Silver Book II: Frailty

Good practice guide
Good practices guides focus on providing information on a clinical topic.
(Section editor)
Ken Rockwood
Date Published:
22 February 2021
Last updated: 
22 February 2021

The Silver Book II was written to address the care needs of older people, specifically those living with frailty, during the first 72 hours of an urgent care episode. This chapter looks at the impact of frailty on the provision of care. 

Frailty is a condition characterised by loss of biological reserves, failure of physiological mechanisms and consequent increased risk of experiencing a range of adverse outcomes, including hospitalisation, longer length of inpatient stay, and delirium.5-7 

A key application of frailty as a method of identifying people at increased risk of adverse outcomes compared to others of the same chronological age is that it enables a shift towards providing care that is more person-centred and goal-orientated.8,9 Alongside risk of hospital admissions, the increased length of inpatient stay is recognised to drive much of the additional healthcare costs associated with frailty, which have been estimated at £5.9 billion annually in the UK.10
Two internationally established approaches to operationalising frailty are the phenotype model11 and the cumulative deficit model.12 Although there are some conceptual differences between the two models, each has been extensively validated in large epidemiological studies, with robust predictive validity.
The phenotype model identifies frailty on the basis of five variables: unintentional weight loss; self-reported exhaustion; low energy expenditure; slow gait speed; weak grip strength. People with three of the five factors are considered 'frail', one or two factors as 'pre-frail', and no factors as 'robust' older people. In the original community-based study, 7% of the over 65 population were identified as frail, 47% pre-frail and 46% not frail. The model has since been validated internationally in a large number of epidemiological studies, demonstrating robust predictive validity for outcomes.
Since original development, the phenotype model has remained more closely suited to research studies as the time required to translate the five variables into clinical practice may be prohibitive. A more feasible alternative might be using instruments based on single items of the phenotype model, for example gait speed, which retain good diagnostic test accuracy in community-dwelling older people,13 or related instruments based on self-report of these items. An aligned issue is the increasing use of the phenotype model in clinical trials of frailty interventions, both to select and stratify the population, and as an outcome measure. Although this ensures robust frailty identification, the future translation of research findings into clinical practice requires additional consideration.
There are particular concerns regarding use of the phenotype model in people who are acutely unwell, as components of the model (for example exhaustion, slow gait speed, weak grip strength) may conflate acute illness with frailty.14 An additional consideration is that the five variables of the phenotype model were fortuitously available and selected from a prospective cohort study that was not designed to investigate frailty. Other potentially important factors such as cognitive impairment, a highly prevalent condition associated with functional decline and disability, were not included as a component of the original model.15
The cumulative deficit model identifies frailty on the basis of the life course accumulation of a range of 'deficit' variables, which can be clinical signs, symptoms, diseases, disabilities and abnormal laboratory test values.12 Thus frailty can be considered pragmatically as the cumulative effect of individual deficits: the more individuals have wrong with them, the more likely they are to be frail.16 The model enables calculation of a frailty index (FI) score as an equally weighted proportion of the number of deficits present in an individual to the total possible in the model (e.g. if 9/36 deficits are present, the FI score = 0.25). The model is useful as it is very flexible - it has been established that a minimum of 30 deficits are required for a model to be valid.17
The statistical distribution of the FI (a gamma distribution) was consistent with a probability model that typically describes systems with inbuilt redundancy. This is attractive as it indicates that the FI has properties that support the central frailty concept of reduced physiological reserve. So, although each individual deficit carries no obvious or imminent threat (e.g. hearing impairment), the deficits cumulatively contribute to an increased risk of experiencing an adverse outcome.
The cumulative deficit model expresses the notion of a gradation of frailty. This is attractive clinically because it allows frailty to be considered as gradable rather than present/absent. Moreover, it is the number of equally weighted deficits, as a measure of accumulated vulnerability that is related to adverse outcomes rather than particular clusters of deficits.
Similar to the phenotype model, the cumulative deficit model as originally operationalised is more closely suited to research studies. However, the flexibility of the model has enabled it to be derived from routine primary care electronic health record (EHR) data, to produce an electronic frailty index (eFI) that automatically calculates an FI score based on EHR information that is already available.6 The eFI is constructed using 36 equally weighted deficit variables, and has good predictive validity for outcomes of nursing home admission, hospitalisation and mortality. This use of existing data to construct the eFI can remove the need for an initial resource intensive assessment to identify frailty, and has supported major NHS policy change through the routine identification and management of frailty in primary care. The eFI is freely available to every general practice in England, and 95% of all UK practices. An additional advantage is that it facilitates the selection and stratification of participants for clinical trials of interventions, and the subsequent translation of findings into clinical practice. A Hospital Frailty Risk Score developed recently uses ICD-10 codes to identify the risk of frailty in hospitalised patients, and this too is being rolled out in the UK.18
The phenotype and cumulative deficit models demonstrate overlap in identification of frailty19 and considerable statistical convergence, supporting recognition of the condition as a unified construct. The continuous FI demonstrated greater discriminatory ability for people with moderate and severe frailty when compared to the phenotype model.20 Use of continuous models may facilitate more precise identification.21
However defined, it is important that frailty does not become the polite face of ageism. For many common procedures, more frail patients do worse, but just how much worse in response to which specific treatments is the matter of considerable inquiry. In the UK, this has given rise to the Specialised Clinical Frailty Network (SCFN). The SCFN is investigating, for example, how the degree of frailty may affect aortic valve replacement, end stage renal disease and renal replacement therapy and use of some chemotherapeutic agents. Recent results from a study of transcatheter aortic valve replacement (TAVR) may offer some clues. There, people who had the least disability impairment did best, whereas those with more severe dependence did worse; response was otherwise guarded across the degree of disability. Whether this foretells responses in other areas remains to be investigated.

Frailty is a pressing issue at all levels of healthcare.22 

First, the essence of frailty is misunderstood. The background risk (the degree of frailty) goes unacknowledged, much less identified and addressed. Acknowledged or not, frailty, and its typical accompaniments of multiple, interacting medical and social problems that produce varying grades of disability, play essential roles in the delivery and outcomes of care. That the complexity of frailty mostly goes unrecognised, is as an unintended consequence of healthcare’s tremendous success from ever greater focus, manifested as diagnosis-based, deeply ingrained, hard-won specialist expertise.
Second, this privileging of a narrow, disease-based focus means that, the idea of the whole person is often seen at best as an issue for other health care providers, and at worst as 'social considerations' not legitimately within the purview of whichever subspecialty is treating their problem. In consequence, how to mitigate the risk associated with frailty is not yet part of the skill sets on offer routinely and at scale.
Third, and most important, especially in systems only coming late to recognising frailty - is that not only do we not routinely mitigate risk from frailty, but many routine health care practices unnecessarily increase it. For example, in the hospital setting it is common to see somewhat less than full vigilance in patients being able to sleep, to eat, to drink, to have intravenous fluid therapies that correspond to their age-related needs, to have pain treatment optimised and medications rationalised.
It is the system that must respond: this is not the work of a few people. Health care systems that have been successful in addressing these challenges – and no system yet can claim complete success - typically emphasise naming and staging frailty.23 Naming frailty without staging it runs the risk of having ‘frailism’ as the polite face of ageism. For example, the needs of people with mild frailty differ vastly from those who are very severely frail. For the former, the case can be made that what is most needed is better routine care. This includes an ability to respond to more than one problem at a time, to take a whole-person view of wellbeing and prevention24 and to improve the routines of hospital care. In contrast, people who are very severely frail – who spend much of their time in bed, with increasing dependence – typically need management suitable to people living in their last year of life.25 These competing needs are not easily reconciled in a single frailty pathway. In any case, every patient will need an individualised care plan. In these, although the degree of frailty is likely to be informative, it cannot be expected to be definitive. Even if care planning is approached in consideration of the degree of frailty, one solution will not fit every patient.
Work on frailty cannot occur in isolation. It must fit with a health care system’s strategic plan. Such plans typically recognise the need to deliver a high-quality and sustainable system. In this context, it is not difficult to cite addressing the complex challenge of frailty as a central means of meeting such an objective. Very commonly too, health care systems espouse the need to strengthen and support a high-performing workforce. Helping them understand frailty and incorporate it into care planning means supplying them with the tools that they need, which for an ageing population will include frailty assessment.
How best to improve care for people living with frailty (all at once, or incrementally) will vary by setting. Everywhere, finding an optimal means of ‘geriatricising’ routine care must be considered and evaluated. Even so, a critical variable will be cultural: how prepared is a system or institution to grapple with the concept of frailty? For example, UK proposals to focus on delivery of Comprehensive Geriatric Assessment within 72 hours of presenting in urgent care settings26 or on hospital-wide geriatric assessment27 do not seem as fantastical there as they would in any North American jurisdiction. Even in North America, system-wide changes delivered through locally contextualised interventions can be effective.28 As reforming care for people living with frailty contains an at least implicit critique of how care is provided now, acceptance is made easier by usable tools that are available on a cost-effective, real-time basis, and are easily communicated. The data arising from them must seamlessly link across elements of health care (public health, primary care, various types of community, prehospital, hospital and long-term care) with the clear aim of improving outcomes.
System change requires extensive discussion and consultation, balanced against the need to get started. If particular groups seem to have made some headway, they often can provide a focus for getting start. That is easiest for teams that might welcome evaluation with a view to more widespread adoption of the techniques which they have found to work in their community of practice. Even so, how to scale will be a function of ease of use, ability to reliably identify people living with frailty, and uptake. Commonly such an orientation – to build on local successes - favours an incremental approach. It also helps to de-risk decision-making, reducing the reluctance to change too much too soon, while reinforcing the knowledge that not changing at all is no longer an option.
Interventions are needed across health care settings. Both comprehensive29 and more local responses (such as the mobilisation initiatives being carried out across Ontario28) can be adopted. Proof of concept consists first in showing that an intervention can work, and then understanding both conditions that lead to success, and conditions in which both system and patient goals are not met. Controlled trials have some role in this, but their ability to exclude the very patients who need to be studied, and the infeasibility arising from time taken to devise protocols, get them funded and have them pass ethical review militate against an exclusive reliance upon them. Into this turbulent surf, Quality Improvement initiatives built on Plan-Do-Study-Act cycles have waded. Their advantages are many, but that does not mean they are without problems; here again, system and institutional cultures will be crucial to getting the best use from such efforts.30
As healthcare practitioners will know, change is difficult. They will also know that populations are ageing, and that what we have done for so long works less well now. A focus on frailty offers a comprehensible way to understand how this has come about, and how we can respond.

Giants are large and powerful. If they are to be confronted, they must be taken seriously. For this, recognition is an essential first step."

- Kenneth Rockwood (2016)

Older adults who live with frailty and become acutely ill are at risk of many adverse health outcomes.31 

