Patient versus Physician Perceptions of Frailty: A Comparison of Clinical Frailty Scale Scores of Older Adults in the Emergency Department

Frailty is a state of vulnerability to impairments in mobility, balance, muscle strength, motor processing, physical function, disability, cognition, nutrition, endurance, and physical activity.¹ Prevalence estimates vary widely from 4.0 to 59.1% (weighted estimate 10.9%), with prevalence increasing with age.² The risk of adverse outcomes for frail older adults including falls, delirium, disability and death is increased compared to similarly aged individuals without frailty, and may be a more important predictor of death than age for older adults.³

Older adults account for nearly 20 million emergency department (ED) visits in the US annually with volumes increasing as the population ages.⁴ Frailty is important for emergency physicians (EPs) to identify and understand because of its with implications on treatment decisions. For patients, being labelled as frail may contribute to the development of a frailty identity and lead to a loss of interest in participating in social and physical activities, increased stigmatization, and poor self-reported physical health.⁵

It is unknown if EPs and patients identify and interpret frailty the same way. Additionally, it is unknown if there is an association between patient or EP perception of frailty and hospitalization which carries risks of complications like infection, delirium, and functional decline. The objectives of this study were to determine (1) if emergency physicians and older adults differ in their perceptions of frailty and (2) if those perceptions of frailty were related to disposition from the ED.

As a planned secondary analysis of a larger study, we performed a cross-sectional survey of community dwelling older adults 65 years old and older at two large U.S. EDs from July 2016 to April 2017.⁶ Patients were eligible if they understood English and were medically and cognitively able to participate, and had not previously participated in this study. Research assistants (RAs) identified patients using the ED’s electronic tracking board and confirmed eligibility with treating clinicians. RAs discussed the survey with patients and assessed patients’ ability to participate. The survey was self-administered or RA-administered based on patient choice. Patients received pamphlets connecting them with local resources. Completed surveys were recorded in Research Electronic Data Capture. The Colorado Multiple Institutional Review Board and the University of North Carolina-Chapel Hill Institutional Review Board approved this project.

The survey included questions regarding demographic, social, and health characteristics including age, gender, living arrangement, social connections, health services use, ED arrival method, ability to perform activities of daily living, and the Clinical Frailty Scale (CFS).⁶ The CFS is a global clinical measure of fitness and frailty which uses pictographs and clinical descriptions to help clinicians stratify older adults on a nine point scale, one being very fit and nine being terminally ill.¹ Self-reported CFS and EP-assigned CFS scores were recorded. Additional variables included anticipated disposition according to the treating clinician, and method of arrival to the ED.

The primary outcome was agreement between patient-reported and EP-assigned CFS score. Secondary outcomes included differences in CFS scores by demographic, clinical, and social variables; and differences in anticipated disposition by CFS score.

Median CFS score was calculated. Wilcoxon rank-sum test was used to test for differences in patient-reported CFS score categories and EP-assigned CFS score categories across categories of demographic, clinical, and social variables. Agreement was determined using weighted Cohen’s kappa. Histograms were created for each CFS score.

Surveys were completed by 266 patients, 194 at site 1, 72 at site 2. Patients reported significantly more CFS scores of 1–2 (very fit or well), while EPs reported significantly more CFS scores of 3 (managing well) or 6–8 (moderate to very severe frailty) (Figure 1). There was moderate agreement (weighted kappa=0.43) of frailty between EPs and patients, site 1 had fair agreement (weighted kappa=0.36), site 2 had moderate agreement (weighted kappa=0.56).

Figure 1:

Patient-reported Clinical Frailty Score compared to Emergency Physician Assigned Clinical Frailty Score

