The Patient Frailty Informing Stratified Healthcare Study (Pfish)

Problem

Health services, and primary care in particular, face a major challenge in how best to provide appropriate care to ever-increasing numbers of frail older adults. Under the new GMS contract GPs are required to identify all patients aged 65 and over who are moderately or severely frail, and provide annual medication reviews and enriched care records for the latter. NHS England suggest the electronic Frailty Index (eFI), recently developed by the NIHR York and Humberside CLAHRC, as an appropriate tool for assessing frailty. The eFI uses over 2,000 clinical codes from a patient’s primary care record to assess the presence/absence of 36 health “deficits”, which are then used to compute the frailty score. The eFI represents an important step forward in addressing frailty in the primary care setting. However, the accuracy of the tool is unknown, and its ability to predict patient outcomes is only moderate. Further methodological work is required to improve and validate the measure.

Approach

The Patient Frailty Informing Stratified Healthcare (Pfish) study aims to produce the “next generation” version of the eFI by addressing some of the limitations of the existing measure. In the first step we use expert review to refine the set of clinical codes making up the health deficits and map them to key theoretical models of frailty. Next we implement these in the Clinical Practice Research Datalink (CPRD), a large representative dataset of GP patient records. Factor analysis methods are applied to determine the model of frailty that best fits the data, in particular whether frailty is a single entity, or if distinct sub-domains of frailty exist, such as physical and cognitive. We will validate our final choice of model on its ability to predict hospitalisation and mortality, and characterise the epidemiology of patient frailty in the UK, including incidence and prevalence and how these vary by age, sex, deprivation, medical conditions and region.

Findings

We are currently mapping deficits to theoretical models of frailty. At the conference we will report our investigation of different models of frailty, our final choice, and the epidemiology of frailty.

Consequences

An improved version of the eFI will help GPs to more accurately identify their frail patients, and information on sub-domains of frailty can help to inform decisions about care.Further methodological work is still required, as in particular Pfish will not address the performance of the eFI compared to reference standard methods of assessing frailty, however we have plans to address this in a future study.The knowledge generated will be used to inform the design of a large-scale trial of an intervention using the frailty index to better target patient care, for which further, external, funding will be sought.