Gong et al. BMC Health Services Research (2020) 20:1110 https://doi.org/10.1186/s12913-020-05967-z RESEARCH ARTICLE Open Access Invitations and incentives: a qualitative study of behavioral nudges for primary care screenings in Armenia Estelle Gong1, Adanna Chukwuma2* , Emma Ghazaryan2 and Damien de Walque2 Abstract Background: Non-communicable diseases account for a growing proportion of deaths in Armenia, which require early detection to achieve disease control and prevent complications. To increase rates of screening, demand-side interventions of personalized invitations, descriptive social norms, labeled cash transfers, and conditional cash transfers were tested in a field experiment. Our complementary qualitative study explores factors leading to the decision to attend screening and following through with that decision, and experiences with different intervention components. Methods: Informed by the Health Belief Model as our conceptual framework, we collected eighty in-depth interviews with service users and twenty service providers and analyzed them using open coding and thematic analysis. Results: An individual’s decision to screen depends on 1) the perceived need for screening based on how they value their own health and perceive hypertension and diabetes as a harmful but manageable condition, and 2) the perceived utility of a facility-based screening, and whether screening will provide useful information on disease status or care management and is socially acceptable. Following through with the decision to screen depends on their knowledge of and ability to attend screenings, as well as any external motivators such as an invitation or financial incentive. Conclusions: Personalized invitations from physicians can prompt individuals to reconsider their need for screening and can, along with financial incentives, motivate individuals to follow through with the decision to screen. The effect of descriptive social norms in invitations should be further studied. Efforts to increase preventive screenings as an entry point into primary care in Armenia may benefit from implementation of tailored messages and financial incentives. Trial registration: The protocol was approved on January 11, 2019 by the Institutional Review Board of the Center of Medical Genetics and Primary Health Care in Armenia (02570094). https://www.socialscienceregistry.org/trials/3 776. Keywords: Screening, Primary care, Financial incentives, Behavioral economics, Cash transfers, Descriptive social norm, Nudge, Hypertension, Diabetes * Correspondence: achukwuma@worldbank.org 2 World Bank Group, Washington, DC 20433, USA Full list of author information is available at the end of the article © The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Gong et al. BMC Health Services Research (2020) 20:1110 Page 2 of 14 Background for free at polyclinics in urban areas, ambulatory facil- Globally, over 15 million people die prematurely every ities in rural areas, and outpatient health centers. How- year due to non-communicable diseases (NCDs) and ever, the costs of outpatient medicines and expensive around 85% of those deaths occur in low- and middle- diagnostic care are not covered except for specific income countries (LMICs) [1, 2]. The rise of NCDs, which groups such as military personnel and their families or includes cardiovascular diseases, cancers, respiratory dis- those classified as socially vulnerable, such as veterans, eases, and diabetes, is due to multiple socio-economic fac- disabled individuals, and children of a certain age or dis- tors that include aging, exposure to tobacco, and advantaged households [12–14]. Hence, one in five Ar- unhealthy lifestyles. At the individual level, these socio- menians still report that financial barriers are reasons economic factors can impact metabolic risk factors like for forgoing care. As a result, Armenians average 4 an- high blood pressure and high blood glucose, which may nual outpatient visits per person, below the European be diagnosed and managed as hypertension and diabetes average of 7.1 visits [15]. Primary care use may be de- mellitus, respectively [2]. Diabetes mellitus, in addition to terred by the perception that primary care is of poor its own health consequences such as nerve damage and quality and lacks regulation [16]. diabetic retinopathy if improperly managed, joins hyper- To increase utilization of primary care, the Armenian tension in increasing the risk of death from cardiovascular government collaborated with the World Bank and other diseases such as heart attack and stroke [3–5]. donors to launch, in 2013, supply-side reforms such as Detecting hypertension and diabetes mellitus is essen- performance-based financing and clinical guideline de- tial to initiating treatment, achieving disease control, and velopment and a demand-side mass media campaign delaying the onset of other NCDs. However, a high pro- educating the public on the benefits of attending screen- portion of people living with hypertension and diabetes ings. However, despite these ongoing programs, screen- mellitus have never been diagnosed, particularly in ing rates were persistently low. In 2016, only 43.5% of LMICs. Of the 1.4 billion people living with diabetes glo- Armenians over 15 years old had their blood pressure bally, only half are aware of their diagnosis and more measured by a health provider and only 24% had their than 80% of undiagnosed cases live in LMICs [6]. Only blood glucose levels measured in the past 12 months 39.2% of people living with hypertension in LMICs have [16]. Focus group discussions with service users and pro- ever been diagnosed, contributing to the low proportion viders highlighted potential barriers to screening, includ- (10.3%) that achieve control of their blood pressure [7]. ing the underestimation of the harmful effects of Screening through medical examinations and tests in- hypertension and diabetes and the high perceived costs creases opportunities to diagnose hypertension and dia- of attending screenings and post-diagnosis care [17]. betes, even when individuals are asymptomatic or have The empirical literature highlights the potential for be- mild symptoms. Furthermore, screening of high-risk havioral “nudges” such as physician reminders, invitation populations can be cost-effectively delivered in primary letters, and financial incentives to increase preventive care facilities or by community health workers even in health care use. For example, studies show that invita- LMICs [8, 9]. tions to attend health checks or screenings for specific In Armenia, a middle-income country in the South diseases such as cancer can increase attendance com- Caucasus, NCDs account for 93% of deaths in the coun- pared to no intervention [18–20]. The contents of invita- try, including cardiovascular diseases (55.2%), cancers tions have been investigated as well, from inclusion of a (20.2. percent), and diabetes mellitus (1.5%) [10]. In this provider signature to descriptive social norms that aim small country with a population of 2.9 million, prema- to influence individual behaviors by referencing peer be- ture mortality rates from diabetes mellitus (560.4 per haviors [18, 21]. Descriptive social norms messaging has 100,000 population) are higher than the average in coun- been shown to influence behaviors in other settings such tries with similar sociodemographic indicators (222.2 per as charitable giving [22], voting [23], and water and en- 100,000 population), including Azerbaijan (465.6 per ergy consumption [24]. In health behavior applications, 100,000 population) and Georgia (429.9 per 100,000 descriptive social norms have been used to encourage population) within the South Caucasus region [11]. The healthy lifestyle habits [25–28] and increased uptake of high burden of NCDs results in part from gaps in access health services like health checks and screenings [21, to and use of high-quality health care, preventing the 29]. Financial incentives such as cash transfers and promotion of healthy lifestyles, early diagnosis, and pre- vouchers have also been shown to increase preventive vention of complications. health care use and even health outcomes in settings of General practitioners and family physicians at the pri- low- and middle-income countries [30–33]. However, mary care level are responsible for health promotion, there is mixed evidence on the relative impact of condi- screening, treatment, monitoring, and referral to special- tional and unconditional cash transfers as well as their ist care in Armenia. Primary care services are provided acceptability among target populations [34–37]. Gong et al. BMC Health Services Research (2020) 20:1110 Page 3 of 14 Given the evidence on messages and incentives to in- the two pharmacy chains where the voucher could be crease preventive care use, the low screening rates in redeemed for themselves or another person, the choice Armenia, and the focus group discussion findings of products, and the monetary value of AMD 5000 (USD highlighting low perceived risk of hypertension and dia- 10). The offer of the voucher was labeled as an encour- betes and high perceived costs, there is a need to assess agement to screen for hypertension and diabetes. whether such interventions may be effective in increas- Intervention group 4 received the same invitation as ing screening attendance in Armenia. Currently, there is group 1 and instructions that screening could be re- a gap in the literature on the impact of behavioral inter- ported to a designated research assistant, who, upon ventions on primary health care use in the Armenian verifying attendance with the local facility, would pro- context. In addition to the potential quantitative impact vide them a pharmacy voucher. The voucher was identi- on screening rates, there is an opportunity for qualitative cal to the one provided to group 3, except that it was investigation to understand user and provider experi- not labeled as an encouragement to screen, but rather as ences with such interventions and determine the mech- a conditional voucher. anism of influence that these interventions have on The control group did not receive any intervention health behaviors in Armenia. under this study, including information and pharmacy This qualitative study complemented a randomized vouchers. However, the ongoing mass media campaign control trial that was conducted in four provinces in funded through the Ministry of Health provided infor- Armenia to evaluate the impact of financial and non- mation on the prevalence of diabetes and hypertension financial demand-side personalized incentives to in- and encouraged adults to screen for free for hyperten- crease screening attendance for hypertension and dia- sion and diabetes in the local primary care clinic. These betes at primary care centers in Armenia [38]. Two campaigns involved broadcasts on public and private thousand individuals aged 35 to 68 who had not television stations, billboards on major roads, posters in screened in the previous year were randomly selected health facilities and post offices, and text messages from from Armenia’s national e-health database and random- the Ministry of Health. ized to one of the four intervention groups or the con- The first three interventions increased the probability trol group. All intervention groups were delivered a of screening for both hypertension and diabetes by about verbal message and written letter that varied by group. 15 percentage points compared to the control group, The verbal message was delivered based on a script in a while the conditional vouchers in intervention group 4 written guide, which all fieldworkers were trained to use doubled the size of that impact. The conditional for consistency in each intervention group. The inter- vouchers were equally cost-effective as messaging inter- vention delivery was verified as part of the process evalu- ventions, while the unconditional vouchers were less ation implemented in parallel with this qualitative study. cost-effective than the other interventions [39]. To in- We describe the content of tested interventions below, form an understanding of the mechanisms underlying while full invitation templates are available in screening behavior in the field experiment, we under- Additional File 1. took a qualitative study. Hence, the objective of the Intervention group 1 received a personal invitation study is to understand the factors influencing the deci- that was delivered in person by study fieldworkers, sion to attend screenings and following up with that de- which consisted of a verbal message and printed letters cision, and experiences with the intervention signed by a physician. The letter invited individuals to components. screen for hypertension and diabetes, highlighted the im- portance of screening for one’s health, noted that Methods screening was free of charge and not time consuming, Conceptual framework and provided the address and phone of the community This study’s conceptual framework was informed by the medical facility where an appointment could be made. Health Belief Model (HBM), which originated in pre- Intervention group 2 received the same invitation as ventive health programs led by the United States Public group 1, but the verbal invitation and letter included Health Service in the 1950s and 1960s [40]. In this additional information on the number of men and model, an individual’s likelihood to take action to avoid women in the individual’s peer group that had screened disease depends on five dimensions: perceived suscepti- in their communities. Peer groups were defined within bility to the disease, perceived severity of the disease, per- three age bands: 35–45 years, 46–55 years, and 56–68 ceived benefit of an action in reducing severity or years. susceptibility, perceived barriers to taking action, and the Intervention group 3 received the same invitation as cue to action. The cue to action is a factor that leads an group 1 and a uniquely numbered, single use pharmacy individual to actually change their behavior, which may voucher and instructions for use. The letter identified or may not be consistent with their perceptions of Gong et al. BMC Health Services Research (2020) 20:1110 Page 4 of 14 susceptibility or severity. The cue to action can be in- perceived barriers related to costs of attending screening ternal, such as the experience of a symptom, or external, and add to a sense of self-efficacy (by reducing the bur- such as a media campaign or encouragement from a den of medication costs) or signal the value (and benefit) friend. In the 1980s, the dimension of self-efficacy was of screening that impacts the perceived severity of or added, as applications of the HBM expanded from one- susceptibility to hypertension and diabetes. time preventative actions (such as vaccinations) to more long-term behaviors to prevent and manage chronic dis- Study design ease [41]. Individuals require a greater sense of self- Our study uses thematic analysis to understand the fac- efficacy, or sense of competency, to carry through more tors impacting the decision to attend screenings and fol- complex behavior changes like smoking cessation or lowing up with that decision, and experiences with medication adherence. different intervention arms. Our conceptual framework Because the HBM provides the conceptual language to informed our approach to data collection and analysis, explain not only if a person can and wants to take ac- namely in the interview guide design and coding tion, but also if they will take action, it is an appropriate process. framework for our study. In our review of other health behavior theories and models, there was significant over- Data collection lap with the HBM [42]. For example, the Theory of We conducted a total of 100 one-on-one in-depth inter- Planned Behavior can be useful in exploring an individ- views: 80 with service users who participated in the trial and ual’s desire and intention to act on a specific health be- 20 with service providers at facilities providing screening havior, where a certain behavior depends on an services for participants in the trial. Service provider per- intention that is informed by individual attitudes, sub- spectives on screening behaviors may differ from those of jective norms, and perceived behavioral control [43]. service users and were included to triangulate findings and Compared to the HBM, individual attitudes and subject- identify areas of convergence or divergence. Service users ive norms contribute to perceptions of susceptibility, se- from all intervention arms were equally represented. How- verity, and benefits of action, and perceived behavioral ever, control group service users were not included because control overlaps with an individual’s sense of self- the qualitative study was conducted before the endline efficacy. quantitative survey for the main study. This was done to Using the HBM, we assume that an individual is more avoid contamination of experimental groups with qualitative likely to attend screening if they perceive themselves as interviews that provided information about other groups. susceptible to hypertension or diabetes, perceive hyper- Interviews were conducted with guides that were de- tension or diabetes as sufficiently severe, and perceive veloped and reviewed by the study team with expertise screening attendance to provide benefits in terms of pre- in behavioral economics, medicine, health services re- vention or management of disease. Whether they attend search, sociology, impact evaluation, social work, and also depends on individual factors such as their sense of qualitative research. Interview guide content was in- self efficacy in traveling to the clinic and potentially initi- formed by concepts in the HBM and were pre-tested ating care, and whatever may be their cue to action. with