Does the frailty index applied to randomised controlled trials really measure frailty?

Cumulative deficit frailty indices from randomised controlled trials (RCT) are increasingly used to assess whether trial findings are applicable to people living with frailty. The aim of this paper was to examine the range and type of deficits included in these frailty indices and compare these to those from cohort studies.

We identified RCTs assessing treatment effect modification using the cumulative deficit frailty index, as well as cohort studies assessing mortality risk associated with frailty, from recent systematic reviews. We extracted the deficits included in the frailty index from each RCT and cohort study. We compared the number of deficits, data sources (e.g. medical history, physical measurements, questionnaires, etc.) and physiological domain (e.g. cardiometabolic, neuro-cognitive, physical function, etc.) of the deficits from each source.

The number of deficits was similar between RCT frailty indices (median 41 deficits, interquartile range [IQR] 35-50) and cohort studies (median 35, IQR 31-45). Broadly similar data sources were used to identify deficits. However, in RCTs of cardiovascular conditions, cardiometabolic deficits made up a greater proportion of deficits (median 47% of included deficits, IQR 38%-51%, compared to 19%, 14%-24%, in cohort studies). Cardiovascular RCTs included fewer physical function measures (median 4% [3%-9%], compared to 16% in other RCTs of other conditions [13%-17%], 17% in cohort [13%-23%]).

In many cardiovascular RCTs, frailty indices focus on cardiometabolic deficits rather than measures of function. These frailty indices need to be validated against outcomes important to people living with frailty before being used to inform treatment. Until then, we would emphasise caution.