Towards improved decision support in the assessment and management of pain for people with dementia in hospital: a systematic meta-review and observational study

Article history

The research reported in this issue of the journal was funded by the HS&DR programme or one of its preceding programmes as project number 11/2000/05. The contractual start date was in October 2012. The final report began editorial review in October 2015 and was accepted for publication in March 2016. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HS&DR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Claire Hulme is a member of the National Institute for Health Research Health Technology Assessment Commissioning Board.

Copyright statement

© Queen’s Printer and Controller of HMSO 2016. This work was produced by Closs et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Dementia and pain are common and poorly managed

It has been estimated that 44.4 million people worldwide have dementia,1 including 670,000 people in England, and that one in three people aged > 65 years will develop the condition. 2 Dementias are chronic neurodegenerative syndromes that are most common in older people. They include Alzheimer’s disease, vascular dementia and frontotemporal dementia,3 and are associated with multiple changes in the brain, causing deterioration in cognitive performance as well as changes in behaviour, personality and communicative functioning. 4,5 Pain is also common among older people, with 45% of people aged > 65 years experiencing chronic pain. 6 A national study in the USA suggested that > 10% of people in the community aged > 65 years have dementia and that, of these, over 60% experience bothersome pain and more than 40% have pain that limits their activities. 7 In the UK the Care Quality Commission (2014)8 investigated people’s experiences of dementia care as they moved between care homes and hospitals, and found that the inability to communicate about pain was one of the most important experiences reported by people and their families. They found that assessments to identify and manage pain were variable, putting people with dementia at risk unnecessarily. 8 The identification and treatment of pain has frequently been poor among people with dementia in both acute9,10 and long-term settings. 11–13

The policy context

The policy imperative for this work is considerable. In England, the 5-year National Dementia Strategy14 (implemented in 2009) prioritised the need to improve the quality of care for people with dementia in general hospitals (objective 8). Then, in 2015, the National Institute for Health and Care Excellence (NICE) published a dementia pathway that emphasised the importance of recognising and treating pain. 15 In addition, two sets of national guidance have been published, one for assessing16 and another for managing17 the pain of older people, each of which considers issues related to cognitive impairment in this group. Although these are useful additions to the information available for clinicians, we do not yet know how to assess and manage the pain of people with cognitive impairment effectively, particularly in acute settings.

The impact of dementia on the experience and expression of pain

Patients with advancing dementia sustain progressive impairments in their cognitive, linguistic and social skills. Nevertheless, they are susceptible to the same potentially painful conditions as those without dementia, and there is little evidence to suggest that they experience less pain than their counterparts without cognitive impairment. 18 While they may experience pain, they may be unable either to understand what they are feeling or to verbalise that they are (or were) in pain. This makes it impossible for health-care professionals (HCPs) to rely on the clinical ‘gold standard’ of self-report for assessing pain in those who are severely cognitively impaired. For those who still have the ability to communicate, their inability to remember, interpret and respond to recent pain may limit their reports to ‘here and now’ experiences. 19

As 88–95% of people with dementia have difficulties with verbal communication,20,21 the recognition and assessment (and therefore management) of pain in this group is a significant challenge for those caring for them. The lack of appropriate pain management may then lead to functional decline, slow rehabilitation, disturbances in sleep routine, depression, agitation, poor appetite, impaired movement, increased risk of falling and a poorer quality of life. 22–24

Pain in hospital patients with dementia

It has been estimated that the cost of health care for people with dementia is around £1.2B, of which hospital inpatient stays account for 44%. 25 Dementia increases the length of hospital admission by an average of 4 days to > 23 days,26,27 resulting in an increase in complications28 and the risk of iatrogenic harm through polypharmacy. 29

An acute hospital ward may be a disorienting and distressing environment for a person with dementia due to heightened/unescapable noise, bright lighting and unfamiliar staff and surroundings. A study undertaken in one UK hospital showed that 95% of patients with advanced dementia were in pain as assessed using three observational pain tools [Abbey Pain Scale, Pain Assessment Checklist for Seniors with Limited Ability to Communicate (PACSLAC) and the Doloplus] as part of a randomised controlled trial in palliative care. 30 Research suggests that hospital patients with dementia are less likely to receive pain control than those without. 31

Poor pain control in the context of the acute environment is associated with neuropsychiatric symptoms, particularly aggression and anxiety,32 as well as behavioural responses such as agitation, vocalisations and withdrawal. 33 Neuropsychiatric symptoms affect over 75% of people with dementia admitted to acute hospitals and can increase the risk of mortality and cognitive decline. 34 Neuropsychiatric symptoms are particularly challenging for clinical staff to manage, and are often associated with suboptimal care or inappropriate prescriptions of antipsychotic medications. 35 Consequently, people with dementia are at higher risk of adverse events during their hospital stay36 and are more likely than their counterparts without cognitive impairment to spend an extended time in hospital. 37–39

There are no behaviours which are exclusively associated with pain, increasing the difficulty of identifying its presence. In people with dementia, many behaviours generally considered to indicate pain may also indicate boredom, hunger, depression, disorientation or other problems. 40 These behaviours lack specificity, some observational pain assessment tools may detect distress as well as pain, and there may be overlap between the two.

As it has been estimated that in general hospitals the prevalence of dementia on acute wards is around 40%,41,42 and more than 45% of those aged > 65 years have chronic pain,6 it is inevitable that hospital staff will regularly provide care for the substantial number of patients who suffer from both of these invisible but highly debilitating conditions. For clinical staff the challenge of interpreting the behaviour of the many patients who have both pain and dementia is considerable, militating against a simple approach to the assessment of pain.

The assessment of pain for people with dementia

In general, because of the subjectivity of pain, self-report is considered to be the gold standard for pain assessment. Although people with mild to moderate dementia are often able to report their pain verbally or use simple visual or numerical pain intensity assessment tools, these options are not feasible for use with people with later-stage dementia in whom communication ability is severely impaired. 43,44 As a result, previous work has shown that pain is frequently underdetected and poorly managed in people with dementia, in both long-term and acute care. 31,45–47

In the absence of accurate self-report it has been necessary to develop observational tools to be used in both research and practice based on the interpretation of behavioural cues as a proxy for the presence of pain. This approach has resulted in a proliferation in the number of pain assessment instruments developed to identify behavioural indicators of pain in people with dementia and other cognitive impairment. The most structured of these are predominantly based on guidance published by the American Geriatrics Society,48 which presents six domains for pain assessment in older adults. These include facial expression, negative vocalisation, body language, changes in activity patterns, changes in interpersonal interactions and mental status changes. However, the interpretation of many of these behaviours is complex when applied to dementia because of the considerable overlap with other common behavioural symptoms or cognitive deficits which may confound an assessment, manifesting from boredom, hunger, anxiety, depression or disorientation. 40 This increases the complexity of identifying the presence of pain accurately in patients with dementia and raises questions about the validity of existing instruments. The psychometric properties, discriminative properties and clinical utility of currently available instruments are as yet unclear. As a result, there is no clear guidance for clinicians and care staff on the effective assessment of pain, or how this should inform treatment and care decision-making. A large number of systematic reviews have been published which analyse the relative value and strength of evidence of existing pain tools. There is a need for guidance on the best evidence available and for an overall comprehensive synthesis.

Most pain tools for use with people with dementia have been developed within long-term care, and more work is required to establish whether or not the use of pain tools is feasible in the acute hospital and if these tools are reliable in detecting pain. There have been no studies in the UK exploring how pain is detected and managed in people with dementia on acute hospital wards. Recognising pain in people with dementia has been described as a ‘guessing game’ by some HCPs,49 and the Counting the Cost: Caring for People with Dementia on Hospital Wards report37 identified that 51% of carers and nurses were dissatisfied with their ability to detect pain and 71% of hospital staff wanted more training in recognising pain.

Dementia, pain and decision-making

The detection and management of pain are cognitive activities associated with decision-making. Pain detection involves identifying information cues (e.g. from a formal assessment tool, patient self-report, observation of behaviour) that would indicate a patient is experiencing pain. Clinicians then evaluate those cues to reach a judgement regarding the nature of a patient’s pain, before making a decision regarding what to do to manage that pain (the decision process). If individuals fail to assess a patient’s pain effectively, or detect that they have pain but then decide not to do anything to manage it, pain can be poorly controlled. The use of decision support tools can aid clinicians in the decision-making process, improve both the processes and outcomes of care50 and subsequently lead to an improvement in the quality of care for patients. In this study we aimed to develop decision support tools to assist clinicians with both the process of judgement (identification of pain) and decision-making (what actions to take on the basis of the judgement made). Figure 1 provides an overview of the theoretical framework that guided this study. 51,52

FIGURE 1.

Theoretical framework for decision-making.

The framework shows the three main sources of information likely to be used by clinical staff as a basis for making judgements about patients’ pain. This judgement then leads to a decision about how to manage the pain in order to produce the outcome of reduced pain and related outcomes for the patient. This theoretical framework suggests that a decision support tool could assist in the acquisition of appropriate information; the accuracy of judgements; and the making of optimal decisions. Taken together, these should improve outcomes for patients.

There is a paucity of research in this area, with little having been reported in the UK. One recent study undertaken in Australia, which examined the complexities of pain assessment and management for hospitalised older people, identified four key aspects of care: communication among nurses and between older patients and nurses; strategies for pain management; environmental and organisational aspects of care; and complexities in the nature of pain. 53 The intricacies of meeting the analgesic needs of older patients were emphasised, especially for those with communication deficits. There has been no similar research undertaken in the NHS in the UK.

Aims and objectives

The aim of the work reported here, therefore, was to inform the development of a decision support tool to be used by staff in hospital settings, to aid in the recognition, assessment and management of pain among people with dementia. In order to achieve this, two studies were undertaken (Figure 2).

FIGURE 2.

Overview of project and preliminary development of PADDS.

The first study was a systematic review of systematic reviews of observational pain assessment instruments, referred to as the meta-review, which had the following three objectives:

1. to identify all tools which are available to assess pain in adults with dementia

2. to identify the settings and patient populations with which they had been used

3. to assess their reliability, validity and clinical utility.

The second was a multisite observational study of current pain assessment and management practices in a range of wards within four hospital sites across the UK, which had the following four objectives:

1. to identify what information is currently elicited and used by clinicians when detecting and managing pain in patients with dementia in acute hospital settings

2. to explore the existing process of decision-making for detecting and managing pain in patients with dementia in acute hospital settings

3. to identify the role (actual and potential) of carers in this process

4. to explore the organisational context in which HCPs operate with regard to this decision-making process.

Development of intervention and follow-on study: health economics and technical specifications

Data from the observational study were analysed from a health economics perspective, with the aim of developing data collection instruments for the feasibility evaluation of the intervention. The focus of the health economics analysis was twofold: first, to identify cost categories and resource use associated with the intervention developed in this project; and second, design of the instruments, which involved exploration of the use of outcome measures to assess proxy issues and generate hypotheses about the domain of impact. A set of health economic data collection forms was developed as the result of this work.

Finally, on the basis of the evidence gathered within this research project, we compiled a technical requirements specification document for the design of a Pain And Dementia Decision Support (PADDS) system.

Study Steering Committee project oversight

A Study Steering Committee (SSC) was established in accordance with Health Service and Delivery Research (HSDR) guidance. The SSC consisted of four experts in the areas of dementia and pain research, together with a representative for carers of patients of dementia.

The SSC and project investigators met three times during the project (June 2014, January 2015 and June 2015) with a final teleconference at the end of the project (1 October 2015). Regular updates on the project were circulated to the chairperson and the SSC by e-mail. Documents shared with the SSC included internal reports, manuscripts for publication, protocol amendments, National Institute for Health Research (NIHR) reports and any significant correspondence with NIHR or sponsor.

Summary

This study was funded by the NIHR HSDR programme (reference number 11/2000/05). The project aimed to generate a theory-based intervention to be used as the basis for providing better-quality care in the assessment and management of pain in people with dementia in hospital settings. The objectives were to identify the best observational pain assessment tool(s) currently available; to understand and support the decision-making processes of HCPs in hospital wards; and to provide insights into how carers’ expertise can be incorporated into the decision-making process. In order to achieve this we undertook two studies, a systematic review of systematic reviews (meta-review) and an observational study, and developed the foundations for the development and implementation of a decision support intervention (see Figure 2).

Structure of the report

The structure and content of each section of this report are outlined below.

Introduction

Patient and public involvement was an important part of the research process. We involved people with dementia and their family members (‘carers’). Patient and public involvement informed the analysis of the findings from our exploratory study on how pain is currently detected, assessed and managed in patients with dementia in hospital settings, and the conclusions we drew from our study. The Alzheimer’s Society (AS) Research Network was instrumental in establishing our connection with the carers of persons with dementia via its research engagement manager.

The lay advisory group

A LAG was established at the beginning of the project. The group was invited to have an oversight and advisory role, informing the running of the project and providing advice on specific aspects of the research if necessary. It was made up of members of the AS Research Network, which is composed of carers and people with dementia who have received training in research methodology.

The LAG was made up of 16 members, one of whom was also a member of our SSC. It was agreed that the group would meet once a year and that they would receive project updates by e-mail, supplemented by further e-mail or telephone contact as and when needed.

More specifically, the involvement of the LAG members took place at three crucial moments in the life of the project:

1. at the beginning

2. on analysis of findings of the meta-review and start of the exploratory study

3. on completion of the exploratory study and initial work towards the development of a decision support intervention.

