Measuring modifiable functional decline and frailty in Primary Care: Is the eFI a suitable tool?

Talk Code: 
Jessica Bollen
Bollen, J.C, Budd, A., Withall, J., Greaves, C.J, Moorlock, S.J, de Koning, J., Stathi, A.
Author institutions: 
University of Exeter, University of Bath, University of Birmingham


Since 2017 General Practices (GP) are required to identify patients over the age of 65 with moderate to severe frailty. Whilst the assessment method is decided by the GP practice, many use the electronic Frailty Index eFI. Interventions targeting older adults who begin to experience some decline in functional mobility and are effective at preventing further decline, could lead to the reduction of health and social care costs and the increase of people’s well-being and quality of life. Identification and recruitment of people with early signs of physical frailty is a key factor for the success of such interventions. This study examined the appropriateness of the eFI as a screening tool for the identification of adults aged 65 and over with modifiable functional decline.


Fourteen General Practice (GP) surgeries whose patients were assessed for eligibility for REACT were approached via a letter. The letter requested eFI scores for their patients who had completed the Short Physical Performance Battery (SPPB) at the REACT baseline assessment. SPPB scores of 4-9, indicating some mobility impairment, were differentiated from scores of >10 indicating non-frail. A score of <3 were excluded. The sensitivity and specificity of the eFI and SPPB outcomes were investigated, and potential participants with SPPB score of 4-9 were identified by determining an eFI cut off score. These data were provided by participating practices and Receiver Operating Characteristic (ROC) Curves were carried out in addition to sensitivity analyses.


Preliminary analysis indicates that the eFI is a poor predictor of SPPB score. An exploratory Spearman correlation identified a weak to moderate negative correlation suggesting as SPPB score increased eFI score decreased. An Area Under the ROC Curve indicated using eFI scores to predict SPPB scores was also poor and therefore referring to a patient’s eFI score to determine physical frailty and functional decline is of limited clinical use. However it may be possible to improve identification of mobility impaired people for research studies using eFI scores (which would achieve a hit rate of roughly 1 in 2)


This study indicates the eFI is not an effective screening tool to identify adults aged 65 and over with modifiable functional decline or physical frailty in a clinical setting. However there may be merit in using eFI to identify who to approach for future studies. Other methods to assess physical frailty and functional decline in primary care need to be identified to enable accurate targeting of appropriate interventions, in addition to standardisation of how to assess frailty.

Submitted by: 
Jessica Bollen
Funding acknowledgement: 
National Institute for Health Research (NIHR)