A multidomain decision support tool to prevent falls in older people: the FinCH cluster RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 13/115/29. The contractual start date was in May 2016. The draft report began editorial review in September 2020 and was accepted for publication in May 2021. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Disclaimer

This report contains transcripts of interviews conducted in the course of the research and contains language that may offend some readers.

Copyright statement

Copyright © 2022 Logan et al. This work was produced by Logan et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaption in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2022 Logan et al.

Why are falls important?

A fall can have a devastating impact on a person, their family and their carers, and can place demand on health and social care resources. Falls are common, with one-third of those aged > 65 years and half of those aged > 80 years falling at least once per year. 2 Ageing societies pose challenges for health and social care systems. The UK has an ageing population; there are nearly 12 million people ≥ 65 years, of whom 5.4 million are aged ≥ 75 years, 1.6 million are aged ≥ 85 years, over 500,000 are aged ≥ 90 years and 14,430 are centenarians. 3 Falls are the most common cause of emergency hospital admissions for older people. 4 In 2017/18, there were around 218,000 emergency hospital admissions related to falls among patients aged ≥ 65 years, with around 149,000 (68%) of these patients aged ≥ 80 years. 5

Consequences of a fall

Falls can cause injury, distress, pain, reduced mobility, loss of confidence or independence, and a fear of falling, leading to reduced levels of activity in daily life and increased mortality. 2 In 2017, 5048 people aged ≥ 65 years died from having a fall, equating to 14 people every day. 4 Hip fracture is the most common serious injury following a fall, and it is estimated that around one-quarter of people who are aged > 65 years and fracture their hip will consequently require long-term care in a care home. Hip fractures are the leading cause of accidental death. 6 Older adults with frailty are less able to cope and recover from accidents, physical illness or other stressful events, including falls. People living in care homes are more frail than community-based populations and their care needs merit specific attention. 7 Delivering comprehensive, consistent and structured enhanced support to those in care homes will ensure that their needs continue to be identified and met proactively. 8

Prevention and management of falls

The National Falls Prevention Coordination Group’s falls and fractures consensus statement advocates a whole-system approach to the prevention of falls that includes risk factor reduction across the life-course, case finding and risk assessment, strength and balance exercise programmes, healthy homes, high-risk care environments, fracture liaison services, and collaborative care for severe injury. 9 Identification of those at risk of falls; assessment of contributory risk factors for falling; and interventions to reverse, reduce or modify those risk factors are recommended by the National Institute for Health and Care Excellence (NICE). 10 However, these recommendations are based on people living in their own homes who have capacity to listen and react to health professional’s advice; they were not written for care home staff or residents.

Falls in care homes

Approximately 421,000 older people were living in UK care homes in 2016–17. 11 Two levels of care accommodation are available, depending on the level of care required. These are personal care and 24-hour support (provided by care homes without nursing, sometimes called residential homes), and additional on-site nursing (provided by care homes with nursing, sometimes called nursing homes). Some homes provide both levels of care (dual-registered homes). Care homes both with and without nursing match the international consensus definition of a nursing home. 12 Around 5500 different providers in the UK operate 11,300 care homes for the elderly, with 95% of beds provided by the independent sector (both for-profit and charitable providers). 13

The majority of older people living in care homes are aged > 85 years; live with cognitive impairment, multimorbidity and limited mobility; and take multiple medications. 7 The rate and risk of falls for residents of care homes is high, with falls being three times more common in care home residents than in older people living in their own homes, and those falling in care homes being 10 times more likely to suffer a serious injury. 14 One in five people living in care homes will die within a year of suffering an injurious fall. 15 Falls may often occur as a symptom of underlying frailty and illness. 16 Falls may engender feelings of anxiety in care home staff, and fear of litigation and complaints, which may have an impact on care staff’s willingness to encourage residents to be physically active. 17

Multiple diverse factors can contribute to the increased risk of falls in care home residents, including frailty, the presence of long-term conditions, physical inactivity, taking multiple medications and the unfamiliarity of the surroundings. The interaction of factors that contribute to an individual’s risk of falling is unique to them; therefore, interventions to reduce falls risk must be individualised and meaningful for individuals. Protocols used to perform risk assessments for falls in hospitals and care homes vary in quality, do not necessarily trigger individually tailored interventions to reduce risk factors and, in some cases, seek only to stratify risk. 18

A number of interventions have been applied to reduce falls in care homes, including multifactorial approaches. However, a Cochrane review found the evidence to be of low quality and inconclusive regarding effective strategies to reduce both falls rates and falls risk. 19

Implementing health-care interventions in care homes

The NICE falls guidelines and quality standards10 do not explicitly provide guidance for care home residents. Instead, care home staff and clinicians are relied on to apply research evidence from hospital in-patient and community falls trials. Interventions are often difficult to implement within a care home and it is unclear whether or not interventions can lead to cultural change that becomes embedded in care home practices so that effects are sustained or even increase after the intervention. 20

Improving the lives and health of older people living in care homes is a major UK government priority and is embedded in the NHS Long Term Plan. 21 The Enabling Research In Care Homes (ENRICH) initiative brings together care home staff, residents and researchers to facilitate the design and delivery of research in care homes. Increasingly, research has looked at providing care home staff with training in an aspect of care from experts, with the aim of increasing carers’ knowledge and expertise in caring for older people with frailty. A recurrent observation has been the need to adapt existing approaches to improvement and implementation to take account of and empower care home staff and organisations. 22 In addition, there is evidence that a number of care home-specific issues affect how ready care homes are to engage with external organisations regarding change. 23 Where care home staff and NHS professionals are required to work together, there is evidence that outcomes will be better if specific activities that encourage shared working between care home staff and visiting health-care professionals are integrated into care delivery. 24

Falls are often a consequence of undiagnosed and untreated underlying health conditions, frailty and environmental factors, including the way that care is structured. Approaches to falls prevention are therefore likely to be multiagency collaborations between care home staff and external health-care providers. Taking into account the lessons learned about effective partnership working and implementation science, adoption of early co-design and strong stakeholder involvement in the planning and design of the intervention and study was essential to create a falls prevention programme that was pragmatic and suitable to subsequent adoption, while also maximising the external validity of the study. The Medical Research Council’s framework for developing and evaluating complex interventions25 provided a framework for the development of our falls prevention programme, feasibility study and multicentre trial. The framework highlights the need to understand the context for delivery of a complex intervention, which is even more relevant in care homes as they pose a distinct challenge for the introduction of complex interventions: they vary in size, funding, workforce and culture, and house vulnerable individuals with far-reaching health and social care needs.

The feasibility study [Falls in Care Homes (FiCH)], which laid the foundation for the current Falls prevention in Care Homes (FinCH) trial, found that an effective falls prevention programme would have to use language that care home staff could understand and identify with, explain the rationale for falls prevention in ways that aligned with care home organisational priorities, and be conducted in a way that allowed for care home schedules and care regimes. 26

The GtACH programme

Our intervention, the Guide to Action for falls prevention in Care Homes (GtACH) programme,27,28 was developed to reduce falls rates by supporting care home staff to identify risk factors for falling that are pertinent for an individual and take action to reduce those risks. It was co-produced by a group of care home staff, clinicians, researchers, public, voluntary and social care organisations and includes care home staff training, support and documentation. Care home staff are trained to implement the programme in their home by an NHS falls lead over 1 hour in small groups. The NHS falls lead is a nurse, physiotherapist or occupational therapist who has specialist training, skills and knowledge in falls prevention and bone health. The homes are asked to identify a falls champion to help maintain implementation. When the GtACH programme is implemented, homes are given a copy of the GtACH manual and are supplied with the GtACH tool. The latter is a paper form, comprising an assessment component (a checklist of falls risk factors) and care planning section supported by suggested actions linked to each fall risk factor (see Appendix 1). After training, it takes an average of 20 minutes to complete the GtACH tool for each resident. Initial proof of concept work29 and a subsequent feasibility randomised controlled trial (RCT), FiCH, showed that the GtACH programme was implementable and changed staff behaviour in line with gold standard practice. 26

Limitations of previous studies

A Cochrane review published in 2018,19 which looked at the effectiveness of interventions designed to reduce falls in older people in care facilities and hospitals, found that the majority of trials were at high risk of bias in one or more domains, mostly relating to lack of blinding. With few exceptions, the quality of evidence for individual interventions in either setting was generally rated as low or very low. Risk of fracture and adverse events (AEs) were generally poorly reported and, where reported, the evidence was of very low quality. The authors concluded that there was a need for further research and, in particular, large RCTs in care facilities to inform practice in falls prevention.

