Frailty Criteria Give Heterogeneous Results When Applied in Clinical Practice

To the Editor: Frailty is an important concept in geriatrics. It is a strong predictor of negative outcomes: new disabilities, institutionalization, and mortality. Interventions targeted at frail elderly patients may prevent or delay these adverse outcomes and are a precondition for efficient geriatric healthcare services. Several frailty criteria have been tested in epidemiological research and are now ready to use in clinical practice, but no criterion standard exists to diagnose frailty correctly in a heterogeneous group of elderly patients.1,2 Four often-used frailty criteria were investigated for their feasibility and effect on the selection of frail elderly patients in clinical practice. Patients admitted consecutively to an acute geriatric or old age psychiatry ward were assessed to determine whether they were frail according to often-used frailty criteria: Rockwood (≥2 items of cognitive decline or disabilities in activities of daily living or urinary incontinence),3 Fried (≥3 items of exhaustion, weight loss, slow gait velocity, low handgrip strength, low physical activity),4 low handgrip strength (women <17.5 kg; men <30 kg),5 and gait velocity less than 1.0 m/s.6 The human research ethics committee at the Radboud University Nijmegen Medical Center approved the study. One hundred twenty-five (72 women) of 201 patients admitted to Nijmegen Medical Center were eligible and participated. Their mean age±standard deviation was 77.3±7.4. Main reasons for nonparticipation were inability to walk 10 meters (n=33), no understanding of simple instructions to walk (n=19), or both (n=13). Eleven patients refused participation. The included patients used on average 6.6 drugs; 63 (50.4%) had dementia, 69 (55.2%) had fallen in the previous year, and 69 (55.2%) had a fear of falling. Mean gait velocity was 0.62±0.30 m/s, and median time to perform the TUG was 16.9 seconds (range 4.9–133.8 seconds). The prevalence of frailty in this population varies from 36% (handgrip strength) to 48% (Rockwood) to 62.4% (Fried) to 88.8% (gait velocity). The patient populations identified by these criteria overlapped partially (Figure 1). Only 29 subjects were considered frail according to all four frailty criteria. The kappas for concordance in classification of patients as frail or not were poor to moderate and ranged from 0.18 (criteria by Fried and Rockwood) to 0.48 (criteria by Fried and handgrip strength). Furthermore, the distributions of other single risk factors for negative outcomes also varied. The prevalence of dementia, for example, ranged from 51% to 73% in the different groups of frail patients, compared with 22% to 45% in nonfrail patients. The differences in prevalence of dementia in frail and nonfrail patients were significant (P<.05) for Rockwood criteria and handgrip strength but not for criteria by Fried and gait velocity. Distribution of subjects (n=113) identified as frail according to four frailty criteria (Rockwood,3 Fried,4 slow gait velocity,6 and low handgrip strength5) in a group of elderly hospitalized patients (N=125). The use of frailty criteria is feasible in a group of geriatric inpatients, but the various criteria give highly different selections of patients. Before frailty criteria can be used to select frail elderly patients, each of the available criteria should be prospectively validated for this purpose. Until then, the choice of the most appropriate frailty criterion should be based on the purpose, the outcome on which the criterion was originally validated, the quality of the validation process carried out so far, and the similarity of the current population to the validation group. Further scientific work on frailty as a targeting criterion to select patients at high risk for adverse outcomes is urgently needed to give this promising paradigm a solid place in medicine for older people. Financial Disclosure: None. Author Contributions: Marianne van Iersel contributed to study concept and design, data acquisition and analysis, and preparation of the manuscript. Marcel Olde Rikkert contributed to study concept and design, data analysis, and preparation of the manuscript. Sponsor's Role: Not applicable.