Accurate identification of a respiratory infection in the presence of other respiratory conditions can be help in the management of disease. Pneumonia is one of the most frequent infections experienced by individuals with other respiratory conditions. The complex nature of the disease and similarities of symptoms with other respiratory conditions often make it difficult for timely detection. Identifying pneumonia events and differentiating them from deteriorations in chronic respiratory diseases (such as exacerbations among patients with chronic obstructive pulmonary disease (COPD) or asthma) can be challenging to clinicians. Uncertainties in the current methods of differentiating pneumonia from other respiratory conditions could lead to misclassification of events. In this study, we will use the clinical practice research data link (CPRD) primary care Aurum data linked to secondary care hospital (HES) and mortality (ONS) data to explore the accuracy of identifying pneumonia events in patients with acute exacerbations of COPD (AECOPD), asthma exacerbations and the general population. Our specifications for accurate identification pneumonia events will involve the use of the following items: symptoms, presence of microorganisms, medication use, and clinical investigation. We will investigate the extent at which these items can be used to accurately identify individuals that have pneumonia and those who do not among specific patients’ group. We will investigate how likely a patient with pneumonia as identified by certain algorithms and codes truly has pneumonia. We will also investigate how likely we are to correctly identify an individual with no pneumonia using the same codes and algorithms.
Routinely collected electronic health and administrative data of patients is a valuable tool for health and epidemiological research. The validity and generalisability of any research findings using patients’ electronic health records (EHR) depends on accurate diagnosis of disease outcomes. Validation of various respiratory disease outcomes (e.g., pneumonia, COPD exacerbations (AECOPD) and asthma exacerbations) have been carried out by many studies. However, there is a paucity of data around accurate differentiation of pneumonia events from AECOPD and asthma exacerbation using EHR. We will investigate the diagnostic accuracy of pneumonia events in people in the general population and among those with COPD or asthma using CPRD Aurum data linked with HES APC, HES A&E, and ONS mortality data. We will validate eligible cases of pneumonia events using data on community acquired pneumonia (CAP) obtained from GP practices for various patient groups using combinations of pneumonia diagnosis code, antibiotics, symptoms, and laboratory results to confirm pneumonia events. We will also look at hospital episode statistics (HES) for patients with and without AECOPD and asthma exacerbations who have evidence of community acquired pneumonia (CAP). To evaluate the accuracy of differentiating pneumonia events from exacerbations, we will calculate sensitivity and specificity and obtain the probability of positive and negative pneumonia diagnosis using CPRD data. We will use PPV and NPV value to compare pneumonia diagnostic accuracy of various clinical features including, symptoms, laboratory investigations and chest X-rays. All calculation for diagnostic test characteristics will be carried out separately for general population, AECOPD and asthma exacerbation.
Pneumonia
COPD exacerbations
Asthma exacerbations
Jennifer Quint - Chief Investigator - Imperial College London
Jennifer Quint - Corresponding Applicant - Imperial College London
Alexander Adamson - Collaborator - Imperial College London
Constantinos Kallis - Collaborator - Imperial College London
Ian Douglas - Collaborator - London School of Hygiene & Tropical Medicine ( LSHTM )
Chukwuma Iwundu - Collaborator - Imperial College London
HES Accident and Emergency;HES Admitted Patient Care;ONS Death Registration Data;Patient Level Index of Multiple Deprivation