Effectiveness of a PMTCT programme in rural Western Kenya

Abstract

We assess the coverage of a Prevention of Mother-to-child Transmission (PMTCT) programme in Busia (Kenya) from 1 January 2006 to 31 December 2008 and estimate the risk of transmission of HIV. We also estimate the odds of HIV transmission according to pharmacological intervention received. Programme coverage was estimated as the proportion of mother–baby pairs receiving any antiretroviral (ARV) regimen among all HIV-positive women attending services. We estimated the mother-to-child transmission (MTCT) rate and their 95% confidence interval (95%CI) using the direct method of calculation (intermediate estimate). A case–control study was established among all children born to HIV-positive mothers with information on outcome (HIV status of the babies) and exposure (data on pharmacological intervention). Cases were all HIV-positive children and controls were the HIV-negative ones. Exposure was defined as: (1) complete protocol: ARV prescribed according World Health Organisation recommendations; (2) partial protocol: does not meet criteria for complete protocol; and (3) no intervention: ARVs were not prescribed to both mother and child. Babies were tested using DNA Polymerase Chain Reaction at six weeks of life and six weeks after breastfeeding ceased. In the study period, 22,566 women accepted testing, 1668 were HIV positive (7.4%; 95%CI 7.05–7.73); 1036 (62%) registered in the programme and 632 were lost. Programme coverage was 40.4% (95%CI 37.9–42.7). Out of the 767 newborns, 28 (3.6%) died, 148 (19.3%) defaulted, 282 (36.7%) were administratively censored and 309 (40.2%) babies completed the follow-up as per protocol; 49 were HIV positive and MTCT risk was 15.86% (95%CI 11.6–20.1). The odds of having an HIV-positive baby was 4.6 times higher among pairs receiving a partial protocol compared to those receiving a complete protocol and 43 times higher among those receiving no intervention. Our data show a good level of enrolment but low global coverage rate. It demonstrates that ARV regimens can be implemented in low resource rural settings with marked decreases of MTCT. Increasing the coverage of PMTCT programmes remains the main challenge.

Keywords:

• prevention of mother-to-child transmission
• HIV
• low resource settings
• effectiveness
• programme coverage

Introduction

An estimated 430,000 new HIV infections occurred among children under the age of 15 in 2008. Most of these new infections are believed to stem from transmission in utero, during delivery or post-partum as a result of breastfeeding (Joint United Nations Programme on HIV/AIDS [UNAIDS] & World Health Organization [WHO], 2009). In the absence of any intervention the risk of mother-to-child transmission (MTCT) is 15–30% in non-breastfeeding populations. Breastfeeding by an infected mother increases the risk by 5–20% to a total of 20–45% (De Cock et al., 2000). In developed countries, highly active antiretroviral therapy (HAART) has reduced the vertical transmission rates to around 1–2% but HAART protocols are not yet widely available in low- and middle-income countries. In these settings several simplified and less costly antiretroviral (ARV) regimens have been offered to pregnant women and/or their newborn babies (Volmink, Siegfried, van der Merwe, & Brocklehurst, 2007). ARV prophylactic regimens in the third trimester of pregnancy, together with intra- and post-partum prophylaxis, can reduce the risk of transmission during pregnancy and childbirth to 2–4% in non-breastfeeding population (Lallemant et al., 2004). When used in breastfeeding setting the regimens of Zidovudine (AZT) or AZT/Lamivudine (3TC) in the last trimester plus a single dose (sd) of Nevirapine (NVP) at labour to mother and new born demonstrated a reduction of the transmission rate to 6–15% (Dabis et al., 2005; Taha et al., 2003; Thior et al., 2006; Volmink et al., 2007). However, translating clinical trial findings conducted in a research setting into practice and increasing coverage of Prevention of Mother-to-child Transmission (PMTCT) programmes remain the main challenge. By 2004, fewer than 10% of HIV-positive pregnant women had access to PMTCT services worldwide and although recent data suggest that coverage increased in 2007 up to 33% it is highly variable between countries and populations (UNAIDS, 2008). Far too few pregnant women are aware of their HIV status. In 2007, only 18% of pregnant women in low- and middle-income countries, where data were available, were tested for HIV (UNICEF, 2008).

