Background: Chronic obstructive pulmonary disease (COPD) is a significant cause of morbidity and mortality in England, however estimates of its prevalence vary considerably. Routinely collected and coded primary care data can be used to monitor disease prevalence, however reliance upon diagnostic codes alone is likely to miss cases.
Methods: We devised an ontological approach to COPD case detection and implemented it in a large primary care database to identify definite and probable cases of COPD. We used this to estimate the prevalence of COPD in England.
Results: Use of this approach to detect definite COPD cases yielded a prevalence of 2.57% (95% CI 2.55-2.60) in the total population, 4.56% (95%CI 4.52-4.61) in those aged ≥ 35 and 5.41% (95% CI 5.36-5.47) in ex or current smokers. The ontological approach identified an additional 10,543 definite cases compared with using diagnostic codes alone. Prevalence estimates were higher than the 1.9% prevalence currently reported by the UK primary care pay for performance (P4P) disease register. COPD prevalence when definite and probable cases were combined was 3.02% (95% CI 3.0-3.05) in the total population, 5.38% (95% CI 5.33-5.42) in those aged ≥ 35 and 6.46% (95% CI 6.46-6.40-6.56) in ex or current smokers.
Conclusions: We demonstrate a robust reproducible method for COPD case detection in routinely collected primary care data. Our calculated prevalence differed significantly from current estimates based upon P4P data, suggesting that the burden of COPD in England is greater than currently predicted.
Keywords: Epidemiologic study; Medical record systems: computerised; Models: theoretical; Prevalence; Primary health care; Pulmonary disease: chronic obstructive.
Copyright © 2017 Elsevier Ltd. All rights reserved.