Death is an obvious consideration, but for many older adults there are things worse than dying, including the loss of autonomy and of dignity associated with relying on others, especially for intimate care.32 Individual fears and life experiences vary, as do expectations of what families and health and social care providers can offer. True person-focused care requires knowledge of the person’s experiences, values, and preferences, especially when they arrive at a place where medical options ‘offer diverging paths, with different important consequences, and lasting implications.’33
Often the first decision an acutely ill older person must make is whether to go to the hospital – often by summoning an ambulance or by going to some sort of urgent care setting. Many healthcare providers and patients believe that ‘the hospital is best’ even for frail older adults who become unwell at home. This is often rooted in the belief that in addition to diagnostics and treatment, it offers them hope to improve, ‘to relieve symptoms, to survive or to achieve personal goals’. Often a consideration, even if less explicit is that it offers the caregiver a much-needed rest from responsibility and worry.34
A more recent alternative viewpoint is rooted in how often hospitalised older adults experience functional decline, or die or are newly admitted to institutional long term care.35 A key challenge, especially at the initial clinical encounter where prognosis is often the least certain, is how to balance the risk of a given course of action with the risk of doing otherwise. Often, once admitted to the acute care setting, decision making by older adults is virtually suspended.36 The healthcare team evaluate the older adult and gather information from their caregivers about how they were functioning, to determine the level of support needed to return home. Unique to the experience of the older adult is the question ‘are they safe to go home?' - the question brings with it the evaluation of risk. The adjudication of risk, and varying risk tolerance is commonly tricky; what can make that even more so is that often in healthcare, providers do not know how to disagree. In consequence, the default is often the position of the most risk averse person on the team, or the outspoken family member focused on the presumed physical safety of a care facility.37
Shared decision making needs to be routine and to balance the patients’ and carers’ points of view with more traditional ways of deciding on the goals of care. Key to shared decision making is that information be given, not withheld and delivered in a manner that is understandable to lay people. A popular approach suggests that choice be introduced, that options then be clearly described and that patients be aided in exploring preferences and making decisions. This model rests upon respecting ‘what matters most’ to patients as individuals, and that this exploration in turn depends on them developing informed preferences.38 Paramount to shared decision making is how the healthcare system is organised and what happens in the decision making interaction related to the imbalance of power in the doctor-patient relationship.36
To enhance patient-centred decision making, an educational workshop was developed to explore risk attitudes of healthcare teams towards older adults.39 It recognised that many older people prefer to live at risk rather than to be cared for in ways that threaten their independence. Beginning the decision making process with the older adult about their wishes and the risks, then negotiating the varying levels of risk with the family, was seen as key. As always, how individuals preferences are expressed requires negotiation, particularly if a given patient’s preference hinges upon someone who cannot provide the level of support expected of them or if their choice is financial unobtainable.
The preference to live at home and at risk is complicated by dementia, especially if the default is to assume that decision making capacity is impaired. That dementia has developed does not oblige removal from familiar and supported surroundings if the risk can be mitigated. Even so, not all substitute decision makers recognise a duty to enact what the patient wished to do when competent. A structured approach to discussing risk, adjudicated by an experienced person charged with balancing competing views can greatly enhance decision-making in ways that respect patient autonomy, while balancing threats of harm, and a duty for beneficence.
The challenge to shared decision making about admission, pits the risk inherent in frailty no matter what the choice, and the presumption that care outside a hospital is riskier than care within it. Navigating this requires a combination of being frank about the limitations of some remedies in the face of the daily reality that risk tolerance varies amongst decision makers.
Tools to help with shared decision making

Increasing numbers of older adults are presenting to urgent care services.

This population of older people is heterogeneous in the nature of the acute problems they present with, their health prior to becoming ill, and their social and economic backgrounds; interacting medical and social problems prove to be a challenge to ‘one-thing-at-a-time’ health care (see Figure 1). Recognising the common geriatric syndromes of delirium, falls, immobility, incontinence, and social abandonment in urgent care is feasible and can help non-specialists at the front door with more accurate risk-stratification, allowing more comfort in the key decision of admission or discharge.
The Clinical Frailty Scale (CFS) is recommended by NHS Improvement, NHS England, the Ambulatory Emergency Care Network and the Acute Frailty Network26 to identify the individual’s level of frailty two weeks prior to presentation. Knowing the state of health, function and support prior to the acute illness or recent decline also helps to guide goal setting: can the baseline state be recovered? How long will it take? Is it feasible that, if recovery occurs, a return to those living arrangements can be effected?
Urgent care providers are typically effective at routinely measuring the intensity of the body’s physiological decompensation as a result of a stressor using indices of acute illness severity. Baseline clinical frailty and acute illness severity may be additive in the prediction of mortality in those admitted.40
When the decision is to admit, higher levels of measured complexity can help channel the admission to inpatient areas where CGA is available. Older patients who receive inpatient CGA are more likely to be alive and in their own homes at follow-up,41 and may also report higher levels of patient satisfaction.42 Rapid CGA in emergency departments may be a good method to triage complex older patients for CGA hospital admission.43 Hospital systems need to take into account that the concentration of higher patient complexity in CGA areas requires a similar concentration of staff and resources (e.g. higher nursing, therapist and social work ratios than in non-CGA areas) to be able to effectively manage such complexity in an equitable fashion.

Feedback on this resource?

This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.