In 2002, Médecins Sans Frontières (MSF) started an HIV/AIDS treatment programme in Busia district, Kenya, which included decentralised care and PMTCT activities in rural areas. Busia district is bordering Lake Victoria, with around 400,000 inhabitants. The area is mostly rural and the HIV prevalence is among the highest of the country, as high as 30% in some areas of the district (UNAIDS & WHO, 2008). The Kenyan National PMTCT programme includes HIV testing in the antenatal care clinics (ANC) to all pregnant women with an “opt-out” strategy and access to ARV drugs to reduce the risk of HIV transmission, following the World Health Organisation (WHO) recommendations for low-resourced countries (WHO, 2006a). In 2006, MSF increased its support to the Ministry of Health (MoH) to improve the decentralisation of the PMTCT services within the district. MSF provided adequate training, supported the integration of PMTCT services in the ANC through getting additional staff and ARV supply when needed.

Our main objective is to assess the coverage of a PMTCT programme in Busia (Kenya) from 1 January 2006 to 31 December 2008 and estimate the risk of transmission of HIV. As secondary objectives, we estimate the odds of HIV transmission according to the pharmacological intervention received by mothers and babies attending ANC clinics, and provide an estimate of the HIV prevalence in the pregnant women attending ANC clinics in the study period.

Methods

Description of the intervention

MSF supported the PMTCT activities in three dispensaries, five health centres, two sub-district hospitals and one mission hospital. The target population were the pregnant women attending their first ANC visit to any of the mentioned health structures between 1 January 2006 and 31 December 2008. After adequate information on the testing package offered in the ANC, those women who did not refuse were tested for HIV using a rapid test. Results were delivered immediately and post-test individual counselling was given. Women found to be HIV positive were invited to be enrolled in the PMTCT programme and information about the options to prevent HIV transmission was provided. HAART was only available in three of the health structures for those in need according with the WHO recommendations, in the other centres short-course AZT, single-dose NVP and post-partum AZT to the new born were the standard of care. Women were able to enter the programme at any stage of their pregnancy or even immediately after delivery and they were offered treatment according to immunological criteria and gestational age. Women were advised to deliver in a health centre attended by a skilled professional but they were free to deliver at the place of their choice. Exclusive breastfeeding with early weaning at six months or artificial feeding was promoted. However, since artificial formula feeding was not provided, this last option was chosen by a minority of women. Exposed babies were tested twice, first time at six weeks of life, or as soon as possible afterwards, and a second time at least 6 weeks after breastfeeding ceased when the first test had been negative. Actual duration of breastfeeding was very variable. Blood samples were collected on dried spots and tested using DNA Polymerase Chain Reaction (PCR) in the Centers for Disease Control and Prevention (CDC) office in the neighbour city of Kisumu. If any exposed baby presented between 12 and 18 months of life, a rapid test was used on first instances, and only after a positive result, confirmation with PCR was performed following the WHO recommendations at that time (WHO, 2006b).

Study setting and participants

The study included all pregnant women attending for their first visit to ANC in Busia, Kenya from 1 January 2006 to 31 December 2008.

Study design

A cross-sectional approach was used to estimate the prevalence of HIV in the pregnant women and the coverage of the PMTCT programme among HIV-positive women attending ANC. To estimate the odds of mother-to-child (MTC) HIV transmission, a case–control study was established among all children born to HIV-positive mothers on which information on outcome (HIV status of the babies) and exposure (data on pharmacological intervention) was available. The cases were all HIV-positive children and the controls were the HIV-negative ones. We assessed their exposure to three different pharmacological interventions described as follows.

Exposures of interest and assessment of outcome

The pharmacological interventions were classified as follows: complete protocol (ARV prescribed in perfect agreement with WHO recommendations, i.e., mother: HAART or AZT four weeks+NVP sd+(AZT-3TC) and child: NVP sd+AZT seven days), partial protocol (does not meet criteria for complete protocol) and no intervention (ARV has not been prescribed to both mother and child). The assessment of the outcome of interest, that is, the HIV status of the baby, was done according to the protocol specification previously described.

Variables and data sources

Data on the number of attendants were collected by MSF staff from the ANC and labour registries. After enrolment in the programme the information collected was that of every visit of the women and the babies to any of the health centres. The information was collected in a specific form, and a self-copied form was sent to MSF offices, where a data manager introduced it in FUCHIA Software (Epicentre, Paris, France).

The variables collected on the women were their age, the gestational age at entry in the programme, type and place of delivery and feeding choice. After delivery additional information was collected in the mother–baby pairs such as the pharmacological intervention received.

