Changing practice in dementia care in the community: developing and testing evidence-based interventions, from timely diagnosis to end of life (EVIDEM)

Demography and impact

There is increasing interest in earlier diagnosis, because of:

• the ageing of the population and the rising prevalence of dementia

• the costs of care for people with dementia, and

• the perceived benefits for the person with dementia (PWD) and their carers of early intervention, with concerns about delays in diagnosis, especially in primary care.

Dementia is one of the main causes of disability in later life; in terms of the Global Burden of Disease, it contributes 11.2% of all years lived with disability, higher than stroke (9.5%), musculoskeletal disorders (8.9%), heart disease (5%) and cancer (2.4%). 1 The total costs of caring for people with dementia in the UK have been estimated at between £17B and £18B per year2 – more than heart disease (£4B), stroke (£3B) and cancer (£2B).

Delayed recognition

Dementia syndromes are underdiagnosed and undertreated in primary care in all countries3^,4 with an estimated 50% of primary care patients of more than 65 years not diagnosed by their primary care physicians. 5,6 The complex reasons for this include patient, family, practitioner and service causes, and are discussed in detail elsewhere. 7

The evidence suggests that there are tangible benefits to earlier recognition of dementia. Early disclosure of the diagnosis seems to be what people with dementia want to have8 and younger professionals want to give. 9 The benefits of making a diagnosis include ending uncertainty about the cause of symptoms and behaviour change, with greater understanding of problems; giving access to appropriate support; promoting positive coping strategies; and facilitating the planning and fulfilment of short-term goals. 10,11 There is also the potential for using cholinesterase inhibitor medication to modify symptoms and delay the need to seek care home moves among people with dementia.

Enhancing skills in primary care

The insidious nature of dementia means that it is most likely to present as a problem within primary care, but there are obstacles to earlier recognition in this setting. Therefore, considerable efforts have been made to provide educational programmes to enhance the diagnostic skills of primary care practitioners. Because of the apparent time constraints in primary care consultations, much research focus has been on the development of brief screening tests for assessing cognitive function. However, despite the availability of user-friendly cognitive function tests, there has been little evidence of improvement in primary care recognition of, and response to, dementia syndromes over the decade since the introduction of cholinesterase inhibitors. The UK evidence is particularly compelling on this point, especially from early educational interventions,12 and recent analyses by members of the EVIDEM (Evidence-based Interventions in Dementia) programme of incidence and prevalence in a large GP data set. 13

We find persuasive the argument that the problem of underdiagnosis is probably due to the interaction of case complexity, pressure on time and the negative effects of reimbursement systems that do not reward time commitment and systematic follow-up. 14,15 However, in our view there is also considerable evidence that the main problem is not that primary care practitioners simply lack diagnostic skills, but that they lack the resources and management skills in both clinical management and in prioritisation of the needs of their patients with dementia. We have presented the evidence for this elsewhere. 16

The response of primary care to the needs of people with dementia may be due, in large part, to the limited availability of support services but this does not mean that an educational intervention is unlikely to have an effect. Improved confidence in clinical knowledge and skills is still needed to allow practitioners to improve their performance with this patient group. It does mean that the effects of an educational intervention would probably be weak, unless other obstacles to earlier diagnosis and better management of dementia were reduced. Education alone does not usually change practice.

Fortuitously, three changes in UK NHS funding, in the law and in NHS priorities, created an environment conducive to changes in practice and thereby opening an opportunity to test an educational intervention in optimal circumstances. In 2006, the introduction of the Quality and Outcomes Framework (QOF), led to financial incentives for primary care to establish a dementia register and carry out annual reviews of patients. This was followed by the implementation of the Mental Capacity Act (MCA, 2005)17 in 2007, which clarified the law on the assessment of capacity to make decisions. The third was the launch of the National Dementia Strategy (NDS) in 2009,18 which prioritised improvements in the care of people with dementia, along the entire disease trajectory, and introduced new resources (such as ‘dementia advisors’ in some areas). GPs came under increasing pressure from primary care trusts (PCTs) to make diagnoses earlier, and to limit the prescription of antipsychotic drugs for behavioural and psychological symptoms of dementia (BPSD). They also began to experience rising demand for advice about memory loss symptoms, and to more frequently encounter situations in which they had to make difficult decisions about the best interests of people with dementia.

The timing for an educational intervention around dementia diagnosis and management in primary care could hardly have been better. Practitioners were under pressure to improve the quality of the care they gave, and were reimbursed for doing so, but knew that they lacked knowledge and skills and so were motivated to learn. 19

Literature reviews

In order to inform the planned trial of an educational intervention designed to change clinical practice, we carried out two literature reviews. The first was a rapid appraisal of the published, English language literature on barriers to earlier diagnosis of dementia. The second was a review of interventions in primary care designed to alter clinical practice with patients with dementia.

Review 1: Investigating Barriers to Good Practice

A full account of this review appears elsewhere. 20

Method

Publications in English, up to August 2009, relating to barriers to the recognition of dementia were identified by a broad search strategy, using electronic databases MEDLINE, EMBASE and PsycINFO. Exclusion criteria included non-English language, studies about pharmacological interventions or screening instruments, and settings without primary care. Figure 1 shows the selection process for papers.

FIGURE 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart for EVIDEM-ED: Review 1, Investigating Barriers to Good Practice.

Results

Eleven empirical studies21–31 were found (see Appendix 1). The main themes from the qualitative studies were lack of support, time constraints, financial constraints, stigma, diagnostic uncertainty and disclosing the diagnosis. Quantitative studies yielded diverse results about knowledge, service support, time constraints and confidence. The factors identified in qualitative and quantitative studies were grouped into three categories: patient factors, GP factors and system characteristics. These are discussed in detail in the published review. 20

Conclusion

Much still needs to be done in service development and provision, GP training and education, and the eradication of stigma attached to dementia to improve the early detection and management of dementia. Implementation of dementia strategies should include attention to all three barriers, and further research should focus on their interaction.

Review 2: Changing Service Provision and Clinical Practice

We conducted a second review of potential solutions to the problem of underperformance in primary care. This second review aimed to identify and appraise empirical studies of interventions designed to improve the performance of primary care practitioners. A full account appears elsewhere. 32

Methods

Search strategy

A rapid appraisal approach was adopted to inform the implementation of the NDS in England,18 introduced in 2009. To avoid delays in the analysis and completion of the review, papers that were not easily accessible were not included. Publications about the detection and management of dementia in the community were searched for using MEDLINE, EMBASE and PsycINFO, without restricting the date or language of publication. Searches were carried out up to February 2010. A broad search strategy was adopted and search terms are listed in Box 1. This was executed as a two-step process with educat* added as the second step, after all of the other search terms. Bibliographies of articles discovered were also examined for additional relevant literature.

BOX 1 - Description of two-step electronic database search process

Dementia OR Cognitiv* Impair* OR Alzheimer’s Disease

AND

Primary Care OR General Practi* OR Family Pract*

AND

Diagnos* OR Manag*

AND

Educat*

Search selection

This search resulted in a total of 4579 articles. To prevent narrowing of the search scope and therefore potentially reducing the search sensitivity, the terms were not refined, but instead each title was reviewed with its abstract if available to ascertain its relevance. A number of exclusion criteria were then applied. All studies of pharmacological interventions were excluded, as were studies of the performance of cognitive function tests. Publications were excluded if they reported on the diagnosis or treatment of dementia in anywhere but a primary care setting (for example, the benefits of respite in long-term care facilities); if they related to a population outside the scope of this review (for example, interventions for caregiver health); or if they were clinical discussions about dementia diagnoses or care. Letters were also excluded, as were publications in languages other than English. Two other relevant articles were found, one directly from bibliographic searching, and a second was sourced following a recommendation from an expert in the field. Figure 2 shows the search process, presented according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. 33

FIGURE 2.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow chart for EVIDEM-ED: Review 2, Changing Service Provision and General Practice.

After applying these criteria, 162 articles remained, but 70 of these were not readily obtainable and these were also excluded for pragmatic reasons; 77 of the remaining 92 papers were excluded after analysis because they were clinical reviews, guidelines, studies of barriers or studies without interventions that were designed to change practice. References in the 92 papers were reviewed to identify other intervention studies.

Data extraction

Data were extracted from each study report to allow comparison of interventions and to assess the quality of study designs. The characteristics chosen for comparison (including location and type of study, size, recruitment process, methodology, outcome measures, results and conclusions) and a detailed description of each intervention can be found in the published review. 32 Randomised trials were assessed for quality using the Physiotherapy Evidence Database (PEDro) scale. 34 The PEDro scale was chosen because its assessment of blinding is very relevant to empirical studies in dementia. In quality-rating scales that consider double-blinding as the central methodological issue, studies would lose points for failing to be blinded. The PEDro rating scale divides blinding into participants, therapists and assessors, recognising that although not all of the components of trials can be blinded, it is preferable that some groups are blinded rather than none at all. 35

Results

Fifteen studies were identified of which one used qualitative methods only and three were unpublished. Of the 11 studies29–31 included in the review, 10 were RCTs (see Appendix 1). Six21–24,27,30 reported educational interventions and five25,26,28,29,31 trialled service redesigns, either by changing the service pathway or by introducing case management. Educationally, only facilitated sessions and decision-support software improved GPs’ diagnosis of dementia, as did trials of service pathway modification. Some of the case management trials showed improved stakeholder satisfaction, decreased symptoms, and care that was more concordant with clinical guidelines.

Conclusion

The quality of the studies varied considerably. Education interventions are effective when learners are able to set their own educational agenda. Although modifying the service pathway and using case management can assist in several aspects of dementia care, these would require the provision of extra resources, and their value is yet to be tested in different health systems.

Developing the educational intervention

The theoretical basis of the intervention is described in full elsewhere. 16 In summary, the educational intervention was constructed using the following principles:

1. Diagnosis in primary care depends upon pattern recognition and the use of ‘illness scripts’ (more or less complex representations of diseases) in non-linear pathways to find the most probable explanations for presenting symptoms.

2. The first stage of diagnosing is not testing, but starting to suspect the possibility that a dementia syndrome may be emerging (the trigger phase). The doctor needs an index of suspicion to construct a diagnosis, which is often ‘triggered’ by a symptom within the patient’s story.

3. A specific problem with dementia is that all involved – patients, families and GPs – are reluctant to diagnose dementia, a serious and largely unmodifiable disease, which carries a huge burden of stigma.

4. The diagnostic task is complex, therefore, because the accumulating cognitive impairments are occurring within complex personalities, themselves embedded in social relationships.

5. Any educational intervention needs to be flexible enough to correct context-specific deficits. In other words, there is no ‘cook book’ or ‘one size fits all’ grand intervention that will alter clinical behaviour in primary care.

A workplace approach

In primary care settings, the most effective educational methods hinge on adult learning approaches – problem-solving, small group work, solution focus, ‘academic detailing’ (on site rather than off site) – that permit flexibility in learning and allow adaptation of guidelines to local circumstances.

A co-design approach to the production of an Educational Needs Assessment (ENA) tool was adopted in order to gain the insights and experiences of a range of practitioners. 37 This involved an expert group of designers and an expert panel of ‘critical friends’ working in an iterative technology development process38 to develop a prototype ENA tool for dementia diagnosis and management. The prototype was refined and subsequently ‘field tested’ with volunteer practices. A full description of this process can be found elsewhere. 39

The expert group

The multidisciplinary expert group was made up of three GPs, an old age psychiatrist, a carer and a psychologist, and was attended by three members of the research team. Members of the expert group were chosen on the basis of their expertise or experience in dementia care, or in professional education, or both. The aim of the expert group was to decide which skills and attributes were essential for a primary care team to possess in order to deliver effective care for patients with dementia.

The expert group was given three objectives: to (1) design a care pathway that would assist practitioners in earlier diagnosis and enhance subsequent clinical management (the care pathway); (2) identify the attributes a practice would need to implement the care pathway (the task matrix); and (3) use the task matrix to derive a set of questions that would identify the practice’s learning needs (the ENA).

The expert group was also asked to frame its work in terms of adult learning approaches, in other words, that learning would be problem-solving, case based and usable by practitioners at different points on the spectrum from novice to expert. 40,41

A modified nominal group technique was used with the expert group to develop the prototype ENA tool. Nominal groups are potentially powerful learning and development tools. 42 They have a particularly useful role in analysing health-care problems43 and can help bridge the gap between researchers and practitioners. 44

The expert group met initially to decide what the elements of a training programme designed to change practice in dementia should be. This meant identifying what tasks were needed to be performed in order to identify patients with dementia, and to care for them appropriately in the primary care setting. The expert group designed a flow chart and identified elements of the diagnostic process outlining the pathway that a clinician might follow once they suspect that a patient has dementia (see Appendix 3, Figure 29). In addition, the elements of a training programme for primary care teams were identified (see Appendix 4). From this, the expert group developed questions that would identify the practice’s strengths and weaknesses in the care of people with dementia.

This cyclical process of adapting and refining ideas took place over one calendar year, and necessitated four meetings of the expert group. The prototype ENA was sent to the expert panel after the fourth meeting of the expert group.

The expert panel

The expert panel comprised a group of external, independent people who had registered their interest in the EVIDEM project by subscribing to a mailing list on the website. The expert panel had 13 members (one of whom dropped out in the course of the development process), with a mix of carers, patients, and professionals, including two GPs, a social worker, a practice nurse and an Admiral Nurse (a community nurse specialising in support of people with dementia and carers). Seven panel members were carers of people with dementia.

Expert panel members were blinded to each other as well as to the expert group members. They received and sent comments on all the documents by e-mail or post. The purpose of the panel was to review the proposals made by the expert group to ensure that no themes were omitted and to decide how comprehensive, valid and feasible the ENA was as a tool. They were also charged with assessing if the development of the prototype concurred with known factors favouring the adoption of an innovation, using literature that had been circulated to them in advance. When this had been completed the expert panel returned their comments and suggestions to the expert group for review.

Field-testing the Educational Needs Assessment tool

After incorporation of the feedback from the expert panel, a prototype version of the needs assessment tool suitable for field testing was prepared. This is shown in Table 1.

This ENA template was then field tested in volunteer general practices. The practices were based in north west and north east London, and were recruited directly by the EVIDEM programme as part of a RCT of an educational intervention. Practices were informed about the process and asked to choose either field-test status or RCT participation. The first five respondents wanting to join the field test were enrolled.

To carry out an ENA, two members of the research team held a group meeting at each practice. A team member experienced in group-based learning in general practice (the expert tutor), facilitated discussion of the needs assessment tool. The other acted as a participant observer, ensuring that all questions were asked and clarifying points where necessary, as well as taking notes about the assessment process. The practices were asked to invite whichever members of staff they thought should participate, including attached staff from community services. The times of meetings were left to the practice to choose. The research project paid for lunch when lunchtime meetings were preferred.

An educational prescription was agreed at the end of each needs assessment process, and up to three follow-up visits were arranged to work through the themes identified for the educational prescription. The educational prescriptions were used to collect and collate learning materials for each practice, and the subsequent workshops were also used to revisit and revise the questions in the ENA. The research team assembled material in written and electronic form for each of the items on the educational prescription workshops led by the expert tutor who had facilitated the ENA process, again with a participant observer from the research team.

Table 2 shows the characteristics of the practices in the field test, the participation of different disciplines, the number of workshops held with each practice, and the themes identified in their ‘educational prescription’.

As a result of the field testing, the variety of learning materials used was broadened to include more reference material for use during sessions. The timing planned for topics was also amended and the expert tutors became more knowledgeable and aware of areas of need that were consistent across individuals and groups. The practices involved in the field test made it clear that they did not want decision-support software embedded in their electronic medical record systems but instead preferred simple, easily accessible guidance in an electronic format that functioned as ‘look-up’ documents.

The educational prescriptions developed through the ENA appeared acceptable and useful in volunteer practices. The time commitment (no more than 4 hours, spread out at the practice’s discretion) appeared manageable. The pilot group of practices prioritised diagnosis, assessment of carers’ needs, quality markers for dementia care in general practice, and the implications of the MCA 200517 for clinical practice. The content of the ENA seemed to be comprehensive, in that no new topics were identified by practices in the field trial. On the basis of this pilot, the ENA tool was used in the full trial.

Developing a learner’s manual

The themes identified by practitioners in the field test were researched by EVIDEM staff and a learner’s manual was constructed (available via www.EVIDEM.org.uk). This manual was designed in a modular form with Microsoft Word version 97–2004 (Microsoft Corporation, Redmond, WA, USA) documents that could be accessed during consultations if necessary. The content of the learner’s manual was developed using the expertise of all members of the EVIDEM team. The manual contains modules on the following:

• dementia diagnosis and cognitive assessment tools

• differential diagnosis of dementia by illness and types

• disclosure of dementia diagnosis and cultural diversity

• use of cholinesterase inhibitors, NICE/Social Care Institute for Excellence (SCIE) guidelines and shared care of medication

• management of BPSD and Department of Health (DH) guidelines for using antipsychotic drugs

• non-pharmacological interventions for dementia

• medication review and dementia

• needs of people with dementia and primary care

• case studies for management of people with dementia

• carer needs and support in general practice

• end-of-life care

• holistic care in dementia – personhood approach

• MCA, advance directives and Lasting Power of Attorney (LPA)

• Independent Mental Capacity Advocates (IMCAs) and Court of Protection

• financial and legal guidance: where to go for help

• dementia and driving.

Trial design

The effectiveness of an educational intervention combining practice-based workshops and computer-based support was tested in a pragmatic, unblinded, cluster RCT, with a pre and post design, and with two arms: usual care compared with educational intervention [see CONSORT (Consolidated Standards of Reporting Trials) diagram in Figure 3]. The researchers were aware of group allocation but carers and people with dementia were not informed. A full description of the trial design – including management structure, participant involvement and patient and public involvement (PPI) – is available in the published protocol. 45

FIGURE 3.

EVIDEM-ED: CONSORT flow diagram.

Standard significance tests assume that random sampling has taken place, and that the behaviour or knowledge of any one individual is not affected by others in the sample. However, this study, like many in the health service field, is based on clusters of individuals who may influence each other – in this case colleagues in a practice who share information or similar views, or patients who receive similar treatment. 46 This ‘clustering effect’ can mean that false conclusions are drawn about relationships in the data. This effect was controlled for by using cluster membership (i.e. respondents in each practice).

Study setting

The study took place within primary care practices in the geographical area covered by the North Thames (NT) Dementias and Neurodegenerative Diseases Research Network (DeNDRoN) – Metropolitan North London, Essex, Hertfordshire and Bedfordshire. Approval for the trial was received from Southampton & South West Hampshire Research Ethics Committee (REC) (A): reference 09/H0502/77.

The educational intervention

The educational intervention consisted of practice-based workshops with a tailored curriculum designed by a multidisciplinary expert group and supplemented by electronic resources. The educational interventions reflected different approaches to adult learning, namely workshops directly relevant to clinical practice, allowing learning to occur through peer reflection about real cases, and electronic resource materials suitable for ‘real-time’ use in consultations.

Recruitment

Outcome measurements

Measurement of unmet needs in patients and carers

Unmet needs were captured using a structured interview schedule with carers, developed by the research team in a previous trial. 21

Consent

People with dementia were identified by practice managers and their lead clinician checked whether or not they fulfilled the inclusion criteria. Before seeking consent from patients to participate in the study, practitioners were asked for their opinion about the capacity of the person with dementia to give informed consent to taking part in the study, using the MCA 200517 as the framework for their judgement.

For patients judged as having decision-making capacity, the following information was posted to the person with dementia: (1) a covering letter explaining the involvement of the practice and signed by the lead clinician; (2) a participant information sheet; and (3) a response letter and prepaid envelope to be returned to the research team. A researcher arranged to see those patients with dementia and their carers who expressed an interest in participating in the study. This encounter took place at the preference of the carer, at the patient’s home, in the GP practice or at the research team’s offices at the Royal Free Hospital. Its purpose was to seek consent from the carer and consent/assent from the person with dementia, where possible, for their participation in the study and for allowing access to patient records.

For those judged by the lead clinician to lack capacity to give informed consent, a consultee (as defined in the MCA 200517) was identified and consulted about possible involvement of the person with dementia in the trial. The relevant clinician wrote to the consultee, providing full information about the trial.

Figure 4 shows the process of identifying participants and how they were approached, recruited and consented/consulted about EVIDEM-ED.

FIGURE 4.

Flow chart of how participants were identified, approached, recruited and consented/consulted about EVIDEM-ED (2009–10).

Statistical analyses

Secondary outcome analysis

An in-depth study of the medical records of 167 people with dementia was carried out to capture the detail of clinical management before and after diagnosis. Nineteen of the 23 practices in the trial invited patients with dementia and/or their carers to permit analysis of medical records; four practices were unable to send invitations at baseline, for organisational reasons. The total number of people invited was 763, of whom 190 expressed an interest (25%) and, after further discussion with researchers, 167 (22%) allowed review of medical records. The medical records of some of the 167 participants were incomplete because they had joined their practice after the index consultation, or after diagnosis, and the previous records had not reached their new practice. One hundred and sixty-one had documentary evidence of a formal diagnosis, and 136 had documentary evidence of the index consultation and a formal diagnosis. Figure 5 shows the characteristics of the secondary outcome sample.

FIGURE 5.

EVIDEM-ED: patient medical record data characteristics.

Symptoms were recorded at index consultation for 101 patients, and, of those, 39% had one or two symptoms, 45% had three or four symptoms, and 17% had five or more symptoms recorded.

Professionals’ knowledge and skills analysis

Carers’ perceptions

Results

Qualitative results

Questions with open-ended responses were included in the interview. Box 2 displays a summary of three themes arising; perceptions of support from primary care, expectations of primary care, and the need for proactive care. Full analysis of the qualitative data will be reported in a separate publication.

Overall, few carers had contact with primary care and had relatively low expectations of their GP. However, those who did have contact with the practice were generally satisfied with the care they received.

Limitations of the study

It is possible that the medical records did not capture important changes in the quality of management after the educational intervention, which were not captured in our evaluation of secondary outcomes, but our creation of a category of ‘opportunistic dementia review’ fitted with clinical practice and allowed a generous interpretation of clinical activity. Many patients with dementia joined or left during the pre and post periods, truncating the data collection time so that the length of follow-up may have been too short to capture a difference. The number of dementia management reviews was higher in the non-intervention group at baseline; this could be the consequence of one practice in the usual care group having medical responsibility for several care homes. Distribution of newsletters and guidelines to usual care arm practices may have had an effect on their patients’ behaviour.

The study took place in south east England with practices that were innovative early adopters, not typical general practices, and local educational programmes that had been developed to implement the NDS may have influenced practice activities, although we found no evidence of this. The volunteer practices were probably different from others, in that they wanted to take part in a pilot educational programme about dementia. However, the results of this study have wider implications, particularly about the value of tailoring educational interventions.

The intervention was developed in ways consistent with current understanding of how effective interventions are made. Nevertheless, there may have been deficiencies in the development process. For example, the views of people with dementia and their family carers may not have had sufficient weight. Some professional perspectives may have been too powerful; the absence of response from expert panel members to the ENA questions themselves could be a sign of this. The expert group may not have used the expert panel’s critical comments sufficiently, resulting in an oversimplification of the ENA. Finally, the expert tutor may have had an effect on the use of the ENA and the ‘filling’ of the educational prescription, even although we used a participant observer to avoid idiosyncratic interpretations of the group discussions.

Conclusions

This study suggests that a tailored educational intervention aimed at GPs does not improve dementia case identification or documentation of clinical management for people with dementia, even when policy pressure encourage changes in clinical practice and the reimbursement system rewards it.

Components of the study

The first question was addressed through a review of the literature; this informed the focus and design of the clinical trial to answer the second and third research questions (ISRCTN01423159). As a consequence of unforeseen difficulties with recruitment, a fourth question arose: Why is it difficult to recruit participants from a seemingly large population and supportive clinical service? To answer this question we added a small substudy that incorporated a focus group with recruiting clinicians.

Literature review

Methods

A rapid appraisal approach was adopted to inform the design and implementation of a RCT of exercise as therapy for BPSD. This review, rather than systematically aggregating data, has adopted a critical interpretive approach,77 the purpose of which is to construct theories grounded in research, develop an understanding about the potential effect of exercise on BPSD and to generate practical methods to evaluate these effects. This broad aim led to the generation of three research questions: (1) Does exercise ameliorate BPSD?; (2) How has exercise been conceptualised and how does this relate to the findings? and (3) What are the main limitations and methodological shortcomings of recent studies? Full details of the methods used in this literature review were published. 78

Inclusion/exclusion criteria

Although the aim of this review was to be as inclusive as possible with the current relevant literature, we recognised the importance of setting boundaries concerning papers to include. 77 Studies and review articles were selected that met the following criteria:

1. Individual studies must measure the efficacy of exercise as a therapy for BPSD. Exercise is defined as physical activity that is a planned, structured and repetitive programme.

2. Review articles must examine intervention studies assessing the efficacy of exercise as a therapy for BPSD.

3. Papers must have been published in peer-reviewed journals using relevant keywords appearing as an important factor in the title or abstract. Studies examining combined interventions (i.e. combining exercise with another psychosocial intervention) were excluded, as these are unable to establish the absolute effect of exercise.

The database search was constructed using the following strategies and searched up to January 2011.

• MEDLINE, EMBASE, PsycINFO and PubMed were searched using a combination of keywords: ‘physical activity and dementia/Alzheimer’s disease’, ‘exercise and dementia/Alzheimer’s disease’, ‘non-pharmacological interventions and dementia’, ‘exercise, sleep disturbance and dementia/Alzheimer’s disease’, ‘exercise, apathy and dementia/Alzheimer’s disease’, ‘exercise, aggression and dementia/Alzheimer’s disease’, ‘exercise, wandering/repetitive behaviours and dementia/Alzheimer’s disease’, ‘exercise, depression/mood and dementia/Alzheimer’s disease’.

• Other sources were also searched including reference lists and book chapters.

Results

Seven hundred and twenty-three articles were screened, of which 16 met the inclusion criteria. Of these, 10 were review articles and six were additional papers not included in these reviews (see the PRISMA chart in Figure 6 for the derivation of the selected papers). We examined each paper included in the reviews to assess their inclusiveness and their contribution to answering our three questions.

FIGURE 6.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow chart describing the search process of finding articles examining the efficacy of exercise on BPSD.

1. Does exercise reduce BPSD? Using very stringent inclusion criteria, the Cochrane review by Forbes et al. 79 concluded that there is insufficient evidence of the effectiveness of physical activity programmes on affective symptoms in older persons with dementia to warrant commissioning new services or decommissioning existing ones. They argue that until further methodologically robust studies are conducted and published, making firm recommendations for clinical practice about the importance, type and duration of exercise as a therapy for BPSD will not be possible. The existing research literature, although thin, does offer insights into potential effects of physical activity on BPSD and hints at the impact (or not) of exercise on behavioural and psychological symptoms not included in the Cochrane review (e.g. sleep disruptions, wandering, repetitive behaviours, anxiety and apathy).

2. How has exercise been conceptualised and how does this relate to the findings? Exercise programmes for people with dementia have been poorly conceptualised and it is unclear which aspects of physical activity (e.g. type and duration) provide better results than others. We need to understand further what behavioural and psychological symptoms respond and to what type of physical activity intervention, and how exercise may work, for whom and under what circumstances. Indeed, exercise may affect various BPSD in different ways. The current evidence cautiously indicates that its effect on some symptoms (e.g. agitation, anxiety) may be more rapid than on others (e.g. depression).

3. What are the main limitations and methodological shortcomings of recent studies? The methodological shortcomings of current work in this area are substantial – very few studies have robust methodologies. Few use randomised controlled designs and few have a sample size with enough power. In addition, many studies have short follow-up periods, which could potentially produce a type 2 error, that is, an effect of exercise on some symptoms may not be evident after only a short period. Many also rely on findings from mixed dementia groups while ignoring the potential differences between different types of dementias. Moreover, the intervention itself was often a heterogeneous one using different modalities of exercise (i.e. strength building, cardiovascular fitness or flexibility) and/or combined with other interventions (e.g. carer education). In such instances it remains unclear which particular components of an intervention are effective.

Trial objectives

Method

Recruitment and baseline assessment of participants

Participants with a clinical diagnosis of dementia or ‘suspected dementia’ with at least one significant BPSD defined by the NPI62 (excluding the domains of delusions and hallucinations) were eligible for the trial (Figure 7). Diagnosis of dementia was confirmed using International Classification of Diseases, Tenth Edition (ICD-10) Diagnostic Criteria for Research (DCR-10). 80 A risk assessment was performed to assess the suitability of participants for the intervention at baseline. This assessment included measurement for risk of falls using the Falls Risk Assessment Tool (FRAT)81 and Timed Unsupported Steady Standing (TUSS) assessment. 82

FIGURE 7.

EVIDEM-E: CONSORT diagram.

Findings

Description of the economic analysis subsample

Complete CSRIs were received from 74 dyads at baseline and 67 dyads at the 12-week follow-up. However, depending on the outcome variable, the matching sample for the economic analysis varied from 49 to 52 dyads. Using multiple imputation techniques we were able to achieve a sample size of 52 dyads (22 control; 30 intervention). Hereafter, our findings on service use, costs and cost-effectiveness are based on the 52 dyads in this subsample, which is just less than half of the sample available for the main outcome analysis (116 dyads).

Table 28 presents descriptive statistics for the economic analysis subsample. We tested if the economic analysis subsample was significantly different from the original sample of 131 dyads by using the Wilcoxon rank-sum test for continuous variables and the Fisher’s exact test (or Pearson’s chi-squared test when appropriate) for binary variables. We found no significant differences between the ‘economic subsample’ and the full sample at 95% level of confidence, although a p-value of 0.07 was found for the difference in MMSE score at baseline. Using the same method we compared the intervention and control groups, and, again, did not find any significant baseline difference at 95%, with the smallest p-value (again equal to 0.07) being associated with the ‘primary education or less’ variable, implying that people included in the final economic analyses sample were perhaps less likely to have a limited education.

Cost-effectiveness analysis

In Table 31 we report the incremental costs and incremental effects for the primary and secondary outcomes.

