Medical Crises in Older People

Introduction

The AMOS study also allowed us to produce patient-based UK NHS and social care costs for this group of older patients (aged 70+ years) who attended an AMU and were then discharged home. We estimated these costs partly to ensure that we had robust methods for our later RCT and partly because this group of patients has not previously been studied greatly yet we were aware of great interest elsewhere about the use of resources and hence service costs in this group of patients using emergency and non-elective care. This study has been published. 45

Methods

Data were collected retrospectively for 90 days from recruitment using Electronic Administration Record information extracted from various health-care services. Hospitalisation data were collected for 644 patients in Leicester and Nottingham (23/667 withdrew consent for their resource use data to be obtained). Hospital care data included inpatient stays, day cases and outpatient and critical care. Social care data were obtained for all participants. In a subset of 456 participants (in Nottingham), further approvals and access were gained to obtain data from general practices, ambulance services and intermediate and mental health care services. Resource use was combined with national unit costs46,47 to derive total patient costs. The costing perspective was NHS and local authority (social services) expenditure.

Results

Data were obtained from 48 out of 118 general practices (250/456 Nottingham participants) despite exhaustive attempts to acquire data from all practices. Thus, costs from all sectors were available for 250 participants. The mean (95% CI, median, range) total cost for this subgroup was £1926 (£1579 to £2383, £659, £0–23,612). Secondary care made up 76% of costs. Other costs were for primary care (10.9%), ambulance service use (0.7%), intermediate care (0.2%), mental health care (2.1%) and social care (10.0%). The 10% of the most costly participants accounted for 50% of the overall costs.

Discussion

Secondary care costs were the main cost driver in this patient group. Despite the expectation that this group would mainly incur ambulatory and community costs, many of these costs contributed little in this patient group. Consideration should be given to focusing primarily on secondary care costs in some research instances, such as when scoping work reveals that secondary care costs are likely to be dominant. Nevertheless, in view of the fact that we aimed to influence community care, we elected to use the same methods in our RCT and to ascertain ambulatory and community costs as well as hospital costs.

Methods

A multicentre, individual-patient RCT comparing the intervention with usual care was undertaken. The intervention was interface geriatrics,48 as described in the previous section. Patients aged ≥ 70 years discharged from two UK AMUs (Nottingham and Leicester) and scoring ≥ 2 on the ISAR risk screening tool were recruited prior to discharge and randomised to receive the intervention or usual care. Carers of participants were also recruited. Follow-up was by postal questionnaire 90 days after randomisation. The primary outcome was the number of days spent at home (for those admitted from home) or the number of days spent in the same care home (if admitted from a care home). Secondary outcomes included mortality, institutionalisation, hospital resource use and scaled outcome measures (including quality of life, disability and mental well-being).

A postal questionnaire was sent at 90 days to carers or family members for whom there was baseline information. Baseline and follow-up carer measures were:

• carer strain: Caregiver Strain Index51

• carer-specific quality of life52

• generic quality of life: EQ-5D. 38

In view of the small numbers, carer outcomes were not compared between groups.

A purposive sample of patient participants and carer participants was selected to have a semistructured qualitative interview at home 30 days after discharge. Analysis was performed in parallel with recruitment and interview and so the content of the interviews developed as data emerged. The emerging findings were not shared with the community geriatricians during the trial. Selection was determined by the researcher on the basis of emerging themes and recruitment continued until data saturation. The interviews covered the problems that led to admission, what participants perceived happened in hospital, what they wanted and expected, what helped and what did not help, discharge arrangements, resettlement at home, impact on everyday activities, transfer of care to community services and ongoing problems.

In the economic study, 417 participants (205 allocated to the intervention arm) were analysed at 90 days’ follow-up. Data were collected retrospectively for 180 days from recruitment using Electronic Administration Record information extracted from various health-care services, as for the AMOS trial, but in addition the cost of the interface geriatricians was also included. Quality-adjusted life-years (QALYs), based on EQ-5D valuations at baseline and follow-up, were obtained for 254 (60.9%) participants (127 per arm). Multiple imputation by chained equations was applied to deal with missing QALY values. Costs and QALYs were adjusted by baseline characteristics using regression methods. The difference in mean total costs and QALYs between arms and incremental cost-effectiveness ratios (ICERs) were estimated, handling uncertainty by non-parametric bootstrapping.

Results

Of 1001 eligible patients, 433 were recruited: 217 in the control group and 216 in the intervention group. The two groups were well matched for baseline characteristics, and withdrawal rates were similar in both groups (5%).

In total, 201 (98%) received the intervention as intended with 133 (66%) having a response beyond the initial assessment; 122 of these were seen at home. The range of actions taken by the geriatricians was largely as intended and as might be delivered in routine practice, most commonly liaison with other practitioners, medication changes, giving health advice and referral for rehabilitation, further diagnostic tests and additional medical follow-up.

The mean number of days spent at home over 90 days’ follow-up was 80.2 in the control group and 79.7 in the intervention group (95% CI for the difference in means –4.6 days to 3.6 days; p = 0.31). There were no significant differences in any of the secondary outcomes.

In total, 65 carer participants (15%) were recruited from the 433 patient participants but only 46 (11%) were true carers (the others being informants). The 46 true carers gave a median of 2 hours per day physical care and 3 hours per day of supervision. Of these, 17 (37%) carers were spouses and 26 (57%) were children, with three (7%) ‘other’ carers. The mean age of the carers was 61.5 years and 31 (67%) were female, 22 (48%) were co-resident with the patient participant and 15 (33%) were in paid employment. At baseline, carer participants had high carer strain [median Caregiver Strain Index 7, interquartile range (IQR) 3–9] and poor carer quality of life (median carer-specific quality of life 8, IQR 5–9) and the median EQ-5D score was 0.81 (IQR 0.69–0.85). There was no significant change in these variables at follow-up.

In total, 18 older patients and six of their informal carers were interviewed. The thematic analysis revealed six themes, some with subthemes:

• Staff recognition (subtheme: dispersal of blame). The majority of the patients wished to express the positive attributes of the staff on the AMU, saying that they felt well looked after on the ward. When problems were identified the patients were keen to point out that they did not blame the staff but rather apportioned blame on external factors.

• Incomplete satisfaction (subthemes: perceived lack of treatment, constant disturbance, waiting, poor communication, discharge uncertainty, carer frustration). Although the patients wanted to portray a positive image of the staff on the AMU, all but two spoke about areas of dissatisfaction and these were clustered around the six subthemes.

• Stoicism (subthemes: ageing assumptions, modest expectations, minimisation of needs, passive acceptance). There was an underlying attitude of stoicism. Patients did not have high expectations around improving the state of their health in light of the ageing process. Similarly, they were tolerant and understanding of any weaknesses experienced on the AMU. The patients had low expectations of hospital care, resulting in passive acceptance of any weaknesses experienced.

• Eager to go home. Although the patients recognised that they needed hospital-based assessment, they did not want to remain on the AMU for any longer than was absolutely necessary.

• Nebulous grasp of the geriatrician role. The patients spoke about the geriatricians possessing a pleasant bedside manner but the majority of patients were unsure what the geriatrician had done for them.

• Outstanding needs (subthemes: unresolved health issues, unresolved daily living needs, impact on informal carer, value of independence). Patients had both outstanding health needs and daily living needs, which were not addressed as part of their stay on the AMU. These impacted on their informal carers. Despite the help received with daily living activities a lot of the patients voiced a desire to complete these activities themselves rather than have others complete them for them.

In the complete-case economic analysis involving the subgroup of 254 patients with EQ-5D valuations at baseline and follow-up completed, the differences in mean total costs and QALYs (intervention vs. control) were +£138.9 (95% CI –£1139.8 to £1434.5) and 0.004 (95% CI –0.012 to 0.020), respectively, resulting in an ICER of £38,583 per QALY, with a 47% probability of the ICER being < £30,000 per QALY. In the adjusted cost-effectiveness analysis, the differences were +£146.4 (95% CI –£60.6 to £340.7) and 0.002 (95% CI –0.006 to 0.011), respectively, resulting in an ICER of £73,200 per QALY, with a 36% probability of the ICER being < £30,000 per QALY).

In the full-sample economic analysis (imputation of missing QALY values), the mean cost of inpatient care was lower in the intervention arm (–£211.7, 95% CI –£1097.9 to £471.6) whereas all other care costs were higher (social care +£220.1, 95% CI –£299.5 to £691.5; day cases +£155.6, 95% CI £31.1 to £280.3; outpatient care +£46.3, 95% CI –£69.5 to £166.2). The intervention cost was +£115.6 per case (95% CI £106.0 to £125.8). In an adjusted cost-effectiveness analysis, the total cost for the intervention group was higher (+£213.4, 95% CI £94.8 to £331.2) with no QALY gain (–0.001, 95% CI –0.009 to 0.007) and so the intervention was dominated by standard care (3% probability of the ICER being < £30,000 per QALY).

Discussion

This specialist geriatric medical intervention applied to a high-risk population of older people attending and being discharged from AMUs had no impact on patient-level outcomes or subsequent use of secondary care or long-term care. It was not cost-effective.

The interview findings indicated some areas in which AMUs could improve patients’ experiences. They also demonstrated that most patients had background conditions and that the trip to the AMU contributed little to their management from their perspectives and was confined simply to the assessment of an acute medical condition (which for these patients was not sufficient to warrant admission to hospital from the AMU). They illustrate that the interface geriatricians seemed to have little impact on the main issues affecting the health and well-being of these patients, which were not the medical crises that had precipitated their presentation but the underlying health conditions in which these crises arose.

Together, the findings support the deduction from the AMIGOS findings that a more integrated follow-up response after an AMU attendance is warranted, involving chronic disease management if health outcomes are to be improved and preventing hospital admission if costs are to be minimised.

Methods

Participants were from two sites at the Nottingham University Hospitals NHS Trust, an 1800-bed teaching hospital providing sole general medical and trauma services for a population of approximately 660,000 people. Individuals aged ≥ 70 years with an unplanned admission to 1 of 12 wards (two trauma orthopaedic wards, three acute geriatric medical wards and seven general medical wards) were eligible for inclusion. Exclusion criteria were unwillingness to be screened, being unconscious or too ill to be interviewed up to the fifth day of admission and an inability to speak English with no available interpreter. Consecutive admissions were identified from the hospital administration computer system and patients were approached between day 2 and day 5 of admission.

A two-stage assessment procedure was used. The first stage identified people unlikely to have a mental health problem. The second stage used more detailed assessments to characterise problems. The first-stage assessment used the Abbreviated Mental Test Score,63 the four-item Geriatric Depression Score,64 the two-item Primary Care Evaluation of Mental Disorders anxiety screen,65 the four CAGE questions for alcohol misuse (Cut down, Anger, Guilt, Eye-opener)66 and a question asking ward staff if there was any other reason to believe that a mental health diagnosis might be present. Participants screening negative for cognitive impairment (Abbreviated Mental Test Score of > 7), depression (four-item Geriatric Depression Score of < 1) and alcohol abuse (CAGE score of < 2) and negative on the mental health diagnosis question, or who scored only on the anxiety questions, were excluded from further study. Patient–carer pairs were recruited from those screening positive for cognitive impairment if a carer could be identified and was willing to participate.

Participants were followed up 180 days after recruitment. Information was collected from the participants, family members and other informal or professional carers. Information on readmissions and total number of days spent in hospital was collected from hospital administration systems. Mortality, and dates and types of care home placements (residential or nursing, permanent or respite) were ascertained from the hospital administration systems, the patients’ GPs, the carer informants or care home. Surviving participants were interviewed at home with a carer or, if this was not possible, by telephone with an informant. Participants were tested for cognitive function and carers provided information on behavioural and psychological symptoms and ADL. Economic data were collected retrospectively for 180 days from recruitment using Electronic Administration Record information, as for the AMOS study (see Chapter 2).

Patient outcomes were survival to 180 days; days spent at home, defined as 180 minus the total number of days spent in hospital, in a care home or dead for patients living in the community at admission and as 180 minus the total number of days spent in hospital, in a new care home or dead for patients living in a care home at admission;67 change in ADL, defined as an increase or decrease of ≥ 2 points on the Barthel index36 at follow-up compared with admission and before the acute illness.

Carer participants were asked at baseline and at 6 months to complete a questionnaire, with help as required, giving demographic and care-giving details. It included the Caregiver Strain Index,51 with a score of ≥ 7 indicating high strain.

Electronic administrative records were sought for 6 months post admission from health services (general practices, hospitals, ambulance transport services, intermediate and mental health-care services) and social care services. Standardised costs were applied to all resource use types.

Of 1004 patients screened, 36% had no mental health problems or had anxiety alone. Of those screening positive, 250 took part in the full study. Adjusting for the two-stage sampling design, 50% of admitted patients aged > 70 years were cognitively impaired, 27% had delirium and 8–32% were depressed. In total, 6% had hallucinations, 8% delusions, 21% apathy and 9% agitation/aggression (of at least moderate severity). Of those with mental health problems, 47% were incontinent, 49% needed help with feeding and 44% needed major help to transfer.

Of the 250 patients recruited to the study, 180 were cognitively impaired and had carers willing to take part. After 6 months, 78 patients (31%) had died and 100 carers were followed up. Carers’ own health, in terms of mobility, usual activities and anxiety, was poor in one-third of cases. At the time of admission, high carer strain was common (42% had a Caregiver Strain Index ≥ 7), particularly among co-resident carers (55%). High levels of behavioural and psychological symptoms at baseline were associated with more carer strain and distress. At follow-up, carer strain and distress had reduced only slightly, with no difference in outcomes for carers of patients who moved from the community to a care home.

The median number of days spent at home for participants was 107.5 days (IQR 0–163 days); 38 (15%) spent > 170 days at home. The mortality at 180 days was 78 (31%), 104 (42%) were readmitted and 46/192 (46%) community-dwelling patients moved to a care home. In surviving participants, half improved in ADL ability at 180 days from admission, but only 24% recovered to their pre-acute illness baseline and 36% showed further decline in function during follow-up.

Health and social care costs were derived for the 247 participants for whom resource use data were available. Primary care data were available for 122 (49%) participants because of the reluctance of some general practices to allow access to data. In this subset with full data, the mean (95% CI, median, range) total cost of care was £9842 (£8573 to £11,256, £7717, £715–48,795). Secondary care contributed > 80% of the costs, with the remaining costs incurred in social care (10.7%), primary care (6.7%) and other sectors (2.1%).

In summary, the Better Mental Health cohort study showed that a large number of older people admitted to a general hospital had mental health problems, particularly cognitive impairment, that their outcomes were poor and that their use of health and social care resources was high.

Methods

Patients were recruited who had been admitted for acute medical care to the Nottingham University Hospitals NHS Trust. Suitable patients were identified on the hospital AMU and were randomly allocated between the MMHU and standard care. Randomised patients were subsequently approached for recruitment to the study. This approach was necessary so that patients could be moved from the admission unit to the wards at any time of the day or day of the week at the pace required for the efficient operation of the hospital, yet allowing sufficient time for patients to be recruited ethically. Participants were aged > 65 years and had been identified by the admissions unit physicians as being ‘confused’. A family member or carer was recruited if available and willing to act as an informant.

Potentially suitable patients were entered into a computerised screening log and, if a bed was available on the MMHU, randomised 1 : 1 between the unit and standard care in a permuted block design, stratified for previous care home residence. Readmitted patients were assigned their original allocation. Regardless of allocation, patients had access to standard medical and mental health services, rehabilitation and intermediate and social care.

