INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Many studies have used the inhalation of histamine or methacholine as a means of measuring airway responsiveness (AR) in children and adults (1). The results of many of these studies suggest that younger (smaller) children are more responsive to these agents than are older children and adults (2, 3, 5). Other investigators have recently questioned the interpretation of these findings because of the possibility that larger doses of agonists, in relationship to body size, are delivered to smaller children than to larger children and adults (5, 8).

On the basis of a study of 818 New Zealand children evaluated at 9, 11, 13, and 15 yr of age, Le Souëf and colleagues reported that AR to methacholine was lower in both males and females in the upper quartile for height than in those in the lowest quartile, independent of age (9). Le Souëf and colleagues proposed that the age-dependent variability in AR was primarily due to older and hence taller children more rapidly developing air flows in excess of nebulizer output, and thus receiving a proportionately smaller dose of agonist (9, 10). If Le Souëf and colleagues are correct, much of the literature concerning changes in AR with age in children would be called into question.

In contrast to Le Souëf and colleagues, Peat and coworkers did not find that height was related to the slope of the dose- response curve for histamine after correcting for lung size and airway caliber in 1,613 Australian children (2). In their analysis, Peat and coworkers used FVC as a surrogate for lung size, and FEV₁/FVC as an indication of airway caliber. Because of the high degree of correlation between height and FVC, it is not surprising that height was no longer significant when FVC was added to a multivariable model; however, the higher r² value when FVC was used led Peat and coworkers to conclude that FVC was a more appropriate variable for adjusting AR than was height (2). Others have also suggested relationships between baseline lung function, such as FEF_25-75% or FEV₁, and AR (8, 11, 12). All of these studies raise the question of whether estimates of AR based on methacholine or other agonists delivered from a nebulizer should be corrected for height or baseline pulmonary function.

As part of an epidemiologic study of asthma in 6- to 8-yr-old children, we had evaluated a birth cohort of 482 children. The purpose of the present study was to analyze our methacholine response data with the same dose-response slope index used by Le Souëf and colleagues, in order to learn whether AR varied by height in children or by baseline values of FVC, FEV₁/ FVC, or FEF_25-75% (2, 9). We also evaluated the slope of the terminal portion of the methacholine dose-response curve, to learn whether this index provided different information from that provided by the dose-response curve slope index.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

The selection of children for the study has previously been described (13). Briefly, all pregnant women living in an area of northern, suburban Detroit defined by contiguous zip codes, and belonging to the Health Alliance Plan, a health maintenance organization, were eligible for recruitment if their babies were due between April 15, 1987 and August 31, 1989. Women meeting eligibility criteria for the study were invited during prenatal visits to participate in it. If a woman agreed to participate, a study nurse collected demographic, health, and lifestyle-related data from her. When participating children were between 6 and 7 yr of age, we attempted to comprehensively evaluate all of them for asthma. All participating families who could be contacted and who remained within driving distance of the study center were contacted and invited to participate in the clinical evaluation.

Children were defined as having current asthma if they had been diagnosed by a physician as having asthma at some point during their lives and if they had had asthma symptoms during the 12 mo preceding the evaluation. If the parent reported that a physician had ever diagnosed a child as having asthma, the child was classified as having had a medical diagnosis of asthma. Children were classified as having possible asthma if the parent responded positively to three or more of the following five questions. Has your child ever: (1) had episodes of prolonged coughing lasting 7 d or more; (2) coughed so hard that it precipitated vomiting; (3) had whistling or wheezing in the chest; (4) had wheezing, whistling, or coughing after exposure to dust, fumes, molds, pollens, foods, pets, or drugs; or (5) experienced wheezing, chest tightness, coughing, or breathlessness while at rest, or with physical activity, emotional stress, exposure to cold air, or infections? If the parent gave a history of the child ever having wheezed, with or without other symptoms, the child was considered positive for wheezing. All aspects of the study were approved by the Human Rights Committee of Henry Ford Hospital, and written informed consent was obtained from the child's parent or guardian at entry into the study and before the asthma evaluation.

Skin testing was performed by the puncture technique on the volar aspects of the forearms, using commercial extracts of Dermatophagoides farinae, Dermatophagoides pteronyssinus, cat, dog, Alternaria tenuis, short ragweed (Ambrosia artemisiifolia), and blue grass (Poa pratensis), with saline and histamine (1 mg/ml) controls (all from the Pharmaceutical Division, Bayer Inc., Spokane, WA). A wheal with a product of perpendicular diameters of 4 mm or more and with a surrounding flare of at least 10 mm was considered positive if saline produced no wheal and less than 4 mm of flare. Alternaria was not added to the skin test panel until after the first 33 children had been evaluated, and the prevalence of atopy therefore varies slightly according to whether all skin-tested children are considered or only those having all seven skin tests are considered. Atopy was defined as one or more positive skin tests.