Each of these three moments of involvement is further detailed below, followed by reflections on how the process informed the development of the project.

The lay advisory group intervention at crucial times of project activities

Evaluation of lay advisory group process

Our project aimed to deliver a decision support intervention to assist HCPs in the care of patients with dementia. A core aspect of this intervention was a decision support tool aimed at staff working in hospital clinical areas. There is a challenge in having meaningful patient and public involvement activities through a research process aimed at developing a working tool for staff – a tool that patients and the public may never use or get access to. They are key stakeholders in the overall process (and the outcomes of patient care delivered with the support of the tool), but not direct end-users of the new tool. Good practices of user-centred design in tool development include involving a varied range of stakeholders, but greater attention is paid often to the direct end-users. We believe that we were successful in overcoming this challenge, and that we were able to involve carers in a meaningful and lucrative way throughout the research.

The members of the LAG provided a complementary perspective over the findings of our study, and important points came out of the discussion that we then took into account at the time of tool development. The timing of the consultation was also important; meeting at ‘transition points’ in the research process enabled feedback from the group to be incorporated into the analysis of findings both for the systematic review and exploratory study, and also into the conclusions of the study, as disseminated in manuscripts for publications.

Face-to-face communication with members of the group happened through yearly meetings. The LAG meetings lasted 3 hours, over lunch, to give enough time for travelling, and they were held at King’s College London, the site of one of the investigators (AC), and a central location. The meetings did not have a formal agenda; rather, they were open and relaxed discussions and provided an opportunity to exchange ideas and inform the group of the progress of the project. Overall, this seemed a good format, allowing space for project presentations to update the group about the progress of the project, as well as substantial time for discussion.

We were aware that family members of persons with dementia are often elderly, possibly with health conditions that make travelling more difficult for them. However, all of the group had access to computers, and we investigated the possibility of using teleconferencing or videoconferencing facilities [such as Skype™ (Microsoft Corporation, Redmond, WA, USA)], as an alternative to meeting in person. We consulted on this with the group, and preference was expressed for meeting face to face. It was therefore agreed that those who could not attend would be consulted by e-mail and their feedback included in the notes of the meeting to be circulated to all members. One of the members who could not attend in person recommended our project to the AS as an exemplar in lay members’ participation, including those who were unable to attend in person. An e-mail was sent to the AS network stating that ‘this project would make an ideal example to take to Society regional meetings to try to explain research projects to the general membership’. Through the process it became apparent that difficulties in attending events in person is a major barrier for patient and public involvement, especially for persons with dementia, and that this is something that needs to be taken into account when planning patient and public involvement activities.

Finally, we never asked the group whether they were happy to be referred to as ‘carers’. Indeed ‘carer’ is not an uncontested label56 and, in retrospect, we should have consulted the group on this. Towards the end of the project one of the group members pointed out that instead of carer she would prefer ‘family supporter (I prefer to call myself that rather than carer)’ and, although this was not raised as an issue, her message was a reminder for future patient and public involvement activities with family members of patients with dementia to consult on this at the start of the research process.

The Alzheimer’s Society Conference

As the research project was coming to its end, in June 2015, we attended the AS research conference. We presented findings from the study to a varied audience including researchers (n = 75), some of whom were clinicians or health professionals, AS staff and trustees (n = 45) and, most importantly for the purpose of patient and public involvement, 72 volunteers, of whom 66 had a personal experience of dementia, including six people with dementia.

Conclusion

Patient and public involvement was very important for the conduct of the research and the conclusions of the study. Through their insight and experiences the members of the LAG helped us to problematise the findings and draw more nuanced conclusions. This chapter, along with the rest of the report, was circulated to the advisory group members, who commented favourably.

Introduction

This meta-review aimed to identify existing assessment tools with validation data concerning people with dementia. It presents a thorough synthesis of current systematic review literature concerning the psychometric properties and clinical utility of pain assessment tools for the assessment of pain in adults with dementia, and provides a detailed picture of the state of the field in the complex task of assessing pain. 54

For ease of reference, in this report we refer to our systematic review of systematic reviews as a ‘meta-review’. We call the systematic reviews considered for inclusion in the meta-review ‘reviews’, and refer to publications included in the reviews as ‘studies’. We use the term ‘records’ to refer to the bibliographic data of publications of reviews (for the most part retrieved through online database searches). The terms ‘scales’, ‘tools’ and ‘instruments’ (pain assessment) are used interchangeably. The process of the meta-review followed guidance from the Cochrane Collaboration57 and the Joanna Briggs Institute (JBI). 58 In undertaking this meta-review, our aims were as follows:

1. to identify all tools available to assess pain in adults with dementia

2. to identify the settings and patient populations with which they had been used

3. to assess their reliability, validity and clinical utility.

Criteria for considering reviews for inclusion

Definitions of criteria for inclusion of reviews in the meta-review followed an adapted setting, population, intervention, comparison, method of evaluation (SPICE) structure. 58 We included systematic reviews of pain assessment tools involving adults with dementia or with cognitive impairment. Dementia and cognitive impairment were defined according to the US National Library of Medicine’s medical subject heading (MeSH) vocabulary. Dementia was defined as ‘an acquired organic mental disorder with loss of intellectual abilities of sufficient severity to interfere with social or occupational functioning’. 59 The dementia MeSH term covered more specific subheadings such as Alzheimer disease or vascular dementia. Cognition disorder was defined as ‘disturbances in the mental process related to thinking, reasoning, and judgment’60 (distinct from, not including, delirium). We did not include learning disorders, defined as ‘Conditions characterized by a significant discrepancy between an individual’s perceived level of intellect and their ability to acquire new language and other cognitive skills’. 61 Examples of learning disorders of this type are dyslexia, dyscalculia and dysgraphia.

We included reviews regardless of setting (e.g. acute or nursing/care homes), type, location or intensity of pain (e.g. acute pain, persistent), and outcomes of the pain assessment (e.g. patients being in pain or not). Reviews were included if they provided psychometric data for the pain assessment tools and were available in English. We excluded publications, such as narrative reviews or case reports, which did not provide psychometric data or were not categorised as systematic reviews62 (Table 2 shows our systematic review definition).

Search methods for identification of reviews

The following databases were searched: MEDLINE, All Evidence Based Medicine Reviews [including Cochrane Database of Systematic Reviews, American College of Physicians (ACP) Journal Club, Database of Abstracts of Reviews of Effects, Cochrane controlled trials reports, Cochrane Methodology Register, Health Technology Assessment, and NHS Economic Evaluation Database], EMBASE, PsycINFO, and the Cumulative Index to Nursing and Allied Health Literature; the searches were all carried out on the same date (12 March 2013). Additional searches included The JBI Library (The JBI Database of Systematic Reviews and Implementation Reports) and the Centre for Reviews and Dissemination database. Further data were retrieved through reference chaining. No grey literature was sought.

The search strategy used a combination of text words and established indexing terms such as MeSH (see Appendix 1). The search was structured by the relevant SPICE concepts. Search terms were identified by comparing published search strategies adopted by reviews in similar areas,64,67 or on the subject of pain or pain management tools, not specifically for the same patient population,68 using the search strategy for retrieving reviews outlined by Montori et al. 69 Detailed search strategies were optimised for each electronic database searched.

Selection of reviews

Four reviewers (DD, MB, VL and PE) screened all search results, initially on the basis of title and abstract and then the full text of potentially eligible papers (see Figure 3). The results of the search were divided into two sets among the reviewers, so that each review was first screened by two of the reviewers independently, then assessed, again independently, by the other two reviewers. When consensus could not be reached, the reviews were referred to a third party (SJC).

Assessment of methodological quality of included reviews

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidance on systematic reviews encourages reviewers ‘to think ahead carefully about what risks of bias (methodological and clinical) may have a bearing on the results of their systematic reviews’. 62 In our meta-review, the risk of bias may reside in each review considered for inclusion, as well as in the original studies that make up that review. We did not access the original studies to be able to accurately judge their quality or risk of bias. For each review, we assessed risk of bias in terms of how the review was conducted and the criteria applied for inclusion/exclusion. Critical appraisal was carried out by two independent reviewers, using the A MeaSurement Tool to Assess systematic Reviews (AMSTAR) systematic review critical appraisal tool. 70 Critical appraisal and evaluation of potential bias were carried out at the time of data extraction, after screening was completed on the basis of the inclusion criteria.

Data extraction and management

Data were extracted by two reviewers independently using a set of data extraction forms which were developed for the meta-review: (1) the AMSTAR checklist;70 (2) two forms for data about the reviews; and (3) one form for data about the tools. The third included a field for data extraction on the user-centredness of the tools, informed by Long and Dixon’s71 work on the development of health status instruments. 71 The data extraction forms were both paper based and built into a Microsoft Access^® database (Microsoft Corporation, Redmond, WA, USA). At the time of data extraction, the reviews eligible for inclusion were screened further on the basis of availability of psychometric data of tools. At this point, we found that some of the reviews initially identified as being eligible for inclusion in the meta-review did not provide psychometric data of tools and were subsequently excluded (this is discussed in detail in Chapter 4). Data about the characteristics of the tool (e.g. tool design and instructions for use) were extracted from the reviews; we did not search for, or retrieve, the original tools.

The search retrieved 441 potentially eligible unique records. After screening titles and abstracts, and removing duplicates, we obtained the full text of 183 records and assessed these for eligibility. We identified 23 reviews as being potentially eligible for inclusion, of which 13 were excluded as they did not provide data on the psychometric properties of the tools. The remaining set included 10 records reporting data from eight reviews (the Schofield et al. 200572 review was reported in three separate studies;72–74 we have combined the results of this) (Figure 3). The tables of included and excluded reviews are listed in Appendix 2. The reviews were synthesised using a narrative synthesis approach.

FIGURE 3.

Flow chart of retrieved sources and screening process. Reproduced from Lichtner et al. 54 © 2014 Lichtner et al. ; licensee BioMed Central. This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Further details about the methods are provided in additional files (see Appendix 3).

Introduction

The findings are structured as follows. First, we briefly summarise the reviews considered at the time of data extraction, but excluded owing to lack of data on psychometric properties of the tools. Second, we describe the reviews included in our analysis, specifically their methods and our quality assessment of them. Third, we describe the findings of these included reviews regarding the characteristics, psychometric properties, feasibility of use and clinical utility of pain assessment tools. We conclude with the reviews’ overall assessment of these tools.

Description of excluded reviews

Thirteen of the 23 reviews were excluded because they did not provide suitable data for extraction (these data were absent or not reported in a format suitable for extraction). Several of these were narrative reviews. They varied in length and details of reporting. Tools were analysed in the broader context of, for example, pain physiology, pain prevalence, and patients’ attitudes and beliefs about pain. For some, the areas of focus were the processes of pain assessment and/or pain management/interventions in the elderly population, including (but not always limited to) persons with dementia (e.g. in Rutledge et al. ,75 Rutledge and Donaldson,76 Andrade et al. 77 and Miller and Talerico78) and/or in specific types of pain (e.g. orofacial pain, Lobbezoo et al. 79). One review80 focused on the barriers to successful pain assessment, identifying non-use of assessment tools as one such barrier.

Two reviews were updated by, or related to, a later review by the same authors (Rutledge et al. 75 is an update of Rutledge and Donaldson76 – both excluded; Herr et al. 81 is related to Herr et al. 82 – the last included in our study – and aimed at reviewing the methods of the previous study).

Two reviews83,84 were reviews of reviews, and they were very specific in their aims, namely to identify pain assessment tools for adults with cognitive impairment recommended for use by paramedics83 and district nurses. 84 They identified two and four reviews, respectively, all of which were among those that we retrieved. Thus, the reasons for exclusion in their case were their nature as reviews of reviews (overviews, as opposed to systematic reviews), that they provided no data for extraction and on the grounds of repetition.

Two other reviews81,85 did report and discuss psychometric data, but not in a form suitable for this meta-review. Their choice of reporting may suggest authors’ methodological concerns for the comparability and presentation of the data in quantitative form, given the heterogeneity of the studies and their methods.

Description of included reviews

Each review included in this meta-review analysed between 8 and 13 tools (Table 3). The most frequently reviewed tools included the Abbey Pain Scale, NOn-communicative Patient’s Pain Assessment Instrument (NOPPAIN), PACSLAC, Pain Assessment for the Dementing Elderly (PADE), Checklist of Nonverbal Pain Indicators (CNPI) and Pain Assessment IN Advanced Dementia (PAINAD). Reviews had searched the literature across a variety of date ranges, from 1980 to 2010. The number of individual studies included in each review varied from 972 to 29,64 although the number of included studies in some reviews was ambiguous. The reasons for this ambiguity were twofold. First, the number of studies included in each review was different for each tool, thus making it difficult to aggregate in one number (‘number of included studies’). Second, the studies included in each review were each found to have reported one or more studies aimed at evaluating a tool. Thus, a number of included studies of ‘1’ may actually have referred to a larger number of studies conducted.

The reviews aimed to summarise the available evidence by means of a comprehensive overview. Three reviews82,86,87 also explicitly aimed at an evaluation of the evidence, that is to critically evaluate the existing tools, or to identify key components and analyse the reported psychometric properties of tools. Two reviews64,82 reported a systematic method for evaluation of the tools.

Not all reviews made explicit their assessment of the quality of the studies or risk of bias, or assessment of the scales considered. When this was done, the reviews highlighted the methodological limitations of both studies and scales. For example, in one review64 the overall assessment was ‘generally moderate’, with 11 points being the highest score out of the 20-point evaluation scale applied, and only four of the 12 tools examined reaching this score [these were Doloplus-2, L’Échelle Comportementale pour Personnes Âgées (ECPA), PACSLAC and PAINAD]. The heterogeneity of study designs and/or inconsistencies made aggregation of findings in the reviews difficult and/or methodologically inappropriate. For example, Zwakhalen et al. 64 stressed the ‘Considerable heterogeneity in terms of design (retrospective vs. prospective), method (pain in vivo vs. observational methods), research population (different types of dementia, different levels of impairment, different settings) and conceptualisation of pain’, making their results hard to compare.

We assessed the quality of the systematic reviews through the use of the AMSTAR questionnaire (original questionnaire adapted to a binary scoring: 1 if item is present, 0 if unclear, absent or not applicable) (Tables 4 and 5). The mean score was about 4.9, with a range of 188 to 1086 out of a maximum score of 11.

Most (six out of eight) reviews presented an a priori design and a comprehensive literature search (question 1; question 3). However, in general, the reporting lacked detail. For example, as shown in Tables 4 and 5, the list of included/excluded studies was provided in only three reviews,67,86,87 the explicit involvement of two or more independent reviewers (question 2) was reported in only one review,86 only three reviews67,82,86 explained the methods used to combine findings (question 9) and only one review86 seemed to have assessed the likelihood of publication bias. This lack of detail in reporting may be because of restrictions on word limits in publications. We did not contact the authors to obtain missing data.

Reviews’ findings: the pain assessment tools

In total, 28 pain assessment tools were assessed in the eight reviews. Nine tools [Abbey Pain Scale, Assessment of Discomfort in Dementia (ADD) protocol, CNPI, Discomfort Scale – Dementia of the Alzheimer’s Type (DS-DAT), Doloplus-2, NOPPAIN, PACSLAC, PADE and PAINAD] were assessed in five or more reviews. One tool [Mobilization-Observation-Behavior-Intensity-Dementia (MOBID) pain scale] was assessed in three reviews. Three tools [Behavior Checklist, Observational Pain Behaviour Tool (OPBT) and Pain Assessment Tool in Confused Older Adults (PATCOA)] were assessed in two reviews. The remaining 15 tools were each assessed in one review.

It should be noted that there seem to be different versions of PACSLAC, with a preliminary tool comprising 60 items, then a later version modified to 36 items. A Dutch version, PACSLAC-D, was also mentioned. There seems to be ambiguity about which version of the tool the data were reported about. 67 Similar ambiguity was found in relation to the PADE questionnaire; it was unclear which version – or which of its subscales – was studied for psychometric properties. Equally, the MOBID scale has been studied in two different versions. It is also unclear how much the Abbey Pain Scale had been refined across the studies carried out to evaluate it (including a Japanese version).

Description of the tools

The reporting of the tools’ content and intended use was undertaken differently by different reviews, making it difficult to provide a comprehensive comparative descriptive summary of all 28 tools. Six of the eight reviews64,66,67,72,87,88 provided summary tables giving an overview descriptions of the tools’ designs, but these summaries focus on a varied range of aspects, including target population,66 number of items,64,66,67 type of behaviours identified,72,82,88 number of dimensions/behaviours,64 presence of the American Geriatric Society (AGS) categories66,82,87 and scoring range. 64,66

From the eight reviews, it appeared that most tools (24 out of 28) were observational, requiring observations undertaken by HCPs. However, reviewers’ classification of these observational tools varied: the Face, Legs, Activity, Cry, Consolability scale (FLACC) was described as a ‘behavioural scale’ as opposed to observational;82 the Abbey Pain Scale, PADE and Pain Assessment In Noncommunicative Elderly persons (PAINE) were classified by one review67 as ‘caregiver or informant rating scales’; the same review67 classified the ADD tool as ‘interactive’ – that is an ‘interactive method’ including ‘a physical and affective needs assessment, a review of the patient’s history, and the administration of analgesic medication’. 67 In a number of reviews no specific classification is made. Among the remaining four tools of the 28, one, Present Pain Intensity (PPI), relied on patient self-reporting, while another, Proxy Pain Questionnaire (PPQ), was described as relying on caregivers reporting, and compared pain experienced currently with pain experienced the previous week.

Twenty-five of the 28 tools appeared to include an assessment of pain intensity. Three tools (ADD protocol, Behavior Checklist and OPBT) aimed at determining simply the presence or absence of pain, with no scoring or rating of pain intensity. In the case of two tools [Doloplus-2 and Rotterdam Elderly Pain Observation Scale (REPOS)] binary scores were summed up and the total score interpreted as presence/absence of pain.

Methods of scoring and rating of pain varied, from scores made of counting checkmarks – that is yes/no binary responses (item present or absent), to a variety of rating systems; total scores ranges varied from 0–6 to 0–60 {i.e. 0–6 [CNPI], 0–9 [PATCOA], 0–10 [FLACC, MOBID, PAINAD], 0–14 [Edmonton Classification System (ECS)], 0–25 [OPBT], 0–27 [DS-DAT], 0–30 [Doloplus], 0–44 [ECPA], 0–54 [Rating Pain In Dementia (RaPID)], 0–60 [PACSLAC]}. Likert scales, binary scores, multiple choice and visual analogue scale (VAS) systems were mentioned in the reviews; in three cases [PADE, Pain Assessment Tool for use with Cognitively Impaired Adults (PATCIA), PPQ] different rating systems are used in the same tool (Likert scales/VASs; Likert scales/binary scores).

Only 2 out of 28 tools (CNPI and NOPPAIN) appeared to be designed explicitly for pain assessment both at rest and during movement, although data about this aspect of the tool’s design may have been missing for the other tools. For one tool (Doloplus-2) the score was reported to reflect the progression of the pain experienced, rather than the patient’s pain experienced at a specific moment in time.

Two tools (REPOS and the ADD protocol) combined assessment with guidelines for intervention (this is discussed further in Settings in which the tools were studied).

Three reviews66,82,87 explicitly analysed the tools in terms of whether or not their design applied the AGS48 guidelines and categories of potential pain indicators in older persons, namely facial expressions, verbalisations/vocalisations, body movements, changes in interpersonal interactions, changes in activity patterns or routines, and mental status. These reviews covered 15 tools [Abbey Pain Scale, ADD protocol, Behavior Checklist, CNPI, DS-DAT, Doloplus-2, Facial Action Coding System (FACS), FLACC, MOBID, NOPPAIN, PACSLAC, PADE, PAINAD, PATCOA and the Pain Behavior Method (PBM)].

Settings in which the tools were studied

The tools were studied in a variety of settings and with varied patient populations. The terminology used to describe settings varied, and those which appeared to be in non-acute settings included long-term care, nursing homes, dementia care units, psychogeriatric units, rehabilitation facilities, aged care facilities, residential care facilities, long-term care facilities and palliative care, but also included geriatric clinics, care homes, residential and skilled care facilities, long-term-care dementia special care units and a residential dementia care ward.

The terminology used to refer to hospital settings also varied, with reference either to patients and/or type of services. For example, terms used included hospital patients in a long-term stay department, psychiatric hospital setting, hospital medical care unit, dementia special care units in hospital, hospital patients and older hospital patients.

Tools’ psychometric data

Feasibility and clinical utility

The feasibility of a tool is ‘its applicability in daily practice’, including aspects such as ease of use and time required to administer it, whereas clinical utility is ‘the usefulness of the measure for decision-making’, that is, to inform further action, such as the administration of analgesics. 66 Data on feasibility and clinical utility of tools were very limited. Often data were not available in the reviews or, when data were available, limitation often pertained to a lack of data in the original studies (e.g. reviewers stating the item was not reported and could therefore not be assessed). More specifically, feasibility data were completely absent for six tools (Comfort Checklist, FLACC, PAINE, PATCOA, PPI and PPQ); clinical utility data were completely or substantially absent for seven tools (ECS, FACS, MPS, PAINE, PBM, PPQ and RaPID). For four tools reviewers explicitly noted that claims of feasibility (e.g. time required to administer the tool) were made from the authors of the study without supporting evidence (Abbey Pain Scale, Doloplus-2, PACSLAC and PADE). There were also two instances (PACSLAC and MOBID) of conflicting data on clinical utility and feasibility from the different reviews, possibly because of an ambiguous reference to different versions of the same tool.

Specific evaluation for feasibility appears to have been carried out only for three tools (CPAT, MPS and PATCIA). In the first two of these cases the evaluation was done by use of questionnaires. It also appeared that users of the Abbey Pain Scale were asked for feedback in the context of the psychometric testing of the tool. In was unclear whether or not the ADD protocol was also assessed for feasibility.

Specific evaluation of clinical utility appeared to have been undertaken for the PATCIA tool, and possibly for the ADD protocol and PAINAD.

It must be stressed that when reviews assessed or mentioned the feasibility and/or clinical utility of the tools, the two aspects were often confounded (reviewers, authors or users typically drawing conclusions from ease of use or brevity of a scale to its usefulness).

Specific dimensions of feasibility assessed were time to complete the assessment (e.g. to complete a checklist), availability of instructions on how to use the tool and/or availability of guidelines on how to score pain, and training needs.

Six tools were reported to be ‘easy to use’ (Abbey Pain Scale, Behavior Checklist, CNPI, CPAT, MPS and NOPPAIN), two were considered manageable/acceptable (ECPA and RaPID) and four were judged to be complex (ADD protocol, DS-DAT, PATCIA and PADE). Conflicting views on the ease of use or complexity of the tools were apparent for five tools (Doloplus-2, MOBID, PACSLAC, PAINAD and PBM).

Instructions for use and/or guidelines on scoring were reported to be available for 13 tools (Abbey Pain Scale, CNPI, CPAT, Doloplus-2, DS-DAT, ECS, FACS, MPS, MOBID, NOPPAIN, PACSLAC, PAINAD and REPOS), with varied assessments in terms of clarity or complexity of the instructions.

Training in the use of the tool was judged as necessary for 10 tools (EPCA-2, MPS, NOPPAIN, PADE, PACSLAC, PAINAD, ADD protocol, CPAT, DS-DAT and MOBID), four of which seemed to require significant training (ADD protocol, CPAT, DS-DAT and MOBID). For six tools it was stated that the authors of studies creators of tools did not report the level and length of training required (Abbey Pain Scale, CNPI, Doloplus-2, PAINAD, PACSLAC and REPOS). For the majority of the tools, however, data about training were not available (Behavior Checklist, Comfort Checklist, ECPA, ECS, FACS, FLACC, OPBT, PATCIA, PAINE, PATCOA, PBM, PPI, PPQ and RaPID).

Specific dimensions of clinical utility were less straightforward. The availability of cut-off scores and interpretation of scores for decision-making appeared to be the two dimensions supporting evidence of clinical utility. The presence of cut-off scores contributes to achieving clinical utility, for example to help discriminate between presence and absence of pain, or to couple the scale with a treatment algorithm. Otherwise, general statements were available related to evidence of use in clinical settings and evidence of clinical utility, the latter being dependent on the former.

Cut-off scores appeared to be available only for REPOS and Doloplus-2, although in the latter case the scores still need to be validated. The availability of guidance on how to interpret the scores and for further action following assessment was variably reported. Data on this were missing for 18 tools (ADD protocol, Behavior Checklist, Comfort Checklist, DS-DAT, ECS, EPCA-2, FACS, FLACC, MPS, MOBID, PATCIA, PAINE, PATCOA, PBM, PPI, PPQ, RaPID and REPOS). For four tools it appeared that interpretations of the scores were available (Abbey Pain Scale, Doloplus-2, CNPI and CPAT), although in two of these they were deemed unclear (CNPI and CPAT). For five tools, it appeared interpretations of the scores were not available (ECPA, NOPPAIN, OPBT, PACSLAC and PADE) and a further four were reported as lacking guidance for further action following assessment (CNPI, CPAT, Doloplus-2 and NOPPAIN). It was unclear whether or not PAINAD does provide interpretation of scores.

Suggestive evidence of use in clinical settings was reported for two tools; the Abbey Pain Scale has been incorporated into the Australian pain guidelines, while the ADD protocol was introduced in 57 long-term care facilities together with an education strategy for 12 months, and the evaluation was done in a study with 32 nurses in 25 facilities. It is otherwise unclear whether any of the other tools were actually used in practice beyond a period of research or testing of the tool. In one example (NOPPAIN), the testing of the tool was done by use of video-recording of an actress portraying a bed-bound patient, and it remains unclear whether further testing was done in clinical setting or with patients at bedside.

With the exception of the ADD protocol and the REPOS, there was no mention of how the tools would inform intervention (e.g. choice of treatment). We found an overall lack of clarity in the descriptions of the ADD protocol, but it seems that one of its strengths is that it links observation of behaviour with interventions. Similarly, one review suggested that the clinical utility of REPOS potentially resides in its combination with a decision tree to assist in determining interventions after pain assessment. 86

Summary of findings

There was a consensus among the reviews that the current evidence on validation and clinical utility of the tools is insufficient. The overall conclusion was that there is a need for further psychometric testing of each tool. Two reviews recommended that the focus should be on studying existing scales rather than creating new ones,64,72 although one review also suggested that there may be a need to revisit the tools’ conceptual foundations. 82

Recommendations for further research and testing of the tools included the involvement of culturally diverse populations64,82,87 and the provision of scoring methods and guidelines for interpretation in the evaluation of the scale. 87 Finally, a need for research emerged to link assessment with treatment algorithms. 66

Some of the reviews also concluded with recommendations for practice; for example, the use of at least two different pain assessment approaches at the same time in clinical practice and two different tools in research;67 the importance of a comprehensive approach to pain assessment beyond the use of tools;82 and the need to involve social workers in regular holistic multidisciplinary pain assessment (in nursing homes), with training in the use of the scales. 87

Among the tools selected by the reviews as possible best candidates, albeit on limited evidence, were the DS-DAT, Doloplus-2, MPS, PACSLAC, PAINAD, Abbey Pain Scale, and ECPA. However, there was no clear consensus on the single best tool. A summary of the conclusions and recommendations from each review is given in Table 6.

