Background NICE draft guidance for the diagnosis of childhood asthma proposes algorithms based on four tests of lung function (FEV1/FVC ratio, bronchodilator reversibility [BDR], FeNO, PEFR variability); a minimum of two tests must be positive to make a diagnosis. For FEV1/FVC ratio, the proposed cut-off for a positive test is <70%, or the lower limit of normal (LLN), which is neither defined nor widely available. In this algorithm, spirometry is the first-line investigation, and children with FEV1/FVC > 70% are not offered BDR. However, the diagnostic test accuracy for FEV1/FVC and BDR is unknown. Within the setting of a population-based birth cohort we investigated the value of FEV1/FVC and BDR in diagnosing asthma.
Methods We assessed study participants at clinical follow-up at age 16 years using validated questionnaires and lung function measurement. Spirometry was measured according to ATS/ERS guidelines. Using the Asthma UK reference equations, we calculated LLN for FEV1/FVC. BDR was considered positive if FEV1 increased by >12% following administration of 400 mg of salbutamol. Current asthma was defined as all three of: (1) doctor-diagnosed asthma ever, (2) wheezing in the previous 12 months and (3) current use of asthma treatment. We assigned children negative to all three features as a non-asthmatic control group.
Results Spirometry was available for 630 children (325 boys, age range 13.1–16.9 years), of whom 74 (11.7%) had current asthma and 403 were assigned as non-asthmatic controls. FEV1/FVC was significantly lower among asthmatics (84.1% vs. 89.2%, p < 0.001, Figure 1). Ten children (1.6%) had FEV1/FVC <70% (two in asthma group). Discriminative ability of FEV1/FVC < 70% was poor (Receiver operating characteristic curve, AUC = 0.70; sensitivity = 2.7% [2/74], specificity = 98.8% [398/403]). For the calculated FEV1/FVC LLN (74.8% for boys, 78.2% for girls), 28 children (4.4%) had FEV1/FVC<LLN (11 in asthma group). Discriminative ability of FEV1/FVC<LLN was poor (sensitivity 14.9% [11/74]; specificity 97.0% [391/403]). BDR was positive in 54 children (8.7%), of whom 12 had asthma. Discriminative ability of BDR was poor (AUC = 0.64, sensitivity = 16.2% [12/74], specificity = 93.5% [373/399]). Combining these two tests did not result in a better diagnostic accuracy (sensitivity = 2.7%, specificity = 99.0%).
Conclusions FEV1/FVC < 70% or <LLN, and BDR > 12% have a poor diagnostic accuracy as tests for childhood asthma.