Figure 3. Time between exposure to single-dose NVP and starting antiretroviral therapy and the probability of treatment failure in the NNRTI Response Study—Zambia, Kenya, Thailand (2005–2008).
Locally weighted regression (LOESS) models of the risk of treatment failure as a function of the time interval between NVP ingestion and starting ART. The left panel defines treatment failure according to the study's primary definition; the right panel defines treatment failure according to a planned secondary definition (see Methods). The horizontal dotted line indicates the failure rate among the women who were not exposed to single-dose NVP. The individual plusses (+) indicated on each panel represent individual patients who were either failing (top) or not failing (bottom).