The ADD protocol was mentioned as an example of a more comprehensive approach for the identification of pain, beyond the use of an assessment tool as ‘a standardised tool is only one step in a complex diagnostic process’. 82 There was also agreement on recommending that patients with dementia can often reliably verbalise their pain, suggesting therefore that the use of observational scales should be limited to patients who demonstrably cannot reliably verbalise their pain.

This meta-review did not cover a new tool recently developed in France by the Doloplus Collective team – Algoplus. 89 To our knowledge, this tool had not yet been included in a systematic review.

Conclusions

The assessment of pain in patients with dementia is challenging for clinicians, because of some patients’ inability to verbalise the nature of their pain, and clinicians’ inability to understand other pain-related cues. This review highlighted the state of the evidence base in relation to pain assessment tools, and provided insights into current gaps in our understanding. We identified a total of 28 tools that could possibly be used in clinical practice to help with this process. However, we cannot at present recommend any particular tool for use in any clinical setting because of the lack of comprehensive evidence on the reliability, validity, feasibility or clinical utility of any one particular tool. These findings are discussed in Chapter 9.

Introduction

This project hypothesised that a rigorously developed decision support tool could help clinicians, carers and people with dementia by improving pain assessment and management in acute hospital settings. To inform the design of such a tool, an exploratory study was conducted to understand how pain is currently recognised, assessed and managed among patients with dementia in representative acute settings in the UK, through the lens of decision-making theory.

Aims and objectives

The overall aim of this exploratory study was to investigate how HCPs and others recognised, assessed and managed pain in patients with dementia in a range of acute settings. This was to provide the basis for the development of a decision support tool to improve the management of pain for this population.

The study addressed the following objectives:

1. to identify what information is currently elicited and used by clinicians when detecting and managing pain in patients with dementia in acute hospital settings

2. to explore the existing process of decision-making for detecting and managing pain in patients with dementia in acute hospital settings

3. to identify the role (actual and potential) of carers in this process

4. to explore the organisational context in which HCPs operate with regard to this decision-making process.

Theoretical framework

Pain is multidimensional, consisting of sensory, cognitive, affective and social components. The focus of this study was physical pain, although we acknowledged that pain and emotional distress are closely linked, as distress may exacerbate pain symptoms and vice versa. Pain experiences are associated with a multiplicity of factors that are unique to each individual and, even in the absence of cognitive impairment, they are very difficult to communicate meaningfully to other people. 90,91

We conceptualised pain assessment and management as involving decision-making processes, such as the accurate interpretation of the patient’s pain experience (an assessment or judgement), and taking appropriate actions to ameliorate the pain (making treatment decisions). This is explained in more detail in Chapter 1 (see Figure 1). There are a variety of theories of judgement and decision-making. 52 A common central element is the existing information that supports judgements and decisions. This includes the type of information, how it is gathered and where it is found. For example, in the hypothetico-deductive reasoning model of decision-making,51,92 individuals process information to make a judgement, defined as ‘an assessment between alternatives’. 93 Information is gathered through ‘cue acquisition’, for example, clinical information about a patient such as the patient’s verbal reports of pain or observation of their behaviour. Hypotheses are then generated to explain and interpret the information, and more information is gathered if needed, until a hypothesis is chosen that is supported by the majority of the evidence or information.

In investigating how HCPs and others recognise, assess and manage pain in patients with dementia, the research design and research instruments for this study were focused on information (types and availability of information sources, clinicians’ information needs and methods of recording information), individuals’ processes of perception, judgement and decision-making, as well as any tools used in the process and documentation of patients’ pain. The model of judgement and decision-making focused on linear processes as shown in Chapter 1 (see Figure 1) was used to guide data collection.

Design

This study was a multiple case site study, with embedded units of analysis (individuals, wards and organisations), approached with ethnographic methods. Case studies involve an empirical design that focuses on describing phenomena within their real-life context94 and appropriate to exploratory objectives.

Setting

Four case sites (NHS hospital trusts, each with one or more hospitals) were sampled to provide varying settings for acute care: one in the south of England, two in the north of England and one in Scotland. One of the four organisations used electronic patient record systems, whereas the others used paper for medical and nursing notes. Criteria for sampling included the type of hospital (tertiary referral centre/secondary care) and the type of service provision available to HCPs in the hospital (e.g. a specialist pain management team, dementia outreach team). In each site two wards were initially selected for data collection, with additional wards approached where access to participants was found particularly challenging. The selection was theoretically driven to ensure that there was representation from a variety of clinical settings in acute care where patients with dementia may be cared for (e.g. orthopaedic, acute medicine, care of the elderly) across the sample, as shown in Table 7. This approach was used to ensure a detailed, comparative overview of how pain is currently detected and managed in patients with dementia in a wide range of acute care settings would be derived.

Data collection

Data collection was undertaken by research fellows at each of the four sites. Data were collected at three of the sites (case sites H1, H2 and H4) by one researcher each (VL, NG and KJ), and by two researchers, in turn, at case site H3 (SC and CS). In each case site, a variety of data collection methods were used to provide multiple sources of evidence for addressing the research questions. Non-participant observation of HCPs and health-care assistants (HCAs) interacting with patients who have dementia was used to identify how information appears to be identified and elicited in order to detect and manage pain, and the care processes that currently take place. This included observing patients at bedside; a focus on how and where pain was discussed and documented; interactions between HCPs/HCAs, patients and carers; interactions between members of the multidisciplinary team (MDT); and availability of resources such as pain specialist services. An observation protocol derived from the theoretical framework was used to guide data collection.

Semistructured interviews lasting approximately 15–60 minutes were carried out with staff (HCAs, nurses, doctors, other members of the MDT) and patients’ family members (‘carers’). Interview guides were used flexibly (see Appendix 4), with the freedom to explore any other relevant issues specific to the site. For the most part these focused on exploring people’s perceptions of how pain was detected and managed in each of the wards, how carers were involved in the process, how the process may be improved and what an effective decision support tool would look like (e.g. format, content and resources). Interviews were recorded and transcribed verbatim, with the exception of those conducted in case site H3 which were recorded using hand-written notes. In site H3 recruitment was particularly difficult, even with the help of a local research (dementias and neurodegeneration network) nurse. Staff were protective of patients and appeared to feel that tape-recording of interviews was unacceptable (even though we had ethical approval). Access to patients had to be negotiated particularly carefully in this site because of the gatekeeping role of ward nursing staff. Note-taking ensured that we did capture relevant data, in a slightly different form.

We looked for any existing policies and procedures in place in the unit and/or organisation that were specifically focused on the detection and management of pain. We also audited patients’ medical and nursing notes for documentation of pain assessment, action taken, pain reassessment and records of prescribed analgesics.

Initial coding of the data was undertaken, collaboratively, with research staff from each site to agree themes and coding categories, and then brought to the whole research team at regular meetings. Analysis was carried out in parallel to data collection and when no new themes emerged from any of the sites, it was considered that saturation had been reached. Data were collected between May 2013 and July 2014.

Participants

Eligible participants were aged > 65 years with a recorded diagnosis of dementia. The degree and type of dementia, and presence of pain, were not recruitment criteria, as we were interested in whether and how potential, as well as actual, presence of pain was addressed by staff in the wards. However, patients in the wards we sampled were likely to have undergone medical procedures, or to be recovering from falls, for example, and it would be highly likely that pain was being experienced.

The sampling of staff and carers for interviews included all members of staff caring for patients in the wards included in the study, in addition to managers and specialists of relevant hospital services. Carer interviews were limited to the family of the patients participating in our study.

The number and length of observation or interviews required to provide an adequate overview of ward-based activities was governed by a notional guide; reaching a specified target (number of participants) was not our concern. Our research was informed by the principles of theoretical sampling and theoretical saturation,95 rather than those usually found in quantitative sampling. Initially, our sampling was done purposefully, on the basis of the aims of the research and the availability of cases (patients who consented). As data collection progressed, we began the analysis and this informed our further data collection activities.

Ethics, consent and permissions

Ethical approval was obtained both in England (National Research Ethics Committee Yorkshire and The Humber – Leeds West – Research Ethics Committee reference number 12/YH/0363) and Scotland (Scotland A Research Ethics Committee, Edinburgh – Research Ethics Committee reference number 13/SS/0006). The process to recruit patients was informed by the Mental Capacity Act 200596 and the Mental Health (Care and Treatment) (Scotland) Act 2003. 97 This recruitment process included written consent by patients or agreement from a carer consultee, patients’ capacity assessment to consent, consultation with staff and assent of carers. 98 Interviewees gave their written consent and were informed of the audio-recording. The NHS trusts participating in this study granted access to the researchers, who complied with local requirements for data collection. Data were anonymised at the time of data collection.

Data analysis

Data for analysis from the observational study consisted of verbatim transcripts of observation sessions, field notes of medical and nursing records, notes and transcripts of interviews. Data were indexed using NVivo (version 10, QSR International, Warrington, UK) qualitative analysis software and subjected to thematic analysis. Both inductive and deductive approaches were applied, with dimensions of decision-making (information/pain cues and documentation; judgement/pain assessment; decision/pain management) providing initial categories for indexing data, but with a variety of other themes (e.g. about the context of care) emerging from the data.

Data from observations and audits of patients’ records were coded separately from interview transcripts. The themes generated from the interview data tended to be related to decision-making processes. These included types of information used, judgments and rationales for decisions. Participants explained their thought processes and actions, and field notes from observations illustrated the context of care and what ‘actually happened’ in terms of patient care and patient–staff interaction. The audit of patients’ nursing and medical notes provided evidence of how and when documentation of pain assessment and management was undertaken. Data from the three sources were compared and contrasted, and then integrated. This provided a more nuanced understanding of the processes and events concerned with the recognition, assessment and management of pain. For example, activities planned for set times actually took place at different times (or not at all), and the documentation of assessment was often too brief to convey the richness of the different aspects of actual pain-related activities and interactions between staff and patients.

The strategy for the multisite qualitative data analysis emerged through a process of team meetings, sharing of documents and reflections among the interdisciplinary team of researchers and investigators. It became apparent that a large number and varied range of themes were emerging from the analysis, but it was agreed that a focus on dimensions of decision-making was necessary to answer the research questions. The senior research fellow in the team (VL) led the multisite analysis, which was then discussed with the research team and verified by other researchers (CS, LR and KJ). Transcripts were scrutinised to identify themes or categories, which were then used to code the data. Subsets of the data set were coded by three of the researchers (VL, KJ and RL). The senior research fellow checked a sample of each subset to verify consistency in the analysis. Data in each theme were examined for negative cases or contradictory findings. 99 To increase transparency in the analytic process, team meetings of the researchers from each of the sites convened with the project analysis group (including also SJC, NA, JK and MB), on a regular basis during, and after, the data collection period. This ensured consistency between the four sites. Emerging themes were compared, contrasted and discussed within the group and with the wider project team until a consensus was reached.

At the end of the study a two-page information sheet summarising the project was sent to participants (see Appendix 5).

Participants’ characteristics

Thirty-one patients with dementia participated in the study and were observed for a total of 170 hours at bedside (Tables 8 and 9). A total of 52 health-care staff and four carers were interviewed (Table 10). Interviews were supplemented with informal conversations recorded in field notes with staff and carers during observations. Among the health-care staff interviewed across the four case sites were seven HCAs, 31 nurses (staff nurses, sisters and specialists), three doctors in training, five medical consultants, a pharmacist, a physiotherapist and four clinical educators. Ward management was informed that all members of staff were invited to take part in interviews, but there were challenges in finding time when they would be available and researchers tried to minimise disruption to normal work activities.

Patients included in the study had a mean age of 88 years (range 75–99 years). A number of challenges were encountered in recruiting patients for all case sites. Among them was the lack of a documented diagnosis of dementia in ward notes despite strong indications that the condition was present, and many patients had no available carer.

Observations took place at the bedside, and included any HCP interacting with the patient. It did not follow specific members of staff. As a consequence, we do not have data about the ‘current shift’ from the HCP perspective, so data collection activities could not be recorded in terms of shifts.

Our initial research protocol planned for ‘40 shifts per case sites’ (i.e. 20 shifts per hospital ward), but this had to be interpreted as ‘periods of observations’. For the purpose of reporting data collection activities, a period of observation was defined as a period of continuous observation at a patient’s bedside in any 1 day. One or more periods per patient could be done per day of data collection.

Our periods of observations ranged from 30 minutes to more than 6 hours (see Table 9).

Themes emerging from qualitative analysis

The analysis identified four overarching themes which are discussed below: communicating pain with dementia; carer–clinician communication; use of analgesics; and putting a picture together.