Statistical analyses

The coverage of the programme was estimated as the proportion of mother–baby pairs receiving any ARV regimen to prevent the HIV transmission risk among all HIV-positive women attending ANC services (with its 95% confidence interval (95%CI). Prevalence of HIV was estimated together with their 95%CI. We used the method described by The Working Group on Mother-to-child transmission of HIV (1995) to estimate the MTCT rate and their 95%CI using the direct method of calculation (intermediate estimate). Additionally, the odds ratios (OR) of MTC HIV transmission according to different intervention were also calculated and their 95%CI. Data were analysed using FUCHIA, SPSS and EPIDAT.

Results

Between 1 January 2006 and 31 December 2008, 22,566 women accepted to be tested for HIV in the ANC clinics and 1668 were found to be HIV positive, thus HIV prevalence was 7.4% (95%CI 7.05–7.73). Of these 1668 HIV-positive women, 1036 (62%) did actually register in the PMTCT programme and the other 632 were lost to follow-up. During the study period, 767 babies were born. Overall, 673 pairs (mother–baby) received any pharmacological intervention to reduce transmission of HIV, resulting in a coverage of 40.4% (95%CI 37.9–42.7) out of the 1668 pregnant women who tested positive for HIV.

Table 1 describes the socio-demographic and obstetric characteristics of the 767 HIV-positive pregnant women who delivered a baby. Mean gestational week at entry was 26 weeks and 425 (58%) delivered at home. Nearly all women, 99%, had vaginal delivery and the majority, 528 (69%) reported they were going to provide exclusive breastfeeding.

Table 1. Socio-demographic and obstetrics characteristics of the 767 HIV-positive pregnant women delivered during the period.

Variables	N (total=767)	Percentage (%)
Age (years)
<16	0	–
≥16 to <18	19	2.5
≥18 to <25	280	36.5
≥25 to <35	401	52.3
≥35	56	7.3
Unknown	11	1.4
Gestational age at entry in weeks, N=726 (mean and SD)	26	7
Place of delivery
Health structure	313
Home delivery	425
Unknown	29
Type of delivery, N=758
Vaginal delivery	749	98.8
Caesarean section	9	1.2
Feeding option
Exclusive breastfeeding	528	68.8
Infant formula	88	11.5
Mix feeding	151	19.7

At the end of the study period, out of the 1668 HIV-positive women initially identified, the final HIV status after breastfeeding was only known for 309 babies (260 were HIV negative and 49 were HIV positive). Using the direct method of calculation (intermediate estimate), the estimate of the HIV MTCT rate after breastfeeding period was 15.86% (95%CI 11.6–20.1).

Out of the 767 newborns, 309 (40.2%) babies completed the follow-up as per protocol, 282 (36.7%) were administratively censored (187 initially HIV negative were pending of the second test while still breastfeeding), 28 (3.6%) died and 148 (19.3%) were lost to follow-up.

Table 2 describes the OR of MTCT according to the protocol received, in those 309 babies. All the mother–baby pairs attended in a health structure received any PMTCT intervention whereas 46 (10.8%) of those giving birth at home did not get any intervention. Of the 309 babies, 80 pairs (25.9%) had completed the protocol, 205 (66.3%) had partially completed the protocol and only 24 (7.8%) received no intervention. The odds of having an HIV-positive baby was 4.6 times higher among pairs receiving a partial protocol compared to those receiving a complete protocol and 43 times higher among those receiving no intervention.

Table 2. Odds of MTCT according PMTCT intervention in 309 babies who completed follow-up as per protocol and final HIV status it is known.

PMTCT intervention	Total	HIV negative	HIV positive	OR (95%CI)
Complete protocol	80	77	3	1
Partial protocol	205	174	31	4.6 (1.4–15.4)
No intervention	24	9	15	42.8 (10.3–176.8)
Total	309	260	49	–

Discussion

In a population of pregnant women with an HIV prevalence of 7.4% the coverage of the PMTCT programme, estimated as the proportion of mother–baby pairs receiving any type of ARV regimen among the HIV-positive women attending ANC services, was 40%. Estimate of the HIV MTCT rate using the direct method of calculation (intermediate estimate) after breastfeeding cessation was 15.86%. The odds of HIV transmission in the sub-group of babies with information available were significantly higher among those receiving partial or no pharmacological interventions.

