Evaluation of Conditional Cash Transfers to Increase Retention in PMTCT Services in Akwa Ibom State, Nigeria

Background:

Nigeria, home to only 2% of the world's population, accounts for 30% of the world's mother-to-child transmissions of HIV. Over the past three years, and in collaboration with bilateral aid agencies, the Nigerian government has strengthened hundreds of free rural clinics, trained thousands of birth attendants, and begun free distribution of antiretroviral (ARV) drugs. Despite this, only 1.8% of low-income women deliver a baby with a skilled birth attendant, only 16% of HIV-positive pregnant women receive ARVs to prevent mother-to-child transmission (PMTCT) of HIV, and only 2.3% of infants born to HIV-infected women received ARV prophylaxis to reduce the risk of virus transmission. Pregnant women are lost at each step along the PMTCT cascade, from attending a prenatal visit to giving life-saving antiretroviral therapy to their newborns.

Operated by New Incentives, a non-profit organization focused on maternal and child health, the All Babies are Equal (ABAE) program was established in June 2014 in Akwa Ibom to improve utilization of health services for PMTCT. The ABAE program provides conditional cash transfers (CCTs) to pregnant, HIV-positive women conditional on obtaining specific ante-, peri-, and postnatal services at public facilities to prevent mother-to-child transmission of HIV. The CCT is primarily a demand-side intervention that aims to reduce the barriers that women may face when seeking care, as well as encourage uptake of and retention in PMTCT services.

Preliminary studies:

A large body of evidence indicates that providing ARV drugs to HIV-positive pregnant women during pregnancy and at the time of birth to both mother and baby can drastically reduce the transmission of the virus to the newborn. However, in Nigeria, even though drugs for PMTCT are free and available, many women do not use them. Additionally, knowledge of mother-to-child transmission of HIV and PMTCT is high among the investigators' target population. Recent research shows that 91% of women of reproductive age have high awareness of PMTCT in southern Nigeria, yet 71% have poor attitudes towards PMTCT treatment and services, resulting in low uptake of services.

To encourage uptake of PMTCT services, the ABAE's CCT program was designed based on research that has shown that CCT interventions have increased utilization of health services and improve child health. CCTs initially used in Latin America provided predictable sums of money to eligible mothers when certain actions were taken, particularly for health services and school enrollment. The cash payouts have been shown to alleviate short- and long-term poverty while building human capital among traditionally underserved populations (i.e. women and children in impoverished households). CCTs have also been effective for HIV prevention. For example, in Malawi, schoolgirls (aged 13 to 22) awarded between one and five dollars for attending school had significantly reduced prevalence of HIV and HSV-2 infection after 18 months. Additionally, several studies have shown that CCTs are an effective way to improve both access and adherence to antiretroviral therapy and to reduce barriers to maternal and newborn health service utilization. A CCT program targeted at increasing births in health facilities in India found a significant increase in uptake of antenatal and intrapartum care services and an increase in facility-based deliveries. As studies continue to evaluate the efficacy of CCTs, ongoing efforts aim to determine the impact of these interventions on transmission of HIV.

This study aims to formally evaluate the ABAE program for increasing retention of HIV-positive pregnant women in PMTCT through a randomized control trial. Through a combination of data extracted from the New Incentives program, facility records, and survey data, the researchers will also investigate the sources of poor attitudes towards PMTCT services, the influence of non-financial barriers on PMTCT retention, religious and cultural norms surrounding delivery outside of clinic settings, and how the CCT intervention should be structured to address the full cascade of PMTCT behaviors without compromising operational feasibility.

Main hypothesis to be tested:

Conditional cash transfers (CCTs) provided by the All Babies Are Equal (ABAE) program carried out by New Incentives, a non-governmental organization (NGO), will improve retention of HIV-positive, pregnant women in prevention of mother-to-child transmission (PMTCT) services in Akwa Ibom, Nigeria.

Intervention procedures:

Participant Screening:

An ABAE program officer will be stationed at each study site on its respective antenatal registration (i.e. "Booking day") day from 8am to 6pm. On Booking Day, all women who test positive for HIV will be recorded on a separate list by the HIV counselor or lab tester on duty in the privacy of his/her office. For each 1-3 women listed, the names will be transferred to the ABAE officer who will then transcribe the list in the order that they appear into a spreadsheet with pre-generated random assignment of women to treatment (i.e. offered to enroll ABAE program) and control (i.e. not offered the program) groups. Individuals randomized to the treatment arm will then be approached by the ABAE officer to screen and verify eligibility in person and against the subject's patient record card.

In-person visit to obtain informed consent:

• Potential subjects that have been randomized to the treatment arm will be directed to a room where the ABAE program officer will conduct the enrollment process. The ABAE officer will then describe the program to the woman, discuss confidentiality and privacy, and see if the woman is interested in participating. ABAE program officers will follow a standardized enrollment script per the ABAE's standard operating procedures for enrollment (see supplementary materials). The ABAE officers will explain both the ABAE program and the accompanying research study, following informed consent procedures, and obtain written consent for program and study participation. If the subject agrees to participate in the program, the ABAE program officer will follow a standardized intake protocol, including survey on background information, to enroll the woman into the program. She will also be given an informational brochure describing the program, including requirements to earn the cash rewards. This brochure is written in a simple language and includes visual graphics to facilitate easy understanding.
• Women receive a small cash payment and mobile phone credits when she completes enrollment to facilitate retrieval of cash from the bank and return phone call verification.

Conditional Cash Transfer 1:

- The participant enrolled in the ABAE program will call the New Incentives hotline the morning after her ANC clinic visit to receive her payment code. If successful, she will receive the first lump sum cash payment.

Conditional Cash Transfer 2:

• A New Incentives staff member will check the data sources containing information about medication pickups in the antenatal care ward in the morning of each participating clinic's Booking Day to check which participants have met the conditions for the second conditional cash transfer.
• During enrollment, participants are informed to call the New Incentives hotline the Friday after retrieving their medications for the second time to receive the payment code for the second lump sum payment.

Conditional Cash Transfer 3:

• A New Incentives staff member will call the participant four to six weeks before her expected due date to remind her of the third cash transfer. The participant will also be reminded about facility delivery by an automated robot call and several text messages.
• A New Incentives staff member will check the Labor Ward Registers in the morning of each participating clinic's Booking Day to check if study participants have met the delivery condition of the third conditional cash transfer.
• During enrollment, participants are informed to call the New Incentives hotline the Friday after birth to receive the payment code for the third lump sum payment.

Conditional Cash Transfer 4:

• A New Incentives staff member will call the participant four to five weeks after she delivers to remind her of the fourth cash transfer. Furthermore, the participant receives a text message regarding the early infant diagnosis test that needs to be taken six weeks after delivery.
• A New Incentives staff member will check the relevant clinic register in the morning of each participating clinic's Booking Day to check which participants have met the conditions of the third transfer.
• During a hotline call, participants are informed to call the New Incentives hotline the Friday after receiving the newborn's early infant diagnosis test to receive the payment code for the fourth lump sum payment.

Evaluation design:

The effect of the conditional cash transfer program on the primary and secondary outcomes will be assessed using a randomized control trial. A total of 550 pregnant women testing positive for HIV will be screened for eligibility and randomized to the 2 arms.

Data collection:

Prior to obtaining consent, the HIV status of all individuals will be collected (this is standard procedure at participating clinics for ANC enrollment). For women in the control arm, additional administrative clinic data, including first name, last name, patient number, pregnant mother number, date of birth, estimated delivery date, month of pregnancy, and phone number will be collected prior to obtaining informed consent to facilitate record tracking and later endline phone survey. Personal identifiers, however, will be deleted once data collection is complete.

There are 3 main sources of data collection:

1. Patient records will be extracted from hospital registers on a weekly basis. Extracted records will reflect information collated from different facility registers, including Patient Card, Laboratory Investigation Report Form, Adult Pharmacy Worksheet, ARV Counselor's worksheet, Daily HIV Test worksheet, Nurses' notebooks, lab testers' notebook, Delivery Register, Labor Ward Daily Report Book, Admissions Register, Child Follow-up Register, Dried Blood Spot Specimen Transport Form, and the Pediatric Pharmacy worksheet.
2. ABAE program data for women in the treatment arm participating in the conditional cash transfer program. These data include the information on program participants obtained through the program enrollment survey, interim telephone calls and questionnaires, and records of cash transfers.

3. An endline phone survey will be administered to all women enrolled in the study for which a valid phone number has been obtained (8-10 weeks after delivery).

   • At this point, the New Incentives employee will verify the subject's identification (i.e. recently was pregnant and registered for antenatal services at the specific participating clinic) and obtain verbal informed consent to participate in the endline survey.
   • This is the only contact that control subjects will have with the study personnel.
   • All individuals completing the phone survey will receive a small amount mobile phone credits as compensation for their time.
   • The endline survey will ask participants about women's birth history, decision-making over place of delivery, perceptions on program participation (if enrolled in the ABAE program), a discrete choice experiment to elicit state preferences over place of delivery (only for women randomized to the control arm), and sociodemographic background.

Data analysis:

1. Overall program effect:

   Our basic analysis will be to calculate point estimates (e.g., achievement of each cash transfer condition) with reasonable precision (i.e., the margin of error) for program participants and non-participants overall and for key sub-categories (e.g., at each facility) based on data abstracted from hospital records. We will first calculate univariate statistics, such as mean retention for each PMTCT condition and standard deviations, for program participants and non-participants, and means and standard deviations for key sub-categories. We will then test whether retention percentages are different between participants and non-participants with 2-tailed chi-squared tests of proportions with significance at α=0.05. We can test for independent effects, confounding, and interactions in multivariate analysis. Logistic regression analyses will be used to test for autocorrelation between subjects enrolled at the same facility, and hausman tests will be used to assess the relative fit of facility random and fixed effects. Possible sources of confounding include differences in HIV testing procedures and collection of patient records across facilities and over time (e.g. due to holidays, agricultural seasons). For example, logistic regression predicting the likelihood of delivering at the facility where ANC was sought can be used to simultaneously assess the independent effect of program participation while controlling for facility site, stage of pregnancy when enrolled, and time. Sensitivity analyses will be conducted to account for known sources of attrition (e.g. participant requesting to be dis-enrolled) and/or potential sources of spillover across study arms (e.g. women in the control group reporting that they had heard about the ABAE program).

2. Assess the differences in program attrition and self-reported attitudes and behaviors between control and treatment arms.

   With responses from the phone follow-up survey, which will be affected by unknown sources of attrition, the study of self-reported behaviors and perceptions is not designed to test one central hypothesis, but will involve exploratory statistical testing of differences in self-reported behaviors and perceptions among program participants vs. non-participants overall, and by key sub-categories (e.g. sociodemographic characteristics, location of residence). Standard multiple linear regression analysis will be used to assess individual determinants of outcomes and changes in outcomes for continuous outcomes. Logistic regression analysis will be used for dichotomous outcomes. For ordered categorical outcomes, count data, or potentially censored outcomes, we will first assess whether a binomial, poisson, or negative binomial distribution is appropriate by examining each outcome's dispersion. Depending on the specification tests in #1 above, additional adjustments of autocorrelation will be included. Wealth indices will be constructed using standardized methods following Filmer and Pritchett (2005).

3. Assess the existence of heterogeneous treatment effects. Multivariate regression also allows for testing interactions, such as whether program participation retention rates are different for different groups of patients (e.g., by stage of pregnancy, facility, a newly diagnosed HIV case, age).
4. Determine women's stated preferences for facility-based deliveries. We will complement our primary analysis with a discrete choice experiment (DCE). The DCE is designed to solicit respondents' preferences over several key childbirth or delivery characteristics. Each DCE question asks respondents to state a preference for hypothetical options A or B. First, for each question, we will use comparison of means (and Chi-squared tests) and regression analysis (and t-tests) to compare the proportion of respondents indicating A to the proportion indicating B. Second, using the same statistical tools, we will pool questions with overlapping characteristics and measure the change in stated choice associated with a particularly delivery characteristic.

Enrollment projection:

Sample sizes were calculated assuming a baseline delivery and drug pickup rate of 50%, which produced the largest possible variance. A conservative estimate of the effect of the intervention would be from a 50% baseline of delivery rates and drug collection to 70% after the intervention. Such results would have 99.9% power if each group consists of 250 people. Contrary to these conservative calculations assuming a 50% baseline, the baseline rate of delivery is approximately 30% at clinic sites and drug pickup rates are even lower, according to the clinic data. We therefore expect the study results to have 100% power. Power calculations were carried out using the DS Research online power calculator. An additional 10% upward adjustment is made to account for possible refusals to participate.

Study sites:

General Hospital, Ikot Ekpene, Akwa Ibom State, Nigeria General Hospital, Iquita Oron, Akwa Ibom State, Nigeria Immanuel General Hospital, Eket, Akwa Ibom State, Nigeria