Risk of Mortality and Loss to Follow-up amongst HIV-Positive Patients Prescribed Stavudine vs. Tenofovir: A matched cohort analysis

Abstract

Background: In April 2010 the South African government replaced stavudine with tenofovir in first-line regimens available in the public sector despite tenofovir’s higher cost. This was done under the assumption that the change would improve treatment outcomes, partly through reduced toxicity and better adherence. We examined the risk of death and loss to follow-up in the first 12-months on antiretroviral therapy (ART) amongst HIV-positive patients initiated on tenofovir- versus stavudine-based first-line regimens. Methods: We conducted a matched cohort analysis amongst HIV-positive patients newly initiated on to either stavudine-based (between January 2009-April 2010) or tenofovir-based (between April 2010-June 2011) regimens at the Themba Lethu Clinic, in Johannesburg, South Africa. Tenofovir patients were matched 1:1 to stavudine patients (on age, sex, TB, hemoglobin, BMI, CD4 count and type of NNRTI) using propensity scores. We evaluated the association of regimen with all-cause mortality over 12-months on ART. Death and transfer were considered competing risks for loss, so we used Fine and Gray’s regression method and calculated cumulative incidence curves to evaluate the relationship between stavudine and tenofovir on loss to follow-up over 12-months. Results: 1871 patients prescribed stavudine were matched to 1871 patients prescribed tenofovir. 130 (7.0%) stavudine and 80 (4.3%) tenofovir patients died, while 191 (10.2%) stavudine and 215 (11.5%) tenofovir patients were loss to follow-up during the first 12-months on ART. Crude rates of mortality were higher for stavudine patients (7.9/100 person-years) compared to tenofovir patients (4.9/100 person-years), while rates of loss were slightly lower in stavudine patients (11.6/100 person-years) vs. tenofovir (13.1/100 person-years). Kaplan-Meier curves  and hazard models showed stavudine patients had a higher rate of death (HR 1.62; 95% CI: 1.23-2.14), while competing risks analysis and cumulative incidence estimates showed stavudine patients with a slightly lower rate of becoming loss to follow-up (HR 0.88; 95% CI: 0.73-1.07) compared to tenofovir patients in the first 12-months on treatment. Conclusions: Although comparable in regards to loss to follow-up, patients initiated onto stavudine have a higher risk of death versus those initiated onto tenofovir. Despite the increased cost, the use of tenofovir could help improve long-term treatment outcomes in HIV-positive patients.

Conference: IWHOD 2013, Cavtat, Croatia