Study protocol: Incentives for increased access to comprehensive family planning for urban youth using a benefits card in Uganda. A quasi-experimental study

Background: The use of contraception is one of the most cost-effective public health interventions and has the potential to prevent about 30% of maternal and 10% of child deaths in developing countries. Voucher-based initiatives for family planning are an effective and viable means of increasing contraceptive use. In this paper, we present a protocol for a pilot study of a novel incentive, a family planning benefits card (FPBC) program to increase uptake of family planning services among urban poor youth in Uganda while leveraging private sector funding. Methods: The study employs both impact and health economic evaluation methods to assess the effect of the FPBC program. We propose a quasi-experimental study design with two separate pre- and post-samples to measure program effectiveness. The main outcome of the impact evaluation is the percentage change in the prevalence of modern contraceptive use and unmet need for contraception. We will also conduct model-based incremental cost-effectiveness and budget impact analyses. The main outcomes of the economic evaluation are the cost per enrolled youth and cost per pregnancy averted, and cost per disability-adjusted life-year (DALY) averted. We will also pilot a corporate social responsibility model of sponsorship for the FPBC program in partnership with local corporations. Budget impact analysis will examine the potential affordability of scaling up the FPBC program and the fiscal implications of this scale up to the corporate social responsibility (CSR) budgets of partner corporations, the government, and the individual taxpayer. Discussion: In this study, we propose an impact and economic evaluation to establish the proof concept of using a FPBC program to increase uptake of family planning services among urban poor youth in Uganda. The results of this study will present stakeholders in Uganda and internationally with a potentially viable option for corporate-sponsored access to family planning in urban poor communities. Trial registration: MUREC1/7 No. 10/05-17. Registered 19th July 2017.

This is a one-year pilot project implementation and evaluation study. We are proposing to design a novel incentive benefits card based system that will be used to increase uptake of family planning services. We will partner with a local insurance firm to design and manage the FPBC system. The research team will manage the evaluation of the effectiveness of the program and conduct other stakeholder engagements. In partnership with the local insurance firm and its partner clinics and pharmacies in the intervention area, the FPBC program will provide family planning services to the youth aged 18 to 30 years. We will recruit and train community health workers (CHWs) on the counseling and guidance necessary to recruit and access the family planning services. The CHWs will mobilize, sensitize, and recruit individuals in the community. Each recruited participant will be provided with a benefits card and a list of hospitals and pharmacies where the card can be used. The benefits card will contain the beneficiary’s photograph, names, and a unique identifier. The card will allow beneficiaries access family planning counseling and guidance, long-acting reversible contraception such as implants or intrauterine devices, hormonal contraception such as contraceptive pills, injections and vaginal rings and barrier methods such as condoms and diaphragms. The FPBC program will also provide free testing and care needed to inform the choice of family planning method. These tests will include among others pregnancy counseling and testing, referrals to primary health care providers for health conditions such as high blood pressure or diabetes, HIV counseling and testing, and screening and treatment of sexually transmitted infections. We will exclude permanent contraception, such as vasectomy and tubal ligation from our card benefits since our target group is expected, for the most part, to be interested in delaying as opposed to discontinuing fertility. The study is being implemented in the city slums located in the suburbs of Kampala, Uganda’s capital city. The two city slums have been purposively identified for the pilot. The study sites are Kifumbira slum as an intervention area and Kalerwe slum as a comparator. Kifumbira slum is located northeast of Kampala city in Kampala Central division and neighbors Mulago to the west, Bukoto to the east, Kyebando to the north and Kamwokya to the south. Kalerwe slum is located north of Kampala city in Kawempe division. The two slums have populations that share the socio-demographic characteristics and hence provide a good intervention-comparison pair for the study. The majority of the two slum’s residents fall in the lowest wealth quintile and record the high levels of unmet need for contraception — 42% as compared to the national average of 34% [5]. Studies also suggest that only about 24% of the slum population is engaged in regular jobs, an indication that majority of the residents are unable to afford basic medical care [39]. The proposed study is a pilot project implementation, and prospective impact and economic evaluation. The project includes three main complimentary sub-studies: 1) an impact evaluation, 2) an acceptability and usability study, and 3) model-based health economics evaluation. The impact evaluation will utilize a pre- and post- quasi-experimental study design with a control group. We propose a household quantitative survey both in the intervention and control sites to assess the study outcomes at 1 and 6 months respectively. The impact evaluation study will have two primary outcomes: percentage change in the prevalence of modern contraceptive use and the percentage change in the prevalence of unmet need for contraception by study group. The secondary outcomes will include the percentage change in the knowledge of different methods of family planning, and satisfaction levels with the services provided among the study groups. The household survey shall only be conducted among females that are aged between 18 and 30 years, residents of the study areas, and willing to provide informed consent to participate in the survey (Fig. ​(Fig.11). Study organogram. Quasi-experimental study design with a comparator, and a pre- and post –FPBC program separate sample surveys We performed a mapping