purposively selected individuals that were represen- Hence, the intervention arms were hypothesized to in- tative of the target population. Wording and cues were fluence screening behavior through mechanisms delin- revised based on testing feedback. The guides are avail- eated in the HBM. able in Additional File 2. A personalized invitation may serve as a cue to action Interviews were conducted in January to March 2020 and increase a sense of self efficacy. It may also increase by a team of five Armenian facilitators who were se- the perceived susceptibility to and severity of hyperten- lected based on education (higher education degree in a sion and diabetes, and subsequently the perceived bene- social science field) and their experience with qualitative fits of screening. The invitation may reduce perceived interviewing (minimum of 3 years). Facilitators did not access barriers by clarifying where and how to attend have prior contact with study facilities or target commu- screening. Learning of peer screening attendance rates nities of the screening intervention. Facilitators also had may influence screening behavior by demonstrating dif- no knowledge of the quantitative outcomes of the trial ferences in peer perceptions of severity, susceptibility, and endline screening rates for each intervention group. and benefits of screening. It may serve as a cue to action Facilitators were assigned to interview service users of and reduce perceived barriers related to social norms the same gender to encourage open discussion during around seeking care, which can also add to a sense of interviews. self efficacy. A pharmacy voucher may serve as a cue to Facilitators contacted selected participants by phone action, with a stronger effect for vouchers conditional on and scheduled an interview. Even though the study service screening attendance. Vouchers may also address any users were already involved in the trial, the facilitators Gong et al. BMC Health Services Research (2020) 20:1110 Page 5 of 14 explained the purpose of their follow up visit using an researcher with a Master of Science in Global Health. introductory script. At the time of interviewing, verbal Two members of the analytical team independently con- consent was obtained for permission to audio record and ducted open coding of all interview summaries, respect- transcribe the interview. For service users, interviews took ively using Dedoose [44] and manual coding. This initial place at home, at work, or in some cases, in the facilita- open coding was done separately to avoid bias in code tor’s vehicle based on participant preference. Despite at- development. Codes were created by assigning phrases tempts to ensure privacy for interviews taking place in the capturing the main concept of the statement that related home, at times this was impossible due to space limita- to central research question: what factors influence the tions and multiple occupants of the household. Whether decision to screen and following up on that decision? others were present for the interview, as well as the gen- [45] Following review by the study team of both inde- eral demeanor of the participant, was noted by facilitators. pendent code trees, the code tree was revised to recon- Provider interviews were conducted at facilities, typically cile differences and grouped codes into themes reflecting in provider offices or another office setting. overarching ideas that explained screening behavior of Interviews were transcribed verbatim in Armenian, service users. The team then revised the themes and re- and then summarized in English with supporting quotes. solved any disagreements, presenting the results below Transcription quality checks were conducted by random in accordance with the consolidated criteria for report- validation of transcripts against audio recordings. All se- ing qualitative research (COREQ) [46]. lected respondents were interviewed despite achieving data saturation before all the interviews were completed. Results Table 1 shows characteristics of the sample of service Data analysis users and service providers. Interviews were conducted The analytical team was comprised of a health policy with 80 service users, who were evenly distributed across specialist with a medical degree and doctoral training in gender and screening status. Ages ranged from 35 to 69, health systems, an Armenian physician with a Master of with about half under the age of 50. Of the 20 inter- Public Health, an economist, and a health services viewed service providers, 55% (n = 11) were above the Table 1 Sample Characteristics Service Users (n = 80) Service Providers (n = 20) Variable Screened (n = 40) Not Screened (n = 40) Gender, n (%) Male 20 (25) 20 (25) 4 (20) Female 20 (25) 20 (25) 16 (80) Age, n (%) 30–39 8 (10) 8 (10) 40–49 12 (15) 12 (15) 6 (30) 50–59 11 (13.75) 11 (13.75) 3 (15) 60–69 9 (11.25) 9 (11.25) 11 (55) Marz, n (%) Ararat 6 (7.5) 7 (8.8) 4 (20) Armavir 8 (10) 9 (11.3) 5 (25) Kotayk 11 (13.8) 7 (8.8) 6 (30) Lori 15 (18.8) 17 (21.3) 5 (25) Years in Service, n (%) 10–19 4 (20) 20–29 3 (15) 30–39 8 (40) 40–49 5 (25) Position, n (%) Facility Director 9 (45) Family Physician, General Practitioner or Therapist 11 (55) Gong et al. BMC Health Services Research (2020) 20:1110 Page 6 of 14 age of 60, 80% were female (n = 16) and 45% (n = 9) held Priority to one’s health and concerns for any devia- the title of ambulatory or polyclinic director. Average tions from full health appeared to be a motivation for duration of interviews was 25 min (range of 10 to 41 undertaking screenings. About a third (30/100) of re- min). No interviews were partially conducted or spondents reflected on attitudes towards one’s health, repeated. with seventeen describing indifference towards one’s There were 26 service users who were contacted for well-being as a reason to not seek care. Over half (55/ interviewing and declined to participate. The reasons for 100) of respondents described delaying care until symp- refusal to participate were being busy (n = 12), being out toms are extreme, with eighteen attributing the delay to of the community at the time of contact (n = 8), having an Armenian mindset or as six of those described, “wait- no desire to participate (n = 4), or being ill (n = 2). ing until the knife hits the bone.” Some reasoned that We identified three major themes that appear to ex- this indifference results from the assumption that any plain the decision to screen (Table 2). First, an individ- symptoms or discomfort will pass with time. A third of ual’s decision to screen depends on the perceived need service users (28/80) expressed that family influence for screening based on how they value their own health could help overcome this indifference and encourage and perceive hypertension and diabetes as a harmful but care seeking, while a similar number (26/80) disagreed manageable condition. In addition, an individual’s deci- and were adamant that “no one” could change their be- sion to screen depends on the perceived utility of a havior—only they can decide for themselves. When facility-based screening, and whether screening will pro- asked about whether gender impacted health seeking, 45 vide useful information on disease status or care man- service users thought women may attend more and pro- agement and is socially acceptable. Finally, if an vided reasons such as women caring more about their individual perceives they need to be screened, and per- health, having more health concerns, or having more ceives the facility-based screening as useful, their follow- time. through depends on their knowledge of and ability to at- tend screenings, as well as any external motivators such In general, I am very indifferent to myself. Even as an invitation or financial incentive. when I want to visit a doctor, there is always an obs- tacle. I don’t care much of myself. Sometimes I want to go, but then I say: ‘Screw this, let others go’. I don’t Theme 1: the decision to attend screening is more likely know. (Female, not screened, Group 1) with greater perceived need for screening Service users described their decision to screen based on whether they perceive it was needed or not. This need Whoever views health another way, they do their appears to reflect how individuals value their own best to frequently go [for checks] and find out... health, if they perceive hypertension and diabetes as a Those people are very few, perhaps only about 10% threat to their health, and if they believe that hyperten- of our society, who actually go. Others – the majority sion and diabetes are conditions to be prevented or – don’t go and don’t even want to, and attend only managed. in extreme conditions. (Male, screened, Group 4) Table 2 Themes and sub-themes Themes Sub-themes Theme 1: The decision to attend screening is more likely with greater perceived Prioritizing one’s health need for screening Underestimating the harmful consequences of hypertension and diabetes Believing hypertension can and should be prevented or managed Feeling cared for and being reminded of preventative screening benefits after receiving invitations Theme 2: The decision to attend screening is more likely with greater perceived Seeking information on disease status and management utility of and access to screenings. Trusting the health system Being reminded of information to be gained in screening and increasing trust in the health system after receiving invitations Theme 3: External motivators increase the likelihood of following through with the Being reminded to attend screenings despite time costs after decision to screen receiving invitations In groups 3 and 4, being motivated to screen due to receiving vouchers as a form of assistance Gong et al. BMC Health Services Research (2020) 20:1110 Page 7 of 14 I’ve always said: ‘God forbids such situations, when commented on the need for people to attend you have to visit a doctor.’ We are Armenians, we screenings for preventative purposes rather than think like that and don’t take care much for our- curative purposes, but that this attitude has only re- selves. They say, ‘God forbids such thing to happen, cently started to change. when the knife has reached the bones and you have to visit a doctor’. Armenians think like that in gen- Of course, the preventive is right. When I heal the eral. (Female, not screened, Group 1) plants in my field we start with the preventive, be- cause when the plant is ill, we’ll have big financial Underestimating the harmful consequences of hyper- losses. The same is for people, just like in nature, if tension and diabetes reduces the perceived need for we don’t do preventive healing, productivity greatly screening. Over half of service users (51/80) recognize decreases. (Male, screened, Group 1) the widespread prevalence of hypertension or diabetes, and either have one of the conditions themselves or If people know they have disease they visit a doctor… know someone who does. However, a similar proportion I know few people who go to medical facilities for (50/80) perceive hypertension and diabetes screening as preventive health care. I don’t go for preventive unnecessary if there are no symptoms, with five service health care, although I realize that it is necessary. providers also describing this sentiment from their pa- (Female, not screened, Group 4) tients. This perception suggests that individuals do not fully appreciate the asymptomatic nature of early hyper- Receiving a personalized invitation can encourage ser- tension or diabetes, where one may feel normal but vice users to place more value on their own health and screen positive upon examination. A third of service remind them that preventative action can be taken to users (28/80) also described the preference to monitor prevent or mitigate disease. A quarter of service users and treat any symptoms at home, either with medication (20/80) mentioned feeling cared for upon receiving the from the pharmacy or with home remedies. invitation. The letter also functioned as a reminder that screening is an opportunity to prevent complications I already know the symptoms of diabetes and hyper- and avoid delaying care until urgent attention is tension. If I notice one of the symptoms, I will go for required. screenings. Otherwise, I won’t go. (Female, screened, Group 2) People’s attitude and trust increased a lot due to the intervention, because they had received the docu- ments, and it was something new for them, a differ- My husband’s blood pressure is high, mine is low, ent format, a better one. I don’t know, probably but we measure it anyway. I measure it in case of people were feeling more appreciated, as someone headache, just want to know if it is high or low. We cared about them and they got an invitation. (Fe- do it just to know it is high or low and what we can male, Therapist) do for it at home at that moment. (Female, screened, Group 4) What can be better than the fact that the state has started to be interested in your health. It makes me I have read that you shouldn’t use chemical soaps or happy. (Male, screened, Group 2) medicine when you have diabetes. Then, some people were saying that legs will rest and hurt less if you put them in salty water […] It is written in the book Sometimes people have diabetes but don’t know that if you put 2 cloves under your tongue for 15 about it, it may be in latent period or may be ac- days, it regulates your blood sugar. But I haven’t quired. If they get such letters, they may attend and tried it yet.” (Male, screened, Group 3) avoid further problems. (Female, not screened, Group 2) Individuals may be more likely to screen if they perceive hypertension and diabetes as conditions Theme 2: the decision to attend screening is more likely that can and should be prevented or managed. with greater perceived utility of and access to screenings About a quarter (24/80) of service users thought Service users who chose to screen often perceive that that it is preferable to prevent conditions such as obtaining information on their disease status was useful, hypertension and diabetes, rather than manage or and that subsequent care would provide some health “cure” it once diagnosed. Five service providers benefit. Gong et al. BMC Health Services Research (2020) 20:1110 Page 8 of 14 Screenings may be perceived as useful if individuals attend a preventative screening but finding no purpose want to know their disease status and seek guidance on in it if they cannot act on new information. Coupled how to manage that disease. A quarter (20/80) of service with the psychological burden of being diagnosed, indi- users described gaining knowledge about their health viduals may dismiss the idea of screening altogether. status by going to screenings, and sixteen service users Four service providers supported this sentiment, echoing also mentioned seeking medical expertise about pos- this mentality among their patients. sible conditions. No service users made note of the sta- tistics on neighborhood screening levels that were People don’t always believe in testing results and provided in intervention group 2 and none remembered pass them in different places, in Yerevan, in Hraz- these statistics when prompted. A proportion of service dan. Test results may be contradictory. People are users (29/80) do not find diagnostic information from disappointed because of this; they don’t know who to screenings helpful, and in fact have a fear of screening. apply to get correct results. Here in Charentsavan To avoid the possibility of knowing that they have a con- there is a lack of equipment, this also affects testing dition, some individuals avoid screenings altogether. results, but in Yerevan I think everything is ok. (Fe- male, screened, Group 2) Screenings are important. There are things you don’t know, and you can’t even imagine, but when you go Whenever we visit, they [referring to the staff] are get screened, something turns out, [proving] you having coffee. Say, a person goes there, opens the should’ve gone [for screening] a while ago. (Male, door and they go ‘Wait on, we’re drinking coffee.’ screened, Group 3) What I’m saying is Nikol [referring to Prime Minister Pashinyan] hasn’t visited here, yet. (Male, not It was difficult at the beginning; we were explaining screened, Group 1) people what screenings are for. Now they realize more and even tell each other, ‘You know, I went for If I want to go for screenings today, some of the ser- screening and found out that I have diabetes, you vices will cost money and I don’t have it. If I go for should also go and be tested.’ Now the attendance is screenings and then cannot afford the treatment, better than at the beginning. (Female, General why should I go. I prefer to treat myself at home. (Fe- Practitioner) male, not screened, Group 1) We always were scared and were avoiding screen- The interventions of personalized invitation and peer ings. We cannot accept anything that is new and information potentially impacted the perceived utility of we were asking why we need this. However, it screening by alerting individuals to the information to turns out that we need it. My blood pressure gets be gained by attending screenings and by increasing high often but I try to ignore it. But when I went trust in the health system. Ten service users and three for screenings, they told me that I should use service providers commented on the effect of receiving medicine. Maybe one day I will be broken as well information that peers are attending screenings. They and will go for screenings on time. (Female, described that learning about others’ screening attend- screened, Group 4) ance can make an individual realize that they may be susceptible too, and also remind them that screening is a A screening that yields in diagnosis is only useful if an socially acceptable and positive behavior. Receiving a individual believes that the subsequent care will be bene- formal and signed invitation to screening also served to ficial and can be accessed. Thirty service users and five increase trust in the health system, which was often bol- service providers discussed levels of trust in the health stered by a positive experience upon actually attending a system, with 22 users expressing doubt towards the ex- screening. pertise of providers or the quality of smaller facilities. Eight service users expressed preference for screening in It is possible that people will think that if others Yerevan due to increased accuracy or better equipment. go for screenings, we should also go, that maybe Four individuals attributed distrust to past experiences this is the age when a lot of problems occur. And of being forced to pay bribes by doctors. Three inter- I think that people will go more. (Female, views described improvements in corruption, giving screened, Group 2) credit to the current political administration. The inabil- ity to afford prescribed treatment and continued care I was thinking that this project will help people visit after diagnosis is a major deterrent to deciding to attend polyclinics and trust the health care system, that screening. Seventeen service users described wanting to they will go for the screenings and will receive the Gong et al. BMC Health Services Research (2020) 20:1110 Page 9 of 14 necessary attention, attitude and will be able to helps giving an opportunity to get some medicine for solve their health issues. (Female, screened, Group 2) free, it motivates people to attend screenings. I have not attended just for the money; I would do it any- Theme 3: external motivators increase the likelihood of way. But you should not do it just for the money, it following through with the decision to screen is secondary, it only motivates people more. On the Receipt of the invitation to screen was itself a trigger to other side, you must have money to purchase medi- consider attending screening, and the formal letter deliv- cine for your health. Both are connected to one an- ered by the fieldworker led to a sense of obligation to other. (Female, screened, Group 4) follow through with screening. Eleven service users de- scribed feeling a sense of responsibility after receiving I have received the gift card and did not attend, as I the invitation, and five service providers corroborated didn’t need it. Those people, who got invitation this in their observation of patients. Service users felt wouldn’t attend in the same way. The card could that they made a promise to fieldworkers or felt the de- not force people to attend for screenings. However, sire to participate in the program they were selected for. people will be glad to get that card. (Female, screened, Group 3) You should do this more often, which will keep us vigilant and trigger to visit a doctor even without Nine service users commented on the level of need such things. If you take a pause, we forget about for the vouchers, and how they could be more bene- it, but when you keep repeating it continually, we ficial to poorer individuals. Further, some service start paying more attention. (Female, not users and one service provider preferred that future screened, Group 2) iterations of the program expand the vouchers to everyone so that they may be able to afford medi- This is a very good program, because when a nurse cines. In contrast, others were wary of continued as- calls people, they may not take it serious. But when sistance and cautioned against creating the they get a call from a different organization and they expectation of aid. get these letters, this is kind of obligatory for them, they feel more responsible (Female, Director of I later learnt that some other people from my neigh- Polyclinic) borhood besides the invitation also received vouchers with 5000 AMD. I got angry and I asked myself Pharmacy vouchers were perceived as motivating whether I was not enough poor to receive it as well. to the service users who received them (26/41) and (Female, not screened, Group 1) were viewed by recipients as a form of financial aid, ameliorating the cost of care, rather than a signal of Any pensioner or social assistance beneficiary needed screening value, that was modest but useful when it more than me, but it may form a habit and then needed. Nearly half (41/100) of service users and pro- they would say: “Why don’t you support me continu- viders believed the conditional voucher would most ally?” (Male, not screened, Group 3) likely motivate screening attendance, while less than half of that (15/100) expressed that the unconditional External motivators such as the invitation and voucher voucher to be adequate in motivating attendance. Ser- were enough for some service users to overcome the vice providers noted that the rumor of financial re- challenge of finding the time to go to the clinic and ward for screening was enough to increase screening waiting in queues, that is the time cost of screening, attendance at the beginning of the program. Further, which was an obstacle for the majority (53/80). Most there were service users in Groups 3 and 4 who re- participants found their local facility easily accessible ceived vouchers and attended screenings but and acceptable, though the sometimes-preferred larger expressed they would have attended anyway, even facilities were often more distant. Service providers in without the voucher. All ten of those in Group 3 who Metsamor also noted that facilities in smaller and more did not screen used their vouchers, or gave them to rural areas, might face more challenges in patient someone who did, and explained that the voucher attendance. was appreciated but they did not feel the need to screen. Many of them were visiting us for the first time. Once I gathered the staff and we visited people at Even without the program I would attend for screen- their homes to make screenings for diabetes, as some ings, and the program was in time, as we purchased people were too busy with their work to attend for necessary things. This also helps people, the program screenings. But this time they found time and came. Gong et al. BMC Health Services Research (2020) 20:1110 Page 10 of 14 They had a strong belief in you [the program officer]. fear of screening and anxiety from being diagnosed with (Female, Director of Ambulatory) a condition that requires unaffordable treatment. Finally, our domain of following through with the decision to Discussion screen identifies the importance of external motivators This study is one of the few that explores the factors and reduction in access barriers including time to attend explaining preventive health care use at the primary care screenings. These concepts are analogous to the cue to level in Armenia. It provides insights that may be rele- action and addressing perceived barriers in the HBM. vant for other contexts with low levels of diagnosis of While we do not aim to build on the HBM, future re- hypertension and diabetes and coincident search on concepts such as the intrinsic valuing of one’s underutilization of primary care. We also describe the health as a predictor of health utilization can be further mechanisms through which demand-side interventions, investigated with careful field-based research. Exploring including messages and financial incentives, may be use- whether this concept is specific to Armenia or ful in improving the use of preventive health care. generalizable to other contexts may be a first step to Three themes emerged that appear to explain the deci- building on the HBM [47]. sion to attend clinic screenings and follow through on Our findings on screening attendance align with other that decision. An individual must first perceive that a qualitative literature on factors influencing preventive screening is needed. This need is informed by the im- care-seeking behavior in a range of settings. For ex- portance placed on their own health, the perception that ample, qualitative investigations into care avoidance in hypertension and diabetes is sufficiently harmful, and the US [48] and preventive care non-attendance in the that the conditions can and should be managed. Should UK [49] both identified low perceived need for care (or an individual choose to take action to prevent and man- relevance of health checks) as a reason given for not age hypertension and diabetes, they may seek out seeking care and found negative past experiences to be a facility-based screening if they believe that the encounter potential contributor to avoidance. Access barriers such will provide useful and wanted information about their as time and financial cost were present in both studies, disease status and treatment. Screenings may not be per- as factors reducing care use, which is consistent with ceived as useful if there is a lack of trust in the health our findings. In our study, service users reported feeling system or there are doubts about being able to afford cared for upon receiving an invitation and that they felt subsequent treatment. External motivators such as a per- obligated to participate after receiving such attention. sonalized invitation increased the sense of trust in the Similar results of feeling a responsibility to attend after health system, while the vouchers were viewed as a form being asked, or feeling obligated to participate, were of aid that was helpful to purchase medicines and motiv- found in qualitative explorations of invitations to pre- ating to make time to attend screenings. ventive health checks in Denmark [50] and the UK [51]. While we used the HBM for data collection and ana- Finally, expensive medication was reported as a reason lysis, our results expand from some aspects of the HBM to forgo screenings in our study, since any treatment in context-specific ways. Novel themes emerged that may be unaffordable following potential diagnosis. In were specific to explaining care-seeking behavior in studies on hypertension diagnosis and treatment in Armenia that deviated from the HBM but better synthe- Bangladesh [52] and Colombia [53], costs of medication sized the experience of service users and providers. Our was also identified as a barrier to maintaining treatment, domain of perceived need for screening as a predictor of and lapses in treatment were reported as well if symp- attendance includes some elements analogous to the toms abated. Our study echoed this potential misunder- HBM’s concept of perceived severity and susceptibility standing of drug therapy, where medication was as influencing health behavior. Hence, we find that the reserved for emergent symptoms rather than regulating perception that hypertension and diabetes is common, blood pressure or blood sugar. potentially harmful, and preventable increases the likeli- Despite the literature on the potential for descriptive hood of deciding to screen. However, in the Armenian social norms to encourage behavior change in applica- context, we also identified the added dimensions of in- tions such as dietary choices [54] or energy-saving be- trinsically valuing one’s health as predictive of screening havior [24], there was no qualitative evidence that behavior. With low priority for one’s health, there is lit- service users factored the statistics on peer behavior into tle motivation to act on prevention or treatment even if their own decision to seek care. No service users could severity and susceptibility is acknowledged. The concept recall the specific contents of the invitation letter and of perceived utility of screening which increases screen- spoke more on receiving the invitation itself. However, it ing use, is analogous to perceived benefits in the HBM is possible that peer screening rates were so low, even in model. In the Armenian context, we highlight that the terms of absolute numbers of people who screen, that perceived benefits or utility of screening are reduced by the information on peer screening habits did not serve Gong et al. BMC Health Services Research (2020) 20:1110 Page 11 of 14 as an adequately motivating social norm. In fact, the let- quantitative data showed higher attendance under ter may have even described and reinforced a social Group 4 and sheds light on the value of conditionality in norm of low screening rates. Additional research into increasing targeted health behaviors [38]. the impact on descriptive social norms on care Providers observed an increase in screening attend- utilization should focus on issues with larger differences ance by those who merely heard of the program and between the target group and peers. promise of financial reward. This observation indi- While service users did not comment on letters cates demand for the voucher program beyond this containing descriptive social norms, participants from trial but also raises questions of ethics for interven- Group 2 and others described learning that neighbors tion scale-up, such as equity in access to vouchers and peers attended screenings as a potential motiv- that improve affordability of necessary medications ator for service users, especially if benefits were evi- and supplies. Both service users and providers com- dent (such as finding out disease status or getting mented on the greater need for such vouchers among treatment). A randomized control trial in the UK that poorer groups, and some proposed that everyone be tested variations of an invitation letter on screening eligible for vouchers if the program were expanded. attendance, including one with descriptive social Any plans to scale up the program should consider norms, found similar results where the letter with de- the acceptability and ethics of using financial incen- scriptive norms did not perform better than other tives to influence health-related decision making, in versions in increasing attendance [21]. In the context addition to the non-financial nudges of invitations of our study, absolute numbers were used to describe and social norms messaging. Further research is peer screening rather than percentages, since needed to determine the acceptability of such nudges highlighting the low percentages of attendance was for health programs in Armenia, in addition to other not considered to be motivating information. If repli- criteria such as sustainability and cost, and address cated, further research should explore potential differ- any concerns on manipulation of decision-making ences in forms of social norm messaging. In addition, that has been raised in debates on the ethics of peers were defined in our study to refer to neighbors nudges [58, 59]. As acceptability of such programs within a similar age range. It may be, however, that varies by country [60, 61], program modifications can these are not the comparators that are relevant to the be made for the Armenian context to ensure that average Armenian. Future research on this subject choices and accompanying consequences are suffi- should include formative research to characterize the ciently transparent and fair to potential service users profile of peers, information on whom may influence in Armenia. screening behavior. Our qualitative investigation suggests that personalized Conditional and unconditional cash transfers, or its invitations from physicians are simple but effective pol- variants such as vouchers, have been tested in multiple icy measures to encourage screening attendance by demand-side interventions aiming to incentivize health prompting individuals to reassess their need for screen- care behaviors [55]. They have been shown to increase ing and consider their local facility as an acceptable op- levels of the desired behavior, with conditional cash tion to do so. Including information on peer screening transfers yielding greater initial impacts than uncondi- rates, if they are not sufficiently different from the indi- tional cash transfers [56]. Perceptions of service users vidual’s screening behavior, may not enhance the effect and providers were in line with this sentiment, where of an invitation perceptibly. An additional pharmacy participants believed that Group 4 (conditional voucher will likely be appreciated and used by recipients, vouchers) would be more motivated to attend screenings but attendance may not appreciably increase relative to to receive the pharmacy voucher. Participants described messages if the voucher is not conditional on it. Hence, that despite the modest amount, the voucher was mean- from a policy perspective, personalized invitations are a ingful and relieved them of significant medication costs. potentially sustainable and acceptable intervention to in- This responsiveness to vouchers confirms that medica- crease screening attendance. While vouchers are effect- tion costs are a burden for many Armenian families, and ive, they require resources to sustain. In the cost- increasingly so for lower income families [57]. However, effectiveness analysis of intervention types, the condi- it should also be noted that while those in Groups 3 and tional vouchers were equally cost-effective as messaging 4 expressed that vouchers motivate attendance, some in interventions, while the unconditional vouchers were Group 3 used the vouchers but did not screen because less cost-effective than the other interventions [38]. they did not feel the need. In contrast, some participants Feedback from participants highlights the importance of of the both voucher groups attended screenings and addressing high out-of-pocket costs of hypertension and claimed they would have done so regardless of the vou- diabetes treatment that follows a diagnosis, including cher. While participants may state this is the case, outpatient medicines. The costs are prohibitive enough Gong et al. BMC Health Services Research (2020) 20:1110 Page 12 of 14 to discourage screening even if there is perceived need, costs and incentivizes attendance if voucher receipt is since treatment would be unobtainable in the event of conditional on it. Invitations and conditional vouchers diagnosis. are recommended policy options to increase screening Our study also highlights the importance of policies attendance, which can be augmented with other policies to expand education on hypertension and diabetes targeting health education and out-of-pocket medication risk and management. As evidenced by sporadic treat- expenditure. ment with medication or home remedies, individuals may not fully understand hypertension or diabetes as Supplementary Information conditions where blood pressure or blood sugar levels The online version contains supplementary material available at https://doi. should be maintained rather than treated emergently. org/10.1186/s12913-020-05967-z. Given the registration of 85 to 100% of the popula- Additional file 1. tion with a primary care provider, and the increase in Additional file 2. screening attendance following messages from a fam- ily physician, these may be effective channels for tai- lored messages to target groups on the risks and Abbreviations HBM: Health Belief Model; LMIC: Lower- and middle-income country; management of NCDs. Our study highlights the im- NCD: Non-communicable disease portance of personalized messages in promoting pre- ventive care use which had additional value in a Acknowledgements context where mass media campaigns had been con- The team is grateful for the guidance and helpful comments from Sylvie Bossoutrot, Tania Dmytraczenko, Volkan Cetinkaya, Gil Shapira, Alaka Holla, ducted for many years. and Gabriel Francis. The team also greatly appreciates the engagement of Our study has some limitations. Because our sample the Ministry of Health, Health Project Implementation Unit, and participating was restricted to participants of the randomized control primary care facilities for their invaluable contributions to the design and implementation of this study. Finally, we acknowledge the excellent trial that selected individuals who had not screened in fieldwork undertaken by Media Model, Armenia. the previous 12 months, our results may not be generalizable to individuals who differ in terms of Authors’ contributions screening behavior. Another limitation is our exclusion DDW and ADC conceptualized and designed the study. ETG and EG analyzed the data. All authors contributed to drafting and revising of control group participants, which may have illustrated intellectual content. All authors read and approved the final manuscript. any differences in care-seeking preferences in the ab- sence of this intervention program. It is also difficult to Funding evaluate any impact of the Group 2 invitation where de- Intervention implementation and data collection was funded through the Health Results Innovation Trust Fund (HRITF) and the Strategic Impact scriptive social norms were used, as the peer screening Evaluation Fund (SIEF) managed at the World Bank. The funding source was rates may have been too low to motivate participants. not involved in the study design, the collection, analysis, and interpretation Future research should further explore the element of of the data, or the manuscript writing. facility size and its perceived acceptability. Individuals Availability of data and materials may believe that seeking care is more burdensome if The datasets generated and analyzed during the current study are not they overlook their local facility and have greater trust in publicly available due potentially identifiable information but are available a larger but more distant (often urban) facility. Trust in from the corresponding author on reasonable request. the health system should also be studied, especially in Ethics approval and consent to participate relation to health seeking behaviors and recent reforms The protocol was approved on January 11, 2019 by the Institutional Review to reduce corruption. Board of the Center of Medical Genetics and Primary Health Care in Armenia (02570094). Under the prior field experiment, all participants had signed written consent forms. The Institutional Review Board approved a verbal Conclusion consent process for the use of an audio recorder and interview transcription This qualitative study explores the mechanisms through under the qualitative study, as the study presented no more than minimal which various demand-side interventions can increase risk to participants and had been applied to similar studies in this context. In this consent process, study facilitators reviewed a printed consent document attendance at facility-based hypertension and diabetes with participants, obtained verbal consent, and recorded in writing that the screenings in Armenia. Deciding to seek health care ser- participant provided verbal consent along with the date. vices such as screening depends on an individual’s per- ceived need to screen, the perceived utility of screening, Consent for publication Not applicable. and any external motivators that impact following up on the decision to screen. Personalized invitations can en- Competing interests courage screening attendance because they alert individ- The findings, interpretations, and conclusions expressed in this paper are uals to the potential need to prevent or manage entirely those of the authors. They do not necessarily represent the views of the World Bank and its affiliated organizations, or those of the Executive hypertension and diabetes. Vouchers are appreciated by Directors of the World Bank or the governments they represent. The authors individuals who often have considerable medication declare that they have no competing interests. Gong et al. BMC Health Services Research (2020) 20:1110 Page 13 of 14 Author details standard letters, risk-personalised letters and telephone invitations. BMC 1 Mount Sinai Health System, New York, NY 10019, USA. 2World Bank Group, Public Health. 2019 Feb 21;19(1):224. Washington, DC 20433, USA. 21. Sallis A, Sherlock J, Bonus A, Saei A, Gold N, Vlaev I, et al. Pre-notification and reminder SMS text messages with behaviourally informed invitation Received: 26 July 2020 Accepted: 25 November 2020 letters to improve uptake of NHS health checks: a factorial randomised controlled trial. BMC Public Health. 2019;19(1). 22. Frey BS, Meier S. Pro-social behavior in a natural setting. J Econ Behav Organ. 2004;54(1):65–88. References 23. Gerber AS, Rogers T. Descriptive social norms and motivation to vote: 1. Cheong A, Liew S, Khoo E, Mohd Zaidi N, Chinna K. Are interventions to Everybody’s voting and so should you. J Polit. 2009 Jan;71(1):178–91. increase the uptake of screening for cardiovascular disease risk factors 24. Allcott H. Social norms and energy conservation. J Public Econ. 2011 Oct; effective? A systematic review and meta-analysis. BMC Fam Pract. 2017; 95(9–10):1082–95. 18(1). 25. Collins EIM, Thomas JM, Robinson E, Aveyard P, Jebb SA, Herman CP, et al. 2. World Health Organization. Noncommunicable diseases. 2018 [cited 2020 Two observational studies examining the effect of a social norm and a Nov 3]. Available from: https://www.who.int/news-room/fact-sheets/detail/ health message on the purchase of vegetables in student canteen settings. noncommunicable-diseases. Appetite. 2019 Jan 1;132:122–30. 3. Islam SMS, Purnat TD, Phuong NTA, Mwingira U, Schacht K, Fröschl G. Non‐ 26. Thomas JM, Ursell A, Robinson EL, Aveyard P, Jebb SA, Herman CP, communicable diseases (NCDs) in developing countries: a symposium et al. Using a descriptive social norm to increase vegetable selection report. Global Health. 2014;10(1). in workplace restaurant settings. Health Psychol. 2017 Nov 1;36(11): 4. World Health Organization. Diabetes. 2020 [cited 2020 Nov 3]. Available 1026–33. from: https://www.who.int/news-room/fact-sheets/detail/diabetes. 27. Koeneman MA, Chorus A, Hopman-Rock M, Chinapaw MJM. A novel 5. World Health Organization. Hypertension. 2019 [cited 2020 Nov 3]. Available method to promote physical activity among older adults in residential care: from: https://www.who.int/news-room/fact-sheets/detail/hypertension. an exploratory field study on implicit social norms. BMC Geriatr. 2017;17(1). 6. Saeedi P, Petersohn I, Salpea P, Malanda B, Karuranga S, Unwin N, et al. 28. Dohnke B, Weiss-Gerlach E, Spies CD. Social influences on the motivation to Global and regional diabetes prevalence estimates for 2019 and projections quit smoking: Main and moderating effects of social norms. Addict Behav. for 2030 and 2045: results from the international diabetes federation 2011 Apr;36(4):286–93. diabetes atlas, 9th edition. Diabetes Res Clin Pract. 2019 Nov;157. 29. von Wagner C, Hirst Y, Waller J, Ghanouni A, McGregor LM, Kerrison RS, 7. Geldsetzer P, Manne-Goehler J, Marcus ME, Ebert C, Zhumadilov Z, Wesseh et al. The impact of descriptive norms on motivation to participate in CS, et al. The state of hypertension care in 44 low-income and middle- cancer screening – evidence from online experiments. Patient Educ Couns. income countries: a cross-sectional study of nationally representative 2019;102(9):1621–8. individual-level data from 1·1 million adults. Lancet. 2019 Aug;394(10199): 30. Lagarde M, Haines A, Palmer N. The impact of conditional cash transfers on 652–62. health outcomes and use of health services in low and middle income 8. Hutchinson B, Small R, Acquah K, Sandoval R, Nugent R, Davidson T, et al. countries. Vol. 2009, Cochrane database of systematic reviews: John Wiley The investment case as a mechanism for addressing the NCD burden: and Sons, Inc. and the Cochrane Library; 2009. Evaluating the NCD institutional context in Jamaica, and the return on 31. Gopalan SS, Mutasa R, Friedman J, Das A. Health sector demand-side investment of select interventions. Husain MJ, editor. PLoS One. 2019;14(10): financial incentives in low- and middle-income countries: a systematic e0223412. review on demand- and supply-side effects. Soc Sci Med. 2014 Jan;100:72– 9. Kane J, Landes M, Carroll C, Nolen A, Sodhi S. A systematic review of 83. primary care models for non-communicable disease interventions in sub- 32. Vlaev I, King D, Darzi A, Dolan P. Changing health behaviors using financial Saharan Africa. BMC Fam Pract. 2017 Dec 23;18(1):46. incentives: a review from behavioral economics. BMC Public Health. 2019; 10. Andreasyan D, Bazarchyan A, Sargsyan S, Torosyan A, Mirzoyan M, Bidzyan L. 19(1):1–9. Health System Performance Assessment, Armenia; 2019. 33. de Walque D, Fernald L, Gertler P, Hidrobo M. Cash transfers and child and 11. Institute for Health Metrics and Evaluation. Armenia | Institute for Health adolescent development. In: Disease control priorities, third edition (volume Metrics and Evaluation. 2019 [cited 2019 Nov 10]. Available from: http:// 8): child and adolescent health and development. Washington, DC: The www.healthdata.org/armenia. World Bank; 2017:325–42. 12. Chukwuma A, Meessen B, Lylozian H, Gong E, Ghazaryan E. Strategic 34. Baird S, Ferreira FHG, Özler B, Woolcock M. Relative effectiveness of purchasing for better health in Armenia [Internet]. Washington, DC: The conditional and unconditional cash transfers for schooling outcomes in World Bank; 2020. Available from: https://openknowledge.worldbank.org/ developing countries: a systematic review. Campbell Syst Rev. 2013 Jan;9(1): handle/10986/34491. 1–124. 13. The Government of Armenia. Decree on State Funded Health Care for Free 35. Akresh R, de Walque D, Kazianga H. Evidence from a randomized evaluation and at Preferential Conditions. 2004. Available from: https://www.arlis.am/ of the household welfare impacts of conditional and unconditional cash documentview.aspx?docid=12227. transfers given to mothers or fathers. Policy Research Working Paper. 2016. 14. Lavado R, Hayrapetyan S, Kharazyan S. Expansion of the Benefits Package: Report No.: 7730. The Experience of Armenia; 2018. 36. Benhassine N, Devoto F, Duflo E, Dupas P, Pouliquen V. Turning a shove 15. World Health Organization, WHO Regional Office for Europe. Health for All/ into a nudge? A “labeled cash transfer” for education. Am Econ J Econ Outpatient contacts per person per year. 2019 [cited 2020 Jun 25]. Available Policy. 2015 Aug 1;7(3):86–125. from: https://gateway.euro.who.int/en/hfa-explorer/#UuCsaHQKBK. 37. Hoskins K, Ulrich CM, Shinnick J, Buttenheim AM. Acceptability of financial 16. Andreasyan D, Bazarchyan A, Manukyan S, Muradyan G, Torosyan A, incentives for health-related behavior change: An updated systematic Chamanyan A, et al. Health System Performance Assessment, Armenia. review. Prev Med (Baltim). 2019;126. 2016. 38. de Walque D, Chukwuma A, Ayivi-Guedehoussou N, Koshkakaryan M. 17. International Marketing Research. Qualitative report on concept testing for Invitations, incentives, and conditions: a randomized evaluation of demand- public relations strategy for screening under the Disease Prevention and side interventions for health screenings in Armenia [Internet]. Policy Control Project. 2018. Research Working Paper. 2020. Report No.: 9346:1-32. Available from: 18. Camilloni L, Ferroni E, Cendales BJ, Pezzarossi A, Furnari G, Borgia P, et al. https://www.ssrn.com/abstract=3666876. Methods to increase participation in organised screening programs: A 39. Chukwuma A, de Walque D, Gong E, Ayivi-Guedehoussou N, Ghazaryan E, systematic review. BMC Public Health. BioMed Central. 2013;13:1–16. Koshkakaryan M. Invitations and incentives for primary care screenings in 19. Bonfill Cosp X, Marzo Castillejo M, Pladevall Vila M, Marti J, Emparanza JI. Armenia. 2020. Available from: http://documents1.worldbank.org/curated/ Strategies for increasing the participation of women in community breast en/373481600188869567/pdf/Invitations-and-Incentives-for-Primary-Care- cancer screening. Cochrane Database Syst Rev. 2001;(1). Screenings-in-Armenia.pdf. 20. Gidlow CJ, Ellis NJ, Riley V, Chadborn T, Bunten A, Iqbal Z, et al. Randomised 40. Rosenstock IM. Historical origins of the health belief model. Health Educ controlled trial comparing uptake of NHS health check in response to Monogr. 1974;2(4):328–35. Gong et al. BMC Health Services Research (2020) 20:1110 Page 14 of 14 41. Rosenstock IM, Strecher VJ, Becker MH. Social learning theory and the health belief model. Health Educ Q. 1988;15(2):175–83. 42. Glanz K, Bishop DB. The role of behavioral science theory in development and implementation of public health interventions. Annu Rev Public Health. 2010 Mar 17;31(1):399–418. 43. Manstead ASR. Attitudes and behavior. In: International Encyclopedia of the Social & Behavioral Sciences; 2001. 44. Dedoose Version 8.0.35, web application for managing, analyzing, and presenting qualitative and mixed method research data [Internet]. Los Angeles, CA: SocioCultural Research Consultants, LLC; 2018. Available from: www.dedoose.com. 45. Hsieh H-F, Shannon SE. Three approaches to qualitative content analysis. Qual Health Res. 2005 Nov;15(9):1277–88. 46. Tong A, Sainsbury P, Craig J. Consolidated criteria for reporting qualitative research (COREQ): a 32-item checklist for interviews and focus groups. Int J Qual Heal Care. 2007;19(6):349–57. 47. Carlile PR, Christensen CM. The cycles of theory building in management research; 2004. 48. Taber JM, Leyva B, Persoskie A. Why do people avoid medical care? A qualitative study using National Data. J Gen Intern Med. 2014;30(3):290–7. 49. Ellis N, Gidlow C, Cowap L, Randall J, Iqbal Z, Kumar J. A qualitative investigation of non-response in NHS health checks. Arch Public Heal. 2015; 73(1):1–8. 50. Broholm-Jørgensen M, Kamstrup-Larsen N, Guassora AD, Reventlow S, Dalton SO, Tjørnhøj-Thomsen T. ‘It can’t do any harm’: a qualitative exploration of accounts of participation in preventive health checks. Health Risk Soc. 2019 Feb 17;21(1–2):57–73. 51. Burgess C, Wright AJ, Forster AS, Dodhia H, Miller J, Fuller F, et al. Influences on individuals’ decisions to take up the offer of a health check: a qualitative study. Health Expect. 2015 Dec;18(6):2437–48. 52. Naheed A, Haldane V, Jafar TH, Chakma N, Legido-Quigley H. Patient pathways and perceptions of hypertension treatment, management, and control in rural Bangladesh: a qualitative study. Patient Prefer Adherence. 2018;12:1437–49. 53. Legido-Quigley H, Lopez PAC, Balabanova D, Perel P, Lopez-Jaramillo P, Nieuwlaat R, et al. Patients’ knowledge, attitudes, behaviour and health care experiences on the prevention, detection, management and control of hypertension in Colombia: a qualitative study. PLoS One. 2015;10(4): e0122112. 54. Huitink M, Poelman MP, van den Eynde E, Seidell JC, Dijkstra SC. Social norm nudges in shopping trolleys to promote vegetable purchases: a quasi- experimental study in a supermarket in a deprived urban area in the Netherlands. Appetite. 2020;151. 55. Wiysonge CS, Paulsen E, Lewin S, Ciapponi A, Herrera CA, Opiyo N, et al. Financial arrangements for health systems in low-income countries: an overview of systematic reviews. Vol. 2017, Cochrane Database of Systematic Reviews: John Wiley and Sons Ltd; 2017. 56. Özler B. How should we design cash transfer programs? Toward Successful Development Policies: Insights from Research in Development Economics. Washington, DC: The World Bank; 2020. 57. Kazaryan I, Sevikyan A, Melikyan M. The prices outpatients have to pay for purchasing medicines in Armenia. J Pharm Policy Pract. 2015;8(S1):P20. 58. Wilkinson TM. Nudging and manipulation. Polit Stud. 2013 Jun 7;61(2):341–55. 59. Hansen PG, Jespersen AM. Nudge and the manipulation of choice. Eur J Risk Regul. 2013 Mar 20;4(1):3–28. 60. Sunstein CR, Reisch LA, Rauber J. A worldwide consensus on nudging? Not quite, but almost. Regul Gov. 2018;12(1):3–22. 61. Sunstein CR, Reisch LA, Kaiser M. Trusting nudges? Lessons from an international survey. J Eur Public Policy. 2019;26(10):1417–43. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.