April 2009–March 2010: Product development; partner identification; field test

Freedom from Hunger started with research to identify consumer education materials currently being used with HMI, and discovered that few tested and applicable materials existed. Using well-tested principles of adult learning, Freedom from Hunger designed the TLCs to be easily delivered by MFIs or NGOs providing financial services using the following six sessions:

• Costs and risks of illness
• How insurance works
• Advantages of health insurance
• Health insurance available to you
• Using your health insurance
• Deciding to enrol

In addition, a seventh session, “Re-enrolling in health insurance,” was developed to be tested for use at the re-enrolment period.

Work began to identify a suitable partner to field-test and revise the education materials. Despite best efforts, a partner could not be found in Latin America, the intended region. By broadening the scope of its search to India and Africa, Freedom from Hunger identified a suitable partner in Ghana—SAT, a well-established MFI with about 100,000 microfinance clients. Additionally, although Freedom from Hunger hoped to work with a private HMI product because it was thought to be more broadly applicable, it again struggled to find a suitable option. The HMI product that is available to the SAT clients is a public one, the Ghana NHIS, and the education focuses on the features of this scheme.

The field test was completed in early February 2010 in Tamale, a poor agricultural region in northern Ghana with low education and health indicators. Freedom from Hunger trained local SAT financial services staff to deliver the TLCs during the field test using role-plays and other interactive exercises, and observed ways to improve each session’s content and delivery. The module was tested over three days, with three groups of about 16 clients participating in two half-day sessions each. The field-test participants were women and were mostly illiterate.  The sessions were conducted outdoors in the usual meeting place for one of the groups. During the field test, Freedom from Hunger staff could closely observe and assess the appropriateness of both the content and the methods used to accomplish the learning objectives for each session. The test yielded important findings regarding several aspects of the NHIS product and the local context for SAT clients, and the most salient gaps in client knowledge about the insurance. These findings were used to modify and improve the content and visual materials in the final module for use in Ghana, and to inform the development of the adaptation guide.

During the field test, new information revealed opportunities for product testing that also expanded the scope of the planned demonstration and evaluation. First, the field-test team discovered that it was not uncommon for clients to have enrolled in the insurance at some point in the past but to have failed to pay their premium the following year, and some were unsure of their status. The problem with re-enrolment was also verified by the local NHIS offices and is a common issue with health insurance for the informal sector. As a result, Freedom from Hunger decided to add a follow-up session to be offered about one year after the initial education was delivered to support re-enrolment in the insurance. Secondly, the team learned that SAT could also provide education in single “one-off” sessions. Since other MFIs had identified challenges in delivering multiple short sessions, the idea to also test the impact of a single intensive training session emerged.

September 2010: Product refinement; preparation for demonstration; baseline study and education roll-out

Following the field test and further assessment of SAT’s capacity and interest in partnering for the broader demonstration, Freedom from Hunger and SAT agreed to proceed with demonstrating the education in the Northern District, to include branches in Tamale, Salaga, Walewale and Bole. Freedom from Hunger and IPA developed a research design based on a randomized control placement of the education treatments and applied for additional funding from two other funders to support more extensive research on the impact of the insurance on client health and financial status. The randomized control trial (RCT) was designed to look at the impact of several variations of the health insurance education on knowledge about health insurance, enrolment, renewal and the ability of clients to access health services.

The product variations that were tested included the following: 

• No education (control)
• Education in six 30-minute sessions, with and without a 30-minute refresher session prior to the annual renewal
• Education in one intensive 2.5-hour session, with and without a refresher prior to the annual renewal

During this time, several challenges arose. First, following the field test, one or more branch managers of SAT did not understand that they should refrain from promoting NHIS enrolment until the demonstration test, and instead proceeded to encourage or even require new loan clients to enrol in the NHIS prior to the delivery of the education. This created the risk of “spoiling” the baseline of active or current HMI enrolment of the target population, which was estimated to be about 10 to 20 per cent.

Secondly, in September 2010, it was learned that a previously announced plan by the current Ghanaian government to change the annual premium for the NHIS to a one-time lifetime premium seemed to be gaining momentum in the Ghanaian parliament, and there was considerable news media coverage. This caused Freedom from Hunger and its partners to carefully assess the potential impact of the change, and the probability of it occurring.

Thirdly, it became apparent that obtaining data on healthcare utilization and enrolment from the NHIS would be problematic due to concerns about data confidentiality, limited staff within NHIS to support such sharing, and overall poor data quality. This information, which was expected to influence the perception of value (and hence enrolment and renewal) by clients would need to be captured through a greater reliance on self-reporting by clients and, therefore, needed to be incorporated in the design of the research to optimize findings and lessons from the demonstration test.

In response, Freedom from Hunger and IPA decided to implement an individual census approach for the baseline survey, in which individuals were asked about their enrolment and, when possible, enrolment was verified with visual inspection of enrolment cards. (See the baseline survey and baseline survey biometrics for more information.) Because clients were sometimes not sure they were enrolled, questions were embedded in the survey that would enable further verification and crosscheck of enrolment status (e.g. asking how certain health services were paid for). Additional questions were also added to query clients about the use and types of services and methods of payment. This longer survey and more cumbersome process to obtain and verify enrolment data added time and cost to this part of the research.

During September and October 2010, the baseline survey was carried out by IPA field staff, trained and supported by IPA. A sample of five clients from each of 300 randomized credit groups was selected to participate in in-depth interviews lasting approximately one to two hours.

October 2010:    Education roll-out with enrolment; knowledge study and baseline survey; first take-up study

The roll-out of education to treatment groups comprising 60 per cent of 1,500 households from SAT credit groups started in October 2010 and was completed in March 2011, nearly two months later than expected. This was due to challenges with identifying and contacting active credit groups within the sample. Other challenges were also experienced along the way. Specifically, while it had been initially believed that credit groups met every two weeks, in fact some only met monthly, making it difficult to schedule the planned set of six biweekly education sessions. In some cases, the monthly meeting was in reality limited to a meeting between the group leader and the credit agent to hand over loan payments. Freedom from Hunger and IPA convened a high-level problem-solving meeting with the SAT director for microinsurance, the regional supervisor and branch managers, to assess and develop plans to complete all of the education for the treatment groups by the end of February.

In addition to the education, the plans included the administration of the post-education knowledge survey and field enrolment visits for NHIS agents.

Initially the goal was to complete education and promotion of the insurance by the end of November to line up with the post-harvest season and a time when household cash seemed more likely to be available. The delay and unevenness in the roll-out of the education described above, the longer time required to complete the baseline survey, and some delay for year-end holidays, pushed the education completion date back to March (the beginning of planting season), which may have had a negative impact on enrolment because of the local seasonality of income.

An Adaptation Guide, designed to enable other MFIs, NGOs and insurance providers to use the education in other settings, was completed in January 2011.

April 2011–October 2011: Analysis of baseline and knowledge studies; first uptake survey

Freedom from Hunger was successful in obtaining additional funding to support a short midline and comprehensive endline survey to assess the impact of the education on insurance uptake, use of health services, health spending, and the impact on the health and financial status of MFI clients.

In March 2011, data was collected to assess the impact of the education on knowledge of insurance with a knowledge survey quiz. An uptake study (which was much shorter than the baseline study) was conducted from July to August 2011. A Final Baseline Report and an Analysis of Knowledge Study were published in September 2011. At this stage findings related to the Knowledge Survey indicated that clients had a relatively high level of starting knowledge, awareness and positive attitudes about the insurance (at baseline). However, findings from the Knowledge Study administered by the SAT financial services officers following the education suggest a positive impact of education on insurance knowledge:

• Clients who received education scored significantly better on the questions that assessed insurance knowledge than those who did not receive education (control)
• Clients in the treatment groups had a significantly greater improvement in insurance knowledge than those in the control groups
• There was not a significant difference between the short session and consolidated session groups with respect to correct answers to knowledge questions
• There were no significant differences between those who received education and those who did not with respect to having generally favorable attitudes towards insurance

Findings from the baseline study related to insurance registration and enrolment into NHIS include the following:

• Demographics
  • At the time of the survey, 57.4 per cent of the clients in the sample were estimated to not be actively enrolled in NHIS, with 32.6 per cent enrolled; another 9.9 per cent may have been in the three-month waiting period
  • Individuals with any level of education are more likely to be registered than those with no education at all. Individuals in rural areas are less likely to be enrolled
  • Among clients who reported that they had not registered for insurance, clients in the treatment groups were significantly more likely to say that the reason was that they had not had time to do it (as opposed to reporting that cost or other barriers were reasons for not registering). The second most prevalent reason given for not enrolling was the premium cost
• Finances and shocks
  • Findings related to income and consumption suggest that premium payments are not a significant barrier to enrolment
  • Registration for insurance is correlated with a lower probability of a food-insecurity event, or removing a child from school for financial reasons. There was no corresponding correlation for either item with enrolment
• Health events
  • While registration and enrolment were not correlated with likelihood of getting treatment, they had a significant relationship to which type of treatment was sought—there was a greater likelihood of consulting a doctor and a decreased likelihood of consulting a chemical seller (drugstore)

Individuals who were enrolled in insurance were more likely to have attended a preventative care visit in the past month. In October, preliminary results from the first uptake survey were made available and a final report provided in March 2012. Overall, the entire sample showed a 30 per cent increase in enrolled clients from baseline to first uptake, and an 11 per cent increase in registration. Enrolment numbers reflect clients who are actively enrolled with up to date premium payments. Registration numbers reflect clients who have registered and who may or may not be up to date with their premium payments. Some surprising results were obtained, as can be seen in Table 3 below:

Table 3.  Results from uptake survey

For enrolled clients, the greatest increases were observed in the control and consolidated or long-session education groups, and the lowest increases were in the groups that received the short-session education. Similar results were seen in the registered clients. None of these differences were statistically significant.

