An alternative method for Frailty Index cut-off points to define frailty categories

Abstract

Purpose

The Frailty Index (FI) is a popular operationalization of frailty. FI cut-off points have been proposed to define, regardless of age, frailty categories with increasing risk. Here, an alternative method is described that takes age into account.

Subjects and methods

29,905 participants aged 50 years or more from the first wave of the Survey of Health, Ageing and Retirement in Europe. The mean follow-up for mortality was 2.4 years. Curve estimation procedures were carried out between age and a FI, and 50% Confidence Intervals (CI) for the regression mean were derived. As opposed to the usual method (FI ≤ 0.08: non-frail; FI ≥ 0.25: frail; rest: pre-frail), the alternative method defines as ‘fit for their age’ those with a FI below the lower 50% CI; ‘frail for their age’ those with a FI above the upper 50% CI; the rest as ‘average for their age’. Using both methods, the prevalence of the frailty categories and their associated mortality rates were compared for each age group.

Results

The best fit between age and the FI was by cubic regression (R² = 0.174, P < 0.001). Among those in their 50s, 5% were frail by the usual method (mortality: 5%) and 14% by the alternative (mortality: 2%). Among those in their 90s, 64% were frail by the usual method (mortality: 43%) and 41% by the alternative (mortality: 48%).

Conclusion

The alternative method may be more sensitive in younger ages and more specific in older ages. This may have implications for population screening.

Introduction

Frailty in older adults is a state of vulnerability to poor resolution of homoeostasis after a stressor event and is a consequence of cumulative decline in many physiological systems during a lifetime [1]. Although there is no international consensus on a definition of frailty [2], [3], two popular operationalizations are the Frailty Index (FI) and the frailty phenotype [4], [5].

According to the phenotypic approach, frailty is defined as a clinical syndrome consisting of unintentional weight loss, self-reported exhaustion, weakness, slow walking speed, and low physical activity [6], [7]. Fried et al. operationalized these criteria in the Cardiovascular Health Study and defined three frailty categories: frail (i.e. three or more criteria present), pre-frail (i.e. one or two criteria present) and non-frail (i.e. none of the criteria present) [6].

The Frailty Index (FI) sees frailty in relation to the accumulation of health deficits. The FI is measured by comparing the ratio of health deficits present within an individual to possible health deficits, using a pre-specified list of 30 or more deficits [4]. A deficit can be any symptom, sign, disease, disability, or laboratory abnormality that is associated with age and adverse outcomes, present in at least 1% of the population, covers several organ systems and has no more than 5% missing data [8]. Age is not included as a deficit, but the FI increases exponentially with age [9].

While the construct validity of the FI is examined through its relationship to chronological age, its criterion validity is examined in its ability to predict adverse outcomes, including mortality [10]. The latter has been the focus of many epidemiological studies [11], [12], [13].

Rockwood et al. have proposed FI cut-off points to define phenotypical population subgroups with increasing levels of frailty. For example, in one of their studies they proposed FI ≤ 0.08 as ‘non-frail’, FI ≥ 0.25 as ‘frail’, and the rest as ‘pre-frail’ [14]. In another of their studies, they proposed FI ≤ 0.03 as ‘relatively fit’, 0.03 < FI ≤ 0.10 as ‘less fit’, 0.10 < FI ≤ 0.21 as ‘least fit’, 0.21 < FI ≤ 0.45 as ‘frail’, and FI ≥ 0.45 as ‘most frail’ [15]. These cut-offs were proposed regardless of age; thus, for example, a 50 year-old and an 80 year-old with a FI = 0.4 would be, according to their scheme, equally frail. This is illustrated in Fig. 2 (Panel A).

A potential problem with the usual FI cut-off method is that it does not take age into account, especially given the fact that the FI increases exponentially with age [9]. In real clinical life, practitioners often operate within a framework of ‘fit for his/her age’ or ‘frail for his/her age’. For example, a FI of 0.1 measured in a 50 year-old could be, at that young age, regarded as unusually high and trigger aggressive interventions to delay the onset of adverse outcomes in that vulnerable young person. On the other hand, a nonagenarian with a FI of 0.1 could be regarded as fitter and more resilient than the majority of his/her peers at that age. Therefore, a FI of 0.1 may not mean the same to people of different ages.

The aim of the present paper was to explore the properties of an alternative method for FI cut-off points that takes into account age and the exponential association between age and the FI. This is a theoretical paper exemplified with real data from the Survey of Health, Ageing and Retirement in Europe (SHARE), a large longitudinal population-based survey.