WEUSKOP6416: Evaluating Pneumonia in Chronic Obstructive Pulmonary Disease (COPD) Subjects

Advantages of the primary care database include the ability to link to secondary care, Hospital Episode Statistics (HES), and to examine risk factors for pneumonia included in the UK Quality Outcomes Framework for COPD that are not collected routinely in most other observational healthcare data sources (e.g., BMI, lung function, smoking history, MRC dyspnea score). Another advantage of this study is the new-user design, as it minimizes biases that can be caused by alternative designs which compare events between prevalent user groups (e.g., survivor bias, covariates altered by prior exposure). There are also known limitations of observational database analyses where treatments are not randomized, including the potential for confounding by severity that may not be fully accounted for in the analysis. ICS-containing medications may be dispensed to patients who have more severe COPD than patients who are receiving long-acting bronchodilators alone. In this study, we will adjust for disease severity and patient characteristics in the period prior to initiation using propensity scores. Medication use in CPRD-GOLD is based on prescribed medications recorded by the primary care physician, which might not have been dispensed at the pharmacy or ultimately utilized by the patient. Finally, diagnostic practices for pneumonia may be different in the UK compared with other countries limiting the generalizability. As with any outcome of interest, identifying pneumonia in databases is imperfect and there may be confusion between diagnoses of pneumonia and influenza or serious exacerbation. To our knowledge, the relationship between pneumonia recorded in these primary and secondary care databases have not been investigated. In addition, there is lack of agreement between pneumonia classification in the absence of chest x-rays, sputum, etc. Despite the limitations, this study will provide insights into risk factors for serious pneumonia, including whether ICS modify the effect of established risk factors for serious pneumonia. The results may identify specific patient groups that are at greatest risk of serious pneumonia and may identify where risk minimization and/or medical recommendations may be appropriate.