from EVIDENCE to POLICY Learning what works for better programs and policies April 2021 ARMENIA: Increasing preventive screening for non-communicable diseases in Armenia More people around the world are dying from noncommunicable Many countries, such as Armenia, have made efforts in recent diseases than ever before. These diseases, which include cancer, years to tackle non-communicable diseases by launching mass me- chronic respiratory diseases, diabetes, and heart disease, prema- dia campaigns and by equipping medical providers to detect and turely kill more than 15 million people between ages 30 and 69 treat these diseases. Despite these efforts, most people still aren’t each year. Many of these health conditions also make individuals getting tested. Policymakers are therefore looking for cost-effective more susceptible to severe forms of other diseases like COVID-19. approaches to get people to go to the doctor and get screened, The largest disease burden of non-communicable diseases is in and they are teaming up with behavioral scientists to answer key questions, such as: Are people more compelled to get tested if they know how many of their peers have done so? Do they respond to a HEALTH personal invitation? What about a small financial incentive? The World Bank’s Strategic Impact Evaluation Fund support- ed a randomized controlled trial to shed light on these questions. Researchers tested the impact of four approaches: 1) a personal invitation for patients to come in for screening, 2) a personal in- vitation that also conveyed statistics on how many of the patient’s peers have been screened, 3) a personal invitation with a pharmacy voucher labeled as an encouragement to get screened, 4) a per- sonal invitation and a pharmacy voucher that could only be used after the patient went for screening. After five months, people in the control group had very low screening rates: a mere 3.5 per- cent of people got screened for diabetes and hypertension. The personal invitation increased this rate to about 18.5 percent, with no additional impact from either the unconditional voucher or from the statistics about peers’ screening. The pharmacy voucher that was conditional on screening, however, was the most effec- low- and middle-income countries, where 85 percent of related tive, nearly doubling the percentage of people who got screened to deaths now occur, putting an extra strain on governments’ health 34.7 percent. Since it was more expensive to implement, however, budgets–– and families–– due to medical expenditures, productiv- the conditional voucher and the personal invitation alone were ity losses, disability, and deaths. While early screening can lead to equally cost-effective. life-saving treatment, screening rates tend to be low and discovery Overall, the findings suggest that very simple personalized of these diseases thus often occurs too late for effective and ef- invitations and conditional financial incentives can lead to more ficient treatment. life-saving health screenings in Armenia. This policy note is based on, “Invitations, Incentives, and Conditions : A Randomized Evaluation of Demand-Side Interventions for Health Screenings in Armenia,” World Bank Policy Research Working Paper (2020) Context Non-communicable diseases account for 93 percent of deaths their blood glucose levels measured in the past 12 months. in Armenia, with heart disease the most prevalent of the causes. This study targeted adults between the ages of 35 and 68 in Since 2010, the government has introduced incentives that reward Armenia who had not been screened in the preceding year. The providers for better performance, including increasing screening baseline data suggests this population was not economically se- rates for hypertension and diabetes. The detection of these two cure: half of participants responded that their income was suf- conditions is essential to initiating treatment, achieving disease ficient for basic family needs, such as food, clothing, and utilities, control, and delaying the onset of other non-communicable dis- but not enough for big purchases like a car, while 35 percent re- eases. Screening also provides a means for identifying the diseases sponded that their income is sufficient for everyday food but not in people who otherwise appear healthy. Clinical guidelines for for clothes and other basic needs. More than half of those in the these conditions have been developed in Armenia and medical study were unemployed. personnel have been trained to conduct screening tests. The gov- ernment has also launched a national mass media campaign to encourage adults to screen for hypertension and diabetes free of Did you know…. charge in their local primary health care clinic, which continued Early detection of diabetes and hypertension delays the onset of other through the study period. non-communicable diseases, but…. Despite all these efforts, the rates of early detection of hyper- • Only half of the 1.4 billion people living with diabetes have tension and diabetes have remained very low. In a nationally rep- been diagnosed resentative survey, 1 out of 3 people above the age of 15 had un- • 84 percent of undiagnosed cases are in low- and middle- income countries diagnosed hypertension and less than half in the same age bracket • Only 39 percent of people with hypertension in these contexts had had their blood pressure measured by a health provider. Simi- have been diagnosed • Among them, only 1 in 10 get control of their blood pressure larly, only 24 percent of individuals aged 15 years and above had Evaluation Researchers designed a randomized controlled trial to measure the were contacted to determine eligibility and to obtain informed impact and the cost-effectiveness of four approaches designed to consent. A large proportion were not available because they were increase the number of people who got screened for hypertension either out of reach, had the wrong contact information, or could and diabetes. The approaches were carefully co-designed with se- not be located (28 percent); or they were temporarily residing nior policymakers in the Armenian Ministry of Health. Research- elsewhere (27 percent). ers also convened focus groups with service users, health care pro- As a result, a total of 2,047 individuals that had been ran- viders, and policymakers in the health sector and used feedback domly assigned to the different treatment arms were eligible to from those discussions to inform the interventions’ designs. participate in the study. To assess the willingness of these indi- To identify potential study participants, the research team viduals to participate in the study, fieldworkers contacted them to used patient records from an e-health database from public health schedule face-to-face visits until they reached a target sample size facilities, where 100 percent of the rural population and 85 per- of 400 individuals per study group. cent of the urban population is registered for care. From an initial Study participants