Chronic Obstructive Pulmonary Disease (COPD) is a complex and long-term condition, the fifth leading cause of death worldwide and a burden on healthcare resources. NHS England has created a strategic plan, called the Long Term Plan, outlining key areas for investment over the next 10 years. Respiratory disease is one of those key areas for investment with an ambition to transform COPD outcomes to be equal to, or better than, those in other countries.
Understanding which patients with COPD who are most likely to be respond to intervention and in whom outcomes such as acute flare-ups and needing lung surgery can be improved is key. Impactibility models, a type of statistical model that looks not only at risk but additional factors, are valuable tools and contain three main components. Firstly, they include ‘gaps in care’, where patients may not have been offered or may not have taken up interventions or diagnostic tests as outlined by national or local guidance. Secondly, they incorporate understanding of how engaged patients are with their care and attempts to suggest appropriate interventions based on their ability and desire to engage. Thirdly, they ‘risk stratify’ people, where existing data will be used to predict how likely patients are to develop further medical conditions. By using these three approaches we will learn how best to predict which patients are most ‘impactable’, therefore helping healthcare professionals to provide better care for their populations and ensure the right interventions are resourced to meet their specific needs.
Risk stratification models are used to decide where and how to allocate limited resources. While risk stratification models may accurately predict future adverse health outcomes, their use has not consistently led to improvements in health outcomes across the population. In 2010 Lewis conceptualised impactibility models, defining them as “refining the output of predictive models by: giving priority to patients with diseases that are particularly amenable to preventive care; excluding patients who are least likely to respond to preventive care; or identifying the form of preventive care best matched to each patient's characteristics”.
To create a COPD impactibility model, we have taken the outputs from our published systematic literature review (SLR) on impactibility to find approaches to refine the predictive model, including understanding an individual’s propensity to succeed and reviewing ‘gaps in care’.
Data will be cleaned and explored with expert consideration given to variable inclusion/exclusion, variable distributions and remedies for treating missing data.
CPRD AURUM linked to ONS Death Registration Data, HES Admitted Patient Care, HES Outpatient, HES A&E and HES DID will be used.
The risk stratification model will be formed for those with COPD and their risk of progression to their first co-morbidity or additional co-morbidity.
Using factors informed by the SLR on COPD and progression to muliti-morbidity such as airways obstruction, age, smoking, and deprivation, this approach will apply competing-risk time-to-event analysis.
This will determine the risk of multimorbidity progression as a function of time from the onset of positive COPD diagnosis. This model will then be further refined by factors outlined in our SLRs
• Inferring propensity to succeed by using Patient Activation Measure score and socioeconomic using available data
• comparing patients’ care pathways with NICE guideline (NG115) to derive a ‘gap score’ to estimate the number of ‘gaps in care’.
• Progression to first or additional co-morbidity as a function of time from the initial COPD diagnosis will be the outcome used to create the predictive model
• To further refine this predictive model to become an ‘impactibility model’ we will also measure the number of care processes missing (as compared with NICE Guidelines) and the number of care processes missing per patient.
Jennifer Quint - Chief Investigator - Imperial College London
Andi Orlowski - Corresponding Applicant - Imperial College London
Alex Bottle - Collaborator - Imperial College London
XIAOXU Zou - Collaborator - The Health Economics Unit
XIAOXU Zou - Collaborator - The Health Economics Unit
HES Accident and Emergency;HES Admitted Patient Care;HES Diagnostic Imaging Dataset;HES Outpatient;ONS Death Registration Data;Patient Level Index of Multiple Deprivation