A Systematic Review of Interventions to Reduce Maternal Mortality among HIV-Infected Pregnant and Postpartum Women

Review Design

Following the PRISMA reporting framework, we conducted a systematic review to identify, analyze, and assess the effect of interventions that prevent death among HIV-infected women during pregnancy and up to one year postpartum.

Definitions

For our review, we define an intervention as a set of actions with a coherent objective to bring about change or produce identifiable outcomes.^[8] We searched for interventions that were applied in clinical or public health settings. Public health interventions are intended to promote or protect health or prevent ill health in communities or populations. They are distinguished from clinical interventions, which are intended to prevent or treat illness in individuals.^[8] ART is defined by the World Health Organization (WHO) as “treatment of people infected with human immunodeficiency virus (HIV) using anti-HIV drugs.” The standard treatment consists of a combination of at least three drugs (sometimes called “highly active antiretroviral therapy,” or HAART) that suppress HIV replication. Three drugs from at least two different drug classes are used in order to reduce the likelihood of the virus developing resistance. ART has the potential both to reduce mortality and morbidity rates among HIV-infected people, and to improve their quality of life.^[9] Therefore, when this review references a study on ART, it refers to the triple therapy regimen described; when it describes studies using HIV therapies with fewer than three drugs, such as single dose nevirapine, we will refer to the medicine by its name.

Study Eligibility

Search Strategy and Selection Process

Study selection

A multi-step strategy was followed to select studies. First, all studies not written in English were excluded. Then, records were excluded based on a title and abstract review. Next, two authors independently read and reviewed full-text articles, compared decisions about inclusion, and resolved discrepancies. Articles were excluded if they did not involve HIV- infected pregnant or postpartum women, did not describe an intervention, or did not include mortality as an endpoint. Figure 1 presents the PRISMA flow diagram for our review.

Close

Figure 1.

PRISMA Flow Diagram. From: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009). Preferred Reporting/terns for Systematic Reviews and Meta-Analyses: The PRISMA Statement. PLoS Med 6(6): e1000097. doi:10.1371/journal. pmed1000097

Export to PPT

Data Extraction and Quality Assessment

Overview of Included Studies

We identified 1,373 records for review. We eliminated l,345 articles during the title and abstract review. We assessed 28 full-text articles for eligibility and excluded nine. See Figure 1 for details of the search.

Nineteen studies met the inclusion criteria for this systematic review. Fourteen were related to the provision of ART, while five addressed the effect of vitamins, micronutrients, and antibiotics on maternal and postpartum mortality. Detailed characteristics of the included studies are summarized in Table 1. We divided review findings into two categories: (1) findings related to ART interventions; and (2) findings related to non-ART interventions. In many cases, studies included important contextual information that requires consideration in the interpretation of their findings.

Findings Related to ART Interventions

Overview of ART studies

The 14 ART studies that met our inclusion criteria focused on a diverse group of study objectives and span a period of more than two decades. Studies assessed the effectiveness of ART in prevention of mother-to-child transmission (PMTCT) of HIV;^[11-13] pregnancy outcomes in women on ART;^[14-17] the clinical impact of postpartum ART discontinuation when used for PMTCT;^[18-20] the impact of incident pregnancy in women on ART for their own health;^[17,21,22] safety and efficacy of drugs, including comparisons of dual versus triple therapy regimens and nucleoside reverse transcriptase inhibitors versus protease inhibitor-based ART;^[16,20,23] and risk factors for maternal mortality.^[24]

Study designs included nine cohort studies (four retrospective^{[14,15,17,18]} and five prospective ^{[12,16,21,22,24]}), three randomized control trials, (RCTs)^[13,19,20] one single arm trial,^[23] and one not stated.^[11] Three studies were conducted in the United States,^[14,16,18] and eleven were conducted in one or more countries in sub-Saharan Africa.^{[11-13,15,17,19-24]}

The timing of ART initiation varied according to study objectives and design. Women started ART at different times during the pre-pregnancy, pregnancy, and the intrapartum periods. One study reported initiation of ART exclusively prior to pregnancy;^[21] four prior to or during pregnancy;^{[14,17,22,24]} three before the end of the second trimester;^[12,15,20] and four studies during the third trimester, including the intrapartum period.^{[11,13,19,23]} Timing of ART initiation during pregnancy is unspecified in two instances^[14,18] and occurred anytime during pregnancy in one study.^[16] Figure 2 depicts timing of ART by study, in detail.

Close

Figure 2.

Timing of ART Initiation and Duration of ART Provision by Study*

Export to PPT

Similarly, ART cessation and long-term continuation varied between studies, according to study objectives. In two studies that focused on PMTCT, ART was provided for the duration of breastfeeding: 24^[12] or 28 weeks postpartum.^[13] In one instance, single dose-nevirapine was provided during the intrapartum period only.^[11] In the studies that aimed to measure outcomes of HIV-infected women on ART, postpartum ART discontinuation occurred at four weeks postpartum^[19] (nonbreastfeeding arm), 24 weeks,^[15,20,23] or 26 weeks^[19] (breast feeding arm). Four studies describe the provision of long-term ART for the mothers’ own health.^[21,22,24] In one study, women taking ART for PMTCT stopped at 26 weeks and those who needed ART for their own health took it long term.^[17] The timing of ART discontinuation was not specified in three studies.^[14,16,18]

Where documented, the duration of the tracking of death as an outcome varied. Three studies tracked mortality to 42 days postpartum based on the standard definition for the maternal mortality ratio.^[15,17,24] Three studies tracked outcomes to 48 weeks or one year postpartum,^[12,13,22] four studies for 24 months postpartum,^{[11,19,20,23]} one study up to four years post-delivery,^[17] and three for the duration of the study.^[14,18,21]

Key findings

ART provided during pregnancy reduces maternal mortality

Of the four studies that analyzed the relationship between ART and maternal mortality, all found that ART reduces maternal mortality.^{[15,17,22,24]} One study was able to estimate a 34% decreased risk of maternal mortality among women on ART at delivery compared to those not on ART.^[24]

ART's impact on maternal mortality is modified by HIV disease stage and CD4+ count

Certain studies measured women's CD4+ counts or assessed their HIV disease stage to determine whether ART affects the risk of maternal mortality at different CD4 counts or disease stages. A Ugandan study suggests that there is an increased risk of mortality in the pregnancy and postpartum period among HIV-infected women initiating ART at advanced disease stages, particularly during early ART.^[22] This finding is supported by additional studies.^[15,17,24] One study found that ART reduces maternal mortality at all CD4+ levels with at least a two-fold greater reduction in mortality among women with CD4+ counts above 350 cells/mm3 compared to those below 350 cells/mm3.^[15]

Longer duration of ART during pregnancy is associated with a reduction in maternal mortality

Four studies found a reduction in maternal mortality associated with longer duration of ART.^{[15,17,22,24]} Marazzi found a 70% reduction in the risk of maternal mortality among women who received more than 30 days of ART prior to delivery compared to those with less than 30 days.^[15] Liotta found the shortterm mortality of women taking ART for less than 30 days was 2.2% compared to those receiving 31 to 90 days of ART (1.1%) and 91 to 270 days of ART (0.6%).^[17] In the same study, maternal mortality was higher with shorter antenatal ART, regardless of w9hether the deaths took place over shorter (< 42 days post-delivery) or longer term (up to four years post-delivery).^[17] More specifically, with respect to long-term mortality (up to four years post-delivery), mortality of women who received less than a month of ART was over two times higher (4.5%) than women who received three or more months of ART (1.8%; p<0.0001).^[17] Li estimated that initiating ART before pregnancy was associated with a 55% decreased risk of maternal mortality, and the risk of mortality decreases by 8% for each additional month on ART.^[24]

Discontinuation of ART for PMTCT leads to disease progression and mortality

Two studies reported on disease progression (including maternal mortality) after discontinuation of ART for PMTCT. In the study by Shapiro, most deaths in the randomized groups occurred after stopping ART.^[20] Eight deaths occurred during six to 24 months postpartum (1.0/100 person years), whereas only one death occurred between ART initiation and cessation at six months postpartum.^[20] Furthermore, among the eight deaths that occurred after ART cessation, three had started taking ART again but five had not. The authors calculated an absolute mortality risk of 1.4% in the 18 months following ART cessation, though not statistically significant.

