Increased risk of adverse pregnancy outcomes in HIV-infected women treated with highly active antiretroviral therapy in Europe

Abstract

Highly active antiretroviral therapy (HAART) may be associated with adverse pregnancy outcomes. Among 4372 live births in the European Collaborative Study, the prematurity rate increased to 24.9% in 2000–2004. Antenatal HAART use initiated pre-pregnancy was strongly associated with prematurity (AOR 2.05, 95% CI 1.43, 2.95), particularly severe prematurity. The implication of increased prematurity is evidenced in high neonatal mortality in these groups (0.66% for infants at 34–36 weeks and 7.37% at < 34 weeks’ gestation). HIV-infected women taking highly active antiretroviral therapy (HAART) could face possible adverse events in pregnancy. Although we and other European observational studies have shown a substantially increased risk of premature delivery associated with antenatal combination antiretroviral therapy [1–3], which was not seen in a study in the USA [4], this was generally felt to be clinically manageable. There is no information documenting the consequences of premature delivery for the infants. Increasing numbers of infected women in Europe are taking highly potent, complex combinations of drugs throughout pregnancy, often initiated before or in early pregnancy. The impact of these changes in HIV treatment on pregnancy outcome is unclear, but we recently noted a worrying increase in severe pregnancy-related adverse events, including neonatal deaths, which are described here. Since 1986 HIV-infected women identified before or during pregnancy and their infants are followed according to standard clinical and laboratory protocols, in the ongoing European Collaborative Study [2,3]. We define prematurity and severe prematurity as delivery before 37 and 34 weeks’ gestation, respectively, neonatal death as death within the first 28 days of life, and perinatal mortality as stillbirths and deaths of live born infants within the first week of life. Data entry was carried out using MS Access 2000 and analyses using SAS statistical software (v8.02, SAS Institute, Cary, NC, USA). There were 4372 live births to women enrolled between 1986 and the end of April 2004, of whom 3131 (74%) were white, 914 (22%) were black and 171 (4%) were of other ethnicity. The median maternal age at the time of delivery was 28.3 years (range 10–47). The overall crude prematurity rate was 18.9% (828/4372), with a significant recent increase from 16.4% (136/832) in 1985–1989, 15.7% (187/1192) in 1990–1994, and 18.0% (205/1142) in 1995–1999 to 24.9% (300/1206) in 2000–2004 (χ2 = 30.1, P < 0.002). Similar trends were observed in the prevalence of low birthweight and very low birthweight. The rate of elective caesarean section in the European Collaborative Study was high (65.4%, 783/1198, in 2000–2004), reflecting its use in the prevention of mother-to-child transmission [5]. The median gestational age among infants delivered by elective caesarean section was 38 weeks, although some centres had a policy of scheduling elective caesarean section at 36 weeks [5]. The increase in prematurity among women delivering by emergency caesarean section or vaginally (thus with a ‘spontaneous’ delivery with rupture of the membranes or initiation of labour) was from 16.1% (108/670) in 1985–1989, 15.0% (147/982) in 1990–1994, to 20.0% (118/591) in 1995–1999 and 37.1% (154/415) in 2000–2004 (χ2 = 66.4, P < 0.002). The prematurity rate among women taking monotherapy (predominantly zidovudine) in pregnancy was 16.8% (118/704), 13.4% (34/254) for dual therapy, and 25.5% (274/1075) for HAART (χ2 = 26.6, P < 0.002). In univariable and multivariable logistic regression analysis, HAART in pregnancy, particularly when initiated before pregnancy, was strongly predictive of prematurity (Table 1). Taking severe prematurity as the outcome gave adjusted odds ratios (AOR) of a similar size for most explanatory variables, but a stronger association with HAART [AOR 2.50, 95% confidence interval (CI) 1.19–5.24, P = 0.016 when initiated during pregnancy, and AOR 4.41, 95% CI 2.06–9.41, P < 0.002 when inititated pre-pregnancy]. Although adjusting for unaccounted centre-associated variation through a variable for random effect at the centre level improved the model's goodness of fit, the AOR for the variables of interest here remained at a similar magnitude. A sub-analysis limited to 1275 women with emergency caesarean section or vaginal deliveries identified the same risk factors for premature delivery as above, although the HAART-associated risk was more pronounced (HAART started antenatally: AOR 3.25, 95% CI 1.99–5.31, and pre-pregnancy: AOR 4.00, 95% CI 2.26–7.08).Table 1: Risk factors for delivery before 37 weeks’ gestation in 2279 mother–child pairs.To assess the clinical consequences of this increase in prematurity and low birthweight we investigated neonatal and perinatal deaths. There were 28 neonatal deaths (eight in the most recent period); with a median gestation of 31 weeks (range 22–40) and median birthweight of 1213 g (range 500–3940). The neonatal morta