pregnancy and continued ARV support for mother and infant during breast feeding ... Cape Town health services were among the first in the region to offer both ...
Tropical Medicine and International Health
doi:10.1111/j.1365-3156.2010.02538.x
volume 15 no 7 pp 825–832 july 2010
Initiation of highly active antiretroviral therapy among pregnant women in Cape Town, South Africa Kathryn Stinson1, Andrew Boulle1, David Coetzee1, Elaine J. Abrams2 and Landon Myer1,2 1 Centre for Infectious Disease Epidemiology and Research, School of Public Health and Family Medicine, University of Cape Town, Cape Town, South Africa 2 International Center for AIDS Care and Treatment Programs, and Department of Epidemiology, Mailman School of Public Health, Columbia University, New York, NY, USA
Summary
objective To investigate highly active antiretroviral therapy (HAART) initiation among pregnant women and the optimum model of service delivery for integrating HAART services into antenatal care. methods We analysed clinic records to reconstruct a cohort of all HIV-infected pregnant women eligible for HAART at four antenatal clinics representing three service delivery models in Cape Town, South Africa. To assess HAART coverage, records of women determined to be eligible for HAART in pregnancy were reviewed at corresponding HIV treatment services. results Of 13 208 pregnant women tested for HIV, 26% were HIV-infected and 15% were HAARTeligible based on a CD4 cell count of £ 200 cells ⁄ ll. Among eligible women, 51% initiated HAART before delivery, 27% received another prevention of mother-to-child transmission (PMTCT) intervention and 22% did not receive any antiretroviral intervention before delivery. The proportions of women initiating HAART between the different service delivery models were comparable. The median gestational age at first presentation was 26 weeks, and early gestational age at first presentation was the strongest predictor of being on HAART by delivery. Of the women who did not initiate HAART in pregnancy, 24% started treatment within 2 years postpartum. conclusions In this setting with clear PMTCT and HAART protocols, services failed to prioritize and initiate a high proportion of eligible pregnant women on HAART. The initiation of HAART in pregnancy requires strengthened antenatal and HIV services that target women with advanced stage disease. keywords mother-to-child transmission, HIV, highly active antiretroviral therapy, coverage, South Africa
Introduction The use of antiretroviral (ARV) prophylaxis during pregnancy and labour can dramatically reduce the risk of vertical transmission of HIV infection. Highly active antiretroviral therapy (HAART) has the greatest efficacy in reducing mother-to-child transmission (MTCT) of HIV and has significant direct benefits for maternal health (Cooper et al. 2002). Results from recent studies of improved regimen combinations, earlier treatment start in pregnancy and continued ARV support for mother and infant during breast feeding in the postpartum period have led to significant changes in the recommendations for managing pregnant HIV-infected women (Chasela et al. 2009; Kesho Bora Study Group 2009; Shapiro et al. 2009; Taha et al. 2009). In the light of new evidence, WHO recently released revised guidelines for the management of HIV-infected women in pregnancy, which recommend
ª 2010 Blackwell Publishing Ltd
initiating lifelong HAART earlier in pregnant women and raising the eligibility criteria from a CD4 cell count threshold of £ 200 cells ⁄ ll to £ 350 cells ⁄ ll or WHO clinical staging 3 or 4, irrespective of gestational age (World Health Organization 2009). Despite the importance of HAART during pregnancy, there are significant patient-related and service-related challenges to the implementation of HAART services as part of antenatal care. In many settings, high HIV seroprevalence among women of reproductive age results in a sizeable proportion of HIV-infected pregnant women accessing limited antenatal and HIV care services (Stringer et al. 2008). Patient-driven barriers include lack of knowledge of serostatus prior to conception; reluctance to test in pregnancy; fear of disclosure and denial postdiagnosis; and under-utilization of antenatal care during pregnancy (Abrams et al. 2007; Bolu et al. 2007; El-Sadr & Abrams 2007; Tlebere et al. 2007). Service delivery 825
Tropical Medicine and International Health
volume 15 no 7 pp 825–832 july 2010
K. Stinson et al. Initiation of HAART among pregnant women in Cape Town
approaches impact on the accessibility and availability of treatment services for HIV-infected pregnant women. Prevention of mother-to-child transmission (PMTCT) services for women who do not have advanced HIV disease, but who require short-course prophylactic regimens, are generally integrated within routine antenatal health services managed by midwives. Presently, there are a few such integrated services for pregnant women who are eligible for HAART (Schneider et al. 2006). As a result, these women may need to access antenatal care and stand-alone doctordriven HAART services that may be geographically separate. The integration of HAART into antenatal services has been proposed and is underway in selected settings, yet little is known about the operational experience of different models of service delivery in resource-constrained contexts (Abrams et al. 2007). Cape Town health services were among the first in the region to offer both PMTCT (1999) and HAART (2001), with protocols providing for universal access to HAART for eligible pregnant women since 2003 (Abdullah et al. 2001; Draper & Abdullah 2008). A number of service models for delivering HAART to pregnant women developed during this period providing an opportunity to explore the relationship between these models and the uptake of HAART. Although HAART has been available for eligible HIV-infected pregnant women in Cape Town since 2003, little is known about HAART coverage in pregnancy. The objectives of this study were to describe HAART coverage in a cohort of eligible women accessing antenatal services and to evaluate different service models for providing HAART.
Methods Study setting This study evaluated three service models for HAART referral and initiation in pregnancy at four primary care antenatal clinics in Cape Town during 2005. The sites were selected to highlight the differences in the proximity of the ARV service to the antenatal site and the service model offered. One antenatal clinic incorporated an ‘integrated’ HAART service, offered once per week by two outreach doctors. The second antenatal clinic referred eligible women to a stand-alone ARV service in a separate building on the same premises, a ‘proximal’ model of care. The third, ‘distal’ model constituted two antenatal clinics, and eligible women were referred to ARV services within a 5-km radius, accessible by foot or public transport. All four clinics operated according to identical midwifedriven care protocols with the same procedures for all aspects of antenatal care (Van Coeverden de Groot 1993; Van Coeverden de Groot et al. 1982) including PMTCT. 826
At each antenatal clinic, HIV counselling and opt-in testing were offered as part of routine care. After group and individual counselling, a rapid HIV test was performed, and results were given with post-test counselling on the same day. For newly diagnosed HIV-infected women, specimens were sent to external laboratories for CD4 cell count enumeration, and results were available at the service within 2 weeks. In the proximal and distal models of care, women eligible for HAART were given a referral letter to each ARV service linked to the antenatal service, where treatment initiation occurred. At all sites, neonatal prophylaxis was administered at birth for HIV-infected women in the delivery room. Study design and data collection This retrospective cohort included all women who presented at the four antenatal facilities between 1 January and 31 December 2005. Antenatal data including demographic information, HIV status and CD4 cell count were extracted from PMTCT registers kept on-site using a standardized data abstraction tool. Field workers fluent in local languages searched paper-based and electronic data sources for patient name variations and supplemented this with other personal identifiers, such as date of birth where needed, to maximize follow-up. Missing data, including CD4 cell counts, were obtained from the National Health Laboratory Services patient database. The subset of HIV-infected women eligible for HAART based on a CD4 count £200 cells ⁄ ll was identified through patient registers and then traced through antenatal care and HIV care and treatment programme records to determine HAART initiation and obstetric and newborn outcomes. HAART coverage was defined as evidence of initiating HAART during pregnancy or evidence of being on HAART at delivery. Patients with missing information on obstetric, PMTCT and ⁄ or HAART-related variables were traced manually through the records and databases of each relevant health facility, including secondary and tertiary referral hospitals as well as every clinic providing HAART in the city. Data were analysed using stata 9.0 (STATA Corporation, College Station, USA). Proportions were estimated for each step of the PMTCT cascade, and HAART coverage was calculated both for the entire cohort and separately for the integrated, proximal and distal service models. Bivariate associations were calculated using chi-squared tests for categorical variables and one-way anova models or the Kruskal–Wallis test for continuous data. Multiple logistic regression was used to examine the independent predictors of HAART initiation in pregnancy. The inclusion of the dependent variables was based upon a priori knowledge,
ª 2010 Blackwell Publishing Ltd
Tropical Medicine and International Health
volume 15 no 7 pp 825–832 july 2010
K. Stinson et al. Initiation of HAART among pregnant women in Cape Town
their significance in univariate analyses and model fit in the regression analysis. Approval for the study was obtained from the University of Cape Town Research Ethics Committee and local government health authorities. Results Patient and pregnancy characteristics A total of 14 987 women presented for antenatal care at the four sites (Table 1). The mean age was 25 years (standard deviation, 6 years). Overall, 88% of women were tested for HIV, and 26% were HIV-positive (Figure 1). Uptake of testing and HIV prevalence varied between service models: the proximal model had the lowest coverage of 79%, and the integrated model had the highest seroprevalence of 30% (P < 0.001 for both associations, Table 1). Overall, 97% of women who tested HIV-positive had a CD4 test completed by external laboratories; 15% (n = 516) had a CD4 cell count of £ 200 cells ⁄ ll and were eligible for HAART. A total of 29% (n = 986) of HIVinfected women had a CD4 cell count >200 cells ⁄ ll but