Conclusion

Pain assessment and management are complex activities that embrace physiological, emotional, cognitive and social dimensions. Pain is often described as a private experience, but in reality it regularly requires public expression in order to obtain relief. When caring for people with dementia, the complexity of communication is exacerbated by the challenges that the progressive cognitive and functional decline present. These complex social processes occur within hospital contexts and routines organised mainly around the needs of the organisation rather than those of patients. This complexity may, at least in part, explain why simple tools available to assess pain in patients with dementia are not well used in practice. Future decision support interventions need to take this complexity into account during their development, reorganising frames and quality of time for communication with patients, making a more varied range of pain management interventions routine, and devising tools that bring together information to provide a ‘picture of a patient’s pain’ accessible to all involved, within an overall framework of person-centred care.

Methodology for the design of a new intervention

In designing our intervention we have followed the guidance for theory-based intervention design,103–106 and have taken inspiration from user-centred design approaches. 107,108

Our aim was to achieve an improvement intervention based on both evidence and theory. 109 The data gathered through the observational study were our main evidence, supplemented with the literature. The main theoretical frame of reference was the decision-making theory51,52 underlying traditional pain assessment models, although other theories were also explored in connection with the range of different themes emerging from the analysis of the data.

The work towards the identification of the intervention was undertaken in parallel with the analysis of the data from the observation study (as described in the Chapter 5). Although the initial aim of the research was the development of a decision support tool, we did not exclude the possibility that other forms of interventions would be as (or more) appropriate. The team was open to the exploration of all possible options, and to then select and prioritise those more strongly supported by the evidence and that had the potential to address the issues with pain recognition, assessment and management identified by our study. Once generated, one (or more) interventions feasible for implementation in NHS hospital settings would then be selected for further development (feasibility being an important aspect of any successful improvement strategy,109 in light of current improvement and policy initiatives in the areas of dementia, pain, and organisation of hospital services (e.g. Department of Health14,110).

The method was iterative and collaborative, involving the whole team of researchers and investigators. Their individual contributions reflected their multidisciplinary areas of expertise which included pain management, mental health of elderly people, occupational therapy, improvement initiatives in acute care settings, decision-making and decision support design, user-centred design and health economics.

In practice, the process unfolded with the following iterative and collaborative activities:

• Identifying in the data areas for improvement, specifically, identifying the data issues, gaps, and problems in current hospital practice.

• Periodically writing-up and circulating analysis documents among the team to support discussion.

• Brainstorming ideas for improvement strategies during analysis meetings.

• Discussion with the SSC (January 2015 meeting). The SSC reminded the team that in designing any intervention, metrics and methods to assess its feasibility and effectiveness would also have to be included.

• Collating all ideas into one whole final table, containing proposals and their rationales (hypothesis/theory). More than 30 intervention components were collated, of various expected impact and level of feasibility (a summary in Table 11). They ranged from systemic health services reorganisation (e.g. involving staff level and skill mix, empowering of HCPs, integration of mental health services into hospital care), changing local practices (e.g. making pain teams routine in all wards, keeping the patient in HCPs’ ‘horizon of observation’), and improving documentation and communication among the MDT.

• A final analysis workshop, with Skype teleconference (with Dawn Dowding participating from Columbia University, New York, NY, USA) was held to prioritise and select the intervention.

  Following guidance about the development of evidence/theory-based interventions,103 the workshop was structured around seven key conceptual stages that summarise the improvement intervention proposals:

  1. how is pain understood and known

  2. the problem(s) identified in pain assessment and management

  3. components of the proposed intervention

  4. mechanisms that would make the proposed intervention work

  5. expected outcomes

  6. metrics to evaluate effectiveness

  7. conditions (contextual factors) necessary for the intervention to work as intended.

• Large flipchart sheets of paper – one for each stage – were laid out on a long table and potential interventions and related elements were mapped onto them with sticky notes (Figure 4). This facilitated making assumptions explicit (e.g. about mechanisms and contextual factors), and prioritising interventions that had the potential to be more feasible in the NHS context. The team agreed that this would be a complex intervention around documentation that would facilitate HCPs in ‘making sense’ of a patient’s pain and support judgements and decisions (later we called this PADDS; we describe this in Chapter 8).

• Finalising a design rationale document, explaining rationales, hypotheses and assumptions for the chosen intervention (see Methods for the design and evaluation of the decision support tool). This was shared via e-mail with the members of the SSC.

FIGURE 4.

Design team analysis workshop: working out components, rationales, expected outcomes and assumptions (the ‘programme theory’). Potential interventions and related elements were mapped with (a) sticky notes and organised into key components; (b) how pain is understood; (c) components; (d) mechanisms; (e) outcomes; and (f) conditions.

In summary, our analysis and exploration of potential interventions led us to a range of intervention components, targeted at different levels of the organisation, from the individual member of staff to ward-level or hospital-wide initiatives. As we acknowledge the potential for such a wide-scope multifaceted complex intervention, the focus of our research was on those elements that aim to support decision-making at the bedside. These include improving the recording of pain assessment/management and creating a more complete pain picture. Therefore, we proceeded with the design of an intervention of this kind.

Methods for the design and evaluation of the decision support tool

As mentioned in the introduction (see Methodology for the design of a new intervention), we approached the design of the intervention on the basis of the principles of implementation science and user-centred design. The former suggests the need for a ‘programme theory’ as the basis of implementation and evaluation; the latter suggests starting with high-level (conceptual) design before moving on to low-level design (the actual working of the system and the ‘look and feel’ of the system). The three phases inform each other as propositions/ideas are discussed, and mock-ups are tested and evaluated (Figure 5).

FIGURE 5.

Three phases for the design of a theory-based intervention.

We completed the first steps of the first two phases by developing the foundations for a programme theory (Table 12, further developed in the proposal for the PADDS system) and the requirements specifications document (version 1; see Appendix 6). These were based on the discussions and analytic work done by the team, written up by the senior research fellow (VL), with documents circulated for comments among the team. The requirement specification document was written with reference to the requirements specifications for electronic prescribing systems published by NHS Connecting for Health. 111

The third stage, the low-level design, including codesign sessions with users and evaluation, required more time and different expertise than those available within the research team and was therefore left to further research. This third stage will incorporate a health economic evaluation, using the outcome measures developed in this study. However, given that the proposed intervention is more complex than anticipated, further development of this part of the evaluation will be essential.

At this point the high-level design envisages a ward-based electronic health record module designed to ‘slot into’ existing systems, augmenting rather than replacing existing electronic or paper forms or records where a patient’s pain is documented. The actual design of this module will be strongly influenced by potential users during the codesign process.

Background

There has been little research about the cost or cost-effectiveness of interventions to identify or manage pain for individuals with dementia. For example, a recent review112 identified only five randomised controlled studies published since 2003 targeting behavioural disturbances and behavioural interventions targeting pain; none of these considered the cost or cost-effectiveness of the interventions. Similarly, a systematic review, of the same year, considered the effectiveness of interventions targeting pain or behaviour in dementia. The study included 16 papers, but again none assessed cost or cost-effectiveness. 113 However, evidence suggests that dementia is costly. The onset of behavioural disturbances such as agitation are often linked with pain114 and result in more frequent medical examinations and hospitalisation, thus higher costs. 115

Aim and objectives

The aim of the economic substudy was to develop health economic data collection forms that may be used in a future economic evaluation of a decision tool for pain management in people with dementia.

The data collection forms would comprise health-care and personal social care resources used by this population over the course of their hospital stay and subsequently, post discharge, in the community, and would be an outcome measure congruent for use in an economic evaluation in line with NICE guidelines. 116 In addition, recommendations for how these forms might be administered was deemed important.

With respect to resource use, while in the UK there is increasing use of NHS administrative records for collection of these data for use in economic evaluation, for personal social care there is considerable variation in the configuration of services and recording systems, so use of this type of tool for collecting detailed information of resource use is likely to continue to be necessary. 117 With respect to the outcome measure, cognitive challenges suggest that in some contexts a proxy measure, completed by a supporter or HCP, might be appropriate.

A first draft of the data collection forms focusing on resource use would be developed drawing on existing literature in the field. These data collection forms would be designed to capture health- and social-care resource use associated with the intervention and its implementation, and any subsequent health and social care used in the community post discharge. The forms would be designed to allow tick-box completion wherever possible. Feeding into the semistructured interviews with clinical staff, patients’ and carers’ opinion would be sought in respect of:

• health- and social-care resource use that should be included

• acceptability/feasibility of draft questionnaires and their content

• the optimal method and timing of data collection.

The data collections forms would be adapted in line with the results from the interviews. The semistructured interviews would also provide a forum in which to assess the feasibility of patient self-completion of outcome measures suitable for use in an economic evaluation. This would include the European Quality of Life-5 Dimensions (EQ-5D), the EQ-5D proxy version (www.euroqol.org/home.html), Dementia Quality of Life Measure (DEMQOL) and the DEMQOL proxy (developed for completion by carers of people with dementia). 118

Within the original protocol a feasibility study was to be undertaken. The decision tool, developed in the overarching study, was to be assessed for feasibility and acceptability in acute hospital care settings. As part of this planned feasibility study it was intended that we would also test the acceptability and feasibility of the economic data collection forms (outcome measures and resource use). The forms could be revised in light of the results from the feasibility study and used in any subsequent randomised controlled study.

However, over the course of the overarching study it became clear that the feasibility study would not be undertaken, and no decision tool was developed. As such, the aim of health economic component of the study became more modest, with focus lying on the development of the health economic data collection forms. The feasibility and acceptability may be tested in any subsequent feasibility study (outside of the study reported here).

Methods

The data collection forms for use in economic evaluation were developed based on a review of the literature and on the ward observations made by the research team (described in Chapter 6). In the case of the latter, a workshop was convened of field researchers and wider research team members including methodologists, clinicians and experts in dementia. A semistructured interview schedule was used that addressed the following areas:

• resource use associated with the patient group while they stay in an acute hospital setting

• the professionals involved in patient care and the tasks these professionals undertake

• the types of assessments or therapy patients have had while in hospital

• the medication types and frequency, specifically with regard to pain

• the reasons for admission, that is, comorbidities.

Notes were taken during the workshop, and content analysis was used to identify elements to include in the design of research instruments.

The focus of the literature review, undertaken in 2013/14, initially lay on pain identification and/or management in individuals with dementia, and cost-effectiveness. Although this was not a systematic review, we used key search terms ‘pain’, ‘dementia’ and ‘cost-effectiveness’. We were unable to locate any previous cost-effectiveness studies, and this was borne out by systematic reviews of the effectiveness of interventions in this area which, it was anticipated, would identify studies in which the cost-effectiveness would be undertaken alongside the clinical effectiveness. As highlighted earlier, Achterberg et al. 112 in their review of interventions targeting behavioural disturbances and behavioural interventions targeting pain, and Pieper et al. 113 in their review of interventions targeting pain or behaviour in dementia, did not report on any studies that included cost-effectiveness. As such, we spread our net over a wider area to identify randomised controlled studies that included cost-effectiveness analyses of dementia treatments.

Based on the findings from the literature review, the results from the workshops and previous data collection forms produced within the Academic Unit of Health Economics (University of Leeds), draft resource-use data collection forms were then designed. The forms also included the two potential quality of life measures that might be used within a subsequent randomised controlled trial: the EQ-5D, three levels (EQ-5D-3L), a measure of health-related quality of life used in the estimation of quality-adjusted life-years, and the DEMQOL,118 which is designed to work across dementia subtypes and care arrangements, and can be used at all stages of dementia.

Once completed, the draft data collection forms were then distributed among the researchers and LAGs for feedback, and comments was received on their acceptability.

Results

The review of the literature revealed that adaptations of the Client Service Receipt Inventory (CSRI) were typically used in economic evaluations alongside clinical trials for this population. 119–121 The CSRI collects details of resources use across health and social care, as well as information on areas such as housing. It has been successfully used in the collection of information on costs, service utilisation, income and related matters. 122 The CSRI in its original form collects data across a wide range of variables including socioeconomic data, education, employment and income. Clearly some of these variables were likely to be surplus to the requirements of the resource use required within this study. We also drew on previous resource-use data collection forms, developed in house for use with similar populations. For example, data collections used within the Pilot Trial of STOP delirium! (PiTSTOP) study; a study of a delirium prevention programme. 123 Finally, the workshop informed the content of the draft resource-use form.

The team analysis workshop was conducted on 7 April 2014 with five project researchers/investigators from across a range of disciplines, and opinion was sought with respect to identifying resource use, professionals’ time, assessment, therapy and medication, as well as reasons for admission. As a result, the following cost categories were identified:

• Resource use, including staff involved in direct patient care-related tasks, such as feeding, clothing, toilet care, supported walking and patient comforting. The supportive family members in some of the tasks challenged this cost category ‘family members are not considered part of the care team’. Resource use included time spent with the patient. For example, it was noted that during the end-of-life stage, the palliative care teams may become involved and staff may find it very useful if a family member is there to observe changes in the patient. If no family members can be present at night, it may be possible for a nurse to stay with the patient if time and workload allow.

• The professionals involved in patient care and the tasks these professionals undertake. Included in this cost category are staff members from a range of specialisms (beyond only doctors and nurses) for example chaplains, HCAs, physiotherapists, social workers and all others covered by overhead figures.

• The types of assessments or therapy patients have had while in hospital. This included general tests such as blood tests and X-rays or specialised assessments such as the Waterloo score to assess risk of pressure ulcers. Therapy may include physiotherapists’ interventions to encourage the patient to move, as some pain is caused by not moving about.

• The medication types and frequency, specifically with regard to pain, including times of drug rounds.

