Prevention of mother-to-child transmission of HIV: a cross-sectional study in Malawi

Abstract Objective To estimate the use and outcomes of the Malawian programme for the prevention of mother-to-child transmission (MTCT) of human immunodeficiency virus (HIV). Methods In a cross-sectional analysis of 33 744 mother–infant pairs, we estimated the weighted proportions of mothers who had received antenatal HIV testing and/or maternal antiretroviral therapy and infants who had received nevirapine prophylaxis and/or HIV testing. We calculated the ratios of MTCT at 4–26 weeks postpartum for subgroups that had missed none or at least one of these four steps. Findings The estimated uptake of antenatal testing was 97.8%; while maternal antiretroviral therapy was 96.3%; infant prophylaxis was 92.3%; and infant HIV testing was 53.2%. Estimated ratios of MTCT were 4.7% overall and 7.7% for the pairs that had missed maternal antiretroviral therapy, 10.7% for missing both maternal antiretroviral therapy and infant prophylaxis and 11.4% for missing maternal antiretroviral therapy, infant prophylaxis and infant testing. Women younger than 19 years were more likely to have missed HIV testing (adjusted odds ratio, aOR: 4.9; 95% confidence interval, CI: 2.3–10.6) and infant prophylaxis (aOR: 6.9; 95% CI: 1.2–38.9) than older women. Women who had never started maternal antiretroviral therapy were more likely to have missed infant prophylaxis (aOR: 15.4; 95% CI: 7.2–32.9) and infant testing (aOR: 13.7; 95% CI: 4.2–83.3) than women who had. Conclusion Most women used the Malawian programme for the prevention of MTCT. The risk of MTCT increased if any of the main steps in the programme were missed.


Introduction
Recent global efforts in the fight against human immunodeficiency virus (HIV) have been focused on the virtual elimination of paediatric HIV infection 1 and improvements in the so-called cascade of care for the prevention of mother-to-child transmission of HIV (PMTCT). This cascade includes antenatal HIV testing, uptake of maternal antiretroviral therapy (ART), infant antiretroviral prophylaxis and early infant HIV testing. 2 In 2011, Malawi adapted the relevant World Health Organization (WHO) guidelines to design a national, publichealth-oriented strategy for PMTCT. 3,4 In this strategy, called Option B+, all pregnant and breastfeeding women found infected with HIV are offered lifelong ART, regardless of their CD4+ T-lymphocyte counts or WHO clinical stage. The strategy's development was supported by emerging evidence that universal ART provision in resource-constrained settings could markedly reduce HIV transmission. [5][6][7] The strategy was designed to remove institutional barriers to care, provide maternal health benefits, reduce maternal mortality and increase ART coverage for future pregnancies in high-fertility settings. 4 During the strategy's early implementation, ART initiations among pregnant women increased sixfold and the proportions of women who, having initiated ART while pregnant, were still receiving HIV care 12 and 24 months later were 72% and 68%, respectively. 8,9 Although a few studies in large Malawian health facilities have indicated that many mother-infant pairs miss one or more of the four main steps in the cascade of care provided, for PMTCT, by the national HIV programme, the results of such studies cannot be considered nationally representative. 10,11 The summary reports produced by Malawi's national HIV programme use data that are aggregated at health-facility level and do not provide insight into individual risk factors associated with the use of the cascade of care for PMTCT. In 2014, the Malawi Ministry of Health began a national evaluation of Malawi's PMTCT programme. Here we present some of the methods, results and conclusions of that evaluation, which aimed to determine the effectiveness of the Option B+ strategy in a nationally-representative sample of mother-infant pairs enrolled at 4-26 weeks postpartum.

