KAREN M. MEGAZZINI DONNA K. ARNETT, COMMITTEE CHAIR ROBERT L. GOLDENBERG DAVID T. REDDEN JEFFREY S. A. STRINGER STEN H. VERMUND A DISSERTATION

Transcription

1 PROVISION OF RAPID HIV TESTING AND NEVIRAPINE ADMINISTRATION IN ZAMBIAN LABOR WARDS TO IMPROVE POPULATION ANTIRETROVIRAL COVERAGE OF HIV-INFECTED WOMEN AND THEIR HIV-EXPOSED INFANTS by KAREN M. MEGAZZINI DONNA K. ARNETT, COMMITTEE CHAIR ROBERT L. GOLDENBERG DAVID T. REDDEN JEFFREY S. A. STRINGER STEN H. VERMUND A DISSERTATION Submitted to the graduate faculty of the University of Alabama at Birmingham, in partial fulfillment of the requirements for the degree of Doctor of Public Health BIRMINGHAM, ALABAMA 2008

2 PROVISION OF RAPID HIV TESTING AND NEVIRAPINE ADMINISTRATION IN ZAMBIAN LABOR WARDS TO IMPROVE POPULATION ANTIRETROVIRAL COVERAGE OF HIV-INFECTED WOMEN AND THEIR HIV-EXPOSED INFANTS KAREN M. MEGAZZINI DOCTOR OF PUBLIC HEALTH ABSTRACT Background: Labor ward-based services for prevention of mother-to-child transmission (PMTCT) of HIV may improve population nevirapine (NVP) coverage compared to antenatal care-based services alone. Methods: We conducted a cluster-randomized clinical in 12 public-sector labor wards in Lusaka, Zambia to determine whether an enhanced labor ward-based PMTCT package would improve population NVP coverage and to determine predictors of HIV testing acceptance. Following a baseline surveillance period, six labor wards offered voluntary counseling and testing to women of unknown HIV status along with NVP administration to those that tested seropositive and conducted a rapid, structured assessment of NVP adherence among known HIV-infected women. Six control labor wards provided the standard of care. Generalized estimating equations were used to compare NVP coverage (i.e. the proportion of HIV-infected/exposed women/infant pairs with confirmed NVP ingestion) and to identify factors associated with HIV testing acceptance. We also modeled the comparative effectiveness of three antiretroviral regimens considering the effects of both efficacy and population coverage. ii

3 Results: During the study, 7664 women gave birth to live born infants. Cord blood was collected from 7592 (99%) of the deliveries, 7438 (97%) of them were tested for HIV, 1618 (22%) were seropositive of which 1279 (79%) were tested for NVP. At baseline, the odds of HIV-infected/exposed women/infant pairs having received NVP in the intervention clinics (42%) was 0.78 times the odds of being covered in the control cluster (53%) whereas in the follow-up period, the odds of being covered in the intervention clinics (52%) was 1.5 times the odds of being covered in the control cluster (43%). This change represents a multiplicative effect of 1.91 upon the odds ratio at baseline (OR 1.91, 95% CI, ). The majority of women tested in the labor ward received treatment at least one hour (85%) prior to delivery. Cervical dilatation less than three centimeters at the time of admission was strongly associated with ingestion of NVP at least one hour (AOR 11.5; 95% CI ) and two hours (AOR 25.2; 95% CI ) before delivery. Conclusion: Labor ward-based PMTCT programs are feasible and can have significant, positive impact on population NVP coverage. iii

4 DEDICATION To Robb and Cruz iv

5 ACKNOWLEDGEMENTS I want to thank my committee for their continued support and guidance throughout the course of my dissertation work. I especially want to thank Dr. Jeffrey Stringer for providing me with the opportunity to conduct this important study and Dr. Sten Vermund for his ongoing mentorship which has been invaluable to me. I would also like to acknowledge my study team in Lusaka, the numerous labor ward midwives that helped to make this study possible and all the women and infants that participated in this study. v

6 TABLE OF CONTENTS ABSTRACT... ii DEDICATION...iv ACKNOWLEDGEMENTS...v LIST OF TABLES... vii LIST OF FIGURES...ix LIST OF ABBREVIATIONS...x INTRODUCTION...1 SINGLE DOSE NEVIRAPINE VERSUS MULTI-DRUG ANTIRETROVIRAL PROPHYLAXIS FOR PREVENTION OF MOTHER-TO-CHILD TRANSMISSION OF HIV: AN OUTCOMES-ORIENTED POLICY REVIEW FOR DEVELOPING COUNTRIES...5 A CLUSTER-RANDOMIZED TRIAL OF ENHANCED LABOR WARD-BASED PMTCT SERVICES TO INCREASE POPULATION NEVIRAPINE COVERAGE IN LUSAKA, ZAMBIA...25 PREDICTORS OF RAPID HIV TESTING ACCEPTANCE AND SUCCESSFUL NEVIRAPINE ADMINISTRATION IN ZAMBIAN LABOR WARDS...55 CONCLUSION...83 GENERAL LIST OF REFERENCES...86 APPENDICES...88 UNIVERSITY OF ALABAMA at BIRMINGHAM INSTITUTIONAL REVIEW BOARD APPROVAL...89 UNIVERSITY OF ZAMBIA RESEARCH ETHICS COMMITTEE APPROVAL...91 vi

7 LIST OF TABLES Table Page SINGLE DOSE NEVIRAPINE VERSUS MULTI-DRUG ANTIRETROVIRAL PROPHYLAXIS FOR PREVENTION OF MOTHER-TO-CHILD TRANSMISSION OF HIV: AN OUTCOMES-ORIENTED POLICY REVIEW FOR DEVELOPING COUNTRIES 1. Characteristics of the three antiretroviral regimens compared in the model Population coverage thresholds necessary for the World Health Organization (WHO) recommended regimen (ZDV TC + SD-NVP), WHO alternative regimen (ZDV-36 + SD-NVP) and WHO minimum regimen (SD-NVP) to prevent equal numbers of infant HIV infections in a hypothetical cohort of 10,000 HIV-infected, pregnant, Zambian women assuming a baseline transmission rate of 21.5%...23 A CLUSTER-RANDOMIZED TRIAL OF ENHANCED LABOR WARD-BASED PMTCT SERVICES TO INCREASE POPULATION NEVIRAPINE COVERAGE IN LUSAKA, ZAMBIA 1. Descriptive characteristics of women according to intervention and control arm assignment during baseline and follow-up surveillance periods Comparison of nevirapine coverage between control and intervention clinics during baseline surveillance and follow-up surveillance study periods Cluster-level effects of labor ward voluntary counseling and testing on population nevirapine coverage Cluster and clinic level effect of the rapid assessment of nevirapine adherence on population nevirapine coverage...54 vii

8 LIST OF TABLES continued Table Page PREDICTORS OF RAPID HIV TESTING ACCEPTANCE AND SUCCESSFUL NEVIRAPINE ADMINISTRATION IN ZAMBIAN LABOR WARDS 1. Characteristics of women eligible for voluntary counseling and testing for HIV in the labor ward (n=278) Factors associated with being offered voluntary counseling and testing for HIV in the labor ward (n=278) Factors associated with HIV testing acceptance in the labor ward (n=146) Factors associated with ingestion of nevirapine or calcium at least one hour prior to delivery following rapid HIV testing in the labor ward (n=71) Factors associated with ingestion of nevirapine or calcium at least two hours prior to delivery following rapid HIV testing in the labor ward (n=71)...81 viii

9 LIST OF FIGURES Figure Page SINGLE DOSE NEVIRAPINE VERSUS MULTI-DRUG ANTIRETROVIRAL PROPHYLAXIS FOR PREVENTION OF MOTHER-TO-CHILD TRANSMISSION OF HIV: AN OUTCOMES-ORIENTED POLICY REVIEW FOR DEVELOPING COUNTRIES 1. Projected number of infant HIV infections prevented in a hypothetical cohort of 10,000 HIV-infected, pregnant Zambian women using three different antiretroviral regimens and assuming population coverage of 47% Projected number of infant HIV infections that occur despite available prevention of mother-to-child transmission (PMTCT) services and the effect of targeted interventions to improve coverage and reduce the number of infant HIV infections in a single-dose nevirapine (SD-NVP)-based PMTCT program...24 A CLUSTER-RANDOMIZED TRIAL OF ENHANCED LABOR WARD-BASED PMTCT SERVICES TO INCREASE POPULATION NEVIRAPINE COVERAGE IN LUSAKA, ZAMBIA 1. Trial Summary Cluster- and clinic-level effect of the labor ward PMTCT package on population nevirapine coverage during the baseline (pre-intervention) and follow-up (postintervention) study periods...52 PREDICTORS OF RAPID HIV TESTING ACCEPTANCE AND SUCCESSFUL NEVIRAPINE ADMINISTRATION IN ZAMBIAN LABOR WARDS 1. Labor ward HIV counseling and testing profile Labor ward HIV testing and tablet administration profile...79 ix

10 LIST OF ABBREVIATIONS AFASS AIDS ANC ANRS AOR CDC CI GEE HAART HIV HIVNET HPLC MIRIAD NNRTI NVP OR PEPFAR PMTCT SD-NVP acceptable, feasible, affordable, sustainable, and safe [breastfeeding] Acquired Immunodeficiency Syndrome antenatal care [French] National Agency for AIDS Research adjusted odds ratio Center s for Disease Control and Prevention confidence interval generalized estimating equations highly active antiretroviral therapy Human Immunodeficiency Virus HIV Network for Prevention Trials High Performance Liquid Chromatography Mother-Infant Rapid Intervention At Delivery non-nucleoside reverse transcriptase inhibitor nevirapine odds ratio President s Emergency Plan for AIDS Relief prevention of mother-to-child transmission single dose nevirapine x

11 LIST OF ABBREVIATIONS (continued) UNAIDS UNGASS VCT WHO ZDV 3TC Joint United Nations Programme on HIV/AIDS United Nations Special Session on HIV/AIDS voluntary counseling and testing World Health Organization zidovudine lamivudine xi

12 INTRODUCTION By the end of 2007, 2.5 million children under the age of 15 were living with HIV/AIDS. 1 Nearly all of these children acquired the virus from their mothers during pregnancy, labor and delivery or through breast-feeding. Despite the tremendous strides made in prevention of mother-to-child transmission (PMTCT) of HIV over the last decade, an estimated 420,000 children were newly infected with HIV in Ninety percent of these infections occurred in sub-saharan Africa where the burden of disease is greatest but access to HIV counseling and testing services, 2 antiretroviral prophylaxis and HIV/AIDS care and treatment still lags far behind that of industrialized nations. In the absence of targeted interventions to reduce the rate of mother-to-child transmission, the risk of vertical HIV transmission approaches 45 percent. Approximately, 5 to 10 percent of vertical transmissions occur during pregnancy, 10 to 15 percent occur at the time of labor and delivery and 5 to 20 percent result from breastfeeding. 3, 4 The risk can be reduced to less than two percent when HIV-infected women have access to preventive measures such as antiretroviral prophylaxis or highly active antiretroviral therapy (HAART) during pregnancy, elective cesarean section, and can safely avoid breast-feeding. 5, 6 While these interventions are often available to women living in industrialized nations, the paradigm is vastly different for women in resource-poor countries. In low and middle income countries, only 11 percent of pregnant, HIV-infected, women receive antiretroviral prophylaxis, 7 access to cesarean 1

13 section is limited, and replacement feeding is infrequently acceptable, feasible, affordable, sustainable and safe (AFASS). 8 In sub-saharan Africa, short course antiretroviral prophylaxis is the backbone of PMTCT programs. A variety of antiretroviral regimens are available with a range of efficacies; 9 some of which also reduce antiretroviral resistance. 10 The most notable of the short course regimens is the HIV Network of Prevention Trials (HIVNET) 012 single dose nevirapine (NVP) regimen. 11 NVP is a non-nucleoside reverse transcriptase inhibitor (NNTRI) that when administered to HIV-infected women at the onset of labor and to their HIV-exposed infants within 72 hours of birth, can reduce transmission of 11, 12 HIV in breast-fed infants by 47%. The high efficacy, low cost, and practicality of this regimen, has facilitated its widespread use and shaped PMTCT program recommendations for HIV-infected women receiving antenatal and delivery care in resource-poor countries. Despite the simplicity of the single dose NVP regimen, bringing nevirapine-based PMTCT programs to scale in urban sub-saharan Africa has been challenging and somewhat disappointing in that lower than expected programmatic coverage (i.e. the proportion of HIV-infected women and their exposed infants with confirmed NVP 13, 14 ingestion) is reported. In order for an HIV-infected woman and her HIV-exposed infant to receive appropriate antiretroviral therapy, the woman must: 1) have access to a program that offers HIV testing and antiretroviral prophylaxis; 2) be offered the services; 3) agree to be tested; 4) receive her test results; 5) receive the prescribed antiretroviral prophylactic regimen; and 6) ingest the prophylaxis; and 6) her infant must also receive 2

14 prophylaxis after delivery. PMTCT programs have repeatedly reported losses at each level of this prevention cascade In Zambia s capital city of Lusaka, a district-wide PMTCT program has been operational since This program offers voluntary counseling and testing for HIV to women who attend any of the 24 public-sector antenatal care clinics. A rapid HIV testing algorithm is employed and women receive their HIV test results the same day the test is conducted. Those who test positive are provided with a single dose of NVP to selfadminister at the onset of labor according to the HIVNET 012 regimen. 1 Back-up doses are provided in the labor ward for those in need of a replacement dose at the time of delivery. Infants born to known HIV-infected women are given a single dose of NVP in the postpartum ward prior to discharge. In 2003, a comprehensive evaluation of this program found the predominant reasons for failed NVP coverage included: a failure to test all women who received antenatal care for HIV (either because they were not offered testing or they declined testing); and non-adherence among women who tested positive for HIV during antenatal care but failed to ingest NVP at the time of delivery. 13 While PMTCT services are ideally based in antenatal care centers, allowing for detection of HIV early in pregnancy when the most efficacious treatment regimens can be implemented, provision of labor ward-based services can provide an additional opportunity to reach a substantial proportion of at-risk women/infant pairs. Labor wardbased PMTCT services can provide women who present in labor unaware of their HIV status a final opportunity to learn their status and reduce the risk of their infant becoming infected. 18 At the same time, simple measures can be taken in the labor ward to target women diagnosed with HIV during antenatal care and improve the odds that they ingest 3

15 NVP prior to delivery. This enhanced labor ward-based PMTCT package, when used in conjunction with antenatal care-based services, can provide an important opportunity to improve programmatic coverage and prevent a greater number of infant HIV infections. The following three manuscripts emphasize the role of labor ward-based PMTCT services in improving antiretroviral coverage of HIV-infected pregnant women and their infants. Single Dose Nevirapine Versus Multi-Drug Antiretroviral Prophylaxis for Prevention of Mother-to-Child Transmission of HIV: An Outcomes-Oriented Policy Review for Developing Countries A Cluster-Randomized Trial of Enhanced Labor Ward-Based PMTCT Services to Increase Population Nevirapine Coverage in Lusaka, Zambia Predictors of Rapid HIV Testing Acceptance and Successful Nevirapine Administration in Zambian Labor Wards 4

16 SINGLE DOSE NEVIRAPINE VERSUS MULTI-DRUG ANTIRETROVIRAL PROPHYLAXIS FOR PREVENTION OF MOTHER-TO-CHILD TRANSMISSION OF HIV: AN OUTCOMES-ORIENTED POLICY REVIEW FOR DEVELOPING COUNTRIES KAREN M. MEGAZZINI, BENJAMIN H. CHI, MOSES SINKALA, JEFFREY S.A. STRINGER and STEN H. VERMUND In preparation for JAIDS Formatted for dissertation 5

17 INTRODUCTION Considerable progress has been made in expanding prevention of mother-to-child transmission of HIV (PMTCT) services in low and middle income countries since the discovery that single dose nevirapine (SD-NVP) given to a mother and an infant could cut HIV transmission in half. In many settings, women can now routinely receive counseling and testing for HIV and those who are infected can receive antiretroviral prophylaxis to prevent infection of their newborn. Much of the progress made to date is attributable to the simplicity of the SD-NVP regimen that has been a mainstay of PMTCT in many developing countries. 1,2 While progress has been made, bringing nevirapinebased PMTCT programs to scale in sub-saharan Africa has been challenging and somewhat disappointing in that lower than expected programmatic coverage is reported. 3-6 By 2005, only 9% of HIV infected pregnant women living in low and middle income countries were receiving antiretroviral prophylaxis for PMTCT, far below the target of 80% set by the original United Nations General Assembly Special Session on HIV/AIDS (UNGASS) Declaration of Commitment on HIV/AIDS in By 2006, this PMTCT coverage estimate had increased to only 11% of the need. 8 Alongside the urgency to expand PMTCT services, there is also a push for programs to implement antiretroviral regimens that are more efficacious than SD-NVP and that induce less antiretroviral drug resistance. The current World Health Organization (WHO) guidelines for HIV-infected pregnant women whose personal health status does not yet qualify them to receive highly active antiretroviral therapy (HAART) recommend a three part antiretroviral regimen for PMTCT: antepartum zidovudine (ZDV) beginning at 28 weeks gestation or as soon as possible thereafter; intrapartum SD- 6

18 NVP coupled with ZDV and lamivudine (3TC) and; postpartum ZDV and 3TC for 7 days for the mother and SD-NVP along with seven days of ZDV for the infant. 9 Although far more complex than SD-NVP alone, this regimen reduces the risk of vertical transmission to about 2% and minimizes development of non-nucleoside reverse transcriptase inhibitor (NNRTI) virus in both the mother and the infant. 10,11 Alternative, less complex, regimens are also suggested for settings unable to offer the recommended regimen due to capacity constraints and for HIV-infected women who present late in gestation or at the time of delivery. 9 Similar regimens to reduce viral resistance to non-nucleoside reverse transcriptase inhibitors have similar efficacy, such as SD-NVP with intrapartum single dose tenofovir-emtricitabine. 12 PMTCT program effectiveness is influenced by both regimen efficacy and population coverage. 5 This analysis assesses possible outcomes when varying increased efficacy on one hand with possibly declining population coverage on the other; we speculate that in the world s poorest nations, the push for more efficacious, yet more complex, regimens may reduce overall coverage of at-risk pregnant women and their infants. We modeled the comparative effectiveness of three antiretroviral regimens for PMTCT considering the effects of both efficacy and coverage. We also highlight ways of improving program coverage to maximize the number of HIV-infected/exposed women/infants pairs that receive antiretroviral prophylaxis. 7

19 METHODS Background PMTCT services have been widely available in Zambia s capital city of Lusaka since The district-wide PMTCT program offers voluntary counseling and testing (VCT) for HIV to women who attended any of the 24 public-sector antenatal care clinics. A rapid HIV testing algorithm is employed and women receive their HIV test results before they return home on the same day the test is conducted. Those who test positive are provided with a single dose of nevirapine (NVP) to self-administer at the onset of labor according to the HIV Network of Prevention Trials (HIVNET) 012 regimen. 1 Back-up doses are provided in the labor ward for those in need of a replacement dose at the time of delivery. Infants born to known HIV-infected women were given a single dose of NVP in the postpartum ward prior to discharge. Both antenatal care and PMTCT services are provided free of charge. [Note: In 2005, we began scaling-up services to screen all pregnant women for eligibility for highly active antiretroviral therapy (HAART). Those that do not qualify for HAART and offered ZDV beginning at 36 weeks gestation along with intrapartum NVP.] In late 2005, we collected umbilical cord blood specimens from all deliveries occurring over four weeks in Lusaka s public-sector labor and delivery centers as part of a clinical trial (results to be published separately). Cord blood specimens were tested for HIV and positive specimens were tested for NVP. The presence of NVP in the cord blood of an HIV-infected woman was used as an objective indicator that the mother ingested NVP. Chart documentation of administration of the infant dose served as an indicator that the infant was treated. To be considered fully covered required that both 8

20 the HIV-infected woman and her exposed infant received NVP. During this four-week cord blood collection period, 47% of HIV-infected/exposed women/infant pairs delivering in Lusaka public labor wards received NVP, or were fully covered. Fifty three percent did not receive NVP or received just one of the two doses of NVP (i.e., the maternal or the infant dose but not both). After excluding seroconverters and transcription errors, the predominant reasons for failed coverage were that women did not know their HIV status ( 50%) and non-adherence among those who did ( 50%). Efficacy-Coverage Model Using assumptions from our Lusaka data, we modeled the number of infant infections prevented in a hypothetical cohort of 10,000 HIV-infected, pregnant, Zambian women in the presence of a SD-NVP program assuming 47% population coverage. We then compared this to the number of infant infections that could be prevented: 1) by implementing one of the more efficacious WHO-recommended regimens assuming the same level of population coverage was achieved, and; 2) by implementing a SD-NVP program that achieves higher levels of coverage by employing targeted strategies (including opt-out HIV testing and labor ward-based PMTCT services) to improve coverage. Thus, one presumes limited funds and decides whether increased efficacy (combination antiretrovirals) or increased coverage (more women and infants taking SD- NVP) is most beneficial in PMTCT. The three antiretroviral regimens in the model were: 1) the WHO recommended regimen with ZDV beginning at 28 weeks gestation, intrapartum SD-NVP coupled with ZDV and 3TC and postpartum ZDV and 3TC for seven days for the mother, and SD- 9

21 NVP along with seven days of ZDV for the infant (ZDV-28 + SD-NVP); 2) a version of the WHO alternative regimen with ZDV beginning at 36 weeks gestation along with SD- NVP with or without the 7-day ZDV/3TC tail for the mother, and SD-NVP along with seven days of ZDV for the infant (ZDV-36 + SD-NVP); and 3) the WHO minimum regimen of SD-NVP to the mother and infant (SD-NVP) with or without the 7-day ZDV/3TC tail (TABLE 1). The ZDV/3TC tail may or may not be given as part of the alternate or minimum regimens depending on the infrastructure of the particular PMTCT program. [Note: The addition of the ZDV/3TC tail does not appear to have an impact on efficacy however it does reduce the development of NVP resistance mutations. 11 ]. We assumed a vertical transmission rate of 21.5% in the absence of antiretroviral intervention based on the 6-8 week transmission rate in the HIVNET 012 trial and the 45 day transmission rate in the National Agency for AIDS Research (ANRS) 049a trial. 1,14 From that we approximated 6 to 8 week efficacy rates of 91% for the recommended regimen, 10 70% for the alternative regimen, 15 and 45% for the minimum regimen. 1 We identified the coverage thresholds that must be achieved for the multi-drug regimens to prevent as many infant infections as SD-NVP regimen. Breastmilk-related infections were not considered in the model although we acknowledge that the estimated transmission rates associated with the alternative and minimum regimens likely include some early breastfeeding transmissions. We also modeled the number of additional infant HIV infections that could be prevented by improving SD-NVP coverage. As in the first model, we used the Lusaka data as our benchmark and assumed that 53% of infected/exposed women/infant pairs were not fully covered either because they were unaware of their HIV status (50%) or 10

22 that they did not adhere to NVP (50%). Among those that did not know their HIV status, we assumed that 74% had not been offered or declined HIV testing during antenatal care and that 26% had either not received antenatal care or did not receive it at a center that offered PMTCT services which is consistent with our unpublished Lusaka data. RESULTS Based on our model estimates, 2,150 of the infants born to a hypothetical cohort of 10,000 HIV-infected, pregnant, Zambian women would be infected with HIV at 6 to 8 weeks of life in the absence of any intervention. A range of infant HIV infections can be averted depending upon the efficacy of the antiretroviral regimen used and the level of population coverage achieved. Given the estimated 6-8 week efficacy of SD-NVP (45%) and a baseline population coverage level of 47%, 455 infant HIV infections would be prevented out of a possible 2,150 infections (10,000 x.215 transmission rate x.47 coverage x.45 efficacy = 455, FIGURE 1). Thus, 1,695 of the 2,150 potentially infected infants born to a cohort of 10,000 HIV-infected women would still be infected with HIV given these model estimates (FIGURE 1). Implementation of either of the more efficacious multi-drug regimens would have prevented more infant infections at the same coverage level, of course, since multi-drug antiretroviral prophylaxis is more efficacious than SD-NVP (FIGURE 1). Assuming the same 47% coverage, the WHO alternative regimen (ZDV-36 + SD-NVP) would have prevented 707 infections and the WHO recommended regimen (ZDV-28 + SD-NVP) would have prevented 919 infections. 11

23 If population coverage were jeopardized by the implementation of a more logistically complex and expensive regimen, superior efficacy may not result in fewer infant infections. However, the diminution in population coverage would have to be substantial for this to occur. Population coverage would have drop from 47% to <30% following implementation of the WHO alternate regimen (ZDV-36 + SD-NVP) and to <23% following implementation of the WHO recommended regimen (ZDV-28 + SD- NVP) in order for the SD-NVP to prevent more infant infections than the ZDV-based regimens (TABLE 2). This does not consider the additional benefits of reducing viral resistance in the mother using the recommended regimen. 16 Significantly more infant HIV infections can be prevented within the existing SD- NVP-based program by improving population coverage. In our model, 5300 (53% of 10,000) HIV-infected women and their infants failed to receive both doses of NVP, i.e., they were not fully covered. Approximately half of the 5300 untreated woman were unaware of their HIV status; most of whom had declined HIV testing during antenatal care (74%) and a quarter of who had not received antenatal care (26%). Implementation of an opt-out HIV testing program in the antenatal care and delivery centers may have resulted in a substantial proportion of these women learning their HIV status and receiving NVP (FIGURE 2). Assuming 80% uptake of opt-out testing during antenatal care and 60% uptake in the labor ward, coverage would increase from 47% to 67% (6,682 of 10,000). If only half of the 80% of women tested during antenatal and the 60% tested in the labor ward adhered to SD-NVP, coverage would still increase from 47% to 57% (5,691 of 10,000). 12

24 The other 50% of HIV-infected, exposed women-infant pairs that were not fully covered consisted of women who were aware of their HIV serostatus and either did not adhere to their dose of NVP or their infants did not receive NVP. If 80% of these women and their infants received NVP in the labor and postpartum wards respectively, coverage would have increased from 47% to 68% (6,820 of 10,000). Together these interventions could potentially increase coverage from 47% to 88% (8,802 of 10,000), nearly doubling the number of infant HIV-infections prevented. DISCUSSION In this report, we have emphasized the dual roles of antiretroviral efficacy and population coverage in maximizing PMTCT program effectiveness. Our model demonstrates that neither efficacy nor coverage alone ensures optimal program effectiveness. Multi-drug regimens can prevent more infant HIV infections than SD- NVP at the same level of population coverage and even at lower levels. Nonetheless, implementation of more efficacious regimens will only benefit the women and infants participating in the program and adhering to treatment. Additional measures are required to identify the women/infant pairs that are not accessing or benefiting from the existing services and to provide services that target these at-risk groups. From a population perspective, both the efficacy of the regimen used as well as the coverage achieved with that regimen determines the bottom line: infant HIV infections prevented. Despite the availability of more efficacious multi-drug regimens, we believe that SD-NVP has a continuing important role in PMTCT, particularly among women who 13

25 learn their HIV status while in labor. 9 Use of SD-NVP to prevent vertical transmission of HIV may also be preferred in settings with substantial capacity constraints due to poverty, refugee populations, civil strife or war, or essential commodities shortages (e.g., the least developed countries as defined by the United Nations). 17 Once a SD-NVP program is in place, providers can strive to implement more efficacious regimens over time while continuing to seek increasing population coverage. As such, SD-NVP programs provide an essential foothold for VCT services as programs work to implement more efficacious regimens. Population coverage can be enhanced in a number of ways including: implementation of an opt-out HIV testing strategy as is recommended by UNAIDS and WHO; 18 use of rapid HIV tests and onsite testing to ensure same-day results; 18 and administration of antiretroviral prophylaxis to HIV-infected women on the day of HIV testing. 4,19 The wider availability of more efficacious antiretroviral regimens challenges all PMTCT programs to use superior multi-drug regimens, 9 some of which should also reduce viral resistance. 11,12 The President s Emergency Plan for AIDS Relief (PEPFAR) and the Global Fund to Fight AIDS, Tuberculosis, and Malaria are making it possible to put more and more pregnant women on potent antiretroviral therapy during pregnancy. The addition of ZDV to the SD-NVP regimen appears to be highly acceptable to patients. 20 It will be important for program managers to assess adherence and population coverage to ensure that population coverage is not jeopardized while scaling up to multi-drug regimens. One limitation of this model is that we have assumed that full coverage (i.e., ingestion of both mother and infant doses) is required in order for the infant to be spared 14

26 infection. However, antenatal and intrapartum zidovudine alone can reduce vertical transmission and intrapartum NVP alone may also be efficacious. 10 Extended postexposure prophylaxis using ZDV and NVP in infants born to untreated HIV-infected mothers has also been shown to be efficacious 24 and is recommended by the WHO. 9 Considering that our model has not accounted for the effect of partial treatments, it is likely that we have underestimated the number of infant infections prevented. In addition, our efficacy estimates are based on studies from different regions and among populations with varying degrees of breastfeeding. Lastly, we do not consider breastfeeding transmissions, an issue that will be compelling once current clinical trials to assess NVP prophylaxis for breastfeeding infants of HIV-infected mothers (e.g., HIV Prevention Trials Network protocol 046) are completed. Ideally, we would like to have available the most efficacious regimen for PMTCT and provide it to all HIV-infected women. 25 Expanded availability of HIV testing is an important first step. 26 In the short term, we may have to strike a balance by offering the more efficacious and complex multi-drug regimens in settings where there are resources and infrastructure to sustain their use while providing the simpler and less efficacious SD-NVP regimen in places with much less support. Ministries of Health must weigh the potential advantages and disadvantages of rapidly transitioning to multi-drug antiretroviral prophylactic regimens vs. expanding simpler (i.e., SD-NVP) PMTCT services as an intermediary step, while building infrastructure and capacity. Components of such programs will include introducing multi-drug regimens into their well established PMTCT program sites, providing SD-NVP to women who learn their HIV status in the labor ward, expanding services with SD-NVP in their least experienced locations, and 15

27 implementing universal opt-out HIV testing with same-day results in antenatal care and labor wards. At a time when many high prevalence sub-saharan African countries are in the process of scaling- up their PMTCT programs, countries can combine these programmatic options to maximize efficacy and population coverage in cost-effective ways. 16

28 Acknowledgements: Our work in prevention of mother-to-child transmission of HIV was supported by the Call-to-Action initiative of the Elizabeth Glaser Pediatric AIDS Foundation. Investigators received support from the U.S. National Institutes of Health (U01 AI47972, D43 TW01035, K23 HD01411, K01 TW06670, K23 AI52481) and from the HIV Prevention Trials Network (U01 AI047972, 1U01 AI068619). 17

29 Reference List (1) Guay LA, Musoke P, Fleming T et al. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: HIVNET 012 randomised trial. Lancet 1999 September 4;354(9181): (2) Jackson JB, Musoke P, Fleming T et al. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: 18-month follow-up of the HIVNET 012 randomised trial. Lancet 2003 September 13;362(9387): (3) Doherty TM, McCoy D, Donohue S. Health system constraints to optimal coverage of the prevention of mother-to-child transmission programme in South Africa: lessons from the implementation of the national pilot programme. African Health Sciences 2005;5(3): (4) Manzi M, Zachariah R, Teck R et al. High acceptability of voluntary counselling and HIV-testing but unacceptable loss to follow up in a prevention of mother-tochild HIV transmission programme in rural Malawi: scaling-up requires a different way of acting. Trop Med Int Health 2005 December;10(12): (5) Stringer JS, Sinkala M, Maclean CC et al. Effectiveness of a city-wide program to prevent mother-to-child HIV transmission in Lusaka, Zambia. AIDS 2005 August 12;19(12): (6) Temmerman M, Quaghebeur A, Mwanyumba F, Mandaliya K. Mother-to-child HIV transmission in resource poor settings: how to improve coverage? AIDS 2003 May 23;17(8): (7) United Nations General Assembly. The Declaration of Commitment on HIV/AIDS: five years later, Report of the Secretary General. Geneva; 2006 Mar 24. (8) United Nations General Assembly. Declaration of Commitment on HIV/AIDS and Political Declaration on HIV/AIDS: focus on progress over the past 12 months Mar 20. (9) World Health Organization. Antiretroviral Drugs for Treating Pregnant Women and Preventing HIV Infection in Infants in Resource-Limited Settings. Geneva; (10) Lallemant M, Jourdain G, Le CS et al. Single-dose perinatal nevirapine plus standard zidovudine to prevent mother-to-child transmission of HIV-1 in Thailand. N Engl J Med 2004 July 15;351(3): (11) McIntyre J, Martinson N, Grey GE. Addition of short course Combivir (CBV) to single dose Viramune (sdnvp) for the prevention of mother-to-child transmission 18

30 (pmtct) of HIV-1 can significantly decrease the subsequent development of maternal and paediatric NNRTI-resistant virus. 3 rd International AIDS Society Conference, Rio de Janeiro, Brazil 2005 Jul 24, Abstract TuFo0204. (12) Chi BH, Sinkala M, Mbewe F et al. Single-dose tenofovir and emtricitabine for reduction of viral resistance to non-nucleoside reverse transcriptase inhibitor drugs in women given intrapartum nevirapine for perinatal HIV prevention: an open-label randomised trial. Lancet 2007 November 17;370(9600): (13) Stringer EM, Sinkala M, Stringer JS et al. Prevention of mother-to-child transmission of HIV in Africa: successes and challenges in scaling-up a nevirapine-based program in Lusaka, Zambia. AIDS 2003 June 13;17(9): (14) Dabis F, Msellati P, Meda N et al. 6-month efficacy, tolerance, and acceptability of a short regimen of oral zidovudine to reduce vertical transmission of HIV in breastfed children in Cote d'ivoire and Burkina Faso: a double-blind placebocontrolled multicentre trial. DITRAME Study Group. DIminution de la Transmission Mere-Enfant. Lancet 1999 March 6;353(9155): (15) Dabis F, Bequet L, Ekouevi DK et al. Field efficacy of zidovudine, lamivudine and single-dose nevirapine to prevent peripartum HIV transmission. AIDS 2005 February 18;19(3): (16) McIntyre J. Preventing mother-to-child transmission of HIV: successes and challenges. BJOG 2005 September;112(9): (17) United Nations Office of the High Representative for the Least Developed Countries. List of Least Developed Countries (18) UNAIDS/WHO. UNAIDS/WHO Policy Statement on HIV Testing (19) van't Hoog AH, Mbori-Ngacha DA, Marum LH et al. Preventing mother-to-child transmission of HIV in Western Kenya: operational issues. J Acquir Immune Defic Syndr 2005 November 1;40(3): (20) Leroy V, Karon JM, Alioum A et al. Twenty-four month efficacy of a maternal short-course zidovudine regimen to prevent mother-to-child transmission of HIV- 1 in West Africa. AIDS 2002 March 8;16(4): (21) Shaffer N, Chuachoowong R, Mock PA et al. Short-course zidovudine for perinatal HIV-1 transmission in Bangkok, Thailand: a randomised controlled trial. Bangkok Collaborative Perinatal HIV Transmission Study Group. Lancet 1999 March 6;353(9155): (22) Wiktor SZ, Ekpini E, Karon JM et al. Short-course oral zidovudine for prevention of mother-to-child transmission of HIV-1 in Abidjan, Cote d'ivoire: a randomised trial. Lancet 1999 March 6;353(9155):

31 (23) Taha TE, Kumwenda NI, Gibbons A et al. Short postexposure prophylaxis in newborn babies to reduce mother-to-child transmission of HIV-1: NVAZ randomised clinical trial. Lancet 2003 October 11;362(9391): (24) Walensky RP, Paltiel AD, Losina E et al. The survival benefits of AIDS treatment in the United States. J Infect Dis 2006 July 1;194(1):11-9. (25) Vermund SH, Wilson CM. Barriers to HIV testing--where next? Lancet 2002 October 19;360(9341):

32 Table 1. Characteristics of the three antiretroviral regimens compared in the model Characteristic Maternal regimen Antepartum Recommended Regimen 1 (ZDV-28 + SD-NVP) ZDV (from 28 wks gestation) Alternative Regimen 2 (ZDV-36 + SD-NVP) ZDV (from 36 wks gestation) Minimum Regimen 3 (SD-NVP) Intrapartum SD-NVP + ZDV/3TC SD-NVP + ZDV SD-NVP Postpartum ZDV/3TC x 7d ±ZDV/3TC x 7d 4 ±ZDV/3TC x 7d 4 Infant regimen SD-NVP + ZDV x 7d SD-NVP + ZDV x 7d 5 SD-NVP Estimated vertical transmission 2% at 6 mo 6.4% at 6 wks 11.9% at 6-8 wks rate Estimated efficacy 6 91% at 6 mo 70% at 6 wks 45% at 6-8 wks References Lallemant, et al., 2004 Dabis et al., 2005 Guay et al, 1999 Infants breast <1% 54% 98% fed 1 World Health Organization (WHO) recommended regimen: ZDV-28 + SD-NVP refers to antepartum zidovudine (ZDV) beginning at 28 weeks gestation, intrapartum single dose nevirapine (SD-NVP) coupled with intrapartum ZDV and lamivudine (3TC), along with a 7-day tail of ZDV and 3TC for the mother postpartum; the infant receives SD- NVP along with seven days of ZDV. 2 WHO alternative regimen: ZDV-36 + SD-NVP refers to maternal ZDV beginning at 36 weeks gestation along with intrapartum SD-NVP (with or without the 7 day tail of maternal ZDV and 3TC); the infant receives SD-NVP along with seven days of ZDV. 3 WHO minimum regimen: SD-NVP refers to single-dose nevirapine to mother and infant (with or without the addition of the 7 day tail of maternal ZDV and 3TC). 4 Intrapartum and postpartum ZDV and 3TC may be given to mothers as part of the alternative and minimum regimens to reduce the development of NVP resistance mutations. The addition of the ZDV/3TC tail does not appear to impact efficacy. 5 ZDV is given to the infant for 4 weeks if the mother receives <4 weeks of antepartum ZDV. 6 Estimated efficacy assumes transmission rate of 21.5% in the absence of antiretroviral intervention

33 Figure 1. Projected number of infant HIV infections prevented in a hypothetical cohort of 10,000 HIV-infected, pregnant Zambian women using three different antiretroviral regimens and assuming population coverage of 47% 10,000 HIV-infected pregnant, Zambian women 47% coverage 4,700 HIV-infected -exposed women-infant pairs receive both doses of the SD-NVP prophylactic regimen 21.5% transmission rate 1,010 HIV-infected infants in the absence of antiretroviral prophylaxis SD-NVP ZDV-36 + SD-NVP ZDV-28 + SD-NVP 45% efficacy 70% efficacy 91% efficacy 455 Infant HIV infections prevented 707 Infant HIV infections prevented 919 Infant HIV infections prevented 22

34 Table 2. Population coverage thresholds necessary for the World Health Organization (WHO) recommended regimen (ZDV-28 + SD-NVP), WHO alternative regimen (ZDV SD-NVP) and WHO minimum regimen (SD-NVP) to prevent equal numbers of infant HIV infections in a hypothetical cohort of 10,000 HIV-infected, pregnant, Zambian women assuming a baseline transmission rate of 21.5%. Infant HIV Infections Prevented Recommended Regimen 1 Population Coverage Needed Alternative Regimen 2 Minimum Regimen 3 (ZDV-28 + SD-NVP) (ZDV-36 + SD-NVP) (SD-NVP) % 17% 26% % 30% 47% % 33% 52% % 50% 78% % 64% 100% % 66% % 100% % None of the above regimens would be likely to protect all 2150 infants 1 World Health Organization (WHO) recommended regimen: ZDV-28 + SD-NVP refers to antepartum zidovudine (ZDV) beginning at 28 weeks gestation, intrapartum single dose nevirapine (SD-NVP) coupled with intrapartum ZDV and lamivudine (3TC), along with a 7-day tail of ZDV and 3TC for the mother postpartum; the infant receives SD- NVP along with seven days of ZDV. 2 WHO alternative regimen: ZDV-36 + SD-NVP refers to maternal ZDV beginning at 36 weeks gestation along with intrapartum SD-NVP (with or without the 7 day tail of maternal ZDV and 3TC); the infant receives SD-NVP along with seven days of ZDV. 3 WHO minimum regimen: SD-NVP refers to single-dose nevirapine to mother and infant (with or without the addition of the 7 day tail of maternal ZDV and 3TC). 23

35 Figure 2. Projected number of infant HIV infections that occur despite available prevention of mother-to-child transmission (PMTCT) services and the effect of targeted interventions to improve coverage and reduce the number of infant HIV infections in a single-dose nevirapine (SD-NVP)-based PMTCT program. 10,000 HIV-infected pregnant, Zambian women 53% not fully covered 5300 HIV-infected -exposed women-infant pairs did not receive both doses of the SD-NVP regimen 2,650 (50%) women of unknown HIV status 2,650 (50%) women with known HIV infection 1,961 (74%) not offered or declined VCT during ANC 689 (26%) did not receive ANC or PMTCT services 2,650 (100%) mother-infant pairs did not adhere to or receive NVP Targeted interventions to improve coverage and prevent more infant HIV infections Opt-out testing during ANC 20%-80% uptake 392 to 1569 HIV+ women would have learned their status 38 to 152 infant infections prevented % transmission * 45% efficacy Opt-out testing in the Labor Ward 138 to 413 HIV+ women would have learned their HIV status 13 to 40 infant infections prevented 20%-60% uptake % transmission * 45% efficacy Provide replacement NVP dose in labor ward and re-train staff to ensure mother and baby receive NVP 51 to 205 infant HIV infections prevented 20% - 80% success success 530 to 2120 HIV+ womeninfant pairs would have received complete course of SD-NVP + + = = 21.5% transmission * 45% efficacy 1060 to 4102 more womeninfant pairs covered (increase of 11% to 41%) 102 to 397 additional infant HIV infections averted 24

36 A CLUSTER-RANDOMIZED TRIAL OF ENHANCED LABOR WARD-BASED PMTCT SERVICES TO INCREASE POPULATION NEVIRAPINE COVERAGE IN LUSAKA, ZAMBIA KAREN M. MEGAZZINI, MOSES SINKALA, STEN H. VERMUND, DAVID T. REDDEN, DANIEL W. KREBS, EDWARD P. ACOSTA, JOYSE MWANZA, ROBERT L. GOLDENBERG, NAMWINGA CHINTU, and JEFFREY S.A. STRINGER In preparation for AIDS Formatted for dissertation 25

37 ABSTRACT Background: Only 11% of HIV-infected women in low and middle income countries received antiretroviral prophylaxis for prevention of mother-to-child transmission (PMTCT) in Provision of PMTCT services in the labor ward may improve population nevirapine (NVP) coverage. Objective: To determine whether an enhanced labor ward-based PMTCT package would improve population NVP coverage in a setting where antenatal PMTCT services were already in place. We defined population coverage as the proportion of HIVinfected/exposed mother/infant pairs with confirmed NVP ingestion. Methods: Between 12 October 2005 and 15 January 2006 we conducted a clusterrandomized clinical trial of enhanced labor ward-based PMTCT services at 12 publicsector delivery centers in Lusaka, Zambia. Following a baseline surveillance period, six labor wards offered HIV testing to women of unknown HIV status along with NVP administration to those who tested seropositive and conducted a rapid, structured assessment of NVP adherence among women known to be HIV-infected. Six control labor wards provided the standard of care. We used generalized estimating equations to compare nevirapine coverage pre- and post-intervention between the two groups. Results: During the study period, 7664 women gave birth in Lusaka public clinics. Cord blood was collected from 7592 (99%) of the deliveries and 7438 (97%) of them were tested for HIV. Of the 7438 specimens, 1618 (22%) were seropositive, of which

38 (79%) were tested for NVP. In the intervention clinics, coverage improved 24% (from 42% at baseline to 52% when the labor ward PMTCT services were available.) At the same time, coverage declined 19% (from 53% to 43%) in the control clinics. At baseline (pre-intervention), the odds of HIV-infected/exposed women/infant pairs having both received NVP in the intervention clinics was 0.78 times the odds of being covered in the control clinics whereas in the follow-up period (post-intervention), the odds of being covered in the intervention clinics was 1.5 times the odds of being covered in the control clinics. This change represents a multiplicative effect of 1.91 upon the odds ratio at baseline (OR 1.91, 95% CI, ). Conclusion: Labor ward-based PMTCT programs are feasible and can have significant, positive impact on population NVP coverage. Key words: HIV, PMTCT, mother-to-child transmission, nevirapine, antiretroviral prophylaxis, coverage, rapid HIV testing, adherence, sub-saharan Africa, Zambia 27

39 INTRODUCTION Provision of antiretroviral prophylaxis to HIV-infected women and their HIVexposed infants is the cornerstone of pediatric HIV prevention. HIV testing during pregnancy is a critical entry point for access to these services. But despite its seeming simplicity, prevention of mother-to-child transmission (PMTCT) programs in low and middle income countries face considerable challenges in successfully identifying and treating all at risk women and their infants. 1-6 In 2003, we conducted a city-wide surveillance in Lusaka. We found that only 30% of the HIV-infected/exposed women/infant pairs who received antenatal care within the Lusaka Urban District public antenatal care centers had ingested intrapartum nevirapine (NVP), i.e. population NVP coverage was 30%. 5 We measured maternal NVP coverage by collecting umbilical cord blood from all deliveries that occurred in the public-sector labor wards during the 3-month surveillance period, testing it for HIV antibodies, and then further testing those that were seropositive for NVP. 7 We used chart documentation of infant NVP administration as an indication of infant NVP coverage. Both the HIV-infected mother and her HIV-exposed infant must have received NVP to be considered covered. Several factors contributed to the low level of population NVP coverage in the surveillance exercise. Among the 8787 women who received antenatal care within Lusaka and gave birth during the surveillance period, 3638 of them (41%) were unaware of their HIV infection status at the time of delivery. Of these,1583 (44%) had not been offered HIV testing during antenatal care and more than half (2055 of 3638, 56%) had declined VCT during antenatal care. 5 A second major factor that contributed to the low 28

40 population coverage was NVP non-adherence. Of the 1112 women who were diagnosed with HIV infection during antenatal care and provided a take-home dose of NVP for self administration at labor onset, 361 (32%) did not have NVP in their umbilical cord blood. 5 This high rate of non-adherence occurred despite replacement doses of NVP being available in the labor wards. With this background, we conducted a cluster-randomized clinical trial designed to determine whether provision of PMTCT services in the labor ward, as an adjunct to antenatal care-based services, would improve population NVP coverage. METHODS Background In Zambia s capital city of Lusaka, a district-wide PMTCT program has been operational since This program offers VCT for HIV to women who attend antenatal care in any of the 24 public-sector antenatal care clinics. Those who test positive are provided with a single dose of NVP to self-administer at the onset of labor according to the HIV Network of Prevention Trials (HIVNET) 012 regimen. 8 Back-up doses are provided in the labor ward for those who report the need for a replacement dose. Infants born to known HIV-infected women are given a single dose of NVP in the postpartum ward prior to discharge. All obstetrical services (including PMTCT services) are provided free of charge. The study was conducted between 12 October 2005 and 15 January 2006 at 12 public-sector delivery centers in Lusaka, Zambia. Enhanced labor ward-based services included separate approaches depending upon HIV serostatus on presentation. Those of 29

41 unknown status were offered labor ward-based HIV testing, and, if seropositive, immediate single-dose NVP administration. Those presenting with known HIV seropositive status were administered a rapid, structured assessment of NVP adherence in the labor ward. Population NVP coverage was the primary outcome of this study. Coverage was defined as the proportion of HIV-infected women and their HIV-exposed infants that received NVP. It was measured by collecting anonymous umbilical cord blood specimens from all deliveries, testing them for HIV antibodies, and testing seropositive specimens for NVP (indicating maternal coverage). Infant NVP coverage was determined by chart documentation of administration of NVP by clinic personnel prior to discharge. Both the HIV-infected women and her HIV-exposed infant must have received NVP to be considered covered. The trial consisted of a baseline cord blood surveillance period (pre-intervention) and a follow-up cord blood surveillance period (post-intervention). Baseline Cord Blood Surveillance Period Between 12 October 2005 and 11 November 2005 we conducted the baseline surveillance phase of the trial to determine the background population NVP coverage level in Lusaka. During this phase, anonymous umbilical cord blood specimens were collected from all discarded placentas at the time of delivery in each of the 12 publicsector labor and delivery centers in Lusaka District. The District midwifery staff in each labor ward was trained to collect the cord blood specimens and to extract non-identifying information from the medical record onto an anonymous surveillance form. The 30

42 surveillance form gathered antenatal and delivery care information including age of the mother, gravidity, number of antenatal care visits, shift of admission, length of time between admission and delivery, whether or not the woman had been offered HIV counseling and testing during antenatal care (and if so, whether she accepted testing), her HIV status and if infected whether or not she received NVP prior to delivery and her baby received NVP prior to discharge. The form contained an anonymous identification number that was also transcribed onto the woman s cord blood specimen linking the data on the surveillance form to the cord blood result. In the event that cord blood could not be collected, the midwife noted the reason on the surveillance form. Single specimens were collected from multiple births. Specimens were not collected from placentas of babies born prior to arrival at the delivery center. Following completion of the baseline cord blood surveillance period, the 12 District labor wards were stratified according to size (based on the number of deliveries per month) and their historical NVP coverage level (based on the 2003 surveillance data) 5 and then randomized to the intervention and control arms of the trial. Follow-up Cord Blood Surveillance Period Between 3 December 2005 and 15 January 2006 we conducted the intervention phase of the trial. As in the baseline surveillance period, cord blood was collected from all deliveries and the District midwifery staff completed an anonymous surveillance form on all women. During this phase of the trial, the labor ward PMTCT package, consisting of rapid HIV testing and a rapid NVP adherence assessment, was provided by the District 31

43 midwifery staff in the six labor wards assigned to the intervention arm. Routine, standard of care services were provided in the six labor wards assigned to the control arm. Rapid HIV Testing Women unaware of their HIV status were offered HIV counseling and testing if they were in the first stage of labor and the midwife deemed there was sufficient time to complete counseling and testing before delivery. All women who chose to be tested for HIV signed a consent form for HIV testing, which was standard procedure in Lusaka District at the time of the study. An HIV counseling flip chart was provided to help standardize the counseling message for the women being counseled in labor. This flip chart was available in English, and the two most common local languages spoken in Lusaka: Nyanja and Bemba. In consideration for the labor ward setting in which this study was conducted, women who chose to be tested were given the option of learning their HIV test results either before or after delivery. Those who choose to postpone learning their results until after delivery could nonetheless consent to NVP for themselves and their babies at the time of testing by indicating their choice on the consent form. The consent form explained that they would be offered a tablet despite the results of their HIV test. Women who tested positive were given a NVP tablet (200mg) and their exposed infants received NVP syrup (2mg/kg). We offered a calcium tablet that resembled NVP to all women who tested negative for HIV so as not to inadvertently disclose a woman s status during labor or stigmatize HIV-infected women by only providing tablets to those who tested positive. 32

44 We used a point-of-care rapid HIV testing strategy in the labor wards. Midwives were equipped with clip-on timers that allowed them to keep track of HIV tests that were in process while they continued with their work. The standardized HIV testing algorithm used in Lusaka District antenatal care centers was employed. Women were screened for HIV using the DETERMINE HIV-1/2 rapid test (DETERMINE HIV-1/2, Abbott Laboratories, Abbott Park, Illinois, USA) and positive results were confirmed with the GENIE II HIV 1/2 rapid test (GENIE II HIV-1/2, BioRad Laboratories, Hercules, California, USA). NVP was administered to all women who tested positive on the DETERMINE test. Confirmatory HIV testing was done on positive specimens using the GENIE II rapid test either before or after delivery as time permitted. Post-test HIV counseling was provided to all women prior to discharge from the center. Women were encouraged to learn their test results and receive post-test counseling after delivery. Those who opted to learn their results prior to delivery could do so assuming there was sufficient time for post-test counseling: a decision left to the discretion of the midwife. Women were informed of their infection status during the posttest counseling session unless they declined to learn their status at that time. Rapid NVP Adherence Assessment The rapid assessment of NVP adherence was conducted among women known to be HIV-infected upon admission to the labor wards that were assigned to the intervention arm of the study. This brief assessment was used to help ensure that women identified as HIV-infected during antenatal care actually ingested NVP at the onset of labor. The practice in Lusaka District Labor and Delivery Centers was to ask women whether or not 33

45 they have ingested the NVP tablet provided to them during antenatal care. At the time of the study, a direct yes/no question was routinely asked (e.g. Did you take your nevirapine tablet? ). Women who reported not having ingested NVP were given a backup dose of NVP. No further action was taken for women who reported that they had ingested NVP. In the intervention arm of the study, we replaced the direct yes/no question with a structured multiple choice question. Midwives in the intervention labor wards were instructed to ask women about NVP ingestion in the following way: Nevirapine is a medication that may help prevent your baby from getting HIV. Which statement best describes your interest in taking nevirapine? 1. I did not take nevirapine before coming here and do not want to take nevirapine. 2. I haven t taken nevirapine yet but would like to take one now. 3. I already took my nevirapine tablet. Midwives indicated on the surveillance form whether or not they used this approach. Umbilical Cord Blood Collection and Processing The umbilical cord blood specimens were collected in 10mL EDTA vacuum tubes and stored in the labor wards at 2 C to 8 C until pick-up by the study team. The study team reviewed all surveillance forms and corrected any errors or discrepancies in consultation with the delivering midwife or midwife in-charge. Forms and specimens were picked up daily and transported to a central laboratory where they were separated and the plasma analyzed for HIV antibodies using the DETERMINE rapid HIV test 34

46 (DETERMINE HIV-1 / 2, Abbott Laboratories, Abbott Park, Illinois, USA). In the event that the cord blood and antenatal care result were discordant, we used the cord blood results. After rapid HIV testing, all plasma specimens were aliquoted and frozen at -70 C. HIV seropositive specimens were shipped on dry ice to the University of Alabama at Birmingham (UAB) pharmacology lab where they were assayed for NVP via High Performance Liquid Chromatography (HPLC). This assay was capable of detecting NVP in plasma at concentrations as low as 25ng/mL 9 and in a previous study in Lusaka was able to detect NVP in the cord blood of 178 of 179 women (>99%) of women in whom NVP ingestion was directly observed. 10 Statistical Analysis Microsoft Excel 2003 (Microsoft Corp, Redmond, WA) was used to randomize labor wards to the intervention and control arms of the trial. All statistical tests were performed using SAS 9.1 software (SAS Institute Inc., Cary, NC, USA). Generalized estimating equations (GEE) were used to develop a model for determining NVP coverage in this study. A GEE model was used because of its ability to adjust for the nonindependence of individuals nested within a clinic. The study was approved by the Institutional Review Boards at the University of Alabama at Birmingham (Birmingham, Alabama, USA) and the University of Zambia (Lusaka, Zambia). RESULTS During the baseline and intervention surveillance periods, 7664 women gave birth in the 12 Lusaka District labor wards. We collected cord blood specimens from

47 (99%) of deliveries and received results on 7438 (97%) of them. Women from whom cord blood specimens were not obtained were similar to those from whom cord blood was obtained in terms of age, gravity, number of antenatal care visits, nursing shift at the time of admission, and place of first antenatal care visit (data not shown). Cord blood testing indicated that 1618 of 7438 (22%) women were HIV seropositive. We tested 1279 of the 1618 (79%) seropositive specimens for NVP by HPLC and detected NVP in 816 (64%) of the specimens. The remaining 339 (21%) specimens were not tested for NVP due to a lack of study resources. Seropositive cord blood specimens that were not analyzed were distributed between the control (145 of 760, 19.1%) and intervention clinics (194 of 858, 22.6%). In order to examine for systematic bias, we modeled the probability of samples being unanalyzed as a function of specific covariates including: age; number of antenatal care visits; nursing shift at time of admission; and place of first antenatal care visit). The analysis did not indicate that any systematic bias due to the samples being unanalyzed existed. During the 4-week baseline surveillance period, 3123 women gave birth to infants in the 12 public-sector labor wards in Lusaka District. Although randomization to the control and intervention arms did not occur until after completion of the baseline surveillance period, we retrospectively assigned the clinics to their respective arms for analysis purposes. During the six-week follow-up surveillance period, 4541 women gave birth to infants in the same 12 public-sector labor wards. Of these, 2435 (54%) deliveries occurred in the labor wards assigned to the intervention arm and 2106 (46%) occurred in the control arm (FIGURE 1). The characteristics of women who delivered in the publicsector labor wards during the baseline and follow-up surveillance periods, according to 36

48 control and intervention arm assignment, are shown in TABLE 1. We built models adjusting for clustering within clinics and did not find any statistical evidence that the proportion of women eligible for either of the labor ward interventions (rapid HIV testing or the rapid assessment of NVP adherence) differed by study period (baseline or followup) or intervention group (control or intervention, data not shown). Population NVP Coverage In unadjusted analysis, implementation of the labor ward PMTCT package had a significant, positive effect on population NVP coverage. In the intervention clinics, coverage among HIV-infected/exposed women/infant pairs improved 24% from 42% at baseline to 52% when the labor ward PMTCT services were available. At the same time, coverage declined by 19% from 53% to 43% in the control clinics (TABLE 2, FIGURE 2). During the baseline surveillance period (pre-intervention), the odds of being covered in the intervention clinics were 0.78 times the odds of being covered in the control clinics whereas during the follow-up period (post-intervention), the odds of being covered in the intervention clinics were 1.5 times the odds of being covered in the control clinics. This change represents a multiplicative effect of 1.91 upon the odds ratio at baseline (OR 1.91, 95% CI, ) with women-infant pairs in the intervention clinics having nearly twice the odds of women-infant pairs in the control clinics of being covered. In bivariate analysis, maternal age, number of antenatal care visits, nursing shift at the time of admission and place of first antenatal care visit were each significantly associated with NVP coverage (data not shown). In a multivariable model adjusted for these four variables, the estimated effect of the intervention remained significant. The 37

49 odds of HIV-infected/exposed women/infant pairs being covered in the intervention clinics was 0.73 times the odds of being covered in the control clinics during the baseline study period and 1.39 times the odds of being covered during the intervention period. As with the unadjusted model, the intervention was associated with a near doubling of the odds of coverage (Adjusted Odds Ratio (AOR) 1.90, 95% CI ). In subset analysis, the rapid HIV testing intervention had a significant, positive effect on population NVP coverage among women who presented to the labor ward unaware of their HIV infection status (TABLE 3). Prior to implementing this intervention, NVP coverage among HIV-infected women who were unaware of their HIV status upon presentation in labor was zero. Following implementation of this strategy, NVP coverage increased to 16%. The rapid assessment of NVP adherence increased NVP coverage by 4% among women attending the intervention clinics (from 63% during the baseline surveillance period to 67% during the follow-up surveillance period) compared to a 9% drop in coverage documented in the control clinics during the study period (from 74% during the baseline surveillance period to 65% in the follow-up surveillance period, p- value=0.14, TABLE 4). DISCUSSION We found that population NVP coverage can be significantly improved by providing enhanced labor ward-based PMTCT services that target women of unknown HIV status (by offering them HIV testing and treatment) and target known HIV-infected women (by helping to ensure adherence to NVP). Four out of the six labor wards in the intervention arm experienced clinically significant increases in coverage ranging from 38

50 18% to 33%. Overall, our labor ward-based package was associated with a 10% increase in population coverage that occurred in the presence of falling coverage in the control clinics. In order for an HIV-infected woman and her HIV-exposed infant to receive appropriate antiretroviral therapy, the woman must: 1) have access to a program that offers HIV testing and antiretroviral prophylaxis; 2) be offered the services; 3) agree to be tested; 4) receive her test results; 5) receive the prescribed antiretroviral prophylactic regimen; and 6) ingest the prophylaxis; and 6) her infant must also receive prophylaxis after delivery. PMTCT programs, including our own, have repeatedly reported losses at each level of this prevention cascade. 1-5 One of the strengths of our labor ward-based strategy was that it targeted women at multiple steps of the cascade. Rapid HIV testing provided women who were unaware of their HIV status with an opportunity to learn their status and to receive prophylaxis if indicated. For the women who were not offered testing during antenatal care, or had not received antenatal care, the labor ward presented itself as their only opportunity to prevent vertical transmission. For those that declined testing during antenatal care, the labor ward testing services provided them with a final chance at prevention. The rapid assessment of NVP adherence targeted women further down the attrition cascade by helping to enhance NVP adherence among women who had been diagnosed with HIV during antenatal care. We had previously documented NVP non-adherence in as many as a third of HIV-infected women diagnosed with HIV in antenatal care clinics in Lusaka so believed an intervention specifically targeting adherence would be germane to improving population coverage. 5 39

51 A second strength of this study was that it was anonymous. The labor ward interventions were offered to all eligible women and receiving the labor ward-based PMTCT services was not dependent on women consenting to participate in a research study. Had this been the case, the study would have been subject to selection bias. Our study was not powered a priori to look at the individual effects of rapid HIV testing or the rapid assessment of NVP adherence. In subgroup analysis, coverage among women unaware of their HIV infection status, therefore eligible for rapid HIV testing, improved from 0% in the absence of labor ward testing to 16% with rapid testing. In five out of six of the intervention labor wards, provision of labor ward HIV testing resulted in an absolute increase in coverage ranging from 17% to 25% (data not shown). In one labor ward, the improvement in coverage was less dramatic (4%). At the same time, the rapid NVP adherence assessment was associated with an overall increase in coverage of 4% among women in the intervention clinics that were aware of their HIV serostatus at the time of admission (63% baseline, 67% follow-up) compared to a 9% drop in coverage among the same group of women in the control clinics (74% baseline, 65% follow-up, p=0.14). This assessment was associated with a positive or neutral effect on coverage (range, 1% to 24%) in all but one of the labor ward where coverage dropped 10% when the rapid NVP adherence assessment was offered. The labor ward that experienced this loss in coverage was the same labor ward that experienced the smallest improvement in coverage associated with the labor ward HIV testing strategy (i.e. 4%). These results were not entirely surprising to us. This particular labor ward was often under staffed compared to the others and the midwives were less enthusiastic about implementing the study strategies. We ve previously 40

52 documented that low staff morale can adversely affect uptake of services in a clinical trial setting and suspect this may have been the case in this study as well. 11 Staff shortages have also been reported by others as an impediment to labor ward counseling and 12, 13 testing. In order for this intervention to be optimally effective, its impact on workload and staffing shortages must be taken into consideration. Overall, only 19% of women presenting to the labor wards in Lusaka during this trial were unaware of their HIV status, a marked improvement over the 50% noted in our 2003 cord blood surveillance study. Even with relatively few women being eligible for this intervention, rapid HIV testing in the labor ward resulted in a significant improvement in population coverage. In settings where women commonly present to the labor ward unaware of their HIV status 4, intrapartum testing and treatment could be expected to have an even more substantial impact on population NVP coverage. At the time of this study, HIV testing in Lusaka was performed using an opt-in testing strategy and women were required to provide informed consent prior to testing. Marked improvements in testing acceptance following implementation of opt-out testing has been demonstrated in both antenatal care and labor ward-based PMTCT programs. 12, 13, 18 We suspect that had an opt-out testing strategy been employed, as is recommended by the CDC and WHO/UNAIDS 19, 20 a greater proportion of women-infant pairs would have participated in this intervention, further improving NVP coverage. Our labor ward intervention had a component that was specifically designed to target women who tested positive for HIV during antenatal care with the intention of improving NVP adherence among this group of women. Previously, we found that as many as one-third of women in the Lusaka PMTCT program were not adherent to their 41

53 NVP dose 5 and in a clinical trial setting we found that 28% of women who were queried about NVP ingestion, falsely reporting taking NVP. 11 We suspected that this level of nondisclosure was in part the result of women wanting to please the midwife or to avoid providing the incorrect answer. A similar problem with disclosure has been 21, 22 documented among women when asked about cigarette smoking during pregnancy. However, Mullen and colleagues found that discloser of smoking status during pregnancy is improved when women are queried about smoking using a multiple choice question that allows them to describe themselves as having cut down rather than forcing a yes or no response. 23 Based on this concept, our approach allowed women to express their interest in taking NVP rather than simply answering yes or no to whether or not they had ingested NVP. Although this study was not powered for secondary outcomes, our findings suggest there was a tendency toward better coverage in the clinics that offered the rapid assessment of NVP adherence. Involving partners in HIV testing and improving women s understanding that antiretrovirals can reduce the risk of their baby becoming infected may also help to improve adherence among this group of women. 24 One limitation of our study was that we were not able to test all specimens for NVP. However, when we modeled the probability of samples being missing as a function of covariates that were associated with our outcome (NVP coverage), the analysis did not indicate that any systematic bias due to missing existed. A closer look at the group of HIV-infected women from whom we do not have HPLC results further support our reported outcome of 16% coverage associated with rapid HIV testing in the labor ward. During the follow-up phase of the study, a total of 100 HIV-infected women, 42

54 that were unaware of their HIV serostatus, were seen at the labor wards assigned to the intervention arm of the trial. Among the 71% (71 of 100) from whom NVP results were obtained, coverage was 16% (11 of 71, TABLE 3). Based on chart documentation of HIV testing and NVP administration, we estimate that coverage among the 29% (29 of 100) of women whose cord blood was not tested for NVP (and their infants) was 17% (5 of 29, data not shown). Had these specimens been tested for NVP, it is unlikely that our reported estimate of coverage, 16%, would have changed (11/71 + 5/29 = 16/100). Given the cluster-randomized design, this study is inherently subject to recruitment bias. It is possible that women attending labor wards assigned to the intervention clinics were systematically different from those attending the control clinics. In order to adjust for this potential imbalance, we stratified the labor wards, based on their population coverage level in 2003 and their size based on the number of deliveries they performed each month, prior to randomization. What is reassuring is that we did not find any statistical evidence that the proportion of women eligible for either of the labor ward interventions differed by study period (baseline or follow-up) or intervention group (control or intervention). However, we were not able to account for potential confounders such as staffing levels, staff morale and staff knowledge of HIV and experience level, all of which may have influenced our results. Inter-clinic contamination is also a reasonable threat with this study design. However, the study period was short which we believe minimized the likelihood of women, who were more apt to participate in the intervention, from migrating to the intervention sites. The short study period also reduces the chance of there being 43

55 significant changes in population coverage due to circumstances other than the intervention strategies. Our study has shown that labor ward-based PMTCT programs are feasible and can have significant, positive impact on population NVP coverage. Four out of six labor wards offering these services had substantial increases in coverage (18% to 33%). At the same time, two clinics experienced insignificant effects emphasizing the need to address staffing issues and minimize testing complexity prior to implementing the program. The labor ward provides an important opportunity to prevent infant HIV infections by offering testing and treatment to women who don t already know their HIV status while taking steps to ensure adherence among those that do. 44

56 Reference List (1) Doherty TM, McCoy D, Donohue S. Health system constraints to optimal coverage of the prevention of mother-to-child transmission programme in South Africa: lessons from the implementation of the national pilot programme. African Health Sciences 2005;5(3): (2) Manzi M, Zachariah R, Teck R et al. High acceptability of voluntary counselling and HIV-testing but unacceptable loss to follow up in a prevention of mother-tochild HIV transmission programme in rural Malawi: scaling-up requires a different way of acting. Trop Med Int Health 2005 December;10(12): (3) Stringer EM, Sinkala M, Stringer JS et al. Prevention of mother-to-child transmission of HIV in Africa: successes and challenges in scaling-up a nevirapine-based program in Lusaka, Zambia. AIDS 2003 June 13;17(9): (4) Temmerman M, Quaghebeur A, Mwanyumba F, Mandaliya K. Mother-to-child HIV transmission in resource poor settings: how to improve coverage? AIDS 2003 May 23;17(8): (5) Stringer JS, Sinkala M, Maclean CC et al. Effectiveness of a city-wide program to prevent mother-to-child HIV transmission in Lusaka, Zambia. AIDS 2005 August 12;19(12): (6) UNAIDS. Report on the Global AIDS Epidemic. Geneva; (7) Stringer JS, Sinkala M, Goldenberg R, Vermund S, Acosta E. Monitoring nevirapine-based programmes for prevention of mother-to-child transmission of HIV-1. Lancet 2003 August 23;362(9384):667. (8) Guay LA, Musoke P, Fleming T et al. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: HIVNET 012 randomised trial. Lancet 1999 September 4;354(9181): (9) Dailly E, Thomas L, Kergueris MF, Jolliet P, Bourin M. High-performance liquid chromatographic assay to determine the plasma levels of HIV-protease inhibitors (amprenavir, indinavir, nelfinavir, ritonavir and saquinavir) and the nonnucleoside reverse transcriptase inhibitor (nevirapine) after liquid-liquid extraction. J Chromatogr B Biomed Sci Appl 2001 July 15;758(2): (10) Stringer JS, Sinkala M, Goldenberg RL et al. Universal nevirapine upon presentation in labor to prevent mother-to-child HIV transmission in high prevalence settings. AIDS 2004 April 9;18(6): (11) Stringer JS, Sinkala M, Stout JP et al. Comparison of two strategies for administering nevirapine to prevent perinatal HIV transmission in high- 45

57 prevalence, resource-poor settings. J Acquir Immune Defic Syndr 2003 April 15;32(5): (12) Bulterys M, Jamieson DJ, O'Sullivan MJ et al. Rapid HIV-1 testing during labor: a multicenter study. JAMA 2004 July 14;292(2): (13) Homsy J, Kalamya JN, Obonyo J et al. Routine intrapartum HIV counseling and testing for prevention of mother-to-child transmission of HIV in a rural Ugandan hospital. J Acquir Immune Defic Syndr 2006 June;42(2): (14) van't Hoog AH, Mbori-Ngacha DA, Marum LH et al. Preventing mother-to-child transmission of HIV in Western Kenya: operational issues. J Acquir Immune Defic Syndr 2005 November 1;40(3): (15) Kowalczyk J, Jolly P, Karita E, Nibarere JA, Vyankandondera J, Salihu H. Voluntary counseling and testing for HIV among pregnant women presenting in labor in Kigali, Rwanda. J Acquir Immune Defic Syndr 2002 December 1;31(4): (16) Moodley D, Moodley J, Coovadia H et al. A multicenter randomized controlled trial of nevirapine versus a combination of zidovudine and lamivudine to reduce intrapartum and early postpartum mother-to-child transmission of human immunodeficiency virus type 1. J Infect Dis 2003 March 1;187(5): (17) Taha TE, Kumwenda NI, Hoover DR et al. Nevirapine and zidovudine at birth to reduce perinatal transmission of HIV in an African setting: a randomized controlled trial. JAMA 2004 July 14;292(2): (18) HIV testing among pregnant women--united States and Canada, MMWR Morb Mortal Wkly Rep 2002 November 15;51(45): (19) Branson BM, Handsfield HH, Lampe MA et al. Revised recommendations for HIV testing of adults, adolescents, and pregnant women in health-care settings. MMWR Recomm Rep 2006 September 22;55(RR-14):1-17. (20) UNAIDS/WHO. UNAIDS/WHO Policy Statement on HIV Testing (21) Russell T, Crawford M, Woodby L. Measurements for active cigarette smoke exposure in prevalence and cessation studies: why simply asking pregnant women isn't enough. Nicotine Tob Res 2004 April;6 Suppl 2:S141-S151. (22) Webb DA, Boyd NR, Messina D, Windsor RA. The discrepancy between selfreported smoking status and urine continine levels among women enrolled in prenatal care at four publicly funded clinical sites. J Public Health Manag Pract 2003 July;9(4): (23) Mullen PD, Carbonari JP, Tabak ER, Glenday MC. Improving disclosure of smoking by pregnant women. Am J Obstet Gynecol 1991 August;165(2):

58 (24) Kiarie JN, Kreiss JK, Richardson BA, John-Stewart GC. Compliance with antiretroviral regimens to prevent perinatal HIV-1 transmission in Kenya. AIDS 2003 January 3;17(1):

59 Figure 1. Trial Summary Assessed for eligibility (12 labor wards) Baseline cord blood surveillance conducted in all 12 labor wards prior to randomization and intervention Excluded: 0 labor wards Stratified randomization (12 labor wards)* 2 extra-large clinics, low coverage 2 large clinics, low coverage 4 medium-large clinics, high coverage 2 small clinics, low coverage 2 small clinics, high coverage Labor ward PMTCT Cluster Received allocated intervention: 6 labor wards Control Cluster Received allocated intervention: 6 labor wards Clusters Analyzed: 6 labor wards Participants: Total of 2435 deliveries analyzed, 2328 (96%) with cord blood Determine results 535 (23%) cord blood seropositive At time of presentation to the labor ward 393 women (16%) unaware of their HIV infection status 413 women (17%) with positive (n=407) or indeterminate (n=6) antenatal HIV test result Excluded from analysis: 194/858 (22.6%) HIV positive cord blood specimens were not tested for nevirapine 170/535 (31.8%) from intervention period 24/323 (7.4%) from baseline period Clusters Analyzed: 6 labor wards Participants: Total of 2106 deliveries analyzed, 2058 (98%) with cord blood Determine results 467 (23%) cord blood seropositive At time of presentation to the labor ward 444 women (21%) unaware of their HIV infection status 340 women (16%) with positive (n=328) or indeterminate (n=12) antenatal HIV test result Excluded from analysis: 145/760 (19.1%) HIV positive cord blood specimens were not tested for nevirapine 116/467 (24.8%) from intervention period 29/293 (9.9%) from baseline period *Stratification: clinic size based on the number of deliveries per month: small <120, medium , large , extra-large >500; 2003 coverage level: low, <32%, high 36%. 48

60 Table 1. Descriptive characteristics of women according to intervention and control arm assignment during baseline and follow-up surveillance periods Characteristics Baseline Surveillance Period (Pre-Intervention), n=3123 Control Cluster (n=1491) Labor Ward PMTCT Cluster (n=1632) Follow-up Surveillance Period (Post-Intervention), n=4541 Control Cluster (n=2106) Labor Ward PMTCT Cluster (n=2435) Median (IQR) Median (IQR) Median (IQR) Median (IQR) Age, y 23 (20-27) 23 (20-27) 23 (20-28) 23 (20-28) Gestational age, wk 38 (36-40) 38 (36-40) 38 (36-39) 38 (36-39) ANC visits or more Gravidity or more Shift of admission 07:30-13:00 13:00-18:00 18:00-07:30 Time between admission and delivery, hr 0:00-3:00 3:01-6:00 6:01-9:00 9:01-12:00 12:01 or more Place of first ANC visit Lusaka Outside Lusaka Unbooked Mother is on zidovudine Yes No n (%) n (%) n (%) n (%) 20 (1) 512 (35) 747 (50) 201 (14) 408 (27) 702 (47) 375 (25) 352 (23) 250 (17) 889 (60) 828 (56) 278 (19) 134 (9) 91 (6) 156 (10) 1374 (92) 95 (6) 22 (2) 25 (2) 1465 (98) 17 (1) 486 (30) 884 (54) 239 (15) 468 (29) 772 (47) 387 (24) 475 (29) 246 (15) 910 (56) 856 (53) 325 (20) 154 (9) 105 (6) 192 (12) 1512 (93) 103 (6) 17 (1) 5 (<1) 1625 (>99) 21 (1) 810 (38) 1019 (49) 248 (12) 579 (28) 974 (46) 543 (26) 490 (23) 383 (18) 1228 (59) 1113 (53) 430 (20) 211 (10) 1156 (6) 234 (11) 1988 (94) 98 (5) 20 (1) 33 (2) 2070 (98) 20 (1) 861 (36) 1271 (52) 273 (11) 693 (28) 1132 (47) 601 (25) 652 (27) 437 (18) 1339 (55) 1267 (52) 494 (20) 237 (10) 135 (6) 291 (12) 2295 (94) 119 (5) 21 (1) 13 (1) 2417 (99) 49

61 Table 1. Descriptive characteristics of women according to intervention and control arm assignment during baseline and follow-up surveillance periods continued Characteristics Day of delivery Weekday Weekend* HIV tested during ANC Yes No HIV serostatus (cord blood) Positive Negative Baseline Surveillance Period (Pre-Intervention), n=3123 Control Cluster (n=1491) Labor Ward PMTCT Cluster (n=1632) Follow-up Surveillance Period (Post-Intervention), n=4541 Control Cluster (n=2106) n (%) n (%) n (%) n (%) 1065 (72) 424 (28) 1197 (80) 294 (20) 1147 (71) 479 (29) 1339 (82) 293 (18) 1406 (67) 697 (33) 1662 (79) 444 (21) Labor Ward PMTCT Cluster (n=2435) 1625 (67) 810 (33) 2042 (84) 393 (16) 1164 (80) 293 (20) 1272 (80) 323 (20) 1591 (77) 467 (23) 1793 (77) 535 (23) IQR, interquartile range; ANC, antenatal care *Weekend defined as Saturday or Sunday 50

62 Table 2. Comparison of nevirapine coverage between control and intervention clinics during baseline surveillance and follow-up surveillance study periods Cluster Baseline Surveillance Period (Pre-intervention) n/total (%) NVP Coverage Follow-up Surveillance Period (Post- intervention) n/total (%) Difference in NVP coverage (postintervention minus preintervention) Control Clinics 1 140/264 (53%) 152/350 (43%) -10% 3 Chainama 1/3 (33%) 4/13 (31%) -2% Chelstone 25/49 (51%) 19/40 (48%) -3% George 19/48 (40%) 33/89 (37%) -3% Kalingalina 11/20 (55%) 11/20 (55%) 0% Kanyama 49/101 (49%) 64/157 (41%) -8% Mtendere 35/43 (81%) 21/31 (68%) -13% Intervention Clinics 2 126/299 (42%) 190/364 (52%) +10% 3 Bauleni 7/14 (50%) 11/16 (69%) +19% Chipata 22/82 (27%) 40/85 (47%) +20% Chawama 34/84 (40%) 31/105 (30%) -10% Ngombe 6/13 (46%) 19/24 (79%) +33% Chilenje 27/42 (64%) 33/49 (67%) +3% Matero Ref 30/63 (48%) 56/85 (66%) +18% 1 Labor Ward PMTCT Services were not offered in the control cluster during the baseline or follow-up surveillance periods. 2 Labor Ward PMTCT Services were offered in the intervention cluster only during the follow-up surveillance period 3 p-value= 0.03, odds ratio (OR) 1.91, 95% CI,

63 Figure 2. Cluster- and clinic-level effect of the labor ward PMTCT package on population nevirapine coverage during the baseline (pre-intervention) and follow-up (post-intervention) study periods Baseline (Pre-Intervention) Follow-up (Post-Intervention) Nevirapine Coverage (%) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Chainama Chelstone George Kalingalinga Kanyama Mtendere Total Control-Arm Labor Wards Baseline (Pre-Intervention) Follow-up (Post-Intervention) Nevirapine Coverage (%) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Bauleni Chilenje Chipata Chawama Ngombe Matero Ref Total Intervention-Arm Labor Wards 52

64 Table 3. Cluster-level effects of labor ward voluntary counseling and testing on population nevirapine coverage Cluster NVP coverage among HIV-infected women, and their exposed infants, who presented to labor wards unaware of their infection status 3 Baseline Surveillance Period (Pre- intervention) n/total (%) Follow-up Surveillance Period (Post- intervention) n/total (%) Difference in NVP coverage (post - intervention minus pre - intervention) Control 1 0/52 (0%) 0/99 (0%) 0% Intervention 2 0/71 (0%) 11/71 (16%) 4 +16% 1 Labor Ward PMTCT Services were not offered in the control cluster during the baseline or follow-up surveillance periods. 2 Labor Ward PMTCT Services were offered in the intervention cluster only during the follow-up surveillance period 3 Coverage results for 293 of 358 (82%) HIV-infected/exposed women/infant pairs unaware of their HIV status upon presentation to the labor ward. Coverage results not included for 65 women-infant pairs (cord blood was not tested for nevirapine in 64 and infant NVP dosing unknown in 1). In the control cluster, 27 had missing coverage results: 7 of 59 (12%) from baseline and 20 of 119 (17%) from follow-up. In the intervention cluster, 38 had missing coverage results: 9 of 80 (11%) from baseline and 29 of 100 (29%) from follow-up. 4 Estimated coverage among 29 HIV positive specimens not tested for NVP was 17% (5 of 29) based on chart documentation of NVP administration. 53

65 Table 4. Cluster and clinic level effect of the rapid assessment of nevirapine adherence on population nevirapine coverage Cluster NVP coverage among known HIV-infected women, and their exposed infants 4 Baseline Surveillance Period (Pre-intervention) n/total (%) Follow-up surveillance Period (Post- intervention) n/total (%) Difference in NVP coverage (postintervention minus preintervention) Control Clinics 1 140/189 (74%) 152/234 (65%) -9% 3 Chainama 1/1 (100%) 4/4 (100%) 0% Chelstone 25/32 (78%) 19/24 (79%) +1% George 19/39 (49%) 33/69 (48%) -1% Kalingalina 11/15 (73%) 11/17 (65%) -8% Kanyama 49/63 (78%) 64/96 (67%) -11% Mtendere 35/39 (90%) 21/24 (88%) -2% Intervention Clincs 2 126/200 (63%) 178/264 (67%) +4% 3 Bauleni 7/12 (58%) 10/12 (82%) +24% Chipata 22/41 (54%) 38/62 (61%) +7% Chawama 34/67 (51%) 30/73 (41%) -10% Ngombe 6/7 (86%) 18/19 (95%) +9% Chilenje 27/31 (87%) 29/33 (88%) +1% Matero Ref 30/42 (71%) 53/65 (82%) +11% 1 Labor Ward PMTCT Services were not offered in the control cluster during the baseline or follow-up surveillance periods. 2 Labor Ward PMTCT Services were offered in the intervention cluster only during the Follow-up surveillance period 3 p-value = Coverage results for 887 of 1117 HIV-infected/exposed women/infant pairs diagnosed with HIV (positive or indeterminate) during antenatal care. Coverage results not included for 230 women-infant pairs (cord blood was not tested for nevirapine in 229 and infant dosing unknown in 1). In the control cluster, 104 had missing coverage results: 19 of 208 (9%) from baseline and 85 of 319 (27%) from follow-up. In the intervention cluster, 126 had missing coverage results: 11 of 211 (5%) from baseline and 115 of 379 (30%) from follow-up. 54

66 PREDICTORS OF RAPID HIV TESTING ACCEPTANCE AND SUCCESSFUL NEVIRAPINE ADMINISTRATION IN ZAMBIAN LABOR WARDS KAREN M. MEGAZZINI, NAMWINGA CHINTU, STEN H. VERMUND, DAVID T. REDDEN, DANIEL W. KREBS, MAUREEN SIMWENDA, BUSHIMBWA TAMBATAMBA, MOSES SINKALA and JEFFREY S.A. STRINGER In preparation for JAIDS Formatted for dissertation 55

67 ABSTRACT Background: Provision of rapid HIV testing in labor wards may provide an opportunity to reach a substantial proportion of the susceptible women and infants currently not receiving adequate prophylaxis. Methods: One of the objectives of a cluster randomized clinical trial we conducted between 12 October 2005 and 15 January 2006 in Zambia was to determine predictors of rapid HIV testing acceptance in a labor ward environment. HIV counseling and point-ofcare rapid HIV testing were offered to women who presented to the labor wards unaware of their HIV status and in the first stage of labor. Nevirapine (NVP) was administered to women who tested positive and a similarly appearing calcium tablet was provided to women who tested negative but chose to learn their HIV results after delivery. Results: Among the 2435 women who presented to the six labor wards assigned to the intervention arm of the trial, 393 (16%) were unaware of their HIV status. Of these women, 278 (71%) met the eligibility criteria for labor ward counseling and testing, 217 of 278 (78%) were offered HIV counseling in the labor ward and 146 of 217 (67%) agreed to counseling. Testing was completed on 82 of the 146 women (56%) counseled. The HIV prevalence among women tested in the labor ward was 28%. In multivariable analysis, women less than 20 years of age were three times (AOR 3.5; 95% CI ) more likely to be tested in the labor ward than women 30 years and older. Most women tested received treatment at least one hour prior to delivery (85%, 60 of 71) and twothirds were treated two hours (66%, 47 of 71) prior to delivery. Cervical dilation of one 56

68 to three centimeters at the time of admission was strongly associated with ingestion of NVP at least one hour (AOR 11.5; 95% CI ) and two hours (AOR 25.2; 95% CI ) before delivery. Conclusion: One third of eligible women were tested for HIV in the labor ward using an opt-in HIV testing strategy. Acceptability may be further enhanced and time between drug ingestion and delivery optimized through an opt-out testing approach. Key Words: HIV, nevirapine, antiretroviral prophylaxis, mother-to-child transmission, labor ward, sub-saharan Africa, Zambia 57

69 INTRODUCTION The number of perinatally acquired HIV infections can be dramatically reduced if pregnant women learn their HIV status and infected women and their infants receive antiretroviral prophylaxis. HIV testing of pregnant women has traditionally been conducted in the antenatal care setting. This approach to testing is optimal because it allows for detection of HIV early in pregnancy when the most efficacious treatment regimens can be implemented. Nonetheless, many women present to labor wards unaware of their HIV status either because they did not receive antenatal care or did not receive HIV testing during antenatal care. 1-5 Providing HIV testing and antiretroviral prophylaxis in the labor ward can provide many of these women a final opportunity to learn their status and reduce the risk of their infant becoming infected. In Zambia, a district-wide program for the prevention of mother-to-child HIV transmission (PMTCT) has been operational in the capital city of Lusaka since This program is antenatal care-based and provides single dose nevirapine (NVP) to HIVinfected women and their exposed infants. In 2003, we conducted an evaluation of this program and found that only 30% of the HIV-infected/exposed women/infant pairs, who received antenatal and delivery care within the Lusaka public clinics had ingested NVP around the time of delivery. 3 The main contributing factor for prophylactic failure was that women had not learned their HIV status during antenatal care, because they were not offered or they declined voluntary HIV counseling and testing. These women could have potentially benefited from labor ward-based PMTCT services. With this background, we conducted a cluster-randomized clinical trial designed to determine whether provision labor ward-based PMTCT services, including HIV 58

70 counseling and rapid HIV testing, could improve NVP coverage of HIV-infected pregnant women and their infants. The results of the parent trial are being published separately. In this article, we report the finding of a sub-study designed to identify factors associated with rapid HIV testing acceptance and successful administration of NVP in a labor ward environment. The elapsed time between NVP ingestion and delivery was estimated to help determine how often the maternal dose was administered at least one hour or two hours prior to delivery. METHODS Background Between 12 October 2005 and 15 January 2006 we conducted the parent clinical trial. As part of the trial, the 12 pubic labor wards in Lusaka were stratified by clinic size and the 2003 estimate of population NVP coverage 3 and then randomly allocated to the intervention and control arms of the study. The six labor wards assigned to the intervention arm of the study offered HIV counseling and point-of-care rapid HIV testing to women who presented to the labor ward unaware of their HIV status. Nevirapine was provided to women who tested positive and to their infants. Women did not provide informed consent to participate in this study. Rather, labor ward-based PMTCT services were available for all eligible women presenting to the six labor wards assigned to the intervention arm of the trial between 3 December 2005 and 15 January The District labor ward midwifery staff was responsible for identifying eligible women and providing them with HIV counseling and testing services. As part of the parent trial, midwives completed a one-page data collection form on each woman delivering in the labor wards. The form contained non-identifying patient information 59

71 that included age, gravidity, number of antenatal care visits, whether or not the woman had been offered and received HIV counseling and testing during antenatal care, her HIV status, and if infected, whether or not she and her baby received NVP. A second onepage form was completed on women who were eligible for rapid HIV testing in labor. This form contained information specific to the labor ward testing services including whether or not the woman was offered voluntary counseling and testing (VCT), whether or not she accepted VCT, her test result, the length of time it took to complete the rapid test and administer NVP, and elapsed time between NVP ingestion and delivery. In order to protect the women s anonymity, we did not collect exact times of admission, HIV testing, NVP administration or delivery. Instead, the midwives indicated on the study form the nursing shift during which the woman was admitted (i.e. morning, afternoon, evening) and selected the length of time between admission and delivery, NVP administration and delivery, and other study activities from multiple choice lists provided on the data collection forms. Rapid HIV Testing Women unaware of their HIV status were offered HIV counseling and testing if they were in the first stage of labor and the midwife deemed there was sufficient time to complete counseling and testing before delivery. All women who chose to be tested provided written consent for HIV testing (which was standard procedure in Lusaka District at the time of the study). An HIV counseling flip chart was provided to help standardize the counseling message for each woman being counseled in labor. This flip chart was available in English, and the two most common languages spoken in Lusaka, Nyanja and Bemba. 60

72 In consideration for the labor ward setting in which this study was conducted, women who chose to be tested were given the option of learning their HIV test result either before or after delivery. Those who chose to postpone learning their result until after delivery could nonetheless consent to NVP for themselves and their babies at the time of testing by indicating their choice on the consent form. The consent form explained that they would be offered a tablet despite the results of their HIV test. Women who tested positive were given a NVP tablet (200mg) and their exposed infants received NVP syrup (2mg/kg). We offered a calcium tablet that resembled NVP to all women who tested negative for HIV so as not to inadvertently disclose a woman s status during labor or stigmatize HIV-infected women by only providing tablets to those that tested positive. We used a point-of-care rapid HIV testing strategy. Midwives were equipped with clip-on timers that allowed them to keep track of HIV tests that were running while they continued with their work. The standardized HIV testing algorithm used in Lusaka District antenatal care centers was employed. Women were screened for HIV using the DETERMINE HIV-1 / 2 rapid test (DETERMINE HIV-1 / 2, Abbott Laboratories, Abbott Park, Illinois, USA) and positive results were confirmed with the GENIE II HIV 1/2 rapid test (GENIE II HIV-1/2, BioRad Laboratories, Hercules, California, USA). NVP was administered to all women who tested positive on the DETERMINE test. Confirmatory HIV testing was done on those who tested positive using the GENIE II rapid test either before or after delivery as time permitted. Post-test HIV counseling was provided to all women prior to discharge from the center. Women were encouraged to learn their test results and receive post-test 61

73 counseling after delivery. Those who opted to learn their results prior to delivery could do so assuming there was sufficient time for post-test counseling: a decision left to the discretion of the midwife. Women were informed of their infection status during the posttest counseling session unless they declined to learn their status at that time. Statistical Analysis Generalized estimating equations (GEE) were used to identify factors associated with being offered VCT in the labor ward, HIV testing acceptance, and NVP ingestion at least 1 and 2 hours before delivery. GEE models were used in this analysis in order to account for the non-independence of individuals nested within each of the six labor wards. All GEE models used in this analysis accounted for delivery site. All variables significant in at the P<.10 level in the bivariate analyses were entered into the multivariable GEE models. All analyses were performed using SAS version 9.1 (SAS Institute, Cary, North Carolina, USA). The study was approved by the Institutional Review Boards at the University of Alabama at Birmingham (Birmingham, Alabama, USA) and the University of Zambia (Lusaka, Zambia). RESULTS During the 6-week trial period, 2435 women gave birth to infants in the six labor wards offering rapid HIV testing and NVP prophylaxis, 393 of the 2435 women (16%) were unaware of their HIV infection status. Among these women, 278 (71%) were in the first stage of labor and were therefore eligible for HIV counseling and testing in the labor ward, 217 (78%) of whom were offered VCT. The reasons eligible women were not 62

74 offered labor ward VCT included: advanced labor and insufficient time to complete counseling and testing before delivery (36 of 61, 59%); midwife felt the mother was in too much pain (17 of 61, 28%); mother was not identified by the midwife as eligible (6 of 61, 10%); midwife on duty was not a counselor or was too busy (3 of 61, 5%); and on one occasion the patient was deaf (1 of 61, 2%). Of the 217 women offered VCT, 146 of them (67%) agreed to pre-test counseling and 82 of 146 (56%) agreed to and were tested for HIV while in labor. The predominant reason women were not tested for HIV following pre-test counseling was that the mother refused testing (54 of 64, 84%). Only five midwives (5 of 64, 8%) reported not being able to perform an HIV test on a woman after counseling her due to advanced labor (n=4) or precipitous delivery (n=1). The HIV prevalence among women tested in the labor ward was 28% (FIGURE 1). Among women eligible for VCT in the labor ward, women offered VCT were similar to women not offered VCT in terms of age, gestational age, nursing shift at the time of admission, and day of delivery (weekday versus weekend) but they differed in terms of gravidity, cervical dilation at the time of admission, elapsed time between admission and delivery, and delivery site (TABLE 1). In multivariable analysis, the odds of being offered VCT were greater among women dilated 0 to 3cm (adjusted odds ratio [AOR] 85.3; 95% CI ) and those dilated 4 to 7cm (AOR 8.4; 95% CI ) than women dilated 8 to 9cm (TABLE 2). The odds of being offered VCT in the labor ward were also greater for primigravida women (AOR 1.9; 95% CI ) as compared to multiparous women (TABLE 2). Among the 217 women who were offered VCT in the labor ward, women who agreed to counseling (n=146) were similar to women who declined counseling (n=71) 63