TABLE 31 - Incremental costs and effects (controlling for baseline costs and outcome)

Note: Positive variations in the incremental effects represent improved outcomes. Minor differences in incremental effects between perspectives are a consequence of the incremental cost-and-effect equations being jointly estimated with the SUR method. 92

Perspective	Incremental cost (£, 2010–11): mean, 95% bootstrap CI			Incremental effect: mean, 95% bootstrap CI			ICER
	Mean	Upper CI	Lower CI	Mean	Upper CI	Lower CI
HSC: 0–12 weeks
NPI	–168.6	–1232.8	895.6	4.07	–4.65	12.79	Intervention dominant
ZBI	–170.8	–1234.6	893.1	1.54	–1.78	4.86
DEMQOL-Proxy	–165.6	–1251.7	920.6	2.87	–1.94	7.68
QALY (DEMQOL-Proxy)	–169.7	–1240.0	900.5	0.0055	–0.0031	0.0140
GHQ	–173.6	–1235.8	888.6	4.19	–0.55	8.93
Societal: 0–12 weeks
NPI	1686.4	–1407.1	4780.0	4.01	–4.72	12.73	421
ZBI	1641.1	–1497.8	4780.0	1.56	–1.75	4.86	1055
DEMQOL-Proxy	1635.9	–1520.9	4792.6	2.82	–1.97	7.61	580
QALY (DEMQOL-Proxy)	1565.8	–1592.6	4724.2	0.0055	–0.0031	0.0140	286,440
GHQ	1657.3	–1471.8	4786.4	4.23	–0.50	8.97	392

Incremental cost-effectiveness ratios would typically be calculated by dividing incremental costs by incremental benefits, but these have not been presented in Table 31 because neither incremental costs nor incremental effects were significantly different between trial arms. From a HSC perspective, we found that incremental costs were negative, that is, the intervention group had lower costs than the control group (about £170 less) and incremental effects were positive, that is, the intervention group achieved better outcomes. Although none of these differences were significant at the 95% level, for illustrative purposes we note that, from a societal perspective, the ICER for the NPI outcome measure was £421. Figures 8 and 9 show CEACs for the primary outcome measure (NPI) from a HSC and societal perspective, respectively.

FIGURE 8.

Cost-effectiveness acceptability curve: exercise regimen vs. usual care; HSC perspective, with effectiveness measured on the NPI scale.

FIGURE 9.

Cost-effectiveness acceptability curve: exercise regimen vs. usual care; societal perspective, with effectiveness measured on the NPI scale.

From a HSC perspective (see Figure 8) the CEAC suggests that, at a willingness to pay of £500 per incremental improvement in outcome (i.e. per 1-point difference in NPI score), the exercise regimen is cost-effective with probability of > 80%. From a societal perspective (see Figure 9), the ICER was £421 per incremental difference in NPI score. If a reduction of at least eight points in the NPI score can be considered clinically meaningful95 then this result suggests that the cost of achieving a meaningful improvement is £3368.

Table 32 presents results for a sensitivity analysis that, in addition to baseline costs and baseline outcomes, also controlled for participant age at baseline, gender, ethnicity, marital status, education level, whether living in a care home, MMSE score at baseline, carer age and gender. This analysis gave results that are consistent with the main analysis.

TABLE 32 - Incremental costs and effects (controlling for sociodemographics, baseline costs and outcome)

Note: Positive variations in the incremental effects represent improved outcomes. Minor differences in incremental effects between perspectives are a consequence of the incremental cost-and-effect equations being jointly estimated with the SUR method. 92

Perspective	Incremental cost (£, 2010–11): mean, 95% bootstrap CI			Incremental effect: mean, 95% bootstrap CI			ICER
	Mean	Upper CI	Lower CI	Mean	Upper CI	Lower CI
HSC: 0–12 weeks
NPI	–159.6	–1267.8	948.5	2.46	–7.59	12.50	Intervention dominant
ZBI	–155.9	–1254.7	942.9	0.56	–3.90	5.02
DEMQOL-Proxy	–156.5	–1256.3	943.3	2.55	–3.32	8.41
QALY (DEMQOL-Proxy)	–156.7	–1257.4	944.1	0.0066	–0.0026	0.0157
GHQ	–155.5	–1250.6	939.5	4.00	–1.92	9.91
Societal: 0–12 weeks
NPI	1,018.8	–2331.0	4368.7	2.42	–7.54	12.39	421
ZBI	992.5	–2384.6	4369.6	0.58	–3.86	5.02	1,711
DEMQOL-Proxy	978.1	–2403.7	4359.9	2.46	–3.40	8.33	397
QALY (DEMQOL-Proxy)	954.0	–2444.5	4352.4	0.0065	–0.0025	0.0155	146,437
GHQ	1,004.8	–2349.2	4358.8	4.07	–1.79	9.93	247

Summary of the main findings

The study comprised three inter-related components: a literature review; a RCT; and some qualitative exploration of experiences of the conduct and participation in the trial. In the literature review we described the limited, but promising, evidence to support the use of exercise as a therapy for BPSD. The main shortcoming in the reported literature was the limitations of the research methods in the existing evidence base, something we sought to address with our RCT. The findings from the literature review were used to design a methodologically robust trial of an exercise programme for the purpose of ameliorating BPSD. Our main finding was that the exercise programme did not produce a significant clinical benefit for BPSD in this population. Importantly, however, we did find that caregiver burden was significantly better at the end of the trial for those in receipt of the exercise programme. Finally, significant – and unexpected – challenges of recruiting to the trial via clinicians were explored in a focus group. Clinicians reported significant anxieties recruiting to our relatively ‘low-risk’ trial that were most readily overcome through the use of a well-defined systematic recruitment and feasibility tool.

Towards a research agenda: from the literature to a randomised controlled trial

Behaviours that challenge caregivers – particularly agitation, psychosis and sleep disturbance – are often central to predicting institutionalisation, probably because of the reported negative impact for caregivers. 96–98 Research that identifies optimal physical activity modalities, frequency, intensity and duration for persons with different types and severity of dementia has been lacking and will be important to inform commissioners of the value of exercise programmes in this population.

Appropriateness, safety and motivation to participate are critical to the success of exercise programmes, and may be particularly complex for older adults with dementia. For example, adherence may be influenced by the nature of dementia itself. Particular symptoms, such as apathy, may have a proportionally greater impact (relative to other BPSD) on adherence, which could be further impacted upon by context76 (e.g. paid carers often view apathy differently to unpaid carers).

Exercise appears to be beneficial in reducing some BPSD, especially depressed mood and agitation, and may also improve sleep and reduce ‘wandering’. However, research into the efficacy of exercise with the therapeutic aim of improving other important symptoms, such as anxiety, apathy and repetitive behaviours, was weak. Although some studies suggest that walking for at least 30 minutes, several times per week, may enhance outcomes, the beneficial effect of exercise type, its duration and frequency are unclear. Indeed, studies examining the effect of daily exercise had more favourable outcomes.

Results from the trial

Although overall BPSD were lower at week 12 in the group that received the exercise in comparison with the TAU group, this difference was not statistically significant. This is not consistent with the meta-analysis reported by Heyn et al. 76 There are three possible explanations for this: first, exercise is not a clinically effective therapy for BPSD; second, our exercise regimen has not had sufficient intensity to effect an impact on the participants; or, third, the effect size is smaller than we anticipated.

Importantly, our secondary analyses revealed a significant benefit for carers by attenuating burden across the trial duration. This is important as caregiver burden is strongly linked to use of health, social and medical services by both the person with dementia and their caregiver. 65–68 Given the absence of impact on behavioural and psychological symptoms, it is unclear why caregivers from the TAU group reported having a significant worsening of burden in comparison with those in receipt of the intervention. The increase may have been due to worsening burden because of BPSD, differential physical burden between the two groups (there was a slightly higher number of physical conditions in the TAU group at baseline) or unhappiness among carers about not being randomised to the intervention. Conversely, participation in exercise in the intervention group may have attenuated the perception of burden.

The economic evaluation looked at costs and cost-effectiveness. At 12 weeks, average costs from a societal perspective for the group following the exercise regimen were non-significantly higher than for the control group getting TAU. From a HSC system perspective, costs for the exercise group were non-significantly lower.

When considering cost-effectiveness for each outcome measure there were no significantly different incremental costs or incremental effects. The sensitivity analysis conducted (adjusting for baseline sociodemographic variables) generated findings that were consistent with the main analyses.

Recruitment to clinical trials

Researchers aim to work to high professional standards of GCP and require all study participants to be consenting volunteers, who are aware of their right of refusal without fear of detriment. When people lack capacity, ethics committees require rigorous protocols to safeguard participants’ wishes and welfare. We believe that all people, whether they have dementia or not, should have the right to participate in research. We recognise that people take part for many reasons, including altruism and a sense of purpose. Within a society that often patronises and demeans older people, many clinicians and researchers are passionate advocates of enlarging patient choice. In our experience people who are not interested in participating in research have a range of direct and indirect ways of showing this. Despite all of these safeguards, clinicians appear apprehensive about recruiting even for relatively low-risk studies like EVIDEM-E. Therefore, we support recent calls to quantify clinicians’ impact on recruitment and test ways of improving accruals. 99 We also agree more is needed to understand the experience of participating as a ‘control’. The response from the teams was positive, but actual recruitment was much slower than anticipated. Six participants were recruited to the trial through memory clinics and community mental health teams over a 6-month period from a patient population exceeding 2000 people with dementia.

There are several ways in which researchers can incentivise clinicians’ involvement in research. The focus group indicated their involvement should bring tangible reward for individual clinicians and participants. It was not enough, in the group’s view, that NHS trusts financially gain through involvement with research – how these resources enhance frontline clinical services should be made explicit. Encouraging a sense of ownership is a well-established tool for engendering investment, and we attempted this by including clinicians on our advisory groups. However, it is likely that a more widespread and involved approach is needed to improve practitioner engagement. Consulting clinicians on dissemination strategies might help them see the pathway from project to practice more clearly. Finally, developing flexible quick guides to recruitment may instil confidence when discussing individual projects with patients.

The approaches suggested by the focus group would all maintain clinicians at the centre of the recruitment process. An alternative could be direct contact between researchers and patients, which could potentially resolve many recruitment challenges for clinician and researcher while improving patient choice. In collaboration with the North Thames DeNDRoN, we helped to develop a registry of people with dementia who were interested in research (‘DemReg’),100 which appears to be an efficient recruitment mechanism. This registry is discussed in detail in Chapter 6 of this report.

Strengths of the trial design

The process of designing the study was challenging to the research team. It saw the project team – supported by an independent steering group – attempt to strike a balance between achieving a feasible and sustainable intervention that could be widely adopted in the target population while maintaining scientific rigour in its evaluation.

Strengths of this study include the relatively large sample size, with low attrition rates; use of well-validated outcome measures; a pragmatic exercise programme that does not require specialist training or equipment; and efforts to control for the impact of therapist contact.

The trial design addressed other important factors: a more homogeneous sample in terms of diagnosis, severity of disease and mobility; a sample size with sufficient power to detect an effect (positive or negative) of clinically significant magnitude; and blinded and objective outcome ratings.

Limitations of the study

Careful consideration was given to blinding the researcher to the allocation status of the participants. The single-blind design left study personnel vulnerable to unmasking group allocation during data collection with participants. To mitigate this impact we decided to include a second, independent, researcher who collected our primary outcome data (NPI) at 6 and 12 weeks. However, a risk remained that the dyad would divulge information about the group to which they have been allocated. The efficacy of maintenance of blinding was assessed at each time point, by personnel collecting data who recorded the arm to which they believed each dyad was randomised.

Separating the effect of psychosocial contact (from the exercise therapist) from the exercise intervention has proved to be the biggest challenge in our study. We addressed this problem by introducing two periods during the intervention: 1–6 weeks, whereby the participants in the intervention group were trained and supported by an exercise therapist, and 6–12 weeks, when participants were encouraged to continue exercising but had no contact with the therapist. Although we tried to control for contacts between participants and the exercise therapist, we could not control for the social contact between the person with dementia and their carer. Dyadic walking may have encouraged not only physical activity, but also psychosocial support from the carer as well.

We used outcome measures that are well validated and widely used in research with individuals with dementia. We also gave considerable thought to the issue of measuring physical activity. Particular consideration was given to the use of equipment by the participants to record their individual activities, for example GPS (Global Positioning System) receivers and pedometers. However, we decided not to use such instruments in order to avoid excessive intrusion, and to prevent the instruments affecting exercise behaviour in their own right. Similarly, participants’ levels of physical activity was evaluated through measurements of the participants’ self-reported RPE. The intervention delivered by the therapist centred on monitoring and increasing activity intensity utilising the RPE, and so could be provided only to the exercise group.

Assessing physical activity with minimum intrusion to study participants is challenging and we relied on participants’ self-reports through daily dairies. Self-report is prone to information bias, especially when participants are not blinded to their group allocation.

The economic evaluation adopted a HSC perspective for some analyses and a societal perspective for others to ensure that unpaid carer inputs were not overlooked. A range of outcome measures was examined, including QALYs generated from a dementia-specific measure. A limitation of the economic analyses was the sample size, which was 50% lower than the sample for the main outcomes analysis because service-use data were not collected. This loss of statistical power limits the conclusions that can be drawn. Another possible limitation is that only 85% of study participants provided information about support received from unpaid carers, yet we know that all sample members had a carer. It is difficult to estimate accurately the amount of time an unpaid carer spends with someone with dementia.

Building research capacity

The exercise therapist (James Lee) used data from this study towards his undergraduate degree; he received the Bob Finney Memorial prize for this work and graduated with first-class honours. The RW (Arlinda Cerga-Pashoja) has registered for a Doctor of Philosophy (PhD) and passed her MPhil–PhD viva in 2011. In addition, David Lowery worked in a postdoctoral capacity and contributed to an expert therapy development group of another National Institute for Health Research (NIHR)-funded research programme (WHELD) on behalf of the EVIDEM group. The study team members have also contributed in delivering educational programmes to the CNWL staff through summer schools.

Voluntary RWs have given an invaluable input in the EVIDEM-E study. Nine volunteers helped with data entering, designing databases and collecting outcomes. They gained research experience and had the opportunity to be trained in GCP.

Changes to protocol

No changes from the original protocol were made.

List of appendices

A systematic review of the evidence about prevalence

Although the presence of incontinence symptoms is well documented in people resident in care homes,116 there is currently little evidence available for clinicians, commissioners and service planners concerning the scale of the problems in the larger population of people with dementia living at home.

Methods

Study design: a systematic literature review

Results

The PRISMA33 flow diagram in Figure 10 reports the results of the search. From 427 references, eight studies129–136 met the inclusion criteria and are summarised in Table 35. No studies reported incidence. Seven studies provided prevalence rates as findings that were incidental to their primary question. Study populations, size of sample and assessment tools were varied, including the Care Needs Assessment Pack for Dementia (CareNap-D),138 Cognitive Performance Score (CPS),139 the ICD-10,80 the UK Office of Population Censuses and Surveys (OPCS) Disability Survey 1985140 and the MMSE. 52 Assessment of risk of bias found that all studies had some aspect that was judged to be a risk; this has been reported in full elsewhere. 137 The reported prevalence rates are presented in Table 36. The prevalence of UI ranged from 1.1% in a general community population to 38% in those receiving home care services. Prevalence of FI ranged from 0.9% in a community population to 27% in a sample attending an old age psychiatry outpatient clinic.

FIGURE 10.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram of search results for prevalence studies.

Discussion

This review found no studies reporting incidence and a wide variation in reported prevalence rates of UI and/or FI in people with dementia living in their own homes. This is explained, in part, by the variety of populations and assessment tools, and the lack of uniformity in criteria for defining incontinence and dementia. These methodological challenges in assessing prevalence rates have been well documented for both incontinence116 and dementia,141 including the issues of underreporting through embarrassment and stigma. Consequently, nearly all of the studies may have underestimated prevalence. The rates were incidental findings in nearly all studies and the one study designed to address the prevalence question of interest had a small population, identified through carers who identified themselves as experiencing high levels of stress. 134

To our knowledge, this is the first review addressing this question of prevalence. The greatest weakness of the study is the search strategy, which may have overlooked other studies with incidental prevalence findings in studies, and the exclusion of studies not reported in the English language. The study is reported in more detail elsewhere. 137

Conclusion

There is currently no definitive evidence of incidence or prevalence of UI or FI in people with dementia living at home, and further research is required to establish population-level data.

Methods

Results

There were 1,246,963 eligible participants from 487 general practices, of whom 54,816 entered the dementia cohort and 205,795 were selected for the non-dementia cohort. During the 10-year follow-up, incontinence (urinary, faecal and double) was newly recorded for 8987 (16%) people with dementia and 23,083 (11%) without dementia. Table 37 reports the incidence of UI and FI per 1000 person-years at risk (PYAR) in men and women with and without dementia. The IRRs, adjusted for age, sex and comorbidity, for UI were 3.2 (95% CI 2.7 to 3.7) in men and 2.7 (2.3 to 3.2) in women. The IRRs for FI were 6.0 (95% CI 5.1 to 7.0) in men and 4.5 (3.8 to 5.2) in women.

Fifteen and 18% of people with and without dementia respectively were prescribed medication for UI. Crude and adjusted IRRs are shown in Table 38.

The incidence of catheterisation could be analysed in 47,066 people with UI over 163,735 PYAR. The IRRs comparing people with dementia to those without are given in Table 38. The results in full are presented elsewhere. 143

Discussion

Dementia is associated with at least a doubling in the incidence rate of incontinence (any type) in community-dwelling primary care patients. To our knowledge this is the first report of incidence of incontinence from the clinical records of community-dwelling people with dementia.

Patients aged 60–89 years with UI and dementia received drug treatments at more than twice the rate of patients without dementia but of the same sex, age and comorbidity. This difference was attenuated with age but remained evident. There is some evidence to suggest that, in general, older adults may not receive optimum treatment for UI. 144,145 Indwelling urinary catheters were recorded at more than double the rate in women with dementia, compared with those without dementia of the same age and comorbidity. The difference was attenuated with comorbidities but not age. Indwelling catheters are associated with discomfort and risk of infection among other problems. 146,147 The clinical reasoning for the greater use of drug treatment and indwelling catheters in this population needs investigation.

The study has a number of limitations. The data are from primary care clinical records, kept for that purpose rather than the questions of this study. The different general practices do not record uniformly all types of information regarding the patient, for example, treatment or non-pharmaceutical supplies from other health-care services. Despite this the data provide insights not offered elsewhere.

Conclusions

This large cohort study establishes the incidence of incontinence among people with dementia living in the community. The incidence of incontinence is more than double – and for some age groups treble – for people with dementia compared with people without dementia of the same age, gender and comorbidity. The clinical reasoning in the primary care management of UI in people with dementia requires further investigation.

Methods

Results

The search results are presented in the PRISMA flow diagram (Figure 11) and three papers148–150 met the inclusion criteria. Each study was assessed for risk of bias. 127 Because of the heterogeneity of the interventions and outcomes it was not appropriate to undertake a meta-analysis, and a narrative summary of findings is presented. The characteristics of the three studies are presented in Table 39.

FIGURE 11.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram of the results of search for conservative intervention studies.

Two studies were described as exploratory and pilot studies, recruiting fewer than 20 dyads. 148,150 All had a control group, although one reported no data on these people. 148 All had risk of bias to the internal validity, presented previously in detail. 150 The interventions were advice and education aimed at the family carer, for them to act upon. Two included education for toileting: prompted voiding (PV)148 and an individualised toileting schedule,150 together with other aspects of continence education. The interventions involved between weekly149 and monthly150 contact with the health professional for the duration of the study. Toileting schedule advice was the least acceptable to carers in the occupational therapist (OT) intervention study. 148 The study of individualised scheduled toileting presented statistics indicating a significant reduction in UI in the intervention arm149 but on our re-analysis of the data provided this was not confirmed. 151 Using the intention-to-treat method the PV study found no statistical differences between groups on any UI outcome but noted there were clinically significant reductions in episodes of UI for some patients. 150 The results in full detail have been published previously. 148

Discussion

Only three studies146–148 of conservative interventions were found, dating from over a decade ago. Two were described as exploratory and pilot studies and all had some risk of bias to internal validity. None provided evidence of effectiveness of the interventions. All three provided evidence as to the difficulties in recruiting to such studies. All three provided evidence as to issues in acceptability and feasibility for some carers in agreeing with and implementing the professional advice or instruction.

The review has limitations in that while the search was designed to be comprehensive, criteria such as English language only may have resulted in the omission of some studies. However, to our knowledge this is the first systematic review addressing the question of evidence of effective interventions for this population group. Previous reviews have not differentiated between evidence from community-dwelling or within-an-institution populations although much of the evidence relates to those within care homes,152,153 or has not drawn on a systematic approach. 154 This lack of attention to the very different environments (both physically and in availability of people to help) of an individual’s home from a care home is apparent in many clinical guidelines. 155,156

Conclusions

There is little evidence of effective interventions for managing incontinence in people with dementia living at home. There is an urgent need for both research and clinical guidance for health professionals tailored to the setting in which the majority of people with dementia live.

Methods

Results

Ninety-eight documents from 38 local community health service specialist continence services, located across all 10 strategic health authority areas, were obtained and analysed.

Of the 38 local services, 19 (50%) made no mention of dementia or cognitive impairment in their documents. Only in the documents of three services were nurses offered detailed guidance about the management of incontinence for people with dementia at home. The remaining 16 organisations provided guidance that mentioned dementia or cognitive impairment in assessment processes, but there was very little evidence of any subsequent advice or mechanism for producing specifically tailored care plans, such as a dementia care pathway. Only one organisation specified that the service user should be given any information about the consultation, and that was a copy of the referral form for incontinence pads with delivery information on it.

