Prediction of one-year mortality by five different frailty instruments: A comparative study in hospitalized geriatric patients

Highlights

• A prospective analysis in a cohort of hospitalized geriatric patients.

• We compared the ability of five frailty scales in predicting one-year mortality.

• All five frailty scales predicted one-year mortality.

• The Clinical Frailty Scale was the most powerful predictor for mortality.

• Using these frailty scales might improve decision making in geriatric inpatients.

Abstract

Background

Data comparing the ability of different major frailty instruments for predicting mortality in hospitalized geriatric patients are scare.

Material and methods

307 patients ≥65 years who were hospitalized on geriatric wards were included in this prospective analysis. A fifty-item frailty index (FI), a ten-domain + co-morbidity frailty index based on a standardized comprehensive geriatric assessment (FI-CGA), the nine category Clinical Frailty Scale (CFS-9), the CSHA rules-based frailty definition (CSHA-RBFD), and the frailty phenotype (FP) were assessed during the patients’ hospital stays. Patients were followed up over a one-year period.

Results

Follow-up data after one year could be obtained from 305 out of the 307 participants. Sixty two participants (20.3%) had died after that time. The FI, FI-CGA, CFS-9, CSHA-RBFD, and FP could all discriminate between patients who died and those who survived during follow-up (areas under the ROC curves: 0.805, 0.808, 0.852, 0.703 and 0.757, all P < 0.001, respectively). The CFS-9 showed a better discriminative ability for one-year mortality compared to the FI, FI-CGA, CSHA-RBFD, and FP (all P < 0.05, respectively). The FI and the FI-CGA did not differ in their discriminative ability for one-year mortality (P = 0.440). The CSHA-RBFD and the FP demonstrated a comparable discriminative ability (P = 0.241) and, when compared to the CFS-9, FI, and FI-CGA, an inferior discriminative ability for one-year mortality (all P < 0.05, respectively).

Conclusion

Among those frailty instruments that were evaluated, the CFS-9 emerged as the most powerful for prediction of one-year mortality.

Introduction

Frailty denotes a condition of increased vulnerability to stressors due to a reduction in physiological reserves of multiple organ systems (Morley et al., 2013). Frail old people show an increased incidence of negative health outcomes, including mortality, compared to non-frail ones (Morley et al., 2013). Several approaches for measuring frailty have been developed and evaluated, which are mainly based on the work of Rockwood and colleagues, as well as Fried and colleagues (Bouillon et al., 2013, Morley et al., 2013). Rockwood and colleagues utilized the classic frailty index based on multiple, individual health deficits (Mitnitski, Mogilner, & Rockwood, 2001; Rockwood & Mitnitski, 2012; Rockwood, Rockwood, & Mitnitski, 2010; Searle, Mitnitski, Gahbauer, Gill, & Rockwood, 2008), the frailty index based on severity of impairment of different functional domains of a standardized comprehensive geriatric assessment, and co-morbidity burden (Jones, Song, & Rockwood, 2004; Jones, Song, Mitnitski, & Rockwood, 2005), the Clinical Frailty Scale (Rockwood et al., 2005), and the Canadian Study of Health and Aging (CSHA) rules-based frailty definition (Rockwood et al., 1999). Fried and colleagues developed and evaluated the frailty phenotype (Fried et al., 2001).

These frailty instruments reflect different concepts. Rockwood and colleagues’ classical frailty index is based on multiple (30 or more) individual potential health deficits, i.e., items (Searle et al., 2008). It represents an arithmetical approach which is independent of pre-set items (Searle et al., 2008) and can be constructed from different data sets, for example, from data of a CGA (Rockwood and Mitnitski, 2012, Searle et al., 2008). Another approach, developed and evaluated by Rockwood and colleagues, is to assess, in a standardized fashion, the level of impairment in the functional domains of a standardized, comprehensive geriatric assessment, together with the co-morbidity burden of the patients (Jones et al., 2004, Jones et al., 2005). The Clinical Frailty Scale (Rockwood et al., 2005) takes physical activity, functional impairment, comorbidity (including dementia severity), and the patient’s remaining life expectancy into account. The CSHA rules-based frailty definition (Rockwood et al., 1999) is a measure based on functional criteria and the cognitive status of the patients. In contrast, the frailty phenotype (Fried et al., 2001) is based on five physical phenotypic criteria.

In geriatric medicine, estimation of the patient’s remaining life expectancy is of relevance to clinical decision making. Patients on geriatric wards often show multiple co-morbidities and functional impairment. In these patients decisions with respect to diagnosis, treatment and intervention are frequently undertaken in absence of strong evidence base (American Geriatrics Society, 2012; Tinetti, Bogardus, & Agostini, 2004). This might result in inappropriate management in some old patients (American Geriatrics Society, 2012). Some patients might be subject to overtreatment or adverse effects of interventions that causes distress at the end of their lives. Other patients, who would potentially benefit from intervention, might, however, not undergo such interventions solely on the ground of their advanced chronological age alone. Thus, in older patients, analysis with respect to frailty and frailty degree, which is an indicator of biological age (Mitnitski et al., 2001), is, amongst other aspects, of interest in terms of estimation of the individual patient’s remaining life expectancy.

Clearly, all aforementioned, major frailty instruments have been found to predict mortality (Fried et al., 2001; Jones et al., 2005; Mitnitski et al., 2001; Rockwood et al., 1999, Rockwood et al., 2005, Rockwood et al., 2010; Wallis, Wall, Biram, & Romero-Ortuno, 2015). However, the predictive power of these frailty instruments may differ. Of note, data comparing the ability of different frailty instruments to predict mortality in hospitalized geriatric patients are scarce (Pilotto et al., 2011, Ritt, Radi et al., 2015, Ritt, Schwarz et al., 2015). In particular, no study, up until now, has compared the ability of all the aforementioned different major frailty instruments, which are based on the work of Rockwood and colleagues, as well as Fried and colleagues, together and in parallel to predict mortality in a cohort of hospitalized geriatric patients. Against this background, our study aim was to analyze and compare the ability of five different major frailty instruments (i.e., a fifty-item frailty index, ten-domain + co-morbidity frailty index based on a comprehensive geriatric assessment (FI-CGA), the Clinical Frailty Scale, the CSHA rules based frailty definition and the frailty phenotype), which reflect aforementioned different concepts, together and in parallel, to predict one-year mortality in old patients who were hospitalized on geriatric wards.