Study setting
Implementation of the Malawi integrated PMTCT/ART guidelines began in July 2011. In theory, this gave all pregnant and breastfeeding women access to HIV testing, HIV counselling and ART. At the time of HIV status ascertainment, each HIVinfected pregnant woman should be given enough nevirapine to provide her baby with six weeks of prophylaxis from birth. She should also be asked to bring her child, for HIV testing, to a clinic for the care of children younger than five years, known as an under-5 clinic in Malawi, as soon as the course of prophylaxis is complete at an age of six weeks. 4 Objective To estimate the use and outcomes of the Malawian programme for the prevention of mother-to-child transmission (MTCT) of human immunodeficiency virus (HIV). Methods In a cross-sectional analysis of 33 744 mother-infant pairs, we estimated the weighted proportions of mothers who had received antenatal HIV testing and/or maternal antiretroviral therapy and infants who had received nevirapine prophylaxis and/or HIV testing. We calculated the ratios of MTCT at 4-26 weeks postpartum for subgroups that had missed none or at least one of these four steps. Findings The estimated uptake of antenatal testing was 97.8%; while maternal antiretroviral therapy was 96.3%; infant prophylaxis was 92.3%; and infant HIV testing was 53.2%. Estimated ratios of MTCT were 4.7% overall and 7.7% for the pairs that had missed maternal antiretroviral therapy, 10.7% for missing both maternal antiretroviral therapy and infant prophylaxis and 11.4% for missing maternal antiretroviral therapy, infant prophylaxis and infant testing. Women younger than 19 years were more likely to have missed HIV testing (adjusted odds ratio, aOR: 4.9; 95% confidence interval, CI: 2.3-10.6) and infant prophylaxis (aOR: 6.9; 95% CI: 1.2-38.9) than older women. Women who had never started maternal antiretroviral therapy were more likely to have missed infant prophylaxis (aOR: 15.4; 95% CI: 7.2-32.9) and infant testing (aOR: 13.7; 95% CI: 4.2-83.3) than women who had. Conclusion Most women used the Malawian programme for the prevention of MTCT. The risk of MTCT increased if any of the main steps in the programme were missed.

Research
Prevention of mother-to-child HIV transmission in Malawi M van Lettow et al.

Study design and sampling
Our aim was to draw a representative sample for national estimates of the ratios of mother-to-child transmission of HIV (MTCT) in Malawi. The sampling frame included all 579 health facilities that provided PMTCT services in Malawi in 2012-2013. We estimated that we would need to enrol at least 3376 HIV-exposed infants to determine the ratio of MTCT at 24 months postpartum reliably. Probability-proportional-tosize selection was used, without replacement, to select the 54 study facilities: 14 rural and nine urban facilities in the North or Central regions and 22 rural and nine urban in the South region. We subjected data obtained at all 54 study facilities to a cross-sectional analysis.

Data collection and laboratory procedures
Between October 2014 and May 2016, women attending under-5 clinics at each of the study facilities were screened for study eligibility. To be enrolled, a woman had to be a mother of or a legal caregiver for an infant aged 4-26 weeks and be willing and able to give informed consent. Information about age, parity, uptake of antenatal care, HIV testing and whether the woman's HIV status had been ascertained during or before the index pregnancy, if ever, was collected in standardized interviews. Whenever possible, interviewers checked the mothers' health booklets to check the accuracy of the mothers' responses. Women who had only discovered that they were HIV-infected through study screening were not asked about their uptake of maternal ART, infant nevirapine prophylaxis or infant HIV testing. After being interviewed, each enrolled woman was tested, within the study facility, for HIV. Maternal HIV testing, which was based on an initial rapid Determine HIV-1/2 test (Alere Medical, Tokyo, Japan) and confirmation with Unigold HIV-1/2 (Trinity Biotech, Bray, Ireland), followed national guidelines. 12 The Joint Clinical Research Centre in Kampala, Uganda, performed qualitative tests for HIV-1 deoxyribonucleic acid (DNA), based on COBAS AmpliPrep and version 2.0 of the COBAS TaqMan assay (Roche Diagnostics, Indianapolis, United States of America), on batched dried spots of blood from all identified HIV-exposed infants.