Ten (26%) of the 38 local organisations referred to a leaflet entitled Continence in the Confused Elderly (anonymous) to be given to family carers. The carers in the advisory group reviewed this leaflet and concluded that it had deficiencies in content, information, tone and style. People with dementia were identified as a special case in the documentation of only one service, which warranted the provision of additional continence products. Only one service made reference to addressing the problems of delivery of NHS continence pads to people with dementia living alone at home.

Discussion

Fifty per cent of sampled organisations had clinical guidance on incontinence which made no mention of dementia. A small minority provided examples of good practice. These good practice examples focused on person-centred assessment and responsiveness to the described problems, an approach advocated for some time in the care of people with dementia generally,115 and by gerontological nurses and dementia specialists. 161 A more detailed discussion of the results has been published previously. 162

To our knowledge, this is the first study exploring the local guidance to primary care professionals in meeting the continence needs of people with dementia living at home. The study has limitations in that the representativeness of the documents is unknown, although the geographical spread, method of requesting from specialist nurses and internet searches would suggest that the variety has been captured.

Conclusion

Local clinical guidance on incontinence management rarely addresses the specific needs of people with dementia living at home or their carers. There is much room for improvement and providing evidence-based guidance to frontline staff on the specific problems of people with dementia and their carers. This review led the advisory group to prioritise for Phase 4: the development and testing of dementia specific continence assessment tools and guidance for the use of primary care nurses working with people with dementia living in their own home. The review also pointed to the importance of the involvement of carers of people with dementia in continence service developments.

The views of people with dementia

We could find no accounts from the perspective of the person with dementia on issues of incontinence. The purpose of this study was to investigate the PWD’s perceptions of managing problems of incontinence.

Methods

The study approach was in the interpretative tradition163 and the design was of qualitative guided conversations, a recommended research technique for use with people with dementia. 164 A purposive sample of people with dementia was recruited through primary care, specialist community mental health services and voluntary organisations. Conversations were taperecorded with permission, transcribed and thematically analysed,165 using NVivo™ version 8, 2008 software package (QSR International, Warrington, UK: www.qsrinternational.com/default.aspx).

Results

It took a period of 1 year to recruit seven people with mild to moderate dementia and UI problems. They described a range of problems from urgency (which they could not respond to in time) to nocturnal enuresis. All described feelings of shame and humiliation as in this example:

There’s always the fear and sometimes I don’t get there in time and I find myself with wet pyjamas and it’s personally very embarrassing.

Man with dementia 04

Strategies included planning activities outside the house only when they knew they could access toilets, accepting help from their spouse (although this sometimes caused tension) and using a variety of absorbent materials for containment, for example towels. Descriptions of the use of incontinence pads included problems of fit and acceptability, as in this example.

I’m so small you know, and I’ve been cutting them [incontinence pads] and the whole flat is covered in bits of, what do you call it, small pieces, fluff, every day fluff . . . they’re so big, I’ve got to cut them.

Woman with dementia 01

Strategies also included using euphemisms in describing the problems, using humour to overcome embarrassment and trying to protect family members from spillage by clearing up or dealing with urine or wet clothes themselves. The detailed results are reported elsewhere. 158

Discussion

To our knowledge, this is the first study reporting people with dementia’s perceptions of these problems of incontinence. The descriptions of embarrassment and shame are no different from those given by other adults. 166 The desire to protect spouses or other family members from the ‘pollution of bodily waste’167 has not been reported before and may have two unintended consequences. The first is that it may increase family carer stress if the person cannot appropriately deal with leakage and excreta. The second is that the family carer may not understand the motivation for such acts and interpret these as the need to protect their dignity by concealment from everyone, including health professionals who could potentially help. Although the study has limitations through size and difficulty in recruitment, it does provide insights that have not been reported previously.

Conclusion

People with dementia are able to describe in detail their experiences and management of incontinence. Their desire to protect their own dignity, but also protect others, may lead to unintended consequences and delays in seeking wider help for such problems.

Views of family carers on the problems of incontinence and effective strategies to manage them

We could find no study investigating family carers’ strategies for managing incontinence. The purpose of this study was to investigate family carers’ perceptions of the range of incontinence problems and their strategies for managing them.

Methods

The research approach was in the interpretative tradition. 164 A purposive sample of family carers of people with dementia, living in their own homes, was recruited through primary care, specialist community mental health services and voluntary organisations. Qualitative semi-structured interviews (SSIs), based on an open-ended topic guide,168 were conducted either face to face or by telephone, recorded by tape, transcribed and thematically analysed,164 using NVivo^™ version 8, 2008 software package.

Results

Thirty-two carers were interviewed. They described a range of problems from supporting the person to remain independent in toileting, through to dealing with inappropriate behaviours, to containing and managing incontinence. Key themes were identified: active problem-solving, protecting the dignity of the person with dementia, problematic responses from health services and unintended consequences. All carers actively used problem-solving strategies but sometimes these were not acceptable or understood by the person with dementia, particularly as the dementia progressed. Most carers reported protecting the person’s dignity by not seeking health professionals’ help, often until the point of crisis.

I didn’t want people to know about her not managing and incontinence . . . I felt her dignity, you know . . . She was so in herself a very dignified lady . . . If she’d have known she’d have been horrified.

Husband c6

Once the carer had decided to seek help, the responses from health professionals were reported to be often less than helpful, and carers reported local health service policies on access to continence products to be inconsistent and often inappropriate to their circumstances. A few carers reported strategies for managing toileting and incontinence that had the potential for distress and harm to the person with dementia. These findings are reported in full elsewhere169 in a paper that was awarded the Royal College of General Practitioners’ Research Paper of the Year 2011, in the category of dementias and neurodegenerative diseases.

Discussion

The reluctance to seek help for toileting and incontinence problems by carers of people with dementia as they seek to preserve their dignity and personhood115 has not been reported before, although the low rate of seeking help for incontinence symptoms is well documented in older adults. 170 The variability and sometimes inappropriate responses of primary care professionals has not been reported previously, to our knowledge. Management strategies were mediated by their acceptability to both the PWD, the carer and other family members. A key influence in carers’ management strategies was to ensure that excreta were well contained, an important goal identified by expert panels for incontinence management in frail older people. 117 Although the study has limitations because of its sample size, the diversity of the population and the sampling until saturation in data ensured depth and breadth in the findings.

Conclusions

There is a wide range of types of incontinence problems that change over time, and carers managing them are often poorly supported by primary care professionals. Primary care professionals could be more proactive in enquiry, repeated over time, about toileting and incontinence problems; alert to strategies that have the potential for negative effects; and develop their repertoire of referrals, advice and information to reduce crisis and problems.

Views of primary care, social care and specialist mental health staff in community settings on effective strategies

We could find no studies reporting HSC professionals’ perceptions of effective strategies for the management of incontinence in this population. The aim of this study was to investigate the perceived range of problems and effective strategies.

Method

The study design was in the interpretative tradition using qualitative individual and group interviews,171 with HSC professionals purposively sampled to ensure a range of generalist and specialist community services, disciplines, grades of staff and those with qualifications regulated through legislation and those without. Participants were approached through organisations that had agreed to participate in the EVIDEM programme and at EVIDEM educational events; these were mainly in north and central London and Hertfordshire. Data were recorded in field notes and in flip charts kept as part of the group discussions, transcribed to electronic documents and thematically analysed.

Results

We recruited 140 participants from a wide range of staff groups (Table 40), and they took part in 30 individual interviews and group interviews. Collectively, they described a wide range of toileting and incontinence problems, with a wide range in the number of problems reported. Some informants mentioned only one or two types of problems, whereas dementia specialist staff, such as the Admiral Nurses, described greater numbers of both physical problems and dementia-related problems.

Collectively, they reported a wide repertoire of strategies but many offered only one or two strategies that reflected their type of service. A commonly reported strategy was to refer the patient to another service, in many cases the district nursing service. The small number of district nurse informants focused mainly on using continence pads as a containment strategy. Many informants considered that the local NHS provision of incontinence advice and absorbent products did not address the needs of people with dementia and their carers; in some instances these deficiencies were seen as contributing directly to decisions to move the person to a care home. Other system-level problems were identified, such as the lack of acknowledgement of toileting support in social care assessments.

Discussion and conclusion

To our knowledge there are no other published studies exploring multidisciplinary and multiagency perspectives on managing incontinence for people with dementia living at home, although there are some that have considered the issue from the perspective of care home staff. 172 The study has limitations in that views offered by a small number of informants may not be representative of others throughout England. However, the intention was not to undertake a representative survey but to understand the breadth of issues and strategies. The findings suggest that many staff groups have relatively limited knowledge of the range of problems that can be experienced with a concomitant limited repertoire of responsive strategies. However, the cumulative responses also point to a mosaic of practical, psychological, behavioural, biomedical, environmental and supportive strategies that can be utilised. One question is whether knowledge alone of this mosaic of strategies can influence primary care professional practice in addressing these types of problems. The respondents also pointed to failures of publicly funded systems in addressing these problems, which raises the question of whether or not new commissioning processes and initiatives, such as personal budgets, will be better able to address them.

Experience, including impact and consequences, of people with dementia and their family carers over time

Dementia follows a course of progressive disablement over a number of years. 173 The experience, impact, cost and consequences over time are poorly described in relation to incontinence and TDs (defined as a range of problems associated with loss of independence in toileting or behavioural symptoms, sometimes resulting in soiled clothing or excreta uncontained in pads or voiding inappropriately). This study investigated the experiences, impacts, costs and consequences of incontinence in people with dementia and their family carers over a period of up to 3 years.

Method

This was a longitudinal mixed-methods174 study investigating the experience of a purposive sample of people with dementia with continence problems living at home, and of their carers. Personal and general invitations to participate were distributed via community organisations, social care organisations, specialist community mental health services for older adults in five London boroughs, and GPs in eight PCT areas. Data collection was by serial qualitative interviews. 175 In addition, validated instruments were used to characterise the person with dementia who met the criteria of the ICD-10. 80 Their mental state and abilities were described by the MMSE,52 the NPI62 and the Disability Assessment for Dementia (DAD),176 their TDs and incontinence by an adapted International Incontinence Modular Questionnaire [International Consultation on Incontinence Questionnaire (ICIQ)]177 and their health-related quality of life using the DEMQOL. 86 Also recorded were the service use and cost consequences using the CSRI Parts 1 and 2,88 as well as the carers’ health-related quality of life [EuroQol: European Quality of Life-5 Dimensions (EQ-5D)],178 and burden (ZBI87 and NPI Caregiver Distress Scale179). Analysis of the qualitative data was thematic,165 through comparisons over time and between cases using the NVivo 8 software package and longitudinal charting. Statistical analyses of quantitative data were undertaken using SPSS version 20 and Stata version 13 software. Costs were derived from the PSSRU Unit Costs of Health and Social Care 201189 and estimates using the Office for National Statistics 2011 Annual Earnings Survey. 180

Results

Thirty-four people with dementia and 29 family carers were recruited over 2 years (1 December 2008 to 31 December 2010). During the period of the study, 10 people with dementia died, six moved to care homes and one withdrew. Because of the long recruitment period the remaining participants were in the study for different lengths of time (see Figure 12) but many family carers provided retrospective data on entering the study to describe their experiences over a longer period of time (Table 41). The characteristics of participants at the first interview are provided in Table 42.

FIGURE 12.

Flow diagram of participants over time.

The types of toileting and incontinence problems are described in Table 43.

The progression of the dementia was evident in participants and was reflected in the DAD scores, which showed a significant reduction over time in ability from a median score of 15 to 10 (p = 0.003 by Wilcoxon signed-rank test). A series of basic regression models were fitted to the data. Inter-participant variability was allowed for by random-effect parameter, and bootstrap CIs were used to account for floor and ceiling effects in the outcomes. The effects on each of the carers’ measures were adjusted for the person with dementia’s NPI measure (Table 44). The carers’ health-related quality of life scores were significantly negatively affected by the presence of FI (–5.3, CI –10.4 to –0.1, significant at the 5% level) but not by TDs or UI. The carers’ distress scores (in response to BPSD) were significantly negatively affected in the presence of TDs (6.5, CI 0.6 to 12.3, significant at the 5% level) and UI (5.2, CI 2.7 to 7.6, significant at the 5% level) but not in the presence of FI.

A number of themes were identified: changes in problems over time; the processes of normalisation, both in managing toileting changes over time and also in managing excreta out of place; the iatrogenic effects of hospital inpatient stays and respite stays in care homes on both the PWD’s toileting abilities and physical health (e.g. the development of pressure ulcers); the interplay of decreased mobility and incontinence; and the perceived positive impact of both paid help in personal care, which the carer controlled, and good excreta containment on reducing carer stress. Figure 13 illustrates some of these themes through the changes and experiences over time of one person with dementia and their family carer. The presence of double incontinence (UI and FI) compared with one type of problem only (TD, UI or FI) was found to significantly increase the costs by 89% from a HSC perspective and by 69% from a societal perspective (Table 45). The HSC perspective includes costs both provided by the NHS or government and also privately purchased. Societal costs include costs of the time of the informal or family carers.

FIGURE 13.

A longitudinal chart of one participant’s experience. PPA, primary progressive aphasia.

Discussion and conclusion

To our knowledge this is the first study reporting experiences of incontinence over time, and impacts upon this patient population and their carers. The small sample is a limitation but provides insights into the experience, types, costs and impact of incontinence that are not provided elsewhere. The results in full are discussed elsewhere. 169 The negative impact of the presence of FI on carer quality of life warrants further investigation. The perceived positive impact of assistance in personal care tasks, which are within the control of the carer, and effective containment of excreta require further investigation, in particular whether or not it influences or defers decisions about moving to a care home.

Method

This was an observational study based on the methods of a previous clinical trial with cognitively intact adults181 and informed by the family carers in the advisory group. The study aimed to recruit a sample of 40 family carers, diverse in their, and the PWD’s, characteristics, for example age, gender, stage of dementia. Family carers were recruited through voluntary organisations across Greater London, specialist mental health services for older people in North West London, and general practices and a community continence service in South West London. At time point 1, the family carers were instructed in observational and diary data collection. They were asked to complete the diary sheets for 1 week. Data characterising the person with dementia (DAD,176 NPI62) and the family carer burden176 and the type of incontinence (ICIQ)177 were also collected. At time point 2, the family carers were interviewed about the experience of data collection and their diaries retrieved. Analysis of quantitative data was undertaken using the SPSS version 20 software package. Analysis of the qualitative data was undertaken thematically. 165

Results

Twenty-five family carers of people with dementia living at home were recruited to the study over 18 months. There was diversity in the sample of family carers, the person they cared for and the toileting and incontinence problems (Table 46).

The range and funding source of aids and adaptations used are described in Table 47.

There was also diversity in the designs of incontinence pads used, which changed for some individuals when outside the home and between day and night-time (Table 48).

Those who were recruited demonstrated high levels of commitment to the study’s aims. Twenty (80%) of the carers were able to complete some days of the diary sheets (mean number of days 8, range 2–16 days). Carers entered data about each change of pads; the number of times they entered data ranged from 4 to 56. One aspect was leakage of excreta outside the pad, for example onto clothes. The rate of leakage by pad design is reported in Table 49.

Analysis of data on the acceptability and effectiveness of different designs of incontinence pads revealed several important factors. Designs of pads were acceptable to the carer when they could safely and efficiently be changed by the carer (and others such as day centre staff). Safety factors related to designs that allowed the carer to change the pad and hold the person steady if necessary (i.e. to minimise risks of a fall). Efficiency factors related to speed in changing, to cause least disturbance to the person and with minimal effort on the part of the carer. Effectiveness related to the ability of the design to ensure containment of excreta.

It was evident that the required specifications of pads changed according to location (e.g. the use of public toilets or at home), the availability of the carer or others to change the pad and whether a pad was to be used in the day or at night. At night, 40% of the sample (n = 10) used either a more absorbent pad than in the day, a different design or two designs in combination. The results are described in full elsewhere. 169

Discussion and conclusion

To our knowledge this is the first study exploring preferences and effectiveness of incontinence pad design when used by people with dementia living at home and their family carers. Considerable endeavour and resources went into the recruitment process, which was much slower than anticipated, and we were unable to recruit the planned numbers: future studies will need to take great care with recruitment strategies. There are limitations due to the sample size but the study provides data that will be used to inform the development of a clinical trial protocol. It also provides insights into the specifications in effectiveness and acceptability of different pad designs for people with dementia living at home not provided elsewhere.

Developing and testing a continence assessment tool and resources

The aim of this study was to develop and test a continence assessment tool and supporting resources for people with dementia, to be used by primary care professionals, primarily community nurses.

Methods

An adapted three-stage Delphi consultation study182 was used. Stage 1 brought together an expert group of carers, dementia specialist workers from the voluntary sector and health professionals. This group examined commonly used continence assessment tools, and discussions were facilitated to describe a broad range of principles and issues that would underpin an assessment tool designed to address the needs of people with dementia. At stage 2 a prototype dementia-focused continence assessment tool was developed using the data generated in stage 1, asking for agreement or disagreement to items plus suggestions for further items. This was used to consult, in writing, both the expert group in stage 1 and also a further group of carers and specialist continence professionals. The prototype was further adapted. At stage 3 a different, wider group of experts (carers and professionals) was consulted in writing. They were sent the draft dementia-focused assessment tool (Figure 14) together with a questionnaire to test its face and content validity. Recipients were asked (1) whether or not the tool would improve recognition of the problems (face validity) and (b) to rate each item for importance and identify missing or unnecessary items (content validity). It was emphasised that the tool’s remit was a supplement to the continence assessment tools and guidance currently used, hence it was called an add-on tool. Participants were also sent supporting resource materials on which to comment (Table 50).

FIGURE 14.

The EVIDEM-C add-on continence assessment tool.

Results

The three stages of the Delphi consultation involved 46 carers and professionals (Figure 15).

FIGURE 15.

The three-stage Delphi consultation and participants.

At stage 1, the key principle agreed by the group was that the typical NHS continence assessment tools should be turned ‘upside down’, and that the first question in the tool should be an enquiry about the problem and its impact as defined by the person (if able) and their carer. At stage 3, 26 experts, of the 50 invited, returned the questionnaire (Table 51).

The majority of respondents (87%) perceived the tool as being likely to improve primary care professionals’ recognition of the toileting and incontinence problems experienced by people with dementia and their family carers. The GPs could see the value of the tool as an educational rather than a practice aid. All agreed that the first two questions, on the problem and the impact, were important. Most agreed on the importance of the items within the tool, although not all items were supported equally. For example, the carers did not rank the question regarding awareness of benefits as important. Caveats to the value of the tool on its own were added, such as the need for training. The results in full are described elsewhere. 183

Discussion and conclusion

The face and content validity of the dementia-focused continence assessment tool were established. The study had limitations in the response rate from some groups but provides a breadth of opinion not provided elsewhere on such an assessment tool. The involvement of family carers (i.e. experts by experience) in the first two stages ensured that the perspective of service users was present. Continence services in the UK have been repeatedly criticised for the absence of service user involvement. 158 To our knowledge there have been no other studies developing continence assessment tools tailored to people with dementia. Using a three-stage adapted Delphi consultation technique we created a tool (see Figure 14) along with supporting resources (Table 50) that is person centred120 rather than constructed from a biomedical perspective. The challenge, as pointed out by a number of participants, is how to ensure that those using the tool also adopt a person-centred approach and have a repertoire of strategies to address the problems identified.

Strengths and limitations

These have been described at each phase. In overview, the study strengths were the presentation of new knowledge not previously reported in each phase. The iterative processes between the phases allowed for greater reflexivity and responsiveness to both findings and the input from carers of people with dementia as well as professionals. The main weakness lies in the failure to always achieve the planned sample size, and the consequences of this for external validity. However, dementia and incontinence are both stigmatised conditions and the discussion of excreta, particularly faeces, is often a taboo subject in all societies. 167 This study was successful in engaging people with dementia and family carers in research on incontinence, and is a strength in both the research approaches and findings. The failure to always engage practitioners in studies indicates that this population and these types of problems are not necessarily perceived as a priority.

Interpretation of the findings

The study has addressed all of the research questions presented at the beginning of the chapter.

First, each phase of the study was able to add to the body of knowledge that primary care practitioners and commissioners can draw upon, because there is very little other evidence or guidance available tailored to this population. 117,118 Toileting and incontinence problems are neglected among those who have been, until relatively recently, an overlooked population: people with dementia living at home.