exercise at the start of the project aimed at delineating the study area boundaries and generating a list of households with members meeting the study’s inclusion criteria. We also identified the clinics and pharmacies in the Kifumbira slum that provide family planning services. A two-stage sampling technique will be used to select the study samples. The first stage involves purposively selecting the two study sites for the evaluation that are similar by both population characteristics and availability of family planning providers. The next stage will be randomly selecting the households that shall work as primary sampling units. From the compiled household sampling frame, a computer-generated random sample will be chosen. Sampled households will then be approached in their homes and invited to participate in the study. In every sampled household, all eligible participants for the survey will be invited to participate in the study. The survey tool will be administered with the help research assistants in a private setting. The same sampling approach will be reciprocated at end-line study leading to two separate samples per study site. According to the 2011 Uganda Demographic Health Survey (UDHS), the prevalence of modern contraception is 26% and the prevalence of unmet need for contraception among women in the lowest wealth quintile is 42%. The study’s aim is to cause an increase in the prevalence of modern contraception use by about 10 percentage points by the end of the program. The study utilizes the sample size formula for behavioral studies [40] stated below. The prevalence of modern contraception has been used to estimate the sample size as described below. Where: n =sample size. d = design effect. p 1 = the prevalence of modern contraception. p 2 = projected prevalence at the end of the study, so that (p 2 − p 1) is the magnitude of change you want to be able to detect. p = (p 1 +p 2)/2. Z 1 − α= the z-score corresponding to desired level of significance. Z 1 − β = the z-score corresponding to the desired level of power. The Table 1 below shows the estimated sample sizes for each group with varying percentage points of the magnitude of change to be detected, different power sizes, and two-sided level of significance of 0.05. A sample size of 669 participants per group will be used for the baseline and end-line surveys in both the control and the study groups, in order to obtain 80% power to detect a 10% change in the prevalence of modern contraception at a two-sided 5% level of significance. The study’s total sample size in both the groups will be 1338 participants. Sample size estimation The survey tool will contain sections on socio-economic and demographic characteristics (such as age, education, household income and expenditure, employment, marital status, religion etc.), reproduction (gravidity, parity, age at first sex, number of under-five children etc.), access to family planning information (sources of information e.g. radio, television, health workers etc.), history of family planning, current use of modern contraception, desire to stop or delay pregnancy (unmet need), reasons for choosing family planning methods, perceived barriers to family planning access and satisfaction levels with provided services and the FPBC program. Questions on socio-economic and demographic characteristics, reproduction, current use and unmet need for modern contraception will be modified from the 2011 UDHS questionnaires. We will carry out data collection for the house survey at both baseline and end-line using the personal digital assistant (PDA) data collection systems. We propose to use open data toolkit (ODK) for questionnaire design and data collection, and a dedicated server to receive data in real time. The electronic system will have data checks and necessary skips in order to improve data quality. The data from the server will be downloaded in excel format and sent to respective software for further analysis. The analysis will be done using R studio (open source version 1.0.136, R Foundation for Statistical Computing, Vienna, Austria) and STATA version 13.0 (Stata Corporation, College Station, Texas, USA). To assess if the control group is as similar as possible to the intervention group in terms of pre-intervention characteristics, we will compare the observed demographic and socioeconomic characteristics by group. Group differences for each variable will be summarized using means and proportions. In the event that the observed baseline characteristics show no similarity, we will use propensity score matching to identify the region of common support and discard observations that are substantially dissimilar in the analysis. The primary analysis of this evaluation is an intention to treat analysis (ITA). The magnitude of change in proportion between the two groups provides a potentially unbiased estimate of the average treatment effect (ATE) of the FPBC program. We will use a two-sample proportion z-test to test the null hypothesis that the two sample proportions are significantly different at six months between the intervention and control groups at 5% level of significance. We will also carry out a difference in difference (DID) analysis to identify the magnitude of change as a result of the program if any. We shall conduct a number of secondary analyses. We propose to assess the effect of the intervention on the participant’s change in family planning knowledge and satisfaction with the services using two-sample proportion z tests. Comparisons of proportions of changes in knowledge of different methods of family planning and satisfaction levels by intervention group will be conducted. Binary logistic regression analysis using modern contraception use as the outcome variable and demographic characteristics, satisfaction, and knowledge as explanatory variables will be conducted. Both adjusted and crude odds ratios with their corresponding confidence intervals and p-values will be reported. All results will be considered statistically significant at a 5% level of significance. Upon finishing the baseline survey, we will embark on the process of mobilizing study participants in the Kifumbira slum to enroll for the FPBC program. The acceptability study will utilize a mixed (quantitative and qualitative) methods study. The quantitative part will use data from the recruitment forms (acceptability forms) and the family planning medical records form (usability form), and the qualitative part will use data from the key informant interviews (KIIs) and Focus group discussions (FGDs). This study will approach study participants from their homes and invite them to enroll into the FPBC program. Recruitment will be done by moving from house to house and screening youth for eligibility and willingness to participate into the program. This will be done with the help of CHWs and the study’s trained research assistants. All persons eligible for the study will first be taken through a counseling and guidance session about family planning before enrolling into the program. Only participants; 1) aged between 18 to 30 years, 2) self-reported non-use of contraception methods, 3) for females, not currently pregnant, and 4) willing to provide consent will be considered eligible to receive the FPBC will be selected to participate into the program. We propose to recruit a purposive sample of 300 participants for the family planning benefits card for this pilot. A total of 250 persons will be females and 50 will be males of the same age brackets. This sample is based on resource availability for the study. The inclusion of men will help the study team to test a way of attracting males to embrace contraceptives use. Persons meeting the stated inclusion criteria and accepting to enroll for the FPBC program will be requested to provide passport photograph and their biodata. All recruited members shall then be provided with the FPBC that shall contain their full names and a photograph, and a list of clinics and pharmacies where the FPBC can be used. They will then be guided on how to use this card to visit the selected licensed hospitals and pharmacies to access free family planning services. The FPBC will have unique numbers that shall be used to track the use of the card at the clinics and pharmacies. An updated list of the beneficiaries shall also be provided to the service providers for cross-validation to avoid misuse of the card by the non-beneficiaries. Trained community health workers will continuously visit the enrolled participants and provide them with counseling and guidance. Acceptability will be quantitatively measured by estimating the proportion of eligible participants who accept to enroll into the program. Persons who refuse to join the FPBC program will be probed further to identify reasons for refusal. Quantitative categories for reasons for refusal will be created for analysis purposes. Usability of the FPBC will be measured by the number of times the card beneficiaries visited the service providers to seek care. The data will be disaggregated by the type of method and sex. This data will be generated from the health utilization reports generated from the medical records. Data for the acceptability will be collected with the help of trained research assistants exactly after the baseline. Approached participants will be requested to complete a study recruitment form. This form shall also be completed by those declining to join the program to help the study understand reasons for declining to join the program. The study will also collect usability data using the family planning medical records form that shall be completed at the clinics and pharmacies. Focus group discussions and key informant interviews will also be carried to provide a detailed understanding of the program achievements and challenges. Trained qualitative research assistants shall be used to collect qualitative data. All enrolled participants will complete recruitment form. The recruitment form will capture participant’s demographics and an acceptability question. For those who will accept to enroll into the program, more personal details such as names, phone contacts, and passport photograph will be collected. In partnership with a local insurance firm, we propose to design the FPBC medical records form (FMRF) which will be given to the partner clinics and pharmacies to collect data on clients such as date, benefits card number, purpose of the visit, services received, new or continuing clients, nature of contraceptive supplies given, need for referral and any notes taken on the client. Each provider will be trained on the implementation of this routine medical data collection form and other project activities. These forms shall be completed in triplicate. A copy shall be retained by clinics and pharmacies, and other copies given to the local insurance firm provider and research team. This will make monitoring the card usage by the research team credible. Data from the FMRF will be used to assess the usability of the card services. We will at the end of the study match the usability and acceptability data sets for analysis of the acceptability and usability of the FPBC program. We propose to conduct 15 key informant interviews (KIIs) with the community health workers and health facility workers implementing the program. The KIIs will be conducted at the end of project implementation to identify the implementation challenges, and document successes realized during the study period. The respondents will also be asked to make recommendations to help the team make improvements on the program. We also propose to conduct 6 focus group discussions (FGDs) of 8-10 participants per group in the intervention group at the end of the program. The FGDs will be disaggregated by sex, age, and acceptance of participation in the FPBC program. The FGDs will inform knowledge of youth in slum communities about family planning, their attitudes, practices, and barriers towards family planning, drivers of youth choices towards the use of certain family planning methods, health workers’ perceptions about the program, including its benefits, challenges experienced, and suggestions improvement of the program. Persons that declined to participate in the program will be probed further to identify reasons for refusal. All the data collected from the recruitment forms and FMRF will be coded and double entered in Epidata version 3.1. Consistency checks and cross-validation will be done at data entry to remove discrepancies in the final datasets. The analysis will be done using R studio (open source version 1.0.136, R Foundation for Statistical Computing, Vienna, Austria) and STATA version 13.0 (Stata Corporation, College Station, Texas, USA) . We will perform descriptive analyses of the demographic characteristics of participants included in the study using means and proportions. Univariate and bivariate analyses will be performed to further characterize the study sample. We will report means and proportions of the different variables such as the mean number of times of the card usage per method, proportions of different contraception methods undergone and will be disaggregated by sex and age group. We will explore the association between different variables and acceptability of FPBC program, usability of the card services and the reasons for declining screening using the chi-square test. Binary logistic regression will also be conducted with acceptability (accepted vs. refused) as the outcome variable. We will also perform poison regression analysis with number of times card used as the outcome variable against the participant’s socio-economic demographic characteristics. Qualitative interviews will be audio recorded and notes will be taken during the interviews. The recordings will be transcribed verbatim and translated into English if the recording is in the local language. Analysis will be done using Atlas Ti software. Different themes will be identified and logged into the analytical frame work and will be used to develop code definitions, code categories and themes. Transcripts will be entered into Atlas Ti scientific software as primary documents and free codes generated based on the codebook while allowing for open coding where new codes emerge. Magnitudes of responses as well as patterns in the data using code frequencies in the codes-primary documents table will be identified. Key themes identified in the discussions will be reported. We propose to conduct a model-based incremental cost-effectiveness analysis (CEA) and budget impact analysis (BIA). We will develop a model of the cost-effectiveness of the FPBC program using Microsoft Excel. The main outcomes of the economic evaluation are the cost per enrolled youth, cost per pregnancy averted and cost per DALY averted. We also propose to perform a static aggregate budget impact analysis (BIA) to examine the potential economic value of scaling up the FPBC program and the fiscal implications of this scale up to the CSR budgets of partner corporations, to the government, and to the individual tax payer of implementing this FPBC program. We will perform a primary micro-costing of FPBC program from the payer’s perspective i.e. include all direct medical and all direct non-medical costs and the modified societal perspective i.e. add indirect costs of lost productivity [41]. The direct medical costs will include all resources used in providing family planning services in terms of the costs of personnel, medical devices, and supplies. The direct non-medical costs will include transport and any upkeep visiting the family planning clinics and pharmacies and opportunity costs. Data on direct non-medical costs will be collected through follow-up studies. Data on personnel resource use will be collected from the benefits card claims records. All hospital and pharmacy visits will be timed to estimate participant time use as well as potential lost work time. Data on wages will be used to estimate indirect costs. A time-motion and costs survey will be carried out among program beneficiaries to estimate the direct non-medical costs. The model of cost-effectiveness will be validated by varying inputs to check if the expected values are logical. We will also set costs and outcomes to 0 separately and check if identical expected values are obtained for the comparators. The baseline analysis will consist of the calculation of the costs, the proportion of participants on family planning by contraception method, and the incremental cost-effectiveness ratios—cost per youth enrolled, and cost per method. Sensitivity analyses will be performed to investigate the impact of parameter uncertainty on payoffs. Empirically derived confidence intervals will be used when available. One-way sensitivity analyses will be performed on all parameter inputs to generate tornado diagrams representing the most influential parameters. Distributions will be assigned to all input parameters and probabilistic sensitivity analysis will be performed using a second-order Monte Carlo simulation. This will enable the generation of a cost-effectiveness acceptability curve to represent the impact of parameter uncertainty on estimates of cost-effectiveness. We will also estimate the aggregate national budget impact as well as the cost per Ugandan taxpayer per year of implementing a possible FPBC program. The BIA will involve the following steps: 1) characterization of the population of Uganda to estimate the potential number of people to be impacted by a possible FPBC program, 2) selection of a time horizon between 1 and 3 years, 3) documenting the current state of family planning nationally, 4) estimating the cost of FPBC program at scale, and 5) presenting budget impacts tailored to different audiences. The main analysis will be the generation of an estimate of the potential annual fiscal implication for the CSR budgets of potential corporate sponsors. The aim of the corporation-sponsored sustainability drive is to pilot test a way for funding a possible FPBC program beyond the immediate pilot program phase. Our vision is that corporate entities, through their CSR budgets, will be willing to fund an effective and cost-effective FPBC program. The first three aims of this study will provide effectiveness and economic efficiency data and the fourth aim will pilot test our idea of corporate partnership. The study will identify corporations with CSR budgets and approach them for potential partnership. The study’s media and publicity team will work with our corporate partners to design a media and community publicity plan to enable their good work to find an audience. We will design different metrics to reward and recognize different corporations depending on the expenditure on the FPBC program. All metrics for recognition will be adjusted based on corporations published book value or market capitalization to incentivize a wide range of small, medium, and large corporations. The study has so far recruited and trained 10 community health workers (CHWs) in the study areas to continuously sensitize the youth on family planning services and the associated benefits. Mapping of the study area to delineate boundaries and compile household lists has been completed. A total of six health facilities have also been enlisted for partnership to provide family planning services. The local medical insurance firm (benefits provider) has also been identified and partnerships discussions finalized.

N/A

• Dima Health