Chronic obstructive pulmonary disease (COPD) describes a group of long-term lung conditions. It affects over a million people in the UK alone, reduces quality of life and costs the NHS a lot to treat. It can be hard to diagnose, but little is known about how patients get their diagnosis, for instance if it’s when seeing their GP, following various tests or during an emergency admission to hospital. Using patient data from GP computer systems linked to hospital admissions records, this project will increase our understanding of how long it takes to get diagnosed and whether some types of patients have longer times to diagnosis than others. This knowledge is the first key step to help the NHS improve care for patients with COPD.
A common consequence of COPD is that breathing can quickly get much harder (known as an acute exacerbation, AECOPD), causing lung damage. If we can predict early on which patients are the most likely to go on and have an AECOPD before too much damage is done, then treatment and monitoring can be adjusted to prevent many further AECOPDs. We will do this by building statistical models that predict the risk of AECOPDs based on factors such as the patient’s age, smoking, blood pressure and medical history. This modelling aims to rank these patient factors in public health importance and thereby give some direction to the NHS to prioritise its care improvement efforts for people with COPD.
Aims and objectives: To describe and model the patient journey from symptom presentation to diagnosis and first acute exacerbation for COPD patients in England. This will include examining variations by GP practice and time period, followed by the construction and internal validation of a risk prediction or risk trajectory model for the first AECOPD.
Methods: Using the Clinical Practice Research Datalink (CPRD) and three cohorts, we will describe the management of the patient following initial presentation with symptoms through to their diagnosis of COPD and their first AECOPD, which for some patients will be the same event. The cohorts will be for 2006, 2016 and the first COVID wave (March to August 2020). Given that COPD can present differently depending on comorbidity, the mapping will be described separately for people with asthma and heart failure in particular. We will assess compliance with NICE guidelines for diagnosis, including practice-level variation in spirometry. The second part will model the risk of the first AECOPD using factors such as airways obstruction, age, smoking, BMI, gender, comorbidities and medications. This will use Cox proportional hazards modelling. Population attributable fractions will be calculated for each predictor. Models will be cross-validated and assessed for discrimination and calibration.
Anticipated impact and dissemination: This study will fill key gaps in our understanding of how patients obtain their COPD diagnosis (their “route to diagnosis”), how they are managed in primary care, and how they get their first AECOPD. Comparisons between the three time periods will highlight what has changed, including during the early part of the pandemic, and inform NHS preparation for future needs regarding COPD. If it performs well, a risk prediction model for first acute exacerbation will aid shared decision-making between GPs and patients and facilitate early intervention; ranking the predictors will suggest priorities for action.
Setting where COPD first recorded (primary or secondary care); NICE guideline compliance for diagnosis; First acute exacerbation (AECOPD)
Alex Bottle - Chief Investigator - Imperial College London
Alex Bottle - Corresponding Applicant - Imperial College London
Alexander Adamson - Collaborator - Imperial College London
Benedict Hayhoe - Collaborator - Imperial College London
Jennifer Quint - Collaborator - Imperial College London
HES Admitted Patient Care;ONS Death Registration Data;Practice Level Index of Multiple Deprivation