Statistical analyses
We focused on five main steps in the PMTCT cascade of care: attendance at an antenatal clinic -known as step 0; ascertainment of HIV status during antenatal care (step 1); uptake of maternal ART (step 2); use of infant nevirapine prophylaxis (step 3); and HIV testing, before the study, of HIV-exposed infants when more than eight weeks old (step 4). The denominators used to calculate weighted proportions for step 1, steps 2 and 3 and step 4 were, respectively, the total number of mother-infant pairs included in the cohort, the total number of known HIV-infected mothers and the total number of known HIV-infected mothers with infants that were more than eight weeks old, i.e. with infants that should have been tested for HIV. Mothers who claimed to be HIV-negative and were subsequently found negative in the rapid test were categorized as confirmed HIV-uninfected. Similarly, mothers who claimed to be HIV-positive and were subsequently found positive in the rapid test were categorized confirmed HIVinfected. The HIV status of the other mothers was categorized either as missed HIV diagnosis, if the mother claimed to be HIV-negative or not know her HIV status, but was subsequently found positive in the rapid test, or as inconclusive, if the rapid test results were inconclusive. We recorded ratios of MTCT at 4-26 weeks postpartum as the percentage of infants tested for HIV-1 DNA that were found positive. We calculated an overall MTCT ratio and also separate ratios for the mother-infant pairs who had missed none or one or more PMTCT cascade steps or who were categorized as missed HIV diagnosis. We report unweighted numbers and weighted categorical proportions with 95% confidence intervals (CI). Missing data were treated as additional categories. We used χ 2 tests to compare weighted MTCT ratios. We used a weighted multivariable binary logistic regression to identify factors associated with missing steps 1, 2, 3 and/or 4 of the cascade of care or a missed HIV diagnosis. In each model, weighted odds ratios with 95% CI were adjusted for region and maternal age, parity and uptake of antenatal care, at the study site or a different site, to give adjusted odds ratios (aOR). In the models for missed maternal ART uptake, missed nevirapine prophylaxis and missed infant HIV testing, we also adjusted for ascertained maternal HIV status. We also adjusted for maternal ART status and timing in the model for missed uptake of nevirapine prophylaxis and for uptake of nevirapine prophylaxis in the model for missed infant HIV testing. All analyses were conducted using SPSS Statistics 23 (IBM, Chicago, USA), and adjusted for the complex design of the whole national evaluation of Malawi's PMTCT programme. Each observation was weighted according to sampling interval and the probabilities of districts, clusters and subjects being selected. 13

Ethics
Ethical approval was provided by Malawi's National Health Sciences Research Committee (#1262), the United States Centers for Disease Control and Prevention (#2014-054-7) and the University of Toronto (#30448). All mothers or caregivers provided written informed consent.

Results
Although 34 637 mothers or caregivers were interviewed and tested for HIV infection, 657 (1.9%) had to be excluded, because of non-eligibility or incomplete data. All 236 (0.7%) caregiver-infant pairs had to be excluded because the caregivers gave inconsistent answers to the questions on PMTCT services. The remaining 33 744 mothers, who attended care with infants aged 4-26 weeks old, were included in our analysis.

Maternal characteristics
The characteristics of the enrolled mothers and their uptake of each step in the PMTCT programme are summarized in Table 1. All percentages and aOR reported below are weighted values. Mothers' ages ranged from 12 to 53 years; 17 931 (53.8%) were under 25 years of age and 6247 (20.5%) were adolescents aged Characteristic Unweighted no. (. . .continued) 12-19 years. Parity ranged from 1 to 14 and 10 943 (30.5%) of the mothers were primiparous. During the index pregnancy, 33 681 (99.8%) of the mothers attended an antenatal clinic (step 0), 33 276 (97.8%) had their HIV status ascertained at such a clinic (step 1) and 26 225 (76.3%) reported that they had attended an antenatal clinic at the site where they were enrolled.

