Development of FastFrail—a rapid frailty screening tool for medical calls: a development study based on cross-sectional data from an urgent care centre in Norway

Abstract

Objective

To develop a rapid screening tool for the identification of frailty in medical calls and other out-of-hospital acute care services.

Design

Development study based on cross-sectional data. A set of potential items were developed based on existing frailty tools and other relevant literature by a panel with geriatric and primary care expertise. The items and the Clinical Frailty Scale (CFS) were administered on a convenience sample of older urgent care patients. Further development of the tool was based on statistical analyses of this data material and final discussions in the panel.

Setting

Urgent care centre in Norway, data collected between January and August 2022.

Participants

All patients ≥70 years were eligible for inclusion, with the exception of patients triaged to the highest urgency level and patients not able to answer questions with no next of kin present.

Primary outcome

Potential items associated with frailty by CFS, measured by explained variance (adjusted R² values from linear regression analyses).

Results

Nine potential items were developed and administered on 200 patients (59% female), of whom 48% were 70–79 years, 38% were 80–89 years and 14% were ≥90 years. The median CFS score was 4 (living with very mild frailty). Receiving help weekly, being homebound and using a walking aid were identified as strong indicators of frailty (adjusted R² values 59%, 48% and 43%, respectively). Together these three factors could explain 74% of the variation in CFS scores.

Conclusions

Receiving help weekly, being homebound and using a walking aid are strong indicators of frailty among urgent care patients. We developed a frailty screening tool for medical calls—FastFrail—consisting of three simple, binary questions (yes/no) on these aspects. We hypothesise that FastFrail can supplement traditional symptom-based triage and enable more accurate assessment of older adults calling for acute medical help. We intend to test the tool in clinical practice.

Correspondence to Isabel Sebjørnsen; [email protected]

STRENGTHS AND LIMITATIONS OF THIS STUDY

• A thorough selection process was conducted by a panel with relevant expertise to identify a first set of potential items for a new frailty screening tool for medical calls, FastFrail, from existing tools and other relevant literature.

• The potential items were administered on an urgent care cohort and analysed for association with scores on the Clinical Frailty Scale (CFS).

• Basing the development of FastFrail on the CFS served as a strength, as the CFS has previously demonstrated good measurement properties and the ability to predict adverse outcomes in the acute care setting.

• However, frailty assessed by the CFS, as well as the inclusion process in the study, may have led to an overemphasis on physical functioning in our results at the expense of other important domains of frailty.

• Data were collected from one urgent care centre only and from patients physically present at the centre, limiting the generalisability and necessitating further research to determine measurement properties in the medical call setting.

Introduction

A significant proportion of older people is living with frailty.¹ Several definitions of frailty exist. Rockwood et al describe the concept as an age-related state of increased vulnerability and eventually decreased ability to perform daily activities due to accumulation of health deficits.² Frail individuals are prone to delays in diagnostics and treatment in the event of an acute illness or injury due to altered disease presentation, often with less pronounced organ-specific symptoms.³ Instead, people with frailty commonly present with non-specific complaints such as falls, delirium or general weakness. These patients are highly susceptible to adverse outcomes, even when affected by conditions that are usually mild and not associated with a high risk of complications.^{4 5} Altered presentation and adverse outcomes occur more frequently, even with very mild frailty, and become more pronounced as severity increases.^6–8 The needs of this growing patient group are not currently being met in acute care services.⁹

Triage is an essential part of acute care services, both out-of-hospital and in-hospital. The process implies prioritising patients based on the severity of their condition after a quick assessment of signs, symptoms and medical history.¹⁰ The aim is to rapidly identify patients with time-critical conditions for immediate treatment. Triage tools that have been assessed for performance are all found less accurate in older adults.^{10 11} The tools do not account (well enough) for how ageing and frailty can alter disease presentation, leading to the misinterpretation of condition severity in this population. Thus, older adults, particularly those living with frailty, are at risk of being assigned an inappropriate triage score.^{10 11} Overtriage prompts unnecessary use of resources and can lead to exposure to potential risks such as hospital-acquired infections and functional decline. Undertriage translates into longer waiting times and delayed treatment. In out-of-hospital services, patients may be inappropriately handled at a lower care level or advised to care for themselves.

The European Joint Action Initiative on Frailty (ADVANTAGE) recommends assessing frailty in all encounters between older adults and healthcare services.¹² For acute care services, combining traditional triage with frailty assessment has been proposed to improve triage decisions and outcomes for older adults.^{5 13–15} Early recognition of frailty can aid interpretation of signs and symptoms, guide clinical decision-making and identify patients for targeted interventions.¹⁶ Multiple frailty screening tools exist, and some have been tested alongside traditional triage tools in emergency departments (EDs), with promising results.¹⁷ A study from the Dutch ED setting published in 2020 found that adding frailty assessment to traditional triage increased the explained variance of 30-day mortality by 5.3% (1.0–6.3%).¹⁴ Ng et al demonstrated a significant improvement in the prediction of critical events among older adults visiting EDs in Taiwan when including the Clinical Frailty Scale (CFS) in the triage process (area under the curve 0.82 vs 0.64).¹⁸ The CFS is one of the most studied tools to assess frailty in the ED setting and is also recommended by the British Geriatric Society as a screening tool out-of-hospital.¹⁹ However, frailty screening is rarely performed, and little research has been conducted in the out-of-hospital acute care setting.²⁰

Out-of-hospital acute care services differ across countries but are typically services with high patient volumes, to which patients can self-refer.^{21 22} With limited diagnostic resources and time for assessment and treatment, such services are involved in the care for a broad range of patients and conditions, spanning from small children to the oldest old harbouring everything from trivial to potentially life-threatening illnesses.^{23 24} In many countries, the public is encouraged to call before attending for an initial assessment over the telephone and referral to the appropriate level of care (telephone triage).²⁵ Telephone triage is generally considered a wise strategy to prevent overcrowding in urgent care centres and EDs. However, triage by telephone can be particularly challenging. Calls must be managed promptly to ensure timely access to care for the public, and assessments are usually made without visual cues. With calls from older adults, frailty and altered clinical presentation pose added difficulty, and studies have demonstrated high age to be a risk factor for undertriage of medical calls.^{26 27} Identification of frailty in medical calls and other out-of-hospital acute care services could allow for more informed decisions on who needs to be seen by a physician without delay, who can be treated safely outside the hospital and who needs to be admitted. For individuals living with frailty, this balance between neither excessive nor insufficient treatment is essential to maintain function and capacity to carry out daily chores.^{5 8}

In Norway, where this study was conducted, the need to assess for frailty in the out-of-hospital setting is underscored by referral-based access to the EDs.²⁸ Also, telephone calls as initial mode of contact are widespread, and more than a third of calls to urgent care centres are managed exclusively by telephone, with no physical consultation.^{25 29} Identifying frailty during these initial telephone encounters can thus be essential to ensure appropriate subsequent care for older adults. However, despite the abundance of frailty screening tools, there is a lack of suitable tools for medical calls. The setting requires a very simple and rapid tool with high sensitivity to distinguish robust patients from individuals with possible frailty. Physical tests and observations are not feasible, and electronic health records from hospital and general practitioner (GP) offices are generally unavailable due to privacy or practical impediments. Consequently, frailty identification must be based on information that can be readily acquired from the caller, who—with increasing age of the patient—is often a caregiver calling on behalf of the patient.³⁰ Also, geriatric expertise or extensive time spent on training to screen for frailty cannot be expected. Hence, any tool to be implemented should be highly intuitive, requiring minimum training.

Among established frailty screening tools, even the more rapid ones are too comprehensive for medical calls or they fail to meet other demands of the setting. Similar barriers to implementing existing frailty screening tools apply to other out-of-hospital settings, such as ambulance services and urgent care centres. The aim of the present study was therefore to develop a new frailty screening tool suitable for urgent and emergency medical calls and other out-of-hospital acute care settings.

Methods

Design

In this article, we describe the development of FastFrail—a frailty screening tool designed for the medical call setting. We drafted a first set of potential items for the new tool through panel discussions. Next, we performed a cross-sectional study in which we administered the first set of potential items and an established frailty screening tool, the CFS, on a cohort of older adults. Further development of the tool was based on statistical analyses of this data material. The primary outcome was potential items associated with frailty by CFS. Ultimately, final items were selected through discussions in the panel.

Development of potential items

Prior to the development, we defined four criteria to meet the needs of the urgent and emergency medical call setting. The tool should:

• Require little training and no special equipment

• Consist of maximum five items—preferably simple and binary questions.

• Contain items that both patients and caregivers can be able to answer over the telephone within the timeframe of a medical call.

• Questions should feel natural for the operator to ask in the acute care setting.

A panel consisting of four of the authors (MSB, AHR, COG, IS) developed a first set of potential items. The potential items were selected based on a review of existing frailty tools, relevant literature, clinical observation and knowledge of the patient group and out-of-hospital acute care setting. MSB and AHR contributed with expertise in geriatric medicine, while primary and urgent care perspective was provided by COG and IS. We relied on Streiner and Norman’s guidelines for creating measurement scales.³¹ For item selection, this included avoiding incomprehensibility, ambiguity, double-barrelled questions, jargon, value-laden words and negative wording. Furthermore, to keep items as short as possible, while still comprehensible. Items in the first set were designed as binary or multiple-choice questions. Items were originally drafted and administered to the study sample in Norwegian and have gone through English translation for the purpose of this article. The items selected for the final version of the new tool have gone through forward and backward translation.

Setting and participants

The study was conducted in Norway, a country with 5.6 million inhabitants and life expectancy of 84.6 and 81.4 years for women and men, respectively.³² The primary assessment and treatment of acute illness/injury usually takes place in primary care services (GP’s office or urgent care centre). The main access point to acute care is a call to one of these services. If deemed necessary by the primary care physician, patients are referred to hospital ED for further assessment and treatment.²⁸ If the condition is critical, the ED can be accessed directly.

Data collection took place at the urgent care centre in Bergen, Norway’s second largest city, between January and August 2022. The population of Bergen per 1 January 2022 was 286 930 inhabitants, of whom 12% were aged ≥70 years.³³ The centre is open 24/7 and conducts approximately 100 000 patient consultations yearly. The public is encouraged to call before attending, but direct attendance is also possible. On arrival, patients are triaged by registered nurses with the five-level triage tool South African Triage Scale (SATS, Norwegian version).³⁴ The red category indicates the need for emergency medical assistance, orange indicates very urgent, yellow indicates urgent, green indicates non-urgent and blue is used for minor issues/administrative enquiries. Nursing home residents and injured patients are cared for through separate services most hours, so patients available for inclusion were primarily acutely ill, independent living older adults. Patients aged ≥70 years were eligible for inclusion. We excluded patients triaged to the highest urgency level (red) and patients not able to answer questions with no next of kin present.

Data collection

We included a convenience sample of older adults visiting the urgent care centre. Two medical students (MNV, NAW) in their fifth year of medical school collected the data, primarily afternoons and evenings (13:00–23:00) weekdays and weekends. The students observed and conducted short interviews with the patients and/or their next of kin after the patients were triaged and were waiting to be seen by a physician. Variables collected included age group, sex, urgency degree by SATS, answers to the first set of potential items and frailty status by the CFS.³⁵ The students had undergone training in administering the CFS through their curriculum prior to the data collection. Each patient was assigned a score between 1 (very fit) and 9 (terminally ill). As recommended for the acute care setting, the frailty assessment was based on the patient’s state 14 days prior to the visit (baseline status).

Steps of FastFrail development

After administration of the potential items and the CFS on the urgent care cohort, a stepwise development through statistical analyses and discussions in the panel was carried out (table 1).

Table 1

Steps of FastFrail development based on data from older adults at the Bergen Urgent Care Centre

Step	Method
Initial exploration of patient data
(1) Testing potential items for association with frailty by the CFS	χ2 tests
(2) Visualisation of the association between patients’ responses to potential items and CFS scores	Box plots
Modification of potential items
(3) Identification of cut-offs corresponding with frailty by CFS* to derive binary versions of potential items originally designed as multiple-choice questions	Box plots; inspection of the median and IQR of CFS scores for each response category Creating new, dichotomous variables
(4) Merging of repetitive items	Panel discussion; identification of repetitive items Creating new, merged variables
Item reduction
(5) Ranking of potential items (original and modified versions) by explained variance of CFS score in two categories: all question types and binary questions only	Linear regression analyses Assessment of adjusted R2 values
(6) Selection of candidate items	Panel discussion; based on explained variance ranking
(7) Ranking of combinations of candidate items by explained variance of CFS score in the same two categories as step 5	Multivariable linear regression analyses Assessment of adjusted R2 values
(8) Selection of the final combination of items	Panel discussion; based on the explained variance ranking and presumed feasibility in the intended setting
Final exploration of selected items
(9) Checking the selected combination of items for multicollinearity	Variance inflation factor
(10) Examination of the selected combination for adherence to frailty assessment by CFS	Calculating sensitivity and specificity

*In this step, we examined for two CFS cut-offs for frailty commonly used in the literature: ≥4 and ≥5.

CFS, Clinical Frailty Scale; IQR, interquartile range.

Statistics

We describe the study population using descriptive statistics, including counts, proportions, medians and means as appropriate. Association between potential items and CFS was measured by explained variance, adjusted R² values, from linear regression analyses; steps 5 and 7 of table 1. Other outputs from the regression analyses were not assessed and are not reported. The data set had few missing values, and cases with missing values were omitted from the analyses. Analyses were conducted using Stata software (V.17.0 and 18.0; Stata Corporation). In tests where significance was assessed, a p value threshold of 0.05 was used.

Patient and public involvement

Representatives for telephone operators at urgent care centres were consulted about the scope and design of the study, the first set of potential items and on the wordings of the items selected for the final version of the new tool. They also contributed with input on how the results should be communicated. Representatives for older adults reviewed the participant materials, such as the information sheets and consent forms, ensuring the language and content were accessible and appropriate.

Results

Development of potential items

The first set of potential items comprised nine items (Q1–Q9) in eight domains: place of living, dependence, mobility, homeboundness, polypharmacy, history of weight loss, history of hospital admission and history of falls (table 2). Q1–Q5 were designed as multiple-choice questions. Q6–Q9 were designed as binary questions (yes/no).

Table 2

Potential items for a new screening tool for frailty for medical calls: FastFrail

Domain	First set	Modified versions
Place of living	Q1 What type of residence do you live in?	Q1a Do you live in assisted living or a nursing home?
Dependence
Help from home care	Q2 Do you receive help from home care regularly, and if so: how often?	Q2a Do you receive daily help from home care? Q2b Do you receive weekly help from home care?	Q2a+Q3a Do you receive daily help? From home care or others Q2b+Q3b Do you receive weekly help? From home care or others
Help from others	Q3 Do you receive help from others regularly, and if so: how often?	Q3a Do you receive daily help from others? Q3b Do you receive weekly help from others?
Mobility	Q4 Do you use a walking aid, and if so: what kind?	Q4a Do you use a walking aid? Q4b Do you use a rollator walker, wheelchair or are you bedridden?
Homeboundness	Q5 How often do you leave your home without assistance from anybody?	Q5a Do you leave your home, more than once a week, without assistance from anybody?
Polypharmacy	Q6 Do you use four or more regular medications daily?
History of weight loss	Q7 Have you recently lost weight?
History of hospital admission(s)	Q8 Have you been admitted to the hospital in the past 12 months?
History of fall(s)	Q9 Have you fallen in the past 12 months?

Study participants

The inclusion process for the study is illustrated in figure 1. Table 3 summarises the patient characteristics of enrolled patients; 200 older adults (59% female). Of the included patients, 48% were 70–79 years old, 38% were 80–89 years old and 14% were ≥90 years. The majority were living alone (58%), and most were residing in own accommodation (93%). Before attendance, 136 (68%) patients had contacted the clinic by telephone. In 89% of the cases, the patients themselves provided the information needed to set a CFS score and the answers to the potential items. The median CFS score was 4 (IQR=1–7). The prevalence of frailty defined as CFS≥4 (process for selection of cut-off explained under Steps of FastFrail development: Item reduction) was 52%, increasing from 30% among patients aged 70–74 years to 82% among patients ≥90 years. By triage at attendance, 1% got assigned blue urgency degree, 19% green, 67% yellow and 13% orange.

Table 3

Characteristics of the patients included in the FastFrail development study

	Total N (%) 200	Robust* n (%) 96	Frail† n (%) 104	P value
Age (years)
70–74	53 (26)	37 (38)	16 (15)
75–79	44 (22)	19 (20)	25 (24)
80–84	43 (22)	21 (22)	22 (21)
85–89	32 (16)	14 (15)	18 (17)
≥90	28 (14)	5 (5)	23 (22)
Sex
Female	118 (59)	53 (55)	65 (63)
Living alone
Yes	114 (58)	44 (46)	70 (69)
Missing	2
Answers to original potential items
Q1 Place of living				0.006
Independent living	186 (93)	95 (99)	91 (88)
Assisted living	12 (6)	1 (1)	11 (11)
Nursing home	2 (1)	0 (0)	2 (2)
Q2 Dependence—help from home care				<0.001
None	152 (77)	94 (98)	58 (57)
Weekly	11 (6)	1 (1)	10 (10)
1–2 times daily	24 (12)	1 (1)	23 (23)
> 2 times daily	11 (6)	0 (0)	11 (11)
Missing	2		2
Q3 Dependence—help from others				<0.001
None	98 (50)	82 (85)	16 (16)
Less than once a week	23 (12)	7 (7)	16 (16)
Weekly	43 (22)	6 (6)	37 (37)
1–2 times daily	19 (10)	1 (1)	18 (18)
> 2 times daily	13 (7)	0 (0)	13 (13)
Missing	4		4
Q4 Mobility				<0.001
No walking aid	116 (58)	86 (90)	30 (29)
Crutch/cane	32 (16)	8 (8)	24 (23)
Rollator walker	41 (21)	2 (2)	39 (38)
Wheelchair	10 (5)	0 (0)	10 (10)
Bedridden	1 (1)	0 (0)	1 (1)
Q5 Homeboundness				<0.001
Leaving home without assistance ≥ 8 times a week	35 (18)	32 (34)	3 (3)
2–7 times a week	110 (55)	62 (65)	48 (46)
0–1 times a week	54 (27)	1 (1)	53 (51)
Missing	1	1
Q6 Polypharmacy‡				0.024
No	67 (34)	40 (42)	27 (26)
Yes	131 (66)	56 (58)	75 (74)
Missing	2		2
Q7 History of recent weight loss				<0.001
No	116 (58)	71 (74)	45 (43)
Yes	84 (42)	25 (26)	59 (57)
Q8 History of hospital admission(s) in the past 12 months				0.004
No	108 (54)	62 (65)	46 (44)
Yes	92 (46)	34 (35)	58 (56)
Q9 History of fall(s) in the past 12 months				<0.001
No	121 (61)	71 (74)	50 (48)
Yes	79 (40)	25 (26)	54 (52)

Significant p values (<0.05) in bold.

*CFS 1-3.

†CFS ≥ 4.

‡Polypharmacy ≥ 4 regular medications daily.

CFS, Clinical Frailty Scale.

Figure 1. Flow diagram of patient enrolment in the FastFrail development study. *Cognitive impairment, somnolent, drug/psychiatric disorder. **Several patients present at the same time, left before seen by a physician, no available rooms, etc.

Steps of FastFrail development

Initial exploration of patient data

All nine potential items (Q1–Q9) administered on the cohort were found to be significantly associated with frailty (table 3). Figure 2 shows box plots of each potential item by CFS scores.

Figure 2. Box plots of scores on Clinical Frailty Scale (CFS) by responses to potential items among patients included in the FastFrail development study.

Modification of potential items

The box plots were assessed to create dichotomous versions of items Q1–Q5 corresponding to frailty; in this step, examined for both CFS≥4 and CFS≥5. This resulted in one new item from each of the items Q1 and Q5 (discrimination at the same response option regardless of frailty cut-off by CFS) and two new items from each of items Q2–Q4 (discrimination at different response options by different CFS cut-off for frailty) (figure 2). The items within the domain dependency were merged after the dichotomisation forming items Q2a+Q3a and Q2b+Q3b. Thus, amounting to 10 new items including 2 corresponding to CFS≥4/5 (Q1a, Q5a), 4 corresponding to CFS≥4 (Q2b, Q3b, Q4a, and Q2b+Q3b) and 4 corresponding to CFS≥5 (Q2a, Q3a, Q4b and Q2a+Q3a). Modified items are all listed in table 2, together with the original items administered on the patients.

Item reduction

Table 4 shows the potential items, original and modified versions, by decreasing explained variance of CFS score. Items from the domains dependency, homeboundness and mobility explained most of the variance in CFS score in both categories. Item Q2b+Q3b (derived from CFS cut-off≥4) was at the top of the ranking in both categories, explaining 59% of the variance in CFS score. When reviewing all question types, combining the top three items (Q2b+Q3b, Q5 and Q4) resulted in an explained variance of 77%. Adding a fourth item added little value (increase in explained variance 0.4%), and there was no added value in combining with yet another item. The top three items in binary questions only, Q2b+Q3b (derived from CFS cut-off≥4), Q5a (derived from CFS cut-off≥4/5) and Q4a (derived from CFS cut-off≥4), could together explain 74% of the variance in CFS score. As for all question types, saturation was reached when combining three items. As modified versions of potential items derived from CFS cut-off≥4 were superior to those from cut-off≥5 in explaining variance of CFS score in the study population, and we opted for a tool with high sensitivity, we chose a frailty definition of CFS≥4.

Table 4

Potential items and combinations of potential items in descending order by explained variance of CFS score in two categories: all question types and binary versions only

Domain	Item	Item type	Explained variance* (%) of CFS score
All question types
Dependency	Q2b+Q3b	Modified	59	72	77	77	77	77	77	77
Homeboundness	Q5	Original	57
Mobility	Q4	Original	49
History of fall(s)	Q9	Original	6
History of weight loss	Q7	Original	6
History of hospital admission(s)	Q8	Original	4
Place of living	Q1	Original	3
Polypharmacy	Q6	Original	2
Binary questions only
Dependency	Q2b+Q3b	Modified	59	69	74	74	74	74	74	74
Homeboundness	Q5a	Modified	48
Mobility	Q4a	Modified	43
History of fall(s)	Q9	Original	6
History of weight loss	Q7	Original	6
History of hospital admission(s)	Q8	Original	4
Place of living	Q1a	Modified	3
Polypharmacy	Q6	Original	2

Item with the highest explained variance within each domain shown.

*Adjusted R² value.

CFS, Clinical Frailty Scale.

Through discussions in the panel, it was highlighted that the original items require considerably more effort by both the asking and the answering part than the modified binary versions. The reduction in explained variance of CFS score by using the modified versions of the homeboundness and mobility items was a mere 3 percentage points. For feasibility purposes, we therefore selected the three questions Q2b+Q3b, Q5a and Q4a to constitute the new frailty screening tool FastFrail (Box 1).

Box 1

FastFrail

• Do you receive weekly help? From home care or others Yes/no

• Do you leave your home more than once a week, without assistance from anybody? Yes/no

• When you walk, do you use a walking aid? Walking stick, rollator walker or similar Yes/no

Response option in bold indicates frailty=positive response.

Final exploration of selected items

Variance inflation factor (VIF) was calculated to be less than two for the final combination of items (Q2b+Q3b, Q5a and Q4a) (online supplemental table 1), thus multicollinearity was considered not pertinent. Of the patients assessed as frail by the CFS in our study sample, 91% had at least one positive response on the FastFrail items, 71% had two or more positive responses and 40% had a positive response to all three items (online supplemental table 2). Among robust patients (CFS 1–3), 20% had a positive response to one FastFrail item, while the remaining 80% had no positive responses to any of the FastFrail items. While these parameters can be thought to indicate the sensitivity and specificity of FastFrail, note that we have not applied the tool on a new patient sample.

Discussion

Principal findings

In this article, we have described the development of a new frailty screening tool for medical calls and other out-of-hospital acute care settings. The tool, FastFrail, consists of simple binary questions about receiving help weekly, being homebound and using a walking aid. These factors were found to be strong indicators of frailty in our sample of older urgent care patients.

Strengths and weaknesses of the study

A key strength of our study is the sound, stepwise approach we used to develop FastFrail. As recommended by Streiner and Norman, we searched existing tools and other relevant literature, combined with clinical experience, to identify a first set of potential items.³¹ We set predefined criteria to foster the tool’s feasibility in the intended setting. We administered the potential items and an established, more comprehensive tool on a relevant cohort and used these data to develop a more rapid tool. This is a well-established method for developing health measurement tools, in general, and frailty screening tools, in particular.^{31 36–39} We had few exclusion criteria to ensure broad inclusion of older adults visiting the urgent care centre.

We included CFS 4 (living with very mild frailty) in our definition of frailty, grounded on findings from steps 3–5 of the FastFrail development, as well as the desire for a sensitive tool. Selection of cut-off≥4 is supported by several studies indicating patients assessed as CFS 4 experience outcomes following an acute illness/injury comparable to CFS≥5,^6–8 as well as O’Caoimh et al’s findings from comparing psychometric properties of various screening tools and optimal cut-offs (balance between sensitivity and specificity) in an ED population.⁴⁰ Cut-off≥4 allows us to include individuals who, while not dependent on others, are showing signs of slowing down and may be at risk of developing frailty-related complications. These patients may benefit from a different approach when contacting acute care services, particularly in an initial assessment by telephone.

Our study also has limitations. The patient sample is small and from one study site only. Also, the FastFrail is primarily intended for the medical call setting, but for feasibility reasons our data were collected from patients physically present at an urgent care centre (most of whom had called before attending). Compared with all older adult callers, our sample likely includes more patients with more moderate-to-severe conditions (that may be more prevalent among frail individuals) and more patients who are fairly mobile (severely frail, homebound individuals are often managed by home visits). It is uncertain how this may have influenced the results. Along with the obvious constraints of the cross-sectional design, this highlights the necessity of follow-up longitudinal studies to demonstrate FastFrail’s feasibility and other test properties in the intended setting.

Important aspects of frailty, such as cognitive impairment and multimorbidity, were not explicitly addressed through the potential items. Furthermore, questions Q6–Q9 were formulated as binary questions, thereby eliminating the chance to find relevant cut-offs with potentially strong associations to frailty by CFS. Moreover, many other tools could have served as a ‘gold standard’ for frailty, yielding different results, so the use of CFS represents a potential area of concern. CFS was chosen because of the large body of evidence that it is suitable and valid in the acute care setting, can predict adverse outcomes in this population, is easy to use and a Norwegian translation was available.^{41 42} CFS’s reliance on subjective judgement can cause high inter-rater variability. Limiting the screening task to two dedicated medical students who had undergone the same training likely mitigated this potential bias.

Interpretation and comparison of results

We found, as expected, an association between all potential items and frailty by the CFS in the urgent care cohort. When examining for degree of association, items closely connected to physical functioning stood out. Several factors could have contributed to this. First, the CFS is highly focused on functional abilities.³⁹ It is thus not surprising that items related to function were found to explain a large percentage of the variance in the CFS score (dependency, mobility and homeboundness; 43–59%). Correspondingly, that the polypharmacy, weight loss, hospital admission and fall items (Q6–Q9) were found to explain only minor variance (2–6%) as such factors are not emphasised in the CFS. Notably, these items were all designed as binary questions in the first set. Multiple-choice versions could have provided different results and enabled us to develop binary questions with more relevant cut-offs (eg, the polypharmacy cut-off may be too low to distinguish frailty). Also, the data collection process favoured inclusion of patients who were assigned a separate room to wait in, which might have introduced a selection bias in favour of patients with physical impairments and/or more severe conditions. We cannot exclude that this may have affected our results, adding to the focus on physical functioning from using the CFS. Particularly, this may be the case for the mobility item. Supportive of our conclusion, though, is that there is a substantial leap from the FastFrail item with the lowest explained variance (mobility, 43%) to the next potential item on the ranking (history of falls, 6%). Furthermore, we believe that the final items are less susceptible to chance occurrences, less dependent on when in the frailty trajectory a patient is asked and less prone to recall bias than the rejected items.

Given the lack of tools for the setting FastFrail is intended for, there is an absence of studies directly comparable to ours. The out-of-hospital setting calls for even shorter methods for assessment than the ED, though it is relevant to draw on related research findings from the latter as the services exhibit several similarities. An international Delphi consensus study was recently published, concluding on central requirements of frailty screening tools to be used in the ED.⁴³ The expert panel reached consensus that tools should be “short, multidimensional and well-calibrated across the spectrum of frailty”. Furthermore, the number of included items should mirror the time available for assessment and that feasibility is paramount to having an ‘ideal’ screening tool. These criteria align well with our predefined criteria for the out-of-hospital setting, and we argue that FastFrail fulfils several of these requirements. The final items in FastFrail can also be placed within the key domains agreed on by the expert panel (functional ability, mobility, cognition, medication use and social factors). While dependency and mobility items occur in several of the question-based tools used in EDs today,^44–47 it is often with a higher detail level than FastFrail.^{46 47} Capturing the full scope of frailty is not possible in such a time-constrained setting, and we assert the level of detail in FastFrail to be appropriate for managing urgent medical calls. Homeboundness is less commonly addressed in established tools. The term is defined by Ko and Noh as being confined to one’s home due to limitations or health issues.⁴⁸ As much as 48% of the CFS score in our material could be explained by this factor (table 4; binary version). We therefore advocate for addressing homeboundness to effectively uncover signs of frailty.

Implications

At present, frailty is often unidentified and unattended to in out-of-hospital acute care services,²⁰ and there has been a lack of suitable tools for medical calls. It is necessary to have carefully selected items to maximise the information that can be obtained from only a few questions. FastFrail provides an opportunity for the assessment of frailty in medical calls and other out-of-hospital services that has not previously been met. Moving forward, the tool should undergo testing in clinical practice to establish its psychometric properties. If the tool demonstrates strong psychometric properties, it could contribute to enhance the precision of initial assessments for older adults in out-of-hospital settings. We wish to highlight that to improve management of acutely ill and injured older adults, frailty assessment tools must be accompanied by increased awareness and knowledge about frailty among acute care providers, including telephone operators in medical call centres. Moreover, services need to be organised to meet the needs of frail older adults, for instance, by facilitating home visits.

Unanswered questions and future research

There is a great need for research on improving triage for older adults in medical calls and other out-of-hospital services to support better management of this rapidly growing population. The findings from this study indicate that the three FastFrail items could potentially be used to detect frailty in an out-of-hospital acute care population. However, FastFrail’s feasibility and validity in the intended setting are yet to be demonstrated. The sensitivity and specificity reported in this article are calculated on the same data as were used to develop FastFrail and should thus be interpreted with caution. Future studies should report on sensitivity and specificity of FastFrail using various cut-offs to clarify what is an optimal threshold for frailty by this tool. How FastFrail should be used and what response the detection of frailty should trigger will likely differ with various models of organisation. Lastly, further research should monitor if implementing frailty screening in the medical call setting has the intended effects.

Conclusion

Feasible frailty screening tools for medical calls are lacking. Through this study, we established that receiving help weekly, being homebound and using a walking aid are strong indicators of frailty among older urgent care patients. We developed a new screening tool, FastFrail, consisting of three simple, binary questions on these aspects. We propose FastFrail could potentially be used to identify frailty among older adults in medical calls and other settings that require a very rapid method for assessment. The tool should be tested alongside traditional triage methods to determine if it can enable more accurate assessment of older adults calling for acute medical help.

We thank the Bergen Urgent Care Centre for facilitating the data collection. We would also like to thank the representatives for telephone operators at the National Centre for Emergency Primary Health Care for feedback during the planning, conduction and reporting of the study. Finally, a thank you to IH Johansen and M Price for the translation of the final items from Norwegian to English language. We also like to acknowledge that this work was presented at the 19th Congress of the European Geriatric Medicine Society in 2023, and the abstract is published in the European Geriatric Medicine.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

The study was approved by the Regional Ethical Committee of Western Norway (reference number 275602). Participation in the study was based on informed consent. To ensure that patients living with cognitive impairment or presenting with delirium were not selectively excluded, patients could participate through proxy consent by a next of kin if unable to provide informed consent themselves. The included patients were not subject to any experimental intervention, and inclusion in the study was considered not to delay necessary medical care. Data were analysed anonymously.

Footnote

Contributors COG, MSB, AHR and IS were involved in the design of the study. VB, MSB, COG and IS planned the analyses. COG and IS conducted and interpreted the analyses with guidance from VB. IS drafted the manuscript. MNV and NAW collected the data and contributed to the 'Methods' section of the manuscript. All authors edited and revised the manuscript and approved the final version of the manuscript. IS is the guarantor. ChatGPT (https://chatgpt.com/) was used in the drafting stage to improve language and shorten paragraphs.

Funding This work was supported by The Norwegian Research Fund for General Practice.

Competing interests None declared.

Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.

