Many patients in the UK have chronic obstructive pulmonary disease (COPD), a lung disease usually caused by smoking that typically causes symptoms such as cough or breathlessness. The main treatment are inhaled drugs, of which 3 main drug types are used. These inhalers prevent daily symptoms and importantly - exacerbations which are a sudden worsening of lung symptoms requiring urgent treatment. Recently there has been concern over side-effects from one of the drug types, inhaled corticosteroids (ICS). Because of this concern,drug trials have investigated the effects of stopping ICS in patients and the consequent effect on their disease exacerbations. These trials have not all agreed with each other.Indeed, trial participants are often not a typical COPD patient. Depite this, doctors sometimes stop ICS in patients. Therefore we want to assess in a real life group of COPD patients (1) how many COPD patients use different inhalers, (2)if these different treatments effect their chance of having an exacerbation and (3) if a patient has their ICS stopped by their doctor, does this effect their chance of having an exacerbation.
We will identify a cohort of COPD patients and carry out two study designs, a cohort study and a self-controlled case series (SCCS). In the cohort, the main exposure will be their combined-inhaler regimen (ICS/LABA, LAMA/LABA or triple therapy); categorised also by incident (new COPD diagnosis or switched to a new regimen) or prevalent regimen use. The main outcome will be exacerbations (in primary care, CPRD, and secondary care, HES, or death, ONS). To estimate the effect estimates we will use a Poisson regression model, confounding will be adjusted for using conventional methods and propensity score methodology. As certain unmeasured patients-factors may still influence why a patient was chosen for a particular regimen, we will also carry out a SCCC; this analysis implicitly controls for any time non-varying confounding as the patient acts as its own control. The stable period will be the time before the new regimen; the risk period will follow the new regimen. An incidence rate comparing the stable and risk periods will be estimated using a conditional Poisson model. Time-varying confounder, age, will be controlled for in the regression model. All potential effect modifiers will be investigated by stratified analysis, including disease severity, and eosinophil count.
Inhaler use and inhaler switching by drug class
Reliever inhaler usage
Exacerbation rates
Hospitalised pneumonia rates
Jennifer Quint - Chief Investigator - Imperial College London
Chloe Bloom - Corresponding Applicant - Imperial College London
Omar Usmani - Collaborator - Imperial College London
HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation