Measures of frailty in population-based studies: an overview

Background: Although research productivity in the field of frailty has risen exponentially in recent years, there remains a lack of consensus regarding the measurement of this syndrome. This overview offers three services: first, we provide a comprehensive catalogue of current frailty measures; second, we evaluate their reliability and validity; third, we report on their popularity of use. Methods: In order to identify relevant publications, we searched MEDLINE (from its inception in 1948 to May 2011); scrutinized the reference sections of the retrieved articles; and consulted our own files. An indicator of the frequency of use of each frailty instrument was based on the number of times it had been utilized by investigators other than the originators. Results: Of the initially retrieved 2,166 papers, 27 original articles described separate frailty scales. The number (range: 1 to 38) and type of items (range of domains: physical functioning, disability, disease, sensory impairment, cognition, nutrition, mood, and social support) included in the frailty instruments varied widely. Reliability and validity had been examined in only 26% (7/27) of the instruments. The predictive validity of these scales for mortality varied: for instance, hazard ratios/odds ratios (95% confidence interval) for mortality risk for frail relative to non-frail people ranged from 1.21 (0.78; 1.87) to 6.03 (3.00; 12.08) for the Phenotype of Frailty and 1.57 (1.41; 1.74) to 10.53 (7.06; 15.70) for the Frailty Index. Among the 150 papers which we found to have used at least one of the 27 frailty instruments, 69% (n = 104) reported on the Phenotype of Frailty, 12% (n = 18) on the Frailty Index, and 19% (n = 28) on one of the remaining 25 instruments. Conclusions: Although there are numerous frailty scales currently in use, reliability and validity have rarely been examined. The most evaluated and frequently used measure is the Phenotype of Frailty.