Provenance and peer review Not commissioned; externally peer reviewed.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

References

¹ O’Caoimh R, Galluzzo L, Rodríguez-Laso Á, et al. Prevalence of frailty at population level in European ADVANTAGE Joint Action Member States: a systematic review and meta-analysis. Ann Ist Super Sanita 2018; 54: 226–38. doi:10.4415/ANN_18_03_10

² Rockwood K, Mitnitski A. Frailty in relation to the accumulation of deficits. J Gerontol A Biol Sci Med Sci 2007; 62: 722–7. doi:10.1093/gerona/62.7.722

³ Simon NR, Jauslin AS, Bingisser R, et al. Emergency presentations of older patients living with frailty: Presenting symptoms compared with non-frail patients. Am J Emerg Med 2022; 59: 111–7. doi:10.1016/j.ajem.2022.06.046

⁴ Pulok MH, Theou O, van der Valk AM, et al. The role of illness acuity on the association between frailty and mortality in emergency department patients referred to internal medicine. Age Ageing 2020; 49: 1071–9. doi:10.1093/ageing/afaa089

⁵ Cunha AIL, Veronese N, de Melo Borges S, et al. Frailty as a predictor of adverse outcomes in hospitalized older adults: A systematic review and meta-analysis. Ageing Res Rev 2019; 56: 100960. doi:10.1016/j.arr.2019.100960

⁶ Bruno RR, Wernly B, Bagshaw SM, et al. The Clinical Frailty Scale for mortality prediction of old acutely admitted intensive care patients: a meta-analysis of individual patient-level data. Ann Intensive Care 2023; 13: 37. doi:10.1186/s13613-023-01132-x

⁷ Park B, Alani Z, Sulistio E, et al. Frailty using the Clinical Frailty Scale to predict short- and long-term adverse outcomes following emergency laparotomy: meta-analysis. BJS Open 2024; 8: zrae078. doi:10.1093/bjsopen/zrae078

⁸ Engvig A, Wyller TB, Skovlund E, et al. Association between clinical frailty, illness severity and post-discharge survival: a prospective cohort study of older medical inpatients in Norway. Eur Geriatr Med 2022; 13: 453–61. doi:10.1007/s41999-021-00555-8

⁹ Theou O, Squires E, Mallery K, et al. What do we know about frailty in the acute care setting? A scoping review. BMC Geriatr 2018; 18: 139. doi:10.1186/s12877-018-0823-2

¹⁰ Puig-Campmany M, Blázquez-Andión M, Ris-Romeu J. Triage tools: a cautious (and critical) view towards their use in old patients. Eur Geriatr Med 2022; 13: 319–22. doi:10.1007/s41999-021-00572-7

¹¹ Grossmann FF, Zumbrunn T, Frauchiger A, et al. At risk of undertriage? Testing the performance and accuracy of the emergency severity index in older emergency department patients. Ann Emerg Med 2012; 60: 317–25. doi:10.1016/j.annemergmed.2011.12.013

¹² Rodríguez Mañas L, García-Sánchez I, Hendry A, et al. Key Messages for a Frailty Prevention and Management Policy in Europe from the Advantage Joint Action Consortium. J Nutr Health Aging 2018; 22: 892–7. doi:10.1007/s12603-018-1064-y

¹³ Boreskie KF, Hay JL, Boreskie PE, et al. Frailty-aware care: giving value to frailty assessment across different healthcare settings. BMC Geriatr 2022; 22: 13. doi:10.1186/s12877-021-02722-9

¹⁴ Blomaard LC, Speksnijder C, Lucke JA, et al. Geriatric Screening, Triage Urgency, and 30-Day Mortality in Older Emergency Department Patients. J Am Geriatr Soc 2020; 68: 1755–62. doi:10.1111/jgs.16427

¹⁵ Mowbray F, Brousseau A-A, Mercier E, et al. Examining the relationship between triage acuity and frailty to inform the care of older emergency department patients: Findings from a large Canadian multisite cohort study. CJEM 2020; 22: 74–81. doi:10.1017/cem.2019.432

¹⁶ De Biasio JC, Mittel AM, Mueller AL, et al. Frailty in Critical Care Medicine: A Review. Anesth Analg 2020; 130: 1462–73. doi:10.1213/ANE.0000000000004665

¹⁷ O’Caoimh R, McGauran J, O’Donovan MR, et al. Frailty Screening in the Emergency Department: Comparing the Variable Indicative of Placement Risk, Clinical Frailty Scale and PRISMA-7. Int J Environ Res Public Health 2022; 20: 290. doi:10.3390/ijerph20010290

¹⁸ Ng C-J, Chien L-T, Huang C-H, et al. Integrating the clinical frailty scale with emergency department triage systems for elder patients: A prospective study. Am J Emerg Med 2023; 66: 16–21. doi:10.1016/j.ajem.2023.01.002

¹⁹ British Geriatrics Society. Silver Book II: Quality urgent care for older people. Available: https://www.bgs.org.uk/resources/resource-series/silver-book-ii-frailty [Accessed 31 Oct 2024 ].

²⁰ Alshibani A, Alharbi M, Conroy S. Frailty Identification in Prehospital Care: A Scoping Review of the Literature. Open Access Emerg Med 2023; 15: 227–39. doi:10.2147/OAEM.S409083

²¹ Kraaijvanger N, van Leeuwen H, Rijpsma D, et al. Motives for self-referral to the emergency department: a systematic review of the literature. BMC Health Serv Res 2016; 16: 685. doi:10.1186/s12913-016-1935-z

²² Steeman L, Uijen M, Plat E, et al. Out-of-hours primary care in 26 European countries: an overview of organizational models. Fam Pract 2020; 37: 744–50. doi:10.1093/fampra/cmaa064

²³ Helsedirektoratet. Legevakt og legevaktsentral. Oslo Helsedirektoratet. Available: https://www.helsedirektoratet.no/veiledere/legevakt-og-legevaktsentral [accessed 31 Oct 2024 ]

²⁴ NHS. Urgent and emergency care services. London National Health Service (NHS). Available: https://www.nhs.uk/nhs-services/urgent-and-emergency-care-services/ [accessed 31 Oct 2024 ]

²⁵ Midtbø V, Fotland S-L, Johansen IH, et al. From direct attendance to telephone triage in an emergency primary healthcare service: an observational study. BMJ Open 2022; 12: e054046. doi:10.1136/bmjopen-2021-054046

²⁶ Inokuchi R, Jin X, Iwagami M, et al. Factors associated with undertriage in patients classified by the need to visit a hospital by telephone triage: a retrospective cohort study. BMC Emerg Med 2021; 21: 155. doi:10.1186/s12873-021-00552-x

