Chronic Obstructive Pulmonary Disease (COPD) is an important cause of long term ill health in the UK and worldwide. Severity varies but more advanced COPD often requires hospital admission and may become sufficiently complex to require management by hospital specialists. However, we don't know what measurements best assess severity and how to use them to determine how many patients have advanced disease. We know that patients with COPD often have other medical problems (multimorbidity) affecting health, life expectancy and risk of hospital admission. However, we don't know if multimorbidity is as important as the severity of COPD itself in predicting future health and need for treatment. Using health data about patients with COPD that is already recorded by GP practices across the UK we will find out how many patients recorded as having COPD actually have advanced, more severe disease. We will learn how best to predict their future needs for healthcare and also to understand how multimorbidity affects the health of people with COPD. It will also help local health organisations plan care for people with COPD, local GPs look after patients with COPD and help us understand which patients should be referred to a specialist.
Using R and SPSS, data will be cleaned and explored with expert consideration given to reasons, extent and remedies for missing data. Statistical models will be formed for COPD health risk, namely risk of all-cause mortality, hospitalisation (from Hospital Episode Statistics) and respiratory death (from Office of National Statistics) and we will establish whether it is best done from indices of disease severity alone or from inclusion of wider multimorbidity (cardiac disease, diabetes, hypertension, psychological disorder). Risk estimates will be derived using exemplar local specialist referral criteria including FEV1, MRC dyspnoea score, Body Mass Index (BMI), home oxygen prescription/use, exacerbations in the preceding twelve months and smoking status. BMI (height in metres divided by weight in kg squared) will be taken from the Additional Clinical file, as will smoking status. (Height, weight and smoking status being entity types 14, 13 and 4 respectively). Exacerbation of COPD will be inferred from hospitalisation due to COPD and/or CPRD Clinical file medcodes 1446 or 7884. Logistic and cox regression models will be evolved with both fixed and random effects considered. Predictive power will be assessed using the C-statistic, Sensitivity, Specificity, Positive and Negative Predictive Power, Harrell's c-index and Gonen and Heller's K statistic.
COPD prevalence
- Exacerbation frequency
- Comorbidities
- Mortality
- Health Resource use
Michael Steiner - Chief Investigator - Glenfield Hospital
Urvee Karsanji - Corresponding Applicant - University of Leicester
- Collaborator -
Claire Lawson - Collaborator - University of Leicester
Jennifer Quint - Collaborator - Imperial College London
Kamlesh Khunti - Collaborator - University of Leicester
Matt Richardson - Collaborator - NIHR Leicester Biomedical Research Centre
Neil Greening - Collaborator - NIHR Leicester Biomedical Research Centre
Rachael Evans - Collaborator - University of Leicester
Sally Singh - Collaborator - University of Leicester
HES Accident and Emergency;HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation