Value of the Bronchodilator Response in Assessing Controller Naïve Asthmatic Children

Methods 

Spirometry 

Pulmonary function testing was attempted in children aged ≥4 years in the standing position. Spirometric results were included in the analysis only when the child completed at least 3 baseline forced vital capacity (FVC) maneuvers that met American Thoracic Society criteria in a maximum of 6 attempts and was able to successfully complete post-bronchodilator (BD) spirometry.15 An observation of the flow expiratory curve was made to ensure that the forced expiratory time (FET) was >1 second in all age groups, particularly in the 4- to 7-year-old population.17 In addition, the software had a computer bell that sounded when the curve reached completion. The best spirometric measures of at least 3 attempts were recorded for analysis, including FVC, FEV1, FEV1/FVC ratio, and the forced expiratory flow (FEF25-75) predicted. Post-BD spirometry was evaluated 10 minutes after administrating 2 puffs (180 mcg) from an albuterol metered dose inhaler with a spacer, or nebulized Albuterol pre-mix (.083%) at a dose equivalent to 2.5 mg of albuterol. The latter was used in younger patients or when the metered dose inhaler technique was not successful. Completed and acceptable spirometric measures were compared with the Knudson Intermountain Thoracic Society normal predicted values and adjusted for ethnic values on the basis of a parent report of ethnicity or race.18 Spirometry was performed by a nurse trained by the asthma specialist. The spirometric equipment was purchased from Creative Biometrics International (WIN DX, version 1.00.54). This instrument gives visual incentives to either blow out candles or blow up balloons. It was calibrated on a daily basis. Because FEV1 is the guideline “gold standard” for assessing airway obstruction,1 it was the only baseline measure used in this analysis. The BDR was calculated as follows15:

Results 

Demographic characteristics of both the asthmatic and non-asthmatic populations that successfully completed both pre- and post-BD maneuvers are shown in Table I. There were 346 children with asthma and 51 children without asthma, with an age range of 4.5 to 17.8 years and 4.2 to 15.5 years, respectively. The mean height for each group was similar. None of the group differences was statistically significant. Evaluation of symptom-based classification of asthma severity1 revealed that 34% of children had mild intermittent (MI) disease, 12% of children had mild persistent (MIP) disease, 26% of children had moderate persistent (MOP) disease, and 27% of children had severe persistent (SP) disease (Table I). The 51 patients without asthma consisted primarily of children with recurrent upper respiratory tract problems, including allergic rhinitis, adeno-tonsillitis, or recurrent sinusitis.

Table I. Characteristics of this controller naïve pediatric asthma population described by means of asthma diagnosis

		Post-albuterol response measured		Significance of asthma Dx group difference
		No asthma (n = 51)	Asthma (n = 346)
	Age, mean (95% CI)	9.0years(8-10)⁎	9.8years(9,10)⁎	t = −1.68, P = .094
	4-7 years	19(37%)	115(33%)
	8-10	22(43%)	113(33%)
	>10	10(20%)	118(34%)
	Sex			χ2(1 df) = 0.11, P = .745
	Male	31(61%)	202(58%)
	Female	20(39%)	144(42%)
	Ethnicity			χ2(1 df) = 0.66, P = .417
	Hispanic	45(88%)	290(84%)
	Other	6(12%)	56(16%)
	Height, mean (95% CI)	134cm(129-138)⁎	138cm(135-140)⁎	t = –1.27, P = .203
	Symptom severity
	Mild intermittent		119(34%)
	Mild persistent		42(12%)
	Moderate persistent		91(26%)
	Severe persistent		94(27%)

DX, Diagnosis.

⁎ 95% CI.

Significant differences in the unadjusted (univariate) and adjusted (multivariate) models were observed in the pre- BD FEV1 of 91% predicted for the children with asthma and 97% predicted FEV1 for the children without asthma, (P = .005, P =.003, respectively; Table II). Comparing children without asthma with children with asthma of different severity, statistical significance was seen starting with those with MIP (P = .037). On average, younger children (4-7 years old) and older children (>10 years old) had higher pre- and post-bronchodilator values compared with children 8 to 10 years old across severity categories. There were no significant differences in the post-BD FEV1 between the non-asthmatic group and asthmatic group regardless of severity.

Table II. Examination of asthma diagnosis and symptom severity group differences in average FEV1%-predicted

	Asthma Dx and symptom severity	Patients, n (%)	FEV1% pre pred, mean (95% CI)	FEV1% post-bronchodilator, mean (95% CI)	BDR-Δ FEV1% initial, mean (95% CI)
	No asthma group	51(12.8%)	97(93-100)	99(95,102)	2.2%(0.2,4.3)
	Asthma Dx—overall	346(87.2%)	91(89-92)	99(97,100)	8.6%(7.5,9.8)
	Mild Intermittent	119(30.0%)	93(91-96)	100(97,103)	7.6%(5.8,9.5)
	Mild Persistent	42(10.6%)	91(87-95)	98(95,102)	7.3%(4.2,10.4)
	Moderate Persistent	91(22.9%)	90(87-94)	99(95,102)	9.1%(6.9,11.3)
	Severe Persistent	94(23.7%)	87(84-91)	97(94,101)	10.1%(7.6,12.6)
	ANOVA
	Asthma Dx and symptom severity
	Group differences	Unadjusted model⁎	F=3.8(P=.005)	F=0.4(P=.796)	F=5.2(P<.001)
		Adjusted model†	F=4.0(P=.003)	F=0.4(P=.824)	F=5.2(P<.001)
	Contrast comparisons (adjusted models)
	No asthma versus asthma Dx		(P=.003)	(P=.779)	(P<.001)
	No asthma versus intermittent symptoms		(P=.107)	(P=.818)	(P=.002)
	No asthma versus mild persistent symptoms		(P=.037)	(P=.701)	(P=.019)
	No asthma versus moderate persistent symptoms		(P=.014)	(P=.957)	(P<.001)
	No asthma versus severe persistent symptoms		(P<.001)	(P=.492)	(P<.001)

Pred, Predicted. Δ = change; analysis performed after removal of outliers for BDR (7 values ≥70.0%). FEV1% pre: (Age, P < .001). FEV1% post: (Age, P = .003). BDR-% change from initial FEV1mls: no factors significant in model that controls for asthma Dx and symptom severity.

⁎ Variable Coding: no asthma, intermittent symptoms, mild persistent symptoms, moderate persistent symptoms, and severe persistent symptoms. † Additional factors considered in adjusted (multivariate) models: age (4-7, 8-10, >10 years), ethnicity (Hispanic and other), sex, and height (cm). Significance of asthma diagnosis and symptom severity variable in each model adjusted for these significant demographic characteristics, described by outcome:

The BDR was found to give a more highly significant differentiation between the asthmatic group and non-asthmatic group (P < .001) than pre-BD FEV1 (Table II), with statistical significance noted even in the MI disease group (P = .002). The BDR in the non-asthmatic group showed a mean of 2.2% (95% CI, 0.2-4.3) response compared with 8.6% (95% CI, 7.5-9.8) in the asthma group (P < .001). The range seen was 7.6% (95% CI, 5.8-9.5) for MI asthma to 10.1% (95% CI, 7.6-12.6) in the SP group. In the multivariate analysis, the only variable that was significant was severity (P < .001). Age, ethnicity, sex, and height did not impact the positive relationship of asthma diagnosis and increased severity to BDR. Some children, however, were unable to perform pre- and post-BD responses and were not included in this analysis. Less than 50% of children <6 years old were able to adequately perform both responses, 74% of children aged 6 to 7 years, 80% of children aged 8 to 10 years old, and 96% of children >10 years old.

ROC analysis to identify asthma by BDR was performed (Figure 1; available at www.jpeds.com). The area under the curve was 0.674 (P < .001). The cutoff points closest to the left-hand border and the top border of the ROC space that provide the best possible tradeoff between sensitivity and specificity are presented below the plot. A cutoff point ≥9% offered optimal balance with a sensitivity rate of 42.5% and a specificity rate of 86.3%, a positive predictive value of 95.5%, and a negative predictive value of 18.1%. Applying a cutoff point ≥12% gave a better specificity rate of 94.1%, but the sensitivity rate was substantially reduced to only 30.4%, the positive predictive value was 97.2%, and the negative predictive value was 16.7%. The percentage of those with ≥9% BDR increased with BA severity regardless of baseline FEV1 (data not shown).

View Large Image Download to PowerPoint

Figure 1. ROC analysis to identify patients with asthma from FEV1 BDR. AUC, Area under the curve; PVP, positive predictive value; NPV, negative predictive value. The triangle denotes change.

