Identifying barriers and potential solutions to improve equitable access to community eye services in central Kenya: a rapid exploratory sequential mixed methods study ======================================================================================================================================================================= * Luke Allen * Sarah Karanja * Michael Gichangi * Cosmas Bunywera * Emmaculate Muturi * Dickson Gachobi * Purity Kathure * Elizabeth Mutile Muasa * Lorna Mutwiri * Lorna Kajuju * Faith Kagwiria * Benjamin Ntabathia * Hillary Rono * David Macleod * Min Jung Kim * Malebogo Tlhajoane * Matthew J. Burton * Jacqueline Ramke * Nigel M. Bolster * Andrew Bastawrous ## Summary **Background** Recent research has found that less than half of people identified with an eye problem in Meru county’s screening programme were able to access care, with younger adults being the least likely to receive the care they needed. We aimed to interview and survey members of this ‘left-behind’ group to explore barriers and identify potential solutions using a rapid mixed-methods approach. **Methods** First, we conducted interviews to explore perceptions of barriers and potential solutions. Next, we asked a representative sample to rank the suggested solutions by likely impact. Finally, we held a multistakeholder meeting to identify which of the top-ranked interventions offered the best balance of impact, feasibility, cost, and potential risks. We used a deductive matrix and thematic analysis to rapidly analyse the interview data. **Results** We conducted 67 interviews. Barriers to access included long queues, conflicting work engagements, and lack of clear information. Proposed solutions focused on reducing queue lengths, providing better counselling and clinic information, holding mop-up clinics, and maintaining adequate stocks & supplies. We conducted ranking surveys with 401 additional people from the left-behind group. All proposed solutions were ranked at moderately-to-highly likely to improve equitable access. Fifteen people attended the multistakeholder meeting, including community representatives. Workshop participants unanimously selected enhanced counselling and SMS reminders as the interventions that offered the best balance of impact, risk, cost, and feasibility. The other proposed solutions were deemed impractical or unaffordable. **Conclusion** Rapid mixed-methods and multistakeholder collaboration were used to identify a range of potential service modifications that will be implemented within the ongoing programme. Our approach was centred on the experiences and perceptions of those who face the highest barriers to care. **Evidence before this study** Previous research in Kenyan community screening programmes has shown that at least half of those found to have an eye health need will not be able to access care at their local treatment clinic, even if the care is provided free. Work in Meru County has shown that younger adults less are likely than any other sociodemographic group to check-in at their local clinic, but it’s not clear what the specific barriers are for this group. Across the African continent, approximately half of all ambulatory appointments are missed across all specialities, and sociodemographic inequalities are ubiquitous. In pursuit of Universal Health Coverage (UHC) and the Primary Health Care principles of equity and justice, health system managers are increasingly focused on identifying, trying to understand, and then address unequal access to care, however the traditional approach to identifying barriers and solutions has tended to centre around expert opinion rather than engagement with affected groups. **Added value of this study** This study builds on previous efforts to introduce routine sociodemographic data collection into the county-wide eye screening programme operating in Meru, Kenya, as well as additional sites in Meru County, Botswana, Nepal, and Uttar Pradesh. Having already identified younger adults as the least likely to receive care in Meru County, this study introduces a novel mixed-methods approach for engaging with members of this left-behind group to rapidly identify barriers and scalable solutions. We used innovative methods to complete interviews and qualitative analysis in under two weeks, followed by a rapid survey to rank the potential solutions that emerged from this work with a representative sample of younger adults who had not been able to access care. Finally, a multistakeholder workshop with strong local and lay representation identified the top-ranked solutions that would be feasible to introduce and test within the ongoing screening programme. In addition to local evidence for action, this study presents an approach that any community-based programme could use to generate robust, non-tokenistic insights from affected communities within a matter of weeks, minimising the research time requirement and number of senior researchers required whilst maintaining rigorous scientific standards. **Implications of all the available evidence** Equitably advancing UHC is predicated on identifying and overcoming unique barriers to care, however existing efforts rarely involve consultation or co-creation with affected communities. Building on existing rapid qualitative and mixed-methods methods, we have developed a cutting-edge approach to identify barriers, prioritise solutions, and identify service modifications that are feasible to introduce. We have applied this approach in Meru County, where younger adults – who were the least likely to access care – suggested a bundle of interventions centring on improving the provision of information and SMS reminders. Our research group will use an embedded RCT to implement and test this bundle, in the context of an equity-focused continuous improvement model that we are also implementing in Botswana, India and Nepal to incrementally improve access for all, with a focus on left-behind groups. ## Introduction Improving equitable access to community health services lies at the heart of Universal Health Coverage (UHC) and ‘leaving no one behind’ is the ‘central, transformative promise’ of the Sustainable Development Goals.1–4 WHO’s *Thirteenth General Programme of Work* states that ‘the main challenge to making progress towards UHC comes from persistent barriers to accessing health services’. Our research collaborative is developing and testing a novel approach to identify and address inequitable access to care using the ‘IM-SEEN’ approach (‘Improvement studies for evidence-based and equitable innovation’). This involves identifying which groups are being left behind in a given programme; engaging with these groups to understand the unique barriers they face and their ideas for service improvements; and then testing these potential solutions with embedded randomised controlled trials (RCTs).5 We are applying this model in the context of community-based eye screening programmes in Botswana, India, Kenya, and Nepal. Uncorrected visual impairment affects over a billion people worldwide, levying major social and economic costs, despite the availability of highly cost-effective treatments like spectacles and cataract surgery.6 Our first set of findings from a cross-sectional equity analysis of over 4,000 people in Kenya’s Meru county found that only 46% of those found to have an eye need were able to access their free local treatment outreach clinic.7 We found that younger adults, males, and those working in sales, services, or manual jobs were the least likely to receive the care they need. Age was the strongest predictor of poor access, with less than a third of people aged 18-44 years receiving care compared to two thirds of those aged >45 years, even after controlling for severity of eye condition and a wide range of other factors. Traditionally, ideas for how to improve programmes come from ‘experts’, service providers, or surveys of service users - rather than affected people themselves.8,9 In the context of renewed interest in Primary Health Care10–12 and the insidious persistence of colonialism and epistemic injustice in global health,13–15 increasing attention is being paid to person- and community-centredness. Simply put, advancing equitable access to health services must be done *with*, rather than *to,* or *on behalf of* left-behind groups.9 In this study we aimed to engage with younger adults who had not been able to access eye care in Meru in order to explore their perceptions of how the local services could be modified to improve access. Working within a live programme, we aimed to deliver robust, non-tokenistic, and generalisable findings within a matter of weeks, with a view to testing suggested service modifications with a subsequent embedded RCT. ## Methods ### Setting Meru is a county with a population of 1.5 million in central Kenya, 110 miles north of Nairobi. It includes Mount Kenya and Meru National Park. The capital, Meru town, is home to a quarter of a million people. Agriculture is the main source of employment, with khat and tea representing important cash crops. Kenya’s Vision Impact Programme (‘VIP’) has been operating in Meru since July 2022, and has reached over 350,000 people to date, according to internal data. Teams of screeners go house-to-house testing all adults’ vision using a simple smartphone-based app developed by Peek Vision.16 Screeners refer people whose visual acuity falls below 6/12; those who have a red eye or another issue upon basic visual inspection; and anyone who feels they have an eye problem, even if there are no clinical signs and their visual acuity is >6/12. Our research team has been working with screeners to gather sociodemographic data from every person who screened positive and was referred to an outreach clinic for further assessment and treatment between April – July 2023. As stated above, we had previously found that younger adults are the least likely to be checked-in at treatment clinics but we did not know what the main barriers were or what could be done about them. ### Research paradigm, theory, and methodology We used a pragmatist philosophical paradigm17,18 and a phenomenological approach19,20 to explore these young adults’ lived experiences and perceptions of barriers to accessing eye clinics, and potential solutions. We grounded our work in the complementary frameworks developed by Levesque et al and Obrist et al.21,22 Both conceptualise access to care in terms of service and service-user characteristics. This distinction is helpful as our ultimate aim was to identify service modifications that improve accessibility for younger adults. We required mixed methods to answer a multi-layered question: what are the main barriers to accessing eye services in each location and what could be done about them? ### Methods overview This study was conducted in three stages. In Stage 1, we used interviews to generate a long-list of perceived barriers and potential solutions. Then, in Stage 2, to move from subjective experiences to generalisable service modifications, we conducted a telephone survey where a representative sample of younger adults who did not receive care ranked each of the suggested solutions by likely impact. Finally, in Stage 3, these ranked solutions were reviewed by a multistakeholder group who identified a package of interventions to test based on likely impact, feasibility, cost, and risks. ### Team composition and reflexivity This project was part of the broader ‘IM-SEEN’ programme of work that seeks to develop a new, rapid, robust, and responsive approach to continuously improving access to care, starting in the field of community-based eye screening programmes in Botswana, India, Kenya, Nepal. LSHTM-based researchers (LA, AB, MB, JR, DM & MK) working with Kenya’s Ministry of Health eye lead MG, AB and NB from Peek Vision - the screening programme software provider, and SK - the local research lead SK based at KEMRI, had already conducted a collaborative equity assessment of Meru’s VIP programme. LA – a mid-career British clinician, policy advisor, and mixed-methods public health researcher - led the development of the methodological approach to be used in all countries to engage with members of the left behind groups. LA worked closely with SK – a mid-career female Kenyan public health social scientist – to tailor the approach for Meru County, supported by the wider team. LA and SK recruited and trained six local, early-career data collectors (DG, EMM, EM, PK, BN and FG) to conduct the interviews and surveys. We were interested in understanding the barriers and solutions as perceived and described by affected people in their own words. SK and LA facilitated the multistakeholder workshop where findings were interpreted by lay representatives, other members of the left behind group, and local programme managers. This local multistakeholder group collectively made the final decisions on which suggested service modifications to take forward for implementation. ### Stage 1: interviews #### Recruitment and sample size Peek Vision – the programme software provider - provided us with a list of every person aged 18-44 years who had not been able to access their clinic appointment in Meru. In random order, we phoned people from this list to invite them to participate in the interviews, and sought recorded verbal informed consent. We tried each person three times before moving on to the next. We planned to use Guest and colleagues’ approach to determine our sample size based on thematic saturation, using a ‘base’ of 12 interviews followed by runs of two interviews with a 0% new information threshold.23 In other words, we aimed to continue recruiting interviewees until no new themes emerged after two interviews in a row, with a minimum sample size of 14 (‘12+2’ approach). #### Interview modality We wanted to use telephone interviews, based on empirical evidence that they can be completed faster at lower cost than in-person interviews, and with equivalent data richness.24–28 However, we were not entirely convinced that the data would be equivalent. As such, we decided to recruit two separate samples and use both modalities, conducting an embedded mode-comparison study29 that will be reported elsewhere. #### Data collection Three pairs of Kenyan data collectors with at least basic qualitative training and fluency in English, Kiswahili, and the local dialect conducted semi-structured interviews using the topic guide summarised in Box 1 (see Appendix 1 for the full script). For the telephone interviews, calls were made on speakerphone in a private space and recorded using the phone’s inbuilt call recording app. As one data collector conducted the interview, the other noted down the times at which each unique barrier and potential solution was mentioned. After the call, the interview recording was immediately replayed and the data collectors entered verbatim quotes directly from the audio into our analytic matrix. The same process was used for in-person interviews, but with an audio recorder instead of a mobile phone. Our decision to use direct-from-audio transcription was based on findings from a background systematic review that we conducted on rapid qualitative approaches.30 Interviewees did not review their transcribed quotes in the matrix. In-person interviews were conducted in private rooms in four different health facilities where interviewees’ responses could not be overheard. Only the data collectors and the interviewee were present for each interview. Box 1: #### Topic guide * Barriers * In your own words, can you talk me through why we didn’t see you at that clinic? * Probing questions * Are there any other factors that prevented you from attending? * Is there anything else you’d like to share? * Solutions * What would make the biggest difference in addressing these barriers? Probing questions * else would help? * other changes could we make to the programme that would make it easier for you to attend? * there any other specific changes that we could make to the way that the programme or eye clinics run? * mentioned [list their proposed solutions]. Some of these may be beyond our control, but if we managed to [list their proposed programme-related changes], do you think that would be enough? That’s the end of my questions. Is there anything else you would like to add? ## Data analysis We utilised an abductive analytic approach,19 whereby data collectors initially entered verbatim quotes relating to barriers and solutions into a deductive framework matrix, nesting each quote under one of ten broad *a priori* themes that had emerged from a literature review that is described in our protocol:31 * Costs * Distance and transport * Desire/priority to seek care * Clinical service quality * Facilities * Awareness & communication * Fear * Norms, values, health beliefs * Empowerment, support & capacity * Other (making room for surprising/unexpected themes) At daily debrief sessions, SK and LA reviewed the matrix with the data collectors and used inductive coding to identify unique barriers and solutions. The decision to use an analytic matrix and collective interpretation was based on the findings of our previous systematic review, which had found these techniques to be rapid and robust.30 Our matrix had one participant per column and one sub-theme per row – with a new row created every time a sub-theme (a unique barrier or solution) was identified. Each sub-theme (e.g. ‘loss of earnings’) was nested under the relevant theme (e.g. ‘costs’) The process of data entry is demonstrated in this short online video ([http://tinyurl.com/29asc6nm](http://tinyurl.com/29asc6nm)) and a blank matrix template is available here. We generated one matrix for the telephone interviews and another for the in-person interviews. This was so that we could compare the themes that emerged from each modality in our embedded study. For our main analysis, presented here, we pooled all barriers and solutions identified using both modalities. We trained the data collectors over two days and performed fourteen pilot telephone interviews before starting data collection. Videocall debriefing sessions were held at the end of each day. ### Additional analysis Our original equity analysis had also indicated that people with the highest incomes and those who owned a car or truck may have been less likely to attend that those reporting no vehicle ownership and lower incomes. We conducted an additional ten interviews with people who reported earnings in the highest income category to assess whether the barriers they reported differed from those reported by younger adults. We hypothesised that richer people did not access VIP services because they had sought private care after being identified with an eye need during eye screening. ### Output and screening We created a summary list of all of the unique solutions that had emerged from the interviews. Before taking these to a representative sample for ranking, we met with the implementing partner to identify any ideas that would be completely infeasible given the constraints of the programme e.g. providing helicopter transportation. Any interventions that were deemed to be completely infeasible were removed from the list. We asked the director of Peek Vision to independently review these decisions. ### Stage 2: telephone survey #### Survey instrument We used the vetted list of solutions to generate a simple telephone-based survey (Appendix 2) where respondents were asked to rank each suggestions from 1-3 on a Likert scale: 1. It would make a big difference - i.e. if we introduced this change then you or people like you would definitely attend, 2. It would make a moderate difference - i.e. it would greatly increase the chances, but it would not be enough to guarantee attendance by itself, 3. It would make a small difference - i.e. it might help a few people, but the impact is likely to be minimal. The telephone ranking survey was piloted with 26 people. #### Sampling and recruitment We used a 95% confidence interval, a 5% margin of error, and a conservative assumption that the total population size was 1 million people, rendering a minimum sample size of 384. We took the same list of 18–44-year-olds who had not been able to access care, and used random numbers to generate a call order, removing those who had already been included in the qualitative interviews. The same six data collectors tried calling each person three times before moving on to the next. #### Data collection and Analysis Data collectors obtained recorded informed verbal consent, and then read through the survey instrument using an online data entry form. Data collectors entered the respondents’ score for each proposed solution. We calculated the simple mean for each solution, and then ranked solutions by mean score. ### Stage 3: multistakeholder workshop Once we had this ranked list of solutions, we convened an online workshop with representatives from the programme implementer, programme funder, the county and national health ministry teams, and our community advisory board. We facilitated a discussion where each stakeholder shared their perceptions of the likely impact, feasibility, costs, and risks associated with each solution. As external public health and research ‘experts’, LA and SK restricted their contributions to presenting the ranked solution scores, facilitating the discussion, and providing information on the general strength of the international research evidence for each of the proposed solutions. At the end of the workshop, we asked the participants to collectively agree on one or more solution to implement in the VIP programme. Figure 1 provides an overview of our entire approach. ![Figure 1:](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/03/15/2024.03.13.24304156/F1.medium.gif) [Figure 1:](http://medrxiv.org/content/early/2024/03/15/2024.03.13.24304156/F1) Figure 1: Overview of the sequential mixed-methods approach This study was approved by KEMRI and LSHTM ethics committees. Those who attended in-person interviews were given a transport reimbursement of KES 500 (USD 3). We used the COREQ checklist to report our study (Appendix 3). ### Findings #### Interviews We made 143 phone calls to invite people to participate in in-person and telephone interviews. Three people declined; 29 did not pick up after three calls; and 34 people agreed but either were not home (13 people) or did not arrive at the agreed interview location (21 people) on the day of their in-person interview; six were not eligible as they told us they had actually received care (i.e. they had not been checked-in properly); and four had moved to a different part of the country. In total we conducted 36 telephone interviews and 31 face-to-face interviews over the course of eight days in September 2023. All our participants were aged 18-44 years old and 53.2% were male. We ended up performing more interviews than were needed to achieve thematic saturation with the 12+2 approach due to the efficiency of our data collectors. At the debrief on day two, they had already completed 24 telephone interviews. Our research leads had not assessed whether saturation had been reached by the time of the call, so – erring on the side of caution - they advised completing a further day of interviews. By the end of day three, 36 telephone interviews had been completed. A detailed retrospective saturation analysis, presented in Appendix 4, concluded that approximately 30 interviews were required to reach thematic saturation (Figure 1). We conducted 31 in-person interviews to enable fair comparison between telephone and in-person interviews for our embedded study. ![Figure 2:](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/03/15/2024.03.13.24304156/F2.medium.gif) [Figure 2:](http://medrxiv.org/content/early/2024/03/15/2024.03.13.24304156/F2) Figure 2: Accumulation of themes as the interviews progressed Supplementary Table 1 (Appendix 5) presents the 21 unique barriers that were identified along with all the quotes from both sets of interviews. Direct, indirect, and opportunity costs; long queues; difficulty getting time off work; and insufficient information about opening times and dates emerged as important themes. We also identified several meta-themes; participants were generally able to access the clinic locations but left after seeing long queues of several hundred people. Many felt they could not ‘waste time’ waiting to be seen, given the associated loss of potential earnings.“I choose to go to work to make money rather than spend my days’ time not knowing whether I will get attended to… If I don’t work, I don’t get money. MFK008, in-person Another important cross-cutting theme was the perceived lack of information about the clinics: where they were, days of operation, opening and closing times, and what services were available. Assumptions around (non-existent) costs and early closures also prevented some people from attending.“I also forgot the exact location where I was to go for the eye check-up and no one followed up to remind me of the place and date.” MT33, telephone > “They did not tell us if we need to come with money or not. Eye treatment, we are usually told to come with money. I assumed I’ll go there and they will ask me for money and I did not have it.” MFK204, in-person One interviewee also told us that the counselling he had received at the point of referral was inadequate. He wanted more information about what would happen at the clinic and on why attending was important, especially given that he did not even realise he had a problem:“They just told me that I have problems with my eyes and I should visit [town name] dispensary so I did not know what I was going to do there, is it surgery, is it being given medication, is it being tested again? And for me I have always known that my eyes are okay, and on that day they told me that they are not okay. They were very brief and I didn’t know what to expect, so that shock of being told that I have an eye problem which I have never had before is the reason why I did not go.” MFI03, in-person In terms of novel barriers, one person told us that they left the queue because they were “an introvert” and didn’t like the crowd (MT772, telephone); another felt their eye problem needed emergency treatment and sought care elsewhere (MFK02, in-person). One interviewee specifically named male health seeking behaviour as the main reason he didn’t attend:“As a man it is very hard to prioritize my health as I am manly focused on my family’s wellbeing and It is easy to forget my health needs” (MT250Z, telephone). Finally, one young man explained that being made to queue alongside women and children was shameful:“There were women on the line. They could have different lines for youths and women for some us to be comfortable because it is shameful to be on the same line with women and children, with worries how they will perceive me as young man. It was a challenge for me to just stand there with women… I had to go back that day without being attended even though right now my eyes have a problem. MT040, telephone The 25 proposed solutions to improve access centred around reducing the clinic queue lengths so that people could be seen quickly and then get back to work. Ideas included adding more clinics, holding them closer to villages or workplaces, increasing staff punctuality and speed, scheduling fewer people to attend each day, and extending the opening days and hours. The other meta-theme related to the provision of more detailed information around clinic services and opening times. Table 1 presents a summary of all 25 suggested solutions along with illustrative quotes. A full list all solution-related quotes is presented in Appendix 6. #### Reviewing feasibility As planned, we presented the list of all 25 suggestions to senior representatives from the implementing partner. We asked them to identify any suggestions that would be completely