The ability of patients to recover from acute illness like an infection does not depend on a patientÂ’s age but their frailty. Frailty is the vulnerability to a more severe response to an acute challenge or illness. The ability to predict who is frail by this definition is critical to understand which patients would benefit from detailed assessment of needs and support, and this would inform future decisions about appropriate interventions or treatments.
Current methods to predict frailty have not modelled frailty as the vulnerability to poor recovery from an acute illness, but instead are risk scores which are based on their ability to predict broad outcomes from a fixed time point. This means that current methods to identify frailty are inaccurate and of limited value. This project will assess the feasibility of defining frailty as a vulnerability to a poor recovery after an acute infection by studying the factors that predict the different patient recovery trajectories following urine or chest infections, using anonymised patient health records. Using this method, the findings will be translatable to patient records commonly used within community healthcare and will lead to the development of better clinical tools to predict patients with frailty.
This proposal will reframe the concept of frailty by assessing the different patient recovery trajectories following a urine or respiratory community acquired infection 2005-2015. This will develop a framework for future frailty research that has true construct validity. Open population based cohorts will be used within routine healthcare data to map the different recovery trajectories across different dimensions. These will include consultation rates, prescription rates, diagnosis rates, specific diagnostic codes, sequences of codes, location of health care interactions (inpatient, community, care home), and mortality. Changes in each dimension from baseline will be assessed over time using within patient case series analysis. To efficiently and systematically identify the most common ordered sequences of events from many potential combinations sequential pattern mining will be used. These different recovery trajectories will then be stratified by pre-existing co-morbidity and frailty measures, and the characteristics of patients in each trajectory will be described. A Cox proportional hazards model will be used to adjust for the competing risks between the multiple different trajectories using cumulative incidence functions. This model will then be used to calculate the association of pre-existing the co-morbidity and frailty measures with each recovery trajectory.
Admission and length of hospital stay
- Clinical consultation rates
- All cause mortality
- Prescription rates
- Discharge or admission to care home
- Increased mobility needs
Colin Crooks - Chief Investigator - University of Nottingham
Colin Crooks - Corresponding Applicant - University of Nottingham
John Gladman - Collaborator - Nottingham University Hospitals
Trevor Hill - Collaborator - University of Nottingham
HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation