Afya Credit Incentives for Improved Maternal and Child Health Outcomes in Kenya

1. BACKGROUND A "Continuum of Care" for reproductive, maternal, new-born and child health (RMNCH) includes integrated service delivery for mothers and children from pre-pregnancy to delivery, the immediate postnatal period, and childhood (1). Assuring continuity of care has become a key programme strategy for improving the health of mothers and new-borns, and is an important measure of service quality for outcomes such as prevention of mother-to-child transmission of HIV (PMTCT) (2). Yet in many countries there remains major gaps in seeking care along this continuum. Retaining women in this continuum even during pregnancy has not been possible to achieve in sub-Saharan Africa (SSA), with only 44% of women making the recommended 4 minimum ANC visits (focused ANC) (3). Focused ANC is a scheduled service, and when clients do not adhere to the recommended timing and repeat visits as required, the very purpose of the service is challenged. A number of treatments provided during ANC have to be delivered in sequence for them to be effective. Post-natal services are equally essential, as half of all postnatal maternal deaths occur during the first week after the baby is born, and one in four child deaths occur during the first month of life (5). In SSA, postnatal care programmes are among the weakest of all reproductive and child health programmes (6). Follow-up treatments and services such as PMTCT; counselling on family planning, breastfeeding and infant feeding practices; and childhood immunizations cannot be achieved if many women are lost to care immediately after delivery, as is the current trend. A major opportunity for altering the low care seeking behavior in SSA exists however, since over two thirds of women come into contact with the formal healthcare system at least once during pregnancy (3). A strategy that can retain these women in care, from their first contact during pregnancy through to birth and childhood period could thus alter considerably the negative trends in maternal and child health in Africa. Personal financial incentives (PFIs) can offer a broad based solution that is needed to tackle the challenge of low care seeking and retention of women in the care continuum. It can tackle both the financial barriers to care seeking, as well as the motivational barriers. A number of studies (30, 31, 32, 33) have demonstrated personal financial incentives (PFIs) can help to increase adherence to lengthy treatment schedules. Empirical evidence supports this rationale; although it is still an emerging field. In developed countries, paying people to change their behaviour has been demonstrated as effective with behavioural problems (10). Notable successes in developing countries are in TB treatment programmes (11) (12). In sub-Saharan Africa, this is still a very new concept.
2. METHODS

2.1 Study setting

The study will be conducted in Siaya County, which is located on the shores of Lake Victoria in Western Kenya. The County is made up of 6 sub counties which are mostly rural, and has a population of 8.3 million people (13).

2.2 Sample size determination

A sample size of 7200 women provides 80% power to detect a difference in the co-primary outcomes if the intra-cluster correlation coefficient (ICC) is either low (0.005) or moderate (0.025), taking the standard 5% significance level. The expected prevalence of these indicators in the control arm ranges between 30 and 80%. The average cluster size (48 clusters) is 150 participants.

2.3. Data collection methods

2.3.1 Abstraction of health records

The primary mode for data collection will be the health records of women held at the facilities they visit during the study period. With the consent of the study participants and approval of the ethics committee, health data that is relevant for answering both the primary and secondary research questions and adjusting for confounders will be abstracted on a monthly basis from participating facilities and on a quarterly basis from non-participating facilities.

2.3.3 Electronic capture of health visits

A secondary method for collecting data on the primary health outcomes is an electronic based system that will be utilized primarily for cash transfers to participants, but with additional capabilities of remotely tracking health visits made by the study participants.

The system is capable of capturing data on health visits remotely and in real-time each time a nurse touches the Afya Card on the NCF terminal reader as a means of validating a health visit. The card readers will be installed in all participating health facilities (24 intervention and 24 controls), as well as all hospitals in the County where most referrals take place.

2.3.4 Enrollment

During enrollment, after informed written consent has been obtained, the nurse at the enrolling facility will enter the details of the participant into the card reader.

2.3.5 Telephone Surveys

A telephone survey will be administered to all participants to generate data on outcomes that cannot be obtained for women who do not make health facility visits.

The second part of the survey will assess health care seeking between the time of enrollment and the time of the interview and will be used to assess quality of care.

. 2.3.6 Data collection in sub-group of participants

Additional telephone surveys

For a random sub-sample of participants (50%), there will be two additional phone surveys to gather more data on the primary and secondary outcomes, as well as establish baseline and end line data needed for cost-effectiveness analyses.

2 weeks after enrollment

The first contact between the research team and a sub-group of the enrolled women will take place two weeks after enrollment, which is the first follow-up period. A short questionnaire will be administered to assess cost of health facility visit.

12 months after the Actual Delivery Date

There will be a final exit survey with the same sub-group of participants to follow-up on health visits and health status of the particpants, and to gather feedback on the experience with the cash transfer project.

High risk groups

High risk groups in this study include mothers with infections that can be transmitted from mother to child, or that could cause severe complications in the mother that makes it impossible to infer any positive benefit of our intervention on their health.

In addition to the methods already outlined for tracking health records of participants, health visits of study participants who have permanently shifted to health facilities outside the evaluation zone will also be tracked. This will be done through extra phone calls to the participants (outside the regular phone surveys) to elicit information on health facilities the study participants are currently receiving care in, and visits made to those facilities to obtain records on a quarterly basis.

2.3.7 Economic evaluation

Cost effectiveness analysis

Economics data will be collected that serves two purposes: a) Enable for an adjustment of service provision when evaluating the impact of the intervention on the outcomes, and b) For cost-effectiveness analyses.

Data will also be collected on health worker's motivation (22) in both arms of the trial.

1. ETHICAL CONSIDERATIONS

The ethical approval for this study will be obtained from the Maseno University Ethics Review Committee (MUERC) and UCL Ethics committee, before data collection commences. The MUERC approved the pilot phase of the trial in which the aims of this main trial were clearly outlined.

3.1 DATA MANAGEMENT AND ANALYSIS PLAN

3.1.1 Blinding

Due to the nature of the intervention, the participants will not be blinded to their study allocation. The research team will however generally be blinded to the intervention allocation status of participants.

The primary outcomes data will be gathered using an electronic system designed by a third party and the research team will remain completely blind to the intervention allocation status of participants and their health visit outcomes until the end of the trial.

The system will randomly select participants in the intervention and control groups for phone interviews. The exception are specific cash transfer questions and usages. Data on the cash transfer would be separated from the rest of the data, and only merged after the rest of the data has been analyzed.

Secondary data will be abstracted by a project implementing partner. The abstracted records will not identify intervention allocation status of the woman. The cards will be mixed so that they do not follow any sequence, to make it impossible to link the ID to an enrolling facility. The trial statisticians will conduct interim and final analyses blind to allocation.

3.1.2 Interim analysis and monitoring by DMC

The investigators do not expect any adverse effects of the intervention but plan to carry out two interim analyses in 2018 and 2019, which will be reported to an independent Data Management Committee (DMC) to be convened according to the DAMOCLES charter (27). For these meetings, the DMC will be provided with a report on the two key safety measures for the trial, presented separately by trial arm, which are infant and maternal mortality. The DMC will decide at each meeting whether to request further analyses, which may include analysis of the study primary outcomes, and will ultimately on each occasion recommend that the study continues or stops.

3.1.3 Final analysis

Final analyses will be by intention-to-treat and include all mothers who were pregnant during the recruitment period (January 2017 - September 2017) and their children, regardless of whether they received the intervention or not. The investigators will test for differences between intervention and control areas in the primary outcomes using logistic regression for the binary outcome, and ordinal logistic regression for the ordinal outcomes, adjusting for clustering using random effect models.

The investigators will also carry out sub-group analyses to examine whether the effect of the intervention differs by wealth/multi-dimensional poverty quintile to understand the equity impact of the intervention.

The final analysis will be presented according to the CONSORT requirements for cRCTs (28, 29). Both the interim and final analyses will be conducted by the trial statisticians.

A detailed statistical analysis plan will be prepared before analysis begins.

3.4 Process evaluation

In order to enable replication and scale up of the intervention, a mixed method process evaluation will be conducted to document the design, implementation and mechanisms of the intervention. Five main sources of data will be used: (a) project planning and monitoring documents; (b) forms to capture attendance, problems and strategies identified by the women and health facility staff; (c) phone surveys data; e) facility costs data and (e) field notes from intervention team members to be reviewed through monthly debriefings will document intervention events.

1. CHALLENGES AND LIMITATIONS

4.1 Data and measurements related challenges

• Certain health outcomes will be difficult to assess in the project; such as prematurity, since the last monthly period (LMP) used to estimate the expected delivery date (EDD) is recall based and may not be accurate. To mitigate this, telephone surveys will be used to obtain the actual birth date, and subsequent follow-ups will be based on actual delivery date.
• There could be infections such as Malaria that occurred before the intervention, but which have important consequences on the secondary outcomes of the study. It will not be possible to make a retrospective identification of these type of health conditions if they have not been captured in health records.
• It will not possible to get detailed data on birth health outcomes for women that deliver at home, who are more likely to be in the control than intervention groups. The investigators will therefore utilize the phone surveys as a secondary method of gathering data on these outcomes.