Visual identifier systems for patients with cognitive impairment in healthcare settings: A survey of practice in UK hospitals

Abstract Background People with dementia and other forms of cognitive impairment form a substantial proportion of patients admitted to hospitals, but problems in their care are persistent. One widely proposed improvement approach involves the use of systems using visual identifiers to help staff quickly recognise people with suspected dementia, with the goal of supporting more personalised care. The aim of this paper is to identify the identifier systems in use and staff perceptions of their strengths and weaknesses. Methods We undertook an online survey of staff providing care for people with dementia in acute and mental health hospitals across the United Kingdom. The questionnaire covered different types of visual identifier systems for dementia. It used categorical and open‐response questions to access staff views of their use in practice. Responses were analysed using descriptive statistics, and the Framework approach for free‐text answers. Results 162 responses were received from staff in at least 48 hospitals. Of these, 128 had direct experience of using visual identifier systems. They reported that multiple identifier systems are in use, including schemes with national scope and locally developed approaches. Most respondents reported that more than one system is in use in their hospital. Different types of identifier were seen to have different strengths and weaknesses. Respondents had a broadly positive view of identifiers, but highlighted risks including lack of reliable and consistent use (linked to competing pressures on staff time), lack of staff training, uncertainty about patient and family views, and unclear consent processes. Conclusions Our study suggests that a wide range of identifier systems is in use in UK hospitals, with many hospitals using more than one. Further consideration should be given to ensuring that multiple perspectives—including those of patients and carers—are drawn on in optimising their design, resolving ethical issues and supporting implementation.


| INTRODUC TI ON
A large proportion of hospital inpatients are affected by cognitive impairment, including both episodic forms (e.g. delirium) and more long-term impairment arising from progressive neurodegenerative disease (e.g. dementia) (Mukadam & Sampson, 2011). Worldwide, estimates of the prevalence of dementia in acute hospital populations range from 15 to 42 per cent (Bray et al., 2015;Jackson et al., 2017).
To avoid harm and distress, it is important that the distinctive needs of this group-for example relating to nutrition, hydration and toilet use, and understanding and consenting to treatment-are met (Jackson et al., 2017;Røsvik & Rokstad, 2020;Scerri et al., 2020;Sillner et al., 2019). However, it may not always be straightforward for those caring for people in hospital settings to recognise readily which patients are living with cognitive impairment (Hermann et al., 2015;Sillner et al., 2019). The systems are various and include stickers, wristbands and notices placed at the bedside or on electronic whiteboards on wards.
The purpose of such identifiers is to serve as a visual reminder to staff that a patient may have additional needs, perhaps unrelated to the reason for hospitalisation. Some act as standalone identifiers of cognitive impairment, while others-such as the Butterfly Scheme and the Dementia Friendly approach (Williams, 2011) National Audit asks only whether a system is in place, and so little is known of which systems are used most frequently across hospitals or how well they work. None of the existing schemes has been subject to full evaluation. Variation in approaches to identification and care of people with dementia or other forms of cognitive impairment may introduce its own risks and burdens, particularly, though not only, when staff and patients move between healthcare organisations (Dixon-Woods, 2019).
A further challenge is the risks that may arise from widespread use of 'technologies of attention', such as visual indicators, particularly in pressured environments . One possible risk is that visual indicators might ironically result in less personalised care, acting as a shorthand for a set of needs to be addressed by routinized responses rather than the preferences and priorities of an individual patient . Use of visual identifiers for confirmed or suspected diagnoses of dementia might also present ethical and legal challenges, for example in relation to obtaining consent from the patient or their advocate, or inadvertent disclosure of personal information to third parties (Brigden et al., 2020). While guidance accompanying some schemes addresses issues of consent and confidentiality (Williams, 2011), there is currently little evidence regarding how such issues are approached in practice.

Summary statement of implications for practice
What does this research add to existing knowledge in gerontology?
• Multiple visual identification systems are used for hospitalised people with dementia, with many hospitals using combinations of two or more systems.
• Staff views of visual identifiers for patients with cognitive impairment are broadly positive, but they also highlight risks such as inconsistency, lack of staff training and unclear consent processes.
What are the implications of this new knowledge for nursing care with older people?
• Good practice in using visual identifiers for patients with cognitive impairment includes considering the wishes of patients and their carers and a clear consent process.
• Nurses and others involved in care of people with dementia should be aware of the different strengths and weaknesses of different systems, and of potential challenges in using multiple systems together.
How could the findings be used to influence policy or practice or research or education?
• Policy and practice around visual identification for patients with cognitive impairment should focus on development of a sound evidence base that addresses effectiveness, cost-effectiveness, training, patient and carer engagement, and ethics.
• An important focus for improvement will be on increasing the interoperability of different identifier systems, but further research into the advantages and disadvantages of standardisation would be of benefit.

K E Y W O R D S
cognitive impairment, dementia, hospital care, identification, patient care, person-centred care, survey

| Aim of the study
The aim of this study was to conduct a survey to map current use of identification systems for patients with confirmed or suspected dementia in organisations in the UK National Health Service (NHS) and to obtain staff views on the systems in use, including their advantages, disadvantages, benefits and risks. The survey was part of a wider study examining the use and design of visual identification systems for people with dementia and other forms of cognitive impairment in institutional settings.

| MATERIAL S AND ME THODS
We developed an online questionnaire in consultation with an expert collaborative group of 20 people, including staff, patient and carer representatives, individuals who had led the development of existing identification and dementia care systems, clinical and nonclinical academic experts in related fields, third-sector organisations, and collaborators in the wider study. We surveyed staff in acute and mental health hospitals in the UK NHS working in areas where adults with confirmed or suspected dementia are assessed or treated. nine existing systems that we had identified as being in use nationally through a scoping review, the expert collaborative group and web searches. The questionnaire also invited participants to provide details of other systems, including locally developed ones, that might be in use. In line with established questionnaire-design principles (Abramson & Abramson, 2008;Streiner et al., 2014), it was user-tested by members of the expert collaborative group and other volunteers to ensure clarity of wording and consistency of understanding, but it was not formally piloted. The protocol and questionnaire were externally peer-reviewed and were given ethical approval by the University of Cambridge Psychology Ethics Committee on 6 April 2020. Approval to conduct the research in NHS organisations was provided by the Health Research Authority on 27 July 2020.
The questionnaire was administered using Qualtrics survey software hosted on the Thiscovery (www.thisc overy.org) platform developed by The Healthcare Improvement Studies Institute at the University of Cambridge, where the participants had to register to take part in the study. Given the wide range of clinical and non-clinical professionals involved in care provision, we sought to include staff of any grade and any occupational group working an NHS hospital with direct involvement in caring for inpatients with dementia. Exclusion criteria covered those who: were not NHS staff, did not have first-hand experience in caring for adult inpatients with dementia, were under 18, or were unable or unwilling to consent to register with Thiscovery or to participate in the study. All participants were provided with written information about the study and assured that responses were confidential and that neither they nor their employing organisation would be named in outputs. They were asked to confirm their eligibility and give consent to participate at the point of accessing the questionnaire.
Due to the exploratory nature of the study, uncertainty over the size of the population of interest and lack of a reliable sampling frame, we did not specify a target sample size. Our approach to recruitment was instead to seek to secure a diverse sample that might be broadly representative of practice in different types of organisation nationally. Participants were recruited using four parallel approaches. First, research and development directorates of participating NHS organisations were provided with publicity materials, including a poster and a letter of invitation, which they were asked to distribute to key individuals involved in the delivery of dementia care and other relevant areas. Second, these materials were also circulated through the Dementias and Neurodegeneration specialty group of the Clinical Research Network in England, which has strong reach to clinicians in relevant areas with an interest in contributing towards research. Third, details of the study were advertised on social media. Fourth, participants were asked to pass on details of the study to potentially interested colleagues (an approach sometimes referred to as 'snowball sampling').
We used descriptive and analytical statistical approaches to analyse responses to categorical questions, and thematic qualitative analysis of responses to open-ended questions, as set out prospectively in our protocol. We produced summary descriptive statistics relating to profile of respondents; frequency with which use of different systems was reported by participants; processes for deciding to use an identifier for a particular patient; and advantages and disadvantages reported. We used Microsoft Excel to support our quantitative and qualitative analyses.
We planned to use analytical statistics to compare respondents' views of the nine different systems we had pre-identified. These systems fell broadly into four different types (systems using stickers placed near the patient's bed and/or on patient notes; those using wristbands worn by the patient; those using electronic whiteboards and similar technologies away from the patient; and those using summary information kept at the patient's bedside). Since the number of responses received for some individual systems was low, we used analytical statistics (chi-squared tests, with a predefined significance threshold of p < .05) only to compare types of system.
On the advice of the expert collaborative group, though not specified in the protocol, we additionally compared respondents' views on the systems pre-identified in the questionnaire with their views on locally developed schemes that were similar but not identical to those systems. Answers to open-ended questions were analysed thematically using methods based on the Framework approach (Ritchie & Spencer, 1994). One researcher read and re-read openended responses, and developed a coding frame that combined codes derived from the questions with codes generated inductively from the responses. Responses were coded accordingly, and coded summaries reviewed alongside the quantitative results to identify key cross-cutting themes and complementary, contradictory or mutually explanatory findings.

| RE SULTS
The survey ran from October 2020 to January 2021, receiving 183 responses, of which 162 met the eligibility criteria. Respondents were asked to optionally identify their employing organisation. Thirty-one did not give a response to this question; the remaining 131 participants gave 48 different employing NHS organisations. Eighteen participants came from one NHS organisation; other organisations provided between one and eight participants each (mode 1; median 3). Respondents came from a range of occupational backgrounds (Table 1). The biggest single group was nursing staff (n = 59/162, 36%), likely reflecting the predominant responsibility for day-to-day care of patients with dementia and other forms of cognitive impairment, including suspected dementia. Allied health professionals (a group that includes physiotherapists, occupational therapists and other professional groups such as podiatrists) (n = 28, 17%) and doctors (n = 18, 11%) were also reasonably well represented among the respondents.
There was a very limited response from non-clinical staff.
Of the 162 responses included in the final dataset, 34 (21%) stated that their hospital did not use visual identifiers, and were accordingly routed to a truncated version of the questionnaire.
Questions regarding the systems used, the approach to administering them and their strengths and weaknesses were addressed to the remaining 128 participants, though response rates to individual questions varied and typically fell slightly short of this figure.

| Systems used and decisions about use
We asked participants which system or systems were used in their hospitals, including established national schemes and locally developed schemes. There were four broad types of system: systems using stickers placed near the patient's bed and/or on patient notes; those using wristbands worn by the patient; those using electronic whiteboards and similar technologies away from the patient; and those using summary information kept at the patient's bedside (Table 2).
The most commonly used systems reported by respondents were national ones. 'This is me' booklets were most frequently reported (n = 59/121, 49%). This system records brief personal details of patients to facilitate more person-centred care for those who may struggle to communicate their preferences themselves, for example due to patient's diagnosis of dementia; the booklet's cover serves as a discreet reminder of the patient's diagnosis. The Dementia Friendly scheme was used by 40% of respondents (n = 48/121). It involves a flower symbol to identify people affected by dementia. Electronic whiteboards that included information on cognitive impairment were reported by 39% (n = 47/121). The Forget-me-not Scheme, a wider system of care and training approach that includes use of a sticker featuring a blue flower placed by the patient's bed and/or on patient notes, was reported by 31% of respondents (n = 38/121).
Participants also reported using systems that were similar but not identical to national schemes, most notably local systems that were similar to the Forget-me-not Scheme (n = 21/121, 17%) ( Figure 1). Free-text responses confirmed that the most common local variation was an alternative image of a forget-me-not flower.

TA B L E 1 Responses to the question 'Which option best describes your role in the hospital?'
More than one system was in use in the majority of organisations. Over a third (36%) of respondents reported that more than one sticker-based system was in place in their hospital; 10 respondents reported three systems; one respondent reported that four different sticker-based systems co-existed in their hospital. It was also common for different types of systems to be used together. For example, 66 respondents (52%) reported that their hospitals used at least one sticker-based system alongside at least one summary bedside information system; 51 respondents (40%) reported use of sticker-based and whiteboard-based systems. Fourteen respondents (11%) reported that at least one system of each of the four types-stickers, wristbands, whiteboards and summary information sheets-was in use in their hospital.
By definition, visual identifiers communicate personal information about individuals to others, including but potentially not limited to staff involved in the individual's care. Ethical concerns, most notably in relation to consent to disclosure, may therefore arise, further complicated by the (potentially fluctuating) capacity of individuals with cognitive impairment to give consent to the user of identifiers (Brigden et al., 2020). We asked respondents who are involved in the decisions to use an identifier for individual patients, including the patient, advocate (i.e. informal [family] carer) and healthcare professional (Table 3; Figure 2). Half of respondents (n = 64/127, 50%) indicated that the decision is taken by a healthcare professional; in approximately a third of cases (n = 44/127, 35%), this was apparently without input from patients or informal carers.

| Strengths, weaknesses, benefits and risks of identification
For each system they reported using, respondents were asked to indicate the extent of their agreement with nine propositions covering issues such as ease of use, impact on quality of care, and views of staff, patients and carers. I work in the elderly medicine admissions ward.
Patients change quickly and these visual cues help when the patient asks the staff for the date, the time or help. Often I will be in the room with a patient and another one will ask me to help them because they feel lost. Knowing that they have dementia helps me understand the underlying cause and I can reorient them and reassure them better, as well as providing care in a way that will help reduce their distress and promote their independence. (Nurse) Challenges of ensuring consistent use of indicators were reported by 14 respondents. Stickers seemed to be well understood, but were prone to 'coming unstuck' from the individual to whom they were intended to relate. Bedside information documents were seen as containing important information that could inform personcentred care, but were apt to be ignored when staff were busy.
Respondents identified situations where stickers placed above the bed were not removed after discharge, creating the risk that staff would assume that the next patient in that bed also had a dementia diagnosis.
If the previous occupant had been moved but the identifier was still over the bed when the next patient arrived, there may be confusion for staff who take a different 'tone'with the new patient.
(Administrative staff) Inconsistency was seen to reduce the usefulness of systems, as staff were left less confident that the presence of an identifier accurately reflected a patient's diagnosis or care needs. Equally, they were wary of assuming that patients without identifiers did not have additional needs.
The identifier is not consistently used in ED; this could lead to staff assuming a patient does not have dementia and therefore not treating them accordingly.
(Allied health professional) Reflecting the quantitative finding that staff were less certain of patients' views, some open-text responses indicated concerns about the acceptability of identifiers for patients or the possible anxieties that use of indicators may cause for patients and carers.
When patients recognise fellow patients do not have the same-coloured wrist band and asking why.

| Strengths and weaknesses of different approaches to identification
We compared respondents' views on the strengths and weaknesses of established national schemes with their views on locally developed schemes ( We also compared respondents' views on the strengths and weaknesses of the four main types of system (stickers placed near the patient's bed and/or on patient notes; wristbands worn by the patient; electronic whiteboards and similar technologies; and summary bedside information documents) ( 'It is easy to find the shape cutter when needed.' For whiteboards, it was phrased: 'Our electronic board has a special option for patients with dementia'.
A well completed 'This is Me' can enable staff to discover likes and dislikes and this can improve many aspects of patient care, particularly as it focusses on the basics.
However, some respondents also noted that completing these documents could be labour-intensive, and that staff did not always have the time to read them and use them to inform their practice on busier days. Third, there was variation in participants' views of the reliability and consistency of use of different types of identifier. There were significant differences in responses about supply issues by type of identifier (p = .010), with fewer participants indicating problems with whiteboard-based systems than others (though this may be an artefact of the phrasing of the question for different types of identifiersee Table 6). Participants also indicated that electronic whiteboards were used more consistently for identifying patients with a diagnosis of dementia than other interventions, with 27/50 (54%) agreeing or somewhat agreeing that whiteboards were used consistently, compared with 81/214 (38%) agreeing or somewhat agreeing with the same proposition for other interventions ( Figure 6); however, the difference between responses to this question did not reach statistical significance (p = .103).
Free-text responses also highlighted the importance of ensuring that use of visual identifiers was complemented by staff training.  Gwernan-Jones et al., 2020), and they therefore warrant scrutiny. This survey provides some of the first insights into how the national recommendation to introduce such systems has been operationalised, identifying a range of schemes and systems in use and characterising staff perceptions of their strengths and weaknesses. Though staff views of visual identifiers for a diagnosis of dementia and other forms of cognitive impairment were broadly positive, they also highlight risks such as inconsistency, lack of staff training, uncertainty about patient and family views, and unclear consent processes. Participants' responses also make explicit some of the challenges, including those relating to the wider institutional determinants of quality of care.
One challenge relates to the need for a supportive infrastructure in high-quality care for people with cognitive impairment (Murray et al., 2019), rather than seeing identifiers as a standalone intervention. This need is emphasised in resources used to support some of the national schemes, such as the Butterfly Scheme and Dementia Friendly. There is some evidence that identifiers deployed as part of a wider approach to dementia care, incorporating training and development for staff, can contribute to improvements in some aspects of care quality (Beattie et al., 2021;Murray et al., 2019). Conversely, a sizeable body of evidence points towards the problems of implementing dementia interventions in isolation, without attention to the wider institutional determinants of care quality, including availability of training Gwernan-Jones et al., 2020). One possible risk is that visual identifiers might contribute to stigmatisation of hospitalised people with dementia, for instance by encouraging taken-forgranted assumptions about their condition and abilities instead of person-centred care (Handley et al., 2017;Houghton et al., 2016).
One potentially useful approach would recognise visual identifiers as elements of socio-technical systems (Holden et al., 2013), and accordingly as requiring purposeful work system design that accounts for training requirements and any additional organising work they may generate (Allen, 2014), including the impacts on the routines of already-busy staff. More broadly, our study findings suggest some difficulties with consistent use of identifiersperhaps reflecting the challenges faced in making reliable use of items that need to be gathered and used for each patient separately, as opposed to systems that are (literally) part of the furniture in a ward.
Future work should also address the challenges associated with the proliferation of different identifiers that we uncovered. We found multiple systems, used in various combinations. Some were established national schemes; some were adaptations of national schemes; and some were locally developed. The common use of multiple systems in combination may reflect the varying advantages and disadvantages of different types of system, as identified in our findings from both quantitative and free-text analysis. By using different types of identifier system in combination, healthcare staff may be able to benefit from their respective advantages and mitigate their respective limitations. A further area of challenge arises in relation to patients and carers' engagement with the systems. With the exception of bedside information documents, staff were not confident that patients and carers liked the systems. Perhaps of concern, and in contradiction with the guidance supporting some schemes, the results suggested that decisions about whether to use an identifier were not always taken in discussion with patients and their informal carers. While the medico-legal ramifications of 'opt-out' approaches to use of identifiers are not entirely clear (given their ambiguous status in relation to the definition of 'medical treatments'), from an ethical perspective there is an argument that identifiers that carry the risk of inadvertent disclosure of confidential information should be subject to an appropriate consent procedure (Brigden et al., 2020). Resolving some of these ethical dilemmas and developing new policies and processes to guide appropriate use of visual identifiers will be an important focus for future work for practitioners and researchers, and must be done collaboratively with patients and carers.

| Strengths and weaknesses
Our study has strengths and weaknesses. A notable strength is that it is the first effort to systematically identify and obtain staff views on a class of interventions that has become commonplace in UK hospitals, but has to date (with a few exceptions ) evaded academic attention. It constitutes a first step towards understanding the broader implications of the use of visual identifiers in UK hospitals. As an exploratory survey, we were able to gain responses from staff in at least 48 organisations, but while the information about the survey was disseminated UKwide, most responses where the employing organisation was identified came from England. The findings therefore should not be taken to be statistically representative of national practice, and comparisons between views on different approaches should be interpreted with caution. A survey of this kind was not able to capture subtle differences in the way different types of identifier were used. We did not collect data on the type of setting in which participants worked (e.g. medical versus surgical wards), and the study suggests that more research is needed on how visual identification systems are used across different wards within hospitals or organisations. Finally, the survey covered healthcare staff only and is only one part of a programme of research that separately includes qualitative work on the views of people with dementia and their informal carers on visual identifiers and their use. This approach recognises that there is no substitute for research directly involving these groups, but also gives voice to staff experiences and perceptions of different systems that may be useful to address in the design and implementation of system.

| Implications for Practice
This study highlights the importance of appropriate staff training and sound procedures to encourage good practice in contexts in which visual identifiers for patients with dementia are used. One important aspect of using visual identifiers for this population is consideration of the wishes of patients and their carers, including a clear consent process. In hospital environments in which multiple systems are in use, staff involved in care of people with dementia should be aware of the strengths and weaknesses of different systems, as well as potential challenges posed by use of multiple systems. Enhancing interoperability across units within the same hospital using different systems may be of benefit here. Another key issue is ensuring that identifiers and their implications for action are recognised by staff across all hospital areas, rather than being confined solely to wards where patients with dementia are typically cared for. Training may play a role in this, but so too may simpler interventions, for example awareness campaigns.
Policy and practice in this area should focus on development of a sound evidence base that addresses effectiveness, cost-effectiveness, work system design, advantages and disadvantages of standardisation, key features of training and other supporting infrastructure, patient and carer engagement, reduction of unnecessary waste, and ethics.

| CON CLUS ION
Though widely used in hospitals and other healthcare settings in the UK, visual identifier systems for patients with dementia and other forms of cognitive impairment are not free of challenge. Little is known about the effectiveness of the schemes, or about which scheme works best-and from whose perspectives. Our study indicates that multiple systems are in use in the UK, with healthcare settings often embracing more than one system. The survey also highlights the structural prerequisites of such systems, including adequate staffing to ensure they can be operated reliably and consistently, good work system design, and clarity about consent and other ethical concerns.

ACK N OWLED G M ENTS
We thank participants in the survey for their time in completing it. We thank colleagues at the NIHR Clinical Research Network for assistance in recruitment. We would also like to thank Gwen Brierley and Bethan Everson for their support with project governance. The Thiscovery team has provided the expertise and infrastructure for the setting up and running of the survey on the Thisc overy.org platform.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.