Table 4. Relationship between exposure interval and treatment failure at 48 weeks in the NNRTI Response Study—Zambia, Kenya, Thailand (2005–2008).

Interval between Single-Dose NVP Exposure and Starting ART	Primary Analysisa			Primary Analysis, Excluding Deaths before 90 Daysb			On-Treatment Analysisc
	No. Failures/N (%)	Unadjusted ORs (95% Exact Confidence Limits)	Adjusted ORs (95% Wald Confidence Limits)	No. Failures/N (%)	Unadjusted ORs (95% Exact Confidence Limits)	Adjusted ORs (95% Wald Confidence Limits)	No. Failures/N (%)	Unadjusted ORs (95% Exact Confidence Limits)	Adjusted ORsb (95% Wald Confidence Limits)
Unexposed	132/523 (25)	Ref	Ref	109/500 (22)	Ref	Ref	38/450 (8)	Ref	Ref
1–6 mo	47/116 (41)	2.02 (1.29, 3.13)	2.16 (1.34, 3.49)	45/114 (39)	2.34 (1.52, 3.60)	2.35 (1.45, 3.82)	33/107 (31)	4.84 (2.85, 8.20)	5.25 (2.87, 9.60)
7–12 mo	25/67 (37)	1.76 (0.99, 3.09)	1.47 (0.82, 2.65)	19/61 (31)	1.62 (0.91, 2.90)	1.42 (0.76, 2.64)	5/51 (10)	1.18 (0.44, 3.14)	1.05 (0.38, 2.88)
>12 mo	42/172 (24)	0.96 (0.63, 1.45)	1.02 (0.66, 1.59)	35/165 (21)	0.97 (0.63, 1.48)	0.98 (0.62, 1.56)	20/152 (13)	1.64 (0.92, 2.92)	1.92 (1.03, 3.58)

^aAdjusted ORs controlling for country, CD4 cell count, viral load, WHO stage, age, hemoglobin, and body mass index (all as categorical variables as in table 1). See methods for definition of primary analysis.

^bAdjusted ORs controlling for country, CD4 cell count, viral load, WHO stage, and age (all as categorical variables as in table 1) – hemoglobin and body-mass index were not retained in the final model.

^cAdjusted ORs controlling for country, CD4 cell count, viral load, WHO stage, and age (all as categorical variables as in table 1) - hemoglobin and body-mass index were not retained in the final model. A separate multivariate model that included self-reported adherence did not alter the adjusted ORs in any appreciable way. See methods for definition of on-treatment analysis.

Ref, reference group.