WHAT IS ALREADY KNOWN ON THIS TOPIC

Assessed frailty has been associated with adverse outcomes in various settings. However, frailty assessments have primarily been made by research personnel. There has been no multicentre evaluation of whether Clinical Frailty Scale (CFS) assessment, conducted exclusively by emergency department (ED) personnel as part of their clinical work, is associated with adverse events in non-selected ED patients.

WHAT THIS STUDY ADDS

This prospective observational study was conducted in three emergency departments in Sweden where staff routinely assess CFS. Frailty (CFS≥5) was associated with increased mortality at 7, 30 and 90 days, higher admission rates and longer ED and hospital length of stay compared with robust patients.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICYBackground

Worldwide, the proportion of adults aged over 65 is expected to increase from 8.5% in 2015 to nearly 25% by 2050.1 This demographic shift affects all parts of the healthcare system, including the emergency department (ED). In the ED, increased numbers of older patients with extensive medical histories and a need for functional support, known as a syndrome of frailty, pose complex challenges.2 Frail individuals are in a state of vulnerability and carry an increased risk for adverse events, including falls or other accidents, hospitalisation and mortality.3 Unfortunately, many EDs fail to identify and address the specific needs of patients living with frailty.4 Importantly, chronological age is not necessarily accompanied by frailty;5 therefore, identification of older patients at risk of adverse events based on frailty assessment is an appealing concept to guide ED care and resource utilisation. Robust older patients may tolerate, and benefit from, advanced medical interventions, while care for those living with severe frailty may focus more on improving the quality of life.6

Out of the many existing tools for frailty assessment,7 the Clinical Frailty Scale (CFS)8 has been recommended for use in the ED due to its practicality in a busy environment.9 The CFS was developed to screen for frailty in patients ≥65 years of age and assess the patient’s morbidity, cognitive status and functional level in daily life. The scale ranges from 1 point (very fit) to 9 points (terminally ill). The cut-off for frailty was set at 5 points;8 those with a CFS≥5 points are considered frail.10 Further, a stepwise increase in the risk of 1-year mortality for each additional frailty level on the CFS has been shown in ED patients,11 but, there is no data on whether, compared with dichotomisation, the CFS score further differentiates patients regarding short-term term mortality.

The ability of the CFS to determine which ED patients are most likely to have poor outcomes has previously been investigated. However, these studies have largely only investigated specific patient groups.12 In studies with non-selected patients, the assessors of frailty have primarily been research personnel;9 13 14 this methodology could affect the generalisability of the results to routine clinical practice. The two pre-existing studies where ED staff exclusively performed the frailty assessments were both single-centre with one of the studies focusing specifically on a patient group living with frailty under low socioeconomic circumstances.15 16

This multicentre study determined whether there is an association between CFS, as assessed by regular ED staff in a non-select population of patients of older age with short and long-term mortality, hospital admission and ED length of stay (LOS).

Materials and methodsStudy design and setting

This was a prospective multicentre study carried out in three Swedish EDs: Linköping ED (university hospital), Norrköping ED (urban community hospital) and Motala ED (rural community hospital). Annual visits in Linköping and Norrköping are approximately 50 000 each while Motala has around 25 000. The collection period was May/June in Linköping, and October/November in Norrköping and Motala.

The council of Region Östergötland in Sweden had previously decided to implement CFS for clinical use, and the instrument had recently been introduced in the participating EDs. During the year leading up to the introduction and study start, all clinical staff at the included EDs, that is, physicians, registered nurses and assistant nurses, were encouraged to complete an e-learning educational course on the CFS as part of their continuing medical education. The content of the course was based on the online training module developed by the Aging Innovation in perioperative Medicine & Surgery (AIMS) research group at Ottawa Hospital, Canada.17 Three clinical vignettes were included, along with the basic theoretical concept of frailty and its consequences for patients’ health and function. Staff began performing the CFS assessments and collecting the data on a worksheet at the start of the study period.

The study was registered on ClinicalTrials.gov.

Selection of participants

Patients ≥65 years of age were eligible for the study and included if the data collection worksheet was complete.

Methods of measurement

A worksheet (online supplemental appendix 1) was added to the ED records for all patients ≥65 years old when admitted to the ED. The CFS assessments were performed at all hours and were done by one of the members of the clinical team responsible for the patient (typically a physician, registered nurse or assistant nurse) and could be carried out at any time during the patient’s stay in the ED. The CFS assessment date, the patient’s social security number and the clinical role of the assessor were all compulsory fields on the assessment sheet for the patient’s inclusion in the study. A member of the research team manually collected and reviewed the worksheets. If the worksheet was incomplete, the patient was excluded from the study.

We used the validated Swedish version of the CFS (CFS-9),18 and included the pictograms from the original version.19 Inter-rater reliability (IRR) was not assessed in this study, but our group has previously evaluated CFS and shown ‘good’ IRR.20

As the assessment was part of clinical care, staff had access to standard clinical information (eg, patient, relative or caregiver, and notes from the electronic medical records) and were allowed to discuss the CFS score within the treatment team.

The worksheets were stored in a locked space without access for unauthorised personnel and the digital data recorded from the documents were kept in a protected network storage space.

Outcome measures

The primary outcome was 30-day mortality. Secondary outcomes were 7-day and 90-day mortality, hospital admission and LOS in the ED and hospital.

Statistical analysis

The sample size was estimated prior to data collection using G* Power.21 The calculation assumed a 90-day mortality of 12% in patients living with frailty (defined as CFS≥5) and 8% in robust patients ≥65 years, based on an overall mortality of 10% at 90 days in patients ≥65 years (calculated from the Swedish Emergency Care Register). With a 95% CI, a power of 0.8, and a 10% margin to account for exclusions, loss to follow-up etc, a sample size of at least 1800 patients was required.

Outcome data based on the patient’s social security number was exported after all patients completed the follow-up period, from the electronic health records, which crosslinks mortality data from the national patient registry. The study cohort was divided into living with frailty (CFS≥5) or robust (CFS<5). Descriptive statistics were reported as medians for continuous variables and percentages for categorical variables. Mortality risk and hospital admission were analysed using logistic regression with and without adjusting predefined confounders; age (continuous), sex (male or female), mode of arrival (ambulance/recumbent patient transport or walk-in) and acuity (1 through 4, with 1 being immediate) based on the Rapid Emergency Triage and Treatment System triage system used in all three EDs.22 The mortality risk is reported as ORs for living with frailty versus robust (1 df) and in a secondary analysis for each score of the scale (8 df).

The association between frailty and LOS in the ED and hospital, treated as medians, was assessed with frailty as a dichotomous variable using the Mann-Whitney U test. CIs for LOS were estimated using bootstrapping with replacement for 1000 iterations. A p value <0.05 with a 95% CI not crossing 1 was classified as statistically significant for the main analysis. In the secondary multivariate analysis, when adjusting for confounders, a p value <0.003 was classified as statistically significant to account for multiple comparisons based on a Bonferroni correction of m=15. Data was imported into Pandas (V.0.23)23 and analysed with Python using the Scipy library (V.1.17)24 and the Statsmodels library (V.0.12).25

Patient and public involvement

Patients were not involved in the planning of this study.

ResultsCharacteristics of the study subjects

There were 2275 ED visits with CFS assessments during the study period. Of these, 435 were excluded, leaving 1840 index visits eligible for analysis (figure 1). The main reasons for exclusion were missing data (234) or return visits (155). Most assessments were made by nurses or assistant nurses (53.5% and 43.0%, respectively), with 3.4% by physicians.

Figure 1

Flowchart describing the inclusion process. CFS, Clinical Frailty Scale; ED, emergency department.

The median age of the study cohort was 78 (IQR 73–85), and 54.6% were women. During the study period, an additional 2240 visits by patients ≥65 years of age were registered at the EDs but were not assessed for CFS. The eligible but missed patients had a median age of 76 (71–82), and 51% were women.

Of the 1840 patients, 606 (32.9%) were living with frailty. Patients living with frailty were older (mean age 83 years vs 76 years in robust patients), higher proportion arrived by ambulance (76.4% vs 41.5%) (table 1). A higher proportion of patients living with frailty had triage levels 1 (immediate) and 2 (very urgent) compared with robust patients.

Table 1

Descriptive characteristics of the study cohort

Main resultsMortality

All patients in the cohort had follow-up until 90 days. The crude 30-day mortality was 3.2%; no patients died during the ED stay. Mortality at 30 days was significantly higher in patients living with frailty compared with robust patients: 7.9% versus 0.9% (diff 7%, 95% CI for the difference 4.8% to 9.2%), as well as at 7 and 90 days (table 2). The unadjusted ORs for mortality in patients living with frailty compared with robust patients were 16.7 (95% CI 3.8 to 72.9), 9.6 (95% CI 4.9 to 18.6) and 7.6 (95% CI 5.0 to 11.7) for 7-day, 30-day and 90-day mortality, respectively. The increased risk persisted after adjusting for confounders, with ORs of 10.7 (95% CI 2.3 to 50.5), 6.0 (95% CI 3.0 to 12.3) and 6 (95% CI 3.6 to 9.1) for 7-day, 30-day and 90-day mortality, respectively (table 3). The OR for 30-day mortality increased by 2.1 (95% CI 1.7 to 2.4, p<0.001), with each additional step in CFS, in the regression analysis (figure 2).

Figure 2

Mortality for each Clinical Frailty Scale score at 30 days.

