Developing feasible person-centred care alternatives to emergency department responses for adults with epilepsy: a discrete choice analysis mixed-methods study

Background Calls have been made for paramedics to have some form of care pathway that they could use to safely divert adults with epilepsy away from emergency departments and instigate ambulatory care improvements. Different configurations are possible. To know which to prioritise for implementation/evaluation, there is a need to determine which are acceptable to service users and likely National Health Service-feasible. Objective(s) (1) Identify configurations being considered, (2) understand service users’ views of them and current provision, (3) identify what sort of care service users want and (4) determine which configuration(s) is considered to achieve optimal balance in meeting users’ preference and being National Health Service-feasible. Design Service providers were surveyed to address objective 1. Interviews with service users addressed objective 2. Objective 3 was addressed by completing discrete choice experiments. These determined users’ care preferences for different seizure scenarios. Objective 4 was addressed by completing ‘knowledge exchange’ workshops. At these, stakeholders considered the findings on users’ stated preferences and judged different pathway configurations against Michie’s ‘acceptability, practicability, effectiveness, affordability, side-effects and equity’ feasibility criteria. Setting This project took place in England. The survey recruited representatives from neurology and neuroscience centres and from urgent and emergency care providers. For the interviews, recruitment occurred via third-sector support groups. Recruitment for discrete choice experiments occurred via the North West Ambulance Service NHS Trust and public advert. Workshop participants were recruited from neurology and neuroscience centres, urgent and emergency care providers, support groups and commissioning networks. Participants Seventy-two services completed the survey. Interviews were conducted with 25 adults with epilepsy (and 5 relatives) who had emergency service contact in the prior 12 months. Discrete choice experiments were completed by 427 adults with epilepsy (and 167 relatives) who had ambulance service contact in the prior 12 months. Workshops were completed with 27 stakeholders. Results The survey identified a range of pathway configurations. They differed in where they would take the patient and their potential to instigate ambulatory care improvements. Users had been rarely consulted in designing them. The discrete choice experiments found that users want a configuration of care markedly different to that offered. Across the seizure scenarios, users wanted their paramedic to have access to their medical records; for an epilepsy specialist (e.g. an epilepsy nurse, neurologist) to be available to advise; for their general practitioner to receive a report; for the incident to generate an appointment with an epilepsy specialist; for the care episode to last < 6 hours; and there was a pattern of preference to avoid conveyance to emergency departments and stay where they were. Stakeholders judged this configuration to be National Health Service-feasible within 5–10 years, with some elements being immediately deployable. Limitations The discrete choice experiment sample was broadly representative, but those reporting recent contact with an epilepsy specialist were over-represented. Conclusions Users state they want a configuration of care that is markedly different to current provision. The configuration they prefer was, with support and investment, judged to likely be National Health Service-feasible. The preferred configuration should now be developed and evaluated to determine its actual deliverability and efficacy. Study registration The study is registered as researchregistry4723. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Health and Social Care Delivery Research programme (NIHR award ref: 17/05/62) and is published in full in Health and Social Care Delivery Research; Vol. 12, No. 24. See the NIHR Funding and Awards website for further award information. Plain language summary Ambulances often attend to people with epilepsy. Most of them are taken to the hospital’s accident and emergency department. This typically has little benefit since most patients are already diagnosed and visit the accident and emergency department with non-emergency states. To change things, National Health Service organisations want an ‘alternative care pathway’ for paramedics to use. It could mean the person is not taken to the accident and emergency department but cared for elsewhere. Our project brought stakeholders together to develop an alternative care pathway that includes things important to patients and carers but is also National Health Service-feasible. Seventy National Health Service organisations first told us via a survey and a workshop which pathways they were considering and which might be feasible. Thirty people with epilepsy and their family members and friends were then interviewed. They explained what is wanted after a seizure and problems with current care. One problem was that going to the accident and emergency department does not lead to them getting a follow-up appointment with an epilepsy specialist to check their treatment is right. Using ‘discrete choice experiments’, around 430 people with epilepsy who recently contacted the ambulance service and 170 of their family and friends were asked to make a choice between alternative packages of care, to say which pathway they would prefer in different seizure situations. The results were clear. People wanted care different from what National Health Service organisations