• The reasons for admission, that is, comorbidities. For example, trauma is a common reason for older patients to be admitted to hospital.

Data collection forms

The data collection forms were designed to capture health- and social-care resource use within the acute sector, and also included community-sector resource use. As highlighted, they were based on results from the analysis of data from the observational study by way of the researchers’ analysis workshops and previous data collection forms, including the CSRI and those produced within house for similar populations. Based on the results, including the importance of input from both formal and informal carers in recording resource use, and acting as a proxy for the person with dementia for completion of the health-related quality-of-life measures, the research instruments were designed to allow tick-box completion wherever possible. Three types of data collection form were developed:

1. a patient questionnaire

2. an informal carer ‘family or friends’ questionnaire

3. a formal carer ‘staff’ questionnaire.

The patient questionnaire was designed to be interview administered. Questions regarding resource use while in hospital are asked, including department, type of assessments and/or operations. The patient is also asked about the health-care services they have used in the last month while not in hospital, and previous hospital attendance prior to their current stay. As previously highlighted (see Aims and objectives) we included DEMQOL and EQ-5D prior to eliciting feedback from the researchers and LAG. The patient is asked to complete a 0–10 pain VAS, the EQ-5D-3L and the DEMQOL. The purpose of the supporter and formal carer questionnaires was to collect and validate information about the health of the person they provide support or care to. The EQ-5D-3L proxy and DEMQOL proxy, a 31-item interviewer-administered questionnaire is included.

Feedback from researchers and lay advisory group

The questionnaires were distributed among the researchers and the LAG. As there was no intervention developed, they could only comment on the acceptability in relation to the patient group.

One member felt that all questionnaires were quite lengthy and questioned whether or not anyone would take the time to complete them. They felt that a person with dementia would have great difficulty answering all the questions, and that completion of more than 10 questions would be unrealistic. Another member felt that background questions for the supporter about education and questions about employment and living situation were intrusive. Another suggested the questions about how many hours worked/sick days were intrusive.

Conclusion

Questionnaires for use in economic evaluations of a cohort of individuals with dementia were developed with input from researchers, HCPs and laypeople (see Appendix 7). Further testing is needed to assess whether or not the data collection forms need to be reduced, as per the concerns voiced in the feedback from researchers and LAG. There is typically a tension in collecting data between participant burden and having a core data set that allows the study question to be answered in a robust manner. It is unfortunate that, within the confines of this substudy, we were not able to undertake further assessment within a feasibility study, as further research is required to test the acceptability and suitability of the questionnaires and the method of data collection (e.g. interview administered or carer/self-completion). The initial feedback suggests that piloting of these research instruments is needed prior to use, and that adjustments may need to be made to make the forms easier to use and more acceptable to users.

The workshop also revealed the intensive staff input to direct patient care-related tasks, such as feeding, clothing, toilet care, supported walking and patient comforting, and the potential, or not, of input of informal carers within the acute sector. This suggests that a micro-costing study to assess this input would be a valuable part of any future feasibility study for this population.

Discussion of meta-review findings

The objective of our meta-review was to identify which tools are available to assess pain in adults with dementia, in which settings and patient population they have been used, and evaluate their reliability, validity, feasibility and clinical utility. We found a relatively large number of reviews on this topic and a considerable number of pain assessment tools available for the population with cognitive impairment. Of the reviews we evaluated, only one86 met all of the quality criteria outlined in the AMSTAR checklist, and would be considered a high-quality review. The review by Herr et al. 82 had an associated website in which 17 tools (specifically for use in nursing homes) are listed and commented on in detail. 124

The systematic reviews commonly situated their choice of interventions to review in the context of the challenges associated with assessing pain in patients who were cognitively impaired. Although there was recognition that the gold standard for pain assessment was the patient’s own assessment, this was unlikely to be possible with this patient population if the person was severely cognitively impaired, thus leading to the use of observational scales. The tools included in these reviews were therefore for the greater part observational, but showed a broad variation of measures and methods of assessing pain in adults with dementia. The foundation and focus of an observational scale is how the pain is manifested or made known, for example, on the basis on the AGS’s pain indicators. 48 An additional conceptual foundation resided on differentiation between aspects of pain, such as ‘the sensory-discriminative and motivational-affective aspects’. 125 Very few tools appeared to have any strong theoretical underpinning to the development.

Overall, there was no single tool that appeared to be more reliable and valid than the others. There was considerable variation in how reliability and validity of the tools were assessed. The majority of reliability and validity assessments were carried out on small samples in one or two different studies, so the applicability of tools across settings has yet to be evaluated. Similar conclusions can be drawn in relation to the feasibility and clinical utility of the tools. These findings have implications for research, which we briefly discuss over four points.

First, given the large number of existing tools identified by our meta-review, it seems inappropriate to develop further tools on the basis of the same conceptual foundations. Instead, researchers need to either envisage new assessment instruments on the basis of different conceptual foundations, or concentrate on extending the psychometric evidence base for the existing tools. There is a need to evaluate the tools on a wider scale, across a variety of patients and clinical settings, using rigorous methods and larger sample sizes. The studies need to ensure that the definitions of cognitive impairment and type of pain that is being assessed are clearly defined, to enable comparisons across populations. Although rigorous research is needed, it must be noted that the accuracy of tools is difficult to assess when an objective biological marker or other gold standard criterion is lacking – such as is the case for pain intensity. 85,86,126

Second, research should be conducted in clinical practice to assess the feasibility and clinical utility of the tools, and thus their potential for use in everyday clinical practice. Tools may be more useful in detecting relative changes in individual patients than differences between patients. 76 Tools that showed high reliability in research may not demonstrate high reliability in routine clinical practice if they are not administered as intended. 127 Pain assessment tools designed for research purposes, in order to aggregate and compare data across patients, do not necessarily transfer easily and effectively to clinical settings for everyday use. In general, it has been suggested that measurement tools developed for the purpose of evaluation of policy-making128,129 or routine general screening127 might have ‘no meaning’ at the frontline of care.

Third, research on the clinical utility of the tools should include evaluation of their impact in terms of choice of treatment and patient outcomes. This question was considered in one of the excluded reviews:126 the reviewers found no evaluations of the effect of protocols such as algorithms or pathways for assessing pain in inpatients (including those with dementia), and they reported ‘a study in cancer patients [that] found opioid-related over sedation and other adverse effects increased substantially after implementation of pain assessment on a numerical scale routinely with other vital signs’ (Helfand and Freeman,126 citing Hurley et al. 130). Herr et al. 82 reported that ‘the use of the ADD protocol was associated with a significant increase in the use of pharmacologic [. . .] and non-pharmacologic comfort interventions’. Alternatively, the use of the wrong pain assessment tool might reduce the likelihood that pain treatment will be initiated or contribute to acute exacerbations of pain (as could be inferred by data from Smith88) or may be found to have no effect on the quality of pain management. 131

Finally, an instrument must be relevant to ‘the condition, setting and participants in the health interaction, in particular the patient and clinical users’. 132 This includes the notion of user-centredness, defined as ‘the extent to which an instrument faithfully captures both the content of the health care user’s views and the ways in which their views are expressed’. 133 As a number of the systematic reviews and other literature makes clear, pain is a subjective experience, thus, the associated measurement gold standard is the patient’s own assessment. 66 However, where the patient is severely cognitively impaired, alternative ways to assess the patient’s level of pain must be found. Resort is made to behavioural indicators and groups of cues. 134 Ideally, these aspects would be used to assess expressed pain when the patient is at rest and when moving. Against this context, user-centredness, in terms of the actual patient’s self-assessment of pain, is unachievable. The next best option, from a user-centred perspective, becomes the assessment of a person who is most familiar with the patient in their everyday life in a hospital or other care setting – what Herk et al. 66 called a ‘silver standard’. It is critical, however, that the assessor has a high familiarity with the patient. It can plausibly be argued that the more this is achieved, the greater the likelihood of a closer fit of the proxy’s view with the patient’s own experience. At the same time, overfamiliarity may lead to an attenuation of the observer’s focus on potential pain clues. However, as Smith88 commented, one is rarely in ‘an ideal situation’ where the direct care providers know, and are thus highly familiar, with the care recipient’s personal habits and history. Thus, rather than seeking to review the degree of the user-centredness of tools in this area, it is important to evaluate the guidance on a tool’s use (that is who should administer the intervention and in what range of situations) and then explore whether or not it is actually used in the indicated way in everyday clinical practice. Exploration and findings on clinical utility and feasibility of use provide suggestive evidence of potential for use in everyday practice. Such evidence is a necessary but not sufficient condition for a tool’s actual use. It is notable that little or no insight is provided into the guidance over who should administer the tools in any of the systematic reviews. To gain insight into this aspect of use and possible additional evidence on a tool’s actual use in practice, one would need to examine the original instrument designers’ papers and validation studies, rather than rely on a systematic reviewer’s observations or summaries. Notwithstanding, it can be recommended that these two aspects of tool use (guidance and actual use) are incorporated into tools reviews and, given this patient group, development of evidence on the actual use of the tools in clinical practice.

Discussion of observational study findings

The overall aim of the observational study was to investigate how HCPs and others recognised, assessed and managed pain in patients with dementia in a range of acute settings.

The main information source was claimed to be the patient, as the sole individual with insight into their pain. However, cognitive impairment, communication difficulties, the organisation and context of the ward, all of which hindered access to this source of knowledge. As shown in other contexts of care,135 the communication difficulties experienced by patients with dementia were interactional in the sense that they ‘arise, in part, from their cognitive deficits’, but were also ‘occasioned by, or contingent on, the other’s contributions in interaction’. The hospital routines and environments generated a modality of interaction that challenged the communication abilities patients with dementia may have had. In fact, a similar finding was discussed about hospital inpatients, in general,136 that ‘the ward social system may also have an effect on patients’ communications [. . .] [in the sense that] the very structure of the work situation discourages patients from communicating clearly’. 136

The social context of the ward environment also critically shaped health professionals’ ability to recognise and respond appropriately to a patient’s communication of pain. The time frames for patient interaction, the number of patients on each ward and the expectations set up by the ward routines all influenced HCPs’ and HCAs’ abilities to perceive, recognise and manage a patient’s pain. Observational studies have revealed the complexity of pain management in acute settings and the barriers to optimal pain relief, such as staff’s attentiveness to pain cues, interruptions when assessing pain and reconciling varying interpretations of pain from multiple sources. 137,138 The typology of patients present in a ward – whether the ward was an admission unit, surgery ward or care of the elderly ward – affected the assumptions and expectations regarding pain and how it should be addressed. This finding echoes that of a study based in a surgical ward unit in the USA where ‘[p]ain assessment was rooted in a reference typology of clients based on surgical procedure’138 and nurses ‘expected’ a ‘certain kind of pain’. The type of patients in each ward influenced staff expectations about their pain and how it should be routinely treated. In turn, this routine also affected what could be expected about a patient’s pain and whether or not pain may be detected and recognised as pain (Figure 6).

FIGURE 6.

Systemic links between HCAs’/HCPs’ (individuals) perceptions and (organisational) routines. Reproduced from Lichtner et al. 100 © 2016 The Author(s). This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Hypothetico-deductive reasoning92 posits that clinicians would initially act on cues given by patients that might suggest pain was present. Being alert and receptive to a pain cue would be the first step in detecting and then managing the patients’ pain. However, the findings from this study showed that cues that were potentially indicative of the presence of pain could be missed or go unrecognised. Clisset et al. 139 defined ‘missed opportunities’ in dementia care as occasions when ‘opportunities presented themselves for HCPs to make some connection with the person with dementia but they seemed unable or unwilling to do so, often by choosing to end the interaction as quickly as possible’. In this study, missed opportunities manifested in several forms. For example, opportunities to detect, and manage, pain were missed when the patient was expressing pain but a clinician was not present (colocated) at that moment, when a pain cue was atypical of that which a clinician might usually interpret as pain, or when the cue did not seem to merit further investigation as it appeared to fall within acceptable limits.

Analysis of data also revealed organisational routines with an almost exclusive reliance on medication for managing pain. This type of approach is reliant on patients being able to communicate the presence of pain and changes in their pain after administration of medication. However, the communication difficulties caused by dementia drastically alter the effectiveness of this process. Other pain management interventions (such as repositioning, physiotherapy or physical activity, or patient engagement in meaningful activities) may offer the patients more opportunities to express and communicate their experience of pain and are included in best practice guidelines. 48 Thus it can be argued that reliance on pharmaceutical interventions – and the absence of any other forms of interventions – not only reduces the opportunities for patients with dementia to have their pain more effectively managed, but greatly limits the opportunities for HCPs to assess pain.

The central finding is that the recognition, assessment and management of pain in patients with dementia in hospital wards is a complex, dynamic, sense-making process, involving a number of information sources and individuals at different times and in different places (Figure 7). Pain cues generated by patients are interpreted by ward staff in light of their own knowledge, experiences and expectations, and are influenced by ward environments, which have their own routines and operate within a wider organisational culture.

FIGURE 7.

Complex, non-linear factors that influence pain recognition, assessment and management.

At present the timing, personnel and recording are disparate and it may be that a system which integrates all of these factors could significantly improve the structures, processes and outcomes of pain management for patients.

In terms of the health economics study, questionnaires for use in economic evaluations of a cohort of patients with dementia were developed with input from researchers, HCPs and laypeople. Further research is required to test the acceptability and suitability of the questionnaires. Depending on the structure of the intervention, any economic evaluation would be required to take account of the time and staff resources involved in pain assessment and delivery of pain relief.