The Kesho Bora study aimed to determine whether stopping ART once the infant is weaned affects 18 to 24 month maternal HIV disease progression.^[19] In this study, progression was defined as WHO clinical stage 4, CD4+ count less than 200 cells/mm3, or death.^[19] Postpartum HIV disease progression risk after ART discontinuation was 15.0% (two deaths, two with WHO stage 4 disease and 39 with at least one CD4+ count less than 200mm3). HIV disease progression varied by CD4+ count. Among women with counts of 200 to 349 cells/mm3 at enrollment, 24.0% (95% confidence interval [CI], 15.7-35.5) were found to have progressed. At counts above 350 cells/mm3, 4.9% (95% CI, 1.6-14.5) progressed.

Incident pregnancy in women on ART for their own health does not have adverse consequences for the mother

Four studies commented on incident pregnancies,^{[14,17,21,24]} that is pregnancies that occurred in a cohort of women taking ART for their own health. Westreich found that incident pregnancy was not associated with an increased hazard and risk of death.^[21] Li estimated that women initiated on ART before pregnancy had a statistically significant 55% decreased risk of maternal mortality.^[24] And Liotta found that mortality within 42 days of delivery was lower among women who started ART prior to pregnancy compared to those who started during pregnancy.^[17]

Two studies described the relationship between incident pregnancy in women on ART and adherence to treatment and abandonment of care. Westreich found that incident pregnancy was associated with a reduced hazard of becoming lost to follow-up (HR=0.62, CI. 0.51, 0.75) compared to the population of women on ART (not pregnant).^[21] Liotta found that loss to follow-up before delivery in the PMTCT group was 9.8% versus 3.9% in women on established ART (p<0.001); abandonment of care was more frequent in the PMTCT group (7.8%) as opposed to women on established ART (5.4%) (p<0.001).^[17]

Findings Related to Non-ART Interventions

Multivitamins, vitamin A, and selenium have a limited or no beneficial effect on maternal mortality

Two studies of vitamin A supplementation met the study inclusion criteria. A randomized controlled trial in urban Tanzania that included four arms (vitamin A, multivitamin without vitamin A, multivitamin plus vitamin A, and placebo) found that the women in the multivitamin group were less likely to progress to stage 4 disease or death compared to the women who received a placebo.^[25] In the randomized control trial study conducted in urban Zimbabwe, a single dose of vitamin A given postpartum had no effect on maternal mortality.^[26] A randomized controlled trial of selenium supplementation in antenatal clinics in urban Tanzania found that selenium supplements had no significant effect on maternal mortality.^[27]

Antibiotics had no effect on maternal mortality

We found two antibiotics studies that met the study inclusion criteria. The randomized controlled trial in Malawi, Tanzania, and Zambia found that erythromycin, ampicillin, and metronidazole given during pregnancy had no effect on maternal mortality.^[28] Likewise, in the double-blind placebo controlled trial in Zambia, cotrimoxazole given after childbirth did not reduce maternal mortality.^[29]

Limitations

Our review has some limitations. Maternal mortality remains a relatively rare event, even in the presence of an extensive HIV epidemic. Our focus on maternal mortality rather than maternal morbidity may have excluded potential interventions that impact mortality, especially when research was focused on morbidity outcomes. The review therefore does not address important questions highlighted in studies that have explored the relationship between ART and pre-eclampsia/eclampsia, TB, and malaria.^[34-39]

Selected studies targeted a diverse set of objectives and study outcomes. The majority of studies did not include maternal mortality as a primary outcome, and in certain studies, mortality was included as part of the descriptive statistics without further analysis or comment.^[11,13] The limited number of studies reduces the basis from which conclusions can be drawn. Since 2009, the number of HIV-infected pregnant women receiving ART for their own health has grown from 25% to 60% in 2012.^[2] Most studies identified through the review were initiated and concluded prior to this rapid period of expansion. It is likely that research initiated during the period of expanded coverage of ART will generate more concrete evidence about the contribution of ART to reduced maternal mortality.

Finally, this review covered only English-language studies and those published in PubMed. It is possible that there are relevant studies published in languages other than English and/or published in other sources of biomedical literature.

Ethical Review:

This study is based on the review of existing studies and biomedical database.