Standard care wards included five acute geriatric medical wards and six general (internal) medical wards. Practice on geriatric medical wards was based on CGA and staff had general experience in the management of delirium and dementia. Mental health support was provided, on request, from visiting psychiatrists, on a consultation basis. The 28-bed MMHU was an acute geriatric medical ward with five enhanced components, as described earlier.

The primary outcome was number of days spent at home (or in the same care home) in the 90 days following randomisation. In addition, a range of health status outcomes was measured: quality of life (DEMQoL,75 EQ-5D,76 London Handicap Scale77), behavioural and psychological symptoms [Neuropsychiatric Inventory (NPI)78], dependency in personal ADL,36 cognitive impairment [Mini-Mental State Examination (MMSE)79], carer strain (Carergiver Strain Index51) and carer psychological well-being (GHQ-1237). Carer satisfaction was measured on 10 dimensions of care (overall, admission, car parking, feeding, medical management, being kept informed, dignity and respect, the needs of a confused patient, discharge arrangements, timing of discharge) using Likert scales (very/mostly satisfied, mostly/very unsatisfied; items taken from Counting the Cost80).

Structured non-participant observations of the experience of care on study wards were undertaken using Dementia Care Mapping. 81 Two trained researchers observed the care of 90 randomly subsampled participants. Observations were made every 5 minutes for 6 hours per patient. Clinical staff were not aware which patients were being observed. Quantified mood and engagement scores, activity, noise and staff interactions that significantly addressed or disregarded patients’ emotional and psychological needs (‘personal enhancers’ and ‘personal detractors’) were recorded, according to strict definitions. Inter-rater reliability was assessed throughout the study and was satisfactory (Cohen’s kappa between 0.50 and 0.85).

Outcome assessments were carried out by research staff who were not involved in recruitment or baseline data collection and who were blind to allocation. Carer satisfaction with hospital care was ascertained through a telephone call 1–3 weeks after discharge. Health outcomes were ascertained during interviews with patients and carers at home 90 days (±7 days) after randomisation. Routine health service records were examined for service use, mortality and readmissions.

In total, 40 family carers were purposively recruited from participants in the RCT, 20 from each setting, and took part in face-to-face semistructured interviews. An interview schedule was constructed to ensure that critical topics were covered, such as patient admission and settling in to the ward; carer relationship with staff; the ward environment; patients’ daily routines such as sleeping, meals, hygiene and activities; privacy and dignity; care and medical treatment; and discharge planning. Participants were encouraged to discuss both what they considered worked well and what they considered worked not so well on wards relating to quality of care. Interviews were conducted in the carers’ homes and consent was obtained to audio record interviews. Participants were reassured that privacy, confidentiality and identity would be protected. Interviews were transcribed verbatim and were coded for themes, which were compared and contrasted between settings to provide a detailed understanding of participants’ experiences and if and how the intervention added to carers’ perspectives of the quality of care.

A total of 22 ward staff from the MMHU were purposively recruited to take part in face-to-face semistructured interviews. The breakdown of the staff interviewed was as follows: two deputy ward managers, six general nurses, three mental health nurses, one student nurse, two occupational therapists, three health-care assistants, two activity co-ordinators, one junior doctor, one receptionist and one cleaner. The mean age of the sample was 37 (range 20–64) years and 15 (68%) were female. Length of experience in the profession ranged from 4 months to 29 years. Interviews lasted between 30 and 90 minutes. An interview schedule was constructed to ensure that the following topics were explored: education and training, job satisfaction, care of patients with dementia, team working, communication with carers and organisational barriers to change in practice and culture.

Twenty-six patients were approached and recruited from a sample of cognitively impaired patients aged > 65 years who had been recruited to the study. A trained dementia researcher assessed whether a traditional semistructured interview was appropriate or not by using MMSE scores combined with a general assessment of patients’ current cognitive function and conversation skills. All interviews were conducted on hospital wards and most were carried out at the bedside because of patients’ levels of illness and mobility or lack of an alternative location. Participants lacking capacity or appropriate communication skills were offered an interview using a Talking Mat. This is a low-tech, alternative and augmentative communication tool that uses images to explore specific topics and provides a visual scale that enables people to express their general feelings about individual options. This tool has been successfully used to assist communication for people with cerebral palsy,82 aphasia,83 learning disability84 and Huntington’s disease. 85 Research suggests that this communication tool can be used effectively with people at all stages of dementia. 86 However, using Talking Mats as a research tool in an acute hospital setting had not been done before. We explored its use as an adjunct to the TEAM interview work.

In the economic analysis, 599 (MMHU n = 309) participants were analysed at the 90-day follow-up, at which point 139 (MMHU n = 68) had died. Health (inpatient stays, day cases, outpatient care, critical care, ambulance service use, mental health trust and primary care) and social care resource use data were collected and combined with unit costs (from the NHS and personal social services perspective, cost year 2012/13) to estimate total costs. Primary care and inpatient resource use data were obtained for 468 out of 599 (78.1%) and 595 out of 599 (99.3%) patients respectively. For the remaining services, resource use data were complete. The per-patient additional cost of the MMHU was calculated as the excess health-care costs incurred at this ward over a period of 90 days (trial follow-up) compared with usual care in a general or geriatric ward, averaged across each patient allocated to the MMHU. QALYs, based on EQ-5D valuations at baseline and follow-up, were obtained for 272 out of 599 (45.4%) patients (MMHU n = 139), including 62 (MMHU n = 30) who had died by the follow-up [assumed baseline utility (EQ-5D valuation) until date of death]. In a complete-case cost-effectiveness analysis, including 209 out of 599 (34.9%) patients with complete QALY and cost data, the differences in mean total costs and QALYs between arms and the ICERs were estimated, handling uncertainty by non-parametric bootstrapping. Costs and QALYs were adjusted by baseline characteristics using regression methods. Additionally, a cost analysis was conducted for the complete-case resource use data set [complete inpatient and primary care data, 466 out of 599 (77.8%) patients].

Results

Between July 2010 and December 2011, 310 patients were recruited from the specialist unit and 290 from standard care. The recruitment rate was slightly higher on the specialist unit (71% vs. 66% of those randomised). Those not recruited were of similar age and sex and were from a similar area of residence (postcode), but care home residents assigned to the specialist unit were more likely to be recruited than those assigned to standard care (73% vs. 56%). In total, 462 participants lacked mental capacity, 227 (73%) assigned to the specialist unit and 235 (81%) assigned to standard care.

There was no statistically significant difference in the number of days spent at home between settings [median 51 days MMHU vs. 45 days standard care; 95% CI for difference –12 days to 24 days; p = 0.3). The median index hospital stay was 11 days in both settings and the mortality rates were 22% and 25% (95% CI for difference –9% to 4%), the readmission rates were 32% and 35% (95% CI for difference –10% to 5%) and the new care home admission rates were 20% and 28% (95% CI for difference –16% to 0%) for the MMHU and standard care respectively. Participants on the MMHU spent significantly more time with positive mood or engagement (79% vs. 68%, 95% CI for difference 2% to 20%; p = 0.03) and experienced more staff interactions that addressed their emotional and psychological needs (median four vs. one per observation; p < 0.001). More family carers in the MMHU group than in the standard care group were satisfied with care (overall 91% vs. 83%, 95% CI for difference 2% to 15%) and severe dissatisfaction was reduced in the MMHU group compared with the standard care group (5% vs. 10%, 95% CI for difference –10% to 0%; p = 0.004). There were no significant differences in any of the other outcomes.

In total, 20 carers each from the MMHU and standard care groups were interviewed. In the MMHU group this included two spouses, 13 daughters, two sons, one brother and two granddaughters and in the standard care group this included six spouses, six daughters, one granddaughter, five sons, one sister and one nephew. Seven of the patients from the MMHU were male and 13 were female, with a mean age of 87 (range 83–97) years, and 11 of the patients from standard care were male and nine were female, with a mean age of 85 (range 69–95) years. The main themes identified in exploring carer satisfaction related closely to met or unmet expectations and included activities and boredom, staff knowledge, dignity and personal care, the ward environment and communication between staff and carers. Neither setting was perceived as wholly good or wholly bad; however, greater satisfaction (and less dissatisfaction) with care was experienced by carers from the MMHU group. Carers were aware of improvements relating to activities, the ward environment and staff knowledge and awareness of the appropriate management of dementia and delirium. However, in some cases communication and engagement of family carers was still perceived as insufficient.

Health professionals suggested that working on the MMHU allowed them to provide better care than they had previously done to cognitively impaired patients. The six main improvements experienced by staff were across the following themes: confidence in competence; working with mental health professionals; increased knowledge of dementia; moving towards a person-centred acute model of care; improving coping strategies; and a positive change in attitudes towards patients with cognitive impairment. Staff commented positively about the skills mix of nursing care available to patients on the MMHU, specifically the introduction of three mental health nurses. Participants highlighted that this helped increase staff confidence and morale when staff were faced with unfamiliar or perceived challenging behaviour. Staff further commented that working on a specialised unit for patients with cognitive impairment had greatly increased their knowledge and awareness of dementia and delirium. Staff generally considered that they had a good understanding of the principles of person-centred care. A few nursing staff felt that the acute hospital setting was too task focused and an inappropriate place to deliver person-centred care. The specialist MMHU was considered a busy and sometimes challenging environment for the majority of staff interviewed. However, staff described a strong ward team spirit and supportive culture, which individuals highlighted helped improve stress-related coping strategies when dealing with unfamiliar situations. Participants acknowledged that their confidence in dealing with this patient group had increased. Staff expressed that this was closely related to the different types of training that they had received (educational and practical) for patients with cognitive impairment. Having a greater understanding of both dementia and person-centred care had helped staff display a more positive attitude towards this group of patients. Themes identified by participants with regard to improving patients’ and relatives’ experiences of care were staff–carer communication; staffing levels and resources; balancing an increased risk of falls against allowing patients to walk around freely; and organisational barriers to change in practice.

The use of Talking Mats increased the total number of patients able to be interviewed from eight to 15, but a substantial minority could not be meaningfully interviewed using either method:

• eight out of 26 (31%) were interviewed conventionally (mean MMSE 19, range 14–24)

• seven out of 26 (27%) were interviewed using Talking Mats (mean MMSE 9, range 1–18)

• 11 out of 26 (42%) were not interviewed conventionally or using Talking Mats (mean MMSE 12, range 0–24).

All of the eight patients interviewed using traditional semistructured methods were admitted to the MMHU. Five were female and three were male. Six had a prior diagnosis of dementia and all had cognitive impairment. Five themes emerged:

1. Feelings. Most of the participants reported positive emotions such as feeling content and enjoying the ward environment, but there were also less positive feelings such as boredom and isolation.

2. Memory and confusion. Of the eight patients interviewed, four appeared to be aware that they were, or had been, confused while in hospital.

3. Activity. Patients noted a lack of activity on the ward, with many references to them sitting by their bed. Only one patient reported that this was because of ill health. Although some patients felt frustrated by this, others did not seem to mind. Of the patients who talked about the activities room on the MMHU, two noted enjoying their time there and the organised activities conducted within. However, some patients spoke of not wanting to go to the activities room or take part in activities, preferring instead to remain inactive or wait for their family to come to see them instead.

4. Communication. This included lack of communication, communication regarding health and availability of staff to talk to.

5. Staff. Feelings towards staff were almost all positive. Care and kindness shown by staff was repeatedly mentioned. All of the patients reported being able to talk to staff when seeking assistance; however, all of the patients noted a lack of communication as well, with staff not communicating on issues such as health, discharge and patients’ likes and dislikes. Three patients talked to family members while they were in hospital, of whom two reported that their family member acted as a liaison between the patient and staff.

All Talking Mat interviews were conducted on the hospital wards (three on the MMHU and three on the standard care ward) and most took place at the bedside because of patients’ levels of illness, their mobility or the lack of an alternative location. Six of the seven participants who were interviewed using a Talking Mat had a previous diagnosis of dementia and most (n = 5) experienced delirium on admission. Functional abilities were poor and all participants had been acutely unwell, with comorbidities. The mean interview duration was 21 minutes (range 13–35 minutes). Five interviews were cut short because of increased confusion, cognitive decline or the effects of physical illness. Participants’ ability to express feelings about different aspects of the ward varied; with between four and 21 questions being answered. However, all participants were able to provide some information about their experiences. Participants on the MMHU placed 13 cards in the positive response (Thumbs Up) category whereas those receiving standard care placed 11 cards in the positive response category. In total, 22 cards were placed in the middle, neutral, section. Participants on the standard care ward placed more cards in the negative response (Thumbs Down) category (n = 10) than those on the MMHU (n = 5).

Five themes emerged from the data about the Talking Mat method of communication:

1. Communication. Participants understood how to use Talking Mats, which enabled them to express their feelings about aspects of care on the ward. The ability to carry out the interview varied between participants.

2. The person as an individual. Information from the Talking Mats coupled with dialogue expressed during the process provided information about individual differences and built a picture of who the participants were and their experiences of hospital care.

3. Cognitive impairment. Attention fluctuated during the interviews with all participants becoming distracted by something, for example the pictures on the Talking Mat or changing topics. ‘Confusion’ increased for most during the interview and was displayed in different ways, such as expressing delusions or becoming agitated.

4. Physical illness. Talking Mats served as a short-term, helpful distracter for those experiencing pain, although symptoms quickly returned and two interviews were terminated as a result.

5. Environment. Noise and lack of privacy was as much an issue as with standard interview techniques, but because of the frailty of the participants it was even harder to take them off the ward to be interviewed.

In the unadjusted complete-case cost-effectiveness analysis undertaken in the subgroup of 209 (MMHU n = 109) participants with complete QALY and resource use data, the mean total cost was non-significantly lower (–£584.8, 95% CI –£2375.9 to £1085.8) and the QALY gain was non-significantly higher (0.007, 95% CI –0.013 to 0.027) in the MMHU group than in the standard care group, giving a 60% probability of MMHU care being dominant and a 85% probability of the ICER being ≤ £30,000 per QALY. In an adjusted analysis, the mean total cost for the MMHU group was significantly lower than that for the standard care group (–£486.7, 95% CI –£854.6 to –£126.5) with no significant QALY gain (0.0003, 95% CI –0.0108 to 0.0117), giving a 53% probability of MMHU care being dominant and a 95% probability of the ICER being ≤ £30,000 per QALY.

In the cost analysis undertaken in the subgroup of 466 (MMHU n = 241) patients with complete inpatient and primary care data, the mean cost for MMHU patients was non-significantly lower than that for standard care patients for inpatient care (–£151.6, 95% CI –£1066.6 to £750.9), social care (–£304.0, 95% CI –£805.1 to £238.3), and all remaining services except for three services for which the costs were non-significantly higher for MMHU patients: primary care (£23.2, 95% CI –£17.5 to £66.0), ambulance service use (£3.5, 95% CI –£18.3 to £25.7) and mental health care (£9.8, 95% CI –£37.8 to £61.2). The cost of care was non-significantly lower in the MMHU arm (–£521.9, 95% CI –£1523.0 to £547.7) and the incremental total cost (taking into account the extra costs of the MMHU) was also non-significantly lower (–£334.2, 95% CI –£1389.7 to £719.2).