infeasible to deliver, given that they are responsible for funding all aspects of the programme. They felt that the programme budget would not permit additional payments for transport reimbursement or attendance incentives. Given that the outreach clinics involve multiple members of staff and large volumes of equipment and supplies, they also felt that it simply wouldn’t be feasible to deliver a door-to-door version of the outreach clinic. These suggestions were removed from the list. The director of Peek Vision agreed with each of these decisions. The remaining 21 suggestions were put to a representative sample of people from the left behind group in a ranking survey. View this table: [Table 1:](http://medrxiv.org/content/early/2024/03/15/2024.03.13.24304156/T1) Table 1: Solutions and implementing partner feasibility assessment #### Survey and selection workshop It took two days to train the data collectors and pilot the survey, and five days to complete ranking exercise. In total, data collectors called 440 people, of whom 401 completed the survey (response rate = 91%). All of the suggestions received a mean score between 2.4–2.9, indicating that all of the potential solutions were felt to offer at least a moderate chance of improving attendance. Table 2 presents the scores for individual-level solutions (that could be tested in an individually randomised RCT) and solutions that would be implemented at the clinic level (requiring testing with cluster RCTs). Our online workshop included representatives from the community advisory board, the screening programme implementing partner organisation, the programme funder, the national eye screening programme office, Kenya Society for the Blind, and the county health department. After our team had presented the survey findings, we facilitated a discussion around each of the proposed solutions in turn. The majority of the suggested clinic-level interventions required additional human resources or clinic locations. The programme funder and implementing partner recognised the issues around long queue times in some locations, but were very clear that inflation had already taken the programme over budget and there were no spare resources to introduce additional clinics or staff. The top-ranked suggestion related to frustration experienced by people who attended the treatment outreach clinic but were found to have a complex eye problem that required onward referral to the local hospital for specialist spectacles (where care is subsidised but not free). The workshop participants also recognised this problem, but agreed that it was not possible to have advanced ophthalmic care services present at each outreach clinic. The group suggested clarifying the process of tiered referrals during counselling. The workshop participants felt that organising separate queues for different ages and genders would be practically feasible, however imposing separation may cause problems for friends/family/colleagues who attend together. After reviewing all of the suggestions, the workshop participants unanimously agreed to implement and test the following bundle of interventions relating to counselling and the provision of enhanced information in SMS reminders: * – Send a reminder text on the appointment day and the day before * – Clarify what services are available at the outreach clinic and costs linked to these services * – Specify clinic opening and closing time at the point of referral * – Explain why attending clinic is important at the point of referral Overall, the ethics review process took four months. The interviews took nine days to complete, and the survey took seven days, including training and piloting. We held the multistakeholder workshop one month after the survey had concluded, with the delay driven by scheduling challenges. View this table: [Table 2:](http://medrxiv.org/content/early/2024/03/15/2024.03.13.24304156/T2) Table 2: Mean scores for each of the suggested solutions and workshop consensus The current verbal counselling script and SMS reminder messages that are used in the VIP programme are presented in Box 2. The SMS reminders are currently sent on the day of referral and the day before the clinic appointment. We drafted a new verbal counselling script and SMS reminder message that included the suggested new elements that were agreed in the workshop. We asked the workshop participants to review the new wording via email, as well as two representatives from the left-behind group. Based on this feedback we made three minor changes. The original script and description of these changes is presented in Appendix 7. The enhanced counselling and SMS reminder will be tested in an individual- level RCT in a subsequent study. Box 2: #### Original and new counselling and SMS reminder wording Usual care counselling, delivered verbally at the point of referral “I have examined your eyes, and you have a problem. I have referred you in the system and you will receive an SMS with where and when you are supposed to attend treatment. You will come for treatment on <> at <>. The examination will be free and you will be informed of anything else on the material day. Current SMS reminder, sent on the day of referral and day before the appointment Dear <>, you were examined and found to have an eye problem. Kindly report on on <> for assessment. For more information contact Meru Referral Hospital. Enhanced counselling script, based on interview, survey, and workshop feedback “I have found a problem with your eyes. I am referring you to the outreach treatment clinic that will be held at <> on <> between <