Background: Current ambulatory approaches for delivering treatment to patients with rntuberculosis in Malawi have significant operational delivery challenges. This led to a clinical rntrial aimed at establishing if guardian administered streptomycin is non-inferior to hospital rnadministered streptomycin. In order to establish effectiveness of guardian administered rnstreptomycin, all integral outcomes including survival need to be assessed. rnObjectives: Based on secondary analysis of existing trial dataset, this study aimed to compare rnsurvival of patients with tuberculosis who received streptomycin administered at hospital or at rnhome by a guardian controlling for potential factors that may confound or mediate effect of rnthe treatments. rnMethods: In the original study - a non-inferiority, parallel, randomised, open label, phase III rntrial â€“204 patients were randomized to receive streptomycin at the hospital or in their homes rnwhere it was administered by a patient nominated guardian. One hundred and one patients were rnrandomized into hospital arm while 103 into home-based group. Patients were followed up for rnten months. In this study, comparison of survival between the treatment delivery options rnemployed Logrank test. Cox proportional hazard model was used to identify factors that rnconfound or mediate survival of patients. Potential factors included age, sex, TB class, TB rncategory, HIV status, smoking status, drinking status and employment status of participants and rneducation level of household head. rnResults: No significant difference was observed in survival of patients between the two groups rn(p-value = 0.726, HR 1.15 (0.52 to 2.55). All covariates except unemployment had no effect on rnsurvival of patients (CI for HR had 1). rnConclusion: Use of patient nominated trained lay carers to administer streptomycin provides a rnpotential convenient and cost-effective approach for treating patients with recurrent and drug rnresistant TB. However, decision for its adoption should be cautious due to small sample size used, switching of critically ill patients from home-based group to hospital arm, inadequate rninformation on how patients selected guardians and lack of details on how TB resistance was rnassessed. A follow up study on assessment of acceptability of the proposed model of care would rnhelp understand if it could be successfully adopted and used to deliver long term TB injectable rnagents to patients.