Justification for current trial

The Cochrane review in 201819 concluded that further research to strengthen the evidence for multifactorial interventions for falls reduction in care homes was required as there were some individual trials that showed potentially important reductions in the rate of falls. The authors noted that a key feature of these multifactorial interventions was the individualised nature of the interventions delivered. The review stated:

[t]his implies that further research with emphasis on an individualised, standardised approach to delivery of interventions with consistent description and application within further trials is warranted, including as a clear description of existing falls prevention practices in the control arm of any trials and the interaction of the intervention arm of the trial with usual care. A mixed methods approach may be necessary to achieve this.

Reproduced with permission from Cameron et al. 19 Copyright © 2020 The Authors. Cochrane Database of Systematic Reviews published by John Wiley & Sons, Ltd. on behalf of The Cochrane Collaboration. This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution Non-Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for non-commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by-nc/4.0/

Research aims

The aims of the trial were to:

• determine the clinical effectiveness and cost-effectiveness of the GtACH programme compared with usual care

• complete a process evaluation to provide insight into the implementation of the GtACH programme and to contextualise trial findings.

Trial design

The FinCH trial was a pragmatic, multicentre, cluster parallel 1 : 1 RCT to evaluate the GtACH programme compared with usual care (an absence of a systematic and co-ordinated falls prevention process) in UK care homes for older people. The allocation was at the care home level. A flow diagram of the trial design is shown in Figure 1.

FIGURE 1.

Flow diagram of the trial design.

Care homes and residents took part in the study. The primary RCT health outcome was falls rates in the 90-day period between 91 and 180 days post randomisation. Secondary outcomes were collected at baseline and at 3, 6, 9 and 12 months post randomisation.

Trial population

The trial population was care homes (with and without nursing) registered with the care home regulator [the Care Quality Commission (CQC)] in England.

Interventions

Baseline and outcome measures and assessment

The following characteristics of care homes were collected:

• number of staff in caring role

• number of beds in care home

• number of residents

• falls monitoring processes.

Ethics and regulatory issues

The trial was not initiated until after the protocol, informed consent forms and PISs received approval/favourable opinion from the REC and the NHS Research and Development department. Approval was received from the NHS Health Research Authority and NHS sites. (Yorkshire and the Humber – Bradford Leeds REC, 11/04/2016, reference 16/YH/0111).

The NCTU governed the trial, monitored data collection, and completed data checking and data cleaning.

The trial was conducted in accordance with the ethics principles that have their origin in the Declaration of Helsinki, 1996;39 the principles of GCP, and the Department of Health and Social Care Research Governance Framework for Health and Social Care, 2005. 40

The trial was registered as Current Controlled Trials ISRCTN34353836 with protocol V6 14 November 2017.

Sample size

The sample size was based on the primary RCT outcome of falls rate during the 90-day period between 91 and 180 days post randomisation. The original total sample size estimate was for 1308 residents to be recruited from 66 care homes (33 in each arm). This assumed a falls rate of 2.5 falls per year (0.625 falls in 3 months) in the control arm,41 80% power and a two-sided significance level of 5%, resulting in the need to recruit 189 residents per arm to detect a 33% reduction in falls rate in the GtACH arm. A reduction rate of 33% was chosen as this was the rate achieved by community-based falls prevention interventions16 and therefore deemed clinically significant. The sample size calculation was based on information obtained from a previous care home study that had a falls rate of 15 falls per year,26 but only recruited those residents who had fallen recently. The adjustment for clustering assumed an average cluster size of 20 residents42 and an intracluster coefficient of 0.1,42 giving a sample size of 549 residents per arm. Incorporating a further 16% into the sample size to account for potential attrition gave a total sample size of 1308 residents (654 in the GtACH arm and 654 in the control arm).

During recruitment it became apparent that the average number of residents per cluster was slightly smaller than expected (19 residents) and the size of the clusters was variable (coefficient of variation 0.5). Based on this new information, the design effect for the revised sample size calculation increased from 2.9 to 3.275, leading to a total sample size of 1474 residents after the adjustment for 16% attrition rate. This led to a need to recruit 39 care homes per arm. Across all sites, it was anticipated that the rate of recruitment was five or six care homes, each with 18–19 residents.

Randomisation and blinding

Care homes were randomised on a 1 : 1 basis to one of two parallel arms: the GtACH programme or control (usual care). Participants, care home staff, site NHS falls leads and RAs undertaking the process evaluation at the care homes were not blinded to allocation.

Randomisation of homes to trial arms occurred after all participants had given consent and all baseline data had been collected. The RA who gathered the baseline information notified the local site NHS falls lead when the care home was ready to be randomised. The NHS falls lead used a remote, internet-based randomisation system to obtain the allocation for each home and informed the falls champion within the care home of the allocated GtACH arm.

Randomisation was based on a bespoke, computer-generated, pseudo-random code using variable block randomisation within strata [site, care home type (nursing/residential/dual registration)] provided by the NCTU via a secure web-based randomisation service.

The sequence of treatment allocations was concealed from the study statistician until all interventions had been assigned and recruitment, data collection and all other study-related assessments were complete.

The Trial Management Group (TMG) and the Data Monitoring Committee (DMC) were unblinded to the intervention. The chief investigators and principal investigators (PIs) had direct contact with the randomised care homes, although not with the participants.

The RAs at the sites, participating residents and staff informants were blind to allocation at recruitment and for baseline data collection because participants were recruited prior to care home randomisation. RAs collecting outcome data were not informed of allocation (occasions of unblinding were recorded). NHS Digital data were extracted blind to allocation.

Interim analyses required to populate recruitment and data monitoring for harm reports for the DMC were conducted on unblinded data by the NCTU.

Assessment of compliance

Care home records for all care homes randomised to the intervention were reviewed by the NHS falls lead during the first 3 months post randomisation to GtACH training and during the implementation period to consider broad compliance with the GtACH programme. As part of this evaluation, evidence was sought that the GtACH manual was accessible, the GtACH poster was displayed, and the GtACH paperwork was attached to care records. The number of care home staff who attended the GtACH programme training was recorded as a proportion of the total number of available staff. Compliance with the intervention was also evaluated as part of the process evaluation. We did not record the number of GtACH forms completed in the homes.

Withdrawal of participants

Residents were able to withdraw from the trial at their own request, their consultee’s request or at the discretion of the investigator. The participants were assured that withdrawal would not affect their future care. Participants and consultees, when appropriate, were made aware (through the information sheet and consent form) that should they withdraw, the data collected up to the date of withdrawal could not be erased and may still be used in the final analysis. Care home managers were able to withdraw support for the trial at their own request or at the discretion of the investigator (residents were also withdrawn following care home manager withdrawal of consent).

Adverse event reporting

Data on AEs (serious and non-serious) were not collected in this study.

This was a low-risk intervention. No specific risks, untoward incidents or AEs were reported during the feasibility work. The GtACH tool recommends that actions are taken but does not stipulate what these actions are, other than to recommend referral to health professionals as appropriate. If residents became distressed during the GtACH assessment or when intervention actions were completed, the process was halted, and the event was recorded and closely monitored until resolution, stabilisation or until it was shown that the study intervention was not the cause. The participant had the right to decline any intervention at any time.

Gentle exercises were 1 out of the 30 activities included in the action checklist. If gentle exercises were recommended after the assessment, staff in the care home were advised to refer the resident to a physiotherapist so that a programme of exercise could be put in place. It was possible that participants may have suffered an injury that they would not have had if they had not taken part in the exercise. These were recorded and monitored by the falls champion in the care home. If there was any concern, the falls champion referred to the NHS falls lead for advice. If required, the exercises were stopped.

Falls rates were monitored for harm and reported to the DMC and TSC every 3 months after they were collected. The DMC and TSC had the ability to recommend changes to the study protocol if falls rates were substantially higher than expected. The DMC reviewed unblinded safety data, including reported falls frequencies, at least yearly. These were provided by the NCTU via secure e-mail.

As the GtACH programme is copyrighted by the University of Nottingham, the University of Nottingham was responsible for any issues that arose because of the design of the intervention, training given to care home staff or any issues with the programme itself. However, in respect of the use of the GtACH programme in care homes, the care home was responsible if the programme as a whole was incorrectly used. Care home managers were asked to confirm that they had indemnity for this and, if the indemnity did not include research, they were asked to seek indemnity from their insurance providers, making clear that all individual components of the GtACH programme were currently being used in routine care but in a consistent or structured manner.

The GtACH assessments and or actions were stopped if the participant showed evidence of distress. This was documented in one participant, but the participant was not withdrawn from the study.

Data handling and record keeping

All trial staff and investigators endeavoured to protect the rights of the trial’s participants to privacy and informed consent, and adhered to the Data Protection Act 1998. 43 The CRF collected the minimum required information for the purposes of the trial only. CRFs were held securely in a locked room, or locked cupboard or cabinet. Study data were collected and managed using REDCap electronic data capture tools32,33 hosted at the University of East Anglia. REDCap is a secure, web-based software platform designed to support data capture for research studies, providing (1) an intuitive interface for validated data capture, (2) audit trails for tracking data manipulation and export procedures, (3) automated export procedures for seamless data downloads to common statistical packages and (4) procedures for data integration and interoperability with external sources. Access to the information was limited to the trial staff and investigators, and relevant regulatory authorities. Data held on computers, including the trial database, were held securely and password protected. All data were stored on a secure dedicated web server. Access was restricted by user identifiers and passwords (encrypted using a one-way encryption method). Information about the trial in participants’ care home records was treated confidentially in the same way as all other confidential medical information. Electronic data were backed up every 24 hours to both local and remote media in an encrypted format.

Statistical analysis

Analyses were undertaken on an intention-to-treat basis in which care homes were analysed in the arm to which they were allocated regardless of their compliance with the intervention. The progress of care homes and residents through the phases of the trial from the screening and enrolment of care homes to the analysis of the outcome data has been summarised in the Consolidated Standards of Reporting Trials (CONSORT) flow diagram (see Figure 6). Those who died between care home randomisation and the 3-month follow-up data collection were regarded as having been exposed to the intervention (GtACH/control) and recorded as lost to follow-up at the time of death.

Data were analysed according to a prespecified statistical analysis plan (SAP) that was finalised prior to the start of the analysis. Full details of all analyses are provided in the SAP (see Report Supplementary Material 1). Analyses were based on available case data. Two-sided tests were used to test statistical significance at the 5% level. The analysis was carried out using standard statistical software, either Stata/MP^® version 16 (StataCorp LP, College Station, TX, USA), SAS^® (SAS Institute Inc., Cary, NC, USA) or R (The R Foundation for Statistical Computing, Vienna, Austria).

Baseline characteristics of care homes and residents as well as outcome measures at baseline and each follow-up time point were summarised by treatment arm using descriptive statistics. The baseline falls rate was expressed as the number of falls per 1000 resident-days for each arm.

The primary outcome, the rate of falls per participating resident during the 90-day period between 91 and 180 days post randomisation, was expressed as the number of falls per 1000 participating resident-days for each arm. This period was chosen to give time for the intervention to be implemented after training, while acknowledging that people in care homes have short life expectancies. The secondary outcomes, rate of falls occurring during the 90-day period between 181 and 270 days post randomisation, and the 90-day period between 271 and 360 days post randomisation were calculated and reported in the same way as for the primary outcome.

The number of falls per resident was compared between arms using a random-effects/hierarchical two-level Poisson model with resident at level one and care home at level two, with length of residence in care home as an offset. The primary analysis adjusted for type of care home (residential, nursing, dual registration) and site.

Two additional models were fitted to assess the robustness of the model. In addition to adjusting for care home type and site, we adjusted for (1) baseline falls rate during the 3 months before the baseline assessment and (2) baseline falls rate and other variables that were thought to be associated with falling.

The falls rates during the 3-month periods prior to the 9- and 12-month follow-up were analysed and presented in the same way as for the primary outcome variable. For other secondary outcomes, arms were compared using multilevel regression analysis for continuous outcomes and multilevel logistic regression for binary outcomes. Regression coefficients and 95% confidence intervals (CIs) were presented.

Compliance with the intervention was calculated as the percentage of care-giving staff in each care home trained to use the GtACH screening and assessment tool. We calculated this as:

The percentage adherence was calculated and presented for each care home in the GtACH arm. The average adherence for all intervention care homes has also been presented.

Site support

The RAs supported the recruitment, data collection, data entry and data cleaning at each site. Delivery models included a combination of National Institute for Health Research (NIHR) CRN delivery staff, NHS research staff, NHS research therapists and academics employed by higher education institutions. RAs received training in trial processes by the NCTU. A RA network was developed to support RAs who were geographically dispersed across the 10 sites. This was led by a RA based in the same site as the chief investigator.

Monthly teleconferences were held to discuss challenges and share good practice, with any ongoing issues raised at the monthly trial management meetings. The format, content and evolution of the network was directed by the RAs. The peer support of the group allowed open discussion of challenges in a supportive environment. The primary foci of discussions were clarification of recruitment and data collection processes, identification of common barriers to recruitment through PCs, and sharing of methods to engage and sustain relationships with care homes. Face-to-face investigator meetings were held for all FinCH trial team members, including the RAs, PPI members and the teams delivering the interventions. Two meetings were held throughout the trial:

• May 2017 (16 RAs, 4 PPI members and 10 therapists attended), which included training in the recruitment of older adults lacking capacity, open discussions on good practice and the opportunity to feed back challenges and solutions to the senior research team.

• July 2018 (23 RAs attended), which included training in abstract and poster design, data quality and awards for sites related to recruitment outcomes.

Recruitment

Recruitment to the FinCH trial opened on 1 November 2016. The first care home was recruited on 16 November 2016 and the first resident was consented on 23 November 2016. In total, 10 sites were recruited to the FinCH trial (Table 1). The last care home was recruited on 29 December 2017 and the last resident was recruited on 31 January 2018; all care homes were randomised by 31 January 2018.

In total, 87 care homes were recruited; 84 of these care homes were randomised to either the GtACH programme or usual care (Figure 2). Over-recruitment was permitted to allow care homes that were actively engaged in the recruitment of residents by 29 December 2017 to continue through to randomisation by 31 January 2018.

FIGURE 2.

Recruitment of care homes.

In the same period, a total of 1698 residents were recruited; again, this was higher than the adjusted target of 1482 individuals (Figure 3).

FIGURE 3.

Recruitment of residents.

Figure 4 shows care home recruitment and resident recruitment by site. An average of 50% of residents from the participating care homes were consented and there was an average of 19.5 participants per care home. There was an average of 45 days between care home consent and randomisation. Figure 5 shows care home recruitment by site.

FIGURE 4.

Care home recruitment by site.

FIGURE 5.

Care home resident recruitment by site.

The CONSORT flow diagram (Figure 6) shows trial screening and recruitment to completion, as well as follow-up loss and completion. During the study there were 490 deaths, of which 459 occurred following care home randomisation. A total of 63 participants moved out of the study care home, 60 of these following randomisation. A total of 24 participants were residents in a care home that withdrew from the trial and a further eight participants were residents in a care home that was closed following a CQC inspection. Primary outcome data (falls occurring between 90 and 180 days after randomisation of the care home) were available for 1342 residents.

FIGURE 6.

The CONSORT flow diagram of the flow through the trial. The CONSORT flow diagram indicates the flow through the trial from the perspective of the availability of primary outcome variable data (falls). Availability of secondary outcome data, including QoL questionnaires is detailed in Secondary outcome analysis. a, Would not provide a falls champion (n = 24), had falls prevention in place (n = 13), had participated in FinCH feasibility trial (n = 4), home exclusively to learning disability/substance misuse (n = 3), in special measures (n = 1), did not wish to participate (no reason given) (n = 10). b, Wished to participate but did not have time (n = 1), initially indicated willingness but then stopped communicating with researcher (n = 8), initially indicated willingness but adopted a local falls intervention prior to consent (n = 3), wished to participate but were not recruited (no reason given) (n = 7).

Intervention adherence

Intervention adherence was defined as the percentage of care-giving staff trained to use the GtACH programme. Table 2 shows intervention adherence for each care home randomised to the GtACH arm. Average adherence to the training per care home was 71% (70% per site; minimum 17%, maximum 130%). A total of 14 (36%) care homes achieved adherence of ≥ 80%.

Baseline characteristics

Primary outcome: falls between 90 and 180 days

A negative binomial regression model [generalised estimating equation (GEE)] showed that the falls rate in the GtACH arm was reduced compared with that in the control arm in both the unadjusted and the adjusted analyses. Over the period of the primary outcome assessment (a 90-day period between 91 and 180 days after randomisation), the falls rate was 6.0 per 1000 residents in the GtACH arm and 10.4 per 1000 residents in the control arm (Table 5). Results of other approaches to the analysis of the primary outcome may be seen in Appendix 2, Tables 22–26.

Secondary outcome analysis

Changes to the analysis plan during the trial

Although we had specified that multiple imputation (MI) would be used to account for missing data, this was not undertaken because the primary reason for missing data was that the patient had died. However, we did collect and analyse the falls data until the date of death. We have presented medians and interquartile ranges for time spent in the care home, rather than means and standard deviations (SDs), because of the skewed distribution of these data.

Overview

The aim of this chapter is to report the within-trial economic evaluation undertaken to estimate the cost-effectiveness of delivering the GtACH programme in care homes from an NHS and personal social services (PSS) perspective. The primary analysis was a cost–utility analysis and presents proxy-reported outcomes as QALYs. A cost-effectiveness analysis (CEA) based on cost per falls averted was also conducted so that the GtACH programme can be directly compared with other interventions aimed at reducing falls.

Methods

The health economics analysis plan (HEAP) was written and approved by the TMG prior to the data being locked. The HEAP is available as Report Supplementary Material 3.

Results

Outcomes

Dementia Quality of Life, proxy complete

The mean Dementia Quality of Life, proxy complete (DEMQOL-P), utilities declined in both arms over time at a virtually identical rate. Thus, accumulated QALYs over time are virtually identical (Figure 7 and Table 17).

FIGURE 7.

The DEMQOL-P-U scores over the trial period. Complete case, n = 1453.

EuroQol-5 Dimensions, five-level version, proxy complete

The EQ-5D-5L-based utilities were consistently higher in the GtACH arm than in the control arm (Figure 8 and Table 17). This leads to larger observed mean QALYs. These raw unadjusted comparisons of the data illustrate the importance of adjusting for baseline utility.

FIGURE 8.

The EQ-5D-5L-P scores over the trial period. Complete case, n = 1453.

Falls

The primary outcome for the trial was based on the number of falls in months 4–6; analysis of the primary outcome showed a substantial difference between arms, with more falls in the control arm than in the GtACH arm. The economic evaluation measures the number of falls over the full 12-month period and, as Figure 9 shows, the difference between arms became smaller in the second half of the trial. Reflecting the RCT results, the intervention appeared to be more effective in the first 6 months than in the final 6 months.

FIGURE 9.

Mean number of falls per quarter over the trial period.

There was a mean of 1.89 (SD 3.66) falls recorded per resident in the GtACH arm and a mean of 2.77 (SD 7.44) falls per resident in the control arm, giving a mean difference of –0.877 (95% CI –1.469 to –0.285).

Cost-effectiveness

Results of the cost-effectiveness analyses are shown in Table 18. The raw incremental cost difference between arms was £19.76. However, this figure is unadjusted for baseline imbalances in cost and potential influences on cost other than the intervention. Adjusting for these yields an incremental cost difference of £108.26 (95% CI –£271 to £488) (‘adj, bs’ analysis). Adding in the cost of additional refresher sessions (sensitivity analysis 1; see Table 6) increases the incremental cost difference to £131.81 (95% CI –£248 to £511). Adding in the cost of mortality (sensitivity analysis 2, see Table 6) increases the incremental cost to £148.52 (95% CI –£245 to £542). The MI sensitivity analysis yields an identical mean cost, but narrower 95% CIs (see the final column, rows 1–4 of Table 6).

TABLE 18 - Cost-effectiveness results

Rows numbered 1–4 show the sample size, mean and SD of cost in each arm, followed by three estimates of the incremental cost, each shown with 95% CIs. The ‘raw’ increment is the arithmetic difference in the means. The ‘adj, bs’ increment is the incremental cost adjusted for baseline covariates using GEE, repeated with 10,000 bootstrapped resamples of the raw data; this is the primary analysis. The ‘MI, adj, bs’ increment is the incremental cost where missing data are imputed 50 times, with GEE run on each set and the results combined using Rubin’s rules. This entire process was repeated 200 times with bootstrapped resamples of the raw data (scenario 3 analysis). The four estimates of cost represent the base case and three scenario analyses around cost. The first scenario analysis adds in estimates of the cost of refresher GtACH tools as per the protocol (which was not observed in the trial). The second adds in the extra cost of mortality. The third adds in both the refresher and mortality costs. Rows numbered 5–7 show the same data for the three outcomes DEMQOL-P-U-based QALYs, EQ-5D-5L-P-based QALYs and number of falls. The ICERs for each analysis are as described, with the first column showing which rows are being divided.

Parameter		GtACH		Control		Incremental mean (95% CI)
		n	Mean (SD)	n	Mean (SD)	Raw	adj, bs (primary analysis)	MI, adj, bs
Cost
1	Base case	732	£3955.29 (£3949.38)	865	£3935.54 (£3879.90)	£19.76 (–£365.88 to £405.39)	£108.26 (–£271.06 to £487.58)	£108.26 (–£232.89 to £449.41)
2	With refresher GtACH tools	732	£3978.20 (£3955.87)	865	£3935.54 (£3879.90)	£42.66 (–£343.32 to £428.64)	£131.81 (–£247.77 to £511.40)	£131.81 (–£209.28 to £472.90)
3	Including extra mortality costs	732	£4103.96 (£4121.02)	865	£4047.89 (£3989.66)	£56.08 (–£343.70 to £455.85)	£124.98 (–£268.68 to £518.64)	£124.98 (–£230.84 to £480.80)
4	Including refresher GtACH tools and extra mortality costs	732	£4126.87 (£4127.10)	865	£4047.89 (£3989.66)	£78.98 (–£321.12 to £479.09)	£148.52 (–£245.40 to £542.45)	£148.52 (–£207.33 to £504.38)
Outcomes
5	DEMQOL-P-U-based QALYs	611	0.578 (0.24)	708	0.581 (0.235)	–0.003 (–0.028 to 0.023)	0.005 (–0.019 to 0.03)	0.005 (–0.018 to 0.029)
6	EQ-5D-5L-P-based QALYs	622	0.266 (0.317)	718	0.232 (0.291)	0.034 (0.002 to 0.067)	0.024 (0.004 to 0.044)	0.023 (0.003 to 0.043)
7	Falls	732	1.889 (3.662)	871	2.747 (7.414)	–0.858 (–1.417 to –0.299)	–0.568 (–0.97 to –0.166)	–0.574 (–0.961 to –0.186)
ICERs	Analysis
1/5	Base case, incremental cost per DEMQOL-P-U-based QALY					–£7226.47	£20,889.42	£20,557.80
2/5	With refresher GtACH tools, incremental cost per DEMQOL-P-U-based QALY					–£15,605.59	£25,433.80	£25,030.04
3/5	With extra mortality cost, incremental cost per DEMQOL-P-U-based QALY					–£20,513.22	£24,115.39	£23,732.56
4/5	With extra GtACH tools and mortality cost, incremental cost per DEMQOL-P-U-based QALY					–£28,892.34	£28,658.26	£28,203.32
1/6	Base case, incremental cost per EQ-5D-5L-P-based QALY					£575.01	£4543.69	£4651.63
2/6	With refresher GtACH tools, incremental cost per EQ-5D-5L-P-based QALY					£1241.73	£5532.14	£5663.56
3/6	With extra mortality cost, incremental cost per EQ-5D-5L-P-based QALY					£1632.22	£5245.37	£5369.98
4/6	With extra GtACH tools and mortality cost, incremental cost per EQ-5D-5L-P-based QALY					£2298.94	£6233.50	£6381.58
1/7	Base case, incremental cost per fall averted					£23.02	£190.62	£188.72
2/7	With refresher GtACH tools, incremental cost per fall averted					£49.72	£232.09	£229.77
3/7	With extra mortality cost, incremental cost per fall averted					£65.35	£220.06	£217.86
4/7	With extra GtACH tools and mortality cost, incremental cost per fall averted					£92.05	£261.52	£258.91

Point estimate, unadjusted, DEMQOL-based QALYs are –0.003 per care home resident. However, adjusting for baseline values and other predictors of QALYs yields an estimate of 0.005 attributable to the intervention (95% CI –0.019 to 0.030). The (adjusted) incremental gain in EuroQol-5 Dimensions (EQ-5D)-based QALYs was larger at 0.024, with the 95% CI excluding zero (0.004 to 0.044). Likewise, the incremental gain in the number of falls was –0.568 per person over the 12-month period (–0.970 to –0.166). Note that a negative value indicates fewer falls in the GtACH arm than in the control arm. As per the cost estimates, imputing missing values does not materially affect the results.

Depending on the costs included (i.e. extra mortality and refresher sessions), point estimate ICERs range between £20,889 and £28,658 per DEMQOL-based QALY gained, with a 53–57% probability that the ICER is below £20,000 (Figures 10 and 11), and between £4544 and £6234 per EQ-5D-based QALY gained, with an 88.8–91.6% probability of being below £20,000 per QALY (Figures 12 and 13). The incremental cost per fall prevented is between £190 and £262, with a 98.6% probability of being cost-effective as long as the willingness to pay to avoid a fall is above £2000 (Figures 14 and 15).

FIGURE 10.

Scatterplot of DEMQOL-P-U-based QALYs: base-case costs.

FIGURE 11.

The CEAC of DEMQOL-P-U-based QALYs: base-case costs.

FIGURE 12.

Scatterplot of EQ-5D-5L-P-based QALYs: base-case costs.

FIGURE 13.

The CEAC of EQ-5D-5L-P-based QALYs: base-case costs.

FIGURE 14.

Scatterplot of falls: base-case costs. Note that negative falls = the GtACH programme prevents falls.

FIGURE 15.

The CEAC of falls: base-case costs. Note that negative falls = the GtACH programme prevents falls.

