Table 2. Independent predictors of VL <80 copies/ml 48 and 144 weeks after ART initiation.

	Week 48 (N=342)		Week 144 (N=343)
	OR (95% CI)	P	OR (95% CI)	P
Main models
sdNVP exposure (yes vs no)	2.34 [1.26-4.34]	0.007	1.75 [0.93-3.29]	0.08
Age at ART initiation (per year younger)	0.70 [0.46-1.08]	0.11	0.72 [0.50-1.05]	0.09
Allocated treatment strategy, vs Arm-A (3TC/ABC/NNRTI throughout)		0.20		0.22*
Arm-B (3TC/ABC/NNRTI throughout, ZDV until week 36)	0.77 [0.43-1.38]	0.38	0.78 [0.42-1.44]	0.43
Arm-C (3TC/ABC/ZDV throughout, NNRTI until week 36)	1.30 [0.74-2.31]	0.36	0.60 [0.33-1.07]	0.08
Pre-ART VL (per log10 higher)	0.55 [0.38-0.79]	0.001	0.57 [0.39-0.83]	0.003
Male (vs female)	0.53 [0.33-0.85]	0.009		0.10 **
Current ART as all syrups (vs tablets)	0.56 [0.31-1.01]	0.05		0.13 ***
Missed doses in last 4 weeks (yes vs no)	0.35 [0.13-0.94]	0.04		0.92

Note: multivariable models based on backwards elimination (exit p=0.1) on complete cases from all factors in Table 1, forcing sdNVP, age at ART initiation and ART-strategy randomisation into the model. Italics shows effect from adding variables into the final model. No evidence of non-linearity in age at week 48 (p=0.9) or 144 (p=0.6).

^*Arm-C vs A and B combined OR=0.67 [0.41-1.10] p=0.12.

^**adjusted (for factors above) OR=0.67 [0.41-1.08] p=0.10.

^***adjusted (for factors above) OR=0.40 [0.12-1.33] p=0.13; only 4% children not taking at least one drug as tablet formulation at week 144.