The aim of this thesis was to link secondary care data from a UK national audit of chronic obstructive pulmonary disease (COPD) care with primary care data from a database of UK electronic health records (EHRs) to explore how variations in patient pathways through healthcare across England affect hospital admissions for acute exacerbations of COPD (AECOPD). This aim was achieved through 6 objectives: (i) a systematic review of the literature on validation of AECOPD definitions in EHRs; (ii) determination of predictors of referral to pulmonary rehabilitation from general practice; (iii) a comparison of the quality of COPD primary care in each UK country, as currently only Wales is assessed; (iv) determination of whether the COPD Best Practice Tariff (BPT) pay-for-performance scheme improves patient outcomes; (v) assessment of the utility of NEWS2 as a severity score measure in AECOPD admissions; (vi) linkage of secondary care audit data with primary care EHR data to explore how management of patients with COPD affects AECOPD hospital admissions.
A summary of the key results is as follows. Firstly, although few studies have validated AECOPD definitions, a validated AECOPD definition was found in a systematic search of the literature that could be used in subsequent objectives. Secondly. while generally appropriate patients appear to be prioritised for PR referral, women were less likely to be considered for referral than men. Thirdly, England, Scotland, and Northern Ireland had substantially lower proportions of patients with confirmed airways obstruction and referrals to pulmonary rehabilitation than Wales. This suggests that completing primary care audits solely in Wales is leading to improvements in, at least, the recording of care that are not happening in the rest of the UK. Fourthly, the combination of interventions financially incentivised by the COPT BPT were not associated with an improvement in 30-day mortality or readmission. One component of the BPT, specialist review, was associated with 31% lower odds of inpatient mortality. Fifthly, NEWS2 was a poor predictor of length of hospital stay, requirement for NIV, and inpatient mortality, with AUC values of 0.7 or less for each outcome. Sixth and finally, 80% of patients admitted for AECOPD had contact with their GP in the 2 weeks prior to admission, suggesting that these admissions could not have been avoided. 86% of admissions were clinically appropriate. Contact with primary care did not appear to affect admission appropriateness. Receipt of a discharge care bundle was associated with receipt of best practice care, however this association appeared to derive from already having received those items of care in secondary care.
Power was limited in the final analyses making it difficult to draw firm conclusions, however COPD discharge care bundles do not appear to be leading to improvements in key patient outcomes.