Identifying patients with first chronic obstructive pulmonary disease diagnosis in routine databases: a descriptive study

Abstract

BackgroundThere were about 850 000 people diagnosed with chronic obstructive pulmonary disease (COPD) in England in 2018, and possibly twice as many remain undiagnosed. Starting with the patient's first record of COPD, this study aimed to identify changes in diagnosis, management, and outcomes in COPD over time. MethodsIn this descriptive study, we identified each patient's first record of COPD in primary or secondary care during April 1, 2005, and March 31, 2007 (cohort 1), and during April 1, 2015, and March 31, 2017 (cohort 2). We used the Clinical Practice Research Datalink (CPRD) GOLD database, which contains anonymised patient-level records from a sample of general practices in the UK that use Vision software. General practice records are linked to the Hospital Episode Statistics database, which captures all admissions to English hospitals. The record of COPD was identified with predefined Read and International Classification of Diseases 10 (known as ICD-10) codes. The research protocol was approved by the Independent Scientific Advisory Committee for the Medicines and Healthcare products Regulatory Agency database research (protocol number 19_116). FindingsWe included 37 268 patients with incident COPD (19 266 males and 18 002 female); 57·4% (n=21 392) were from cohort 2. For 24 649 patients (66·1%), the diagnosis was first recorded in secondary care: 40·9% (n=6489 of 15 876) in cohort 1 compared with 84·9% (n=18 160 of 21 392) in cohort 2. However, 14·5% (n=2295) in cohort 1 and 43·0% (n=9207) in cohort 2 transferred out from their general practice before their first COPD record; most of these patients had their first COPD diagnosis in secondary care. More than half of patients with their diagnosis in secondary care had COPD recorded not as a primary diagnosis (51·6% [n=3347] in cohort 1 vs 63·3% [n=11 495] in cohort 2), casting doubt on the diagnosis. For 11 242 patients from cohort 2 with COPD in secondary care, general practices stopped submitting data to CPRD before the COPD diagnosis; consequently, an initial diagnosis could have been made between the time that the practice stopped submitting data and the hospital admission. InterpretationThe background and context of data entry or coding can have an impact on the interpretation of diagnostic data in administrative datasets. When diagnostic codes are used as a starting point for examination of trends in COPD diagnosis and management with CPRD, care is needed to overcome various potentially misleading data artifacts. FundingThis study is funded by the National Institute for Health Research Health Services and Delivery Research stream. The views expressed in this Abstract are those of the authors.