HTB

Pregnancy outcomes in Zambia

GlobePolly Clayden, HIV i-Base

Three presentations at 8th INTEREST were from Zambian studies looking at birth outcomes among women receiving antiretroviral treatment (ART) in pregnancy.

The first study found no significant association between ART duration and risk of low birth weight (LBW) <2500g.

This was a retrospective review of data from the Zambian Electronic Perinatal Record System (ZEPRS). The investigators looked at the association between timing of ART initiation and LBW in women who were ART-naive, eligible for treatment (CD4 count <350), and delivered singleton infants at >28 weeks at a health facility between 1 January 2009 and 1 September 2013.

They categorised duration on ART as <8, 9-20 or 21-36 weeks before delivery and compared these groups to women who were eligible but never started treatment.

To assess the effect of missing data, the investigators looked at predictors of missing ART initiation date and performed multiple imputation for missing dates. They used log-binomial regression to estimate risk ratios (RR) for the association between duration on ART and LBW and adjusted for multiple confounders (preterm delivery was not included as a confounder). Data on World Health Organisation (WHO) clinical stage and viral load were not available.

A total of 9,276 women met the inclusion criteria for the analysis: 5,746 (64%) never initiated ART, 1432 (16%) received ART for 1-8 weeks, 1672 (19%) for 9-20 weeks and 192 (2%) for 21-36 weeks.

Of women included in the analysis: 5,744 (62%) were missing confounder, exposure or outcome information; a further 2,154 (19%) reported being on ART at delivery, but had no start date.

There were 1,267 (14%) LBW infants – this was greatest in the ART for 1-8 weeks group, 235 (18.6%). In the complete case analysis (reference, never initiated ART and adjusted for: number of ANC visits, age, BMI, CD4 count, education, hemoglobin, malaria prophylaxis, parity, syphilis screening and tuberculosis status) RRs for 21-36 and 9-20 weeks on ART were respectively 0.48 (95% CI 0.21- 1.13) and 0.87 (95% CI 0.68-1.12). For 1-8 weeks RR was 1.21 (95% CI 0.97-1.51). No associations were statistically significant.

In the imputed data analysis, the investigators found that RRs for 21-36 and 9-20 weeks on ART moved closer to the null and the RR for 1-8 weeks on HAART increased and became significant (p-value not reported). They suggested this association might be due to the higher number of preterm births among women on HAART for 1-8 weeks (18.5% vs 12.3% HAART 9-20 weeks and 0.5% HAART 21-36 weeks).

The investigators noted that the limitations of the study included: understanding the impact of preterm birth, whether or not earlier initiation of ART was a marker for better health seeking behaviour and that the data was observational and collected in the previous Option A programme.

The second study also evaluated data from the ZEPRS and used regression discontinuity (a “quasi-experimental approach” that can minimise such confounding with observational data when the probability of being treated is dependent on an arbitrary threshold) to look at CD4 threshold for starting ART in pregnancy ART in pregnancy and possible associations between ART and birth outcomes. This analysis did not find significant associations.

The investigators identified newly diagnosed HIV-positive pregnant women in Lusaka, with CD4 counts 300 to 400 cells/mm3. They modelled the association of ART initiation in pregnancy with birth weight (BW), LBW, and stillbirth. They also conducted sensitivity analyses with wider and narrower CD4 windows around the threshold of 350 cells/mm3: +/-75 and +/-35 cells/mm3.

Between Jan 2009 and May 2013, 3,660 of 31,795 (12%) newly diagnosed pregnant women had CD4 counts of 300-400 cells/mm3, including 1,924 with 301-350 and 1,736 at 351-400 cells/mm3. When the women were stratified according to CD4 category, there were no not statistically differences in age, socioeconomic status, parity, gestational age at first ANC, and obstetrical risk factors.

The analysis revealed that women with CD4 300-350 cells/mm3 did not have worse birth outcomes, although they were over twice as likely to start ART vs those with CD4 351-375 cells/mm3: 37% vs 15%, p<0.001. Both the intention to treat and the as-treated analyses suggested that ART initiation is associated slight decrease in probability of LBW (-0.04, 95% CI -0.53 to 0.45 for as-treated) and of stillbirth (-0.13, 95% CI-0.38 to 0.13) and increase in BW (396 grams, 95% CI -345g to 1137g). None were significant and sensitivity analyses using wider and narrower CD4 ranges gave similar results.

The third study looked at a pilot programme that offered universal ART to HIV-positive pregnant and breastfeeding women at the Adult Infectious Disease Centre, University Teaching Hospital, Lusaka from 2008 to 2011. This programme extended beyond the HIV treatment guidelines at the time (ART for those with CD4 count < 350 cells/mm3 or WHO Stage 3 or 4). The data source for the analysis was also the ZEPRS.

This study described the characteristics associated with starting antenatal ART stratified by CD4 count (> or < 350 cells/mm3), and those associated with postnatal ART initiation. It also compared pregnancy and HIV outcomes between women who started antenatal ART and those who did not. Outcomes include: mode of delivery, infant birth weight and newborn vital status, Apgar scores, NICU admission, infant feeding method, and infant HIV status at 6 weeks and 6 months.

In this cohort, CD4 count <350 cells/mm3 was associated with increased ART initiation. Most pregnancy and HIV outcomes did not differ among women on and not on ART stratified by CD4 count.

The analysis included 353 women with pregnancy outcomes of which 70 (19.8%) initiated ART before pregnancy. The remaining 283 women were offered ART, 169 (59.7%) had a CD4 count <350 cells/mm3, so were eligible for ART according to Zambian guidelines. Of these, 144 (85.2%) started treatment. A further 114 women with CD4 count > 350 cells/mm3 were also offered ART, and 88 (77.2%) started treatment.

Of the 51 women who declined antenatal ART, 25 (49.0%) started after delivery. For women with CD4 count <350 cells/mm3, higher gravidity was the only characteristic associated with starting antenatal ART, p=0.04.

For women with CD4 count > 350 cells/mm3, the median maternal weight was significantly higher in those starting antenatal ART than those who did not: 71.0 vs 63.5 kg, p<0.05. For women who declined antenatal ART, a recent CD4 count <350 cells/mm3 was associated with starting postnatal ART, p<0.05.

Most pregnancy and infant outcomes were similar among women on and not on ART stratified by CD4 count. Most recent CD4 < 350 cells/mm3 preceding post natal initiation was associated with formula feeding, p<0.01. Fewer women who did not initiate antenatal ART formula fed than those who did: 13 vs 33%, p=0.05. Overall, the majority of women chose to breastfeed.

In this cohort women who were not treatment eligible were less likely to initiate ART even when it was offered to them. The investigators noted that messages about Option B+ to pregnant and breastfeeding need to be strengthened and this is a critical role for the community.

Comment

The data on outcomes presented here seem reassuring.

References:

  1. Bengtson A et al. Estimating the association between duration of HAART before delivery and low infant birthweight using data from the Zambian Electronic Perinatal Records System. 8th International Workshop on HIV Treatment, Pathogenesis and Prevention Research in Resource-Poor Settings (INTEREST). 5-6 May 2014. Lusaka, Zambia. Oral abstract O_10.
    http://regist2.virology-education.com/2014/8INTEREST/48_Hancock.pdf (PDF)
  2. Zanolini A et al. Using a regression discontinuity (RD) approach to investigate the effect of combination antiretroviral therapy (cART) in pregnancy on birth outcomes in Zambia. 8th International Workshop on HIV Treatment, Pathogenesis and Prevention Research in Resource-Poor Settings (INTEREST). 5-6 May 2014. Lusaka, Zambia. Oral abstract O_11.
  3. Liu KC et al. Characteristics and outcomes of HIV-infected pregnant women accepting combination ART for PMTCT in Zambia. 8th International Workshop on HIV Treatment, Pathogenesis and Prevention Research in Resource-Poor Settings (INTEREST). 5-6 May 2014. Lusaka, Zambia. Oral abstract O_12.
    http://regist2.virology-education.com/2014/8INTEREST/50_Liu.pdf (PDF)

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