Frailty is a condition of increased vulnerability to major changes in health and is common in older age, affecting around 10% of those over 65. It develops as our bodies change with ageing and lose their inbuilt reserves such as muscle strength.
People with frailty are at increased risk of falls, disability, loneliness, and therefore hospitalisation and nursing home admission. These reduce quality of life and are costly for the NHS and social care.
Recently a scoring method: the electronic frailty index (eFI) score has been developed for use in primary care to help identify older people living with frailty, to better target packages of care. The eFI score will change over time, and our study aims to identify different groups of older people with frailty based on how quickly or slowly their score changes. For example, a rapidly increasing score might suggest that a person needs extra support from the NHS or social care services.
We will use two very large GP databases for our study, CPRD Gold and Aurum. These will enable us to identify subgroups of people with different patterns of frailty, and whether the latter are at higher risk of hospital admission, and death.
If successful, the study will give us the opportunity to further support GPs to target older people with frailty who are at higher risk of worsening health, compared to those whose health is more stable. This information will help GPs target the limited resources available to those most at need.
This work, funded by an RfPB grant, aims to extend the utility of the electronic frailty index (eFI), recently introduced into the NHS.The eFI score currently is a simple count of the number of deficits (maximum 36), which include symptoms/signs (e.g. dizziness), abnormal lab values of anaemia, disability (e.g. housebound) to diseases (e.g. arthritis). Using this score, we aim to identify patterns of trajectories of developing frailty in the population aged over 65. We will mine routine primary care data to identify such trajectory patterns and validate these against the subsequent risk of a number of adverse outcomes.
The work comprises 3 phases:
1. Anonymised primary care electronic health record data from the CPRD GOLD database of 225,264 participants with available data from January 1 2009, aged 65+, of which approximately 60% are linked to HES, will be extracted. We will then calculate the cumulative crude number of deficit variables based on the eFI rules. In each of the successive four years, any increase in the cumulative number of the deficits based on the new recording of any previously unrecorded variable will be captured. Distinct patterns of the five year trajectories will then be derived using both latent class growth analysis and growth mixture modelling.
2. These trajectories will be tested in a second primary care database, CPRD Aurum, to examine their reproducibility and, we will also explore the types of the individual patient profiles fit with these
3. We will then explore the association between the emerging distinct frailty trajectories with the adverse outcomes of mortality, and unplanned hospitalization, using both cohorts over the subsequent three years of follow-up.
Outcomes:
- Primary: the primary outcome will be all cause mortality, defined as death recorded in the ONS Death Registration Data.
- Secondary: unplanned hospital admission. The decision to use CPRD or HES records of unplanned admission will be assessed after a comparison of the reporting in the two datasets.
Alan Silman - Chief Investigator - University of Oxford
Victoria Y Strauss - Corresponding Applicant - University of Oxford
Andrew Clegg - Collaborator - University of Leeds
Antonella Delmestri - Collaborator - University of Oxford
Daniel Prieto-Alhambra - Collaborator - University of Oxford
Danielle Robinson - Collaborator - University of Oxford
Leena Elhussein - Collaborator - University of Oxford
Sara Khalid - Collaborator - University of Oxford
HES Accident and Emergency;HES Admitted Patient Care;ONS Death Registration Data;Practice Level Index of Multiple Deprivation