There were no significant differences in results of any variables, including CFS scores by site. Median patient-reported CFS scores were higher for patients who were older (65–74 = 3, 75–84 = 3, 85–92 = 4, p=0.007), did not work or volunteer (yes=3, no=4, p<0.001), did not go into the community regularly (yes =3, no=4, p<0.001), did not drive (yes=3, no=4, p<0.001), ate alone (yes=4, no=3, p=0.006), , had difficulty paying bills (yes=4, no=3, p=0.042), were hospitalized in the past month yes=4, no=3, p=0.007), needed help with routine care (yes=4, no=3, p<0.001), needed help with personal care (yes=6, no=3, p<0.001), needed special equipment (yes=4, no=3, p<0.001), or did not live in private housing (private=3, other=4, p=0.006). There were no significant differences in patient-reported CFS score with regards to: gender, difficulty affording food, having a primary care physician, ED arrival method, and patient disposition. Median EP-assigned CFS scores were higher for patients who were older (65–74=3, 75–84=4, 85–92=4.5, p=0.001), did not live in private housing (private=3, other=5, p=0.009), did not work or volunteer (yes=3, no=4, p<0.001), did not go into the community regularly (yes=3, no=4, p=0.001), did not drive (yes=3, no=5, p<0.001), were hospitalized in the past month (yes=4, no=3, p=0.001), needed help with routine care (yes=4.5, no=3, p<0.001), needed help with personal care (yes=5, no=3, p<0.001), needed special equipment (yes=4, no=3, p<0.001). There were no significant differences in physician-assigned CFS score with regards to: gender, living situation, seeing family or friends regularly, eating alone regularly, difficulty affording food, difficulty paying bills, having a primary care physician, and ED arrival method.

There was no significant association between planned disposition and self-reported frailty (p=0.08). However, higher EP-assigned CFS was associated with planned hospitalization or transfer to a nursing or rehabilitation facility (admission=4, nursing/rehabilitation facility=5.5, home=3, uncertain=3, p=0.014)

We identified moderate agreement in assessment of frailty between EPs and ED patients age 65 and older as measured by the CFS. We found that EPs were less likely to rate older adults as very fit or well and more likely to report patients as moderately to very severely frail compared to patient self-report. Living alone, eating alone regularly, or having difficulty paying bills was associated with increased self-reported frailty, but not EP-reported frailty. Conversely, self-reported CFS was not associated with disposition, while while higher EP-assigned CFS was associated with hospitalization or nursing home or rehabilitation facility transfer.

Measures of frailty like the CFS may help to identify older adults in the ED at risk of poor outcomes and who may benefit from additional services. It is unclear however, who is the most appropriate rater of frailty. A study of older adults in the ED with blunt trauma found poor agreement between the patient-reported and EP-assigned CFS. Patient-reported CFS had better agreement with the patient-reported Fatigue, Resistance, Ambulation, Illnesses, & Loss of Weight (FRAIL) scale, which is also patient-reported.⁷ However, it is not clear whether patient-reported CFS is more accurate at predicting important short term outcomes like functional decline than EP-assigned CFS.

As EDs face additional pressure from an aging population and continued ED crowding, quick assessments of frailty like the CFS may help identify patients who would benefit from a more thorough assessment by a multi-disciplinary team and may be useful in providing insight into patients’ environment (including potential challenges) and sense of well-being. It is currently unclear which of the many frailty scores is best for use in the ED. Since there are many different constructs of frailty, identifying the best measure to use and the optimal assessor depends on the outcome of interest. The Study of Osteoporotic Fracture (SOF) frailty index has been shown to have at least fair prediction of functional decline.⁸ Additionally the CFS, Fried Frailty Index, and the Stable, Unstable, Help, Bedridden (SUHB) scale all had good predictive properties for poor outcomes after an ED visit, however the CFS was easier to administer in the ED.⁹ Additional study is needed to determine if patient or clinician reported frailty scores are more accurate at predicting poor outcomes.

The major limitation of this study is that it is unknown whether EP-assigned CSF scores or patient-reported CSF scores are more accurate at predicting short term outcomes after an ED visit, and researchers did not separately administer a “gold standard” objective assessment of frailty. Additionally, there are questions as to the reliability and validity of the CFS.¹⁰ Other limitations include convenience sampling, though RAs made every effort to identify all eligible patients, it is possible that bias was present in the selection process. Disposition may have changed after anticipated disposition was recorded. Finally, the secondary outcome of hospitalization is ultimately determined by the emergency physician, who also peformed the CFS. Future studies should use clinical outcomes which are independent of the rater’s decision making such as funcational decline, return to the ED, re-hospitalization, and nursing home placement over a predermined followup time.

Agreement between patient-reported and EP-assigned CFS is moderate for older adults in the ED; discordance was particularly notable at the less-frail end of the spectrum. Hospital admission or transfer to a nursing home or rehabilitation facility appeared associated with increased EP-assigned CFS but more detailed and controled analyses would be useful to determine whether patients underestimate frailty or EPs overestimate it.