ART status and timing of ART initiation among confirmed HIV-infected
Of the women who reported that they had had their HIV status ascertained at an antenatal clinic, 1637 (6.4%) and 1596 (5.1%) reported that they had been found HIV-positive before and during the index pregnancy, respectively.
HIV testing at the time of our study identified 3519 (12.1%) mothers with HIV infection, including 3233 (11.3%) confirmed HIV infections and 286 (0.9%) missed HIV diagnoses. The latter represented by 244 mothers who had claimed they were HIV-negative and 42 who had claimed not to know their HIV status.
Of the confirmed HIV-infected mothers, 2676 (75.9%) reported giving their infant nevirapine prophylaxis from birth to an age of six weeks (step 3) and 323 (16.4%) had given such prophylaxis for less than six weeks. Although, when they were interviewed, 139 of the 323 had infants that were under six weeks of age. Of the 1465 identified HIV-exposed infants that were older than eight weeks when their mothers were enrolled, 790 (53.2%) had been tested for HIV-1 DNA before the study screening (step 4).
Of the 3519 HIV-infected women identified at screening, 286 (7.1%) had missed being diagnosed earlier. Such missed diagnosis was associated with having attended an antenatal clinic  (Table 3). Not having started ART was associated with having attended an antenatal clinic somewhere other than the study site (aOR: 4.7; 95% CI: 1.5-14.8) and with having been diagnosed with HIV during, rather than before, the index pregnancy (aOR: 3.2; 95% CI: 1.3-8.0).
Overall, 675 (46.8%) of all known HIV-exposed infants that were older than eight weeks when their mothers were interviewed had reportedly not been tested for HIV-1 DNA at that time (Table 4). Mothers with infants that had not been tested were more likely to be 3) than to be on ART and to have given no nevirapine prophylaxis to their infants than to have given such prophylaxis (aOR: 2.0; 95% CI: 1.2-3.4; Table 4).
The MTCT ratio among mothers who had missed HIV diagnosis was significantly higher than that among mothers who had completed steps 1-3 (P < 0.01) or 1-4 (P < 0.01) or missed step 2 (P = 0.04). None of the other differences seen between MTCT ratios, e.g. between mothers who had completed and missed steps or between mothers who had missed HIV diagnosis and those who had missed steps 2-3 or 2-4, reached statistical significance.

