Efavirenz use in pregnancy and birth outcomes

Polly Clayden, HIV i-Base

Of all the antiretrovirals, efavirenz attracts the most controversy with regard to use in pregnant women.

Two posters at CROI looked at birth outcomes among mothers who received efavirenz in pregnancy. [1, 2]

Daniel Westreich and colleagues analysed prospective data from Themba Lethu Clinic in Johannesburg. They examined the risk factors for pregnancy after initiation of HAART using Cox proportional hazards regression and looked at birth outcomes in women receiving efavirenz.

The investigators reported that of 5011 women initiating HAART between April 1 2004 and March 31 2007, 351 (7%) became pregnant giving a rate of 4.1 pregnancies per 100 woman years (95% CI 3.7–4.5).

They found that this was less among older women (>35 years) and those with lower CD4 counts or employed.

The investigators noted that although women who initiated HAART with efavirenz based regimens were less likely to become pregnant that than those starting with nevirapine, HR 0.6 (95% CI 0.4–0.8), 68% (n=238) of pregnancies occurred in women receiving efavirenz.

The investigators looked at 136 of the women receiving efavirenz for pregnancy outcome. They reported three maternal deaths, 39 women lost to follow up and one who refused to be interviewed. In addition 12 women were still pregnant at the time of analysis.

Out of 81 pregnancies analysed, 8 were voluntarily terminated and 13 miscarried. Of the remaining 60 live births, 41 were examined using the Denver Development Screening Test. According to this scale, 30 infants were classified as “within normal limits” and 11 (>25%, 95% CI 14–43) were  “suspect” for developmental delay. The investigators did not find any congenital abnormalities among the infants examined.

The investigators rightly suggested that further study including a comparison group is required to evaluate whether suspected neurodevelopmental delays or miscarriages are associated with efavirenz use. And they wrote, “These results suggest that the risk of efavirenz in pregnancy may be less catastrophic than feared”.

A related poster authored by Daniel Conway and colleagues reported findings from an analysis of prevalence of congenital abnormalities among antiretroviral-exposed infants in IMPAACT P1025.  This US cohort study enrolled all women and infants in P1025 born between 2002 and 2007 (n=1306) during pregnancy or up to two weeks post-partum.  No information was provided regarding prospective or retrospective enrollment, that is, whether any women/infants could be enrolled after an anomaly had already been diagnosed.

A total of 1112 infants were eligible for analysis. Infants were examined for congenital anomalies and reviewers were blinded to in utero antiretroviral (individual drugs and classes) exposure. In utero exposure was classified as 1st, 2nd/3rd trimester and unexposed. The investigators used logistic regression to estimate the relationship between congenital anomalies and antiretroviral exposure.

The investigators reported 80 congenital anomalies among 61 infants, a rate of 5.49/100 live births (95%CI 4.22–6.99). These were broken down into organ systems involved: cardiovascular (n=33), genitourinary (n=7), renal (n=8), musculoskeletal (n=15, including accessory digits of hand or foot, n=7), craniofacial (n=3), central nervous system (n=3), chromosomal (n=3), eye (n=3), gastrointestinal (n=5).

This study reports that first trimester exposure to efavirenz was associated with a significantly increased risk of congenital anomalies (OR 2.89; 95%CI 1.15–7.25) compared to 2nd/3rd trimester exposure. No information was provided regarding the types of anomalies associated with efavirenz exposure.

They did not find any other classes of antiretrovirals nor individual drugs (only data for efavirenz and AZT were shown) to be associated with increased risk of anomalies. Covariates including ethnicity, maternal age and substance use (tobacco, alcohol and recreational drugs) were not significantly associated with increased risk of congenital anomalies.


Because of the uncertainties about efavirenz use in pregnancy, all new information on this subject is worth reporting. 

Unfortunately neither of the posters summarised here showed enough data to make their interpretation straightforward and both studies seem underpowered. 

The Themba Lethu Clinic analysis does not report baseline data and has high loss to follow up. Only the abstract is on the conference website so it was not possible to double-check these data.  

The P1025 numbers are also small, about 1/10 of the APR, including those with efavirenz first trimester exposure (47 vs 501). [3] They do not report what the 6 anomalies with first trimester efavirenz exposure were, which seems very important in view of the monkey data.  The only other antiretroviral they report on individually is AZT.  

Some of the P1025 data are already in the APR though it is not clear which, and not all women were enrolled into this study before pregnancy outcome was known which would make them ineligible for the APR. Retrospective observations of birth defects cannot be used to determine birth defect prevalence, because of differential reporting bias, that is, that a foetus or infant with a birth defect would be more likely to be enrolled than one with an unknown or normal outcome.

Good data to guide recommendations in this area are urgently needed including other important outcomes such as spontaneous abortion and termination of pregnancy. 


Unless stated otherwise, all references are to the Programme and Abstracts of the 17th Conference on Retroviruses and Opportunistic Infections. 16-10 February 2010, San Francisco.

  1. Westreich D et al. Pregnancy, efavirenz, and birth outcomes in Johannesburg. 17th CROI, San Francisco, 2010.  Poster abstract 922.
  2. Conway D et al. Prevalence of congenital anomalies in infants with in utero exposure to antiretrovirals: IMPAACT P1025. 17th CROI, San Francisco, 2010.  Poster abstract 923.
  3. Antiretoviral pregnancy registry.

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