Toggle Main Menu Toggle Search

Open Access padlockePrints

Frailty trajectories to identify end of life: a longitudinal population based study

Lookup NU author(s): Dr Daniel StowORCiD, Professor Fiona MatthewsORCiD, Professor Barbara Hanratty



This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).


BackgroundTimely recognition of the end-of-life allows patients to discuss preferences and make advance plans, and clinicians to introduce appropriate care. We examined changes in frailty over one year, with the aim of identifying trajectories that could indicate where an individual is at increased risk of all-cause mortality and may require palliative care.MethodsElectronic health records from 13,149 adults (cases) age 75 and over who died in a one-year period (01/01/2015-01/01/2016) were age, sex and general practice matched to 13,149 individuals with no record of death over the same period (controls). Monthly frailty scores were obtained for one year prior to death for cases, and from 01/01/2015-01/01/2016 for controls using the electronic frailty index (eFI: a cumulative deficit measure of frailty, available in most English primary care electronic health records, and ranging in value from 0 to 1). Latent growth mixture models were used to investigate longitudinal patterns of change and associated impact on mortality. Cases were reweighted to the population level for tests of diagnostic accuracy.ResultsThree distinct frailty trajectories were identified. Rapidly rising frailty (initial increase of 0.022 eFI per month before slowing, from baseline eFI of 0.21) was associated with a 180% increase in mortality (OR 2.84 95%CI 2.34 to 3.45) for 2.2% of the sample.Moderately increasing frailty (eFI increase of 0.007 per month, with baseline of 0.26) was associated with a 65% increase in mortality (OR 1.65 95%CI 1.54 to 1.76) for 21.2% of the sample. The largest (76.6%) class was stable frailty (eFI increase of 0.001 from a baseline of 0.26). When cases were reweighted to population level, rapidly rising frailty had 99.1% specificity and 3.2% sensitivity (PPV 19.8%, NPV 93.3%) for predicting individual risk of mortality.ConclusionsPeople over 75 with frailty at highest risk of death have a distinctive frailty trajectory in the last 12 months of life, with a rapid initial rise from a low baseline, followed by a plateau.Routine measurement of frailty can be useful to support clinicians to identify people with frailty who are potential candidates for palliative care, and allow time for intervention.

Publication metadata

Author(s): Stow D, Matthews FE, Hanratty B

Publication type: Article

Publication status: Published

Journal: BMC Medicine

Year: 2018

Volume: 16

Issue: 171

Pages: 1-7

Online publication date: 21/09/2018

Acceptance date: 06/08/2018

Date deposited: 07/08/2018

ISSN (electronic): 1741-7015

Publisher: BioMed Central Ltd.


DOI: 10.1186/s12916-018-1148-x


Altmetrics provided by Altmetric


Funder referenceFunder name
Medical Research Council