Purpose
Heterogeneous findings exist on antiretroviral treatment (ART) use in pregnancy and preterm delivery (PTD) or infants born small-for-gestational age (SGA). Whether reported differences may be explained by methods used to ascertain gestational age(GA) has not been explored.
Methods
We enrolled consecutive pregnant women attending a large primary care antenatal clinic in South Africa. Public-sector midwives assessed GA by last menstrual period (LMP) and symphysis-fundal height (SFH). Separately, if clinical GA was <24weeks, ultrasound (US) was performed by a research sonographer blinded to midwife assessments. In analysis, the impact of measurement error on the association between HIV/ART status and birth outcome by GA method was assessed, and factors associated with clinical GA under- or over-estimation identified.
Results
In 1787 women included overall, estimated PTD incidence was 36% by LMP, 17% by SFH and 11% by US. PTD risk was higher for HIV-infected than HIV-uninfected women using US-GA (adjusted odds ratio (aOR) 1.95; 95% CI 1.10–3.46); for LMP/SFH-GA the associations were smaller and not significant. These findings persisted after adjustment for age, parity, height and previous PTD. PTD risk did not vary by timing of ART initiation (before or during pregnancy) for any method. Elevated BMI and older age were associated with decreased risk of under-estimation by both LMP and SFH; HIV status and obesity were associated with increased risk of over-estimation by SFH. There were no differences in SGA incidence across GA methods.
Conclusions
Findings for an association between HIV/ART and birth outcomes are substantially influenced by GA assessment method. With growing public health interest in this association, future research efforts should seek to standardize optimal measures of gestation.