told us was available. The choice experiment showed everyone prefers pathways where paramedics have access to their medical records, an epilepsy specialist is available to advise the paramedic, the general practitioner gets a report and they get an appointment with an epilepsy specialist in the future. Everyone wants to avoid long episodes of care (6 hours) and after a typical seizure people with epilepsy want to stay at home. Three workshops were run with paramedics, epilepsy specialists and managers. They said the alternative care pathway wanted by users could be National Health Service-feasible. There is a need to implement and evaluate it now. Scientific summary Background Emergency department (ED) visits for epilepsy are common, costly, often clinically unnecessary and typically lead to little benefit for epilepsy management. An alternative care pathway (CP), which diverts people with epilepsy (PWE) away from ED when ‘999’ is called and leads to care elsewhere, could potentially generate savings and instigate improvements in the person’s ambulatory care. This project sought to identify the optimal configuration, ensuring the CP is both acceptable to service users and NHS-feasible. A key method used was discrete choice experiments (DCEs). They involved users being presented with hypothetical seizure scenarios and making a number of choices to say which CP configuration, described according to a fixed set of characteristics (attributes), they would prefer. The project’s aims were as follows. Aims and objectives Work Package 1 Systematically identify alternative CPs being considered by NHS services and describe extent to which service users have been involved in their design. Understand decision-making processes of PWE and their significant others (SOs) for seeking ED care and their concerns and expectations regarding alternative CPs. Identify attributes of postseizure care that PWE and their SOs (close family and friends) consider important. Work Package 2 Determine users’ and SOs’ preferences for postseizure care using DCEs and identify subgroup differences. Estimate utility of different CP configurations to service users and subgroup differences. Work Package 3 Hold expert workshops at which stakeholders, informed of the DCE evidence, would identify the optimal alternative CP/s for seizures that could be recommended for implementation and evaluation. Methods Work Package 1 Work Package 1a. Survey of service providers England’s ambulance services, epilepsy services and a random sample of its EDs were surveyed (April–June 2019). They were asked what alternative CPs they were considering and the extent to which they had consulted users. Work Package 1b. Interviews with service users Semistructured interviews were conducted in Southeast England with a purposive sample of PWE (April–September 2019). They also completed a ranking exercise to identify the characteristics (attributes) of postseizure care they considered most important. Participants were recruited from third-sector user groups and via a research newsletter. To be eligible, the person needed to be (1) aged ≥ 18, (2) have been diagnosed with epilepsy for ≥ 1 year, (3) had contacted the emergency services in the previous 12 months and (4) be able to provide informed consent and complete an interview in English. People with epilepsy were invited to be interviewed along with a SO. Work Package 1c. Knowledge exchange event Work Package (WP) 1b and 1c findings were used to select a set of candidate attributes that could be used to describe CPs within the DCE. To ensure attributes were features of alternative CPs and the levels likely NHS-feasible, a 1-day ‘knowledge exchange’ event was run (October 2019). Participants were representatives from ambulance and epilepsy services. They were identified for participation via WP1a. Work Package 2 Work Package 2a. Discrete choice experiment design panel and pilot A DCE design panel used evidence from WPs 1a–c to generate the DCE survey, using a D-efficient experimental design (Ngene 1.2.1, ChoiceMetrics, Sydney, Australia). The survey’s purpose was to obtain data of users’ care preferences for three separate scenarios, namely, a ‘typical seizure at home’, a ‘typical seizure in public’ and an ‘atypical seizure’. It, and vignettes describing the different seizure scenarios, was iteratively refined and finalised based on piloting. Work Package 2b. Formal discrete choice experiment A sample of n = 348 adults with epilepsy (and their SOs) from the target population was sought to complete the DCE. They filled in the DCE via an online platform. Participants were recruited via two routes between July and November 2020: (1) from the Merseyside area via the North West Ambulance Service and (2) nationally via public advert. PWE needed to (1) be aged ≥ 18 years, have a diagnosis of epilepsy, (2) have been seen by the ambulance service in prior 12 months and (3) be able to provide informed consent and independently complete the survey in English. Significant others needed to be aged ≥ 16. Data were analysed using a random-effects logit model. Preference weights were estimated for each attribute. Data from PWE and SOs were analysed separately, thus creating six seizure contexts for which preferences were modelled (three seizure scenarios × two participant types). Work Package 3 To identify which CP configuration(s) represented the optimal balance between user preference and NHS feasibility, three online ‘knowledge exchange’ workshops were run. Attendees were representatives from the ambulance service, from epilepsy services and commissioning. Service user representatives were also present. Commissioning and managerial representation was secured via national professional bodies. The DCE findings from WP2 were