OBJECTIVE

To evaluate the association between HIV antiretroviral therapy (ART) and preterm birth (PTB), when defined by gold standard antenatal ultrasound versus newborn exam.

DESIGN

A secondary analysis of the PROMISE 1077BF/1077FF randomized controlled trial, which compared antiretroviral strategies to reduce perinatal HIV transmission and improve maternal health. The trial used newborn exam (i.e. New Ballard Score, NBS) to assess gestational age. This analysis included liveborn singleton pregnancies with both newborn exam and ultrasound data. The primary exposure was the trial's antiretroviral strategies: zidovudine with intrapartum nevirapine ('ZDV alone'); zidovudine/lamivudine/lopinavir-ritonavir ('ZDV-based ART'); or tenofovir/emtricitabine/lopinavir-ritonavir ('TDF-based ART'). The primary outcome was PTB less than 37 and less than 34 weeks based on the gold standard of ultrasound dating. We evaluated the association between antiretroviral strategy and PTB. We fit multivariable logistic regression models, adjusting for maternal characteristics, obstetric history, and HIV disease severity.

RESULTS

Among 720 assessed pregnant women, PTB less than 37 weeks was 15.4% by NBS and 18.3% by ultrasound. The NBS was specific but not sensitive for PTB less than 37 weeks (92.0% and 48.5%). Women receiving ZDV-based and TDF-based ART had significantly higher odds of PTB less than 37 by ultrasound compared with ZDV alone (adjusted odds ratios: 1.68; 95% confidence interval 1.10-2.57, and 2.71; 95% confidence interval 1.39-5.29), as well as for PTB less than 34 weeks. These results held for ultrasounds performed less than 24 weeks, and were generally consistent with prior analyses from the PROMISE trial using the NBS.

CONCLUSION

Women starting HIV ART in pregnancy remained at higher risk of PTB when determined by ultrasound, consistent with prior data using newborn exam. However, newborn exam misclassified cases of PTB compared with gold standard ultrasound.