Seema Pant, Manoj Kumar Varshney, Gurprit Grover and Seema Pant*
Background: Owing to the new treatments and medicines, many cancer patients get cured of the disease and they do not experience the event of interest (death). Such patients constitute the cure fraction. To analyze survival data related to diseases with cured fraction, cure rate models have been found to be more appropriate as compared to the standard survival models. Promotional Time Cure Rate Model is one such model and it assumes that the patient death may have been caused due to some latent competing causes. In our case we have assumed that the number of competing causes follow either Binomial or Poisson or Negative Binomial Distribution.
Material and Methods: Parameter estimation has been done by Bayesian approach, using Markov Chain Monte Carlo (MCMC) technique. A real dataset from a breast cancer data of 85 patients is used to illustrate the proposed methodology. The software’s Open BUGS and STATA is used for the analysis purpose.
Results: The DIC value of binomial distribution is 143.8 which is least among the three distributions which we have considered for analysis. Also, the predictors Age, tumor size and tumor Grade are found to be significant. The cure rate is found to be 11.58 using the Binomial distribution as the distribution of the latent variable N. The overall cure rate is found to be 13.94 in the presence of predictors.
Conclusion: The findings revealed that Binomial – Exponential distribution with a cure fraction can be an interesting option to explain/predict the survival time and distribution of latent variables in Promotional Time Cure Model as compared to Negative Binomial and Poisson distribution in breast cancer patients.
Published Date: 2022-01-29; Received Date: 2021-12-29
