Introduction
The rising prevalence of diabetes over the past 30 years presents challenging health impacts and costs to individuals, health care systems and wider society. Prevalence rates in the UK rose from 3.2 million people in 2013 to 4.7 million in 2019, and they are expected to rise to 5.5 million by 2030 [
1,
2]. Prevalence rates are increasing more rapidly in low and middle income countries [
3]. Having diabetes can involve a number of related health issues, including diabetic retinopathy (DR). DR is a major cause of vision impairment among adults worldwide and is the second most important cause of visual loss in England and Wales [
4].
In the UK, the National Health Service (NHS) eye screening programmes have offered annual screening to all people with diabetes (PWD) over the age of 12 years for around 10 years. These programmes aim to detect sight threatening diabetic retinopathy (STDR) before it affects a person’s sight and when timely, effective treatment can be provided. Evidence suggests that it may be safe to screen low-risk people at longer intervals [
5‐
11] and the interval has been extended in some countries [
12,
13]. However, this evidence is not conclusive and is based largely on modelling rather than experimental research. In those countries, such as the Netherlands, Iceland, and the city of Hong Kong, with extended intervals the population being covered is significantly different to the UK. The shift towards varying screening intervals is not restricted to DR. For breast cancer there are moves to identify risk-stratified screening strategies to lower the rates of over diagnosis and to prevent deaths [
14]. Such directions illustrate a general move within medicine to personalised health care and potentially to re-allocate resources to those most in need; in the case of DR screening focusing on non-attenders. Risk estimating equations have been developed to allow this personalisation in DR [
15‐
17] and in other specialties [
14,
18]. Nevertheless, there has been little work on the impact on PWD of changing eye screening intervals and concern amongst HCP about safety including reduced attendance and loss of diabetes control [
9].
An intervention, such as changing eye screening intervals, can be considered to be implementing evidence-based practice. The aims of an intervention are to promote the uptake and optimal use of effective clinical services, along with modifications to health-related behaviour. It can be anticipated that there may be negative as well as positive outcomes from an intervention, therefore effective development and implementation is essential. Understanding enablers and barriers to change and then putting in place effective strategies to encourage or mitigate against their effect is crucial. Models of behaviour change can be a useful theoretical lens to explore behaviour and how to effect positive change. Such models have been used extensively within clinical and public health arenas to understand illness and health-seeking behaviours [
19‐
21]. There have been moves away from a deficit model, where primarily patients are perceived as lacking in their understanding and simply needing “more education” about their condition to resolve any issues. The Behaviour Change Wheel (BCW) is cognisant of the many components involved in changing health related behaviours, it recognises that the sources for behaviour can be found within three areas and usefully applied to changing eye screening intervals: capability (is the individual able to attend eye screening?); opportunity (does the eye screening service make it as easy as possible to attend an appointment?); and motivation (can an individual manage any changes to their eye screening appointment?) [
22]. The BCW approach also offers screening service commissioners and providers a range of interventions and policy approaches to align with PWD and HCP capability, opportunity and motivation to change eye screening intervals. The BCW has been successfully used in a number of other clinical arenas [
23,
24].
As part of the ISDR randomised controlled trial (RCT) [
25], we undertook a qualitative study with PWD and HCP designed to investigate and uncover enablers of and barriers to behaviour change of moving from annual to personalised risk-based variable-interval screening and to gain wider insights into perceptions amongst PWD and HCP. Our aims were to develop detailed understandings of the acceptability and enablers for, successful implementation of personalised screening in England and other countries with similar systems. We followed Medical Research Council (MRC) guidance on developing and evaluating complex interventions [
26].
Methods
Setting
A programme of applied research developed an enhancement to screening for STDR by introducing and testing an individualised or personalised approach based on measured patient-centred risk. A novel intervention was developed comprising variable-interval screening determined by a risk calculation engine (RCE) informed by real-time demographic, retinal and clinical data from the individual, referenced to local historical data. Intervals were allocated at 6, 12 or 24 months for high, medium or low risk respectively and recalculated at each screening appointment. A RCT was designed to compare the efficacy and cost-effectiveness of this individualised approach to standard fixed interval screening [
25]. A public involvement (PI) group was embedded in the programme. This setting allowed, for the first time, the real rather than theoretical investigation of the enablers and barriers around implementation of varying intervals, and the use of a risk calculator, in a population from an established screening programme and in a geographical location where annual fixed interval screening is already established.
Design
Semi-structured interviews [
27,
28] were conducted to gather views on risk-based variable-interval screening. Interviews with PWD were conducted prior to its implementation (Phase 1, baseline) in the setting of the parallel RCT and subsequently with a second group (Phase 2) during the RCT. All interviews with HCP took place prior to implementation.