received either 1) a personal invitation list of 6,934 individuals, researchers randomized people to five from a physician, 2) a personal invitation that included informa- groups. After the randomization, potential study participants tion about screening rates of peers in participant’s age group, 3) a personal invitation with a pharmacy voucher worth 5,000 Arme- graphic survey with the study participants between July and Sep- nian drams (about US$10), described as an encouragement to get tember 2019. screened, and 4) a personal invitation with the pharmacy voucher redeemable only after screening (conditional incentive) or 5) no additional intervention at the time of study (the control group). The personal invitations were delivered in person by the study’s fieldworkers and consisted of a verbal message and printed let- ters signed by a physician, inviting the person to be screened for diabetes and hypertension. The invitations also highlighted that the screening visit was important for the individual’s health and the fact that it was free, and included details on how and where to go for screening. Five months after rolling out the programs, the research team measured whether patients came into a clinic to get screened for diabetes and hypertension, using administrative data from an e- health database. They also conducted a short social and demo- Findings The personal invitation led to a statistically significant The personal invitation was most effective when ac- and economically meaningful increase in screening. companied by a pharmacy voucher that could only be redeemed after screening. Without any intervention, only 3.5 percent of participants of the control group went for both screenings during the study pe- In the fourth group, which received the personal invitation riod. Patients in the treatment groups, all of whom received the with a conditional pharmacy voucher, 34.3 percent of partici- personal invitation (either alone, with information about peers, pants obtained both screenings – nearly 10 times as many as or with the pharmacy vouchers), screening rates increased by those who were screened in the control group. about 15 percentage points (or more than 400 percent). While combining the conditional voucher with the per- The peer information and the unconditional pharmacy sonal invitation was by far the most effective, it was voucher did not augment the impacts of the personal also more expensive to implement. invitation. Generally, the costs incurred for the interventions including The impacts were about the same in the first three treatment a voucher were the highest. For the unconditional vouchers, groups, regardless of whether people received the personal invi- this was because the vouchers were given to all, irrespective of tation alone or received it with a voucher or peer information. screening status. For the conditional vouchers, this was because Specifically, the rates for receiving both screenings were 18.5 fieldworkers verified which participants in the group had satis- percent in group 1 (personal invitation), 18 percent in group fied the screening condition and would then receive a voucher, 2 (personal invitation with peer group information), and 17.7 which required more visits and communication with partici- percent of participants in group 3 (personal invitation with an pants. unconditional pharmacy voucher). Nevertheless, the personal invitation alone and the The statistics about screening among peers may have personal invitation with the conditional voucher were failed because screening rates were so low and thus equally cost-effective. The unconditional pharmacy didn’t effectively “nudge” people to follow their voucher was the least cost-effective. peers, or it may be because of the type of statistics that were shared. The personal invitation improved screening at a cost of $62.20 per additional person screened for both diabetes and hyperten- The researchers used absolute numbers instead of percentages sion. When the invitation was accompanied by a voucher that because those percentages were relatively low, and they were not could only be redeemed after screening, each additional person sure whether communicating low rates of screenings would be screened for both tests cost $64.20. Because it was equally effec- perceived as encouraging. Other studies have conveyed social tive as the personal invitation alone but more expensive because norms using percentages. it included the cost of the voucher, the unconditional voucher For a peer-effect intervention to work, it’s possible that rates was about twice as expensive per additional person screened need to be relatively high to persuade individuals to follow what ($127.2). their peers are doing. Conclusion Overall, this research finds that conditional incentives and per- if screening is the main objective. Similarly, showing people sonalized invitations can substantially increase screening for statistics on peer screening (at least as designed here) didn’t diabetes and hypertension for those who haven’t been recently have any additional impact on top of sending out a personal screened. Adding a conditional incentive to the personal invita- invitation. tion doubled its effectiveness. The two approaches were equally When considering scalability, the personal invitations cost-effective, however. It’s likely these interventions would would be relatively straightforward to scale-up, and indeed the also be effective in other settings where screening rates are low Ministry of Health in Armenia will send these invitations to and where people haven’t responded to the usual mass com- every household in major cities starting in early 2022. Scaling munication campaigns to go for preventive health screenings. up the highly effective conditional incentive, while more ex- Importantly, the study also revealed that unconditional incen- pensive, may well be worth the cost given the enormous finan- tives – in this case, through a pharmacy voucher labeled as cial and health burden of non-communicable diseases. Further an encouragement for screening – may not be worth the cost research may be needed to evaluate these interventions at scale. The Strategic Impact Evaluation Fund, part of the World Bank Group, supports and disseminates research evaluating the impact of development projects to help allevi- ate poverty. The goal is to collect and build empirical evidence that can help governments and development organizations design and implement the most appropriate and effective policies for better educational, health, and job opportunities for people in low and middle income countries. For more information about who we are and what we do, go to: http://www.worldbank.org/sief. The Evidence to Policy note series is produced by SIEF with generous support from the British government’s Foreign, Commonwealth and Devel- opment Office and the London-based Children’s Investment Fund Foundation (CIFF). THE WORLD BANK, STRATEGIC IMPACT EVALUATION FUND 1818 H STREET, NW, WASHINGTON, DC 20433