Table 3.

Regression analyses of length-for-age z-scores (LAZ) at birth and their evolution over time by exposure to antiretroviral drugs (ARV).

	Univariable linear regression of LAZ at birth N=531		Multivariable linear regression of LAZ at birth* N=319
	Difference in LAZ (95%CI**)	P-value	Difference in LAZ (95%CI**)	P-value
ARVs in the first trimester	−0.56 (−0.25 to 0.06)	0.11	−0.05 (−0.41 to 0.30)	0.76
No ARVs during pregnancy	−0.21 (−0.50 to 0.08)	0.15	0.26 (−0.23 to 0.75)	0.30
ARVs in the second or third trimester	0 (reference)	-	0 (reference)	-

	Univariable mixed-effects regression of LAZ over time (N=577)		Multivariable mixed-effects regression of LAZ over time (N=356)
	Difference in LAZ (95%CI**)	P-value	Difference in LAZ (95%CI**)	P-value
ARVs in the first trimester	−0.27 (−0.52 to −0.01)	0.04	−0.35 (−0.63 to −0.08)	0.01
No ARVs during pregnancy	−0.23 (−0.45 to −0.001)	0.05	0.12 (−0.25 to 0.49)	0.53
ARVs in the second or third trimester	0 (reference)	-	0 (reference)	-

^**CI, confidence interval; ARV, antiretroviral therapy; LAZ, length-for-age z-score

^*Adjusted for mother's age (younger than 25 years or older), CD4 cell count (square root transformed), viral load (undetectable or detectable), year of birth, and family income.N=356 children