The research team and the PPI group created interview topic guides. With PWD these covered understanding about diabetes, self-management, health services contact, responsibility for monitoring diabetes, links between diabetes and eye health and screening intervals. With HCP the guides focussed on diabetes services, current eye screening, future changes to eye screening and DNA rates. All participants received a brief overview of the individualised risk-based variable-interval screening intervention. Most patient interviews were conducted in participants’ homes, though some chose to complete them in the researcher’s university office, and one participant completed their interview in their own work office. HCP were interviewed at their place of work. Interviews lasted between 30 to 90 min, with most lasting around 45 min.
Participants
PWD aged over 16 years attending the eye screening programme were identified in two General Practices in a city in the North West of England. Suitability for participation was confirmed by a General Practitioner prior to a letter of invitation and patient information pack with reply slip being posted out. PWD who were interested in participating returned the reply slip or contacted the research team to make arrangements for an interview.
Sixty PWD were recruited, 30 to phase 1 (baseline) and 30 to phase 2 (post implementation of risk-based variable-interval screening). Thirty-four of the 60 were men and 8 had type 1 diabetes. The age range was 19–83. Times since diagnosis were: 1–5 years n = 10, 6–10 years n = 7, 11–15 years n = 4 years, > 15 years n = 9; range 1–40 years. For phase 2, participant allocation to risk based screening intervals was: 6 months n = 4, 12 months n = 5, 24 months n = 21.
PWD participants reported a range of social situations; the occupations described by the sample were very mixed, including a range of professionals as well as students and manual workers, retired and unemployed as well as one person who was unable to work due to long-term ill health.
HCP were identified by personal and professional local networks of the research team. To help with time commitments HCP could participate in individual interviews, in groups, or by completing an open-ended questionnaire to be returned by email. Interviews were conducted at the HCP’ place of work in a private office (one joint interview was completed at a participant’s home). Most interviews lasted around 40 min. Six of the HCP participants were interviewed as three colleague pairs.
Twenty-one HCP were recruited. Sixteen participated in interviews whilst five elected to complete questionnaires via email. Professional roles were: screener/grader n = 7, consultant ophthalmologist (retina specialist) n = 5, eye screening service manager n = 4, optometrist providing DR screening n = 2, public health specialist n = 2, general practitioner n = 1.
Analysis
Interviews were audio-recorded and transcribed verbatim to enable detailed analysis. Semi-structured case summaries were produced by a researcher upon completion of each interview to provide a summary of key themes to enable identification of emerging themes to inform further data collection and analysis. After reading and re-reading the transcripts, data were analysed to identify sections of text that informed understandings of the issues [
29‐
31]. Each concept was assigned a descriptive or analytical code, which was then combined into conceptual categories and broader themes. Each of the datasets were coded using NViVO software, which enabled searching and retrieval of specific data.
When key themes were identified, a number of charts were created, based upon the thematic analysis, to enable comparison of the datasets, taking a framework approach [
32,
33]. Data from individual participants were entered into cells within the chart to enable comparison on each theme at the level of individual participant and whether they were a PWD or an HCP.
PI
Our PI group contributed to the research questions and topic guides, informed analyses and further exploration, and one member joined the authorship.
Ethics
The two phases of the study were conducted under the following research ethics committee approvals: 13/NW/0287 and 16/NW/0061.
Discussion
Our findings show general support by PWD and HCP for the introduction of risk-based variable-interval screening. Key factors for our participants were the increasing prevalence of diabetes and finite resources for healthcare. Support was more clearly expressed by HCP, likely related to a better understanding of the aims and current pressures in screening. Our findings are reassuring for policy makers and service providers who are considering introducing variable interval screening, either stratified based on retinal grading or variable intervals based on risk estimation. Our findings also have relevance to other screening programmes with fixed intervals and help to mitigate in part the concerns raised recently in cancer screening [
34]. Against these generally supportive findings are a number of important concerns attached to the processes around risk-based allocation and to the screening intervals themselves, clearly expressed by both user groups. These need to be comprehensively addressed to ensure successful implementation, where success can be considered to be early detection of DR, attendance at eye screening and increased understanding of the relationships between diabetes, self-care, and eye screening.
Any changes in health services provision can be problematic for individuals and organisations and can be amplified by routinised practice and behaviours, systemic factors, and local discrepancies in service provision and practice [
35‐
37]. Implementing evidence based practice, such as risk-based variable-interval eye screening, is challenging and needs to address societal, political, cultural, individual and organisational barriers. To take into account all of these variables and their interaction requires a sophisticated theoretical model. Implementation science [
38] is emerging as a broad theoretical umbrella with a range of frameworks and strategies, including the BCW. As already mentioned, the BCW identifies three sources of behaviour: capability; opportunity; and motivation. It can then map each of these to appropriate intervention factors for each (and any interactions between them), with a final layer of policy categories to firmly support an intervention.