Providing treatment to all groups initially assigned to receive treatment proved to be a challenge. In Bole, Salaga and Walewale, about 19 client groups were assigned to receive the consolidated sessions, and 19 client groups were assigned to short sessions; in Tamale, twice that number were assigned to each of the treatment groups. However, monitoring by IPA suggested that a number of these groups did not actually receive the education. Once the initial data from the uptake was analyzed and the lack of effect observed, IPA and Freedom from Hunger determined that to fully understand the results, better measures of which groups actually received treatment were required. IPA worked with SAT to interview the loan officers who implemented the education to capture information on which groups actually received the education. Agent recall was checked with additional quick interviews of clients and judged to be fairly reliable.

As expected, there were large differences in treatment rates across branches. In two branches, nearly all groups assigned for treatment received it. In one other branch, most of the short-session treatments were provided, but fewer of the consolidated-session treatments. And in the third branch, where more groups were assigned, treatment had the lowest implementation rate; only one short-session group received five or more of the six short sessions.

To further understand whether there were differences across groups assigned to treatment who either got no or only partial education, the data were then disaggregated to the level of the credit group and credit agent and analyzed to look for differences in enrolment and registration. However, the registration and enrolment variables again did not show significant differences between treatment and control groups, or for those individuals in the treatment group that SAT reported actually received treatment as compared to all individuals assigned to the treatment groups.

The final endline uptake survey was planned to take another look at the outcomes analyzed in this report, as well as a range of additional variables related to health and financial outcomes. In addition, the survey would include a qualitative component, aimed at helping understand why individuals do or do not sign up for insurance.

January 2012–April 2012

Freedom from Hunger provided a refresher training to SAT staff in January 2012 to prepare them to deliver follow-up reminder sessions to their clients. The reminder sessions were designed to be administered about a year after the initial education to review the information taught in the initial sessions and to remind clients that after a year they must re-enrol in insurance to maintain active enrolment status and be able to access covered health services. Half of the groups who received initial education were scheduled to receive the reminder sessions, enabling comparison of outcomes between groups that got initial education plus reminder sessions and groups that got initial education but no reminder sessions.

Meetings were also held with management to provide feedback about the lack of consistent and complete delivery of the initial education and to encourage greater monitoring. SAT was asked and agreed to conduct the reminder sessions during February, and monitoring reports and systems were put in place with incentives for prompt reporting of the status of education delivery and completion of the education to the assigned groups. However, by the end of February IPA and Freedom from Hunger became aware of a major staffing change in Tamale, the largest branch in the region. SAT management indicated that the entire branch staff (manager, credit agents and support staff) had been “transferred” and that another branch manager at one of our sample branches had been dismissed because of irregular activity. A new branch manager was assigned to Tamale, an individual who had participated in all of the Freedom from Hunger trainings for SAT and was knowledgeable about the importance of the study and the research design. These staffing and management changes delayed the completion of the delivery of the refresher sessions until early April.

May 2012–September 2012: Endline Survey and Qualitative Study

The endline survey and qualitative studies using focus groups were completed during May and June.

The midline and endline data on insurance enrolment found no significant differences in health insurance enrolment rates between the treatment groups and control group. (See Table 1 for final estimated enrolment numbers.) In addition, there were no significant differences in enrolment rates between those who received reminder sessions and those who did not. The results of the knowledge test administered after the initial treatment sessions suggest that the education did improve knowledge among those who received treatment; however, by the time of the endline (conducted over one year later) there were no significant differences in insurance knowledge across treatment and control groups.

The education may not have had a large impact because baseline knowledge of insurance was already high, suggesting that knowledge was not a barrier to enrolment. Rather, it appears that convenience of registration and a tendency to put off the decision to enrol despite understanding benefits of doing so, and the availability of cash at the time of making premium payments are more important barriers to enrolment and re-enrolment. In addition, the treatment was not implemented with all groups assigned to receive it, and enrolment rates were already relatively high in the sample (32.6% vs. about 20% in the local area) at the beginning of the study. These factors made it more difficult for the treatment to have an impact large enough to be statistically observable. The high enrolment, high knowledge about insurance, and incomplete implementation were unique factors that suggest this study may have limited external validity; in environments where knowledge and enrolment are low, educational programmes may have more impact.

While there were no significant differences in enrolment rates between the treatment and control groups, respondents’ enrolment increased notably over the course of the study, at a rate that appears to be much greater than the increase in enrolment in the general population in the areas where the study was conducted (a 50 per cent increase in the sample compared to 14 per cent increase locally). While activities of the NHIS aimed at increasing enrolment likely account for some of the increase, it is also possible that the repeated surveys, the checking of cards and expiration dates by surveyors to collect enrolment status, and the education sessions, might have together served as “touch points” that prompted clients to take action to register or enrol in insurance.

Having insurance was not associated with a higher likelihood of getting treatment in the case of health events, but those with insurance were more likely to go to doctors, while those without were more likely to go to less highly trained chemists. Those with insurance also incurred somewhat lower out-of-pocket costs for an incident of illness, and were more likely to have attended a preventative care visit. Although no causal link can be established, being registered in insurance is associated with a lower likelihood of food insecurity, a lower likelihood of removing children from school for financial reasons, and a lower likelihood of selling assets to get money.