One-in-four deaths in pregnancy due to HIV in worst-affected countries

This article is more than 8 years old. Click here for more recent articles on this topic

One-in-four pregnancy-related deaths is due to HIV in countries with a high HIV prevalence, a meta-analysis of 23 studies shows.

HIV-infected women have eight times the risk of a pregnancy-related death compared to uninfected women (pooled relative risk [RR]: 7.75, 95% CI: 5.37-11.16), according to results from the meta-analysis published in the advance online edition of AIDS.

Researchers from the London School of Hygiene and Tropical Medicine found that, at the population level, a very high proportion of pregnancy-related deaths can be attributed to HIV. This means the considerable proportion of excess deaths due to HIV (mostly in the absence of ART) among HIV-infected pregnant and postpartum women (pooled attributable risk [AR]: 994 per 100,000 pregnant women, 95%, CI: 677-1310) has a significant impact on all causes of pregnancy-related deaths even in areas where HIV prevalence is low.



When the statistical data from all studies which relate to a particular research question and conform to a pre-determined selection criteria are pooled and analysed together.

case-control study

An observational study in which a group of people with an infection or condition (called ‘cases’) are compared with a group of people without the infection or condition (called ‘controls’). The past events and behaviour of the two groups are compared. Case-control studies can help us understand the risk factors for having an infection or a condition. However, it is difficult both to accurately collect information about past events and to eliminate bias from case-control studies.


Relating to pregnancy, childbirth and the first few weeks after birth.


Studies aim to give information that will be applicable to a large group of people (e.g. adults with diagnosed HIV in the UK). Because it is impractical to conduct a study with such a large group, only a sub-group (a sample) takes part in a study. This isn’t a problem as long as the characteristics of the sample are similar to those of the wider group (e.g. in terms of age, gender, CD4 count and years since diagnosis).

sample size

A study has adequate statistical power if it can reliably detect a clinically important difference (i.e. between two treatments) if a difference actually exists. If a study is under-powered, there are not enough people taking part and the study may not tell us whether one treatment is better than the other.

“In areas where HIV prevalence among pregnant women is as low as 2%, 12%  of all pregnancy-related deaths may be attributable to HIV. This figure rises to 50% in areas with an HIV prevalence of 15%”, the authors write.

Based on 2011 UNAIDS’ estimated HIV prevalence rates among adults of reproductive age, an estimated 5% of all pregnancy-related deaths worldwide, and one in four (25%) in sub-Saharan Africa, are attributable to HIV.

These findings, they note, highlight the importance of integrating HIV and reproductive health services and monitoring trends in maternal mortality.

While HIV is the leading cause of death among women of reproductive age worldwide, women in sub-Saharan Africa also experience the highest rates of maternal death.

The effect of HIV on pregnancy and vice versa is poorly understood. Evidence is lacking to support the argument that pregnancy increases HIV progression and conversely that the risk of obstetric complications among HIV-infected women is increased.

While much is known, the authors write, about the contribution of HIV to adult mortality, little is known about how HIV contributes to mortality during pregnancy and the postpartum period.

The authors cite two approaches used to estimate the proportion of maternal deaths attributable to HIV. The first, a systematic review of the causes of maternal deaths, based on only eight studies and 'verbal autopsies' (a research method that helps determine probable causes of death in cases where there is no medical record, relying on reports from family members or others)  with no defined criteria for classifying a maternal death as HIV/AIDS-related, attributed 6.2% of maternal deaths to HIV in 2006 in Africa.

The second approach uses mathematical models in the absence of empirical data. Two models dominate, offering vastly different estimates for 2008, the latest year both models provided estimates. While the Institute for Health Metrics and Evaluation estimated 17.9% of maternal deaths worldwide were attributable to HIV, the Maternal Mortality Estimation Inter-agency Group model gave an estimate of only 5.9%.

The main difference between the models, the authors note, is in the assumptions made about the number of deaths among HIV-infected pregnant and postpartum women attributed to pregnancy and so classified as maternal deaths. In the former, all are classified as maternal, while in the latter only half.

The authors proposed an alternative approach.

Basing calculations on empirical data from a systematic review of studies comparing death during pregnancy and the postpartum period in HIV-positive and negative women, they reported on the risk ratio and prevalence of HIV.

Eligible studies included those comparing death during pregnancy, delivery and/or up to 365 days postpartum between HIV-positive and negative women using a cohort, census or case-control study design. Death could be defined as “pregnancy-related” (including all deaths) or “maternal” (excluding deaths which were accidental or incidental to the pregnancy). Only studies where HIV status was determined by HIV testing and which had a sample size of at least 30 women in each study group were included.

Summary estimates of relative (RR) and attributable risks (AR) for the link between HIV and death during pregnancy and the postpartum period were calculated through meta-analyses.

The authors predicted the effect of HIV on pregnancy-related death at the population level by calculating population-attributable fractions for each study individually and in scenarios with varying HIV prevalence using the pooled RR from the meta-analysis.

Of the 18,949 potentially relevant articles identified, 17,640 were excluded through abstract and title screening. Twenty-three studies out of 1291 full texts had data on the risk of pregnancy-related death in HIV-positive and negative women.

Study populations were from South Africa, Tanzania, the Republic of Congo (Congo-Brazzaville), the Democratic Republic of Congo, Malawi, Zimbabwe, Rwanda, Uganda, Kenya, India, Spain, USA and Mexico.

Excess mortality due to HIV among HIV-infected pregnant and postpartum women, the authors note, is not surprising. Most women were not on ART, and many would be at an advanced stage of illness resulting inan increased risk of death. However, the extent of the excess is higher than expected.

The authors believe their approach has two main advantages over previous studies. First, their findings are based on empirical, not modelled, data. Their summary estimate is based on 23 studies from across the world. However, heterogeneity between the studies suggest the results be interpreted with caution. So the summary estimate should be considered “an average RR about which the true study RRs actually vary”.

Second, estimation of the contribution of HIV to pregnancy-related death, not maternal death, means no assumptions need be made about HIV being indirectly related or coincidental to pregnancy.

Most pregnancy-related deaths are among HIV-infected women in sub-Saharan Africa where verbal autopsy is routinely used to report cause of death.

The authors cite a recent WHO document suggesting deaths in HIV-infected pregnant and postpartum women be categorised into direct obstetric deaths, “AIDS-related indirect maternal deaths” (those who die because of effect of pregnancy on HIV), and “HIV-related deaths” (who die of a fatal complication of HIV or AIDS that is coincidental to the pregnancy). Distinguishing AIDS-related indirect deaths from HIV-related coincidental deaths is limited and no guidance is given.

Severe anaemia and tuberculosis, note the authors, can be considered as both causes of indirect maternal deaths and HIV-related deaths. They suggest reporting pregnancy-related – and not maternal – death will resolve the issue and allow for more reliable monitoring of causes.

These findings, conclude the authors, have important implications for integrated service delivery.

Future research, they add, “should focus on how to identify HIV-related deaths using verbal autopsies, so that pregnancy-related mortality can be monitored including and excluding HIV-related deaths”.


Calvert C, Ronsmans C The contribution of HIV to pregnancy-related mortality: a systematic review and meta-analysis. AIDS 27, advance online edition, doi: 10.1097/qad.0b013e32835fd940, 2013.