Date of Approval
Application Number
21_000379
Lay Summary
Chronic obstructive pulmonary disease (COPD) is a common lung condition that causes breathing difficulties such as breathlessness, coughing and wheezing. As these symptoms get worse, it can lead to the patient needing to be treated in hospital and may lead to death. The severity of the disease is measured on a scale ranging from mild, moderate, severe to very severe. This study focuses on those with moderate-to-very-severe COPD. The study aims to investigate risk of death from all causes, understand which factors influence the risk of death, and describe medication used in patients diagnosed with moderate-to-very-severe COPD in England.
First, we will investigate whether the population in the CPRD database is representative of the population of England with respect to mortality (i.e. death rates) and explore the availability of information needed for our study among all COPD patients in the CPRD database, such as data on lung function tests.
Dependent upon these results, we will specify criteria to select the subgroup of COPD patients to be included in our analyses. We will calculate the risk of death, e.g. by comparing the risk of death to a group of patients with similar characteristics and investigate whether factors such as age, socioeconomic status, lung capacity and prescribed treatments affect the probability of death. We will describe treatments that those patients received and investigate what patient and clinical characteristics led to patients receiving particular types of treatment. With this knowledge, doctors can monitor COPD patients more appropriately and manage their treatment more effectively.
First, we will investigate whether the population in the CPRD database is representative of the population of England with respect to mortality (i.e. death rates) and explore the availability of information needed for our study among all COPD patients in the CPRD database, such as data on lung function tests.
Dependent upon these results, we will specify criteria to select the subgroup of COPD patients to be included in our analyses. We will calculate the risk of death, e.g. by comparing the risk of death to a group of patients with similar characteristics and investigate whether factors such as age, socioeconomic status, lung capacity and prescribed treatments affect the probability of death. We will describe treatments that those patients received and investigate what patient and clinical characteristics led to patients receiving particular types of treatment. With this knowledge, doctors can monitor COPD patients more appropriately and manage their treatment more effectively.
Technical Summary
This is an observational retrospective cohort study using CPRD-HES-ONS linked data aiming to describe treatment patterns and mortality among moderate-to-very-severe COPD patients in England, and to investigate risk factors for receiving particular treatments, death and other COPD-related outcomes. The study will include COPD patients who are eligible for linkage to HES and ONS. HES data will be used to determine COPD exacerbations, while ONS data will be used to obtain details on deaths.
Exploratory analyses will assess the representativeness of the entire CPRD-HES-ONS linked population in terms of mortality, by comparing age-standardised sex-specific all-cause mortality rates in this population to published mortality statistics. We will assess availability of COPD parameters among COPD patients and assess the impact of applying specific selection criteria on the sample size, to help inform the main analyses.
In the main analyses, only moderate-to-very-severe COPD patients will be considered. We will describe demographic and clinical characteristics, calculate mortality rates and estimate either absolute mortality, through a Weibull regression, or mortality relative to a ‘general population’ comparison group within the CPRD dataset, by calculating the standardized mortality ratio and hazard ratio. A survival model will be applied to assess the association between COPD-related factors and mortality. The COPD Galaxy model will be fitted to the COPD cohort to derive predicted probabilities of death and compare observed versus expected deaths. We will describe COPD treatments at baseline and at first switch. Patient characteristics will be compared based on baseline treatment using chi-squared tests, and one-way or Kruskal-Wallis ANOVA as appropriate. The association between risk factors and treatments will be assessed with logistic regression models. Kaplan-Meier time-to-event analyses will be performed for first treatment switch. Treatment pathways will be visualized using a Sankey diagram. Risk of COPD outcomes will be compared by treatment using a Cox proportional hazards model.
Exploratory analyses will assess the representativeness of the entire CPRD-HES-ONS linked population in terms of mortality, by comparing age-standardised sex-specific all-cause mortality rates in this population to published mortality statistics. We will assess availability of COPD parameters among COPD patients and assess the impact of applying specific selection criteria on the sample size, to help inform the main analyses.
In the main analyses, only moderate-to-very-severe COPD patients will be considered. We will describe demographic and clinical characteristics, calculate mortality rates and estimate either absolute mortality, through a Weibull regression, or mortality relative to a ‘general population’ comparison group within the CPRD dataset, by calculating the standardized mortality ratio and hazard ratio. A survival model will be applied to assess the association between COPD-related factors and mortality. The COPD Galaxy model will be fitted to the COPD cohort to derive predicted probabilities of death and compare observed versus expected deaths. We will describe COPD treatments at baseline and at first switch. Patient characteristics will be compared based on baseline treatment using chi-squared tests, and one-way or Kruskal-Wallis ANOVA as appropriate. The association between risk factors and treatments will be assessed with logistic regression models. Kaplan-Meier time-to-event analyses will be performed for first treatment switch. Treatment pathways will be visualized using a Sankey diagram. Risk of COPD outcomes will be compared by treatment using a Cox proportional hazards model.
Health Outcomes to be Measured
Exploratory analyses
Descriptive variables: smoking status, spirometry (forced expiratory volume in one second (FEV1) pre-bronchodilator and post-bronchodilator, FEV1 predicted, FEV1 % predicted pre-bronchodilator and post-bronchodilator, FVC, FEV1/FVC, FEV1/FVC predicted); COPD-related scores and biomarkers (Medical Research Council (mMRC) Dyspnea Scale, COPD Assessment Test (CAT), Global Initiative for Chronic Obstructive Lung disease (GOLD) stage, 6-minutes’ walk distance (6MWD), Fractional exhaled Nitric Oxide (FeNO), fibrinogen and eosinophil (EOS) count); COPD medication (double or triple therapy regimen flag); COPD exacerbations (moderate and severe exacerbations); COPD-related Accident and Emergency (A&E) visits; COPD-related inpatient hospitalisations; non-comorbidity-related QRISK2 score components (systolic blood pressure (SBP), body mass index (BMI), total cholesterol, high density lipoprotein cholesterol ratio, family history of coronary heart disease (first-degree, aged<60)). We will also describe age-standardised all-cause mortality rates by sex.
Main analyses
Descriptive variables as at the index date (the date by which all inclusion criteria are met with an appropriate look-back period applied): demographic, clinical and lifestyle characteristics (e.g. age, sex, patient index of multiple deprivation, BMI, smoking status, comorbidities (ischaemic heart disease, heart failure, stroke, heart arrhythmia, bronchiectasis, diabetes, osteoarthritis, osteoporosis, inflammatory bowel disease, depression, anxiety, panic attack, rheumatoid arthritis and peptic ulcer); COPD-related factors and biomarkers (e.g. mMRC, FEV1, CAT, GOLD stage, 6MWD, fibrinogen and EOS count); COPD-related medication (e.g. inhaled corticosteroids (ICS), long-acting beta agonists (LABA), long-acting muscarinic antagonists (LAMA) and their combinations); QRISK2 score; pneumonia risk score; COPD exacerbations (moderate and severe exacerbations); calendar year of the index date. We will also derive mortality rates and standardised mortality ratio.
Descriptive variables during the follow-up: COPD-related medications (as described in detail later in the document) and FEV1.
Outcomes for the regression models: all-cause and COPD-related mortality, major adverse cardiovascular events (MACE), pneumonia event, moderate exacerbation event, severe exacerbation event.
Descriptive variables: smoking status, spirometry (forced expiratory volume in one second (FEV1) pre-bronchodilator and post-bronchodilator, FEV1 predicted, FEV1 % predicted pre-bronchodilator and post-bronchodilator, FVC, FEV1/FVC, FEV1/FVC predicted); COPD-related scores and biomarkers (Medical Research Council (mMRC) Dyspnea Scale, COPD Assessment Test (CAT), Global Initiative for Chronic Obstructive Lung disease (GOLD) stage, 6-minutes’ walk distance (6MWD), Fractional exhaled Nitric Oxide (FeNO), fibrinogen and eosinophil (EOS) count); COPD medication (double or triple therapy regimen flag); COPD exacerbations (moderate and severe exacerbations); COPD-related Accident and Emergency (A&E) visits; COPD-related inpatient hospitalisations; non-comorbidity-related QRISK2 score components (systolic blood pressure (SBP), body mass index (BMI), total cholesterol, high density lipoprotein cholesterol ratio, family history of coronary heart disease (first-degree, aged<60)). We will also describe age-standardised all-cause mortality rates by sex.
Main analyses
Descriptive variables as at the index date (the date by which all inclusion criteria are met with an appropriate look-back period applied): demographic, clinical and lifestyle characteristics (e.g. age, sex, patient index of multiple deprivation, BMI, smoking status, comorbidities (ischaemic heart disease, heart failure, stroke, heart arrhythmia, bronchiectasis, diabetes, osteoarthritis, osteoporosis, inflammatory bowel disease, depression, anxiety, panic attack, rheumatoid arthritis and peptic ulcer); COPD-related factors and biomarkers (e.g. mMRC, FEV1, CAT, GOLD stage, 6MWD, fibrinogen and EOS count); COPD-related medication (e.g. inhaled corticosteroids (ICS), long-acting beta agonists (LABA), long-acting muscarinic antagonists (LAMA) and their combinations); QRISK2 score; pneumonia risk score; COPD exacerbations (moderate and severe exacerbations); calendar year of the index date. We will also derive mortality rates and standardised mortality ratio.
Descriptive variables during the follow-up: COPD-related medications (as described in detail later in the document) and FEV1.
Outcomes for the regression models: all-cause and COPD-related mortality, major adverse cardiovascular events (MACE), pneumonia event, moderate exacerbation event, severe exacerbation event.
Collaborators
Caroline O'Leary - Chief Investigator - IQVIA Ltd
Minouk Schoemaker - Corresponding Applicant - IQVIA Solutions B.V. (Nederland)
Clare Flach - Collaborator - IQVIA Ltd
Heloísa Galante - Collaborator - IQVIA II Technology Solutions Portugal, Unipessoal LDA
Joshua Warden - Collaborator - IQVIA World Publications Ltd.
Mar Pujades Rodriguez - Collaborator - IQVIA
Nelly Ly - Collaborator - IQVIA Ltd
Sander Van Olst - Collaborator - IQVIA Solutions B.V. (Nederland)
Stephanie Castello - Collaborator - IQVIA
Linkages
HES Accident and Emergency;HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation