INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Assessment of bronchodilator responsiveness is a well-recognized method in the armamentarium of diagnostic tests for childhood bronchial asthma (1). By tradition, asthma is often determined from an increased measured flow or reduced airway resistance exceeding the 95th percentile of bronchodilator response in healthy subjects. Typically, a 9% to 15% increase in FEV₁ is indicative of asthma in both children and adults (1, 4). The response in healthy young children to an inhaled ₂-agonist has rarely been evaluated, and the sensitivity and specificity of the response to a ₂-agonist as a means for discriminating asthmatic from healthy children aged 2 to 5 yr have not previously been reported. Children under 6 yr of age cannot reproducibly perform lung function measurements that require active cooperation. Therefore, the diagnosis of asthma in this population is largely based on the history as provided by parents, and on physical examination, the exclusion of other diseases, and the response to pharmacologic manipulation. Objective measurements would be desirable to increase the specificity of diagnosis, minimize over- or undertreatment, and improve the monitoring of disease activity. We recently suggested specific airway resistance (sRaw) (by whole-body plethysmography), respiratory resistance measured with the interrupter technique (Rint), reactance and resistance at 5 Hz measured with the impulse oscillation technique [IOS] (Xrs5 and Rrs5, respectively), as objective measurement of lung function during tidal breathing in nonsedated preschool children.

The aim of this study was to delineate the abnormal bronchodilator response in young asthmatic children from the normal bronchodilator response in healthy young children by using sRaw, Rint, Xrs5, and Rrs5 as methods of lung function testing. We also wished to determine and compare the sensitivity, specificity, and predictive values of each method.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Asthmatic children. Children aged 2 to 5 yr and attending the outpatient pulmonary service clinic of our hospital with a diagnosis of asthma were eligible for the study. The asthma diagnosis was defined empirically and based on typical asthma symptoms such as a recurrent wheeze, cough, and breathlessness; symptom relief with inhaled corticosteroids (ICS) or an inhaled ₂-agonist; and relapse during interruption of such treatment.

No signs or symptoms of acute asthma were present on the study days. Current medication for asthma was left unchanged during the last month before the study visit, and short- and long-acting ₂-agonists were withheld for at least 8 and 24 h, respectively, before the reversibility test was done.

Control subjects. Children without any prior lung illness, atopic symptoms, atopy in first-degree relatives, or smoking at home were considered eligible for recruitment into the study as controls. The control group was recruited through a health questionnaire mailed to families of young children aged 2 to 5 yr and living in the greater Copenhagen area. The families to whom the mailings were sent were randomly selected from the municipal population data base for Copenhagen.

The parents of all children in both study groups gave written informed consent to the study protocol, which was approved by the local ethics committee of Copenhagen (KF 02-117/97).

Pulmonary Function Testing

Pulmonary function testing was done with a Master Screen (Erich Jaeger GmbH, Würzburg, Germany). Methods and equipment have previously been described in detail (7). Measurements were made through a face mask (Astratech No 2; ASTRA, Albertslund, Denmark) equipped with a flexible, noncompressible mouthpiece that ensured support of the cheeks, mouth breathing, and stable access to the airways, and which prevented nose-breathing. Flow and volume were measured with a heated screen type pneumotachograph with a screen resistance of 36 Pa · s · L¹, flow range of 0 to + or 20 L/s, accuracy of +2%, and frequency response from 0 to 50 Hz.

sRaw

sRaw was calculated as the relationship between simultaneous variations in respiratory flow and variations of the volume recorded by a constant-volume whole-body plethysmograph. The calculation was based on the maximum changes in plethysmographic volume during inspiration and expiration (9, 10). Compensation for body temperature and barometric pressure at water vapor saturation (BTPS) was achieved electronically (9). The child undergoing testing was coached to reach a respiratory rate of 30 to 45 breaths/min. When needed, an adult accompanied the child (9). The relationship between flow and pressure was displayed on-line. The median value of five sequential breaths resulting in resistance loops of similar appearance was retained.

Rint

Rint is defined as the ratio of the difference between mean alveolar pressure and airway opening pressure to flow measured at the mouth. The technique for calculating this is based on the assumption of immediate equilibration between mouth and alveolar pressure and a simple relation between this pressure and the airflow after reopening of the shutter valve attached to the pneumotachograph. The shutter valve is electrically controlled by a motor with a short occlusion time of approximately 7 ms. The mechanical component that closes the flow channel is a rotating flap. After inspiration of 50 ml of air, the shutter occludes the airway opening for 80 ms. The pressure is read at the end of the occlusion period. A pressure plateau is not reached because the subject is breathing, which therefore increases the alveolar pressure sensed at the mouth. After opening of the shutter and an additional dead time of 70 ms, the exact flow fitting the registered end-pressure is measured. A plateau would be considered an artefact. Acceptance of the measurement assumes a variance of less than 15% of the Rint parameter and identical pressure and flow curves (7).

The mean value of five consecutive, technically satisfactory measurements as just defined (identical pressure and flow curves) was saved as the result.

Xrs5 and Rrs5

Xrs5 and Rrs5 were measured with the IOS. Rectangular impulses containing the entire frequency spectrum were generated mechanically by a loudspeaker and applied to the respiratory system through the mouthpiece of the spirometer. The resulting pressure and volume signals were analyzed for amplitude and phase difference to determine Xrs and Rrs of the respiratory system at a frequency of 5 Hz. Thirty seconds of undisturbed breathing at a frequency of 20 to 40 breaths/min and resulting in technically satisfactory measurements was used for outcome (7, 8, 10). Coherence limits were not applied as quality criteria, since this did not add to the clinical usefulness of the IOS method in young children, who often exhibit low values of the coherence function (10).

Reversibility Test

Duplicate measurements were made for each lung function method in the sequence: Rint, IOS, and sRaw measurements were done at baseline and 20 min after administration of 500 µg of terbutaline (2 puffs of 250 µg each) or placebo (2 puffs) inhaled from a pressurized metered dose inhaler (pMDI) connected to a metal spacer (Nebuchamber; ASTRA, Lund, Sweden) and a face mask. After each puff the subject took 10 tidal breaths through the face mask.

All pre- and postbronchodilator measurements were made by the same observer.

Study Design

The healthy volunteer subjects visited the outpatient clinic on two separate occasions not more than 1 wk apart, and received their inhalations in a randomized, double-blind, placebo-controlled fashion.

The asthmatic subjects were scheduled for a single visit to the outpatient clinic, where lung function and reversibility testing was performed exclusively with terbutaline, as described earlier.

Data Analysis

Baseline data were compared for the two study groups and also with reference values calculated as percents of predicted values according to height (10). Bronchodilator responsiveness as measured by changes in each lung function parameter was calculated in four different ways: (1) as an absolute value ( abs); (2) as a percentage of the initial value (% init); (3) as a percentage of the predicted value (% pred) (10); and (4) as the number of intrasubject SD units (SDw units) using the relevant SDw for each lung function test. Calculation of the relevant SDw was made from the difference between paired baseline measurements divided by (i.e., for sRaw: 0.109 kPa · s; for Rint: 0.078 kPa · s · L¹; for Xrs5: 0.104 kPa · s · L¹; and for Rrs5: 0.131 kPa · s · L¹) (10). Therefore, calculation of the bronchodilator response was: (Postbronchodilator value  baseline value)/SDw.

The aim of this SDw index is primarily to allow comparison between measurements with different repeatability. Indices of bronchodilator responsiveness were plotted against baseline lung function and the results were analyzed by linear regression analysis. The response decision level for the various parameters was determined in steps of 0.5 SDw for the range of changes from 0 to 7 SDw, and analysis of receiver operating curve (ROC) characteristics was performed by plotting sensitivity versus 1  specificity for each possible cutoff level. By this method it is possible to identify the optimal cutoff point that discriminates most efficiently between the absence or presence of disease. The point on each ROC curve that is closest to the top left-hand corner is the optimal cutoff point. Comparison of lung function tests for sensitivity, specificity, and predictive value was done with the optimal cutoff point as determined from the ROC curve analysis.

Central tendency was expressed by the arithmetic mean and 90%, standard deviation (SD), or range. The 5th and 95th percentiles were used to define the lower and upper limits of normal response. The repeatability of baseline measurements was investigated through Altman-Bland plots (11), calculation of bias, 95% limits of agreement, and intraclass correlation coefficients. The paired t test was used to compare values before and after medication. The unpaired t test was used to assess differences between groups. A value of p < 0.05 was taken as statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Baseline characteristics of patients and healthy control subjects are presented in Table 1. Reversibility testing was done on 55 asthmatic children ranging in age from 2.3 to 5.9 yr. They had used regular inhaled antiasthma medication for a mean period of 20 mo (range: 2 to 57 mo). Thirty-three (60%) had a first-degree relative with atopic disease and 19 (35%) were exposed to passive smoking at home. Twenty-six (47%) of the patients had other manifestations of atopy (atopic dermatitis and/or rhinitis), and 20 (36%) had at least one clinically relevant positive skin prick or radioallergosorbent test. At the time of the study, 40 (73%) patients were treated with inhaled budesonide at a mean daily dose of 387 µg (range: 100 to 800 µg/day) and a short acting ₂-agonist as required, delivered via a pMDI and Nebuchamber. The rest of the patients were being treated with a short acting ₂-agonist as required; one was additionally being treated with the long acting ₂-agonist salmeterol on a regular basis. All asthmatic children were fully acquainted with the use of a face mask from their regular asthma treatment, and complied with the protocols for the study measurements at the first visit. Forty-one healthy children were invited for testing, but three 2-yr-old children and one 3-year-old child failed to comply with the measurement techniques, mainly because of anxiety in using the facemask; the remaining 37 children, aged 2.5 to 5.9 yr, completed the measurements.

Data on baseline and postbronchodilator lung function in asthmatic and healthy subjects are given in Table 1. Asthmatic subjects had significantly increased values of sRaw (p < 0.001), Rint (p < 0.01), and Rrs5 (p < 0.001), and decreased values of Xrs5 (p < 0.05) at baseline as compared with healthy control subjects. When comparison was made with the reference values from a previous study (10) (Figure 1), 23 (42%), 22 (40%), three (6%), and one (2%) asthmatic patients had abnormal baseline values (outside the 95% reference range) for sRaw, Rint, Rrs5, and Xrs5, respectively. Correspondingly abnormal results were found for one (3%), six (16%), none (0%), and two (5%) healthy control subjects.

View larger version (22K): [in this window] [in a new window]   Figure 1.   Absolute values of sRaw, Rint, Xrs5, and Rrs5 in relation to height in centimeters at baseline (before terbutaline) in healthy and asthmatic children. The middle line in each graph represents the predicted mean value, and the parallel upper and lower lines represent the upper and lower 95% predicted levels. Asthmatic subjects = open squares; Healthy subjects = closed triangles.

Analysis of the repeatability of measurements in healthy control subjects, within the two replicates at baseline in one day, between baseline replicates on different days, and for replicates after treatment, did not show any significant bias (data not shown).

The indices of bronchodilator response (abs, %init, and %pred) after administration of 500 µg of terbutaline are given in Table 2 for each lung function parameter. The bronchodilator response as reflected by each of the four measures used in the study and expressed as %pred is shown in Figure 2. Since %pred expresses the absolute response independent of the initial value, this index seemed preferable. Healthy young children exhibited significant improvement with terbutaline in all four measures (p < 0.001), but also with placebo in all measures (sRaw: p < 0.001; Rrs5: p = 0.002; Xrs5: p = 0.02) except Rint (p = 0.47). However, the bronchodilator response with terbutaline was significantly greater than with placebo when measured with sRaw (p < 0.001), Rint (p < 0.001), and Rrs5 (p < 0.05), but not with Xrs5 (p = 0.39).

View larger version (23K): [in this window] [in a new window]   Figure 2.   Bronchodilator response in healthy and asthmatic subjects and response to placebo in healthy subjects expressed as percent of predicted values (%pred) for sRaw, Rint, Rrs5, and Xrs5. Means are indicated by connecting lines (healthy subjects) and horizontal bars (asthmatic subjects).

Lung function of the asthmatic children in the study improved significantly after inhalation of terbutaline as compared with the terbutaline response in healthy controls. This was reflected in sRaw (p < 0.0002), Rint (p < 0.003), Xrs5 (p < 0.01), and Rrs5 (p < 0.0005), and was independent of the method used to calculate the index of bronchodilator response. No significant difference between asthmatic and healthy control subjects was seen in postbronchodilator lung function measurements (Table 1). Analyses of data for subgroups with and without an atopic predisposition or allergic or other atopic diseases revealed no differences in bronchodilator responsiveness.

Bronchodilator responsiveness showed an unequivocal tendency in all measures toward a greater response in subjects with reduced baseline lung function parameters, as reflected by slopes of regression lines that were significantly different from 0. Except for Rrs5, this relation was also found in healthy subjects (data not shown). Omitting all asthmatic subjects whose prebronchodilator baseline lung function was outside the normal reference interval excluded Rint as a tool for discriminating between asthmatic and healthy subjects (p = 0.99), whereas differences remained significant for sRaw (p < 0.02), Xrs5 (p < 0.01), and Rrs5 (p < 0.0005).

sRaw had the best discriminative capacity as judged by ROC curve analyses (Figure 3); this was followed by Rrs5, Rint, and Xrs5, in the order given. The optimal cutoff level for sRaw was 3 SDw units, giving a sensitivity of 66% and a specificity of 81%. Three SDw units of sRaw correspond approximately to a change of 25% relative to the predicted value.

View larger version (17K): [in this window] [in a new window]   Figure 3.   ROC characteristics showing relationship between sensitivity and 1  specificity of bronchodilator response calculated from changes expressed as intrasubject SD units for sRaw (closed squares), Rint (closed circles), Rrs5 (open circles), and Xrs5 (closed diamonds).

For comparisons of methods, the optimal cutoff level for each test was chosen for calculation of sensitivity, specificity and predictive values. The results are shown in Table 3. The measure with the best overall discriminative capacity at the individual optimal cutoff level was sRaw, followed by Rrs5, Rint, and Xrs5 (Table 3).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We assessed bronchodilator responsiveness in asthmatic and healthy young children applying different measures of lung function, and found that it was possible to segregate asthmatic from healthy control subjects with reasonable reliability. Assessment of bronchodilator responsiveness is widely applied as a routine procedure in the diagnostic workup for asthma in adults and schoolchildren. The present study is the first to quantify and compare bronchodilator responsiveness as measured by whole-body plethysmography, the interrupter technique, and impulse oscillometry in asthmatic and healthy children aged 2 to 5 yr. Healthy control subjects exhibited significant improvements in sRaw, Rint, and Rrs5 after terbutaline as compared with placebo, thereby documenting the presence of a certain bronchomotor tone even in healthy nonatopic controls, whereas improvement in Xrs5 was identical after terbutaline and placebo. Still, the mean changes in all lung function measures in response to terbutaline were significantly greater in asthmatic than in healthy control subjects. The discriminative capacity of whole-body plethysmography (sRaw) exceeded that of the other tests with a sensitivity of 66%, specificity of 81%, and predictive value of a positive test of 84%. Assessment of bronchodilator responsiveness using sRaw as a measure of lung function may therefore help in defining asthma in this clinically difficult group of young, wheezy children.

The baseline measures of lung function in asthmatic children showed a significantly greater airway resistance than in healthy controls, which is in agreement with our findings in previous studies of a random group of young asthmatic children (12) and of selected groups of young asthmatic children (13- 15), probably indicating that these children's asthma disease activity is not revealed by history and clinical examination.

Bronchodilator response may be expressed in different ways, depending on the choice of physiologic test for and purpose of the assessment (16). The controversies over the subject of expression of bronchodilator response in adults and schoolchildren have been extensively covered and critically reviewed in the literature (16). The study by Waalkens and associates (17) recommended using the change in percent of the predicted value (%pred) for a particular measure as the index of bronchodilator response in children, since this method offered independence of age, stature, and initial lung function. We chose to present different indices because of the explorative nature of the study and because comparative data on measurements of bronchodilator response with these methods of lung function testing in preschool children were lacking. However, %pred was preferred as a description of the bronchodilator response, as recommended (17), and also because this index offered a wider range of responses than the other two indices for most of the tests in both healthy and asthmatic subjects.

Only a few reports of bronchodilator responsiveness in young children are available (19, 20), and only one study (19) has taken the normal response to an inhaled ₂-agonist into account. Bridge and colleagues (19) used the interrupter technique in an ambulatory setting in studying 2- to 5-yr-old children with and without respiratory symptoms. They found a baseline level of Rint comparable to our results, but suggested an absolute change in Rint that exceeded 0.21 kPa · s · L¹ as a reflection of significant reversibility. We found that an absolute change of > kPa · s · L¹ reflects the 0.35 limit of normal response for healthy children (Table 2), implying a much greater normal response.

Hellinckx and coworkers (19) recently reported a study of bronchodilator response in healthy and asthmatic children up to 6.5 yr of age using the IOS. They did not find any difference between the two study groups, either in baseline or in bronchodilator responsiveness (19). In contrast, we found differences in both baseline and bronchodilator responsiveness using the IOS. This may be explained by differences in study populations. Our study population consisted of well-defined groups of asthmatic and healthy subjects at each end of a spectrum ranging from disease to health. Hellinckx and coworkers (19) defined asthmatic subjects from a standardized questionnaire that probably has a low diagnostic specificity, and there may have been inadvertent inclusion of nonasthmatic subjects in the asthmatic study group; in fact, only half of their asthmatic subjects were treated for asthma at the time of investigation. They found baseline values for Xrs5 and Rrs5 in healthy children that were identical to the values in the present study. Furthermore, they suggested a cutoff level for a positive bronchodilator response of 40% (%init) for Rrs5, which is slightly above the level in the present study, in which we found cutoff values of 29% when expressed as %init and 27% when expressed as %pred (Table 2). We found no reports in the literature presenting comparable figures for bronchodilator response based on sRaw in young healthy children.

The sensitivity and specificity of a diagnostic method are critically dependent on how well asthma can be defined, as well as on the reliability of the history provided by healthy controls. The asthmatic children in our study were carefully selected on the basis of a history of recurrent asthmatic episodes, the need for current antiasthma therapy, and recurrence of asthmatic symptoms during interruption of treatment with ICS. In this study, the sensitivity of reversibility testing based on sRaw seems to have been comparable to the sensitivity found in a study of asthmatic schoolchildren done with spirometry and a cutoff level of 15% as a significant response (17).

Our results are in agreement with previous observations of a certain bronchomotor tone existing even in healthy, nonatopic preschool children (1, 5, 6). Unexpectedly, we also showed a significant response to placebo as measured with sRaw, Xrs5, and Rrs5. Since no bias between baseline replicates was seen, these observations suggest better compliance with the use of a face mask and mouthpiece in measurements made after 20 min, but also the observer's expectation of a certain change in lung function. This observation emphasizes the need for blinding and placebo control even in "objective" measurements, the lack of which may have hampered the results and conclusions about the bronchodilator response in asthmatic subjects in the present study as well as in previous studies of bronchodilator responsiveness in general. Still, the bronchodilator response with terbutaline was highly significantly more pronounced with all of the measures used in the present study, except Xrs5, in healthy control subjects. In asthmatic subjects, the response to terbutaline was even more pronounced, reflecting the significant difference in baseline lung function as compared with healthy control subjects. The latter was demonstrated by the loss of discriminative power of Rint when subjects with abnormal baseline values were omitted from calculations with this measure. Even after omitting patients with abnormal baseline values, we found significant difference in bronchodilator response between healthy and asthmatic subjects, except when using Rint. We were unable to find comparable subanalyses in this respect in other studies. The finding of a negative correlation between different indices of bronchodilator response and baseline lung function is expected and in keeping with results of other studies of children (17, 19).

From a clinical viewpoint, a positive response to a bronchodilator in anyone suspected of having asthma may only be meaningful in two different scenarios: (1) baseline lung function is within normal levels, but the response is above the cutoff level; and (2) both baseline lung function and response are abnormal. Both of these scenarios are indicative of asthma, but in the first case the reversibility test is of more help than in the latter case. In both cases further characteristics of asthma should be investigated. Demonstration of bronchial hyperresponsiveness (BHR) would strengthen the suspicion of asthma, and in fact we have previously reported the usefulness of sRaw and cold, dry air challenge in the diagnosis of BHR and asthma in young children (13). Thus, a diagnosis of asthma is strongly supported by measurements of sRaw that show an abnormal baseline value and a bronchodilator response of > 25% of the predicted value (3 SDw units), and is further supported if BHR is revealed, even though we do believe that bronchodilator responsiveness and BHR represent different underlying mechanisms of asthma, and that these mechanisms do not necessarily coexist.

In conclusion, this study demonstrates that assessment of bronchodilator responsiveness in young children is possible through the use of sRaw, Rint, and IOS measurements, and that such assessment may serve as a useful tool in the study of bronchial asthma in young children. Whole-body plethysmography (sRaw) has the highest discriminative capacity and serves as the most sensitive method in assessing bronchodilator responsiveness.