Discussion

Specialist care for people with delirium and dementia provided by the MMHU improved the patient experience and carer satisfaction although there were no convincing benefits in terms of health status or service use outcomes. The qualitative findings confirmed and elaborated on these findings and indicated causal links between the interventions that included the MMHU and the observed outcomes. The economic study showed a high probability that the MMHU was cost-effective, in part because of lower inpatient (despite the cost of the intervention itself) and social care costs.

These findings are valuable because patient experience and carer satisfaction may be more appropriate measures of success for frail older people approaching the end of their life, as these patients were. Not only do these findings support the notion of further development and testing of MMHUs, they also illustrate the broad principle that investment in, and delivery of, best practice in dementia care leads to demonstrable changes in patient experience. These results provide the justification for further investment in and evaluation of such units. They also show that investment in and delivery of best dementia care can be both cost-saving and cost-effective.

The qualitative findings from carers enrich the understanding of what carers identify as important domains of good general hospital care in this patient group. An important finding is that the amount of communication required by family carers cannot be underestimated. We found the extent of this surprising and beyond what we had planned for. New approaches to engagement with family carers are required, including the assessment of expectations and the giving and receiving of information. Meeting this need will require major changes to the way that acute wards operate and the re-prioritisation of staff time to enable this activity. Facilitating more hands-on care by family members may provide the quid pro quo to enable it within resource-constrained health-care systems. Organisational development methodologies should be explored in future attempts to implement such changes, alongside more staff-directed education and training interventions and incorporation in pre-registration education. Nurse leaders will play an important role in creating the conditions for delivering, and fostering a culture that rewards and raises the demand for, relationship-centred care for this population.

The findings of the staff interviews contrasted markedly with those observed only a few years earlier, as reported in the Better Mental Health study, with staff ill-prepared to look after patients with cognitive impairment. 9 Better-educated staff and the development of dementia-friendly environments will help nurture staff confidence and morale in working with this client group as well as improve patients’/family members’ experiences of and satisfaction with care. Developing a unique skill mix in this setting (the introduction of mental health specialist staff) and encouraging multidisciplinary working alongside increased education and training has allowed MMHU staff to share and deliver best practice. Staff who lack the knowledge, skills and confidence to care for dementia patients are unlikely to be able to undertake and support patients’ ongoing personal care needs, especially when faced with challenging or unfamiliar patient behaviour such as aggression, wandering and disrobing or refusing/not wanting to eat, particularly if communication is difficult. The major outstanding need, despite the efforts of the intervention to encourage staff to be proactive towards carers on the MMHU, is communication. Although it can be argued that staff on the MMHU had made positive changes towards the delivery of person-centred care, less success was achieved in the delivery of relationship-centred care. Relationships between staff (especially nurses) and relatives still need reforming, with more partnership and collaboration. Collaboration, in terms of shared decision-making and exchange of knowledge and information, has been shown to be particularly important for relatives’ satisfaction with hospital care of the elderly. 87 Organisational factors identified previously as impeding the development of effective nurse–family collaboration include a task-focused culture and workload; shift patterns and length; lack of training; education given and poor supervision; resistance to change; and bureaucratic issues. 9,88

The five themes arising from the patient interviews (feelings, memory and confusion, activity, communication and staff) illustrate the major domains of hospital experience as contemporaneously elicited. Given that these are the main themes of patient experience, they are the main domains that staff and providers should focus on to improve experience and hence satisfaction. The first two point to the need for staff to acknowledge and understand the feelings experienced by patients, including their own awareness of their cognitive problems, which contrasts with a more task-focused style of care that focuses on personal ADL. The third theme validates the importance of the focus on promoting activity as part of the development of the MMHU. Although the importance of communication and staff kindness is hardly a new observation, this study reminds us (if such reminding should be needed) that people with cognitive impairment remain sensitive to these issues – there is no sense that once one becomes confused one becomes so unaware of surroundings that communication and other behaviours are not appreciated.

The use of Talking Mats increased from only one-third to just over half the proportion of participants who the researchers felt would be able to undergo a meaningful interview. This tool appeared to enable people with more severe cognitive impairment to be interviewed. Thus, although this tool does not provide a solution for eliciting information from all patients in hospital, it appears to offer an improvement over conventional interview approaches.

Introduction

The first element of the care home workstream was to undertake a systematic review of the RCT evidence for interventions specifically for the residents of care homes. This had not been done previously and it was important to scope the extent of the evidence base to guide the eventual implementation of evidence-based interventions. This work has been published. 89,90

Methods

Appendix 31 shows the databases searched and the search strategy.

Results

In total, 3226 abstracts were identified and 291 articles were reviewed in full. Most were recent (median age 6 years) and from the USA. A wide range of targets and interventions was identified. Targets included behaviour (n = 44 studies), prescribing (n = 20), malnutrition (n = 20), influenza (n = 19), quality of life (n = 18), depression (n = 17), mobility (n = 13), oral health (n = 13), falls (n = 12), quality of care (n = 12) and urinary incontinence (n = 12). Interventions were often mixed and included pharmacological, educational, physical therapeutic and managerial interventions. Appendix 32 shows selected data extraction (results) tables.

Discussion

This study was the first to collate data from all RCTs conducted in care homes and represents an important resource for those providing and commissioning health care for this sector. The evidence base is developing rapidly. Several areas – influenza, falls and mobility – are appropriate for systematic review. For other topics researchers need to focus on outcome measures that can be compared and collated.

Introduction

The second element of the care home workstream was to survey residents of care homes. UK care home residents are understood to be poorly served by existing health-care arrangements. The aim of this study was to describe the health, functional status and health-care resource use of a representative cohort of UK care home residents to help estimate their needs and hence the health-care services required to meet these needs. This study has been published. 90,91 Appendix 33 shows the baseline data collection form, Appendix 34 the baseline interview form, Appendix 35 the follow-up data collection form and Appendix 36 the follow-up interview form.

Methods

An 180-day longitudinal cohort study was undertaken of 227 residents across 11 UK care homes (five nursing homes and six residential care homes), selected to be representative of nursing/residential status and dementia registration. 91 The Barthel index measuring dependency in personal ADL,36 MMSE,79 NPI,78 Mini Nutritional Assessment,92 EQ-5D76, GHQ-1237, diagnoses and medications were recorded at baseline and the Barthel index, NPI, GHQ-12 and EQ-5D were recorded at follow-up after 180 days.

A costing study was also performed. Data were collected retrospectively for 180 days from recruitment using information extracted from various health-care services including general practice (including direct and indirect patient-centred events, medication and wound management), hospital care (including inpatient stays, day cases, outpatient care and critical care) and wider secondary care (including the ambulance service, intermediate care and mental health care). The perspective for applying costs was that of the NHS. The differences in mean costs between care home types (nursing/residential) were assessed using Student’s t-tests with non-parametric bootstrapping employed to manage the non-normality of cost data.

Results

In total, 227 out of 323 (70%) residents were recruited. The median (IQR) Barthel index score was 9/20 (2.5–15.5), the median (IQR) MMSE score was 13/30 (4–22) and the median (IQR) number of medications was eight (5.5–10.5). The mean (standard deviation) number of diagnoses per resident was 6.2 (4). In total, 30% of the residents were malnourished and 66% had evidence of behavioural disturbance. Thirty-seven (16%) participants died and one left the area without forwarding details before the follow-up at 180 days. Sixteen of the participants who died were from residential care homes and 21 were from from nursing homes. The death rate varied significantly between individual homes (range 0–32% of respondents; χ² test p < 0.05).

The median (IQR) Barthel index score fell from 5 (1.5–8.5) at baseline to 3 (0–7) at follow-up for nursing home residents (Wilcoxon signed-rank test p < 0.01) but did not change significantly for residential care home residents or the cohort as a whole over time. There was greater behavioural disturbance in all groups at follow-up, with the median (IQR) NPI score increasing from 3 (0–10.5) to 5 (0–13), from 2 (0–6.5) to 4 (0–10) and from 6 (0–19.5) to 8 (1.5–14.5) for the whole cohort, residential care home residents and nursing home residents respectively (Wilcoxon signed-rank test p < 0.01 for all).

Secondary care resource use data were collected for all 227 participants (103 residential care home and 124 nursing home residents) over the 180 days. Primary care resource use and medication use were collected for a subset of 209 participants (90 residential care home and 119 nursing home residents) whose GPs allowed access. Out of a total of 227 participants 110 (48.5%) used secondary care services and out of a total of 209 participants 181 (86.6%) used either primary or secondary care services. Over the 180-day follow-up period there were 41 hospital admissions that resulted in an overnight stay, comprising 503 inpatient-days, and a further 11 day-case admissions. There were 763 general practice contacts, comprising 264 in-practice consultations and 499 consultations at home.

In the full cohort the mean (95% CI, median, range) total hospital and secondary care costs were significantly higher for patients residing in nursing homes than for those residing in residential care homes (£1254, £855 to £1858, £113, £0–17132 vs. £535, £331 to £922, £0, £0–9781; p < 0.02) and when accounting for all health-care services in the subset the mean (95% CI, median, range) cost was significantly higher for nursing home residents than for residential care home residents (£1669, £1277 to £2316, £556, £0–17566 vs. £945, £691 to £1410, £383, range £0–10399; p < 0.03).

In the full cohort, during the trial period there were 52 inpatient or day-case events across 43 participants; 96% (50/52) of these events were classified as an emergency and 82% (41/50) of the emergency events resulted in an inpatient stay. In the subset, hospital care contributed 59.4% of costs. Other sectors contributing costs were primary care (30.9%), the ambulance service (0.6%) and mental health care (9.0%); no participants used intermediate or critical care. The 10% most costly participants accounted for 51.5% of the overall costs.

Discussion

This cohort study demonstrated highly prevalent dependency, cognitive impairment, mild frequent behavioural symptoms, multimorbidity, polypharmacy and frequent use of NHS resources across both residential care home and nursing care settings. Effective care for such a cohort requires broad expertise from multiple disciplines.

The key cost-generating events external to the care home were hospital inpatient stays, and the results of this study suggest that care home-residing older people mainly require hospital care in emergency situations and that nursing home residents consume external health-care services at a significantly higher cost than residential care home residents. In general, however, the study participants were associated with lower levels of health-care costs from health-care services external to the care home than in the other two studies within the Medical Crises in Older People programme (see Chapter 2, Patient-based health and social care costs of older adults discharged from acute medical units, and Chapter 3, The Better Mental Health cohort study). These findings suggest that the most cost-effective model to provide health care for care home residents would most likely need to focus on preventing admissions to hospital and that it might be more cost-effective if targeted at nursing home residents.

Methods

This study was a qualitative interview study using a grounded theory approach. It was set in six UK care homes and included primary care professionals serving the homes. In total, there were 32 participants: seven care home managers, two care home nurses, nine care home assistants, six GPs, three dementia outreach nurses, two district nurses, two advanced nurse practitioners and one occupational therapist. Appendix 37 provides further details of the methods.

Results

Five themes were identified:

1. Complex health needs and the unstable and unpredictable nature of residents’ illness trajectories (illustrative quote from GP informant: ‘Because one day, they can be fine, the next day, they stop eating, and then they could linger for months, or the next day, they could die’).

2. A mismatch between health-care requirements and GP time.

3. Reactive or anticipatory health care (illustrative quote from GP informant: ‘In the past, we used to try and do anticipatory things like a little ward round once a week. And I think we just found that it wasn’t making a lot of difference to just letting the staff call us when they needed help. So we were putting more hours in without seeing very much for it’).

4. A dissonance in health-care knowledge and ethos (illustrative quote from GP informant: ‘The average general practitioner isn’t experienced enough . . . and you need a, basically another specialism going in and I think that would deliver better care to the patient’).

5. Tensions in the responsibility for the health care of residents (illustrative quote from a district nurse informant: ‘As a district nurse is a bit of an issue, because there are times when we have to go into a residential home to administer insulin when there are nurses there, trained nurses, and they will not administer the insulin because they’re saying we’re not insured, so that piles even more pressure, even more visits onto the district nurses’).

Care home managers and staff were pivotal to health-care delivery for residents despite their perceived role in social care provision. Formal health care for residents was primarily provided by one or more GPs, often organised to provide a reactive service that did not meet residents’ complex needs. Deficiencies were identified in training required to meet residents’ needs for both care home staff and GPs. Misunderstandings, ambiguities and boundaries around roles and responsibilities of health and social care staff limited the development of constructive relationships.

Discussion

The health care of care home residents was found to be difficult because the residents’ needs were complex and unpredictable. Neither GPs nor care home staff had enough time to meet these needs and many lacked the prerequisite skills and training, irrespective of the model of organisation employed. Anticipatory care was generally held to be preferable to reactive care, but attempts to structure care to make it more anticipatory were dependent on effective relationships between GPs and care home staff and their ability to establish common goals. Roles and responsibilities for many aspects of health care were not made explicit and this risked poor outcomes for residents. These findings help explain the concerns noted about the quality of health care for care home residents in the UK. Missed opportunities for partnership working were described, which should give rise to more rational approaches to service development.

Research preparation and conduct

Losses to follow-up

There is also a risk of missing follow-up data. There are several common reasons for this in studies of frail older people. Obviously, scaled health outcome data at follow-up will be missing in participants who have died over follow-up. Some participants will formally withdraw from the research but many will not formally withdraw but will simply decline to participate in outcome assessment, even if offered a postal questionnaire, telephone support or a home visit. Some will explain that they are too ill to complete questionnaires. This can be seen as implicit withdrawal. Great care has to be taken to strike an acceptable balance between the requirement to gather as much data as possible and placing an undue burden on vulnerable participants. From the experience of the Medical Crises in Older People studies (and previous studies conducted by the research team) the following processes produce the best results:

• Participants should be informed, when consenting, that outcome assessments will be requested in due course and those who are not willing to undertake these assessments at the outset should not be recruited.

• With cognitively impaired participants, recruiting a carer is recommended. Such a person can act as an advocate for the participant, can also be an informant and may often be a research subject in their own right. Ethically, including cognitively impaired people in research without a carer remains a challenge.

• Just before follow-up, hospital records and GPs should be consulted to establish whether or not a participant has died and if any change of address has occurred, such as to a care home.

• The use of postal questionnaires is adequate for many self-report health outcomes. It is an inexpensive method and avoids researcher bias. Questionnaires can be sent with a reply-paid envelope and covering letter. Participants who do not respond within 2 weeks can be reminded by telephone and those who have not completed all parts of the questionnaire or who give ambiguous answers can be asked to clarify these by telephone. The telephonist can be kept blind to allocation in controlled studies. Non-responding participants can be offered a repeat mailing or a home visit from a researcher (who can also be blinded to allocation in controlled studies).

• For very frail or cognitively impaired participants, with whom it was not possible to carry out face-to-face interviews, flexibility and initiative was required, such as the collection of as many data as possible from a family member by telephone.

• A small number of participants will not return a further mailed questionnaire. Some may indefinitely put off visits from researchers or be absent or decline to answer when visited. This should be interpreted as meaning that some do not wish to continue with the study but do not wish to be impolite by openly withdrawing. Some, despite verbal and written assurances, may be unsure about their rights to withdraw or if doing so will have adverse consequences for them. Thus, any further attempts to gather data from them goes beyond what is ethically acceptable and could be seen as harassment. It is important for researchers to discuss individual cases with senior researchers and to not be put under pressure to obtain data ‘at all costs’ or go beyond the process for follow-up that has been approved by the research ethics committee. Some degree of missing data has to be expected in ethically conducted studies of frail older people. The exact process of data collection in those who do not respond but who have not formally withdrawn needs to be clear in the research ethics committee application.

Table 1 shows the number of participants for whom data were missing and the reasons for this in all of the studies referred to in this report.

Specific health items

The most overwhelming measurement issues related to the effects of cognitive impairment, and to a lesser degree physical health, on outcome ascertainment:

• In the care home outcome study GHQ-12 data were often missing. Only the minority with a high MMSE score completed this outcome. It is not a suitable measure of mood or well-being for use in care home residents.

• In the TEAM study a decision was made that participants with a MMSE score of ≤ 10 were not required to complete the DEMQoL as answers were deemed to be not meaningful; this is specified in the instructions for use, although DEMQol proxy can be used.

• Care homes often did not record weight and patients who were immobile were hard to weigh, making body mass index and other physical measures difficult to complete. Many participants did not know if they had lost weight.

• Many participants did not know their height, could not stand to be measured and found it hard to comply with demi-span measures.

• Simple physical measures such as arm or leg circumference or grip strength were hard to measure because of injuries, dressings, drips, oedema and amputations.

• Walking tests even in mobile participants could be limited by the environment.

• Participants were often unclear about the paid carers and other services that they received.

Even using informants could be difficult. Many informants did not have accurate knowledge of personal details such as items in the Barthel index scale (measuring independence in personal ADL) or accurate knowledge of whether or not the participant experienced the symptoms recorded in the NPI (measuring behavioural disturbance).

Days at home: an outcome measure in studies of specialist services providing care for older people67

A common problem in designing research studies into services for older people, particularly those who are likely to be in the last months or years of their life, is to find an outcome measure that adequately reflects their problems and the objectives of care and is sensitive to interventions. Surveys of older people suggest that remaining at home is highly valued and government policy promotes this (improvement area 3.6 of the NHS Outcomes Framework,98 ‘Helping older people to recover their independence after illness of injury’, is assessed by measuring the proportion of older people still at home 91 days after discharge into rehabilitation). ‘Days at home’ was investigated as a potential outcome measure for studies of older people using hospital services. A similar concept of ‘home time’ has recently been explored as an outcome measure for stroke trials. 103,104

Days at home was created as a continuous variable by counting the number of days not dead, in an institution (long term or respite), in hospital or in other health-care facilities. This could also be summarised as the proportion of days spent at home in the follow-up period. It was hypothesised that such a variable would be more sensitive to change than a categorical outcome and would enable more powerful statistical analyses to be performed.

The limitations of using days at home as an outcome measure are that hospital or long-term care is not necessarily a bad outcome and, if widely adopted as an outcome measure, health services could be encouraged to target the measure at the expense of patients’ quality of life. A strength of this measure is that data collection does not require direct patient contact, therefore reducing the burden on participants and their carers.

For the AMOS study, the Better Mental Health cohort studies, the AMIGOS study and the TEAM study, data on date of death and overnight stays in hospital were easily extracted from routine hospital databases. However, information on care home admissions is not collected on one central database and required intensive follow-up by researchers to ascertain and verify dates provided by participants, carers, care homes and general practices. In the AMIGOS study this meant that stays in respite care could not be included in the days at home calculation. In future studies careful planning will be required to ensure that data can be collected for all days not spent at home in the study follow-up period, especially for those participants who do not wish to complete assessments at the end of the study.

The number of days at home was calculated for 598 out of 600 participants in the TEAM study and for 417 out of 433 participants in the AMIGOS study (using the revised definition). This included 77 and 78 participants in the two studies, respectively, who did not complete outcome assessments at follow-up. The mean and median numbers of days at home were greater for participants in the AMIGOS study (mean 80 days, median 90 days) than for participants in the TEAM study (mean 43 days, median 48 days), suggesting that days at home is a valuable measure to discriminate between sicker older patients who are admitted to hospital and a ‘healthier’ group who undergo a short assessment on an AMU and then return home. In both studies associations in the expected directions were observed with baseline variables, which would be considered a priori to be associated with adverse outcomes. This included ADL and cognitive impairment scores (measured in both studies), delirium and behavioural and psychological symptom scores (measured in the TEAM study), and comorbidity and ISAR scores (measured in the AMIGOS study). There was evidence in both studies that the number of days at home was associated with change in ADL score (calculated for the participants not lost to follow-up). Days at home values were smaller and more variable for groups experiencing a decline in ADL score at 90 days’ follow-up than in those with no decline. There remained a large amount of overlap in days at home values between participants with no decline in ADL score and participants with decline in ADL score. The observed associations between number of days at home and change in cognitive impairment and behavioural and psychological symptom scores (in the TEAM study) and EQ-5D health status and GHQ-12 scores (in the AMIGOS study) were much weaker.

Within each study, however, there was a large proportion of participants with the same value for number of days at home: 0 days in the TEAM study for 28% of participants who were not able to return to their usual place of residence after the initial admission and 90 days in the AMIGOS study for 55% of participants who were not readmitted during the follow-up period. This, combined with the left-skewed distribution in both studies, meant that two-part modelling and non-parametric techniques had to be used to evaluate the effects of the interventions, including using bootstrapping to calculate 95% CIs for the intervention effect. No significant differences were found in either study between the intervention group and the standard care group. This does not rule out the possibility that number of days at home can detect a difference as this finding was consistent with there being no significant differences between groups when component outcomes, such as mortality and care home placement, were analysed individually. In the TEAM study there were two distinct parts observed for the days at home distribution and so an overall summary measure (based on the mean or median) may not be that meaningful. For this study, summarising the effect of the intervention in terms of the proportion of patients able to go home and the amount of time spent at home for those returning home may be more relevant to patients. There may still be difficulties in interpreting an intervention effect that is summarised in two parts, especially if there is evidence of a difference between the two groups in the proportion of patients able to go home.

The experience of using number of days at home in the Medical Crises in Older People studies of older people using acute hospital services suggests that it may be a useful overall summary measure to compare different populations but may have little added value over single outcomes when used as an outcome measure in RCTs.

Introduction

Economic evaluations require patient-level resource use information to estimate patient costs. Several methods are available, including questionnaires, diaries and electronic record searches. The National Programme for IT105 prompted UK health and social care services to record patient-level resource use using Electronic Administration Record systems. Because of lack of interoperability between health and social care sectors, and the requirement for multiple service approvals, retrieving Electronic Administration Record information in the UK is labour intensive. However, this method may provide better information than self-report methods, such as the Client Service Receipt Inventory (CSRI), particularly in cognitively impaired people. For this reason, extraction of data from Electronic Administration Records was the primary method for obtaining resource use data across this programme. The CSRI was used as a back-up method and so comparison of the two methods would be possible. Lessons learnt can be divided into obtaining and coding data from multiple sources.

Obtaining data

One of the biggest challenges was gaining access to the service data sets. Consent to access health and social care records was obtained from participants as part of the overall consent process. Gaining access to services subsequent to patient consent, however, was a significant task and highly time-consuming. Anonymity and the protection of personal data are important; the process for anonymising data is not difficult and can be carried out by the service before the data are sent to a researcher if an encrypted patient list and patient identification codes are sent securely to the service for consenting patients. However, it proved particularly difficult to get general practices to agree to share their data, despite researchers being experienced in this area.

The delays between patient consent being gained and obtaining data for analysis varied between 3 and 6 months. This is a significant consideration for project planning. Once access was gained, the data extraction process itself was generally not time-consuming for most services.

Once access was obtained, data were extracted from a wide range of Electronic Administration Record systems. With the exception of primary care, the researchers did not need to have any particular prior knowledge about the electronic system itself, just the parameters available within the system that described resource use at each service. Parameter lists that described resource use at each service and which were important for the research objectives of the Medical Crises in Older People programme were created through discussions with the research team and an analyst at each service. Most, but not all, services employed analysts who knew how to extract the data required. At least one researcher required extensive knowledge of primary care systems to train a larger research team. After training and distribution to the research team of an extraction protocol for primary care systems, most researchers became familiar with the extraction process. By the end of the extraction period all researchers could visit a practice and extract the data by themselves.

Data coding

The data sets obtained from the services were of different designs and used a range of different coding systems. To develop a unified data set it was necessary to code and process the data fairly extensively.

Secondary care data were processed through the Healthcare Resource Group (HRG) toolkit. 106 The data that we obtained are similar to the parameters required to assign the most appropriate HRG4 code to patient-level resource use information. As suggested by Geue107 in ‘Spoilt for choice’, we assigned unit costs to secondary care inpatient and day-case episodes using the HRG4 casemix costing method. HRG4 codes are available in the NHS reference costs46,108 and are suggested by the NHS Health & Social Care Information Centre [www.hscic.gov.uk/ (last accessed 7 April 2015)], Geue107 and ourselves as the gold standard method for attaching unit costs to inpatient and day-case care. We did not agree with the costing algorithm suggested by Geue,107 which used a ‘dominant’ HRG4 code to assign a unit cost (believing that this underestimated the cost of care of frail older people with multiple episodes per spell), and so designed our own costing algorithm.

Primary care data processing was carried out using our specially designed Visual Basic for Applications scripts [the Visual Basic for Applications scripts were originally designed within Microsoft Excel 2003 but were updated to work in Microsoft Excel 2007 and Microsoft Excel 2010 (Microsoft Corporation, Redmond, WA, USA)]. Although not perfect, the scripts provide a simple, efficient method for converting the data from Microsoft Word (Microsoft Corporation, Redmond, WA, USA) format into parameter categories within Microsoft Excel and then into Stata version 11 (StataCorp LP, College Station, TX, USA) for analysis. When possible, we replicated the costing methods described in Curtis47,109 [Personal Social Services Research Unit (PSSRU) costs]. Various extrapolations for time assumptions were used from the PSSRU costs because the variety of different tasks and professionals associated with patient care that were elicited from the primary care electronic system was greater than that listed by Curtis. 47,109 Many of these extra professionals and tasks were not included in the final analysis because this extra information was available in only one of the primary care systems (SystmOne) from which resource use data were extracted. For those professionals and tasks that were kept in the final analysis and for which a time assumption could not be extrapolated from the PSSRU costs, we consulted the wider literature or used expert opinion. In some cases the costs may still be an underestimate of the time taken to perform some tasks but are probably a closer approximation of the true cost of primary care than if only accounting for consultations and ignoring all other patient-centred tasks.

Social care data were plentiful but not always in a standardised format. We processed data on contacts, assessments and care plans only as these are the categories for which we could make assumptions to add unit costs. A number of assumptions were made for the time taken and for who performed a contact or assessment. For social care, costs for care plans were sourced from Curtis. 47,109 For contacts and assessments, the time spent with the professional who performed the contact and assessment was used as the main aspect of the cost for this task. If a unit cost for contact and assessment could not be sourced from Curtis,47,109 the same method as that used to extrapolate a time assumption from the PSSRU costs was used, as has been described for extrapolating a time assumption for primary care resource use.

All other data from the other services were relatively easy to process as the data sets were already in Microsoft Excel format and categorised and assigning a unit cost was relatively simple.

Discussion

In summary, Electronic Administration Records provided more complete patient costs. Although the CSRI can be modified and is simple to administer, poor recall and inadequate detail about the nature of contacts prevented accurate patient-level cost estimation. Using Electronic Administration Records reduces the burden on research participants, which is especially important for frail and cognitively impaired people. Gaining access to Electronic Administration Record systems is labour intensive; however, once access has been gained, these systems provide useful information that can be used for health-care research. If interoperability between services’ electronic systems improves, this will also improve the ability to access this data for research and clinical purposes. 107

The medical and mental health unit patient and public involvement group

The development process was led by a researcher paid by the Medical Crises in Older People grant. An initial information search of the internet and the INVOLVE (a NHS organisation set up to promote the involvement of patients and public in research) and NIHR websites provided information on what PPI was but not how to go about forming such a group. A number of meetings were held with staff with general and research PPI responsibility at the local NHS trust. The trust already had a database of experienced PPI members who had been involved with other PPI groups within the hospital and potential members of the new PPI group were sought from this database. People in this database had already gone through a process of becoming aware of the roles and responsibilities of PPI in research.

Five people attended the first meeting: two of these were sisters and previous carers for their mother who had dementia; the other three had previously worked in mental health services. This meeting involved spending time getting to know one another, learning what experiences or knowledge they had of dementia and of PPI and explaining what they wanted from the group. This programme and further studies were discussed. The PPI lead was overwhelmed by the sense of enthusiasm, friendliness and genuine interest in the research. Members discussed and agreed the times of future meetings.

Great effort was made to remain in regular contact with the members over the following month, as this would be an important time in terms of retaining or losing these members. During this month, other people showed an interest in joining the group, either people known to the original five PPI members or people volunteering from the MMHU under study in the TEAM trial.

In the third month posters and leaflets were designed to hand out around hospital and community buildings promoting the group. These were reviewed by the PPI members before distribution. We also started to talk to local community groups that were supported by the local mental health NHS trust [such as the Black and Minority Ethnic Dementia Community of Interest] in an attempt to gain new members and to raise awareness. Contacts and suggestions were followed up to find suitable groups who might be interested in becoming involved.

Within a few months the group had 14 interested members; not everyone had attended a meeting but interested members had stayed in touch by e-mail and wanted to be part of the group, helping where they could. Members discussed how the they wanted the group to develop and be run – we wanted them to be in a position to offer value to research rather than demanding tasks from them. The group agreed on most things but a couple of questions threw up discussion, namely how wide the scope of the research should be (dementia-only studies or open the research up to all older people) and whether the focus should be on hospital research or should be expanded to the community. These discussions were constructive even though they could have potentially led to a loss of focus.

An example of the approaches needed to encourage involvement is illustrated here. One lady who had looked after her mother who had Alzheimer’s disease until her death was interested in joining the group. However, she lived in a neighbouring county, too far away for her to attend the monthly meetings. The PPI lead e-mailed her regularly on an individual basis as well as sending her the group e-mails and also telephoned her every so often. Including this lady in the group has been of benefit. She always responded to tasks e-mailed out to the group and her input has been just as valuable as the input of those who attended meetings. E-mailing and telephoning her has not really been much of an effort and sending material by post is required for many members who do not use e-mail. However, this is resource intensive.

The initial members of the group were predominantly white, British, middle class and retired. One Pakistani health-care worker joined the group later. He believed that finding black and minority ethnic (BME) members for such a group would be difficult and that BME carers would probably feel anxious about being in the group. He felt that there may be barriers between white and BME races and at the time of writing this issue was not resolved.

The PPI lead attended specific PPI in research conferences hosted by INVOLVE, finding a large and growing community committed to this research contribution. An early priority was to provide general PPI training to group members to enable them to better understand research processes and how things work. Currently, training for PPI personnel tends to be locally arranged and the content of such training is not clearly defined. INVOLVE did not provide training for the group and it has been difficult to find training. This remains a challenge.

After 6 months the group had 20 members. Meetings were usually attended by 10–12 people and tended to focus on the first part of the research cycle: defining problems and priorities from the carer perspective.

Regarding outputs so far, the group has discussed, supported and submitted two NIHR Research for Patient Benefit programme research applications. It has also given advice on two other applications and on PhD research. Group members attended a focus group to plan a research study into continence issues in people with dementia, the outcome of which changed the direction of the research. Interestingly, the findings were so useful that they have been used by the University of Nottingham School of Nursing to rewrite the continence modules for their Bachelor of Science and Master of Science degrees. This illustrates that PPI for research purposes can have unexpected effects beyond the original remit. Group members have also been involved in the analysis phase of current Medical Crises in Older People programme work, assisting with the analysis of observations of staff–patient interactions in hospital. The group has helped organise two dissemination conferences, one for professionals and the other for the public, and has contributed by speaking at them. One member has also had an article relating her experience of her father’s admission to the MMHU published in the BMJ. 110

These outputs illustrate what a viable PPI group can achieve. It should be noted that this required considerable resource input. A lot of time was needed to set up this group and the month-to-month running of the group was also time-consuming. Members of this group, and other PPI groups, stressed the importance of remaining in touch with the research team between meetings rather than just being picked up as and when needed. If ongoing interest is not shown through regular contact, members might not get the acknowledgement that they want and might leave the group. This observation also indicates the sorts of skills required to run a PPI group. Organising the monthly meetings also took time because of having to liaise with the researchers who wanted to present at the meetings and then feeding back to them afterwards. Administrative tasks were also important such as typing up agendas and the minutes of meeting and organising refreshments and lunches.

Finally, this group may still be in the ‘honeymoon’ period and issues about replacing and sustaining the group have not yet been addressed. This is largely related to securing funding but also finding a suitable person to lead the group.

The Leicester patient and public involvement group

In Leicester, the PPI group was formed in 2011, 3 years into the Medical Crises in Older People programme, and was supported by the local NHS research support system (the Comprehensive Local Research Network). Possible members were targeted to represent a range of professional, clinical and lay members from as broad a spectrum as possible: the NHS, academia, patients, social care, the local authority and charities. Local community groups were also approached, including the Leicestershire and Rutland Minority Ethnic Forum. A more formal approach was taken with defined terms of reference, including aims and purpose, roles and responsibilities of the members and methods of communication.

The group consisted initially of 19 members with typically 7–10 members attending each meeting. The lay members were always well represented. Pre-meeting reading was forwarded to all group members with active feedback from those who could not attend. These comments were fed into the meeting and reflected in the minutes, which were then circulated for approval and any study questions. The group usually discussed two proposals at each meeting. Members of the group are co-opted onto project management boards for individual studies that are brought to the group. Professional group members often allow the use of their resources to advertise studies and arrange events such as focus groups for individual researchers who come to the group looking for support. In conjunction with the NIHR Age and Ageing Specialty Group a video was produced featuring the AMIGOS study to encourage older people to participate in research. This can be viewed at www.crncc.nihr.ac.uk/about_us/ccrn/specialty/age_and_ageing (accessed 27 November 2014).

Legal framework for the inclusion of people lacking mental capacity in research

The ethical principles of informed consent date back to the Nuremberg Code (1948) and state that the voluntary consent of subjects is essential in medical research and this requirement has been reiterated through the Helsinki Declarations (1964, 2008). However, it has also been argued that using research findings derived from an unrepresentative sample, for example in the case of the exclusion of large numbers of participants who are unable to provide consent, is itself unethical: some findings may not be properly extrapolated from relatively well, non-cognitively impaired people to those with cognitive impairment and multiple morbidities; there are well-recognised broader benefits of being involved in research that should not be denied to people with cognitive impairment; and people with cognitive impairment who may have wished to have undertaken an altruistic act such as participate in research should not have that wish denied. 111 There has been little research conducted on the implications of excluding patients who lack capacity from research studies but a clinical example of the consequence of not including people with dementia in hypertension trials is that there is genuine uncertainty about the risks and benefits in this group.

The Mental Capacity Act 2005112 came into force in England and Wales in 2007 and provided a legal framework for the inclusion in research of participants who lack capacity. The act stipulates that, whenever possible, consent should be obtained from the participant but that when this is not possible at the time a relative or friend should be consulted. This consultation is aimed at ascertaining the wishes and feelings of the person lacking capacity about taking part in the research. There is no requirement under the Act for this consultation to take place face-to-face and a telephone consultation, as often necessitated in an acute situation, is deemed appropriate. 113 In situations in which a friend or relative is not available, the Act makes provision for the researcher to seek permission from a registered practitioner who is not involved in the research project. These consultees can only guess as to a patient’s wishes in respect of research. Most consultees would reason that an individual would wish to improve medical treatment and provide consent, but this may not actually reflect the prior wishes of the patient. 113

The Mental Capacity Act112 stipulates that studies including patients lacking capacity must be approved by a research ethics committee. There are five basic issues for research ethics committees to consider: ensuring that there is an adequate research design, an acceptable participant selection method, a favourable risk–benefit ratio, documented free and informed consent and compensation for research-related injuries. Committees will approve research projects involving participants who lack mental capacity only if there are reasonable grounds for believing that ‘research of comparable effectiveness’ cannot be carried out if confined to participants who have the capacity to provide consent:114 there has to be a good reason for including people in studies who lack the capacity to give informed consent. Examples would be specific treatments for severe dementia, which obviously should not be tested on people without dementia or on those with only mild dementia.

Practical issues

The Medical Crises in Older People studies were undertaken within the legal framework described above. Several practical ethical issues made them challenging.

There are issues regarding the separation of research from clinical practice: research participants must be aware of, and not be misled about, the difference between research processes and clinical processes. The issue of who first makes contact with potential participants can be difficult. The guidance states that the clinical staff responsible for the patients and not the researchers should be the ones who make first contact with the patients about the research. This means that clinicians are aware of and complicit in the research (assuring safety) and avoids information governance problems (researchers having access to information to see whether or not a patient is suitable before having proper permission to do so). However, busy staff may not have the time to identify suitable patients, leading to under-recruitment. The more that clinical staff are involved, the more there is a risk that patients will be inclined to see participation in research as part of their care, or to feel under an obligation to participate. This may similarly occur in care homes when care home staff make the first approach to residents.

Information governance brings other issues. Research staff have no right to see or use data without the appropriate permission to do so and this includes routine data from those who decline to enter a study that might help characterise those who did not enter and hence estimate any recruitment bias. In the TEAM study, patients were randomly allocated to the MMHU or usual wards under clinical governance procedures and were recruited to the study afterwards. This meant that some patients were randomised who did not become participants, which could have introduced bias. This bias could be examined by looking at routinely collected data (e.g. length of stay, death rates) from all patients randomised, not only those recruited to the study, as might be carried out as part of an audit study. In fact, the officer within the hospital responsible for adherence to legislation pertaining to information governance permitted this under appropriate clinical governance procedures.

In the AMU studies there was a short time during which to recruit participants and assess their capacity. Many potential participants did not have their spectacles or hearing aids and they had little time and a reduced ability to assimilate study information. Research ethics committees tend to require that many points are included in patient information sheets but this can make them difficult to comprehend quickly. However, we were pleased to note that the research ethics committee reviewing the TEAM study gave permission to use single-sided patient information sheets for exactly this reason, with supplementary information available if required. Many patients were anxious and in a state of mental shock such that they felt unable to make decisions without family advice. Despite explicit reassurance, some feared that research involvement might affect their state benefits, that the research was commercially driven or that their details would be made public, and it is therefore important to recognise quite how scared and vulnerable these patients are and feel. Together with the embedding of research staff in the AMU, these issues could have affected recruitment and may have predisposed to a degree of coercion of potential participants; a proportion of potential participants with mental capacity did not give consent and a large proportion of included participants subsequently implicitly withdrew. Recruitment after discharge from the AMU was not appropriate as the AMIGOS study required intervention to begin on the AMU.

Potential participants among care home residents were vulnerable in other ways that could have affected participation: they suffered from fatigue, pain, cognitive impairment, limited functional status, sensory and speech deficiencies, depression and dependency. However, these could be overcome with time and patience.

The use of family or carer consultees has challenges in practice. In the Better Mental Health cohort study 10% of potential participants without capacity had no family member or friend who could act as a consultee and a further 12% of potential participants without capacity had family members or friends who could not be contacted, partly because 6% of potential participants were discharged from hospital quickly, before an appointment with the researcher could be arranged, partly because some families visit in the evening or at weekends only and partly because some family members do not visit at all because of commitments of children or other relatives. Some carers were so distressed by their family member’s admission that they did not want to spend time being interviewed by a researcher. Family consultees may hesitate to agree consent involving their elderly relative in the belief that it is not in their relative’s best interests; on assessing the benefits and risks they may not believe that the benefits of participating will be sufficient because of the imminent end of the elder’s life or because they may not want their relative to be bothered or undergo undue stress.

For the AMOS study, the research ethics committee would not give approval to allow the use of registered medical practitioners as consultees on the basis that it assumed that few people without mental capacity would be discharged without the involvement of a potential consultee. Such an assumption was not unreasonable given that the clinical management of patients is also subject to the same legal framework: decisions on patients lacking mental capacity should be undertaken in their best interests, which involves consultation with members of the family or others who know the patient. In practice, however, many people who the research team assessed as lacking mental capacity to participate in research were discharged from the AMU without the opportunity to identify a consultee. This resulted in recruitment of a lower-risk group than intended.

Findings from the AMOS study were used to advise the research ethics committee for the AMIGOS study that patients lacking mental capacity and without a consultee should be included after consultation with the responsible registered medical practitioner on the AMU, and the ethics committee approved this. In the TEAM study the research ethics committee agreed that professional consultee agreement from the nurse in charge of the ward was acceptable if potential participants had lacked mental capacity to give consent and had no available carers.

A common issue in discussion with research staff is about the disclosure of information given by participants and a duty of care. A non-clinical researcher might respond to a participant describing a physical or mental problem with ‘Oh dear’ or might respond to a request for help with ‘You had better see your GP’, but may wonder if he or she should actively pass the information on to the GP, for example if the patient is housebound or otherwise unable to do so themselves. The sense of a duty of care may be even higher for clinical researchers in such circumstances. Researchers undertaking observations may observe aspects of care that are harmful or substandard, such as observing untended patients at risk of falling. In the Medical Crises in Older People studies, specific arrangements were in place to enable discussion of such concerns and for specific cases to be dealt with by the workstream lead or overall principal investigator. A general rule is that researchers should feel empowered to do what they feel is right under such circumstances, even if it means a non-participant observer becomes a participant observer.

A specific related requirement, stipulated by research ethics committees, was that participants in interview studies should be advised prior to giving consent that, although their replies would be confidential, if they disclosed evidence of harm or wrong-doing the researcher was bound to disclose this information to other relevant people. It is impossible to know if this led to significant self-censorship, but research staff tended to find this clarification helpful.

General considerations

Research of any sort should, hopefully, eventually lead to ‘impact’ or some sort of consequence that makes the research worthwhile. Impacts can be in health, economic, social, cultural, political or other dimensions. They will usually be indirect: research findings are seen in the light of existing and other emerging findings and it is often the sum of all of these findings that brings about change. Health research, especially when publicly funded, should be expected to lead to patient benefits, but there are other possible impacts of health research such as developing skills in the workforce, stimulating further innovation or supporting industry or improving nations and their citizens in more subtle but still valued ways.

Researchers and those who fund research are concerned that the ‘investment’ in research should ideally reap returns that are obtained as soon as possible and which are as large as possible. There is increasing recognition that much research does not get rapidly taken up and used. This is often referred to as the ‘know–do gap’115 and in biomedical circles is also part of what is known as the ‘second translation gap’. 116 There is a focus on closing the know–do gap by explicit and active means rather than leaving it to chance. The assessment process used to evaluate British universities in 2014, and hence determine their future funding [the Research Excellence Framework – see www.ref.ac.uk/ (accessed 9 February 2015)], included an assessment of research impact, which is intended to be an incentive in this direction.

This applied research programme has generated much new knowledge as summarised in earlier chapters. The purpose of this section is to outline the research team’s understanding of the research impact and to outline the steps taken to date and future plans that will optimise the impact of the Medical Crises in Older People programme.

The nature of impact

A theoretical analysis of impact is beyond the scope of this chapter but one commonly cited classification, which shows the complexity of the nature of impact, is as follows:117,118

1. conceptual impact – opinions, attitudes and knowledge (hearts and minds) are changed as a result of the research

2. instrumental impact – tangible changes occur as a result of the research, for example health outcomes, economic, social, cultural, policy, workforce

3. process impact – conducting the research leads to distinct changes in services or practice

4. symbolic impact – the research evidence is used as a political mechanism to change services or practice.

Demonstrating the impact of research is difficult. It may be hard to show attribution: a single study may provide only part of the information base that leads to a change or impact. There may be a considerable delay between the generation of a piece of new knowledge and its ultimate impact. Seventeen years has been suggested as the average time lag between beginning research to reaching clinical practice, but measuring and defining these gaps is problematic and varies across scientific disciplines. The impact arising from negative intervention studies or simple descriptive studies is harder to characterise than the impact of positive intervention studies that have been implemented. Negative studies can prevent ineffective or unsafe interventions from being implemented, but the effects of this are impossible to measure. Similarly, descriptive studies may improve understanding, but this too is hard to measure. For these reasons, the concept of the ‘pathway to impact’ is used: ultimate and total impacts of research may be impossible to ascertain, but it is possible to enumerate the tangible steps that are expected to lead to impact.

Pathways to impact

Methods and frameworks exist for operationalising the pathway to impact and the ultimate health research impact119–121 and these enable researchers to track the steps towards change and provide evidence of attribution. Understanding these processes allows researchers and those involved in the translation and implementation processes to attempt to speed up and streamline the process of closing the know–do gap. The NIHR, when it funded this research in 2008, defined impact as ‘demonstrable change in NHS practice, service delivery or policy. Effective translation of research findings into improved outcomes for patient and carer benefit’. This was a narrow definition, but the NIHR’s tool to assess steps along the pathway to impact (described later in this paragraph) implied that a more sophisticated definition is understood in practice. The NIHR in 2015 stated that ‘By “impact”, we mean the contribution to benefits to society resulting from the research we fund, including patients, populations, the NHS, health services, the economy and academia’ [see www.nets.nihr.ac.uk/impact (accessed 11 March 2015)]. The Research Excellence Framework 2014 (see www.ref.ac.uk/) used to assess the impact of the work of British universities describes ‘an effect on or changes to the activity, attitude, awareness, behaviour, capacity, opportunity, performance, policy, practice, process or understanding of an audience, beneficiary, community constituency, organisation or individual’. The criteria used to assess impact are ‘reach’ and ‘significance’. Ultimately, the health impacts of research are expected to stretch beyond the NIHR’s 2008 notion of demonstrable ‘patient and carer benefit’ to include increased research capacity; assistance to implementation; contributions to policy or guidance; changes to service delivery; influences on commissioning; stimulation of innovation; benefit to the economy, society, culture or public policy, health, the environment or quality of life; reduced risk or harm; and improved public understanding or behaviour. Capturing evidence of impact is complex and requires systematic identification and subsequent gathering and recording of the elements leading towards it. As part of the NIHR Award Assessment Tool the NIHR developed a checklist for monitoring auditable outputs and events that are assumed to represent stages along the pathway to impact (note that since 2014 the NIHR has used an external system to replace the NIHR Award Assessment Tool). The domains of this tool were:

• further research funding

• research collaborations both in the non-profit and the industry sectors

• research training (undergraduate, postgraduate and professional)

• academic promotion or establishment

• generation of research resources (databases, tools, measures, methods, models, etc.)

• academic publications and presentations

• non-academic publications, presentations and other outputs (e.g. media)

• citation in guidelines or reviews

• research team members contributing to guidelines and reviews

• use in teaching or training

• interventions or products (including intellectual property)

• other.

Knowledge translation and implementation are important elements of the processes by which applied research findings lead to impact. Research organisations with a responsibility for implementation research have examined this area and would include the development of implementation networks and the extent of their reach as an important domain beyond these NIHR domains.

Using this approach, the following paragraphs summarise and reflect on progress along the pathway to impact of the Medical Crises in Older People programme, recognising that impact will continue beyond the lifetime of the research programme.

Further research funding

The Research Excellence Framework explicitly excludes academic impact from its assessment of impact but clearly it was an intention of the Medical Crises in Older People programme that a viable critical research mass was developed, and the award of further research awards is evidence of this. Six further research awards were obtained during the life of the programme, three for a further programme looking at the prevention of falls in people with dementia, one for the evaluation of care home services, one to augment the Medical Crises in Older People programme itself and one to examine the merits of antihypertensive medication in patients with dementia.

Research collaborations

Research collaborations improve research opportunity, are associated with better, less parochial and hence more generalisable research and potentially have greater reach. Significant collaborations arising from the Medical Crises in Older People programme included links to colleagues studying care home medicine in the Netherlands and dementia care in Canada. The strength of the research and the critical mass of researchers was a major factor in the local research implementation organisation [the Collaboration for Leadership in Applied Health Research and Care – Nottinghamshire, Derbyshire and Leicestershire (CLAHRC-NDL); see www.clahrc-ndl.nihr.ac.uk/clahrc-ndl-nihr/research/older-peoples-health-and-wellbeing/index.aspx (accessed 9 February 2015)] adopting an Older People’s Health and Wellbeing theme, with the Medical Crises in Older People programme principal investigator as its lead. This has led to important opportunities to enhance impact.

Research training

Eleven postgraduate students were supported by work associated with the Medical Crises in Older People programme, with many co-supervised by academics outside the host academic division. This has the potential effect of developing a broad research workforce that is skilled and informed about ageing and older people. Numerous undergraduate medical and nursing students also took roles in different aspects of the work, such as literature reviewing, data checking and analysis. This has the potential to affect their future career choices while providing real-life research experience. Nursing staff were involved in data collection and analysis, bridging the gap between academia and clinical practice and promoting theory into practice and evidence-based care. Members of the Nottingham dementia PPI group were trained to support and review new grant proposals, critique research documents, analyse qualitative data, publish documents and present at conferences. The impact of these training and educational opportunities will extend far beyond the lifetime of the Medical Crises in Older People programme.

Academic promotion and establishment

Several formal appointments and promotions of staff occurred during the lifetime of this award, using evidence from participation in this programme. An important step was significant recognition by the associated NHS trust that research into older people was a core part of its research and innovation strategy, with the commitment of significant core funding. This funding was matched by a local wealthy philanthropist to support a clinical/academic post for a Medical Crises in Older People programme researcher to continue knowledge translation and implementation of learning developed during, and beyond, the programme. This will include developing ‘advanced nurse practitioners’ for frail older people. The potential impact of this is considerable; the evidence and good practice from the Medical Crises in Older People programme, developed through this additional funding, could be applied in other national and international health-care settings.

Generation of research resources

Perhaps the main research resource produced by the programme was the MMHU. It was not only an intervention to be tested but can also continue to act as a research and innovation hub, in the same way that coronary care units in the last four decades not only provided acute cardiac care but also facilitated many of the large cardiac studies that transformed cardiac care over this period.

Academic publications and presentations

Publications arising from the Medical Crises in Older People programme are listed on the Medical Crises in Older People Discussion Paper Series website (see www.nottingham.ac.uk/mcop). All publications are open access to increase their potential impact. Numerous presentations have been made across the UK and internationally, both invited and following submission to research conferences. Such presentations are usually limited to impact within the academic sphere. Those with a published abstract are also shown on the Medical Crises in Older People Discussion Paper Series website. An important research output and publication worth noting is one authored by a member of the Nottingham PPI group published in the BMJ. 110 This is a moving personal narrative that describes the carer’s experience of the positive impact of the MMHU: it has both reach and significance.

Non-academic publications, presentations and other outputs

One of the purposes of the Medical Crises in Older People Discussion Paper Series was to disseminate findings that might not be suitable for peer-reviewed research publications but which might be useful for many potential research users (which includes clinicians). An important example is the paper describing the development of the MMHU. Web publishing allows people around the world to freely read about this work.

Other key activities that enable knowledge transfer to audiences who might not read academic journals include presentations and workshops delivered to NHS development organisations and relevant charities, including the Alzheimer’s Society and Age UK.

Another method of knowledge transfer specifically designed to help busy practitioners become aware of research findings is the use of BITEs (Brokering Innovation Through Evidence) [see www.clahrc-ndl.nihr.ac.uk/publications/bites.aspx (accessed 9 February 2015)], which are short, accessible summaries of research developed by CLAHRC-NDL, put into context and circulated directly to relevant practitioners. Several of these arising from the Medical Crises in Older People programme can be seen on the CLAHRC-NDL website.

The use of the media, newspapers and television becomes increasingly possible and important for research that deals with topics of high public relevance. Engagement of the media can enhance public understanding of the science of the research or the general areas that it concerns. Garnering public approval and stimulating interest can enhance the degree to which research findings can influence policy. It can also generate important research consequences such as people wishing to collaborate or participate in the research or study it in more depth. The hospital care of people with dementia became an important priority in the UK mass media during the period of this programme, which afforded many opportunities for radio, newspaper and television engagements. For example, a national newspaper described the MMHU as ‘leading the kindness revolution’ [see www.dailymail.co.uk/health/article-2277169/Inside-hospital-thats-leading-kindness-revolution-Concluding-series-crisis-compassion-nursing.html (accessed 9 February 2015)], reporting examples of good practice and evidence from the Medical Crises in Older People programme, promoting the research to a wide audience and generating much public and political interest. To a lesser extent, the acute care of older people has afforded similar opportunities. Health care in care homes did not have the same opportunities, reflecting lower levels of media interest. The extent to which research findings can drive media priorities is unclear, but it is likely that areas of research with little media interest will find it hard to make use of the media for knowledge transfer.

A final area of knowledge translation and exchange explored in the Medical Crises in Older People programme was the development of a DVD documentary about the MMHU, called ‘Today is Monday’, to illustrate the aspects of care that were tested. A professional film-maker was commissioned to make the documentary and the result was a short high-quality film showing a moving reconstruction of a day in the life of the MMHU. Early feedback from viewers is that its honesty is incredibly powerful, working as a piece of drama as well as being suitable as an educational tool; in fact, the two functions are seen as intertwined. At the time of writing over 500 people had seen this video (and plans are made for many more to do so), the majority of whom would be unlikely to have read our peer-reviewed reports. This is an example of the use of the arts to aid the dissemination, and particularly public understanding, of science. The potential impact of this unique research output is unknown. Thus far, requests have been received to use it as an educational tool for psychiatrists, medics and nurses and to inform policy-makers and commissioners. International interest in this research product extends to Canada, the USA and Australia, providing opportunities for global comparisons and learning. Its potential reach and significance are considerable.

It is not yet clear how important formal academic outputs, non-academic outputs and other activities are to research impact. It seems reasonable to argue that direct impact is unlikely or may be delayed without employing as many different knowledge transfer approaches as possible, but at the same time the ultimate authority of the messages will depend on their scientific rigour, which is tested by being published in peer-reviewed journals. It is not one or the other but both that are necessary.

Guidelines or reviews

The research itself has produced several review articles, as listed on the Medical Crises in Older People website (www.nottingham.ac.uk/mcop), but at the time of writing this report it was too early for research papers from the programme to be highly cited elsewhere.

However, two potentially influential documents were developed during the period of the programme, with significant involvement from Medical Crises in Older People programme personnel who drew on this research. These were the Royal College of Physician’s Silver Book99 on the acute care of older people and the British Geriatrics Society’s Quest for Quality122 on health care for the residents of care homes. This work with the British Geriatrics Society also influenced its Commissioning Guidance for High Quality Health Care for Older Care Home Residents [see www.bgs.org.uk/campaigns/2013commissioning/Commissioning_2013.pdf (accessed 27 November 2014)] and the response from the Royal College of Physicians and the British Geriatrics Society to the Older People’s Commissioner for Wales formal review into the quality of life and care of older people living in residential care in Wales.

Use in teaching or training

At the time of writing the findings of the programme have been used by local educators but there has not been time for this to result in any significant changes to the specific content of undergraduate or postgraduate training. However, emerging collaborations with the East Midlands Local Education and Training Board, whose priority in 2013/14 was frail older people, have provided ongoing knowledge translation opportunities.

Interventions or products (including innovations)

At the time of writing this report, whether the MMHU is seen as a complex intervention that could be replicated or a unit demonstrating a set of principles remains to be seen. Similarly, it is not yet clear whether or not the documentary made on the ward is a product that could be used to guide the development of similar units or the application of the set of principles. However, both have been identified as examples of innovation and dignified compassionate care by clinicians, academics, politicians, journalists and commissioners. Political windows of opportunity exist, allowing initiatives such as these to be shared through considered knowledge translation strategies.

Several examples of innovation seem to have sprung from this work: new ideas for incontinence in people with dementia, enhanced care planning to reduce persistent vocalisation in elements of dementia care,123 interventions to reduce falls in people with dementia, procedures to optimise the use of antihypertensives safely in people with dementia and better models of care for frail older people in emergency settings. 99

Other (including network and reach)

In 2013, England witnessed a major transformation of organisation of the NHS. One development was the creation of Academic Health Sciences Networks. These are regional organisations with several higher-level transformational functions (promoting research, promoting life science industries and developing health informatics, as well as assisting implementation, improving services and developing the workforce) involving health, social care, industry and third-sector providers in the region. As the Medical Crises in Older People programme dealt with high-priority issues through scholarly and evidence-based interventions and evaluations in clinical settings, the principal investigator was invited to lead the frail older people’s clinical theme in the East Midlands Academic Health Sciences Network. 124 Knowledge from the Medical Crises in Older People programme will form the basis of the initial strategic direction of this clinical theme. Developing a network of clinical expertise and evidence-based research through stakeholder engagement and effective knowledge translation and implementation provides ample opportunities to close the know–do gap, building sustainable health services for frail older people across the East Midlands and beyond.

The acute medical unit

The impact section of this report shows that this workstream has demonstrated leadership in the area of transforming the care of older people on AMUs in the UK. The very notion that the AMU is a focal point for the initiation of CGA has been widely accepted. The research work showed that even the best-evidenced simple clinical risk stratification tool, the ISAR tool, is insufficient alone to guide clinical management. The concept of the interface geriatrician has been developed, implemented and tested so that others can develop it further. Such development is required because the addition of an interface geriatrician alone is not sufficient to make a major change to the health and well-being of higher-risk patients discharged from AMUs.

The medical and mental health unit

This workstream showed convincingly the plight and challenges faced by cognitively impaired people in hospital, providing evidence to guide innovation and development elsewhere. It showed that it is possible to develop, in a publicly funded health-care system, a specialist unit for such patients that is demonstrably different from usual care. This will help others to innovate. The unit was tested in a rigorous trial, the first ever in the world. The quality of patients’ experiences was improved and their carers were more satisfied. In other words, their experience of care was more dignified. The effects on ultimate health outcomes were modest but a preliminary cost analysis showed that the costs of developing the ward were offset by subsequent resource use savings across the health and social care system. The economic analysis concluded that the MMHU was likely to be cost-effective. Not only does this provide powerful evidence to counter any lingering sense of hopelessness about the inevitability of loss of dignity for such patients, it also justifies further examination of the use of MMHUs as a means of improving the care of people with cognitive impairment in general hospitals. It also illustrates the general principle that investing in best practice can be affordable and beneficial.

Health care in care homes

This work has been valuable despite the major changes to the research plan. A huge and valued repository of care home literature has been developed, demonstrating convincingly that care home medicine is far from ‘an evidence-free zone’. The complexity of health care in care homes has been illustrated, showing the vital health-care roles that the care home staff play (despite being seen as social care practitioners), and a light has been shone on current primary care arrangements, which a range of national organisations find unacceptable. The findings can help policy-makers, commissioners and practitioners to appreciate the policy, managerial and clinical steps that need to be taken to improve matters, recognising the importance of partnership working.

Programmatic considerations

The programme developed a critical mass of researchers and has increased research capacity in the field of the care of older people. This developed alongside a growing research capability within the NHS, including local trusts and research networks. The crucial mass of researchers held within it a large body of expertise in measurement in frail older people and in the ethical conduct of research in such people. Success in the involvement of patients and the public in research was demonstrated. Early success in the process of closing the know–do gap was demonstrated so that early patient benefits from this work can be achieved.

What is comprehensive geriatric assessment?

A large team of researchers has considered the nature of CGA, from the time when the grant proposal was written through the development phases of the funded work to the evaluation, analysis and reporting of the studies. The team’s understanding of the CGA process deepened over this period. A summary of this follows and in the following section we use this understanding to comment on why neither of our flagship trials, the AMIGOS and TEAM trials, yielded the hoped-for benefits in terms of improved health outcomes at follow-up.

The team came across frequent misunderstandings. CGA was remarkably unknown to people who were not specialists in geriatric medicine (and sometimes to specialists), even those who were unknowingly contributing to it and even though the evidence base had been robustly shown 20 years before the date of this report and it offers a contribution to one of the most pressing health-care problems of the age. The principles of CGA need to be known by all who deal with frail older people and not just specialists in one professional discipline. CGA is a misnomer, which can make it difficult for non-specialists to understand; it is more than assessment, implying a complex process beyond the assessment. CGA can be misunderstood by many as simply another phrase for geriatric medicine or specialist geriatric medical input. Such misunderstandings are understandable in a health service with many unhelpful acronyms and we noted that many people found the acronym an off-putting piece of jargon. Synonyms such as geriatric evaluation and management are likely to offer no advantage.

The assessment part of CGA, to be comprehensive, needs to assess physical and psychiatric conditions (diagnoses such as delirium or dementia and health problems such as falls or immobility), functioning (impairments, activity limitations and participation restrictions), the social environment (such as the social network) and the physical environment (such as the home). Implicitly, the staff carrying out the assessment need to be knowledgeable about these complex areas, which requires good levels of training. Experience in the AMIGOS trial showed that assessments can be comprehensive and brief even if they are not deep. Assessments need to be tailored to the patient group and setting. Patients presenting with falls need assessments targeted towards their likely constellation of problems (such as strength and balance) whereas those with confusion will need assessment focusing more on their cognition, mood and behaviour. Patients in emergency settings require brief high-level assessments whereas those in hospital settings require more detailed assessments. It is inevitable that most instances in which CGA is employed will require a team of different professionals to ensure that all domains have been assessed by a suitably trained person.

Comprehensive geriatric assessment needs to be targeted at those for whom such a complex process is required and justifiable. Frailty describes a state in which patients are vulnerable to deterioration, in which such deterioration can be extreme in the face of challenges that would not trouble a more robust person and in which the recovery after such deterioration can be prolonged or never materialise. CGA should be focused on those who are frail, but it remains difficult to identify these people.