Strengths and limitations

Health economics evaluation analysis

The work undertaken was important, generating preliminary data collection forms. The forms need piloting and refinement prior to use, but they constitute a strong base from which to further develop the economic evaluation of the proposed PADDS system intervention.

Implications of the studies

For research

The vast majority of the research concerned with managing pain in people with dementia has taken place in care homes, with most being undertaken in the USA, Europe and Australia. Very little research has considered issues encountered in busy acute hospital settings. Furthermore, most research in the field has focused on the development and validation of observational pain instruments. Less effort has been dedicated to the contextual factors influencing their use. Not only should suitable assessment tools be used, but also the relative contributions of factors such as improved use of time, improved multidisciplinary communication and a more varied range of resources for managing pain need to be understood.

There is a need for a valid and reliable observational pain tool with good feasibility and clinical utility to be available for use with people who have dementia. Many have already been developed, but rather than creating yet another, further research should be conducted on existing observational pain assessment tools to establish their psychometric properties, clinical utility and feasibility in acute settings and, provide much needed guidance on their appropriate use in clinical practice.

Realist studies of clinical practice where decision support tools have been implemented could explore the relevance of contextual contributory factors. This should be followed by clinical trials of the effectiveness of the interventions likely to have the most positive impact.

There is also a need for research to evaluate integrated approaches to pain management, considering behavioural symptoms, prescription patterns and institutionalisation, in order to embed pain management within the overall treatment and care a person receives throughout their journey with the condition. In particular, the role of carers in the interpretation of pain cues has considerable potential, and evidence on how this potential may best be developed in acute settings would be of value. The strong reliance on analgesic medications to manage pain is unsurprising in acute settings, but greater use of evidence-based complementary approaches may be effective in improving pain and quality of life. An evaluation of methods of implementation of such approaches, including education about, and structured use of, such interventions with patients who have dementia would be a helpful addition to knowledge in this field.

The key findings of the observational study were the three prerequisites of time, interdisciplinary communication/documentation and the availability of a range of pain management resources (Figure 8). Together, improvements in these areas should facilitate the creation of an ‘overall picture of pain’ to support clinical decision-making for optimum pain management. Recognising and assessing pain involves a degree of guesswork, underpinned by medical knowledge and experience (‘what is usual for this medical case?’), the most common ‘typologies’ of patients in each ward, and the type of ward (e.g. surgery or elderly care).

FIGURE 8.

The key elements to obtain a dynamic, patient-specific, overall picture of pain. Reproduced with permission from SAGE from Closs SJ, Lichtner V. Management of pain in peple with dementia: time for a change of approach. Pain News 2016;14:76–8, box 2. 156 © Authors.

When pain is recognised, it is understood through a dynamic sense-making process shared across individuals in the MDT in what staff, in this study, referred to as ‘a picture’ of the patient. This key finding has implications for the development of decision support interventions, in that they would need to ensure that they assist with staff identification of pain cues, in part through allowing sufficient time and adequate location to do so. Notably, individuals may have their own unique time-related patterns of pain, related to circadian rhythms, which are known to be disrupted in dementia, and the timing of the assessment would need to consider this.

Further research is needed to ascertain whether or not staff education to improve the understanding of the importance of giving sufficient/quality time to patients and patient-centred communication could improve pain recognition (and therefore management).

The creation and development of an overall picture of pain, specific for the patient, is both a prerequisite for planning and undertaking a trial of therapy and a result of this therapeutic trial. In the sites we studied there was a lack of connection between documentation of the information about decisions to act (decision to prescribe, decision to administer), the feedback from the action (e.g. outcome of administration of analgesics) and the reassessment that informs the ensuing decisions and/or actions. This lack of connection represented a considerable challenge to the understanding of the patient pain and effectiveness of therapy. On the part of staff, it involved the cognitive process of reassembling the pieces, requiring memory, ‘mental computation’ and time. Research is needed to ascertain whether or not centralising how all the relevant pain information is documented and displayed can save time and energy in reaching an overall picture of each patient’s pain.

Given a ward’s division of labour and organisation of work over shifts, effective documentation is essential for composing this ‘overall picture’ over time, each information item acquiring value when aggregated with and compared with others. Decision tools would need to assist in the integration of the distributed pain information, allowing a central point where the picture of a patient’s pain is created, over time and across individuals. Furthermore, it would have to assist with action and decision taking in a dynamic fashion that supports the trial and error nature of pain management and options for intervention.

The paper-based information systems currently in use for the distributed work processes involved in pain assessment and management did not facilitate the perception of such a holistic ‘picture of pain’. Single pain assessments at one point in time, even if they were done with the aid of one of the many pain assessment tools available, did not provide sufficient information about a patient’s pain; even when repeated they risked remaining as isolated information points because they were not connected in a common, accumulated picture of a patient’s pain and the effectiveness of any pain mitigating care given. Much of the literature concerned with the assessment and management of pain for people with cognitive impairment has focused on the development and validation of observational instruments designed for the purpose of assessment. 54 A number of studies have shown that these tools are not often used in clinical practice, a finding that was supported in our data. This may be because the tools do not facilitate the rapid creation of the ‘overall picture of pain’ and, therefore, do not assist clinicians with their decision-making about pain in this vulnerable group of patients.

Conclusions

The main conclusions derived from the project reported here which could guide future research studies were the following:

1. There are no existing observational pain assessment tools for patients with dementia that have yet been demonstrated to have good validity, reliability, feasibility and clinical utility. No single tool can be recommended in preference to any other for general use in hospital settings.

2. Future assessment tools should:

   1. have a strong theoretical underpinning

   2. elicit self-reports and identify cues from the patient first (at rest and on movement), then from those who are familiar with them.

3. Clear opportunities for interactions between patients, carers and staff are needed, allowing time for the identification and understanding of pain.

4. The influence of the social context of wards should be recognised, incorporating assumptions about pain according to patient type, etc.

5. The present reliance on medication provision (using trials) to alleviate pain should be supplemented with other non-pharmacological interventions.

6. Clear and effective communication between all the individuals involved in the care of the patient is needed.

7. Centralised records of all pain assessment and intervention information are needed.

8. Guidance on the use of assessment and/or decision support tools should be available.

Each of these conclusions needs to be evaluated in future research. It would be helpful to know how these suggestions for change may influence the recognition, assessment and management of pain for patients who have dementia. They are interlinked and while each may have its own impact and place within patient care, the interactions between them may also be significant.

The way forward

Our findings suggest that a reconceptualisation of how we approach the assessment and management of pain for patients with dementia in hospital is needed. Rather than relying on the traditional linear concept of assessing pain, providing an intervention and reassessing, a broader, more systemic approach is needed. We hypothesise that systemic thinking needs to replace the current focus on improving individual tasks (e.g. history-taking, pain intensity assessment, analgesic prescribing). This requires thinking in a different way about care, where, for example, a ward may be understood as a complex system containing many components (staff, patients, equipment, visitors, etc.) and there are linkages and interactions between the components that constitute the entirety of that defined system. Systems thinking of this kind has been defined in many ways,157 but should be goal-orientated (e.g. to ensure good pain management).

The patient’s self-report should still be used as the gold standard wherever possible, including careful conversations with patients about their pain and the judicious use of self-report pain scales if they are appropriate, usually for acute pain. This requires staff to ensure that they spend sufficient time with patients, and where feasible their carers, to identify the presence of different types and locations of pain; and that staff learn to elicit and understand patient self-reports of pain and to recognise individuals’ unique pain cues.

The pain-related information elicited by different staff and informal carers is not currently effectively communicated between all those involved, so strategies are needed to ensure that everyone has easy access to all the necessary information. Records of such information are currently kept in different forms in different locations and, consequently, individual records tend to be incomplete. These need to be centralised in some way so that comprehensive accounts of a patient’s pain are rapidly accessible by all staff.

The almost exclusive use of medication to alleviate pain should be documented and linked with pain reports over time. Analgesics could be supplemented with other non-pharmacological approaches, some of which may be benign and soothing (e.g. gentle massage) and others which may have a positive effect on rehabilitation, such as exercise therapy.

The use of observational pain assessment tools needs to be pragmatic, using the best (in the top seven of those reviewed), preferably locally available tools, which staff may be familiar with, increasing the likelihood of their use. Over time, one of the many tools available may be demonstrated to be superior to the others, at which point its adoption into practice should be considered.

The disparate activities described above need to be integrated into the complex, dynamic and multidisciplinary sense-making activity of hospital care. This led us (following analysis and debate reported in Chapter 7) to a proposal for a new kind of pain decision support intervention, using a more systemic approach, as outlined in the next section.

Proposal for a new decision support intervention

Given that structured, observational pain assessment tools do not appear, on their own, to provide the information needed for clinicians to understand and manage the pain of people with dementia, we are suggesting a broader approach.

Three key practical problems related to effective pain management for people with dementia, which are amenable to change, have been identified. These include patients’ pain not being noticed and understood as pain; a lack of integration of relevant information required for collective sense-making of the patient’s pain; and poor feedback mechanisms to inform the empirical (trial) approach to pain management.

It would be ambitious to attempt to change an entire ‘ward system’, so we are proposing an intervention which focuses on these three problem areas and makes links and interactions between key elements of the processes become visible, and rapidly accessible. This is the PADDS system tool.

The PADDS system tool would be a module designed to ‘slot into’ existing centralised electronic health record system systems, which would be strongly influenced by potential users during the codesign process. It would be accessible to all ward staff involved in the care of patients with dementia (and others, especially those with communication difficulties). It would be used to record different types of pain-related information elicited by different staff at different times. It would integrate all this information and produce detailed graphical displays of chronological events including assessments and interventions, alerts and reminders for assessments and interventions, options for interventions other than medications and narrative information from staff, family and friends. It is intended to be flexible enough to be implemented in a range of different contexts and organisational cultures, but requires a pre-existing electronic platform into which it may be integrated. A preliminary specification for the essential elements of the PADDS system is given in Appendix 6.

At this stage the focus is on the PADDS tool, which comprises an electronic information module to integrate and display all the relevant pain-related information. This is one part of the PADDS intervention, which is intended to include not only the PADDS tool but also an educational package to support staff, patients and carers in understanding (each to an appropriate extent) the rationale for and functions of the PADDS tool: the importance of knowing the patient; strategies for effective pain communication; the use of non-pharmacological approaches to pain management to augment the use of analgesics; and an implementation plan for incorporating the PADDS tool into routine practice.

Before wide implementation of the PADDS system in NHS hospitals can take place, its feasibility and impact should be evaluated, through initial pilot and then wider clinical trial. We propose that the PADDS system pilot and evaluation would comprise part of a complex intervention, being codesigned with ward staff in order to accommodate contextual issues, then introduced into the existing electronic ward health record alongside an education and implementation package. This would require a realist evaluation which would need to ascertain first, whether the PADDS system is associated with improved pain recognition, assessment and management for patients with dementia, and, second, whether or not staff find that the PADDS system is a useable and useful augmentation of existing electronic documentation systems. Within such a study the objectives would be to:

1. codesign the PADDS electronic pain module with NHS clinical staff and carers

2. codesign an implementation strategy for PADDS with users, with representation of each professional group and managers in the selected wards/units

3. develop an educational package to support the implementation of the PADDS tool by staff, patients and carers, along with skills in the use of non-pharmacological pain interventions

4. introduce the PADDS using the implementation strategy [designed in (b) above]

5. assess the feasibility of outcome measures of the PADDS, for use in a future trial of the intervention’s effectiveness

6. identify potential mechanisms through which the PADDS leads to these outcomes

7. understand potential contextual factors of PADDS use that are necessary for the mechanisms to be triggered

8. develop a programme theory to support a trial of a refined version of the PADDS, including the identification of contextual factors that need to be in place, and the contexts in which the PADDS can provide most benefit, to inform wider implementation and evaluation.

The meta-review and observational study have provided the evidence for a theoretical basis for the PADDS system which can then be developed and tested following Medical Research Council guidance for evaluation of complex interventions. 109 The implementation of the PADDS system has the potential to move practice towards widespread improvements in how the pain of people with dementia (and others) have their pain managed in hospital.

Lay advisory group

We are particularly grateful to our LAG, who were contacted via the AS. They were a participative group, ready to speak out, whose commitment was evident, and whose input was crucial in improving various aspects of the project as we went along. We would like to thank (alphabetically) Mrs Rosemary Copsey, Sandra Duggan, Andra Houchen, Mr U Hla Htay, Matt Murray, Andrew Pepper, Mrs Lynne Ramsay, Sylvia Wallach Squire, Sue Tucker and Barbara Di Vita. Three other members did not give permission to be named here.

Contributors

Our thanks to Sara Campbell and Nita Gorasia for contributing to data collection for the exploratory study and Lesley Jones (Advanced Nurse Practitioner), Greater Manchester West Mental Health NHS Foundation Trust, for providing expert advice and input on pain and dementia.

Study Steering Committee

We would like to thank the chairperson (Peter Crome) and members (Sandra Duggan, Roger Knaggs, Bridget Penhale and Pat Schofield) of the SSC for their wise counsel and support throughout the project.

Contributions of authors

S José Closs (Professor, Nursing Research, Pain Assessment) was the chief investigator for the project, contributed to securing the funding, the overall conduct of the study and led the writing of the final report for publication.

