Induced sputum differential cell counts provide important information about airway inflammation and future risk in asthma, but are not universally available. We set out to ascertain the sensitivity and specificity of the Juniper Asthma Control Questionnaire (ACQ) and/or the peripheral blood eosinophil count to detect a sputum eosinophil count >3%.

Methods We performed a retrospective, cross-sectional study of 165 subjects with asthma, aged 18–80, prescribed as-required bronchodilators, long-acting β-agonists, 0–4000 mcg inhaled beclometasone diproprionate equivalent or maintenance oral steroids (0–20 mg prednisolone) from our database. Current smokers were excluded. Spirometry, FeNO at 50 ml flow, sputum induction, Asthma Control Questionnaire (ACQ) and blood eosinophils (BEos) were recorded.

Induced sputum eosinophils (SEos) defined eosinophilic (EA, SEos ≥3%) and non-eosinophilic (NEA, SEos <3%) groups. BEos, SEos and FeNO were log₁₀transformed and groups were compared with t-tests or Mann-Whitney-U using STATA and GraphPad. Receiver operating characteristic (ROC) curves determined cut-points of ACQ and BEos that identified SEos ≥3%, and these were tested retrospectively in a second population of adults with asthma (n = 48, 40% EA, mean FEV₁ 82% predicted).

Results The 31% with EA had a lower FEV₁% predicted and FEV₁/FVC ratio, a higher FeNO, BEos and ACQ compared to NEA (Table 1). There was a significant correlation between ACQ and sputum eosinophils (Pearson r 0.32, p < 0.0001).

The optimal cut-points of ACQ and BEos identifying a SEos of 3% were an ACQ >1.57 (sensitivity 55%, specificity 78%) and BEos ≥0.22 × 10⁹/L (sensitivity 84% and specificity 68%) respectively. Individuals with both ACQ of ≥1.57 AND a BEos of ≥0.22 the sensitivity was 45% and specificity of 100% for a SEos of 3%.

Testing in the second population the ACQ≥1.57 had a sensitivity 47% and specificity 86% for SEos >3% while BEos ≥0.22 had sensitivity 68% and specificity 45%. The combination of ACQ ≥1.57 AND BEos ≥0.22 had a lower sensitivity of 37% but a high specificity of 93% and could be useful to exclude rather than identify the presence of sputum eosinophilia.

Conclusion Combined testing using ACQ and BEos only identifies eosinophilia in 37% and can’t replace induced sputum, but may be useful to exclude sputum eosinophilia and prospective study is warranted.