Modeling survival rate of drug abusers in Kenya

Abstract

Drug and substance abuse is a serious health problem in many countries. In Kenya drug abuse is one of the leading causes of mortality. The government and other stakeholders have made efforts to fight the problem of drug abuse. However there are no significant results that have been drawn from these efforts. This study sought to model the survival rate of drug users in relation to drug and substance abuse. The objectives of this study were to formulate survival model for drug users using a Kaplan-Meier and Cox proportional hazards model, to establish the recovery rates of drug users under medication, to perform sensitivity analysis on the model parameters to determine the significant predictors of drug use and compare survival rates based on significant predictors. The dependent variable was survival time to recovery of the subject and the independent variables were age, gender, residence, marital status, job status, mode of drug abused and the type of drug abused. The study used secondary data on drug use obtained from Mathari National Hospital. Data was collected from specialized registers containing the drug users’ medical information provided by the hospital. Data was fitted to the survival model using R statistical software. Kaplan-Meier and Cox Proportional Hazard methods were used to formulate a survival rate model for drug users. Sensitivity analysis of the model parameters was performed to determine an optimal model for the study. Kaplan-Meier model was used to establish the rate of recovery of drug users. The optimal model revealed that age, gender, marital status and job status were significant predictors .Female drug users had higher survival rates (80.95%) compared to male drug users (19.05%).The overall survival rate was 36.37% recovery rate increased with progression of treatment. The study recommended that campaigns against drug abused should be more focused towards treating male subjects since they have lower survival rates compared to female subjects.