75 with respect to all key demographic characteristics listed in TABLE 1 with the exception of shift of admission (data not shown). Women admitted to the labor ward during the evening shift were less likely to agree to VCT when offered compared to those admitted during the morning shift (odds ratio [OR] adjusted for clinic of delivery, 0.6; 95% CI, ). There was also a tendency for women admitted during the afternoon shift to decline VCT when offered (OR adjusted for clinic of delivery, 0.7; 95% CI ). Among the 146 women who received HIV counseling in the labor ward, women less than 20 year of age were three times more likely to agree to HIV testing in the labor ward than women 30 and older in multivariable analysis (TABLE 3, AOR 3.5; 95% CI ). HIV testing and ingestion of NVP or calcium prior to delivery Of the 82 women tested for HIV in the labor ward, 71 (87%) of them received either a NVP tablet or a calcium tablet prior to delivery. Among them, 21 (30%) were HIV-infected and received NVP and 50 (70%) were uninfected and received calcium. Eleven women did not receive a tablet prior to delivery (reasons shown in FIGURE 2). Tablets were administered to 85% (60 of 71) of women more than one hour prior to delivery and to 66% (47 of 71) of women more than two hours prior to delivery (FIGURE 2). Women who received a tablet more than 1 hour prior to delivery were similar to those that received a tablet less than an hour before delivery with respect to the following covariates: age, gestational age, number of antenatal care visits, gravidity, shift of admission, elapsed time between blood collection and tablet administration, and day of delivery (data not shown). However, women who received a tablet more than one hour 64

76 prior to delivery were more apt to be in the latent phase of labor (cervical dilation 1 cm to 3 cm) at the time of admission (AOR 11.5; 95% CI , TABLE 4) compared to women who delivered within an hour of tablet ingestion. Cervical dilation at 1 to 3 cm was also associated with tablet ingestion more than two hours prior to delivery (AOR 25.2; 95% CI , TABLE 5). In multivariable analysis, younger age (<20 and 20-24), having received two or fewer antenatal care visits, and admission during the afternoon shift were associated with reduced odds of ingesting NVP more than two hours prior to delivery. The length of time between blood collection and tablet administration was not associated with timing of NVP ingestion in multivariable analysis. In bivariate analysis, the length of elapsed time between admission and delivery was associated with being offered VCT in the labor ward (TABLE 2), acceptance of VCT (TABLE 3) and ingestion of NVP more than one hour (TABLE 4) and two hours prior to delivery (TABLE 5). However, due to the presence of moderate co-linearity between the cervical dilation and time between admission and delivery variables, we did not include time between admission and delivery in the multivariable models. Rather we chose to use cervical dilation because we believe it is the more clinically relevant variable. DISCUSSION We found that rapid HIV testing is feasible in busy Lusaka labor wards and that many Zambian women were receptive to opt-in labor ward-based VCT services. The District midwifery staff successfully identified 98% (387 of 393) of women who presented unaware of their HIV status and offered VCT to 71% (217 of 278) of eligible 65

77 women. Two-thirds of women agreed to be counseled and just over half of the women who received counseling were tested for HIV. The rate of testing acceptance among women counseled in the labor ward (56%) was considerably lower than the acceptance rates seen in public-sector antenatal care centers in Lusaka, 3 or that is characteristic of many other antenatal care-based PMTCT programs in sub-saharan Africa (95% - 99%). 4, 7, 8 Overall, 29% (82 of 278) of eligible women were tested for HIV in the labor ward. In order for labor ward-based HIV counseling and testing services to be optimally effective, women must: 1) have access to HIV testing services in the labor ward; 2) agree to be tested for HIV and; 3) if infected, must receive antiretroviral prophylaxis ideally at least one hour prior to delivery. In our study we identified considerable losses at the first two steps in this cascade. Among the 278 women who were eligible for VCT in the labor ward, 22% (61 of 278) were not offered VCT and 49% (135 of 278) did not agree to either HIV counseling or testing in the labor ward. Overall, 71% (196 of 278) of eligible women did not benefit from the labor ward-based VCT services available to them. At the time of this study, HIV testing in Lusaka public-sector clinics was performed using an opt-in testing strategy which involved lengthy pre-test counseling and required that women sign an informed consent form prior to testing. In the present study, we employed the same opt-in testing strategy with some modifications to facilitate its use in a labor ward setting, e.g. we provided flip charts to aid pre-test counseling and trained midwives to perform rapid HIV tests in the labor ward (i.e. point-of-care testing). Others have shown that HIV testing acceptance can be improved by using an opt-out testing strategy, 9, 10 which involves systematic HIV testing of all patients unless they decline 66

78 testing. We suspect that had opt-out testing strategy been in place in Lusaka public labor wards at this time of this study, considerably more women may have benefited from the HIV testing and treatment services offered. In this study, women who were tested for HIV in the labor ward were offered a tablet for ingestion despite their HIV status. This strategy allowed women to be tested during labor but postpone learning their test results until after delivery. It also was meant to avoid inadvertently stigmatizing HIV-infected women by only providing tablets to those who tested positive. Nonetheless, we found that most women (83%) received posttest counseling and learned their HIV status before delivery suggesting that it may not be necessary to wait until after delivery to inform Zambian women of their test results. Among those who were tested in the labor ward, 87% (71 of 82) ingested a tablet (either NVP or calcium) prior to delivery and 85% were treated at least 1 hour prior to delivery. The timing of tablet administration was similar among HIV-infected and -uninfected women. The predominant reason women did not receive a tablet was that they refused: no further explanation was indicated. There was only one occasion in which the woman delivered her baby before the tablet could be administered. This woman was not infected with HIV. Two factors have likely contributed to this high rate of treatment success, i.e. tablet administration at least one hour prior to delivery. First, most women who received HIV testing were in the early stages of labor. This provided a wide window of opportunity for midwives to complete testing and administer treatment before delivery was imminent. Second, this study utilized a point-of-care HIV testing strategy in which HIV testing was performed by midwives in the labor ward as opposed to sending the 67

79 specimens to the laboratory for testing. In the Mother-Infant Rapid Intervention At Delivery (MIRIAD) study, point-of-care testing was shown to significantly shorten testing time and improve the proportion of test results that were received prior to delivery over laboratory-based testing. 11 In our study, 75% (53 of 71) of the women tested for HIV received their tablet (NVP or calcium) within 45 minutes of the time their blood was drawn for HIV testing. We suspect that had opt-out HIV testing strategy been utilized, our treatment success rates may have been higher because opt-out testing would have theoretically shortened, or eliminated, the time needed for pre-test counseling resulting in women being tested and treated earlier in labor. The proportion of eligible women who were offered VCT varied significantly across the six labor wards ranging from 58% to 100%. In five out of the six labor wards, 73% to 100% of women were offered VCT. Throughout the study, it was apparent to the study team that the labor ward which ultimately offered VCT to the smallest proportion of eligible women (58%) had experienced the most difficulty with staffing. Although there were only two instances during the study when midwives reported not offering VCT to an eligible woman because there was not a midwife counselor on duty, both occurred at this site. Staffing shortages were also sited as major reasons for not offering counseling and testing in the MIRIAD study in the US and in a hospital-based 12, 13 intrapartum HIV counseling and testing program in Uganda. Labor ward testing will undoubtedly increase the workload for labor ward staff, particularly when point-of-care testing is implemented. There is a need to address staffing shortages prior to implementing labor ward-based HIV testing services. 68

80 We estimated the elapsed time between tablet administration (NVP to infected women or calcium to uninfected women) and delivery and found that 85% women tested during labor ingested the tablet more than one hour prior to delivery and 67% ingested NVP more than two hours prior to delivery. Although sub-therapeutic concentrations of NVP (<100 ng/ml) have been reported in infants born to women who received NVP less that two hours before birth, 14 maternal NVP dosing less than two hours before delivery was not associated with vertical HIV transmission in a phase III trial of infant NVP versus NVP plus zidovudine at birth. 15 Maternal dosing less than one hour before delivery was associated with vertical transmission in a study in Zambia. 16 We found that NVP was successfully administered to the majority of women according to either dosing standard (i.e., one or two hours before delivery). Cervical dilation at the time of admission was an important and practical predictor of treatment success. We found that women dilated 1 to 3 cm were significantly more likely to receive treatment more than one hour (and more than two hours) prior to delivery than those dilated 4 cm or more. Unlike the MIRIAD study, we did not find that admission to the labor ward during the evening shift was associated with lower testing acceptance but did find that admission during the afternoon shift was associated with lower chance of ingesting NVP at least two hours before delivery. This finding may be related to staff workloads during the afternoon shift. In Lusaka, women are routinely discharged from the postpartum ward at 16:00 daily. Because midwives are busy with discharges they may be delaying HIV testing and treatment of women admitted during that shift. 69

81 The success of the MIRIAD study has led the Centers for Disease Control and Prevention (CDC) to recommend that opt-out rapid HIV testing be routinely conducted 17, 18 among all women of unknown HIV status delivering at hospitals in the U.S. In less developed countries, labor ward testing is becoming increasingly more popular and is recommended by UNAIDS and WHO. 10 Labor ward-based programs in Brazil and Nigeria have reported acceptance rates of 98% and 99%, respectively, demonstrating that 19, 20 women in these countries are highly receptive to rapid HIV testing in the labor ward In Nairobi s Kenyatta Hospital, HIV testing uptake in the labor ward using an opt-in testing strategy (40%) was similar to the percent of eligible women tested in our study (29%). 2 Our study has a few limitations. Midwives were tasked with the responsibility of identifying women eligible for VCT in the labor ward and whether or not women were offered VCT was left of the discretion of the midwives. There was a tendency for midwives to offer VCT services to those in the earlier stages of labor and not offer services to those in more advanced labor. This may have inflated our reported acceptance rate and treatment success rate. Also, we estimated treatment success based on the timing of tablet administration: NVP to infected women and calcium to uninfected women. Given that only 21 women tested positive for HIV in the labor ward, we based our estimates of treatment success on the cohort of 71 women who received a tablet (NVP or calcium) in the labor ward. Although there was not a statistical difference in the timing of NVP or calcium administration, it is possible that there was less urgency among the midwives to administer calcium to an uninfected woman, particularly when most women (83%) learned their HIV status before delivery. This lack of urgency may have 70

82 resulted in fewer women receiving calcium more than one or two hours before delivery, biasing our results towards the null. This study has shown that rapid HIV testing is feasible in busy Lusaka labor wards and that NVP can be successfully administered to women who test positive. Onethird of eligible women were tested for HIV in the labor ward using an opt-in testing strategy and 85% of them received treatment more than one hour before delivery. Alternative HIV testing strategies, such as an opt-out approach, may enhance testing acceptance and should be considered when implementing labor ward HIV testing and treatment services. 71

83 Reference List (1) Kowalczyk J, Jolly P, Karita E, Nibarere JA, Vyankandondera J, Salihu H. Voluntary counseling and testing for HIV among pregnant women presenting in labor in Kigali, Rwanda. J Acquir Immune Defic Syndr 2002 December 1;31(4): (2) Ong'ech JO, Kiarie JN, Mbori-Ngacha D, Gachoki AW. Improving effectiveness of a PMCT program through a multi-pronged approach in a large public hospital in Nairobi, Kenya. XV International AIDS Conference, Bangkok, Thailand 2004 July 11, Abstract ThPeB7059. (3) Stringer JS, Sinkala M, Maclean CC et al. Effectiveness of a city-wide program to prevent mother-to-child HIV transmission in Lusaka, Zambia. AIDS 2005 August 12;19(12): (4) Temmerman M, Quaghebeur A, Mwanyumba F, Mandaliya K. Mother-to-child HIV transmission in resource poor settings: how to improve coverage? AIDS 2003 May 23;17(8): (5) van't Hoog AH, Mbori-Ngacha DA, Marum LH et al. Preventing mother-to-child transmission of HIV in Western Kenya: operational issues. J Acquir Immune Defic Syndr 2005 November 1;40(3): (6) Stringer EM, Sinkala M, Stringer JS et al. Prevention of mother-to-child transmission of HIV in Africa: successes and challenges in scaling-up a nevirapine-based program in Lusaka, Zambia. AIDS 2003 June 13;17(9): (7) Homsy J, Kalamya JN, Obonyo J et al. Routine intrapartum HIV counseling and testing for prevention of mother-to-child transmission of HIV in a rural Ugandan hospital. J Acquir Immune Defic Syndr 2006 June;42(2): (8) Manzi M, Zachariah R, Teck R et al. High acceptability of voluntary counselling and HIV-testing but unacceptable loss to follow up in a prevention of mother-tochild HIV transmission programme in rural Malawi: scaling-up requires a different way of acting. Trop Med Int Health 2005 December;10(12): (9) HIV testing among pregnant women--united States and Canada, MMWR Morb Mortal Wkly Rep 2002 November 15;51(45): (10) UNAIDS/WHO. UNAIDS/WHO Policy Statement on HIV Testing (11) Jamieson DJ, Cohen MH, Maupin R et al. Rapid human immunodeficiency virus- 1 testing on labor and delivery in 17 US hospitals: the MIRIAD experience. Am J Obstet Gynecol 2007 September;197(3 Suppl):S72-S82. (12) Bulterys M, Jamieson DJ, O'Sullivan MJ et al. Rapid HIV-1 testing during labor: a multicenter study. JAMA 2004 July 14;292(2):

84 (13) Homsy J, Kalamya JN, Obonyo J et al. Routine intrapartum HIV counseling and testing for prevention of mother-to-child transmission of HIV in a rural Ugandan hospital. J Acquir Immune Defic Syndr 2006 June;42(2): (14) Mirochnick M, Dorenbaum A, Blanchard S et al. Predose infant nevirapine concentration with the two-dose intrapartum neonatal nevirapine regimen: association with timing of maternal intrapartum nevirapine dose. J Acquir Immune Defic Syndr 2003 June 1;33(2): (15) Taha TE, Kumwenda NI, Hoover DR et al. Nevirapine and zidovudine at birth to reduce perinatal transmission of HIV in an African setting: a randomized controlled trial. JAMA 2004 July 14;292(2): (16) Stringer JS, Sinkala M, Chapman V et al. Timing of the maternal drug dose and risk of perinatal HIV transmission in the setting of intrapartum and neonatal single-dose nevirapine. AIDS 2003 July 25;17(11): (17) Rapid HIV antibody testing during labor and delivery for women of unknown HIV status: a practical guide and model protocol. [Atlanta, GA]: Centers for Disease Control and Prevention; 2004 Jan 30. (18) Advancing HIV prevention: new strategies for a changing epidemic--united States, MMWR Morb Mortal Wkly Rep 2003 April 18;52(15): (19) Sagay AS, Musa J, Adewole AS et al. Rapid HIV testing and counselling in labour in a northern Nigerian setting. Afr J Reprod Health 2006 April;10(1): (20) Nogueira SA, Lambert JS, Albuquerque AL et al. Assessment of a rapid HIV test strategy during labor: a pilot study from Rio de Janeiro, Brazil. J Hum Virol 2001 September;4(5):

