Introduction

Pre-existing cognitive impairment in hospitalised older people is estimated at between 15% and 42%.1–5 Many of these patients may not have had a previous diagnosis of dementia.3 6 This has led to concerns about providing adequate care for older people with dementia, and policy development to improve the identification of dementia in older people in hospital.7–10 In the English National Health Service (NHS) initiatives were introduced to increase awareness and improve the identification of dementia in secondary care.11–14 In 2012–2013, the Department of Health introduced a mandatory policy requiring that all hospitals across England routinely carry out cognitive assessments with ≥75-year olds who have had an emergency admission and that those identified as potentially having dementia are appropriately assessed and referred. The rationale for case finding in hospital was that early identification of cognitive impairment would lead to a timely diagnosis, informed patient care and improved health outcomes for older people.15 16 This was implemented in the absence of any randomised controlled trial evidence to know the impact of such an approach.

Although cognitive assessment tests can detect undiagnosed dementia,17–19 dementia does not fit the criteria for screening programmes and there are concerns over the impacts of case finding as well as screening for dementia.20 21 Moreover, in practice the distinction between screening and case finding is often not clear to clinicians and the public.22 23 Healthcare professionals have been reported as viewing case finding as a financially and politically driven policy with little or no evaluation of patient outcomes.24–27 In a previous paper, we reported how the variation in approaches to dementia case finding has meant that information communicated to primary care is inconsistent and has an impact on how well general practitioners (GPs) can effectively plan patients’ treatment and care once they are home in the community.27

Little is known about the wider impacts of dementia case finding on patient and carer experiences and on their access to dementia-related or other care services. The overall aims of this study were to explore the impacts of dementia case finding on older people identified by hospital case finding as warranting further investigation, their family carers and their use of health service and care services. For example, were they being reassessed in primary care and, if appropriate, receiving specialist referral, being given a diagnosis or receiving new services?

Methods

This study involved in-depth exploratory interviews with former hospital patients and their family carers. A range of qualitative and quantitative data was collected. This was the second phase of the CASCADE study (CASe finding in hospitals—impacts on CAre for people with DEmentia).28 Phase I results are reported in full elsewhere.27

We recruited former patients of two participating hospitals in the East of England. Eligible patients were aged at least 75-year olds, had been acutely admitted 6–12 months previously, had been flagged by dementia case finding as warranting further investigation about their cognition and were community dwelling. Family carers of these patients were also recruited for interviews. Invitation letters were sent to eligible participants from the hospital elderly medicine consultant under whose care they had been admitted, together with an information sheet and response form.

Semistructured interview schedules were developed to gather patient and family carers’ experiences and perceptions of dementia case finding and the impacts of case finding on their subsequent treatment and care (online supplementary file 1). Demographic data were collected and the Mini-Mental State Examination (MMSE)29 and EuroQol (EQ-5D-5L)30 validated scales were used to characterise patients’ cognition and health-related quality of life, respectively. For the analysis, those with a score below 24 on the MMSE were classified as cognitively impaired and those with a score of 24 and above were categorised as cognitively intact. EuroQol’s validated proxy version of the EQ-5D-5L was undertaken with carers to characterise patients in the minority of cases when a carer were interviewed separately from the former patient. Completed EQ-5D-5Ls were scored against the published algorithm and results from the scale’s five domains were used to characterise participants’ levels of difficulty in each.31 Data were collected on patients’ self-reported use of inpatient and outpatient services, day activity services since the index hospital admission, community care services and current medication. Service use was collected using relevant sections of the Client Service Receipt Inventory (CSRI)32 using ‘Visio’ timelines33 to facilitate recall and recording of events such as referrals or the start of new services. Two female researchers (A-MB, JF) carried out the interviews in the patients’ and family carers’ homes across four counties in the East of England (Bedfordshire, Cambridgeshire, Essex and Hertfordshire). Interviews lasted about an hour. Participants gave signed consent before the interview and interviews were tape recorded, transcribed and anonymised.

An initial thematic framework was developed by the research team (A-MB, FB, JF) which was informed by the research questions and coding a selection of transcripts. The codeframe and the anonymised transcripts were transferred into NVivo V.11 and the data analysed using thematic content analysis.34 This enabled the key features of patients’ and carers’ experiences to be elicited from the data. Quantitative data were entered into an Excel spreadsheet.

Patient and public involvement

Two well-established patient and public involvement groups were involved in the study; the University of Hertfordshire Patient Involvement in Research Group, and the University of Cambridge’s Public Involvement in Research into Ageing and Dementia Group. Four members of these groups acted as a User Reference Group for the study. They provided feedback on the aims of the study, study documents such as patient information sheets, and commented on preliminary findings from the qualitative analysis.

Findings

We conducted a total of 28 interviews involving 49 participants. Of those, 24 were former patients who had been identified as having a concern about their memory during an unplanned hospital admission and 25 were carers. Nineteen interviews were held jointly with the patient and their family carer(s). All patients had received a memory assessment during their hospital stay which indicated that there was a cause for concern over their cognition which warranted further investigation.

Table 1 summarises descriptive characteristics of the former patients in our study. Both their median and mean age was 85 (range 79–94) and just over half (53%) were women. The median MMSE score was 23.0 (IQR 16.5–26) indicating mild cognitive impairment, but there was considerable variation with scores ranging between 6 and 30. Patients had considerable levels of physical impairment and difficulty with activities of daily living, for example, nearly half were unable to walk and over a third had severe problems washing and dressing themselves. The EQ-5D-5L scoring system31 can provide scores between 1 and −0.594. This range has anchor points where 1 is considered to be full health and 0 is considered to be equivalent of dead. The scoring algorithm allows some scores to be <0, that is, worse than dead. Twenty-four former patients provided a value for their own health state and the mean value for these individuals was 0.39 (95% CI 0.25 to 0.52). There were 25 carers who provided a valuation of their own health with a mean value 0.78 (95% CI 0.69 to 0.86). For comparison, population norms are available for those aged over 75 years for the EQ-5D-3L39 which give values of 0.75 and 0.71 for males and females, respectively. The value obtained here for former patients is considerably below these values.

Impacts of dementia case finding on service use

Nine people had been assessed postdischarge either in a memory clinic or by an old age psychiatrist but two of these had already been referred by their GP prior to hospital admission. For the seven people who had been referred to memory services since hospital discharge it was not clear whether these referrals were a result of hospital case finding or not. Patients referred for investigation since discharge (n=7) generally had lower MMSE scores than those not referred (MMSE of 17 (IQR 15–22) versus 25 (IQR 19–26)). In the non-referred group, two patients had MMSE scores that indicated severe cognitive impairment and five had scores indicating moderate cognitive impairment. Five patients who were referred since case finding had been started on dementia medication (eg, donepezil, memantine), but two of these had stopped taking this medication within a fortnight because of side effects.

Estimates of the costs incurred by former patients of dementia case finding for the period between index stay discharge and interview are given in table 2. We had available data for 28 former patients, 24 where the patient was present at interview and 4 where only a carer was present. Precise costing was not generally possible, so these costs should be taken as indicative. This group have a mean estimated cost of £5180 (95% CI £2394 to £7967) in the period following the index stay. There is also considerable uncertainty around these cost estimates as indicated by the CIs. The major drivers of cost were inpatient and outpatient services, district nursing and home care worker. Comparatively few costs were specifically related to cognitive services, apart from memory clinics and some spending on memory-related drugs. There was also one GP home visit that was specified as related to memory assessment.

Two main themes emerged from the qualitative analysis (1) attitudes and experiences of dementia case finding and (2) patients’ and carers’ follow-up experiences. Examples quotes (Q=quotes) are given in tables 3 and 4.

Discussion

Dementia case finding in acute hospitals involves the routine cognitive assessment of older people during an unplanned hospital stay and aims to improve the identification of patients with dementia in hospitals so that they could be flagged to their GP for further investigation and referral. This study aimed to explore patients’ and their families’ experiences of how this worked in practice. It was conducted in two acute hospital settings in the East of England where dementia case finding was implemented with little or no evaluation of its effectiveness in terms of improving patient outcomes. To our knowledge, this is the first study to explore the views and experiences of older patients and their family carers to case finding and to estimate the economic impact for the health service.

We spoke to patients and carers of patients who had been identified as having a concern about their cognition during their hospital admission. Although the median MMSE score of included participants at interview indicated that most patients had mild cognitive impairment, MMSE scores ranged from 6 to 30. The findings from this study suggest that there was little follow-up from patients’ GPs postdischarge and few further investigations or referrals or medication instigated due to dementia case finding outcomes; although in some cases, patients had already been seen by their GP prior to their hospital stay about their cognition. The lack of follow-up may not be surprising in light of the fact the outcomes of dementia case finding are often poorly reported to GPs.27

Patients and their families had mixed attitudes towards dementia case finding but many did not remember the assessment and/or did not appear to have been informed about what this would mean in terms of their subsequent care and treatment. There was also concern that the acute hospital setting might not be the best place in which to assess cognition. Indeed, 10 former patients had MMSE scores which indicated no cognitive impairment despite having been picked up by the case finding process when in hospital. In a previous paper, we reported that dementia case finding has raised awareness about dementia within hospitals.27 However, delirium is often undetected in hospitals40 and can increase the likelihood of reporting false positives and result in misdiagnosis of dementia.41

The results of the costing evaluation showed that participants/patients had average costs of £5180. The mean length of time over which these costs were incurred was 283 days, excluding the initial stay. Although these individuals had been identified from dementia case finding, the proportion of those costs attributed directly to services related to cognition appear to be small (~6%). This is not surprising as we found that few services had been instigated due to case finding. While the costs of services initiated in the community following case finding did not appear to be high, from health professionals’ reports of how case finding has been implemented in the previous phase of the CASCADE study27 it is likely that case finding is associated with significant costs within the hospital itself, though to our knowledge no research to date has quantified these yet. This suggests that an evaluation of the costs of such a policy, including the opportunity costs within the policy implementations system (Public Health England, NHS England) are warranted.

Conclusions

This study highlights that dementia case finding in hospitals did not necessarily lead to a GP follow-up or referrals for further investigation or lead to new supportive services being put in place. Often patients and their family carers were not informed that memory assessments had been carried out while in hospital or what the outcomes of the assessment would mean in terms of their future care and treatment. There is a need for a more evidence-based approach to the initiation of mandatory initiatives such as case finding. This includes evaluation of the costs of such a policy, including the opportunity costs within the policy implementations system.