Background
Frailty is a state of vulnerability to poor resolution of homoeostasis after a stressor event and is a consequence of loss of reserve across multiple physiological systems which occurs across a lifetime [
1]. Frailty can be used to identify older adults who are at increased risk of mortality and functional decline when they are hospitalized [
2], but there is no consensus on the most appropriate way for non-geriatricians to identify frailty at the time of hospital admission [
3]. Traditionally subjective opinion has been used by non-geriatricians to identify frailty, but this correlates poorly with objective measures of frailty [
4].
The frailty phenotype [
5] and the frailty index [
6] have both been validated against adverse outcomes in large community cohorts. Fried’s frailty phenotype requires measurement of gait speed [
5], which is likely to be affected by an acute illness. Comprehensive geriatric assessment (CGA) is a multidimensional patient assessment examining medical, psychological, nutritional, cognitive and functional domains [
7]. CGA can decrease mortality and length of stay for hospitalized older adults [
7]. The frailty index based on a comprehensive geriatric assessment (FI-CGA) at the time of hospital admission predicts increased risk of mortality and need for residential care [
3,
8], but requires geriatrician input. CGA is time and resource intensive and it is not feasible to provide this for all patients who present to hospital at this time. Screening for frailty by non-geriatricians may identify patients most likely to benefit from a CGA [
3]. It may also help non-geriatricians with prognostication.
The Clinical Frailty Scale (CFS) [
9] was developed to enable frailty to be measured in the outpatient clinical setting [
10]. It has demonstrated very good inter-rater reliability [
10,
11]. When used by trained assessors it predicts short-term and long-term mortality in acutely hospitalized older adults [
11‐
16] grouped as frail or not frail. A large retrospective cohort study demonstrated that increasing frailty on the CFS has a linear relationship with inpatient mortality and increased length of stay [
17]. The CFS is an attractive tool as it can be completed based on routine clinical admission and there is no need for extra equipment, so there are minimal barriers to its’ implementation. Although other frailty measurement tools, such as the Reported Edmonton Frailty Scale (REFS) have been validated against the Geriatrician’s Clinical Impression of Frailty in the inpatient setting, the REFS has features which limit its’ use in patients who do not speak English, or who are hearing or vision impaired [
18].
The objective of this study is to determine the predictive validity of the CFS when used by untrained junior medical staff in the acute general medical setting using only routine clinical information.
Discussion
This study demonstrates the feasibility of using the Clinical Frailty Scale in the acute general medical setting. The CFS correlates with the important outcomes of death and functional decline. This is the first study the authors are aware of, where no training was provided to junior doctors in order to examine how the CFS functions in a real world setting. This study also shows that this scale is highly acceptable to medical staff as there was a 95 % completion rate. It was completed with information obtained on routine assessment at the time of admission, so the additional workload for junior medical staff was minimal. The combination of acceptability and prognostic guidance supports the role of the CFS as a tool to identify patients most suitable for comprehensive geriatric assessment.
Screening for frailty may act to decrease age related discrimination by identifying robust elderly patients. Screening can also identify the most frail and trigger discussions regarding limitations of treatment.
Other studies have looked at the CFS as a predictor of mortality in the acute hospital setting [
11‐
13,
15,
17]. This study also examines functional outcomes. Failure to return to pre-morbid functional status predicts mortality [
2] and institutionalization [
21]. Patients who are at high risk for functional decline are also at high risk for mortality [
2].
There was no association found between length of stay and frailty score, which is not consistent with other studies [
15,
17]. This may be due to the high prevalence of frailty, meaning that discrimination was lost, as other studies that have examined this association have had lower proportions of frail patients [
3]. In the studies by Wallis et al. [
17] and Evans et al. [
3] it is not clear whether length of stay included subacute care, which was not included in our study.
There is a positive relationship between the degree of frailty and the risk of mortality and functional decline when frailty is measured by FI-CGA [
3]. This has also been demonstrated with the CFS in other studies [
15].
Consistent with previous findings, female gender conferred protection against mortality [
22].
Wallis et al. conducted a retrospective study to determine the association of the CFS with patient characteristics and outcomes. The CFS was completed for all patients aged 75 or older as part of routine care by junior medical or nursing staff, who were provided with training at induction. Despite the lack of training provided for the junior medical staff in our study, the OR for inpatient mortality was 1.6, 95 % CI 1.48 1.74, which was comparable to the three month mortality rate in our study of 1.82, 95 % CI 1.14, 2.91. Similar to our findings, Wallis et al. [
17] demonstrated that the least frail patients had slightly higher mortality than the moderately frail patients. This may be due to patients who are more robust only needing to be hospitalized for more a more severe interceding illness. In a similar population to ours, Basic and Shanley [
15] also found a higher risk of mortality with increasing frailty identified on the CFS.
The study has certain strengths. A high proportion of eligible patients were included. Since individual consent/assent was not required and the CFS could be completed at any time during the hospital stay there were no barriers to recruitment of patients with communication, language or cognitive difficulties or those admitted outside routine working hours. This increases the generalisability of our findings to patients who have barriers to communication.
We also acknowledge methodological weaknesses. This is a single centre study, and so the results may not be applicable to other sites.
The measure of functional decline was indirect, as it was the need for subacute care, the need for increased services on discharge or the opinion of an allied health team member that the patient was below pre-morbid function. As we relied on routine clinical data, there was no direct measurement tool available. Although this is not a validated measure, the proportion of patients who experienced functional decline was similar to other studies in similar settings [
2,
23]. Other studies have used a count of activities of daily living and instrumental activities of daily living as a marker of functional decline [
24], and this is similar to the assessment performed by physiotherapists and occupational therapists. A new need for residential care has also been used as a marker of functional decline [
25]. In this hospital setting it is rare for a patient to be newly discharged to residential care without being admitted to subacute care, so this was deemed a more appropriate measure.
Patients who were from residential care were excluded from the analysis of functional decline, as some of the criteria used to define functional decline were not applicable to this group. The measure used may have lacked sufficient sensitivity to detect functional decline in those who already had low baseline function, for example people receiving full time care from family members. These limitations could be overcome by conducting further research with an objective measure of function at the time of hospitalization and the time of discharge.
We were unable to include some potential confounders in the multivariate analysis. Only information that was routinely collected for patients as part of standard medical, allied health and nursing care was available, so we were unable to obtain a measure of nutrition, cognition or delirium.
As this study was conducted in a real world setting, we were unable to obtain inter-rater reliability. The CFS has previously been demonstrated to have high inter-rater reliability [
11,
26].
A general limitation of frailty measurement in the acute setting, it that it is possible that the level of frailty is over-stated due to the effect of the antecedent illness. Many patients experience functional decline prior to hospital admission [
27], which will lead to a higher frailty score. If the antecedent insult (such as infection, new drug) causes a functional decline resulting in a higher frailty score this still may represent a bad prognostic factor. The only way to examine this would be to look at prospective cohorts recruited in the community.