Factors associated with survival of cancer patients receiving palliative care: a case study of Hospice Africa Uganda.

Abstract

This study concludes that HIV status, Smoking habits of the patient, alcohol consumption and age of the patient are significant factors associated with the survival of cancer patients receiving palliative care. In addition, the study notes that for survival data analysis, the Cox proportional model and the censored quartile regression models are both essential and can augment each other. Overall, they reflect the same information about the variable effects but with different details and insights. The Cox proportional model indicates a constant effect for the risk variables for this result, whilst the censored quantile regression model captures and better characterises the variability in effect of different risk factor variables on survival time across the distribution. Further the study has shown that the proportional hazards model could only indicate that the risk factors were significant in affecting survival times, the CQR model showed at what point of the survival time distribution were these risk factors significant. The CQR model also highlighted the dynamic nature by which the risk factors can affect survival times through showing that the effect of a risk factor can be significant in the early periods of initiation and yet after a certain period of continuous treatment, the risk factor ceases to affect survival time in a significant manner. The advantage of such an insight is that decision makers can plan based on the pattern of risk factor effectiveness, thereby helping positively those that are on treatment to aid in their quest to survive longer.