Comprehensive geriatric assessment is more than assessment. Not only is a team usually required to assess the patient, a team is also required to undertake the interventions that arise from that assessment. This means that a care plan should be derived from that assessment, team members should know their responsibilities in terms of delivering the care plan and the effects of doing so need to be monitored by repeated assessments. These steps require some form of case management and team working. The model for CGA working in hospitals is well rehearsed. Medical consultants usually lead defined teams in defined settings with a defined patient case load. Multidisciplinary team meetings enable information flow for the production of coherent care plans, adherence to them and monitoring the response to intervention. The AMIGOS trial experience illustrated the problems of conducting CGA across the secondary–primary care interface and the difficulty of doing so in primary care. In the AMIGOS trial the assessment was carried out by a geriatrician, without direct access to or authority over a multidisciplinary team. As the clinical report about interface geriatrics showed, a far more comprehensive assessment was made after a home visit. However, the care plan advised by the geriatrician was likely to be not as good as one devised with the help of a full multidisciplinary team. Most importantly, there was no clear multidisciplinary team in the community: there were multiple GPs and multiple community teams, all with differing patient caseloads. Few patients used the small intermediate care services and there were no overt mechanisms for multidisciplinary communication. There remains a need for models of community CGA to be demonstrated. The British Geriatrics Society and the Royal College of General Practitioners have proposed such a model,138 and ‘virtual wards’139 represent an attempt to implement such a model, although evidence is still required to determine whether or not these are effective.

The research team offers the following questions to identify whether or not a system is delivering CGA:

• Is the patient frail?

• Does the patient undergo an assessment that includes all domains of a CGA?

• Are the assessments suitable for the type of geriatric condition?

• Are the assessments suitable for the setting?

• Is a care plan made based on the CGA?

• Are there mechanisms to share the production of the care plan and to allocate responsibilities to deliver interventions?

• Is there evidence that these mechanisms are used?

• Are the interventions as planned delivered?

• Is there evidence or later reassessment?

• Is there evidence of repeated review by the team?

• Is there evidence of further cycles of planning and action?

Such a framework may help guide the development and evaluation of CGA interventions in novel settings and quality assure those in more familiar settings.

Understanding the findings of the randomised controlled trials

The positive aspects of the AMIGOS trial were that we were able to develop and test a novel intervention targeted at a high-priority group of patients and, along with the AMOS trial, shine a light on this group of patients. This was appreciated by the AMU staff. The positive aspects of the TEAM trial were that we were able to operationalise and test best practice in dementia care in a general hospital setting, conduct the first ever trial of a unit delivering such care in a general hospital and demonstrate that the quality of experience for patients and carers could be improved by doing so, in a cost-effective manner. However, neither RCT significantly improved health outcomes.

There are several possible explanations. It is possible that the TEAM trial was underpowered for some outcomes. There was an 8% absolute reduction in the percentage of patients who moved into long-term care. This could have occurred by chance but might have been statistically significant if the sample size had been larger and would be clinically valuable. It is possible that poor targeting of the intervention in the AMIGOS trial, because of the difficulty of identifying the frailest patients, could have diluted any benefits of the intervention in that study.

It may be that, in the AMIGOS trial, the right variant of CGA was not delivered. In this report we described the intervention as a specialist geriatric medical intervention and not CGA per se. To have made it CGA there would have to have been far greater integration with community services and possibly greater understanding of specialist geriatric care in the community than at present. Similarly, maintaining any advantages brought about by the MMHU after discharge from hospital would have required community services to continue to support patients and carers using similar expertise. We cannot say whether or not this occurred. System-wide change is a challenge for any trials of CGA-like interventions.

We have increasingly wondered whether or not the patients in the Better Mental Health cohort study and the TEAM study were too close to the end of their lives to benefit from CGA. The benefits of CGA in previous studies are typically restorative ones: increased survival and better health. However, many patients in the Better Mental Health cohort study and the TEAM study were clearly close to death and perhaps beyond the scope for recovery. In such patients a palliative care framework would be more suitable. In such a framework, the quality of care prior to death would be the primary outcomes; restorative outcomes such as survival and disability would be secondary outcomes. Days at home would probably not be chosen as the primary outcome measure in, say, a trial of palliative and terminal care. If a palliative and supportive care approach had been taken, the primary outcomes of the TEAM trial might have been patient experience and carer satisfaction, in which case the result would have been seen as a positive one, not a negative one.

Research-commissioning priorities

Further innovation is required in the use of CGA in AMUs and such innovations need to be evaluated. Examples include the use of frail elderly care units on AMUs and integrated services delivering a whole package rather than the more limited one tested in the AMIGOS study. The health problems experienced by those who attend AMUs may not be solved by interventions in the AMU but by interventions in other parts of the wider system either before or after episodes of medical crisis. System changes might require changes to the training and education of the workforce to ensure that staff in all professions and at all levels of seniority are trained to deal with the frail older people that they meet; the identification of at-risk people in the community and developing community-based services that meet their needs before they present to hospital with a medical crisis; developing effective and efficient models of vertical integration so that patients discharged following a medical crisis receive effective and efficient services to meet their underlying problems, and this will also require the development of models of horizontal integration, taking into account the need for the statutory (health and social care) sectors and the voluntary and private sectors to work together.

Further evaluation of other MMHUs is warranted, to examine the generalisability of the TEAM study, and this might enable meta-analyses to see if any modest but not statistically significant findings of the TEAM study are real and if cost-effectiveness can be seen in other settings and contexts. The development and evaluation of different models for enhancing hospital staffing and training and the hospital environment are warranted. However, a MMHU is far from a panacea. Given the high proportion of cognitively impaired people in general hospitals, MMHUs can only ever be part of a complete service, and so there will always be a need for organisational research into the best configurations of service components in a range of different contexts. Furthermore, clinical problems that remained despite the MMHU, such as patients who repetitively shout out on a ward, require the development of new interventions and their evaluation.

Existing innovations in models of health care for the residents of care homes need immediate evaluation. An example in the UK context would be the use of specialist nurses instead of GPs to provide first-line primary care. Such nurses and other models such as comprehensive teams provide the ability to deliver a form of CGA and are likely to prove preferable over ordinary, usual primary care. Complex research designs are required to evaluate these developments and to guide future innovation. There is scope for a subspecialty of care home health care to develop, and complex research studies will be required to evaluate the effect of such a development.

1. Common methodological issues

Certain aspects of this research will be common to all four strands. It is through these common processes that the secondary objectives (1–5, Full Form Section 8) will be achieved. These include:

• a single research programme board, comprising all grantholders

• a larger reference group, including user groups, Trust personnel, clinical teams & other researchers

• the development of a common user involvement mechanism

• a common core baseline and outcome dataset

• a common intervention ideology (Comprehensive Geriatric Assessment)

• a shared methodology requiring a preparatory stage, a stage in which a cohort or register study is performed from which risk stratification factors can be identified to target interventions upon those most likely to benefit and from which data to design evaluative studies can be drawn, a stage of service optimisation and characterisation of the complex intervention to be delivered, and a final phase in which an exploratory controlled trial is performed

• a co-ordinated dissemination and implementation strategy for rapid patient benefit, including a programme web-site

These common aspects will enable the programme to meet its 5 secondary objectives, and justify the need for long term and large scale programme support.

2. Primary objective 1: Evaluation of Medical and Mental Health Unit (MMHU)

3. Primary objective 2: Evaluation of Acute Medical Unit Comprehensive Geriatric Assessment

4. Primary objective 3: Evaluation of hip fracture Comprehensive Geriatric Assessment

1. Sub-package three is not to be supported

We accept that workstream 3 should be dropped.

In dropping this workstream, this allows us to spread the three RA posts we have requested across the remaining three workstreams. There is clearly a practical advantage of having one RA per project instead of them being split across projects. It also increases the amount of RA time per workstream from 0.75 WTE to 1.0 WTE, and therefore deals with the general issue about having inadequate resources for these ambitious projects.

In dropping this workstream it also releases the consultant research PA sessions that had been intended for the workstream lead. These have been recycled to respond to the further issues raised by the Research Selection Panel, the main costs going to:

• the research costs of care homes in Workstream 4 (see point 9 below)

• the research costs of General Practices involved in Workstream 4 (see point 9 below)

• costs for University staff and facilities involved in this programme

The issue of costs to the University, although not mentioned in the specific list by the Research Selection Panel, was raised by one reviewer, and indeed has been a concern of the research team. Whilst each individual and the University are absolutely enthusiastic about this prestigious award, they are mindful of the need to cover a proportion of the real staff costs the University will incur.

2. Named qualitative researcher

Although the existing study group has considerable expertise in qualitative work of the nature required in this programme (for example the qualitative studies of the dementia services we have studied), we recognise that our most well known work has been our complex RCTs. Nevertheless, the Research Selection Panel’s request gives us the opportunity to invite two of our qualitative research colleagues and collaborators to join. These are Professor Justine Schneider (Professor of Mental Health and Social Care, Institute of Psychiatry, University of Nottingham/Nottinghamshire Mental Healthcare Trust) and Dr Jane Dyas (Primary Care Lead, Trent RDSU). Their CV’s are appended, together with letters confirming their support and anticipated involvement.

Justine Schneider has broad experience of applied research on health and social care for older people. Before taking up a research career she was a social worker with adults for ten years and an ‘informal’ carer for a similar period. She brings a social care perspective to the programme of research. Her research on older and their carers people employs mixed methods, including qualitative and observational approaches: an experimental study of occupational therapy in residential homes for older people; a longitudinal study of informal care for people with dementia; a cross-national study of carer ‘burden’ in dementia; and a large study of quality of life in residential care. She is about to commence an ethnographic study of inpatient care for people with dementia (subject to contract with the SDO). In the present study, she will participate in the reference group, with particular expertise in qualitative methods and social perspectives.

Jane Dyas has broad experience in qualitative and other research particularly in primary care. She is based in the Trent Research Development Unit. She has worked with the applicants in the past and does so now through a study of the use rehabilitation for older people with non-injurious falls who call an ambulance, and she also supports the Designated Research Team which will work with this programme.

3. Qualitative work to be carried out not only during the trials but during the development work

We quite agree that phase I and phase II work is required prior to the formal RCTs. This is planned for the first two years of the programme. We have described this developmental work, and although we have not used the word ‘qualitative’ for this work (whereas we did use this word to describe work running alongside the RCTs), you will see from the description taken from the proposal that it includes both quantitative and qualitative approaches:

1.2 The common approach taken to optimise and characterise the services in each strand will be closely steered by the research programme board to ensure consistency.

The objectives in this phase for each strand are:

• to develop a service that delivers comprehensive geriatric assessment (CGA)

• to demonstrate that a generalisable service is in place that is fit for purpose & evaluation

An iterative, complex human systems approach will be employed to develop services [10.11]. In this approach a service is seen as a complex human system, and defined (in a ‘root definition’) by the interactions between six factors:

• the patients

• the staff

• the interventions made by staff and the changes made to the patients

• the managerial systems that control the services

• the national context that influences services

• the local context that influences these services.

The lead clinician for each strand will develop a root definition for each service, discussing it with the research programme board, and the relevant parts of the programme reference group. Doing so will lead to the identification of obstacles to service implementation including staffing issues, involvement of key personnel, training needs, development of operational protocols, etc. Close working between the research programme board and the NHS Trust is crucial, and has been assured (see Annex 4). The developmental stage will be highly user-sensitive and will make use of public involvement mechanisms in the acute Trust (Nottingham University Hospitals NHS Trust) and a community Trust (Nottinghamshire Country Teaching PCT).

Although the root definition of a service is necessarily largely descriptive, the description will be generalisable by virtue of specific research outputs including:

• clearly defined and justified criteria to select patients

• description of staffing numbers, their expertise, and working patterns

• an observational cohort study and case studies describing evidence of the delivery of comprehensive assessment, appropriate interventions, and outcomes compatible with clinical effectiveness

• user and staff views.

We agree that we should also have included in this brief description that this same process will be used to test the trial processes (recruitment, for example) and choice of outcome measures. We would argue that what we plan is precisely ‘how to pilot the intervention’, although we did not use these exact words. We therefore fully agree that qualitative work should be done prior to the trial stage, and argue that this is what we already intend to do, albeit with some quantitative work which we also think is necessary.

The key object of the developmental work in all three workstreams is, to repeat what we have already stated, to ensure that we have shown that we have developed a service that is fit for evaluation. To re-frame our original protocol, there are three sub-phases that make up the developmental phase.

3.1 The first part of the work towards that will be the register study (MMHU workstream), ISAR evaluation study (AMU CGA workstream) and cohort study (Care Home CGA workstream). These are all required to examine the nature and number of people with the worst outcomes or the greatest use of resources. For the MMHU workstream, the study will show our colleagues in the Trust where the patients are, how they can be identified, how many we will have at any one time, and what the likely throughput will be on the new MMHU. This will allow us together to design our recruitment processes to the number of beds we have available (we anticipate a 20-bedded general medical ward adjacent to a psychiatry ward will be chosen). Similarly the ISAR study will help us to establish which of the very many elderly patients attending and being discharged from the AMU are most at risk and hence most likely to need some sort of intervention focussed upon them (or, conversely, to establish which are stable and low users of resources). The Care Home cohort study will describe patterns of care, identify just how much out of hours care is taking place and the characteristics of which sort of residents this care is focussed upon. The register, ISAR evaluation and cohort studies therefore provide the quantitative information upon which services can be developed.

3.2 The development of the services themselves is when the actual staff, locations and working practices are chosen and assembled. This sub-phase will require a particularly close partnership between local service personnel and the research team, with emphasis upon the service personnel.

For the MMHU it requires deciding precisely on a venue, assembling a team of people to work on it, in-service training, and agreement of procedures and protocols. This is why we have asked, via NHS Support Costs, for mental health nursing staff to help at this phase. Our research team will obtain important views to obtain from staff, users (patients or their carers), co-patients (those without mental health problems), advocates and outside bodies such as the Alzheimer’s Society. Interviews, case studies and focus groups will illuminate problems and potential solutions from these differing perspectives. This is where we will undertake qualitative work and require expertise. Our earlier and on-going literature review will provide the evidence base for proposed treatments and interventions, and also identify and review relevant good practice guidelines and protocols. By these means we expect to develop a realistic, logical intervention, and to have described it and justified its working practices.

For the AMU CGA study, we will need to clarify the job timetable of a geriatrician to ensure the right amount of time is available for clinics, domiciliary visits, etc. Working relationships will need clarifying between secondary, intermediate and primary health care, and social services. The working practices of the existing multidisciplinary geriatric team will need to be adjusted, and again we will draw upon the mental health expertise of the mental health nurse (from NHS support costs). Again, our research team will obtain important views to obtain from hospital staff but also community staff, patients and their carers. Interviews, case studies and focus group will illuminate problems and potential solutions from these differing perspectives.

The Care Home CGA study is more complex than the other two workstreams. This is because we will have two sets of service professionals to deal with: GPs and their primary care teams, and care home staff. Furthermore, we will have more descriptive data to collect. For example, we will run a survey of how care homes record their assessments, and a survey of the nature of GP cover to care homes.

In response to point 9 we have described the financial arrangements we have proposed to incentivise the developmental work. To develop the intervention we intend in this study for a small group to be formed of local GPs and the care homes to which they provide cover. We will examine individual cases as a group of clinicians to arrive at the best way to undertake assessments, record them, and what to do when a trigger observation is made. This process will have a large educational element to it. Our earlier and on-going literature review will provide the evidence base for proposed treatments and interventions, and also identify and review relevant good practice guidelines and protocols.

To aid the later implementation of the intervention we aim to develop a larger care home research group, comprising GPs and primary care staff, care home staff, and representatives of the residents. This group will meet partly for educational purposes, but we will also use them to gather their views and suggestions: some simple surveys, focus groups and interviews will be undertaken.

