Inclusion of children with airway disease for the development of spirometry reference data

To the Editor:

When establishing normative data for the development of spirometric reference equations, generally measurements have to fulfil internationally accepted criteria [1, 2] and should be derived from data taken from healthy subjects. In their recent publication Lum et al. [3] discussed how “healthy” children should be when selecting reference samples for spirometry. They investigated this in the context of a study designed to explore ethnic differences in the lung function of school children aged 5–11 years from London. They recommended including children with current respiratory tract infection, a history of prior asthma or minor pre-existing risk factors, such as prematurity and low birth weight in normative analysis.

In contrast, we have come to a different conclusion, and we advocate that recommendations for patient samples should be less inclusive in the development of reference data. Our group analysed data from our LUNOKID study (LUng function NOrmal values for KIDs in Germany), where we measured lung function (N=5104; aged 4–18 years) between 2007 and 2009 in three German communities under field conditions [2, 4, 5]. Spirometric reference values were developed with the same regression model used by the Global Lung Initiative (GLI) [6]. For the reference data set, the following subgroups (not overlapping) were excluded. 1) Subjects with asthma diagnosis ever (n=417); a) with no current asthma medication (n=195) and b) with current asthma medication (n=222). 2) Upper respiratory tract infection (RTI) on the day of the investigation (n=734) (without asthma). 3) lower RTI within 6 weeks prior to testing (n=180) (without asthma and without upper RTI). 4) Children who have ever been diagnosed with wheezy bronchitis (“obstructive”, “asthmatic” or “spastic bronchitis”) (n=629) (without asthma and either an upper or lower RTI).

From the total group, 3205 children fulfilled American Thoracic Society (ATS)/European Respiratory Society (ERS) quality criteria. Compared to healthy children, the proportion of visually acceptable manoeuvres was lower in children with upper RTIs on the day of investigation (72% and 62%, respectively; p<0.01), whereas it was higher in asthmatics (80%; p<0.01). No further statistically significant differences related to the fulfilment of quality criteria were observed.

Using all acceptable tests, mean z-scores were then calculated for the subgroups (table 1). As suggested by Hall et al. [7] and Thompson et al. [8] a difference in z-score of 0.5 was considered relevant. The mean LUNOKID based z-scores for the healthy reference children are zero by definition [4]. Children with a history of physician diagnosed asthma had relevantly lower mean z-scores for forced expiratory volume in 1 s (FEV₁)/forced volume capacity (FVC) irrespective of current treatment (−0.52 and −0.66). Furthermore, the standard deviation for this group was higher than the expectation (=1), and it was higher than in the other groups. No relevant mean differences were found for the other subgroups or for the total study population when all subgroups were included. These findings are in accordance with the observations of Lum et al. [3]. However, the percentage of values below the lower limit of normal (LLN) for FEV₁/FVC is larger than the expected 5% in the total group, if these additional subgroups are included.

Many children with an upper RTI on the day of testing were excluded from our reference population because they had not been able to produce technically acceptable manoeuvres. This finding is in accordance with the findings of Lum et al. [3] who also showed that children who were symptomatic on the day of investigation had a higher failure rate in performing technically acceptable manoeuvres. Acceptable tests, however, were indeed not different from our healthy reference population.

We agree that in order to establish the reference data the reference population should be as large as possible. Lum et al. [3] propose to include children with current upper respiratory symptoms and/or a history of asthma. Our data confirm that spirometry data from children with upper RTIs and no asthma diagnosis may not be different from healthy children's data in cases where the ATS/ERS quality criteria (including careful visual control) are fulfilled. Statistically, there is also no reason to exclude children without an asthma diagnosis but a history of lower RTIs within 6 weeks of testing, or a history of wheezy bronchitis from the reference population. However, clinical assessment may differ between investigators, thereby potentially influencing the results. The correct diagnosis of an RTI as upper or lower, with or without obstruction, may be difficult for field staff to ascertain; therefore, strict criteria to define a healthy population should be adhered to. While success rates in children with a history of asthma may be higher, the differences in results (especially FEV₁/FVC) are clinically relevant.

Although the mean reference data would only slightly change when including asthmatics because of the low prevalence of asthma, for clinical reasons, we would like to challenge the recommendation generated by Lum et al. [3] to include children/adolescents with a present or past history of asthma or current respiratory tract infection in a reference population for spirometry. Inclusion will result in higher standard deviations and lower LLNs, which may finally negatively affect diagnostic accuracy.