Introduction

Frailty is a physiological state where small perturbations in health result in disproportionate adverse effects due to an underlying decline in reserve of multiple physiological systems [1,2,3]. It is common in older Emergency Department (ED) patients with reported prevalence rates between 21 and 62 [4,5,6,7]. Frailty is associated with a wide range of adverse outcomes, including mortality [8], hospitalization [9], delirium [7] and diminished quality of life [10]. People often present to the ED due a change in health status, this offers a unique opportunity to alter their health trajectory. To meet the needs of the growing population of older adults with frailty presenting to the ED, there is advocacy for the integration of ED frailty evaluation [11, 12]. However, the benefit and harms associated with frailty screening in the ED are largely unknown [13, 14]. Furthermore, frailty identification in the ED is not common [15]. Cited barriers included feasibility of tools in the time pressured ED environment, lack of formal clinical frailty guidelines for the ED and geriatric expertise [11, 13, 15, 16].

Previous scoping reviews on frailty in the acute care setting have included multiple medical disciplines including geriatrics, emergency medicine, general medicine, cardiology and orthopedics [14, 17]. Van Dam et al. recently completed a narrative review of frailty assessment in the ED [18]. They focused on the predictive accuracy of frailty screening tools, the use of clinical gestalt to determine frailty, and the rationale for and implementation of frailty assessment in the ED. However, some of included studies have used tools that were initially designed to predict risk of adverse outcome (ie ISAR, TRST) and not frailty specifically [5, 19].

There are 89 different measures that have been used to evaluate frailty in the acute care literature [20]. The Clinical Frailty Scale (CFS) is one of the most commonly used tools. The CFS was initially a seven-point scale used as a judgment-based tool to assess frailty [21]. In 2007, it was expanded to a nine-point scale, from 1 (very fit) to 9 (terminally ill) (Fig. 1). Compared to other frailty tools, the CFS seems to be the ideal choice for measuring frailty in emergency medicine, because it is easier and faster to use, without giving up any prognostic accuracy [22]. There are no studies that exclusively synthesize information on the use of CFS in emergency medicine. This scoping review is intended to fill this gap, by focusing strictly on the CFS literature in the emergency medicine setting. We aimed to describe the current evidence and identify gaps in knowledge including: version of CFS, timing of CFS evaluation, who is completing the evaluation, goals of frailty evaluation, the prevalence of frailty, and the outcomes associated with frailty identification using the CFS.

Fig. 1
figure 1

The Clinical Frailty Scale

Fig. 2
figure 2

Flow diagram

Materials and methods

A protocol for this scoping review was developed and published on the Open Science Framework, where the study was registered before performing the search strategy (https://doi.org/10.17605/OSF.IO/W2F8N) [23]. We have followed the PRISMA-ScR Statement for reporting scoping reviews [24].

Eligibility criteria

Based on the population, concept, and context (PCC) framework for scoping reviews [25], inclusion criteria were: (1) adult (≥ 18 years) population; (2) use of the CFS; (3) emergency medicine setting (intra-hospital or pre-hospital); and (4) original research. We did not language restrict.

Studies not reporting frailty or reporting frailty using another tool (such as Fried [26], ISAR [27]) exclusively were excluded. We also excluded conference abstracts, editorials, commentaries, position papers, narrative and systematic reviews, and case studies, that did not report on original research.

Search strategy

The MEDLINE search strategy was developed by a health science librarian and peer-reviewed by another librarian [28]. Databases searched were MEDLINE(R) ALL via Ovid, Embase Classic + Embase via Ovid, EBM Reviews—Cochrane Central Register of Control Trials via Ovid, CINAHL via EBSCOhost, Ageline via EBSCOhost, and Scopus. The main search concepts were comprised of terms related to emergency department or pre-hospital settings and frailty. The date of publication was limited from 2005 to 2021. This limit was applied as the Clinical Frailty Scale (CFS) was introduced in 2005. The search strategy was developed in MEDLINE (Appendix 1) and translated to other databases. All databases were searched on July 6th, 2021. Additionally, a manual search of all eligible articles’ reference lists was completed to identify any additional literature.

Selection of source of evidence

Search results were imported into Covidence and de-duplicated [29]. Screening and data abstraction were also completed in Covidence. First, team members screened a sample of 50 citations. Conflicts were reviewed and discussed. As the agreement on the pilot test was low (< 90%), another pilot was performed, with success. Then, two reviewers independently screened all remaining citations. Disagreements were resolved by consensus. Second-level screening was performed using a similar strategy (pilot, double independent screening). The study screening form can be found in Appendix 2.

Data charting process and data items

Data were abstracted, using a pre-specified data abstraction form. To ensure consistency between reviewers, all reviewers initially abstracted the same five citations. Any discrepancies were resolved by consensus. The form was then adapted (Appendix 3), and data abstraction was completed independently by two reviewers. We collected data on publication characteristics (authors, country, year of publication, journal), study characteristics (design, sample size, setting, patients’ age and sex), frailty [version of CFS used, cut-off used to define frail people, type of categorization of CFS, purpose of the assessment (outcome, screening, descriptive, exposure, covariate, potential predictor), assessor, prevalence of frailty] and outcomes under study. When composite outcomes were studied, we collected each outcome of the composite outcome individually.

Critical appraisal of individual sources of evidence

As the main goal of this study was to report on the contextual features of frailty in emergency medicine literature, no critical appraisal was performed on the individual studies.

Synthesis of results

Results of the search and the screening process are presented using a flow diagram. Outcomes were grouped according to essential themes for the purpose of analysis.

Results

Figure 2 presents the study flow diagram. From the 7164 records, we identified 4757 unique citations after deduplication. Sixty-one studies were also identified from references of included articles. Following first-level screening, 4575 were deemed irrelevant. Second-level screening excluded a further 209 citations. Thirty-four manuscripts (33 full manuscript and one research letter) underwent complete data abstraction and are presented in this manuscript (Appendix 4). No potentially relevant studies were excluded.

Table 1 presents characteristics of the included studies. All studies were published in English and the primary author affiliation was mainly from North America [7, 30,31,32,33,34,35,36,37,38,39,40,41,42,43] (44%) and Europe [44,45,46,47,48,49,50,51,52,53,54,55] (35%). No papers had been published before 2015, and most of the papers (76%) were published beginning 2019. Studies were published in emergency medicine journals (41%) [7, 30, 32, 33, 36, 40, 46, 48,49,50, 53, 54, 56, 57], geriatric journals (38%) [31, 34, 35, 38, 39, 41, 42, 45, 47, 56, 58, 59] or other types of journals (21%) [37, 43, 51, 52, 55, 60, 61].

Table 1 Summary of study characteristics, N = 34

Two-thirds of the studies were prospective cohorts [7, 30, 33, 35,36,37,38, 40, 42, 43, 47,48,49,50, 52, 54,55,56,57,58,59, 62], while the remaining were retrospective cohorts (24%) [34, 39, 41, 45, 46, 53, 60, 61], intervention studies (9%) [31, 44, 51] or cross-sectional studies (3%) [32] (Table 1). One study [45] was performed in pre-hospital setting only, and another one [43] included both pre-hospital and ED patients. Overall, the median sample size was 612, with an important variability from one study to the other (IQR 330–1309). The median or mean age varied between 75 and 85, while the proportion of female patients varied between 36 and 77%. Patient consent was required in 20 studies and not required in four studies [36, 45, 53, 54]. The 10 remaining studies [32, 34, 41, 43, 46, 50, 51, 59,60,61] did not mention patient consent.