In Figure 2A, we compare the percentage of abnormal spirometry results, either baseline FEV1 <80%, BDR ≥9%, or both, between the asthmatic group and the non-asthmatic group and across severity. The percentage of children with a FEV1 <80% either alone or in combination with BDR ≥9% ranged from 19.7% to 24.7% in children with MI BA to children with SP BA (P = .704). In the non-asthmatic group, the FEV1 <80% alone was 7.8%, and none had the combination criteria. In contrast, the BDR >9%, either alone or in combination with FEV1 <80%, ranged from 13.7% in the non-BA group to 51.1% in the group with SP BA (P = .002). Using either/or criteria, children with asthma were significantly more likely to have abnormal spirometry results (49.3%, compared with 21.5% of the children without asthma; P = .001). Although non-significant, the percentage of abnormal spirometry results increased with severity, reaching 57.1% in the SP group (P = .323). However, 44.4% of children with MI BA also had abnormal spirometry results. The BDR ≥9% contributed to most of the abnormal findings across severity, most of which were seen when the baseline FEV1 was >80% (normal). In Figure 2B, with the BDR cutoff point ≥12%, we observed the same pattern, but abnormal spirometry results were found less often in the non-BA group (13.7% compared with 21.5%), reflecting greater BDR specificity, with decreased sensitivity particularly noted in the MI group in which abnormal spirometry results were seen in 34.2%, compared with 44.4%. This was mainly caused by decreased positive BDR from 35.9% to 21.3% comparing the ≥9% with the ≥12% cutoff point criteria.

View Large Image Download to PowerPoint

Figure 2. A, Percent of patients in each category defined by baseline FEV1 <80% and BDR ≥9% described by asthma diagnosis status and symptom severity. FEV1 ≥80% BDR <9%, FEV1 ≥80% BDR ≥9%, FEV1 <80% BDR <9%, FEV1 <80% BDR ≥9%. B, Percent of patients in each category defined by baseline FEV1 <80% and BDR ≥12% described by asthma diagnosis status and symptom severity. FEV1 ≥80% BDR <12%, FEV1 ≥80% BDR ≥12%, FEV1 <80% BDR <12%, FEV1 <80% BDR ≥12%.

Discussion 

We have shown in a controller naïve, inner city pediatric population that those in whom asthma is diagnosed on a clinical basis by an asthma specialist had significantly greater mean BDR, even at the mildest level, compared with those who were deemed non-asthmatic, regardless of age, sex, height, or ethnicity. This distinction was clearly better than that shown by means of baseline FEV1 (Table II; Figure 2A and B). Several reports have shown that the mean BDR can differentiate BA from non-BA by using impulse oscillometry in young children20, 21 and with conventional spirometry,10, 22, 23, 24 even when baseline values were in the reference range.22, 24

However, the major usefulness of the BDR in clinical practice requires a cutoff point that best distinguishes children with asthma from children without asthma. This has previously been estimated by epidemiological data in healthy children25 and by evaluating variability limits on repeated measurements of baseline FEV1.26 The Dales study found the ≥9% cutoff point included the 95th percentile confidence limits of healthy children, aged 9 to 11 years.25 Strachan reported that the 95th percentile for variability of FEV1 within the same day and between days in children 7 years old was 10.2%.26 The most direct approach, however, was taken by Dundas et al, who, in the only earlier report we could find, compared several BDR cutoff points in healthy children and children suspected to have mild, step 1 asthma assessed clinically by a physician.10 Similar to our finding, a BDR of ≥9% gave a better sum of sensitivity and specificity, 50% and 86%, respectively, than ≥12%, which had a sensitivity rate of 35% and a specificity rate of 98%. In our step 1 asthmatics (MI), the sensitivity rate was 37% at the >9% cutoff point, perhaps because we used 180 mcg, which is a frequently used dose in children,11 compared with 400mcg in the other study. Dundas also observed that baseline FEV1, although reduced, was neither sensitive nor specific in identifying the child with asthma.10