For example, in our study, both groups of participants expressed concerns about the potential safety of the RCE linking to it being a more complex system, and uncertainty around the reliability of the source data. The BCW allows us to put these concerns into the capability area, and to allay anxieties, there can be a mixture of interventions and policy work. An intervention can be in the form of making transparent the failsafe mechanisms for the RCE to HCP, along with guidelines and policy by the screening programme/Public Health England to ensure that data from primary care practices are embedded with their systems. The use of champions, experts, who are respected in their field, to present and discuss the risk engine to other professionals, where regular updates on the stability and accuracy of variable screening can be fed back to clinicians, can also be a powerful intervention. As we have seen from elsewhere, lessons learnt from the lengthening of screen intervals in other screening programmes, such as breast screening in 2018 [
34,
39] show that to address these concerns will require careful explanation of the RCE and transparency about the systems involved. There also had to be opportunities for both HCP and PWD to refer or self-refer back into the eye screening programme, if for example for the latter group, there have been changes in the diabetes severity.
The introduction of the shorter interval for those at greatest risk was widely welcomed by PWD and to a greater extent HCP. PWD perceived the interval as reassuring, while recognising that it meant a higher risk of developing sight threatening disease for them personally. For the former group of PWD, they can be allocated to the motivational area of the BCW, where their emotions around reassurance will probably lead to them attending 6 monthly screening. However, for those PWD who saw a 6 month appointment as high risk, then interventions to encourage attendance would include increased training for HCP on how to approach and engage PWD around monitoring and preventive strategies to lessen the chances of developing STDR. Aligned policy approaches would provide increased service provision for these identified PWD.
The 12 month interval was perceived as standard care, but there were misunderstandings about its purpose allied to an important theme of needing to disentangle health and illness beliefs in PWD around diabetes and eyes, and to develop a clearer understanding of the aims of screening. There continues to be conflation and misconstruction of these concepts along with their interconnectedness and purpose, as previously reported [
40‐
47]. The BCW can usefully place such health beliefs (and potential health behaviours), within all three of its components; capability, opportunity and motivational issues. For example, a PWD who has negative emotions about having diabetes (motivational), will need support (interventions) from HCP, family and friends, perhaps in the form of co-produced information, self-help groups and online forums to increase awareness and confidence in their self-care. In addition, policy changes will be needed to support increased service provision for PWD, not just in primary care but also within communities, such as roadshows. Furthermore, changes to the eye screening services could potentially fracture the trust that PWD have in the current system and in HCP, and by default impact upon diabetes self-care. Thus, introduction of any changes requires thought and consideration, further complicated by the annual screening programme having been in place for over a decade. Old habits die hard, implementing large changes to embedded clinical practice, beliefs and understanding comes with significant challenges and any modifications will require effort by the diabetes care team and the screening programme to explain and consolidate pertinent information.
Potentially, the move to the 24 month interval was controversial for PWD with mixed responses from reassurance to rejection. Most HCP were supportive. Both groups expressed apprehension about the potential impact on care of diabetes by PWD. The BCW suggests that behaviours related to opportunity, such as feeling rejection, is outside of an individual’s control. To allay and soothe such high emotions, which could well lead to non-attendance at screening, can be managed by persuasive relationships (interventions) between a HCP and PWD, where clear explanations are provided that are tailored to the PWD. These conversations can also be backed up with good communication and marketing policy by Public Health England.
Our qualitative study ran in parallel with the ISDR RCT which has shown that risk-based variable-interval screening is safe and effective when compared to annual screening (manuscript under review). Cost-effectiveness is greatly improved allowing resources to be reallocated, including for hard to reach and vulnerable groups. 22% of people declined to participate in the RCT study explicitly stating that they wished to remain in annual screening or did not want a change of interval, this is further evidence of a level of resistance to changing intervals. However dropout in the 2265 PWD randomised to variable interval screening was very low.
Strengths and limitations
Our PWD participants represented a broad range of individuals with diabetes giving a variety of views from their experiences. Our HCP informants were multidisciplinary and with a range of ages and experience. This affords a breadth of experiences and views, enhanced by our purposive sampling approach and co-produced topic guide. The study was informed by our PI group who commented on all aspects of the research process and were able to inform the content of the topic guides based on their own experiences of diabetes and eye screening.
Our participants may have been motivated to take part in interviews as a way of expressing their views on particular topics related to diabetes care, or gaining social contact. PWD participants in Phase 1 were not the same individuals as in Phase 2.
Conclusions
Extending intervals and introducing a fully personalised approach is gathering momentum in screening for diabetic retinopathy, and in other areas of disease prevention in the UK.
Our qualitative work sheds new light on the issues around implementing risk-based, variable-interval, and stratified screening using the BCW, with PWD and HCP generally supportive. For successful implementation, a range of issues must be addressed: interpretable and clear safeguards for individual PWD are required against increasing non-attendance, loss of diabetes control and system failures; alternative referral pathways are needed for those lost to follow-up or whose risk factors change substantially over longer intervals; and, for risk calculation systems, reliable monitoring and clear communication is necessary. Utilising the frameworks for changing health services provision in the BCW is likely to improve implementation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.