Discussion

Conclusion

The evidence suggests that, depending on the choice of HRQoL measure, the GtACH programme was either of borderline cost-effectiveness or well within conventional thresholds of cost-effectiveness. Implementation of the GtACH programme was associated with a reduction in falls in care homes and improvements in EQ-5D-based QALYs. Future research could explore the relative validity and appropriateness of different health-related QoL measures in care home residents.

Introduction

Background

Care homes pose a distinct challenge for the introduction of complex interventions: they vary in size, funding, workforce and culture, and house vulnerable individuals with far-reaching health and social care needs. This heterogeneity of organisational context and uncertainty of individual need is an inherent (and unavoidable) barrier to effective innovation. 24,66–70 Although the delivery of programmes such as the GtACH programme is intended to be consistent, with justifiable variation only, it may be that the needs of local residents and the preference of local staff create situated and specific variations in how programmes are delivered.

Our starting point in this (as with all realist evaluation) was to consider how the GtACH programme was intended to work. In the realist tenet, programmes are not simply treatments or interventions, but rather ‘Programmes are “theories incarnate”. Every programme has a theoretical underpinning, whether it is made explicit or not’. 65 A programme such as GtACH rests on some theorised causal relationship that has a broader reach and application than its specific components; previous realist research illuminates the type of programme theories that might underpin initiatives in care homes. 71

In a recent realist review,24 three broad programme theories were recognised in the delivery of health care to care home residents. Incentives, targets and sanctions might motivate GPs to engage more routinely in the delivery of health care to this group; greater involvement of experts might make for more appropriate provision of elderly health care; and health outcomes might be improved by better relational working that spans care home staff and external health professionals. (Relational working may already exist; this is more about improving relational working.)

More recent research by the same authors66 identified similar programme theories. Improved relational working might underpin better outcomes in the delivery of health care; dedicated (financial) investment can trigger more appropriate provision; and wrap-around care, manifest in referral networks for external services, can support care home staff in accessing appropriate specialist care. The benefit of dementia-specialist services is a final focus for improved care home health care.

The PEACH (Proactive hEAlthcare of older people in Care Homes) programme67 looked to programme theories derived from the quality improvement collaborative literature to examine the implementation of comprehensive geriatric assessment (CGA) in care homes.

The FIRE (Facilitating Implementation of Research Evidence) evaluation68 speculated that the implementation of incontinence recommendations is mediated by organisational context, a pertinent theory of action, and staff support for change.

The GtACH programme had its own programme theories (see Chapter 5, Initial programme theories), but these examples demonstrate how broader theories and causal relationships might underpin the specific components of an intervention. They highlight that it is not (simply) the adequacy of the incontinence recommendations, the appropriateness of the CGA approach or the health care delivered, but rather that it might be incentives, better relational working,24,66 sharing best practice67 or the fit with organisational context68 that govern the impact of an intervention.

Methods

Results

Discussion

Introduction

Patient and public involvement was embedded in the FinCH trial to enhance the design, conduct and dissemination of the trial, and future implementation findings. The PPI team was instrumental in securing funding, influencing the trial set-up and advocating for care home residents throughout the trial. An adaptation of the research cycle (see Appendix 9, Figure 17) was examined to plan PPI involvement in each stage of the study, as advocated by INVOLVE. 82 The aim was to ensure that the trial was relevant to care home residents and care home interested parties, as well as to the public. The GRIPP2 (Guidance for Reporting Involvement of Patients and the Public, revised version) short form83 framework was used to ensure consistency. This approach captures the unique perspective of patients and the public experience at each stage of the research cycle. This co-designed model with people skilled and willing to enhance the FinCH trial had a PPI budget of £21,252.00. This chapter first describes and then explores the acceptability and appropriateness of the model.

Methods

Description of the hub-and-spoke approach

A hub-and-spoke organisational approach was used,84 in which there is a central anchor (hub) and spokes located at trial sites (Figure 16). The success of this approach is evidenced in health-care practice and services. 85,86

FIGURE 16.

Hub-and-spoke approach. Orange indicates subgroup.

Hub role in the FinCH trial

A job description focusing on this leadership role was produced and research networks were approached to identify interested people with the appropriate skills and training. A PPI member who had previous experience of being a carer for a care home resident and experience of a range of research projects, including RCTs, expressed an interest in the hub role. They were appropriately trained and had leadership experience. They contributed to the monthly TMG meetings to oversee the conduct of the study from a PPI perspective and to represent the broader PPI team. This role was supported and managed by a named PPI researcher. Specific support for the role included communication throughout the trial by e-mail, by telephone, at one-one meetings (both formal and informal), and through open-door access to the wider trial team (including the chief investigator and a trial administrator who provided administrative support, including the processing of expenses) to foster professional relationships.

Spoke role in the FinCH trial

The research team recruited ‘spoke’ members. The PPI researcher asked the FinCH trial networks and PIs to identify individuals who were willing to contribute their local trial perspectives for a minimum of 1 hour bimonthly. The PPI lead worked with the PPI researcher to recruit four spoke members with a range of experience. The majority had worked on assessing grant proposals and had PPI roles on research committees. A retired care home manager and nurse (female); a retired medic with experience in a caring role (male); a carer who had previous experience of caring for a partner with dementia (male); and a patient research ambassador (female) who was a lay chairperson for NIHR joined the FinCH trial team. The PPI researcher completed a telephone interview with these people and appropriate training was provided in line with the national standards for public involvement. 87

Patient and public attendance at FinCH trial meetings

Annual investigators’ meetings were organised as a communication mechanism to engage all trial team members, including the PPI team, by providing an opportunity to interact and connect with the national trial research team, keynote speakers and invited guests. These events included research methods training on topics such as gaining consent and writing an abstract.

Monitoring the GtACH programme’s delivery

Two members of the PPI team observed the GtACH training in care homes and fed back their perspectives to the process evaluation team in the form of written and vocal reports, and to the trial team through the TMG.

Undertaking data analysis for the process evaluation

The PPI team participated in the qualitative data analysis with the trial process evaluation team. The PPI members were asked for their perspectives on emerging themes to broaden data validation. These perspectives were then embedded in the process evaluation findings. The lead PPI member received training organised at the University of Nottingham (Nottingham, UK). Spoke members received training materials by e-mail/post and received support from a qualitative research expert (JD) by telephone and e-mail. All five members contributed to the data analysis.

Patient and public contribution to dissemination of results

The PPI team attended the investigators’ meeting to reflect on the results and celebrate successful completion, and supported workshops on dissemination and ways to add impact. The PPI team contributed to resources to ensure that plain, accessible language was used when the key findings were relayed to forums, for example at ENRICH, a research-ready care home network. The Plain English summary for the final report was written by the hub lead and this chapter was co-written with the PPI members.

Findings and emergent themes

Discussion

What this trial shows

Care homes in which the GtACH programme was implemented experienced significantly lower falls rates than care homes in which the programme was not used. The primary RCT outcome result showed an unadjusted incidence rate ratio (IRR) of 0.57 (95% CI 0.45 to 0.71; p < 0.01) in favour of the GtACH programme. The falls rate over this period was 6.0 per 1000 residents in the GtACH arm and 10.4 per 1000 residents in the control arm. This translates to a falls rate per participant per year of 2.2 for the GtACH arm and 3.8 for the control arm. The secondary RCT results saw a significantly lower falls rate in the GtACH arm for the 1- to 3-month period, but not in the 7- to 9-month or 10- to 12-month periods. There were no differences between arms in any of the other secondary outcomes. In the base-case analysis, the mean cost per resident was £3955 in the GtACH arm and £3935 in the control arm, with an adjusted mean difference in cost of £108 (95% CI –£271.06 to £487.58) representing the mean additional cost per resident in the GtACH arm. 96

In the base case, the DEMQOL-based QALYs were 0.578 in the GtACH arm and 0.581 in the control arm, with incremental QALYs of 0.005 (95% CI –0.019 to 0.03). The EQ-5D-5L-based QALYs were 0.266 and 0.232 for the GtACH and control arms, respectively, with incremental QALYs of 0.024 (95% CI 0.004 to 0.044). The incremental cost per DEMQOL-based QALY was £20,889 and the incremental cost per EQ-5D-5L-based QALY was £4544. The base-case incremental cost per fall averted was £191.

The GtACH programme was feasible: 69% of care homes achieved acceptable levels of staff training, with 80% of staff who worked in a caring role being trained. The programme was also likely to be affordable: the mean cost per resident was £88. The process study showed that the GtACH programme was widely accepted by stakeholders. Care home staff valued the way that the GtACH programme helped them address specific falls risks directly, and the emphasis on training and peer support. They also valued the fact that different elements of the GtACH programme could work independently of each other and be of benefit to the care home residents. The improved knowledge acquired through training or the increased awareness of falls gained from taking part in the GtACH programme did not always trigger completion of the GtACH tool. However, action was taken to limit falls risks, which is the likely mechanism for the reduction in falls rates.

Comparison with other studies

Prior to the FinCH trial, there was an indication in the Cochrane care home and falls prevention review19 that multifactorial interventions could prevent falls, but the evidence for effectiveness in care homes was limited and of low quality. FinCH was the largest care home RCT completed to date to evaluate a multifactorial falls prevention intervention and it has added a large number of data to the evidence base (increasing it by over 50%).

The Cochrane review19 recommended further research into falls in care homes, with an emphasis on evaluating an individualised, standardised approach to the delivery of interventions that are adequately described and, therefore, easily replicated or implemented, and on using a mixed-methods approach. The FinCH trial met these recommendations by using a published and standardised falls prevention programme, successfully replicating delivery across 87 care homes (control homes were given the training at the end of the study), as well as measuring effectiveness through a RCT that followed CONSORT guidelines; a process evaluation; an economic evaluation; and resident, care home staff and public collaboration.

In closer comparison to the existing evidence, the Cochrane review19 identified 13 studies that trialled multifactorial interventions in care homes. This included studies ranging from 31 to 682 participants. The FinCH trial was considerably larger, with 1657 recruited participants. The FinCH trial reported the rate of falls in the same way as the Cochrane review19 by measuring the total number of falls per unit of person-time. The studies reported a range of falls rates, from 1.7 falls per person per year to 2.51 falls per person per year. 19 This FinCH trial found slightly higher falls rates, with the GtACH arm having a falls rate of 2.2 falls per participant per year and the control arm a falls rate of 3.8 falls per participant per year. These falls rates were similar to those seen in the FiCH feasibility study,26 and could indicate that residents in care homes are falling more than in previous years, that care home residents in the UK are more likely to fall than those in Canada or the USA (where the Kennedy97 and Rubenstein98 studies were conducted) or that some intrinsic selection factors recruited those residents who were at a greater risk of falls.

To be able to compare the results of the FinCH trial to those of other studies, it is useful to compare the populations on a range of outcomes. Hospital admission rates usually indicate that a person is unwell and, although it is a blunt measure, it can be objectively measured and is needed for health economic evaluations. The mean number of hospital inpatient days for participants in the FinCH trial was 1.8 days per year, which is lower than that reported by Gordon et al. ,7 who found that a cohort of 227 participants had a mean of 2.2 days in hospital. This could indicate that the FinCH trial participants were more stable than those recruited by Gordon et al. 7 or that interventions undertaken across the UK to reduce the number of admissions of care home residents to hospital may be working.

The Barthel Index was used in the FinCH trial to measure participants’ ADL ability and although there was no difference between arms, the mean score of 8 out of 20 points indicates that most of the population could walk around the home with the use of aids, could feed themselves and were continent. Another large care home study of a rehabilitation intervention, the Occupational Therapy in Care Homes Trial (OTCH),99 which recruited 1042 residents with stroke from 227 care homes, found that 70% of the participants recruited were classified as severely limited on the Barthel Index at baseline, with 50% in the very severe (0–4) category. However, OTCH was primarily focused on patients with stroke-related disability, with most of the patient participants immobile and dependent, as less impaired patients with stroke are discharged home. The FinCH trial, meanwhile, focused on residents at risk of falls who were, by definition, ambulant, with this being the likely explanation for the difference between the cohorts recruited for these two large studies.

Strengths and limitations

To our knowledge, this is the largest RCT of a falls prevention intervention in care homes in the UK to date. The study was well powered to detect meaningful differences in falls rates, even when allowing for inflation for the power calculation and adjustment of recruitment targets during the study. We used clinically plausible outcome measures that were likely to be affected by the GtACH programme. We adhered to best practice for resident and public involvement, randomisation, allocation, outcome assessment and analysis. We took overt steps to reduce contamination bias. 31 RAs collecting outcome data were blinded and less than one-third of residents (29%) were accidentally unblinded as the study progressed.

A potential limitation was that the care home staff and participants were not blind to allocation owing to the nature of the intervention.

As with all falls trials, a limitation is the possibility of falls ascertainment bias: care home staff may have been sensitised to falls and have been more likely to record them in the GtACH arm because of the awareness-raising aspect of the GtACH programme. We believe that we minimised this risk by recruiting only those care homes in which there was a well-established falls recording system in place before randomisation. It is not clear if such bias occurred or, if it did, what effect it may have had on our results. We assume that if there was an effect, it would have increased falls reporting in the homes allocated to the GtACH programme and thereby attenuated our observed effect size. For this reason, we believe that our primary study’s positive finding is unlikely to be due to bias.

Another potential limitation was that the two trial arms were not completely balanced at baseline with regards to the mean falls rate. However, as the analysis included adjusting for baseline falls rate, we do not feel that this confounds the results.

The care homes we recruited were generally representative of UK care homes and, hence, our findings are likely to be generalisable within the UK and settings with similarly sized and run care homes. Ten sites in England took part, spread across urban, suburban and rural locations, and 87 out of the 186 homes that we approached took part. However, we acknowledge that the impact of the GtACH programme may differ in other countries and settings, such as nursing homes where medical and therapy staff are on site and regularly contribute to care. In such settings, the GtACH programme may contribute little, but it is likely that some kind of structured and evidence-based approach to falls management will be associated with a reduction in the incidence of falls. Although the GtACH programme was implemented across all residents in an allocated care home, we were not able to recruit all residents in all homes. Residents who lacked the mental capacity to provide consent and for whom no suitable consultee was available were a particularly under-recruited group, as with other care home studies. 100 However, we have no reason to think that the programme would be more or less effective in the residents we were unable to recruit.

A limitation of the GtACH programme itself is that it served as a prompt to care home staff, rather than the wider health-care team. We did not specifically collect data on lying and standing blood pressures, nor did we collate data on medication changes during the study in a way that would have enabled us to understand the effect of the intervention on polypharmacy. There are, however, prompts in the GtACH programme to ensure that care home staff trigger a medical assessment of residents who fell and we did not detect any evidence that staff were unable to follow this advice or trigger such assessment. If such assessment was not triggered, then this represents an opportunity to further improve the treatment effect seen with the GtACH programme and we will focus on this issue in future research studies.

A further issue is that because we did not collect data on lying and standing blood pressures at baseline, our intervention and control arms could have been imbalanced with regard to these risk factors. We hope that randomisation will have dealt with this issue. There was no evidence that the prescription of medications was different between intervention and control arms.

Another strength of this study is that our process evaluation was undertaken by a research team that was independent of the RCT team. This not only helped to illuminate the mechanisms by which the intervention led to the improved outcomes, but also identified where the intervention’s implementation could be improved. The findings of the process evaluation will be valuable in optimising the subsequent adoption of this intervention beyond this trial setting.

We recognise that measuring QoL in care home residents sufficiently for economic analyses is challenging, largely because of the high prevalence of cognitive impairment in this population that makes it difficult for them to complete assessment schedules asking questions about abstract concepts such as QoL. For this reason, we chose two approaches to measuring QALYs: the widely used EQ-5D-5L and the more recently developed DEMQOL. Our estimates of cost utility differed between these two approaches – using the EQ-5D-5L, the GtACH programme was conventionally cost-effective, but using the DEMQOL it was only of borderline cost-effectiveness. When we finalised our analysis plan, we chose the DEMQOL over the EQ-5D-5L, although this decision was finely balanced. Since then, limitations of the DEMQOL have led to the development of a care home-specific version, the DEMQOL-CH (Dementia Quality of Life Care Home version). Although there are limitations of using the EQ-5D-5L in this setting, it has been consistently evaluated better than other QoL measures in care homes. 59 Given the uncertainty of the methods available to assess cost utility in care home residents, while we cannot conclude that the GtACH programme was unequivocally cost-effective, we have presented evidence that suggests it is soundly cost-effective when using the EQ-5D-5L, but of borderline cost-effectiveness when using the (original) DEMQOL. We would argue that decision-makers should not rely solely on one effectiveness or economic statistic to make funding decisions, and that the overall evidence presented here supports a decision to implement the GtACH programme more widely.

Adoption and implementation

Given our findings, we believe that further trials of falls prevention interventions versus usual care in care homes are no longer required or justified, although this and any other trials recently and yet to be completed should be added to existing meta-analyses. We believe that it is now important to put these findings into practice as widely and swiftly as possible so that the GtACH programme, or other programmes derived from it, become part of usual care in care homes. The GtACH programme was extensively developed and delivered before being evaluated in this trial and, hence, is inherently designed to be suitable for implementation. Our process evaluation sheds a light on how it could be implemented even more effectively, for example through greater engagement with care home provider organisations to encourage the adaptation of the GtACH programme and documentation to the systems and processes of different care homes.

