SURVIVAL ANALYSIS OF PROGNOSTIC FACTORS ASSOCIATED WITH CERVICAL CANCER

Abstract

Cervical cancer is a major public health menace to women of reproductive age. It is the most common cancer in women in Nigeria. Survival refers to the life of a person after diagnosis of a disease and survival studies measure the overall performance of patients in terms of quality and quantity of life after diagnosis and treatment. This study investigate the data of patients diagnosed with cervical cancer from January 2010 to December 2020 at National Hospital Abuja (NHA). The Kaplan-Meier estimator was used to estimate the survival function and median time of the patients. Result showed that each patient has a 50% chance of surviving at least 13 months and a minimum of 10 months but not more than 17 months. The Log-rank test was used to test the differences in the survival curves. Result showed significant difference in survival times for International Federation of Gynaecology and Obstetrics (FIGO) stages and recurrence with a p-value of 5e-08 and 0.007 respectively. Classification and regression tree (CART) was used to predict the chance of survival of the patients. Findings from the CART model revealed that the model had 82.5% of correctly classifying cervical cancer patients and also showed menopause as the most important predictor of cervical cancer. Finally, Accelerated Failure Time (AFT) model was used to determine the prognostic factors associated with cervical cancer. AFT models used were Exponential. Weibull, Log-logistics and Lognormal distributions and based on Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC), all models were compared for performance. The Log normal with the minimum AIC and BIC values of 1821.70 and 1844.66 was the best model for the data and was subsequently used for further analysis. Results obtained establish that recurrence and patients who had adenosquamous (ADQ) histological type significantly prolonged the survival time of patients while parity, occupation house wife (H/W), tribe (Yoruba), tumour grade well differentiated (WD) and treatment received (chemotherapy) significantly shortened survival time of patients. The findings of this study showed that Lognormal AFT model described the survival time of the cervical cancer patients dataset better than other distributions used. Furthermore, the study found a high percentage diagnosed at advanced stage, which had negative effect on survival and stressed the need for improving early detection.