treatment. The interpretation of all the other factors that proved to be significant was applied to all the population. The strongest projection weight in the model was with the factor of decease phase, followed by the type of medical treatment and, in equal proportions, by the age and the type of surgical operation. The results of checking the global test could demonstrate the feasibility of predicting the survival rate for the overall model with the significance level equal to 0.05. It was proved that adjuvant and chemical therapies could reduce the risks of the recurrence, but this result had not an ultimate statistical validity. Despite of the positive non-recurrence result, the survival rate by groups still reduces, being an obvious effect of the abovementioned groups of medical treatment.

Keywords

• breast cancer, survival rate, adjuvant therapy, cox model, lls curves, relative risk.