There are very few published papers of care home research and implementation. A recent rapid review and consensus study23 around implementation in care homes recommended consistent, regular, fluid conversations across diverse care home settings, paying close attention to understanding when care homes are ready for change and what measures have to be taken to enable them to be ready. An increasing body of evidence suggests that implementation and improvement in care homes must be led by care home staff, with NHS colleagues playing a facilitatory role, with the aim being to support the development of improvement capacity in care homes in parallel with implementation. 23

Stacey et al. 101 looked to determine the use of patient decision aids in clinical practice following RCTs. Only 44% of trial authors indicated some level of subsequent use of patient decision aids following trials. In 2019, Douglas and Affoo102 explored the difficulty of translating evidence-based health-care innovations for clinical practice settings (skilled nursing facilities), and examined the barriers to and facilitators of implementation, finding that the engagement of managers was crucial and time for investment was needed.

A key issue for implementation of the GtACH programme is if the short-term effect on reducing falls rate could be extended to reducing falls rates over a longer term or if the short-term effect of a relatively inexpensive intervention is enough to persuade people to implement the GtACH programme. It is important to consider the fact that the reduction in falls rates was not maintained beyond 6 months after randomisation. The process evaluation findings suggested that the training was beneficial, increasing staff knowledge and awareness of falls risks and providing the skills to reduce these risks, but it also indicated that one component of the programme (i.e. the paper assessment) was not completed as regularly as expected. This appeared to be because of a number of barriers, including care staff not being allowed to write in care records and a lack of ongoing support and training from the GtACH programme trainers. By not reinforcing the learning through physically completing the paper assessment, the longer-term implementation of the GtACH programme may have been lost. Investigating ways to better embed learning into longer-term practice will be an important component of follow-on work.

The study identified a number of contextual features that are pertinent to how the GtACH programme is received and used. Existing falls management processes, prior staff training in falls, strongly demarcated staff roles, external management systems and the high proportion of residents with dementia might have an impact on how the GtACH programme is implemented. Subsequent long-term implementation of the GtACH programme will need to address these issues, but not through a series of external interventions by NHS staff. Rather, in keeping with care home implementation research, they can only be addressed if care home staff see value in doing so. Dissemination materials must look at packaging the important learning from the FinCH trial in a way that is accessible and immediately useable for care home staff, and in a way that the work can be seen to align with their organisation and personal priorities, as well as achieving immediate, recognisable results. Normalisation process theory103 suggests that minimising additional work is essential for implementation. How this is done depends on leadership by the care home sector, as it is care home staff who understand their own organisational processes and routines.

Essentially, if the GtACH programme is to be implemented, the following actions should be considered:

• The care home community should be supported to lead the implementation process and any implementation research. The GtACH programme had a greater impact when it was championed by local staff (formally or informally).

• Information aimed at care home owners, managers, staff and residents should highlight the ways in which the benefits of the GtACH programme align with individual and organisational priorities.

• The context of the care home should be assessed in a sensitive way to understand any barriers to training or use of the GtACH programme. For example, in care homes that were willing to be recruited to the study, care home staff indicated a willingness to take part in the research, attend training, explore quality improvement concepts and assist with data collection. It may be that these homes had a different context to homes that did not agree to take part in the research study. In addition, when no prior training had been received or where training had been solely managed internally, care home staff were more keen to adopt the GtACH programme. When not all care home staff were allowed to write in care records, the GtACH paper assessments were not completed as regularly as in other homes. But, again, the emphasis should be on enabling care home managers and staff to assess and respond to their own organisational contexts.

• The role for NHS staff would be championship, a long-established approach for work in care homes. NHS falls leads may be well placed to do this, but will be particularly enabled when care home work is recognised and protected in their job plans.

• The GtACH programme consists of awareness, training, a tool and action; although we acknowledge that its elements may be independently important, it is the combination of the components that makes it a programme. The evidence suggests that the only justifiable conclusion based on the FinCH trial is that the GtACH programme should be delivered as a whole to gain similar results.

• The strongest impact of the GtACH programme was seen following its introduction, when it remained a novelty and was a focus of attention. Beyond this period, existing systems may re-emerge as the normal way of working. For future implementation, this means that the distinctiveness and benefit of the GtACH programme (beyond standard falls management) need to be better communicated to encourage stronger engagement.

There are some important areas for further research to be conducted in tandem with implementation. The GtACH training increased knowledge and awareness in the short term; however, to maintain knowledge, we know that learning is better established when it is enacted on a regular basis, but we don’t know how to reinforce such activities. The GtACH programme content may need to be provided in different formats, as some homes use electronic records only and some use paper only. Online, electronic or smartphone digital platforms that host the components of the GtACH programme may need to be developed. For future implementation, the role of the GtACH programme’s associated paper resources should be considered – it may be that the GtACH programme can make a difference without all staff completing dedicated paperwork. Alternatively, it could be that revised or simplified versions of the GtACH programme are created for less experienced staff to complete. Given the above points about minimising effort to support implementation, minimising the labour associated with the GtACH programme will be an important consideration. In homes with a high number of residents with dementia, falls were considered inevitable and initiatives such as the GtACH programme were considered ineffectual and not worth implementing; therefore, future implementation requires appropriate adaptation of the GtACH programme for residents with dementia or clearer communication of the pertinence/application of the GtACH programme to residents with dementia. Targeted, co-designed work with care home staff would be a powerful means by which to achieve this.

Contributions of authors

Philippa A Logan (https://orcid.org/0000-0002-8585-478X) (Professor of Rehabilitation) was the chief investigator from the concept of the study through to lead authorship of the final report.

Jane C Horne (https://orcid.org/0000-0003-0985-8110) (Senior Research Fellow and Occupational Therapist) was the intervention trial manager, providing clinical advice and support to the NHS falls leads delivering the intervention. She also co-trained the NHS falls leads (therapists), managed the PPI members, co-authored Chapter 6 and contributed important intellectual content to the report.

Frances Allen (https://orcid.org/0000-0001-7265-0676) (Research Fellow) jointly analysed qualitative interviews for process evaluation, jointly prepared the qualitative results for publication and contributed intellectual content to the report.

Sarah J Armstrong (https://orcid.org/0000-0002-6789-6169) (Professor of Medical Statistics) oversaw the statistical analysis of the trial, jointly prepared the statistical results for publication, contributed to the development of the trial application and trial protocol, and contributed important intellectual content to the report.

Allan B Clark (https://orcid.org/0000-0003-2965-8941) (Senior Statistician) jointly prepared the statistical plan, analysis and results for publication; advised the oversight committees on statistical procedures and analysis; and contributed important intellectual content to the report.

Simon Conroy (https://orcid.org/0000-0002-4306-6064) (Professor of Geriatric Medicine, Geriatrician) contributed to the grant application and protocol, and contributed important intellectual content to the report.

Janet Darby (https://orcid.org/0000-0002-6702-6956) (Research Fellow) jointly conducted the qualitative interviews and analysis for the process evaluation; jointly prepared the qualitative results for publication; and contributed to the grant application and trial protocol, and contributed intellectual content to the report.

Chris Fox (https://orcid.org/0000-0001-9480-5704) (Professor of Clinical Psychiatry) contributed to the study design and interpretation of data, and contributed intellectual content to the final report.

John RF Gladman (https://orcid.org/0000-0002-8506-7786) (Professor of the Medicine of Older People) contributed to the grant application and protocol, and contributed important intellectual content to the report.

Maureen Godfrey (https://orcid.org/0000-0001-8081-2188) (Patient and Public Representative) contributed to the grant application and protocol, observed the intervention, contributed to the qualitative data analysis, co-authored Chapter 6 and wrote the Plain English summary for the final report.

Adam L Gordon (https://orcid.org/0000-0003-1676-9853) (Professor of the Care of Older People) contributed to the study protocol and contributed important intellectual content to the report.

Lisa Irvine (https://orcid.org/0000-0003-1936-3584) (Research Fellow) jointly conducted the health economics analysis and contributed to the preparation of the health economics results for publication.

Paul Leighton (https://orcid.org/0000-0001-5208-0274) (Associate Professor of Applied Health Services Research) conducted the process evaluation and contributed to the data analysis, jointly prepared the qualitative results for publication, contributed to the trial protocol and contributed important intellectual content to the report.

Karen McCartney (https://orcid.org/0000-0003-2027-6031) (Research Fellow) prepared the protocol for publication and contributed to the preparation of the final report.

Gail Mountain (https://orcid.org/0000-0002-5417-7691) (Professor of Applied Dementia Research) contributed to the study protocol and contributed important intellectual content to the final report.

Kate Robertson (https://orcid.org/0000-0002-0922-3126) (Consultant Occupational Therapist and Honorary Assistant Professor) contributed to the grant application and study protocol, designed and jointly delivered training to the fall leads (therapists) and contributed important intellectual content to the final report.

Katie Robinson (https://orcid.org/0000-0003-1458-8186) (AHP Clinical Academic Lead) contributed to the study protocol and contributed important intellectual content to the final report.

Tracey H Sach (https://orcid.org/0000-0002-8098-9220) (Professor of Health Economics) jointly conducted the health economics analysis and prepared the health economics results for publication, contributed to the development of the grant application and trial protocol, and contributed important intellectual content to the final report.

Susan Stirling (https://orcid.org/0000-0001-6663-7846) (Statistician) jointly conducted the statistical analysis of the trial, prepared the statistical results for publication and contributed important intellectual content to the report.

Edward CF Wilson (https://orcid.org/0000-0002-8369-1577) (Senior Lecturer in Health Economics) jointly conducted the health economics analysis, prepared the health economics results for publication and contributed important intellectual content to the final report.

Erika J Sims (https://orcid.org/0000-0002-7898-0331) (Operations Manager at NCTU) was the Trial Manager, contributed to the development of the trial protocol, led the trial delivery and contributed important intellectual content to the final report.

Publications

Horne JC, Darby J, Godfrey M, Riley P, Jabber R, Stockton B, et al. Using a Hub and Spoke Approach to Enhance a Trial. British Geriatric Society Spring Conference, Nottingham, 2016.

Horne JC, Darby J, Godfrey M, Riley P, Jabber R, Stockton B, et al. Using a Hub and Spoke Approach to Enhance a Trial. Public, Patient Involvement International Conference, Manchester, 2017.

Izza MAD, Lunt E, Gordon AL, Gladman JRF, Armstrong S, Logan P. Polypharmacy, benzodiazepines, and antidepressants, but not antipsychotics, are associated with increased falls risk in UK care home residents: a prospective multi-centre study. Eur Geriatr Med 2020;11:1043–50.

Robinson K, Allen F, Darby J, Fox C, Gordon AL, Horne P, et al. Contamination in complex healthcare trials: the falls in care homes (FinCH) study experience. BMC Med Res Methodol 2020;20:46.

Logan PA, Horne JC, Gladman JRF, Gordon AL, Sach T, Clark A, et al. Multifactorial falls prevention programme compared with usual care in UK care homes for older people: multicentre cluster randomised controlled trial with economic evaluation. BMJ 2021;7:e066991.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to available anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

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

Hip fractures

• S72.0 Fracture of neck of femur.

• S72.00 Fracture of unspecified part of neck of femur.

• S72.1 Pertrochanteric fracture.

• S72.10 Unspecified trochanteric fracture of femur.

• S72.2 Subtrochanteric fracture.

• S72.20 Not found in ICD10 (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision), but occurs seven times in data set.

Wrist fractures

• S52.5 Fracture of lower end of radius.

• S52.50 Unspecified fracture of the lower end of radius.

• S52.6 Fracture of lower end of both ulna and radius.

• S52.60 Unspecified fracture of lower end of ulna.

• S62.0 Fracture of navicular (scaphoid) bone of hand.

• S62.1 Fracture of other carpal bone.

• S62.4 Multiple fractures of carpal bones.

• S62.40 Not found in ICD10, but occurs once in data set.

• S62.8 Fracture of other and unspecified parts of wrist and hand.

• S62.80 Not found in ICD10, but occurs twice in data set.

Any fracture

• S02.0 Fracture of vault of skull.

• S02.00 Not found in ICD10, but occurs twice in data set.

• S02.1 Fracture of base of skull.

• S02.10 Unspecified fracture of base of skull.

• S02.2 Fracture of nasal bones.

• S02.20 Not found in ICD10, but occurs twice in data set.

• S02.4 Fracture of malar and maxillary bones.

• S02.40 Fracture of malar, maxillary and zygoma bones, unspecified.

• S02.5 Fracture of tooth (traumatic).

• S02.50 Not found in ICD10, but occurs once in data set.

• S02.6 Fracture of mandible.

• S02.7 Multiple fracture involving skull and facial bones.

• S02.8 Fractures of other skull and facial bones.

• S02.9 Fracture of skull and facial bones, part unspecified.

• S22.0 Fracture of thoracic vertebra.

• S22.00 Fracture of unspecified thoracic vertebra.

• S22.1 Multiple fractures of thoracic spine.

• S22.10 Not found in ICD10, but occurs once in data set.

• S22.2 Fracture of sternum.

• S22.3 Fracture of one rib.

• S22.30 Not found in ICD10, but occurs 11 times in data set.

• S22.4 Multiple fractures of ribs.

• S22.40 Not found in ICD10, but occurs 13 times in data set.

• S22.5 Flail chest.

• S22.8 Fracture of other parts of bony thorax.

• S22.9 Fracture of bony thorax, part unspecified.

• S32 Fracture of lumbar spine and pelvis.

• S32.0 Fracture of lumbar vertebrae.

• S32.00 Fracture of unspecified lumbar vertebra.

• S32.1 Fracture of sacrum.

• S32.2 Fracture of coccyx.

• S32.3 Fracture of ilium.

• S32.30 Unspecified fracture of ilium.

• S32.4 Fracture of acetabulum.

• S32.40 Unspecified fracture of acetabulum.

• S32.5 Fracture of pubis.

• S32.5 Not in ICD10, but occurs five times in data set.

• S32.50 Unspecified fracture of pubis.

• S32.7 Multiple fractures of lumbar spine and pelvis.

• S32.8 Fracture of other and unspecified parts of lumbar spine and pelvis.

• S42.0 Fracture of clavicle.

• S42.00 Fracture of unspecified part of clavicle.

• S42.1 Fracture of scapula.

• S42.2 Fracture of upper end of humerus.

• S42.20 Unspecified fracture of upper end of humerus.

• S42.3 Fracture of shaft of humerus.

• S42.30 Unspecified fracture of shaft of humerus.

• S42.30 Not in ICD10, but occurs three times in data set.

• S42.4 Fracture of lower end of humerus.

• S42.7 Multiple fractures of clavicle, scapula and humerus.

• S42.8 Fracture of other parts of shoulder and upper arm.

• S42.9 Fracture of shoulder girdle, part unspecified.

• S42.90 Fracture of unspecified shoulder girdle, part unspecified.

• S52.0 Fracture of upper end of ulna.

• S52.1 Fracture of upper end of radius.

• S52.2 Fracture of shaft of ulna.

• S52.3 Fracture of shaft of radius.

• S52.4 Fracture of shafts of both ulna and radius.

• S52.5 Fracture of lower end of radius.

• S52.6 Fracture of lower end of both ulna and radius.

• S52.7 Multiple fractures of forearm.

• S52.8 Fracture of other parts of forearm.

• S52.9 Fracture of forearm, part unspecified.

• S62.0 Fracture of navicular (scaphoid) bone of hand.

• S62.1 Fracture of other carpal bones.

• S62.2 Fracture of first metacarpal bone.

• S62.3 Fracture of other metacarpal bone.

• S62.30 Unspecified fracture of other metacarpal bone.

• S62.4 Multiple fractures of carpal bones.

• S62.40 Not in ICD10, but occurs once in data set.

• S62.5 Fracture of thumb.

• S62.50 Fracture of unspecified phalanx of thumb.

• S62.6 Fracture of other finger.

• S62.60 Fracture of unspecified phalanx of finger.

• S62.61 Displaced fracture of proximal phalanx of finger.

• S62.7 Multiple fracture of fingers.

• S62.8 Fracture of other and unspecified parts of wrist and hand.

• S62.80 Not in ICD10, but occurs twice in data set.

• S72.0 Fracture of neck of femur.

• S72.00 Fracture of unspecified part of neck of femur.

• S72.1 Pertrochanteric fracture.

• S72.10 Unspecified pertrochanteric fracture.

• S72.2 Subtrochanteric fracture.

• S72.20 Not in ICD10, but occurs seven times in data set.

• S72.3 Fracture of shaft of femur.

• S72.4 Fracture of lower end of femur.

• S72.40 Unspecified fracture of lower end of femur.

• S72.7 Multiple fractures of femur.

• S72.8 Fractures of other parts of femur.

• S72.9 Fracture of femur, part unspecified.

• S82.0 Fracture of patella.

• S82.00 Unspecified fracture of patella.

• S82.1 Fracture of upper end of tibia.

• S82.10 Unspecified fracture of upper end of tibia.

• S82.2 Fracture of shaft of tibia.

• S82.20 Unspecified fracture of shaft of tibia.

• S82.3 Fracture of lower end of tibia.

• S82.31 Torus fracture of lower end of tibia.

• S82.4 Fracture of fibula alone.

• S82.40 Unspecified fracture of shaft of fibula.

• S82.5 Fracture of medial malleolus.

• S82.6 Fracture of lateral malleolus.

• S82.7 Multiple fracture of lower leg.

• S82.8 Fractures of other parts of lower leg.

• S82.9 Fracture of lower leg, part unspecified.