Discussion
We estimated that 12.1% of infants of women attending under-5 clinics across Malawi were HIV-exposed. This value is higher than an aggregated estimate (8.1%) based on data collected in antenatal-care facilities across Malawi, 8 but consistent with the prevalence of HIV infection reported among women aged 15-49 years (12.1%) as part of the recent HIV impact assessment in Malawi. 14 The high uptake of PMTCT services we report aligns with estimates produced by the health minstry. 14, 15 The PMTCT uptake was similar across regions or between rural and urban sites. Our results add to the evidence indicating that, within Malawi, the scale-up of the implementation of the Option B+ strategy has provided widely decentralized and equitable coverage of PMTCT services. 8,16 The elimination of paediatric HIV infections will probably depend on the full use of PMTCT services. [17][18][19] In South Africa, a third of early infant HIV infections were attributed to the missing of one or more steps in the PMTCT cascade of care and, as observed in our study, the MTCT ratio increased as more steps were missed. 10 We found that young women and adolescent girls were particularly prone to missing antenatal HIV testing. Data previously collected in Kenya, 20 Malawi, 21 and South Africa 10,22 indicated that adolescents were especially likely to miss parts of the PMTCT cascade. 10,[20][21][22] In our study, adolescents found HIVpositive usually started ART and infant nevirapine prophylaxis, but often missed infant HIV testing. Further research is needed to improve our understanding of the retention of, and outcomes among, young women and to assess the potential benefits of youth-friendly initiatives such as peer-support groups for pregnant adolescents.
We identified a substantial number of women who missed HIV diagnosis during the index pregnancy and these were associated with a relatively high MTCT ratio. Some of these women may have had the very high viral loads during acute HIV infection that are strongly associated with MTCT. 23 We need more research on the identification and engagement of this high-risk subgroup.
The impact of service integration and timing of HIV diagnosis on uptake of ART was demonstrated by the fact that HIV-infected women were relatively unlikely to be on ART if they had received antenatal care at a different facility to the one hosting the under-5 clinic they attended and if they had been found HIV-infected during, rather than before, the index pregnancy. In earlier studies in Malawi, a facility's model of care was found to influence mothers' engagement in the PMTCT cascade 24 and retention on ART has been found to be relatively poor in facilities where women were asked to initiate ART on the same day that they discovered they were HIVinfected. 25 For the women involved, HIV diagnosis during pregnancy and the concept of life-long treatment are both challenging. 26,27 Further understanding of the impact of a HIV diagnosis during pregnancy and of the optimal model of care is required.
In our study, around a quarter of HIV-exposed infants did not receive the full six-week course of nevirapine prophylaxis and almost half of the HIVexposed infants above eight weeks of age had not been tested for HIV-1 DNA. Delayed or missed infant HIV diagnoses, which have previously been highlighted as key challenges in Malawi,8,28 lead to delayed ART and place HIV-infected infants at high risk.
If we project our findings to the Malawian population and burden of HIV, we estimate that, in 2017, about 16 874 Malawian women, i.e. 2.2% of the 767 000 Malawian women who gave birth, 29 did not receive antenatal HIV testing. If we assume a 12.1% prevalence of HIV infection among the fertile women, 14 we can estimate that 92 807 pregnant Malawian women were HIVinfected, including 6589 (7.1%) who missed HIV diagnosis in pregnancy. Among the women who knew that they were HIV-positive when pregnant, an estimated 1035 (1.2%) will not have started ART, 6639 (7.7%) of their infants will not have received any nevirapine prophylaxis and 40 350 (46.8%) of their infants will not have been tested for HIV-1 DNA after they were aged six weeks. An estimated 69 327 (74.7%) of the mothers will have received antenatal HIV testing and, if found infected, started maternal ART and given nevirapine prophylaxis to their infants, of whom an estimated 2496 (3.6%) will have been infected. Among the 6589 mothers who missed HIV diagnosis, 1450 (22.0%) will have transmitted HIV to their infants. Based on our findings and these projections, the focus of future targeted interventions needs to be on reducing the numbers of missed HIV diagnoses in pregnancy and increasing the proportion of HIV-exposed infants tested for HIV-1 DNA.
The large sample size and the study's national representativeness are strengths of this study. However, by screening at under-5 clinics, we may

Missed HIV diagnosis
Programme utilization 95% CI CI: confidence interval. Notes: Ratios were recorded for the mother-infant pairs who missed or did not miss some of the main steps along the cascade of care provided by the national programme for the prevention of motherto-child transmission of human immunodeficiency virus.
Step 1: ascertainment of HIV status during antenatal care; step 2: uptake of maternal ART; step 3: use of infant nevirapine prophylaxis: step 4: HIV testing, before the study, of HIV-exposed infants when more than eight weeks old. The values were weighted according to sampling interval and the probabilities of districts, clusters and subjects being selected.

Research
Prevention of mother-to-child HIV transmission in Malawi M van Lettow et al.
have biased our sample towards women who usually take up health care. We made no attempt to recruit women who do not attend clinics and those with infants who died younger than four weeks. As a result, our estimates of the use of the PMTCT programme may be too high. We also acknowledge the potential for responder bias in terms of reported uptake of services, although interviewers checked mothers' health booklets to confirm responses. Lastly, as some of the numbers in the subgroup analyses were small, the reported confidence intervals are wide and need to be interpreted with care. The same small numbers limited the statistical significance of the between-subgroup differences seen in MTCT ratios.
In conclusion, the scale-up of Option B+ services appears to have provided universal access to PMTCT care. The findings that young maternal age, high parity and use of services at different clinics were associated with incomplete care should help to guide the development of interventions to accelerate progress towards the elimination of MTCT from Malawi Conclusion Une majorité de femmes a eu recours au programme malawite pour la prévention de la TME. Le risque de TME augmentait lorsque l'une des principales étapes du programme n' était pas suivie.