Rationale: Data from the INSPIRE study (AJRCCM 2008: 117: 19-26) suggests LABA/ICS, compared to a LAMA, reduces exacerbations requiring treatment with oral corticosteroid. Thus, non-purulent exacerbations that generally do not involve antibiotic therapy could be an indication for prescribing LABA/ICS.
Methods: COPD patients enrolled in the London COPD cohort between 3/10/2005 and 1/3/2016 recorded worsening in respiratory symptoms and oral exacerbation treatment on daily diary cards. Exacerbations were defined as two or more consecutive days with increase in two symptoms, one or more was major (major increased dyspnoea, sputum purulence or sputum volume; minor: cold, sore throat, increased wheeze or cough). Treatment was also recorded at clinic visits.
Results: We examined data from 282 patients who had completed diary cards for at least one year. They were 69.9 (SD 8.6) years old, with a predicted FEV., = 50.1% (SD 16.1) and FEVi/FVC ratio of 47.0% (SD 12.2); 173 (61.4%) were male. 26 patients never experienced an exacerbation. The remaining 256 patients experienced 2531 exacerbations, over a median 1225.5 days (IQR 813-1967), of which 1247 (49.3%) did not involve symptoms of sputum purulence, 1217 (48.1%) did, and symptoms were unknown for 67 (2.6%).
The annualized rate of exacerbations for the 256 patients was 2.14/year (IQR 1.28-3.49) and of treated exacerbations 1.58/year (IQR 0.84-2.65). The rates for non-purulent exacerbations were 0.99/year (IQR 0.59-1.68) and those treated exacerbations 0.63/year (IQR 0.26-1.16). Purulence was absent for 130 of 252 first exacerbations (data missing for 4).
Sub-divided by the 1st exacerbation, rates for non-purulenct exacerbations were 1.18/year (IQR 0.68-1.80), if the first exacerbation was non-purulent, and 0.87/year (IQR 0.45-1.53) if the first exacerbation was purulent (Mann-Whitney; p=0.0059). For treated exacerbations, rates were 0.90/year (IQR 0.49-1.47) and 0.55/year (IQR 0.15-0.90) respectively; p<0.0001. However, whether patients fell into the top 50% of non-purulent exacerbators (> 0.99 exacerbations/year) was not predicted well. 72 of 130 in the top 50% (55.4%) and 69 of 122 (56.6%) in the bottom 50% were identified correctly based on the first exacerbation (p=0.058; ROC AUC=0.56) For treated exacerbations, 63 of 99 (63.4%) of the top 50% of non-purulent treated exacerbators (>0.63 exacerbations per year) and 81 of 142 (57.0%) in the bottom 50% were predicted correctly; p=0.002, ROC AUC=0.60.
Conclusion: Absence of purulent sputum at exacerbation is predictive of future exacerbation character but sensitivity and specificity are poor.