The prevalence of HIV in the study population was similar to the estimated 7% HIV prevalence among adults in 2007 in Kenyan rural areas (UNAIDS & WHO, 2009). Our coverage results, 40%, are similar to other published data (Dabis & Ekpini, 2002; National AIDS Control Council [NASCOP], 2008; Perez et al., 2004; Temmerman, Quaghebeur, Mwanyumba, & Mandaliya, 2003; van't Hoog et al., 2005) but still far from the 2010 United Nations General Assembly (UNGASS) target of 80% of pregnant women having access to appropriate HIV counselling, testing and ARV treatment to prevent MTCT of HIV. Most worrisome is that almost 38% of the women found to be HIV positive were not registered in the programme; even if there were other actors in the area offering PMTCT services, this is a main bottleneck for the implementation of efficacious PMTCT programme. It is reassuring that nearly 90% of mother–baby pairs from whom information was available at the time of delivery have received PMTCT intervention. This means that if we are able to maintain the women in the programme after the first visit, the adherence is very good. This supports the importance of adequate counselling and information about the treatment options to avoid HIV transmission, which has been corroborated in group discussion with community members in the area. Since integration of PMTCT services within the ANC was not universally available in all the health centres pregnant women may be reluctant to be seen in the waiting area with HIV/AIDS patients. The social stigma and the potential consequences in case of disclosure of their HIV status to their family may prevent some women to attend for follow-up to the clinic (Bwirire et al., 2008).

After delivery, a significant percentage of babies (19.3%) were also lost to follow-up showing a second bottleneck in the programme. At that time, given that the pharmacological intervention has finished, it is likely that the perception of the mothers about the advantages of attending for follow-up decreases. The majority of the babies would be healthy and the fear of a theoretically positive result in the baby may prevent the women to bring them for testing. Again, stigma and inadequate counselling about the treatment options for the women and babies may play a roll in this defaulting rate.

The mortality rate is similar to other reports (Coovadia et al. 2007; Petra study team, 2002) but unfortunately we can not assure the quality of this information because there was not a good defaulting trace system and we do not know how many among those babies considered lost were actually dead. We do not have information on the cause of death of these babies.

The estimated MTCT rate of HIV in our data is similar to previous published data in low resource settings where breastfeeding is the option for most of the women (Jackson et al., 2003; Lallemant et al., 2000; Petra study team, 2002) and we consider it as a success taking in consideration the programme conditions. The ARV regimens actually taken by the pairs mother–baby have been various and heterogeneous; there was not universal access to blood collection for CD4 cell counts, initiation of HAART was not possible in all the health centres and the gestational age at entry into the programme was very variable with an important number of women attending just on labour. The high odds of HIV transmission found among those few pairs who did not receive any pharmacological intervention may be confounded by socio-economic and clinical status. Unfortunately, we can not adjust by this as information on CD4 count and viral load data was not available.

As recommended by WHO, the first HIV test to the babies should be done about 6 weeks after birth while most of them are still breastfeeding but the variable timing of test performance does not permit us to establish exactly when the transmission occurred and corroborated the relevance of the breastfeeding period in the transmission as other authors have highlighted (Kilewo et al., 2009; Kourtis, Lee, Abrams, Jamieson, & Bulterys, 2006).

The compliance of the regimens prescribed is a main concern because self-reported adherence to ARVs among pregnant women has been proven higher than actual levels of drug measured in cord blood (Albrecht et al., 2006; Bii, Otieno-Nyunya, Siika, & Rotich, 2007; Carlucci et al., 2008). More than 50% of the women delivered at home and in 10% of these cases the pair mother–baby did not get any ARVs, as in other studies home delivery results in poorer adherence to treatment (Kasenga, Hurtig, & Emmelin, 2007).

The data collection in the field and the assurance of good quality information were arduous because it was a functioning programme and not a research setting.

Our results show that the implementation of a PMTCT programme in a low resource setting is feasible with a significant reduction in the MTC HIV transmission in a real-life situation. The actual impact of the programmes depends on the universal access to the PMTCT package: from HIV testing to post-natal treatment and follow-up as much as possible integrated within the ANC. To minimise the MTC HIV transmission risk, it is necessary that PMTCT programmes can provide universal access to CD4 counting and triple therapy for those women in need but the main problem remains the low coverage achieved. Addressing the cultural barriers and social constrains to reduce the lost of follow-up, before and after delivery, and increase adherence to the programme should be a priority. Although more sophisticated ARV regimen has shown more efficacy to reduce HIV transmission, the main problem is those women and babies who do not access to the services and/or do not receive any treatment. To significantly reduce the burden of paediatrics HIV infection universal access to PMTCT services is crucial.