85 Figure 1. Labor ward HIV counseling and testing profile 2435 women gave birth to infants during the 6- week period when labor ward VCT was offered 393 (16%) women presented to the labor ward in labor and unaware of their HIV status 278 (71%) were eligible for labor ward VCT 2042 aware of their HIV status: 1626 (80%) seronegative, 416 (20%) seropositive ineligible because they were in the second stage of labor (dilated 10cm) 2 7 Not identified by midwife as eligible (includes one deaf woman) 36 Advanced labor (midwife s opinion) / precipitous labor 17 Mother in too much pain 217 (78%) Women offered HIV counseling and testing 3 No trained counselor on duty / midwife too busy 71 Declined HIV counseling 146 (67%) Women agreed to HIV counseling 54 Mother refused testing 5 Labor advanced after counseling 82 (56%) Women tested for HIV 2 Mother was in too much pain to complete testing 23 (28%) Seropositive 59 (72%) Seronegative 1 Includes 5 women with indeterminate HIV status; 2 Includes 5 women who were 10 cm dilated upon admission to the labor ward but still offered counseling and testing 74

86 Table 1. Characteristics of women eligible for voluntary counseling and testing for HIV in the labor ward (n=278) Characteristics Age, y < > 30 Gestational Age, wk < ANC visits or more Gravidity* or more First ANC visit Lusaka Outside Lusaka or unbooked Cervical Dilation, cm* 0-3 (latent) 4-7 (active) 8-9 (active-advanced) Shift of admission Morning (07:30-13:00) Afternoon (13:00-18:00) Evening (18:00-07:30) Time between admission and delivery, hr* 0:00-3:00 3:01-6:00 6:01-9:00 9:01-12:00 12:01 or more Day of Delivery Weekday Weekend Offered VCT Not Offered VCT n=217 n=61 n (%) n (%) 35 (16) 69 (32) 55 (25) 58 (27) 98 (45) 118 (55) 82 (38) 100 (46) 35 (16) 64 (30) 99 (46) 51 (24) 177 (82) 40 (18) 130 (60) 69 (32) 17 (8) 59 (27) 38 (18) 120 (55) 77 (36) 59 (27) 27 (12) 18 (8) 36 (17) 145 (67) 72 (33) 7 (12) 16 (27) 23 (38) 14 (23) 32 (52) 29 (48) 30 (50) 24 (40) 6 (10) 13 (21) 26 (43) 22 (36) 48 (79) 13 (21) 4 (7) 15 (27) 36 (66) 15 (25) 15 (25) 31 (50) 42 (69) 8 (13) 4 (7) 2 (3) 5 (8) 42 (69) 19 (31) 75

87 Table 1. Characteristics of women eligible for voluntary counseling and testing for HIV in the labor ward (n=278) continued Characteristics Labor Ward 1 (Bauleni) 2 (Chawama) 3 (Chilenje) 4 (Chipata) 5 (Matero Ref) 6 (Ngombe) Offered VCT Not Offered VCT n=217 n=61 n (%) n (%) 11 (5) 36 (17) 40 (18) 58 (27) 47 (22) 25 (12) 2 (3) 26 (43) 8 (13) 21 (34) 4 (7) 0 (0) VCT, voluntary counseling and testing; ANC, antenatal care * p <.0001 for at least one category in the generalized estimating equations model accounting for clinic of delivery p <.01 for at least one category in unadjusted generalize estimating equations model 76

88 Table 2. Factors associated with being offered voluntary counseling and testing for HIV in the labor ward (n=278) Adjusted for Delivery Site Full Model* Characteristic No./Total (%) Estimate Odds Ratio (95% Cl) Estimate Intercept Gravidity 1 64/77 (83) ( ) /125 (79) ( ) or more 51/73 (70) 1.0 Cervical Dilation, cm 0-3 (latent) 4-7 (active) 8-9 (active-advanced) Time between admission and delivery, hr 0:00-3:00 3:01-6:00 6:01-9:00 9:01-12:00 12:01 or more 130/134 (97) 69/84 (82) 17/53 (32) 77/119 (65) 59/67 (88) 27/31 (87) 18/20 (90) 36/41 (88) ( ) 8.6 ( ) ( ) 3.5 ( ) 6.1 ( ) 4.7 ( ) Odds Ratio (95% Cl) 1.9 ( ) 1.4 ( ) ( ) 8.4 ( ) 1.0 CI, confidence interval *Generalized estimating equations model accounting for delivery site and statistically significant covariates in the bivariate model at p<.10 (gravidity and cervical dilation) was used to create the full model. Omitted from the full model due to the presence of moderate co-linearity with cervical dilation. 77

89 Table 3. Factors associated with HIV testing acceptance in the labor ward (n=146) Adjusted for Delivery Site Full Model* Characteristic No./Total (%) Estimate Odds Ratio (95% Cl) Intercept Age, y <20 18/21 (86) ( ) /47 (62) ( ) /39 (36) ( ) /39 (54) 1.0 Gravidity 1 2 or more ANC visits or more Cervical Dilation, cm 0-3 (latent) 4-7 (active) 8-9 (active-advanced) Time between admission and delivery, hr 0:00-3:00 3:01-6:00 6:01-9:00 9:01-12:00 12:01 or more 28/42 (67) 53/102 (52) 37/56 (66) 36/66 (55) 9/24 (38) 52/88 (59) 24/45 (56) 4/12 (33) 30/51 (59) 25/39 (64) 8/20 (40) 7/12 (58) 12/24 (50) ( ) ( ) 1.6 ( ) ( ) 2.1 ( ) ( ) 0.9 ( ) 0.9 ( ) 1.0 ( ) Estimate Odds Ratio (95% Cl) 3.5 ( ) 1.4 ( ) 1.0 ( ) ( ) ( ) 1.9 ( ) ( ) 2.0 ( ) 1.0 CI, confidence interval; ANC, antenatal care *Generalized estimating equations model accounting for delivery site and statistically significant covariates in the bivariate model at p<.10 (age, gravidity, number of ANC visits, and cervical dilation) was used to create the full model. Omitted from the full model due to the presence of moderate co-linearity with cervical dilation. 78

90 Figure 2. Labor ward HIV testing and tablet administration profile 82 women tested for HIV in the six intervention labor wards 23 (28%) Seropositive 59 (72%) Seronegative 1 refused 1 unavailable 5 refused 1 nurse forgot 1 precipitous delivery 2 reason unknown 21 (91%) Ingested NVP 50 (85%) Ingested calcium Time between tablet ingestion and delivery 2 (10%) 1 hr 9 (18%) 1 hr 5 (24%) >1 hr but <2 hr 8 (16%) >1 hr but <2 hr 14 (67%) >2 hr 33 (66%) >2 hr NVP, nevirapine 79

91 Table 4. Factors associated with ingestion of nevirapine or calcium at least one hour prior to delivery following rapid HIV testing in the labor ward (n=71) Adjusted for Delivery Site Full Model* Characteristic No./Total (%) Estimate Odds Ratio (95% CI) Estimate Odds Ratio (95% CI) Intercept First ANC visit in Lusaka 46/55 (84) Yes ( ) ( ) 14/16 (88) No Cervical Dilation, cm 1 to 3 4 to 8 Time between admission and delivery, hr 0:00 to 3:00 3:01 or more 42/44 (95) 18/27 (67) /24 (63) 45/47 (96) ( ) ( ) ( ) 1.0 CI, confidence interval; ANC, antenatal care *Generalized estimating equations model accounting for delivery site and variables significant at p<.10 in a bivariate model (place of first ANC visit and cervical dilation) were used to create the full model. Omitted from the full model due to the presence of moderate co-linearity with cervical dilation. 80

92 Table 5. Factors associated with ingestion of nevirapine or calcium at least two hours prior to delivery following rapid HIV testing in the labor ward (n=71) Adjusted for Delivery Site Full Model* Characteristic No./Total (%) Estimate Odds Ratio (95% CI) Estimate Intercept Age, y < 20 9/16 (56) ( ) /26 (65) ( ) /11 (73) ( ) > 30 13/18 (72) 1.0 ANC visits 0 to 2 3 or more First ANC visit in Lusaka Yes No Shift of admission Morning Afternoon Evening Cervical Dilation, cm 1 to 3 4 to 8 Time between blood collection and drug administration, min 45 > 45 17/31 (55) 30/40 (75) 36/55 (65) 11/16 (69) 17/24 (71) 7/13 (54) 23/34 (68) 37/44 (84) 10/27 (37) ( ) ( ) /50 (66) 14/18 (78) ( ) 0.8 ( ) ( ) 1.0 Odds Ratio (95% CI) 0.1 ( ) 0.2 ( ) 0.3 ( ) ( ) ( ) ( ) ( ) 1.0 ( ) ( ) ( ) 81

93 Table 5. Factors associated with ingestion of nevirapine or calcium at least two hours prior to delivery following rapid HIV testing in the labor ward (n=71) continued Adjusted for Delivery Site Full Model* Characteristic No./Total (%) Estimate Odds Ratio (95% CI) Estimate Intercept Day of Delivery Weekday 36/49 (73) 1.0 Weekend 11/22 (50) ( ) Time between admission and delivery, hr 0:00 to 3:00 3:01 or more 6/24 (25) 41/47 (87) ( ) Odds Ratio (95% CI) ( ) CI, confidence interval; ANC, antenatal care *Generalized estimating equations model accounting for delivery site and statistically significant covariates in the bivariate model at p<.10 (age, number of ANC visits, place of first ANC visit, shift of admission, cervical dilation, time between blood collection and drug administered, and day of delivery) were used to create the full model. Omitted from the full model due to the presence of moderate co-linearity with cervical dilation. 82

94 CONCLUSION The number of perinatally acquired HIV infections can be dramatically reduced if pregnant women learn their HIV status and infected women and their infants receive antiretroviral prophylaxis. While antenatal care centers offer an optimal setting for diagnosis and treatment of HIV among pregnant women, challenges remain in successfully identifying and treating all HIV-infected women and their HIV-exposed infants. The labor ward provides an important opportunity to improve antiretroviral coverage among HIV-infected/exposed women/infant pairs by providing HIV testing and treatment to women who are unaware of their HIV status at the time of admission and by implementing strategies to improve antiretroviral adherence among women diagnosed with HIV during antenatal care. We found that rapid HIV testing can be performed by midwives in the labor ward using a point-of-care approach and NVP can be successfully administered before delivery to the majority of women who test positive. Zambian women were receptive to rapid HIV testing in labor using an opt-in testing strategy but it is likely that acceptability could be markedly enhanced using an opt-out approach to 19, 20 testing. Ensuring adherence to NVP among women who learn they are infected with HIV during antenatal care can be challenging. Although there was tendency towards better coverage among women approached using a multiple choice style question rather than a 83

95 direct yes/no question, the present study was not powered to detect clinically significant improvements in coverage resulting from each of the individual labor ward interventions. Given the success of the opt-out HIV testing strategy compared to the traditional opt-in approach, future operational studies may consider looking at offering replacement doses of NVP to known HIV-infected women in a similar manner. Such an approach would require that midwives systematically identify all known HIV-infected women upon presentation to the labor ward and query them about NVP ingestion but while doing so they would have a replacement NVP tablet in hand for women to take if needed. This approach may make it easier for women to consent to NVP while still protecting their right to decline treatment. PMTCT program effectiveness is influenced by both regimen efficacy and population coverage. 4 In order to optimize their effectiveness, programs should strive to implement the most efficacious and safe prophylactic regimens while also working to enhance coverage. This will involve identifying the reasons for failed coverage within their programs and implementing strategies that target women-infant pairs currently not accessing or benefiting from the existing services. From a population perspective, both the efficacy of the regimen used as well as the coverage achieved with that regimen determines the bottom line; infant HIV infections prevented. Provision of PMTCT services in the labor ward, as an adjunct to antenatal carebased services, can provide an important opportunity to reach a substantial proportion of at-risk women/infant pairs and significantly improve programmatic antiretroviral coverage. While antenatal care-based PMTCT services are ideal in that they allow for detection of HIV early in pregnancy when the most efficacious treatment regimens can be 84

96 implemented, labor ward-based services can specifically target at-risk women and infants that may have otherwise slipped through the cracks. Labor ward services can effectively improve population NVP coverage ultimately saving the most infant lives. 85

97 GENERAL LIST OF REFERENCES (1) UNAIDS and WHO. AIDS epidemic update. Geneva; 2007 Dec. (2) Vermund SH, Wilson CM. Barriers to HIV testing--where next? Lancet 2002 October 19;360(9341): (3) De Cock KM, Fowler MG, Mercier E et al. Prevention of mother-to-child HIV transmission in resource-poor countries: translating research into policy and practice. JAMA 2000 March 1;283(9): (4) Mofenson LM, McIntyre JA. Advances and research directions in the prevention of mother-to-child HIV-1 transmission. Lancet 2000 June 24;355(9222): (5) Mofenson LM. Technical report: perinatal human immunodeficiency virus testing and prevention of transmission. Committee on Pediatric Aids. Pediatrics 2000 December;106(6):E88. (6) Mofenson LM. Advances in the prevention of vertical transmission of human immunodeficiency virus. Semin Pediatr Infect Dis 2003 October;14(4): (7) United Nations. The Declaration of Commitment on HIV/AIDS: five years later, Report of the Secretary General. Geneva; (8) World Health Organization. HIV and Infant Feeding Guidelines for Decision Makers. Geneva; (9) World Health Organization. Antiretroviral Drugs for Treating Pregnant Women and Preventing HIV Infection in Infants in Resource-Limited Settings. Geneva; (10) McIntyre J. Strategies to prevent mother-to-child transmission of HIV. Curr Opin Infect Dis 2006 February;19(1):33-8. (11) Guay LA, Musoke P, Fleming T et al. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: HIVNET 012 randomised trial. Lancet 1999 September 4;354(9181): (12) Jackson JB, Musoke P, Fleming T et al. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: 18-month follow-up of the HIVNET 012 randomised trial. Lancet 2003 September 13;362(9387): (13) Stringer JS, Sinkala M, Maclean CC et al. Effectiveness of a city-wide program to prevent mother-to-child HIV transmission in Lusaka, Zambia. AIDS 2005 August 12;19(12):

98 (14) Temmerman M, Quaghebeur A, Mwanyumba F, Mandaliya K. Mother-to-child HIV transmission in resource poor settings: how to improve coverage? AIDS 2003 May 23;17(8): (15) Doherty TM, McCoy D, Donohue S. Health system constraints to optimal coverage of the prevention of mother-to-child transmission programme in South Africa: lessons from the implementation of the national pilot programme. African Health Sciences 2005;5(3): (16) Manzi M, Zachariah R, Teck R et al. High acceptability of voluntary counselling and HIV-testing but unacceptable loss to follow up in a prevention of mother-tochild HIV transmission programme in rural Malawi: scaling-up requires a different way of acting. Trop Med Int Health 2005 December;10(12): (17) Stringer EM, Sinkala M, Stringer JS et al. Prevention of mother-to-child transmission of HIV in Africa: successes and challenges in scaling-up a nevirapine-based program in Lusaka, Zambia. AIDS 2003 June 13;17(9): (18) Bulterys M, Jamieson DJ, O'Sullivan MJ et al. Rapid HIV-1 testing during labor: a multicenter study. JAMA 2004 July 14;292(2): (19) HIV testing among pregnant women--united States and Canada, MMWR Morb Mortal Wkly Rep 2002 November 15;51(45): (20) Homsy J, Kalamya JN, Obonyo J et al. Routine intrapartum HIV counseling and testing for prevention of mother-to-child transmission of HIV in a rural Ugandan hospital. J Acquir Immune Defic Syndr 2006 June;42(2):

99 APPENDICES 88

100 APPENDIX A UNIVERSITY OF ALABAMA at BIRMINGHAM INSTITUTIONAL REVIEW BOARD APPROVAL 89

101

102 APPENDIX B UNIVERSITY OF ZAMBIA RESEARCH ETHICS COMMITTEE APPROVAL 91

103