Table 1.

Ecologic studies of community viral load and treatment as prevention of new HIV infections.

Setting	Time Period	Evaluation	Outcomes	Reference
Taiwan	1984 – 2002	National HIV surveillance data. Transmission rate estimated by use of exponential model.	Transmission rate 0.391 new cases/prevalent cases pre-ART Transmission rate 0.184 new cases/prevalent cases post-HAART. Overall decrease 53%.	[52]
Vancouver, British Columbia, Canada	1996–2007	Prospective cohorts of injection drug users. Median CVL*. Cox regression model to association with HIV incidence.	CVL associated with time to HIV seroconversion (Hazard ratio 3.32 per log10 increase). After median viral load fell to <20,000 copies/mL, no statistical association with HIV incidence observed.	[34]
San Francisco	2004–2008	HIV/AIDS public health surveillance for new diagnoses and calculated HIV incidence. Mean Community viral load. Poisson models for CVL and new HIV diagnoses.	Significant decline in mean CVL 2004–2008 (p=0.037). Reduction in CVL associated with decrease in new HIV diagnoses (p=0.003).	[53]
British Columbia, Canada	1996–2009	HAART† coverage from centralized registry. HIV public health surveillance for new diagnoses. Median CVL. Poisson models for CVL, ART coverage and new HIV diagnoses.	547% increase in ART uptake. 52% decrease in new HIV diagnoses. For every 100 additional individuals on ART, new cases decreased by factor of 0.97. For every 1 log10 drop in CVL, new cases decreased by factor of 0.86.	[39]
Hlabisa sub-district, KwaZulu-Natal, South Africa	2004–2011	Cohort of 16,667 HIV uninfected individuals. ART coverage and HIV prevalence within the surrounding community assessed, and rate of new seroconversions captured.	As ART coverage within a community expands, risk of acquisition decreases: for communities with 30–40% ART penetration, risk of new infection dropped 38% compared to communities with <10% ART uptake.	[54]

^*CVL = community viral load

^†ART = highly active antiretroviral therapy. Adapted from [55] with permission.