Menu 返回 Poster-Presentations-Details

M-24: Factors Associated with Treatment Outcomes for Patients with Extremely Drug Resistant TB: A Random Survival Forest Approach

Poster Presenter

      Justine Nasejje

      • Lecturer
      • University of Kwazulu-Natal
        South Africa


Identifying factors strongly associated with favorable and unfavorable long-term treatment outcomes for patients diagnosed with extremely drug-resistant TB (XDR-TB).



The study was conducted in South Africa between March 2008, and August 2012. A total of 107 adult patients with microbiologically confirmed XDR-TB from three provinces in South Africa, were hospitalized on treatment. We used Random Survival Forest models for data analysis.


By the end of the study, 75% of the patients had died. The results have shown that the factors responsible for predicting survival among patients with extremely drug-resistant TB include, net conversion, net reversion, HIVART status, treatment with Isoniazid, Amoxicillin, and Clofazamine.


Apart from the interaction between Ethambutol and Isoniazid that was associated with net reversion, most of the drug interactions were not important. However, we recommend that further investigations on existing drug interactions in the treatment of XDR TB should be made a priority in a country like South Africa where XDR TB cases are on the rise and yet new drugs such as Linezolid which could be effective in the treatment of XDR TB are inaccessible.