Conflicting findings with HAART use in pregnancy and prematurity

Polly Clayden, HIV i-Base

Two oral abstracts presented conflicting and worrying findings on the association of prematurity and HAART use during pregnancy.

Dr Claire Thorne from the Institute of Child Health in London presented data from the European Collaborative Study in which HIV-positive pregnant women and their infants are followed up prospectively in 25 clinics in nine European countries [1].

As of January 2004, 4,377 HIV-positive women and their children had been enrolled in the study. Dr Thorne evaluated data from 1994 to 2003 and reported an increase in prematurity associated with more widespread use of HAART in pregnancy.

The authors found that during the period 1994 to1997, 54% of women had received ART during pregnancy. This increased to 88% in 1998 to 2000 and 90% in 2001 to 2003 (p<0.0001). Only 11% of women delivering in the earliest time period were already receiving ART when they became pregnant, increasing to 19% in 1998 to 2000 and 34% in 2001 to 2003 (p=<0.0001).

In the later period they also noted that there had been a decline in the rate of elective Caesarean section – generally seen to add no benefit with the use of complex therapy – from 73% in 1999 to 62% in 2002, and therefore a rise in vaginal deliveries.

During the period 1995 to 1997, 17.1% of infants were born prematurely (defined as before 37 weeks of gestation) which rose to 35.2% and 46.1% during 1998-2000 and 2001-2003 respectively (p=0.0001). There was also an increase in very premature infants (defined as before 34 weeks of gestation): 5.94%, 16.1% and 21.2% in the same three time periods respectively (p=0.0001).

Low (defined as below 2,500 grams) and very low (defined as below 1,500 grams) birth weight incidence also increased:

Low (<1,500g) Very low (<2,500g)
1995 to 1997 14.9% 0.48%
1998 to 2000 29.4% 4.7%
2001 to 2003 36.9% 8.05%

(trend p<0.0001)

The authors cited maternal age 35 years and above (OR 2.24, p=0.002), maternal IDU (OR 2.52, p=0.001), maternal CD4 below 200 cell/mm3 (OR 2.26, p=0.0001) and HAART use in pregnancy (OR 3.41, p=0.0001 without protease inhibitor and 4.17 p=0.0001 with protease inhibitor) as risk factors for prematurity.

Additionally they reported a high rate of neonatal mortality (defined as dying within 28 days of birth) from 1999 to 2000, 21.00 per 1,000; and from 2001 to 2002, 24.5 per 1,000. The range in the general population in Western Europe is 4.4 to 9.2 per 1,000. However, this finding is hard to interpret as data on pre-labour caesarean section were censored in this study, and general population data excluding pre-labour caesarean section is not available.

There were too few deaths to perform a multivariate analysis, but in univariate analysis the authors found prematurity to be associated with infant death: 74% who died were born prematurely vs 21% who survived (0.0001). Protease inhibitor containing HAART was moderately associated with an increased risk but the association did not reach statistical significance.

The authors concluded: “Although the benefits of HAART in reducing MTCT vastly outweigh the increased risk of premature delivery and possibly in perinatal and neonatal mortality, monitoring for adverse pregnancy outcomes should be recommended in ART programmes.”

A presentation in the same session from Dr Karen Beckerman from New York University used data from the Antiretroviral Pregnancy Registry (APR), an ongoing registry since January 1989 [2].

The objective of this study was to examine the association between prematurity, low and very low birth weight (defined differently from the European Collaborative Study: 32 weeks in the APR and 34 weeks in the ECS) and the use of combination ART. The authors compared women receiving monotherapy with those receiving combination ART. Additionally, protease inhibitor containing HAART was compared with combinations that did not.

Of the 3,782 evaluable pregnancies, 440(13%) received monotherapy, 1,368 (41%) received HAART with no protease inhibitor and 1,539 (46%) received protease inhibitor containing HAART. Dr Beckerman reported no significant differences in prematurity between the various treatment groups (see table elow).

Mono (N=440) Any Combo (N=2907)
Gestation Age
<37 weeks 55 (13%) 353 (12%)
<32 weeks 9 (2%) 57 (2%)
Birth Weight
<2500 g 57 (14%) 443 (16%)
<1500 g 7 (2%) 53 (2%)

However, the study did find a modest increase in low and very low birth weight among women receiving protease inhibitor containing HAART particularly in the first trimester: low birth weight – 258 (17%) vs 185 (14%) and very low birth weight – 34 (2%) vs 19 (1%) for protease inhibitor containing and not containing HAART respectively.

Dr Beckerman hypothesised that other risk factors such as viral load, disease stage and prior premature delivery, which could not be controlled in this analysis, could be responsible for this effect. She explained that women with more advanced HIV, may be more likely to receive a protease inhibitor (women with CD4 counts of less than 200 cells/mm3 were 2.8 times more likely, and women exposed in the first trimester were 2.4 times more likely to receive a protease inhibitor) and more likely to have low birth weight babies.


Neither study resolves this issue, but despite the differences in emphasis, both support earlier observations that antiretroviral therapy in pregnancy can increase obstetric risk for babies. This does not however diminish the considerable benefit of antiretroviral therapy for mothers and their babies.

The data from the ECS, which overstate the risk of pre-term delivery by excluding deliveries by pre-labour Caesarean section, need to be re-evaluated with all deliveries included (with the understanding that elective pre-term Caesarian sections will also bias the data).

Conversely the data from the Anti-retroviral Pregnancy Register may underestimate any effect, as only women exposed to any antiretroviral therapy are included.

In both studies an effect of antiretroviral therapy on severe pre-term delivery should be analysed according to the time of starting therapy ie including women starting therapy later than 32 weeks will reduce the likelihood of detecting a genuine effect of antiretrovirals on severe pre-term deliveries and severe low birth weight.


  1. Thorne C, Newell M, and European Collaborative Study. Pregnancy outcome in ART-treated HIV-infected women in Europe. 11th CROI 2004, Abstract 98.
  2. Beckerman K, Covington D, Garcia P et al. Association between antiretroviral therapy during pregnancy and prematurity/low birth weight. 11th CROI 2004, Abstract 98.

Links to other websites are current at date of posting but not maintained.