disseminated, and attendees were introduced to Michie et al.’s affordability, practicability, effectiveness, acceptability, side-effects and equity (APEASE) feasibility criteria. Nominal group techniques were then utilised, with stakeholders sharing views on the extent to which they considered the attribute levels favoured by users to be NHS-feasible. The groups then specified the levels for each attribute they believed represented the optimal balance. Results Work Package 1 Work Package 1a. Survey of service providers Clinical leads and managers from 72 (82.8%) of the invited services responded. Sixty per cent were considering or using an alternative CP change; only 21.2% had consulted service users on it. A range of CP configurations were reported. Some involved caring for PWE at home, ‘on scene’ or in some instances conveying them to an urgent treatment centre. The potential of configurations to improve ambulatory care varied. Work Package 1b. Semistructured interviews with service users Thirty service users (25 PWE and 5 SOs) were interviewed. The average age of PWE was 37 years; mean years was diagnosed 21. Participants provided insights into the challenges service users experience and disruptions they experience due to inadequate information transfer and co-ordination between ambulatory and urgent and emergency care services. Interviews highlighted some initial concerns users had with some possible CP configurations. The ranking exercise provided an approximation of the factors contributing to the ‘decision calculus’ users use when considering postseizure care options. Six attributes appeared most important: (1) whether the ambulance crew would have access to their medical records or care plan, (2) where the CP would take them, (3) how long it would take for them to be assessed, monitored and treated during the emergency episode, (4) extent to which the emergency care provider could request advice from an epilepsy specialist (e.g. epilepsy nurse, neurologist), (5) whether the CP would mean their usual care provider/s would be informed of the incident and, finally, (6) whether any follow-up from the epilepsy specialist would be instigated. Work Package 1c. Knowledge exchange event Thirteen representatives attended the workshop (n = 6 paramedics, n = 4 neurologists, n = 2 epilepsy nurse specialists and n = 1 user group representative). Participants agreed the six attributes identified by WP1b were all characteristics of an alternative CP, plausible and could be ‘traded’. Much of the event focused on identifying attribute levels that were considered feasible. Work Package 2 Work Package 2a. Discrete choice experiment design panel and pilot The panel decided that the six attributes identified in the formative work should be used to describe the CP options within the DCEs. Accordingly, DCEs for the three seizure scenarios were developed with the intention that participants would be randomised to each complete DCEs for two of three seizure scenarios, achieving a 1 : 1 : 1 allocation ratio. The six attributes and their associated levels (between 2 and 4) had the potential to generate 288 possible CP configurations. Use of an efficient experimental design meant that for each of the seizure scenarios a person would consider 24 CPs that were paired into 12 binary choices. The presentation of the DCE was finalised following pilot interviews. The final labels used for the attributes were as follows: (1) ‘The paramedic has access to medical records or a care plan’; (2) ‘What happens next’; (3) ‘Time’; (4) ‘Epilepsy specialists today’; (5) ‘GP told’; (6) ‘Additional contact with an epilepsy specialist’. Work Package 2b. Formal discrete choice experiment Useable submissions to the DCE survey were secured from n = 594 participants (n = 427 PWE; n = 167 SOs). Most (81.1%) were recruited via the public advertisement route. When participants from the two recruitment routes were merged to form a single sample, it was broadly representative of the target population. Two exceptions were its younger age and higher proportion reporting recent contact with an epilepsy specialist. The main ways that the SO participants knew someone with epilepsy was that they were the patient’s parents. The characteristics of the PWE known by SOs differed from those of the PWE taking part themselves (e.g. more learning disability). Across the three seizure scenarios, both PWE and SOs preferred CP configurations that differed from the way care is currently provided. Using coefficients from the random-effects logit models, the expected utility to service users for all possible CP configurations was calculated for the six seizure contexts. Care pathways were then ranked: rank 1 = most preferred; 288 = least preferred. The highest rank that the CP configuration approximating current care achieved was 220/288. There was similarity across the seizure contexts in terms of which attribute levels featured in the top-ranked configurations. In fact, three of the six contexts shared the same top-ranked configuration. For all six contexts, the highest-ranked CP configurations included (1) the paramedic having access to medical records or a care plan, (2) an epilepsy specialist (e.g. epilepsy nurse, neurologist) being available to advise the attending paramedic, (3) the patient’s GP being notified of the incident and (4) an appointment with an epilepsy specialist within 2–3 weeks being booked for the patient. The only attributes that varied within the top-ranked configurations were ‘What happens next’ and ‘Time’. For five of them, the top configuration involved the patient staying ‘where they were’. For the remaining ones – namely, for an ‘atypical seizure’ when considered by a SO – it involved the patient being conveyed to an urgent treatment centre. Different attributes had different levels of effect on preference. Providing crews with access to the medical records or care plan of the patient being attended to and having a specialist able to advise the paramedics had the strongest and consistent positive effect. Not all intended subgroup analyses could be completed due to insufficient cases. Those which could, showed social deprivation, recruitment route, presence of an intellectual impairment, contact with an epilepsy specialist in the prior 12 months, familiarity with the seizure scenario and whether the person with epilepsy had a care plan were not significantly related to preference. Work Package 3 Twenty-seven stakeholders attended a workshop. They included 10 ambulance clinicians, 8 epilepsy specialists, 5 commissioners and 4 patient and public involvement (PPI) representatives. Stakeholders believed the attribute levels preferred by service users were broadly feasible in the next 5–10 years. They identified those where most investment/work was required for them to be deliverable. Stakeholders were of the view that there was sufficient commonality in users’ preferences across the six seizure contexts for a single CP to be deployed. The groups were of the same position as to which configurations represented the optimum. They comprised of the following: (1) ambulance clinicians having access to medical records, (2) the person typically staying where they were, (3) the time taken being < 6 hours (be it 1, 2 or 3 hours), (4) crews being able to be advised by a specialist (an epilepsy nurse or neurologist) on the day, (5) for the person’s GP to be notified and, finally, for the incident to result in an appointment being made for the patient to have a follow-up appointment with an epilepsy specialist (be it within 1 week or 2–3 weeks). When attribute levels were restricted to these, the number of CP configurations for consideration reduced from 288 to 18. Evidence from WP2b showed these 18 included the configurations most preferred by service users for each scenario. Moreover, the evidence indicated that all 18 would be expected to hold more utility to users than the configuration representing current care. Conclusions By working collaboratively with service users from the target population, clinicians and commissioners, this project has provided clear answers to a pressing service delivery question. It provides evidence indicating that for common seizure scenarios, users appear open to paramedics not conveying them to ED and how poorly current care aligns with their preferences. Our project has also shown what postseizure care service users appear to want and that limited differences exist for different seizure scenarios. This appears to support the deployment of a single CP configuration. The study also provided evidence on the estimated improvement in utility that would result from individual attribute changes. This could be used to prioritise CP changes. Importantly, service providers and commissioners were found to be of the view that the CP configuration favoured by users could be NHS-feasible within 5–10 years, and they identified which attribute levels require the most work. It is important to now take the evidence this project has captured on people’s stated preferences and views and seek to implement one of the possible 18 CP configurations and evaluate it to determine its actual feasibility and efficacy in practice. Implications for NHS service commissioning, policy and practice The project identified a refined set of CP configurations that are considered optimal and the most promising candidates for formal evaluation. Pressures on acute services have intensified since our project was completed. This may incentivise ambulance services to use our findings and implement some form of alternative CP before an evaluation has occurred. Not all ambulance services would have the infrastructure and/or relationships with partner organisations to immediately implement the entire CP configuration considered optimal. Commissioners, policy-makers and service providers could support the development of the systems – such as shared record initiatives – to permit it to be offered. Identifying eligible people for the DCE survey via the NHS ambulance service was time-consuming compared to using public adverts. This could be made for efficient by embedding processes within NHS records systems to temporarily ‘strip’ them, including free-text fields, of confidential data. This would enable trained individuals not involved in a person’s care to screen anonymised records. Recommendations for research Using the attribute levels specified by stakeholders as representing the optimum, one of the possible 18 CP configurations should be evaluated for its efficacy. Any evaluation should consider short- (e.g. rates of recontact, death) and long-term outcomes (e.g. proportion of PWE ‘unknown’ to specialists brought to their attention). A cluster-randomised controlled trial would likely provide the most rigorous evidence. However, an alternative, faster evaluation approach may be needed to ensure evidence is generated in an acceptable time frame. Other ways to reduce unnecessary ED attendance warrant research attention. One is how to address ambulance crews’ requests for support in identifying persons suitable for non-conveyance. Attention is also needed on users’ seizure first aid confidence and knowledge. Study registration This study is registered as researchregistry4723. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Health and Social Care Delivery Research programme (NIHR award ref: 17/05/62) and is published in full in Health and Social Care Delivery Research; Vol. 12, No. 24. See the NIHR Funding and Awards website for further award information.