The majority (74%) of the studies used the nine-point CFS [32,33,34, 36, 39,40,41, 44,45,46, 48,49,50,51,52,53,54,55,56,57,58,59,60,61,62]. For three studies [30, 43, 47], it was not possible to assess which CFS version was used. Only two studies excluded patients with CFS score of nine. [33, 49] Thirteen studies reported frailty prevalence, with a median (using authors’ cut-off) of 36.8% (IQR 31.8–57.6). Frailty was assessed mostly during patient work-up (65%) [31,32,33, 35,36,37,38, 40, 42, 44, 45, 47,48,49, 51, 53, 55, 56, 58, 59, 61, 62], while some authors assessed it at triage (18%) [41, 46, 50, 52, 54, 60], at patient disposition (9%) [7, 34, 57] or at other times (9%) [30, 39, 43]. Table 2 shows the different types of assessors. Research staff (35%) [7, 30, 31, 35, 37, 38, 43, 47, 49, 56, 59, 62], nurse (32%) [36, 40, 41, 44, 46, 48, 50, 52, 54, 58, 60] and ED physician (20%) [32, 33, 36, 40, 42, 46, 57] were the most frequent.

Table 2 Person completing Clinical Frailty Scale assessment

CFS was most commonly used as a main exposure (44%) [7, 33, 37,38,39, 41, 42, 46, 49, 52, 53, 55, 57, 60]. Other frequent purposes included potential predictor (15%) [30, 35, 45, 56, 62] and outcome (15%) [32, 36, 40, 43, 48]. Only two studies used it as an eligibility criterion. When CFS was used as a main exposure or a predictor (20 studies), the most frequent studied outcomes (either alone or in composite) were mortality (10 studies, 50%) [33, 39, 46, 49, 55,56,57,58, 60, 62] and hospital admission (7 studies, 35%) (Table 3) [33, 35, 41, 49, 53, 55, 60]. For mortality, several time points were used, including 1 month [33, 39, 49, 55, 57, 60], 3 months [56, 62] or 1 year. Three papers used it as a time-to-event variable [39, 46, 49]. Four papers considered patient-oriented outcomes (alone or included in a composite outcome), such as quality of life [37, 58], functional decline [38, 42] or need for community service following discharge [58]. In the case of use as the main exposure, a sample size calculation was reported only in three studies [7, 49, 52]. Different methods to deal with the CFS variable as exposure or predictor were used for the statistical analysis: binarization (35%) [7, 33, 38, 55, 56, 58, 62], categorisation in 3 or more groups (30%) [35, 37, 39, 42, 46, 49] or continuous (20%) [41, 45, 53, 57]. One study [60] used different methods and two studies [30, 52] did not mention their analytic approach. Among the 15 studies looking for an association between a main exposure and an outcome, only 3 (20%) mentioned a sample size calculation [7, 49, 52]. Finally, these 15 studies found a statistically significant association. Three studies did not incorporate any covariate in the model [41, 42, 52]. For the other ones, age (10 studies [7, 33, 38, 46, 49, 53, 55, 57, 58, 60]), sex or gender (9 studies [33, 38, 46, 49, 53, 55, 57, 58, 60]) and comorbidities (7 studies [37, 38, 46, 53, 57, 58, 60]) were the most frequent covariates used for adjustment (Table 4).

Table 3 Reported study outcome measures
Table 4 Adjusting variables,

Discussion

We conducted a scoping review that explored the use of the CFS in adult patients in emergency medicine. We found there is increasing use of the CFS in the emergency setting. Most of the studies using it have been published in recent years. The revised version of the CFS with nine points was the most frequently used; however, the purpose and timing of the CFS, who performed the assessment and the analytic approach differed between studies. The cut-off used to define frailty not reported in almost half of studies and the most frequent use of CFS was as an exposure, to look at an association with an outcome.

Our study adds to the work of Church et al., and van Dam et al. [18, 63]. Van Dam et al. completed a narrative review of frailty assessment in the ED. Their study evaluated multiple tools and only included three studies that used the CFS. Church, on the other hand, focused exclusively on use of the CFS, but only six were in the ED. While there are some similarities, including trend over time, assessors and outcomes under study, our findings contribute significantly to our understanding of the current use of the CFS in the ED, as we focused on the ED setting and we examined additional characteristics, such as consent and statistical analysis.

This research showed that consent was required for study inclusion most of the time. While we acknowledge the importance to seek patient consent to participate in a study, studies looking at the impact of frailty assessment or association with outcomes that exclude patients that cannot give informed consent are at risk of, in the very least, limiting the generalizability of the results but in the worst case biasing their results. The impact of patient selection based on consent on study results has been shown in other vulnerable populations, including patients with delirium and stroke [64, 65]. As there appears to be a relation between frailty and ability to give informed consent, the risk of bias in this patient population is high [66]. Therefore, it would be optimal to get a waiver of consent for minimum risk studies.

Another important finding of this study is suboptimal reporting regarding CFS. It was occasionally difficult to determine who completed the CFS assessment, when the assessment took place, which version of the CFS was used or how the CFS was considered in the analysis. A lack of standardized reporting is a crucial issue in research as it could impact interpretation and reproducibility of results [20].

Regarding the analysis, our study highlights several issues that should be mentioned. Studies that reported frailty prevalence or used frailty as a binary variable in their analysis, did not use a consistent CFS cut-off, some authors used four and more whereas other authors used five and more, likely because of the recent change of wording (“vulnerable” to “very mild frailty”). Although binarization is never the best solution, there needs to be consensus regarding a standardized cut-off if the CFS is to be dichotomized. While many studies consider frailty as a binary variable, some authors used it as a continuous one. Such analysis should be performed with caution as it is unlikely that regression fundamental assumptions would be met, such as linearity of the log-odds. Using categories, or even more advanced methods such as restricted cubic spline, could improve the rigor in this part of a study [67]. Almost all authors chose to adjust the main association. Age and comorbidities were frequently chosen. It can be argued that, because the CFS is a multi-faceted tool, incorporating already such aspects, there is a risk of collinearity.

Some limitations of this scoping review should be acknowledged. Our search strategy was developed for our specific question, however there is the possibility that studies could have been missed, especially studies with CFS used as inclusion criteria, baseline characteristics or covariates as they are frequently not mentioned in the abstract. Therefore, the results regarding the purpose of the CFS assessment in the ED could be biased, with a risk of underestimating the use of CFS for those purposes. We decided a priori to include only studies with patients, as our goal was to see how the CFS was used in the ED. There are, however, some papers on the reliability of the CFS that were based on clinical vignettes. Those studies were excluded. Finally, to ensure the homogeneity of our results, we excluded papers that included both ED patients and ward patients, as the finding could have biased our results, if the CFS was not assessed in the ED environment.

This scoping review has strengths. To our knowledge, this is the first exhaustive review on the CFS in the ED. The results from this review will help to define future research questions. Secondly, we used rigorous methodology for the sources (several databases, published papers and conferences abstract), the search strategy (more comprehensive than previous studies), the screening (pilot testing, double independently review) and the data extraction. This process reinforces the internal validity of our results. Finally, this scoping review was registered, its protocol is available, and all amendments to this protocol are listed to increase the transparency of our work.

Based on this review, we identify several gaps that could be considered in future research projects. From a global perspective, there needs to be a move toward common data elements (including cut-off point where appropriate) and core outcome measures [68]. Consensus on data elements and outcome measures for the CFS in the ED could be achieved using the Delphi methodology [69]. We identified multiples studies that looked at the association between CFS level and outcomes. Robust synthesis, including bias assessment and meta-analysis should be performed. From a clinical perspective, there are currently few studies looking at the added value of the systematic use of the CFS in the ED. Evaluation of the impact of ED frailty screening with this tool is therefore needed. Studies comparing frailty screening to no screening are required before advocating for a large implementation of frailty screening. Other important questions include who should complete the frailty evaluation and what is the optimal timing of frailty assessment during the ED course. While it has been shown in the ICU that assessment based on chart review, with family or directly to the patient were quite similar [70], the research on this issue within emergency medicine is scarce. It is likely that assessing frailty at triage versus at disposition could have a different impact. Finally, we found only one study performed exclusively in the pre-hospital setting. When paramedic attend at patients’ home, they could have a better perspective of their environment and could therefore have a more accurate assessment of their frailty.

In summary, this scoping review found increasing use of the Clinical Frailty Scale in studies with adults presenting to the ED. The majority of studies used it as a predictor for adverse outcomes, most commonly admission to hospital and mortality. The quality of the reporting in future studies must be improved. Future research should look at how patients can benefit from its use in the ED and when, how and by whom the CFS should be used.