3.3 A third sub-phase is required to check that the new intervention has been implemented and is working as intended. Clinical effectiveness techniques will be used (for example, audit) to determine whether guidelines for referral are followed, whether protocols and procedures are followed, whether outcomes are compatible with those intended, and to gauge user and staff views.

We shall request approval from the research ethics committee and R&D departments to undertake this developmental work as an entire package of study for each workstream. We will seek approval in a form which allows the work to be done flexibly, according to the local conditions at any moment.

4. Consent and the Mental Capacity Act

This has been an area to which we have devoted considerable thought. We welcome the Research Selection Panel’s request for us to elaborate.

We accept that many of the potential participants of this programme will not have the mental capacity to give real consent to participate in the research – just as they as patients will often not have the capacity to consent to the medical and social interventions that are proposed for them.

We will, of course, submit the research for approval by the appropriate body (the Ethics and R&D Committees) in relation to our Strategic Health Authority under the DoH, comprising the appropriate authority (relating to MCA, 2005, s30(1–6)). We will endeavour to ensure that all our procedures, as outlined then to them, and as implemented, will be in conformity with MCA, 2005, s31, s32 & s33. All the interventions we propose could be argued to be ‘best practice’ and none involve new drugs, surgery or technical procedures. The ‘risks’ would more appropriately be described as the burdens of the data collection processes. We will strive during pilot phases to ensure that the assessment burden is as brief as necessary to answer the questions, and hence minimise this burden, Thereby we will strive to make it more acceptable to potential recruits.

The Panel wisely draws our attention to how gaining consent from older people (who may be cognitively impaired) would be challenging and place pressure on the delivery of the programme. All the research will be overseen by senior clinicians dealing with older people, for whom these ethical and practical dilemmas, related to the consent seeking process and problems with lack of capacity, are literally an everyday occurrence in clinical practice. We are often consulted on these matters by other clinicians and asked to teach on these topics. We would also like to point to our record in research with frail older people to illustrate that we have considerable expertise in doing this at a practical level and delivering results. For example, in our recent RCT of a variant of nasogastric feeding in acute stroke (where most patients did not have capacity, and where the ethical issues and the emotional context were complex) we achieved recruitment rates of 74% of all nasogastric tube-fed patients (abstract to be presented to the British Geriatrics Society April 2008). This expertise will be strongly used to support the researchers with such problem areas.

We re-affirm our plans to recruit researchers with an appropriate previous health professional background such that they have facility with working with these problems, and to ensure that researchers are carefully trained and closely supervised for these areas.

There is no doubt that there will be a sufficient supply of appropriate patients in the clinical setting but it is reasonable to be concerned that delays in going through consent/capacity processes could slow the research plans. As a result of dropping workstream 3, this will give us more researcher time than we had previously allowed, and this additional factor we believe will considerably strengthen our ability to overcome any such threatened delay problems.

We believe our research will be fully consistent with the provisions of the Mental Capacity Act. In relation to Mental Capacity Act issues it may be helpful for you to see what we have drafted to accompany the appropriate section of our Ethics Committee application. Thus, regarding Section 30, supplementary information:

1. What impairing condition(s) will the participants have?

   Whilst some participants will be capable some participants will be incapable as a result of suffering from dementia or delirium, or from other disorders such as severe depression or serious learning disability.

2. Justifying the inclusion of participants unable to consent for themselves. It should be clear why the research could not be carried out as effectively if confined to adults capable of giving consent.

   Quite clearly, since such people are the main group within our research, to exclude them would invalidate the work almost entirely. The study focuses on how older people, both with impaired capacity as well as those who are capable, are best dealt with in medical crises, to achieve the best outcomes for themselves and their carers, and it is therefore essential that all such patients are included. In many cases mental health or mental capacity problems, perhaps varying, will considerably complicate the medical crisis. This again makes it essential to include all such cases to help to establish the best ways for services to achieve the best results for all involved.

3. How will the capacity of potential participants to consent to the research be assessed? Who in the research team will make the assessment and what knowledge of the participant or relevant training/experience will they have to enable them to undertake it?

   The researchers conducting the interviews will have primary responsibility for assessing the capacity of the individual to participate in research, under the direct supervision of the workstream lead. They are expected to be recruited from health professional backgrounds and to already possess considerable experience in these areas. They will receive training from members of the research team, including Professor Gladman and Assoc. Professor Jones, who are experienced in this field, and have given specialist clinical advice on and taught on such matters. Additionally, the researchers will be informed by the clinical teams of all necessary and appropriate issues and information regarding the participant.

4. Does the research have the potential to benefit participants who are unable to consent for themselves?

   1 Yes 0 No

5. Will the research contribute to knowledge of the causes or the treatment or care of persons with the same impairing condition (or a similar condition)?

   1 Yes 0 No

   If Yes, please explain how the research will achieve this:

   The study will give us considerable knowledge on how clinical services can be improved to cater better for and to achieve better outcomes with vulnerable older people in medical crises, and especially with those with mental incapacity. The research will have great potential importance for informing improved practice in this area to the benefit of incapable people.

6. Will the research involve any foreseeable risk or burden for these participants, or interfere in any way with their freedom of action or privacy?

   1 Yes 0 No

   If Yes, please give an assessment below. Highlight any risk, burden, restriction or invasion of privacy specific to these participants and say what will be done to minimise it:

   This will be minimal in that all patients will anyway receive normal clinical management, either the standard treatment approach or an intervention package approach expected to be revealed as achieving better results, but all patients will be subject to structured clinical assessments which may be somewhat more extensive than would always be practice. There would be no different impact on their freedom of action or their privacy than would be involved in ordinary clinical practices but the information collection might in some cases be somewhat longer and potentially tiring. The researchers will be trained to be sensitive to such issues and will adapt their approaches to avoid a burdensome approach in interviewing.

   The project information would be passed to family carers (or consultees) assenting to participation on behalf of incapable research subjects, under s32 of the Mental Capacity Act, in advance of the involvement of the person lacking capacity, creating another mechanism for the views of the person lacking capacity to be taken into account.

7. What arrangements will be made to identify and consult persons (‘consultees’) able to advise on the inclusion of each individual participant and on their presumed wishes and feelings?

   Consultees will be identified by the referring clinician at the time participants are considered for participation. For persons without an obvious family member or similar non-professional carer to serve in this role, an independent advocate, planned to be either from the Alzheimer’s Society or Age Concern, will be approached to act as consultee. We would, of course, also be pleased to work with any such consultees already identified and made available by the Ethics Committee.

8. Is it possible that a participant might need to be treated urgently as part of the research before it is possible to identify and consult a consultee?

   0 Yes 1 No

   If Yes, say whether arrangements will be made instead to seek agreement from a registered medical practitioner and outline these arrangements. Or, if this is also not feasible, outline how decisions will be made on the inclusion of participants:

9. What arrangements will be made to consult consultees during the course of the research where necessary? What burden could this place on consultees?

   The consultee would be approached for involvement once, prior to contact being made with the participant. The consultee could, of course, approach the research team with new information at any time, but it is expected that the interviews with participants would take place promptly following agreement of the consultees. It is therefore not expected that it would be necessary routinely to involve consultees after the initial agreement.

10. What steps will you take, if appropriate, to provide potential participants who are unable to consent for themselves with information about the research, and to consider their wishes and feelings?

    This will be approached sensitively, as appropriate, at the pace and to the depth the individual wishes, at the beginning of the interview with the participant, including the presentation of written information.

11. Is it possible that the capacity of participants could fluctuate during the research? How would this be handled?

    Individuals’ capacity may vary, for example, according to the time of the day interviews occur. We will seek information on this from the clinical team and we will endeavour to conduct the interview at a time when the participant is at his or her most competent. However, it is essential that the relevant clinical data is collected at the time of the medical crisis rather than sometime later. For all those with incapacity we are involving a consultee from the initial stages of the research for all such participants and this will be on-going, and we will involve such a consultee at any later stage with any individual who subsequently loses capacity during the duration of his/her involvement with the research.

    Thereby, we intend that for participants who lose capacity at any stage the research will thus continue to meet the terms of the Mental Capacity Act 2005.

12. What will be the criteria for withdrawal of participants?

    Capable individuals will withdraw by indicating their refusal at any stage. For those lacking capacity withdrawal will occur: either: via a communication from the consultee that the individual would wish to be withdrawn from the study, consistent with s 32(5) of the Mental Capacity Act; or, if the participant indicated in any way that he or she wished to be withdrawn from the study, consistent with s 33(5) of the Mental Capacity Act.

13. Describe what steps will be taken to ensure that nothing is done to which participants appear to object (unless it is to protect them from harm or minimise pain or discomfort)?

    The views of the participant will be sought at the time the interview is to occur, and if the individual wishes it, additional time can be made available for the individual to consider matters. In addition, the consultees are asked in the information sheet to notify us if there is reason to believe that the participant does object or would object if he or she had capacity to anything we propose.

14. Describe what steps will be taken to ensure that nothing is done which is contrary to any advance decision or statement by the participant?

    The information sheet to the consultee directs his or her attention specifically to advance views and decisions of the participant.

5. Co-location and sharing of expertise

6. Detailed study design for the Medical Mental Health Unit and issues of contamination

The detailed study design is appended. The Research Selection Panel in particular was concerned that if the workstream lead was also the clinician involved, that this might introduce ‘contamination’ or some form of bias. Again, we had thought long and hard about this, and found it difficult to provide a brief justification and description of the issues in the final proposal.

We accept that any open (unblinded) service trial is open to biases – such as our trials of a stroke unit or rehabilitation services. A concern is that undue involvement of a researcher to one group of participants only, such as the trial participants, might evoke some sort of response bias either by biasing the proportion of respondents, or by encouraging a more favourable response. The approach we have taken towards this in the past and which we shall use here are:

• Recruitment should be by a researcher who is not part of the clinical service.

• This research should recruit participants in a way that establishes the clinical equipoise of the trial: we do not know which is better. This reduces expectation bias, or at least does not encourage it.

• Follow up should be independent and blinded to allocation status, and independent of the service. This can be achieved by several means including the use of telephone follow up where possible and the use of a researcher who is unaware of allocation. The exact words used when approaching patients to obtain follow up information must be carefully designed to avoid invoking bias. Researchers will be carefully trained and supervised.

• Outcome measures should be chosen that do not inherently or unfairly favour the intervention, or refer to it.

• Data should be recorded by someone blinded to the allocation status and independent of the intervention.

• Data should be analysed blind to the allocation status.

Another concern is that the expertise being developed for the treatment might ‘leak’ into routine, control, practice, and this ‘contamination’ might reduce the observed treatment effect. It is noteworthy that despite this possibility, the benefits of stroke units were still observed in trials. We argue that the correct trial to perform, the correct research question to answer, is ‘what is the benefit of a Medical and Mental Health Unit when compared to optimised ordinary care’. Therefore, if the control treatment is enhanced simply by the presence of the trial, then this is the correct control condition to test. Thus ‘contamination’, to some extent, is not a bad thing. In fact, we suspect that however much ‘contamination’ occurs, it will still not enable non-specialised services to develop the clinical skills and expertise, or the management and organisational practices that would be possible in a specialist unit. By analogy with the stroke unit studies, where the benefit is attributed to ‘co-ordinated care’, we suspect that it will be these elements, the ones that cannot leak and cause contamination, that will be important.

With these safeguards in place, we expect biases to be minimised. We see no reason for the fact that the workstream lead will also be the lead clinician from introducing bias.

7. Workstreams 1 & 2: further details including justification of sample sizes

Full protocols for workstreams 1 and 2 are attached.

Owing to lack of space in the original proposal document, we did not fully explain our sample size calculations. The Research Selection Panel requested more information about the sample size calculations for workstreams 1 and 2. Each of these workstreams has two quantitative elements: a cohort study used largely to prepare for each trial, and the trials themselves. Fuller descriptions of the sample size calculations and reasoning for each part are given below.

8. Detailed health economic study plan for each workstream

For brevity, we described in the final proposal the overall approach across all workstreams:

The perspective for costs taken will be that of the third party payer: NHS, personal social services and private care home sector. Resource use data will be collected for the index admission or care episode, along with subsequent secondary care (inpatient and outpatient episodes), primary and community (geriatric day hospital, GP and social services home care services) contact and specialist accommodation requirements. GP records are a good source for primary care data, but patient or carer reports are more reliable for other service contacts [10.12]. A modified version of the Client Service Receipt Inventory [10.13] will be developed to capture these data across all workstreams. Unit costs will be taken from routine local or national sources, as appropriate. Depending on the level of skewness in the data, differences in costs will be analysed using t-tests or non-parametric bootstrapping. In line with standard practice, the EQ-5D will be used in the generation of QALYs in economic studies to generate conventional incremental cost effectiveness ratios (ICERs). Where appropriate, patients will provide EQ-5D data, and where this is not possible, carers will provide proxy EQ-5D ratings. In view of the limitations of this instrument in this population group, we will also use disease specific quality of life measures in each strand, and examine their relationship to the EQ-5D. QALYs will also be extrapolated to lifetime, and discounted at 3.5% to estimate the cost effectiveness of the interventions over the lifetime of the target population. Uncertainty around ICERs will be expressed using appropriate probabilistic methods including bias-corrected bootstrapping methods. Cost effectiveness acceptability curves and estimates of expected value of information will be generated to inform decision-making and to quantify the costs and benefits of further research.

Fuller descriptions of the economic approaches we shall use for each workstream, taking into account specific comments by individual reviewers are given below.

The overarching aim of the economic analysis is to assess the relative value for money of each of the applications of CGA. To maximise the transferability of the methods and results obtained, the economic analysis protocol will have a standard structure across the three workstreams, with common data collection and analysis strategies used whenever appropriate. In addition to this, the individual workstreams will have specific issues regarding perspective, resource use and outcome measurement that will need to be managed. The parameters and protocol for collection of those parameters will be piloted, finalised and applied during the cohort study phases, and will then be applied to the evaluative phases. The MMHU and AMU workstreams will some overlap in perspective and resource use categories. The care home workstream will take a similar approach, but has more complex data handling needs due to the multi-level nature of data collected, and the increased number of care models.

Approaches to characterisation and valuation of resource use will be workstream specific due to the complexity of each intervention under evaluation. Details of the approaches are provided in the next section. A consistent methodological approach is presented for the three workstreams, with adaptations for each intervention where necessary. Common approaches will be used to assess utility of outcome for the economic evaluation, and to carry out the economic analysis, and are provided in this section.

9. Re-evaluate costs for workstream 4

We welcome the opportunity to improve the funding of this workstream, and agree that the lack of costs here would have rendered it vulnerable. We still believe that this project is possible, and will depend greatly upon clinical leadership – from the study team and local colleagues.

10. Summary of changes to the finance form

11. Conclusion

We trust that in this response letter, and in the further attached information, we have satisfactorily answered each of the queries raised by the Research Selection Panel. We are now confident that we will have the resources to complete this programme successfully.

The main service development ideas in our proposal are already part of the longer term service development strategy for the Nottingham University Hospitals Trust. We have already started discussions to prepare for the staff and service changes. We have already identified some of the key research staff and will shortly begin the public and user engagement process. We will be ready to start on proposed start date of 1 August 2008, subject to the contractual process.

Yours sincerely

Professor John Gladman, on behalf of the co-applicants.