Comparative Study on Survival Analysis Models for Obstetric Fistula Patients in Nigeria
Musa Aishat O, S.I.S. Doguwa and A. Yahaya
Keywords: Weibull Accelerated failure time model, Kaplan Meier, Obstetric Fistula, Weibull proportional hazard model, Cox proportional hazard model, Exponential model.
Abstract
Survival analysis was designed for study of time to event; where event here are not limited to: contacting a disease, equipment failures, earthquakes, automobile accidents. This paper investigates survival recovery time and risk factors for fistula disease in Nigerian patients, aiming to inform healthcare managers on timely decisions and provide prompt care. The study used data from the Demographic and Health Survey (DHS) to estimate obstetric fistula patients’ parameters, formulated predictive models using maximum likelihood methods, and analyzed using R statistical software. The time recorded in weeks indicates that patients whose marital status type is married have higher survival probability than patients who are divorced. Patients with formal education have a higher probability of recovering from obstetric fistula than patients with no formal education. Patients with less than two days of labor have shorter survival recovery time than those patients who had more than two days of labor. Patients with no abortion had a higher survival recovery probability compared to those with abortion. Patients with height greater than 150cm have higher survival recovery probability than patients with height less than 150cm; finally, patients who receive antenatal care have a lower survival recovery probability than those who do not receive antenatal care. The results for Weibull AFT model indicate that Survival time recovery of obstetric fistula patients were significantly related with age at first marriage, number of deliveries, educational status, mode of delivery, number of abortions, height and follow-up of antenatal care. Also, the average recovery time of obstetric fistula patient is 5.16 weeks. The Accelerated Failure Time (AFT) Weibull model and Weibull Proportional Hazard model has the smallest AIC and BIC values but the Proportional Hazard assumption was violated Hence, the AFT Weibull model was found to fit the data best compared to other models for modeling Obstetric fistula disease.