Dawn Dowding (Professor, Nursing, Clinicial Decision Making, Nursing Informatics) conceived the study, secured the funding, contributed to the systematic review, the analysis of the data from the observational study and the design of the decision support intervention.

Nick Allcock (Professor, Nursing and Pain Research) was the principal investigator leading research in one of the four sites where the exploratory study took place, contributed to the analysis of the data of the exploratory study and to the overall conduct of the project.

Claire Hulme (Professor, Health Economics) led the health economics aspects of the project.

John Keady (Professor, Older People’s Mental Health Nursing) was the principal investigator leading research in one of the four sites where the exploratory study took place, contributed to the analysis of the data of the exploratory study and to the overall conduct of the project.

Elizabeth L Sampson (Reader, Old Age Psychiatry and Dementia) was the principal investigator leading research in one of the four sites where the exploratory study took place, contributed to the analysis of the data of the exploratory study and to the overall conduct of the project.

Michelle Briggs (Professor, Nursing and Pain Research) led the systematic review, contributed to the analysis of the data of the exploratory study and to the overall conduct of the project.

Anne Corbett (Senior Lecturer, Dementia Research) led the user involvement, contributed to the writing of previous related publications and to the overall conduct of the project.

Philip Esterhuizen (Lecturer, Nursing Ethics) contributed to the systematic review and the exploratory study, with data collection and analysis.

John Holmes (Senior Lecturer, Liaison Psychiatry of Old Age) contributed to analysis of the data of the exploratory study.

Kirstin James (Occupational Therapist and Researcher, Acute Care Medicine; Older Person’s Well-Being) contributed to the exploratory study, to data collection and analysis, to the writing of previous related publications and the findings report, and to the overall conduct of the project.

Reena Lasrado (Research Associate, Mental Health Social Work) contributed to data analysis in the exploratory study, to the writing of previous related publications and to the overall conduct of the project.

Andrew Long (Professor, Health Systems Research) contributed to the systematic review, to the analysis of the data of the exploratory study and to the overall conduct of the project.

Elizabeth McGinnis (Nurse Consultant, Tissue Viability) contributed to the exploratory study by facilitating access to setting and recruitment.

John O’Dwyer (Research Officer, Health Economics/Economic Evaluation) contributed to the health economics aspects of the project and designed the health economics questionnaires.

Caroline Swarbrick (Research Fellow, Dementia and Ageing Research) participated in the design of the study, contributed to the exploratory study with data collection and analysis, to the writing of previous related publications and to the overall conduct of the project.

Valentina Lichtner (Senior Research Fellow, Health Informatics and Decision Making) led the project management of the four research centres and co-ordinated all project activities, contributed to the systematic review, the exploratory study, the data collection and analysis, and led the design of the decision support intervention.

Publications

The research described in this report has been presented in a range of publications and at a variety of conferences.

Data sharing statement

Data may be accessed via the corresponding author.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HS&DR programme or the Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HS&DR programme or the Department of Health.

List of included reviews

List of excluded reviews and reason for inclusion

Additional file 1: literature search

Details of the retrieval process, including databases searched, adjustments to the search strategy first optimised for the MEDLINE (via Ovid) database, for use in other databases, and detailed search strategies for each database.

Additional file 2: notes on screening titles abstracts

Details of the screening process on the basis of title and abstract of the records retrieved, how it was shared among the reviewers so that each record was screened by two independent reviewers, and the number of records referred for a third reviewer.

Additional file 3: notes on screening full text

Details of the screening process on the basis of the full text of the records retrieved, how it was shared among the reviewers so that each record was screened by two independent reviewers and the number of records referred for a third reviewer.

Additional file 4: data extraction forms

The data extraction forms used to build a data extraction database and extract data for the meta-review.

Additional file 5: summary of reviews

Overview of the included reviews, with a description of the aims and the methods used for including and analysing the studies, their quality assessment of the studies, and the tools examined.

Additional file 6: sample of studies’ methodological weaknesses identified by the reviews

A sample of studies’ methodological challenges and weaknesses identified in/by the reviews, organised by type of issue (e.g. sampling, tool design).

Additional file 7: summary of the tools – characteristics

Summary of the data extracted from the reviews providing a classification and description of the tools.

Additional file 8: summary of the tools – settings of use

Summary of the data extracted from the reviews pertaining to the setting, where the tools had been studied, tested or used for clinical practice.

Additional file 9: summary of the tools – reliability and validity

Summary of the data extracted from the reviews regarding the reliability and validity of each tool.

Additional file 10: summary of tools – content-validity-user centredness

Analysis of the data extracted from the reviews regarding conceptual foundation of each tool and ways content validity was explored. These aspects also pertain to the user-centredness of the tools.

Additional file 11: tools concurrent and criterion validity comparison table

Summary of the data on the concurrent and criterion validity of the tools, extracted from the reviews.

Additional file 12: summary of tools – feasibility

Summary of the data on the feasibility of each tool, extracted from the reviews, analysed in terms of ‘dimensions’ of feasibility: ease of use, time to complete, availability of instruction and guidelines, training required.

Additional file 13: summary of tools – clinical utility

A summary of the data on the clinical utility of each tool, extracted from the reviews, analysed in terms of ‘dimensions’ of clinical utility (availability of cut-off scores and interpretation of scores for decision-making) and overall evidence.

Contents

Summary

This document contains the specifications for a set of functional requirements for an electronic decision support system to assist with the recognition, assessment and management of pain, specifically for patients with dementia. We call this a Pain And Dementia Decision Support (PADDS) system. The system as outlined in these requirements is intended to be used primarily in hospital settings.

The requirements outlined in this document emerged out of an in-depth exploratory study conducted in three acute NHS trusts in England and one acute NHS board in Scotland, through research funded by the NIHR HSDR programme (HSDR – 11/2000/05). These specifications are intended as a first set of requirements to guide software development for the PADDS.

It is expected that in the first instance the PADDS system will be developed within a single hospital organisation which has ward-based EHRs. The PADDS system will then be tested for usability and feasibility.

These requirements specifications are to be revised, amended and completed once the system has been put in place and its use, usability and clinical utility evaluated.

1. About this document

2. Introduction

There has long been recognition that the assessment of pain in patients with dementia in hospital is challenging and difficult to manage effectively. Challenges are linked to the subjective nature of pain as an experience, patients’ difficulties in recall and interpretation of pain, behavioural signs of pain being altered in unexpected ways in patients with dementia, and barriers in communication. The context of hospital wards contributes to further complexity, as patients may not be well known to staff, and both the illness and the environment may create distress for the patient that adds to and confounds signs of pain. There are significant consequences related to the inadequate management of pain in hospital settings, including slower functional rehabilitation, longer hospital stays and lower quality of life, making improvements in this area of care urgent and important.

There is currently no single reliable mechanism or method that could be recommended for staff to use in hospitals for identifying and assessing pain in patients with dementia. It should be noted that recognising absence of pain is of equal importance in assessment, in order to avoid the unnecessary use of pain medications with associated side effects.

Decisions support systems (DSS) have been shown to improve care processes and patient outcomes in other clinical areas and are one of the ways in which pain assessment and management could be better supported in hospital settings. Traditionally DSS provide guidance on the basis of data about the patient and pre-set rules or algorithms, such as ‘cut-off’ points for generating alerts. In pain assessment and management of patients who are able to verbalise their pain, the patient pain intensity scores may be used, for example, to trigger an increased or reduced use of analgesics.

In the case of the assessment and management of pain in patients with dementia in hospital settings, given the inherent uncertainties about their pain and the limited staff knowledge of the patient, decision support systems cannot be based on standardised pre-set automated algorithms on the basis of inferred data points. Instead, the system needs to support HCPs judgements (recognition/assessments of pain) and decisions (management strategies) through:

1. The progressive, cumulative building of a ‘picture of pain’ that is specific for that patient that staff can use to ‘make sense’ of the patient’s signs as signs of pain (or not pain). This involves the meaningful display of an aggregated summary of all pain-related information gathered at different points in time by the different HCPs and HCAs involved in the care of the patient, through graphical displays that support individuals’ cognitive processes of pattern recognition.

2. Providing a reminder of the options for pain management that are inclusive of non-pharmacological interventions.

3. Providing a set of alerts on the basis of trends and unusual signs rather than activated by single individual data points.

This is what we call the PADDS system. In the following sections we first describe brief use cases for this system, to identify users, other systems, and activities involved with the PADDS; we then present what is required in the design of the system, for the system to work as intended (functional requirements specifications). Evidence and rationales in support of each of the requirements listed is provided in the appendix. Further information about the underlying rationale for PADDS is available in the documents listed at the front of this document (see Table 15).

PADDS is recommended for use with patients with dementia, but may be used with all patients, especially those with communication difficulties.

PADDS is recommended for use with patients with or without identified pain or established presence of pain, as a reminder at any point in time that patient’s signs ‘could be pain’.

3. Use and interconnections with other systems

The PADDS system will be used in two modalities:

1. PADDS will be used for the care of individual patients, to support recognition, assessment and management of their pain. This includes the identification of patient specific pain cues, the establishment of pain intensity scores for EWS systems, the assessment of the presence, causes, quality and intensity of pain, the decisions about analgesics or other pharmacological or non-pharmacological pain management strategies, the assessment of the effectiveness of these strategies.

2. PADDS will be used for the management of activities pertaining to all patients present in the ward at a given time for which PADDS is used; PADDS will work as a system HCPs can use to monitor the quality of care for these patients. In prioritising the implementation of systems requirements, those pertaining to this second modality can be considered optional.

PADDS is an electronic system embedded and interconnected with existing hospital EPR and ePrescribing/EPMA systems. It is envisaged that most of the data displayed in PADDS will be entered through other systems in use for the care of the patient. These include: ePrescribing/EPMA systems for prescription and administration of pharmaceutical pain management strategies; EWS systems for periodic observations of a patient’s pain; clinical notes for observation and assessment of patients’ that may involve signs of pain such as specialist nurse assessments, physiotherapists’ notes, medical rounds. Should any of these systems be paper-based at the time of implementation of PADDS in the hospital wards, there should be temporary arrangements for data to be copied from these paper charts to PADDS in near real-time. However, it is an essential requirement that data entry in PADDS is not proposed to HCPs as a documenting task in addition to existing practices of documentation to be continued in parallel; rather PADDS should replace or augment existing documentation tasks.

PADDS can be used together with ePrescribing systems by displaying the interconnection between the prescription (by doctors), the administration of the medicine (by nurses) and the patient’s response to it (as assessed by HCPs or HCAs). Developers of ePrescribing systems may want to consider the possibility of alerts generated through data available in PADDS appearing in the ePrescribing system at the time analgesics are prescribed. As this pertains to the design of the ePrescribing systems, further details on this are outside the scope of this document.

It is assumed that access to PADDS will be controlled in the same way that access to other hospital systems is controlled (e.g. with NHS Smartcards, username and password combinations). Access to PADDS should be granted through these information governance/access systems to all HCPs and HCAs who care for the patient. It is also assumed that actions performed within the system will be auditable.

Relevant information about the patient’s pain provided by family members/informal carers (‘carers’) MUST be included in PADDS. However, it is not envisaged at this stage that carers, nor patients, will have direct access to PADDS. The possibility of patient/carer access to PADDS SHOULD be explored as a further development of the PADDS system at a later stage, as a new release or system update.

It is assumed that the PADDS screen will display all the necessary demographic information about the patient as per all other hospital systems and that this information will be obtained via interfacing with other clinical or PAS systems. The specification for these remains outside the scope of this document.

Figures 9 and 10 illustrate the expected direct users of the system together with their goals in using this system, and the interconnections with other systems.

FIGURE 9.

The PADDS technical specifications: use cases (modality 1). IT, information technology.

FIGURE 10.

The PADDS technical specifications: use cases (modality 2). IT, information technology.

Note: These diagrams present a simplified generic range of users of PADDS and its connected systems identified through generic professional roles (e.g. nurses, doctors). A sample of possible interconnected systems are shown in boxes. Text displayed in ovals presents the main purposes for use of these systems. All these are not intended to be comprehensive; they need to be interpreted in the light of specifics staff roles and systems in use in each hospital context. Uses and use as represented in the diagram are not sequential (as in a workflow) but can happen in any order at any time of the patient stay in hospital. The dotted line for carers’ input to the system indicates carers’ access to the system should be granted – though this may be done on a subsequent version of the system.

4. Requirements

Each of the requirements listed below has a priority level assigned to it. Throughout this document, the following apply:

• Requirements containing the word MUST are mandatory; the PADDS system cannot be expected to work as intended if they are not implemented. Delivery priority: 1

• Requirements containing the word SHOULD are recommended; the PADDS system can begin to be put into use without them, while possibly implemented at a later stage. Delivery priority: 2

• Requirements containing the word MAY are intended as optional. Delivery priority: 3

5. Hypotheses and rationales for system design

Table 19 provides a summary of the supporting evidence and rationales associated with each of the requirements listed in this document. These rationales may serve as hypotheses to be verified with the evaluation of the PADDS system.

(PDF download)

Reproduced with permission. 174

(PDF download)

Reproduced with permission. 174

(PDF download)

Reproduced with permission. 174

(PDF download)

(PDF download)

(PDF download)

Reproduced with permission. 174

(PDF download)

Reproduced with permission. 174

(PDF download)

Reproduced with permission. 174

(PDF download)

(PDF download)

(PDF download)

Reproduced with permission. 174

(PDF download)

Reproduced with permission. 174

(PDF download)

Reproduced with permission. 174

(PDF download)

(PDF download)