Purpose:  With increasing antiretroviral therapy (ART) use in resource-limited settings (RLS), concerns about drug resistance are rising.  Sentinel testing programs for HIV drug resistance were established to help inform ART policies and drug-sequencing decisions, with a non-nucleoside reverse transcriptase inhibitor (NNRTI) resistance threshold of 5% chosen to trigger concern.  Our objective was to evaluate how an increasing prevalence of primary NNRTI resistance (NNRTI-R) might influence recommendations for population-based ART choices in RLS.

Methods:  We adapted a state-transition model of HIV disease to simulate HIV clinical care in Côte d'Ivoire.  At alternative prevalences of NNRTI-R (which decreases NNRTI efficacy), we evaluated the incremental cost-effectiveness of: 1) no ART, 2) Co-trimoxazole (TMP-SMX) alone, 3) TMP-SMX + ART beginning with an NNRTI-based regimen (75% HIV RNA suppression, 48 wks), followed by a boosted protease inhibitor (PI)-based regimen (58% HIV RNA suppression, 48 wks), and 4) TMP-SMX + ART beginning with a boosted PI-based regimen (75% HIV RNA suppression, 48 wks), followed by an NNRTI-based regimen (30% HIV RNA suppression, 48 wks).  In sensitivity analyses, we varied drug costs (base case NNRTI regimen = $427/yr, PI-regimen = $580/yr) and treatment efficacies both with and without resistance.  Life years and costs were discounted at 3%/year. 

Results:  Results from the base case, with 5% NNRTI-resistance, are shown:

Strategy	Mean per person lifetime cost (2005 $US)	Life expectancy (months)	Cost-effectiveness ($/YLS)
No ART	1,060	31.4	-----
TMP-SMX alone	1,090	33.3	210
PI-based --> NNRTI-based	4,970	80.5	dominated
NNRTI-based  --> PI-based	5,210	87.5	910

Addition of two regimens of ART to TMP-SMX alone more than doubled life expectancy.  Up to a population prevalence of NNRTI-R of 31%, starting with an NNRTI-based regimen weakly dominated a strategy starting with a PI-based regimen. At a cost per YLS <3X GDP per capita (Côte d'Ivoire = $2,409) – a threshold considered good value – results consistently favored initiation with an NNRTI-based regimen, regardless of the prevalence of NNRTI resistance in the population (up to 76%) and the efficacy of an NNRTI-based regimen in the setting of resistance.  

Conclusions: Drug costs and treatment efficacies, but not NNRTI resistance levels, were most influential in determining optimal HIV drug sequencing. Results of surveillance for NNRTI resistance should not be used as a major guide for treatment policy in RLS.