Second, the depth and breadth of the description of the problem of incontinence and its impact, provided in this study, expand and illuminate accounts of people with dementia in care homes186 and studies of older adults with incontinence but without cognitive impairment. 166,170

Third, a key message from this study for primary care professionals was to proactively enquire at each contact about toileting/bladder and bowel problems. The range of problems (from those amenable to medical treatment such as urinary tract infections through to voiding in inappropriate places) suggested a stepwise approach to identifying suitable and acceptable management strategies. The findings indicated that primary care professionals need a repertoire of strategies that draws on knowledge of cultural, psychological, behavioural, biomedical, technical and environmental aspects, as well as local systems, in order to help solve practical problems. This study has been able to develop resources for primary care practitioners that take cognisance of all factors (see Figure 14 and Table 50). There is little other research detailing this mosaic of interlocking incontinence problems, although it is evident in the report of eight case studies in Australia. 187 Expert opinion tends to focus on physical practicalities117 or behavioural techniques188 as solutions. It has highlighted for the first time that some strategies adopted by carers (family and paid care workers) may be unintentionally detrimental to the person and that health-care professionals need to be alert to this possibility.

Fourth, the study has developed and tested the first person-centred tool for continence assessment by primary care nurses, tailored to people with dementia living at home. This requires further field testing and evaluation.

Fifth, the study has demonstrated the high priority that family carers place on adequate containment of excreta, and its link with carer quality of life. The evidence of less-than-helpful responses from some NHS continence services supports other accounts from service users about variation in service quality and lack of professional help. 145 This study provides further evidence to support a continence paradigm shift from cure and prevention to very good containment as proposed by the International Continence Society expert group on UI in frail elders. 117 A pilot study has been conducted that investigated the acceptability and effectiveness of different continence pad designs with people with dementia living at home and their carers. The results indicate that a clinical trial of designs of pads is feasible, although with caveats on recruitment, as discussed below.

Sixth, the study highlighted the difficulties in recruiting people with dementia and their family carers to observational studies that address issues of incontinence. This has also been reported in studies attempting investigations of interventions in the USA149,150 and Australia. 187 The influence of embarrassment and stigma on clinical practice and research for both dementia and incontinence is well documented. 4,116

Impact

In addition to the impact through the presentations at events attended by HSC professionals, people with dementia and their family carers (see Tables 70 and 71), the study can point to two other types of impact. The first was the award to the study by the Royal College of General Practitioners for Research Paper of the Year 2011 in the category of dementias and neurodegenerative diseases. 169 The second relates to the invitations to provide materials on incontinence and dementia for the continuing professional development of:

• GPs by the British Medical Journal Quality group184

• nurses by the Royal College of Nursing-published Nursing Standard185

• adult care staff by the South East England DH and the Social Care & Local Partnerships Programme.

Background: the need for this study

People with dementia have a limited life expectancy and the type of end-of-life care that they need has been widely discussed. 190–193 These discussions focus particularly on the place of care and on the adaptation of existing end-of-life care pathways, but people with dementia experience different trajectories of functional decline and have different needs. 194,195

Communication problems, confusion, functional loss, nutritional deficiencies and complications, such as infections and incontinence, can compound the distress of a person who is dying, challenge clinicians’ knowledge and expertise,196 and result in inappropriate medical interventions and referral to hospital. 197–199

There are significant methodological challenges in researching end-of-life care for people with dementia. 200,201 Most research in this area tends to rely on proxy or retrospective accounts of particular symptoms and events. Fewer studies have addressed the environment and organisation of care, or decisions by different practitioners on individual patient outcomes.

In the UK, the quality of care of people with dementia in care homes is variable. 202–204 The cognitive ability and dependency of residents vary between care homes, and there is a considerable overlap in residents’ needs between homes with on-site nursing and those without. 205 Many factors influence how care homes work with primary health-care services. These include ownership, staff turnover, the qualifications of the care home manager, access to specialist palliative care, financial incentives offered to GPs, and the history of innovation within the home. 206–210 There is little evidence that care home residents, relatives and staff are involved in identifying priorities for end-of-life care for people with dementia, or in developing models of care that go beyond the individual patient-focused encounter. There is a need to develop and test dementia-specific approaches to end-of-life care for this often marginalised population. 211,212

The NHS End of Life Care Programme for care homes209 seeks to ensure that when residents are dying they can avoid emergency hospital admissions and be supported by care home and NHS staff who are skilled in providing end-of-life care. Care homes are being encouraged to use end-of-life care support tools, such as the Gold Standards Framework (GSF),213 Preferred Priorities for Care,214 and the Liverpool Care Pathway (LCP). 215 These interventions increase staff confidence and knowledge, reduce unplanned hospital admissions, and allow more people to die in their preferred place of care. 216,217

Study design

Phase 1 used a prospective, mixed-methods design including case notes analysis, interviews, mapping of service use and economic evaluation. Data from residents’ care notes were extracted at three time points (Table 52). Care notes of residents who had either died or left the home before the end of the data collection period were reviewed to the date of death or leaving the home.

Unstructured conversational interviews were undertaken with a purposively selected sample of older care home residents with dementia. The manager and, where possible, deputy manager, as well as two NHS professionals (e.g. GP and district nurse) with ongoing involvement with each home were interviewed. Emergency care practitioners and paramedics were also interviewed to establish how they interpreted their responsibilities when called to an older person with dementia in a care home. Care home notes of older people who died during Phase 1 were reviewed to document services received prior to death, the place of death and the care provided. These reviews were complemented by interviews with care home staff. Interview schedules and data extraction forms are shown in Appendices 37–46. Interview information sheets are shown in Appendices 49–51.

Inclusion and exclusion criteria

Inclusion and exclusion criteria for care homes and OPWD are detailed in Appendix 35. Inclusion criteria for care home residents were age ≥ 65 years, and a documented diagnosis of dementia or an assessment by the senior care worker that their level of cognitive impairment and behaviour were consistent with a diagnosis of dementia. Exclusion criteria were people with dementia who did not speak English, those whom the care home manager thought it inappropriate to approach (e.g. people in the terminal stage of the disease) and individuals who lacked capacity to consent and for whom a consultee could not be identified.

Sample

Details of the recruitment process used for both the care homes and the OPWD are reported elsewhere210 (see also Appendix 35). A total of 214 residents were eligible for participation in the study across the six care homes. Care homes were identified using the CQC (previously known as the Commission for Social Care Inspection) directory of care homes. Inclusion criteria were that the care home was for older people, provided personal care and specialist support in dementia care, had a minimum of 20 beds and there was no onsite nursing provision. To optimise the opportunities for participation, a further inclusion criterion was that the last CQC inspection report was favourable and that care home staff considered that they have a good working relationship with their local primary care services. Of the 10 care homes approached, six agreed to participate. The six care homes had a mix of ownership and geographical location.

A staged approach to the recruitment of OPWD was adopted, involving separate meetings with staff, residents and relatives. Individual consent was secured with residents with dementia who could consent in the moment and for those without capacity, through letters and follow-up contact with consultees.

Minimum data set

Areas of data collection are detailed in Appendix 35. Participants’ health and care needs over time were documented, as well as service utilisation (using an adapted version of the CSRI). 88 Additional data on residents were obtained using the DAD,176 the Barthel ADL Index,218 the Cohen–Mansfield Agitation Inventory (CMAI),219 and the Cornell Scale for Depression in Dementia,220 completed by home managers and senior carers.

Analytic methods

Costing analysis

Study design

Phase 2 used a co-design approach (AI)225,226 to bring together care home and visiting NHS staff to identify what worked well in end-of-life care for people with dementia, and to use that as a basis to plan and implement change. The exploration and enhancement of what is working well in an organisation is often framed as a ‘4-D cycle’ (Box 5). The cycle is often conceptualised as discrete steps, although the process is non-linear, iterative and ongoing.

The 4-D cycle was implemented over three facilitated hour-long AI sessions over 6 months. The AI method and session design are detailed in Appendix 47 (for further detail regarding the design of the intervention, see Nicholson et al. 227). Phase 2 tracked the events and care experienced by OPWD living in the care homes who participated in the intervention phase. Data were extracted from residents’ care notes at two time points, staggered over 7 months of data collection that ran from January to July 2011.

Baseline data extraction coincided with the start of the intervention (Table 54). Researchers reviewed participating residents’ care notes over the 3 months prior to the start of the intervention or closely thereafter and again in July 2011 after the third and final round of intervention sessions in the care homes. Baseline data were collected in this way for 90% of the study sample. Baseline data for the remaining residents (n = 8) were collected between mid-March and mid-April 2011 owing to delays in obtaining consent.

Impact of the Appreciative Inquiry intervention

The impact of the AI intervention on integrated working in end-of-life care was evaluated through analysis of resident deaths following the intervention and resource use before and after the intervention.

Resident deaths were examined through care note reviews, complemented by interviews with care home staff, for changes in services received prior to death, place of death and care provided following the intervention.

Resource use was examined among residents who participated in both phases of the study (see Costing Analysis, below, for further details).

Finally, postintervention interviews with all participants following the third and final AI session (see Appendix 47, interview schedule) were undertaken to examine perceived changes in working practices.

Minimum data set

A minimum data set of information was generated, as in Phase 1. Additional data on residents were obtained using the Clinical Dementia Rating Scale,228 completed by care home managers and senior carers.

Sample

Analytic methods

Interviews and care note data were analysed as in Phase 1. The facilitated meetings were digitally recorded, anonymised and analysed thematically using NVivo version 8 to organise and manage the analysis. Two researchers were involved in the analysis and discussion of key themes and issues.

Costing analysis

Care home characteristics

Care homes were a mix of provider type, size, location, building structure and religious affiliation (Table 55).

Resident funding status and care home fees, as well as staff numbers, working patterns and qualifications are shown in Tables 56–58.

Organisation of care

All of the care homes operated a key worker system with a named care worker. Most operated teams headed by a senior care worker/team leader and tried to keep the same teams for continuity of care. Only one care home had permanent night staff. Training in most care homes was ongoing and in-house (see Appendix 55, Table 106). Frameworks for structured end-of-life care, such as the LCP215 or GSF213 for Care Homes, appeared rarely, if ever used (see Appendix 55, Table 107).

Resident needs

Resident needs are shown in Appendix 56, Table 108. Overall, care home 1 is characterised by the highest concentration of dependent older people with advanced dementia. Residents at care homes 3 and 4 had similar levels of dependency but less of dementia-related dysfunction. Residents in care homes 5 and 6 were more heterogeneous, evidenced by wider ranges of cognitive and functional ability.

Reported number and location of resident deaths in the year prior to the study are shown in Table 59. Between 30% and 50% of residents who had died in the year prior to the study died in hospital or in a hospice.

Access to primary and specialist palliative care

Phase 1 resident baseline characteristics (n = 133)