²⁷ Graversen DS, Pedersen AF, Christensen MB, et al. Factors associated with undertriage and overtriage in telephone triage in Danish out-of-hours primary care: a natural quasi-experimental cross-sectional study of randomly selected and high-risk calls. BMJ Open 2023; 13: e064999. doi:10.1136/bmjopen-2022-064999

²⁸ Blinkenberg J, Hetlevik Ø, Sandvik H, et al. Reasons for acute referrals to hospital from general practitioners and out-of-hours doctors in Norway: a registry-based observational study. BMC Health Serv Res 2022; 22: 78. doi:10.1186/s12913-021-07444-7

²⁹ Haraldseide LM, Sortland LS, Hunskaar S, et al. Contact characteristics and factors associated with the degree of urgency among older people in emergency primary health care: a cross-sectional study. BMC Health Serv Res 2020; 20: 345. doi:10.1186/s12913-020-05219-0

³⁰ Sortland LS, Haraldseide LM, Sebjørnsen I. Eldre i den akuttmedisinske kjeden. Rapport nr. 1-2021. Bergen Nasjonalt kompetansesenter for legevaktmedisin, NORCE Norwegian Research Centre, avdeling Helse; 2021. Available: https://norceresearch.brage.unit.no/norceresearch-xmlui/bitstream/handle/11250/2729947/Rapport%2bnr.1-2021%2bEldre%2bi%2bden%2bakuttmedisinske%2bkjeden.pdf?sequence=1&isAllowed=y

³¹ Streiner DL, Norman GR. Health measurement scales: a practical guide to their development and use. 1st edn. Oxford: Oxford University Press, 1989.

³² Statistics Norway. Døde. Oslo Statistics Norway. Available: https://www.ssb.no/befolkning/fodte-og-dode/statistikk/dode [accessed 31 Oct 2024 ]

³³ Statistics Norway. Table 10826: population by sex and one-year age groups, 1986–2023. Oslo Statistics Norway. Available: https://www.ssb.no/statbank/table/10826/ [accessed 13 Aug 2024 ]

³⁴ Helse Bergen HF. Brukerveiledning SATS-Norge: Barn og voksne. Versjon 2.01. Bergen Helse Bergen; 2015. Available: https://www.helse-bergen.no/48f300/siteassets/seksjon/mottaksklinikken/documents/2015.09.14-brukerveiledning-sats-norge-barn-og-voksne-versjon-2.01-ny.pdf [accessed 13 Aug 2024 ]

³⁵ Klinisk skrøpelighetsskala (Clinical Frailty Scale). 2005-2020 Rockwood, Version 2.0 (NO). Translated with permission by Rostoft S, Engstad TT, Sjøbø B, Flaatten H. n.d. Available: https://www.legeforeningen.no/contentassets/21ef25cf569d44749573de21a8d6b043/cfs_norsk_horisontal_2021.pdf

³⁶ Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001; 56: M146–56. doi:10.1093/gerona/56.3.m146

³⁷ Mitnitski AB, Mogilner AJ, Rockwood K. Accumulation of deficits as a proxy measure of aging. ScientificWorldJournal 2001; 1: 323–36. doi:10.1100/tsw.2001.58

³⁸ Tsiouris A, Hammoud ZT, Velanovich V, et al. A modified frailty index to assess morbidity and mortality after lobectomy. J Surg Res 2013; 183: 40–6. doi:10.1016/j.jss.2012.11.059

³⁹ Rockwood K, Song X, MacKnight C, et al. A global clinical measure of fitness and frailty in elderly people. CMAJ 2005; 173: 489–95. doi:10.1503/cmaj.050051

⁴⁰ O’Caoimh R, Costello M, Small C, et al. Comparison of Frailty Screening Instruments in the Emergency Department. Int J Environ Res Public Health 2019; 16: 3626. doi:10.3390/ijerph16193626

⁴¹ Dejgaard MS, Rostoft S. Systematic assessment of frailty. Tidsskr Nor Laegeforen 2021; 141. doi:10.4045/tidsskr.20.0944

⁴² Church S, Rogers E, Rockwood K, et al. A scoping review of the Clinical Frailty Scale. BMC Geriatr 2020; 20: 393. doi:10.1186/s12877-020-01801-7

⁴³ Moloney E, O’Donovan MR, Carpenter CR, et al. Core requirements of frailty screening in the emergency department: an international Delphi consensus study. Age Ageing 2024; 53: afae013. doi:10.1093/ageing/afae013

⁴⁴ Raîche M, Hébert R, Dubois M-F. PRISMA-7: a case-finding tool to identify older adults with moderate to severe disabilities. Arch Gerontol Geriatr 2008; 47: 9–18. doi:10.1016/j.archger.2007.06.004

⁴⁵ McCusker J, Warburton RN, Lambert SD, et al. The Revised Identification of Seniors At Risk screening tool predicts readmission in older hospitalized patients: a cohort study. BMC Geriatr 2022; 22: 888. doi:10.1186/s12877-022-03458-w

⁴⁶ Vandewoude MFJ, Geerts CAM, d’Hooghe AHM, et al. A screening tool to identify older people at risk of adverse health outcomes at the time of hospital admission. Tijdschr Gerontol Geriatr 2006; 37: 203–9.

⁴⁷ Rolfson DB, Majumdar SR, Tsuyuki RT, et al. Validity and reliability of the Edmonton Frail Scale. Age Ageing 2006; 35: 526–9. doi:10.1093/ageing/afl041

⁴⁸ Ko Y, Noh W. A Scoping Review of Homebound Older People: Definition, Measurement and Determinants. Int J Environ Res Public Health 2021; 18: 3949. doi:10.3390/ijerph18083949