Table 2

Mortality, admission and ED and in-hospital length of stay

Table 3

Multivariate logistic regression results for mortality

Admission and length of stay

The overall admission rate was 43.5% in the cohort. Patients living with frailty had significantly higher admission rates compared with robust patients (58% vs 36%, diff 22%, 95% CI 17% to 26%) with an unadjusted OR of 2.4 (95% CI 2.0 to 3.0)(table 2). Patients living with frailty had a higher admission rate and longer lengths of ED and hospital stay (figure 3A,B).

Figure 3

(A) Emergency department length of stay in hours by CFS score; (B) Hospital length of stay in days by CFS score, of all admitted patients in the study cohort. CFS, Clinical Frailty Scale.

Patients living with frailty had longer ED LOS and hospital LOS compared with robust patients (table 2).

Discussion

In this multicentre study conducted in three EDs in Sweden, patients living with frailty (CFS≥5) had a significantly higher risk of death at 7, 30 and 90 days compared with robust patients. Patients living with frailty also had a higher admission rate and longer lengths of ED and hospital stay. These results are in line with previous data when trained research personnel assessed frailty14 26 and confirm that frailty assessments can be made by regular ED staff.

The ORs for mortality changed little when adjusting for confounding predictors of death, like triage acuity, arrival by ambulance and age. This suggests that CFS carries prognostic information beyond common indicators for worse outcomes and may be missed if not considered. Furthermore, the mortality risk increased with the CFS score. Hence, the concept of frailty, using the CFS, are potentially useful tools to risk-stratify older patients in the ED that current triage tools may miss. This has been investigated in a study from Taiwan where Ng et al found the risk of undertriage in older adults in the ED decreased when incorporating CFS into the triage assessment.23 Kaeppeli et al also found that CFS outperformed the Emergency Severity Index model in predicting 30-day mortality.26

A considerable proportion of the observed mortality is likely to be non-preventable.24 Therefore, interventions implemented based on frailty should not mainly aim to prevent death but to possibly preserve function and improve quality of life. Higher frailty scores could point out the need for Comprehensive Geriatric Assessment (CGA) in the ED, which assesses fall risk more accurately25 and has been shown to reduce the progress of frailty through early initiation of rehabilitation.27 Early identification of the ED patients with the highest risk of mortality, may allow ED staff to prioritise establishing realistic goals of care and avoid possible over-investigation and inappropriate interventions28 and perhaps also earlier initiation of medical workup.

Further research should focus on investigating ways to decrease barriers to frailty assessments, possible interventions in the ED to improve the quality of care for older patients living with frailty and potential interventions to reduce ‘avoidable’ mortality, possibly by targeting high-risk older patients by means of adding frailty assessment to triage scores.

Limitations

Although 50% of eligible patients arriving during the study period did not have a CFS, the completion rate of assessment (50%) compares well to previous feasibility studies.29 We previously found that high workload, critical illness and simply forgetting to do the assessment were the most common reasons patients were not assessed.30 There may also be selective recruiting by staff. In this study, eligible but not assessed patients had a slightly lower mean age, similar to the age of robust patients. Staff may be less inclined to include younger or robust-appearing patients as the intention of the CFS is to identify frailty rather than robustness. Since the majority of included patients were robust, including more robust-appearing patients might have affected the overall mortality rate, but not the difference between groups.

The mortality rate was lower compared with the estimated rates in the sample size calculation. However, the difference between the groups was also larger than expected, and so the net effect on the power of the study was negligible.

We did not collect data on staff experience with the CFS prior to our study, and we did not have an expert reference for the assessments, which could affect the validity and reproducibility of CFS scores. However, the aim of this study was to investigate the association of CFS with adverse outcomes in a clinical ED setting where the amount of staff training familiarity, and awareness of specific procedures will inevitably vary.

The CFS assessment was not done at a prespecified time in the ED. If completed prior to admission, the CFS could have potentially affected the admission decision. However, the CFS was not a routine part of the admission process before or during the study and the overall admission rate in the study did not differ from the historical admission rate in this patient group which indicates little or no impact.

Conclusion

In this multicentre study from Sweden, patients living with frailty, as assessed with CFS by ED providers, had notably higher hospital, 7-day and 30-day mortality. The rate of admission and ED and in-hospital LOS were also higher compared with robust patients. This suggests that frailty can be assessed routinely in the ED and may be a useful additional tool in risk-stratifying patients.

Abstract translationThis web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.Data availability statement

Data are available upon reasonable request. There is no plan to share individual participant data. Personal data related to this study is available upon request. Electronic data is stored in a protected network storage space. The worksheets are stored in a locked space without access for unauthorised personnel.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

Name of the Ethics Committee: Swedish Ethical Review Authority. Reference no: 2021-00875. Informed consent was waived by the Ethical Review Authority as Clinical Frailty Scale assessment already was introduced into clinical routine.