Designing a nurse‐led assessment and care planning intervention to support frail older people in primary care: An e‐Delphi study

Abstract Aim To identify and establish expert consensus on important and feasible components of a nurse‐led, comprehensive geriatric assessment (CGA)‐based intervention for community‐dwelling older people who live with frailty. Design A three‐round modified e‐Delphi survey. Methods An expert panel of 33 UK specialist older people's, primary and community care nurses participated in the three‐round e‐Delphi survey over a 12‐month period in 2017–2018. Data from round 1 were analysed using content analysis. Descriptive statistics were used in the subsequent two rounds to demonstrate convergence of panel opinion and consensus. Results In round 1, experts proposed 30 CGA components that were combined with six additional components from a literature review and clustered into six domains. In round 2, components were rated for importance and feasibility. Rating scores for importance were high across all domains, with lower scores for feasibility. Round 3 revealed that 36 components achieved consensus on importance and 11 out of 36 components reached consensus on feasibility. Conclusion Based on expert panel opinion, the content of a nurse‐led CGA‐based intervention was established, with the aim of future feasibility testing in a randomized controlled trial. Impact This study provides feasible components of a CGA‐based intervention that can be implemented in clinical practice by nurses in partnership with older people who live with frailty. Following further testing and evaluation, the components have the potential to improve clinical outcomes, maximize independence and improve the quality of life for community‐dwelling frail older people.


| INTRODUC TI ON
The population is ageing and, although this is undoubtedly a success for improved public health and welfare leading to longer life expectancy, it brings with it the challenge of meeting the health and social care needs of higher numbers of older people. By 2050, 1 in 6 people in the world will be over age 65 (16%), up from 1 in 11 in 2019 (9%). In 2018, for the first time in history, persons aged 65 or above outnumbered children under 5 years of age globally. The number of persons aged 80 years or over is projected to triple, from 143 million in 2019 to 426 million in 2050 (United Nations, 2019). In the United Kingdom, remaining life expectancy at age 65 is currently 19.5 years; however, many people experience 10 years of diminished quality of life due predominantly to limiting disability and illness (Mortimer & Green, 2015). Much of this disability and loss of function can be attributed to the development of frailty.
Frailty is a clinical syndrome associated with ageing, which develops through cumulative cellular damage over the life course and leads to progressive disability and loss of independence (Clegg et al., 2013). Biomedical assessment of frailty focusses on the diagnosis and treatment of the clinical syndrome (Hoogendijk et al., 2019), however, this approach fails to capture individuals' differences and can cause clinicians to neglect peoples' abilities to participate in their own care and support (Rahman, 2018). The World Health Organisation advocates for an asset-based model of assessment and support including a holistic, multidimensional approach to managing frailty as a means of preserving function, personhood and independence (World Health Organization, 2015).
Assessment of frailty is mostly undertaken in acute hospitals using a Comprehensive Geriatric Assessment (CGA), led by a geriatrician (Clegg et al., 2013). This assessment and care planning process is acknowledged as the gold standard for the management and prevention of deterioration in frailty (Gladman, 2016). It is a multidimensional, interdisciplinary diagnostic process to determine the medical, psychological and functional capabilities of a frail older person to develop an individualized care plan for treatment and longterm follow-up in partnership with the patient and their families (Ellis et al., 2011). However, many older people who live with frailty do not access hospital services, and there is no evidence to indicate that the acute hospital CGA is immediately transferable to community or primary care delivery.
It is not clear whether clinicians in this community setting (including nurses) possess the specialist skills and knowledge to deliver CGA. In addition, concerns have been raised across Europe about the time taken to identify the frail population, conducting a CGA and the additional cost in time and resources to primary care (Shaw et al., 2018). Implementation of primary care frailty management can be problematic because primary care doctors may view frailty screening as a burden in an already challenging workload (Reeves et al., 2018). A holistic, flexible intervention is required that can be delivered by primary care professionals other than GPs and adapted for the individual and their needs. In addition, evaluation of the role of nurses in leading this care model is required. Beswick et al. (2008) found that the delivery of complex interventions (based on CGA) for older people at home reduces care home and hospital admissions and falls. A systematic review investigating the implementation of one primary care CGA-based approach noted a lack of an agreed implementation model and concerns of workforce capacity in primary care (Craig et al., 2015). Another review attempted to identify approaches to CGA in primary care and although there were several in existence, the authors highlighted the need for more research into what is feasible for large numbers of the population (Morley et al., 2017).

| Background
Professional organizations, such as the British Geriatrics Society (BGS), advocate for a multi-professional approach to supporting frail patients and point to evidence that nurses and allied health professionals can successfully lead and provide input into the assessment and care planning process (Schadewaldt et al., 2013). Although some authors have evaluated healthcare professionals' attitudes to frailty assessment and management and found more positive engagement among nurses than other clinicians (Moffatt et al., 2018), the nursing contribution to frailty management is poorly developed, with nurse-led approaches showing mixed outcomes (Bleijenberg et al., 2017;Schein et al., 2005;Taube et al., 2018).
Reviews of some of these studies report a lack of specialist older persons' knowledge and advanced assessment skills which affected delivery and fidelity to the intervention (Hertogh & Bastiaans, 2016;Hoogendijk et al., 2016). Some authors have suggested that primary care teams require the support of specialist services, such as geriatricians (Hertogh & Bastiaans, 2016), whereas others have employed nurses with advanced assessment and case management skills and reported more positive effects on outcomes (Kono et al., 2016;Rockwood et al., 2000). In addition to advanced clinical skills, several studies have highlighted the importance of a goal-orientated intervention focussing on person-centeredness and self-management. This approach should be built on a caring, supportive relationship between the nurse and patient (Imhof et al., 2012). Therefore, whilst a CGA-based intervention may be appropriate for primary care delivery, the practicalities of its implementation require further exploration, including which components can and should be led by nurses rather than doctors. This formed the focus of this Delphi study, which is part of a larger programme of research related to the development and implementation of a nurse-led, CGA-based intervention.

| Aim
This study aimed to identify and obtain expert consensus on important and feasible components of a nurse-led, CGA-based intervention for community dwelling older people who live with frailty.

| Design
The e-Delphi survey was conducted as the first phase of a mixedmethods feasibility study to develop and test a nurse-led assessment and care planning intervention for frail older people in primary care. A modified e-Delphi technique was used (Foth et al., 2016) with a literature review, expert opinion and achievement of prespecified levels of consensus (Keeney, Hasson, & McKenna, 2011).
An a priori definition of consensus was agreed by the research team (Jünger et al., 2017) and defined as 75% expert panel agreement that a component met the criteria of 'fairly important' or 'very important' and 'fairly feasible' or 'very feasible' was required at round 3 of the survey.
Methods and results are reported in line with the 'Guidance on Conducting and Reporting Delphi Studies' (CREDES) (Jünger et al., 2017), which promotes consistency and quality in conducting Delphi studies. Figure 1 summarizes the Delphi process. to provide rigour and transparency in methods, study procedures were planned in detail and piloted whenever possible.

| Participants
The research team aimed to recruit at least 20 expert panel members,

Results Analysis
Literature review results 6 components Round 1:

| Survey design and administration
Three members of the research teams designed the round 1 survey.
To assess ease of use and understanding of content, the survey was completed initially by two specialist community nurses. They sug- any missing components that could be included in the next round.
The round 3 survey listed the 36 components from round 2 with the aggregated results (frequency and percentage) for each component from round 2. Data were presented back to the panel along with the same rating scales so that panel members had the opportunity to re-rate based on the group response in round 2.

| Ethics considerations, confidentiality and data security
Ethics approval was granted by the University of Plymouth Faculty Research Ethics and Integrity Committee (Reference Number: 18/19-1027). The online platform used to administer the surveys has data security and privacy policies in place (SurveyMonkey, 2020). Information systems and technical infrastructure are hosted in accredited data centres. Access to technology resources is only permitted through secure connectivity and requires multi-factor authentication. All data are encrypted, and all responses to surveys are private by default.
Expert panel members were assured of anonymity in the participant information sheet. Internet protocol addresses were used to contact panel members who could not be identified in the process, and individual responses were unknown to other panel members.
Panel members were asked to sign an informed consent form including agreement to the use of email addresses to contact with subsequent survey rounds.

| Data analysis
Statistical analysis and definition of consensus were planned and agreed prior to data collection. Panel members' demographic characteristics were reported by frequencies and percentages.
Data from the open-ended question in round 1 were analysed using content analysis (Hasson et al., 2000). In round 2, frequencies and percentages for all ranking scales were calculated prior to being presented back to the panel in round 3. In addition to frequencies being derived, means and standard deviations for ranking scores were calculated to assess convergence of opinions from round 2 to round 3.

Final consensus figures (percentage consensus for each component)
were calculated for reporting after round 3.
There is no specific guidance available for acceptable response rates in Delphi studies. Some Delphi studies relating specifically to older people's care have not reported response rates (Goldberg et al., 2016;Mahoney et al., 2017). However, others have reported between 75% (Rodríguez-Mañas et al., 2013) and 92% (Jeffs et al., 2017). Because of the iterative process of Delphi studies, there is the potential for panel members to withdraw after subsequent rounds, which can lead to response bias if attrition is significant (Evans, 1997). Some authors recommend that a 70% response rate is necessary for each round to maintain rigour (Sumison, 1998). In this study, a response rate of 70% was anticipated to the rounds 2 and 3. To encourage consensus, three reminders to complete the survey were sent to the panel members.

| e-Delphi round 1
Content analysis generated an initial 35 components suggested by the expert panel. These were aggregated into 30 components and grouped into six domains, which were as follows:

| e-Delphi round 2
The mean of all expert panel scores was calculated for each component and then combined to give a mean score for each domain. When analysed in the six domains, mean scores for importance were high across all components and lower for feasibility except for the frameworks/care structures domain, which were high for both importance (mean 4.8; SD 0.06) and feasibility (mean 4.8; SD 0.06). All other domains had mean scores ranging

| e-Delphi round 3
Domain mean scores for both rounds are presented in Figure 2a

| Panel consensus
Following round 3, all 36 components met consensus on importance, but only 11 out of the 36 components reached consensus on feasibility at the pre-defined level of 75% panel agreement (Table 3).
In the frameworks/care structure domain, all components met the consensus threshold for importance (range 90.5%−100%), but four out of the six did not reach consensus on feasibility. These were a competent, well-trained workforce (47.6%), a system for data/information gathering (47.6%), a shared care record (57.8%) and a timely response to crisis (19.0%).
In the home/family/safety assessment domain, all components met the consensus threshold for importance (range 81.0%−100%), but five out of the seven did not reach consensus on feasibility.
In the domain of personalized care and support planning, all components met the consensus threshold for importance (range 85.7%−100%), but nine out of the eleven did not reach consensus on feasibility. These were formulating a personalized care and support plan (PCSP) (57.1%), documenting in a co-created PCSP (33.3%), monitoring response (42.9%), review of the PCSP (61.9%), empowerment and self-management (33.3%), determining advanced care preferences (71.4%), assessment of resilience and coping mechanisms (33.3%), escalation/contingency planning (61.9%) and establishing the narrative (52.4%).
All four components in the long-term condition management domain met the threshold for consensus on importance (range 90.5%−100%) while three of these did not achieve consensus on feasibility; advanced clinical assessment skills (51.2%) and problem/ deficit identification and optimizing long-term condition management (both 71.4%).
In the domain of physical health assessments, all components met the consensus threshold for importance (range 80.1%−100%), and two out of the six did not reach consensus on feasibility; assessment of vision, hearing and dentition (66.7%) and sexual health assessment (28.6%). Finally, in the domain of mental health assessments, both components of assessment of cognition and assessment of mood met the consensus threshold for importance (100%), but not for feasibility scoring 71.4% and 66.7%, respectively.  (Garrard et al., 2020;Phelan et al., 2007). The reality in primary care is still far from the strategic vision of integrated care for older people.

| DISCUSS ION
Other components that were thought not feasible relate to the possession of specific skills by primary care nurses. The panel thought nurses would not be able to assess carer's needs, conduct environmental assessments and determine preferred place of care.
They also doubted the feasibility of nurses with advanced assessment skills. This is borne out by the literature where some studies have reported a lack of specialist older people knowledge and skills amongst primary care nurses (Hertogh & Bastiaans, 2016;Hoogendijk, 2016).
Notwithstanding the concerns already discussed, it was encouraging to see that specific components from other domains were thought to be important and feasible in a nurse-led approach. These were establishing the diagnosis and severity of frailty and assessment of functional ability including re-ablement, falls risk, pain, medication adherence and optimization, nutritional status (including hydration) and bladder and bowel function. This is important information in informing the design of a nurse-led, CGA-based intervention as,  (Elkan et al., 2001). They also echo results of other studies that highlight the perceived challenges to the delivery of primary care-based CGA (Craig et al., 2015;Monteserin et al., 2010;. This may be an opportunity to examine which clinicians are best placed to provide care and support to frail older people in a way that can increase capacity in the primary/ community care team and provide a more convenient approach for patients who would struggle to attend secondary care. A new model may include the substitution of nurses where care and treatment has previously been provided by doctors, for example, as an alternative to geriatrician-led CGA. Ensuring the most appropriate clinician delivers care and support is an ongoing debate, with increasing acceptance that care for older people with complex needs can be led by nurses. Three systematic reviews have reported that care provided by nurses is of equal quality to care provided by primary care doctors (Horrocks et al., 2002;Laurant et al., 2005;Martínez-González et al., 2014). Recently, the BGS affirmed that nurses can lead CGA-based interventions in primary care as they are 'are well placed to manage the complexity of assessment in an efficient way drawing together the different strands to coordinate a personalized treatment plan' (Turner et al., 2019, p. 4). Turner et al. also emphasize that nurses have a duty to act as patient advocate set out in their codes of conduct and are expert in enabling shared decision-making. Given that CGA is a multi-dimensional assessment and care planning process, it has been advocated that multiple clinicians should be involved and that nurses should coordinate and lead the process ensuring best use of scarce resources and targeting of this approach at those who will most benefit (Schadewaldt et al., 2013).

| Study limitations
Older people and their carers were not included in the expert panel, and this was a limitation of the study as the expert panel did not necessarily reflect their views on what was important to them. To address this deficit, the Delphi findings were later shared with a research stakeholder group made up of older people, carers and clinicians to consult with them the results of this study and on the final content and delivery methods. Unfortunately, there are no qualitative data that can explain the reasons why participants felt that several components were not considered to be feasible and so were not included in the emergent nurse-led CGA model.

| CON CLUS ION
This e-Delphi study developed consensus on important and feasible components of a nurse-led, CGA-based intervention in primary care.
The study indicates which components of traditional CGA can be effectively delivered in primary care, by non-medical practitioners, as well as those elements which may not be feasible in practice. The intervention now requires further evaluation in real life clinical practice and will be tested in a feasibility randomized controlled trial in the next phase of this research.