Most countries now recommend initiating HIV patients on tenofovir (TDF) as the standard NRTI in first-line therapy to reduce toxicities associated with the NRTI stavudine. Exploiting national guideline changes in South Africa (SA) and Zambia, we assessed the causal impact of a policy to initiate TDF on ART outcomes using regression discontinuity.
Prospective cohort study of ART-naïve, non-pregnant, HIV patients >16 years who initiated first-line ART in SA or Zambia (IeDEA-SA). Patients initiating ART +/-12 months around the national guideline changes were included: SA-1 April 2010 and Zambia-1 July 2007. We implemented a regression discontinuity, a quasi-experimental design, using the timing of national guideline changes as natural experiments. Patients initiating just before/after guideline change are similar but receive different regimens. Comparing those patients, we estimated the intent to treat (ITT) effect of guideline change on single-drug substitution (SDS), death, loss to follow-up (LTFU), CD4 response and virologic failure (VF, SA only) in the first 24-months on ART on a risk difference (RD) scale using local linear regression. We excluded patients initiating +/-14 days of the date of the guideline change in all estimates due to imprecision in the implementation of the guidelines. We then collapsed across country to estimate combined ITT effects.
16,773 South African and 44,399 Zambian patients were eligible. The probability of initiating TDF increased in both countries for patients starting ART after the guideline changes(Figures A and B). ITT estimates showed a significant decrease in the risk of SDS in SA(RD:-14%; 95%CI:-18%,-11%)(Figure C), while we saw no difference in Zambia(Figure D). In both countries we saw no effect on mortality(SA RD:1.0%; 95%CI:-2.2,4.0%; Zambia RD:-0.3%; 95%CI:-2.2,1.5%), LTFU(SA –RD:2.6%; 95%CI:-6.8,1.5%; Zambia RD:-1.0%; 95%CI:-3.4,1.3%), mean CD4(SA RD:10.1; 95%CI:-43.5,23.2; Zambia RD:5.9; 95%CI:-9.1,20.9), or VF in SA(RD:0.0%; 95%CI:-2.1,41.9%). Combined ITT estimates showed a significant increase in TDF (RD:37%; 95%CI:25%,48%) and no difference in outcomes.
Guideline changes led to an impressive increase in tenofovir initiation in SA and Zambia. Initiating patients on TDF led to reductions in SDS in SA, suggesting that a global policy to initiate TDF may have resulted in fewer patient-years spent on sub-optimal therapy and fewer patients experiencing side effects/toxicities. No change was observed in other outcomes.