Perhaps the major value of the BDR for the clinician was shown in those with step 1 MI BA on the basis of symptom frequency criteria (Figure 2A,B). By using combination criteria of FEV1 ≤80%, BDR ≥9%, or both, we found that 44.4% of the children classified as having MI BA had abnormal spirometry results and thus could be candidates for controller therapy. Although guidelines currently do not recommend controller therapy for MI BA, we recently reported that 24.3% of those classified as having MI BA had significant exacerbations resulting in hospitalization or emergency outpatient visits in the previous year.27 Robertson et al found that as many children with clinical evidence of mild BA die as those who have more serious disease.28 Furthermore, children with mild disease experience far less morbidity when appropriately treated.29

For diagnosis, a stepwise algorithm has been suggested in children characterized by a careful history of chronic cough, recurrent wheezing, and dyspnea relieved by bronchodilator and exclusion of alternative diagnosis.30 To confirm the diagnosis, an objective test of airway obstruction is recommended.1, 30 Furthermore, when obstruction is demonstrated, a BDR ≥12% is suggested as more definitive evidence of BA diagnosis.1, 15, 30 We would argue that these spirometric assumptions may not be valid in the pediatric population. First, as seen in our data (Table II), most children have baseline FEV1 in the reference range (≥80%), with approximately 10% to 20% of children having levels <80% regardless of severity.7 We found that a BDR ≥9% or ≥12% can be present even with a baseline ≥80% (Figure 2A,B). In addition, a ≥9% BDR increases sensitivity by 35% overall and by 68.5% in those with MI disease in which sensitivity is needed most; 44.4% of the MI group would have been under-treated if we assume that children with abnormal spirometry results should receive controller medications even if BA is clinically mild. To substantiate this assumption, several investigators have shown that patients with a low FEV1 in childhood often continue to have more severe airway obstruction in adulthood31 and have an increased morbidity rate.32 In addition, an increased BDR has been shown to be associated with biomarkers of inflammation including exhaled nitric oxide11, 12, 13 and bronchial eosinophilia.14 Furthermore, the BDR is a predictor of future lung function33 and correlates significantly with responsiveness to inhaled corticosteroids.14, 34 We have demonstrated that spirometry including the BDR can be successfully performed in most children ≥6 years of age. Although in many primary care settings the peak expiratory flow is used in place of spirometry to assess pulmonary function, it has been shown to correlate poorly with the FEV16 and, therefore, it can be misleading in evaluating the child suspected of having BA.

Several limitations need to be addressed in this observational, retrospective study. Our pediatric population was predominantly Hispanic, from the inner city with poor access to medical care. Although we did not find any significant spirometric differences on the basis of ethnicity, one cannot necessarily generalize our findings to other populations. Furthermore, our non-asthmatic population did not consist of “normal” prescreened healthy control subjects because they were referred, usually by the school nurse on the basis of perceived respiratory problems. Approximately 8% had a FEV1 <80%, and approximately 14% had a positive BDR ≥9%. In addition, we found that approximately 33% of subjects were atopic on the basis of positive skin test results (data not shown). The atopic state itself has been associated with an increased BDR.35 One would expect that the differences between patients in whom BA is diagnosed and prescreened healthy non-atopic subjects might be even greater than reported here. Our cohorts might better reflect the “real world” situation faced by the clinician in practice, where the asthmatic child must be differentiated from a population of children with a variety of respiratory symptoms, not healthy children with no symptoms. We also did not account for albuterol use 4 to 6 hours before spirometry. None of the subjects had received long-acting beta agonists. However, those reporting albuterol use ≥3 times per week had lower baseline FEV1 (P < .014) and greater BDR (P < .001; data not shown). One could assume that our results would have been even better had we excluded children receiving albuterol 4 to 6 hours before spirometry. Patients were also not excluded on the basis of earlier upper respiratory infection within 2 weeks of spirometry. However, this would apply to both patients diagnosed with BA and the non-BA population. The standard we used to determine the value of spirometry was the diagnosis by the asthma specialist on the basis of recommended guideline clinical criteria alone,1 which has frequently been the standard against which more objective tests have been evaluated.10, 13

Our data suggest that spirometry including the BDR, although objective, provides only modest sensitivity in confirming the diagnosis of BA. Of greater potential is our observation that the BDR in combination with baseline FEV1 can detect a population of children with asthma with only mild clinical manifestations who might benefit from inhaled corticosteroid therapy. Although confirmation is needed, we suggest that a BDR ≥9% be considered a positive response in children. We recommend that the BDR be performed in all children ≥6 years old who are suspected of having BA as a practical tool that may help the physician decide which therapeutic strategy is most appropriate.