Spirometry and Methacholine Challenge

Lung function was recorded with a KoKo spirometer (Pulmonary Data Service, Louisville, CO) connected to a personal computer and calibrated daily with a 3-L syringe. The children were coached to engage in maximal expiratory maneuvers while standing and without nose clips. Spirometry was performed in accordance with American Thoracic Society standards (14). Predicted values were based on the equations of Polgar and Promadhat (15). Spirometry was considered acceptable if the child made a good effort and if two forced exhalation maneuvers showed reproducibility (± 5% for both) FVC and FEV₁. If the child's FEV₁ was  70% predicted and reproducible, the child was challenged with the normal saline diluent and then with five sequential doses of methacholine (0.025, 0.25, 2.5, 10, and 25 mg/ml) administered with a DeVilbiss 646 nebulizer (DeVilbiss Health Care Inc., Somerset, PA) connected to a French-Rosenthal-type dosimeter (Pulmonary Data Service, Louisville, CO). The nebulizer was driven by compressed air at 20 psi. The dosimeter was set to deliver methacholine for 0.6 s at the initiation of inhalation during tidal breathing. Spirometry was repeated 3 min after each dose of methacholine. Increasing concentrations of metacholine were administered until the FEV₁ fell to less than 80% of the best postsaline value and remained below this level for at least 2 min, or until the maximum concentration of methacholine was reached. A positive methacholine challenge was defined as a decrease in FEV₁ to less than 80% of the postsaline value after inhalation of methacholine at concentrations up to 10 mg/ml. Children who could not undergo the challenge were excluded from the analysis of the challenge test only.

Statistical Analysis

The variable of primary interest was the slope relating the changes in FEV₁ to increasing inhaled doses of methacholine, calculated as suggested by Le Souëf and colleagues and O'Connor and coworkers (9, 16). This slope was defined as: slope = log ([% change in FEV₁/last dose] + 0.55), where the % change in FEV₁ = 100 × (baseline FEV₁  last FEV₁)/last FEV₁. The last dose is the dose that resulted in a 20% decrease in FEV₁ from the postsaline value, or as the 25 mg/ml dose if the FEV₁ had not declined by 20%. If the provocative dose of methacholine causing a 20% decrease in FEV₁ (PD₂₀ methacholine) was bracketed by two of the predetermined doses, the value was determined by linear interpolation. The constant of 0.55 was added to insure that all values were positive, allowing meaningful log transformation of all values. A negative value could occur if the child's FEV₁ had improved during the challenge sequence. As pointed out by O'Connor and coworkers, the value of using an expression of the methacholine-versus-FEV₁ slope is that this represents AR on a continuous scale for all subjects (16). Those with higher values for the slope are more responsive to methacholine.

When the records of the methacholine challenge studies were examined visually, it seemed that the FEV₁ of some children declined gradually with each successive dose of methacholine, whereas the FEV₁ of other children remained relatively constant for several doses and then declined precipitously. In an attempt to analyze this apparent difference in the shape of the methacholine dose-response curve, we examined the percent change in FEV₁ between the next-to-last and last doses of methacholine. This variable, which we termed "end FEV₁ change," was defined as: ([baseline FEV₁ FEV_1B]/FEV_1B)  ([baseline FEV₁  FEV_1A]/FEV_1A), where FEV_1A is the last FEV₁ and FEV_1B is the next-to-last FEV₁. Negative values for this variable indicate a decrease in FEV₁ between the last two methacholine doses, with a larger negative number indicating a greater decrease.

To assess the relationships of the slope of the methacholine dose- response curve and the end change in FEV₁ with other variables of interest, we used two-sample t-tests with nominal variables, and estimated correlations for continuous variables. To assess possible multivariate models, we used a stepwise linear regression approach. Multivariate methods were also used to examine the relationship of methacholine response and height before and after adjusting for FVC % predicted, FEF_25%-75%, % predicted, and FEV₁/FVC in the models. Values of p < 0.05 were considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Four hundred eighty-two children appeared for evaluations for asthma and were therefore available for spirometric testing (58% of the 833 children enrolled at birth). Participating children were predominantly Caucasian (96.3%), and their parents were relatively well educated (61.4% of mothers and 68.3% of fathers had education beyond high school). Table 1 provides other descriptive statistics for the participating children. Methacholine responsiveness was measured in 471 (98%) of the 482 children. Slopes of methacholine dose-response curves could not be estimated for 11 children because of inability or refusal by six to perform reproducible forced exhalation maneuvers and an FEV₁  70% predicted in five. No child had a 20% or greater change in FEV₁ after inhalation of normal saline. By study design, the age range of the subjects was narrow, at 6.1 to 7.7 yr (6.7 ± 0.2 yr [mean ± SD]). The subjects' heights and weights were similar to those expected for children of comparable age (17). The sexes (n = 237 males and 245 females) were approximately equally represented.

As seen in Table 2, 7.1% of the subjects had current asthma, as compared with 10.9% who had ever had a physician diagnosis of asthma, 19.9% who possibly had asthma, and 23.3% who had a PD₂₀ methacholine  10 mg/ml. Atopy was found in approximately one third of the children, with a 6.1% higher prevalence when Alternaria was included among the skin-test antigens.

Table 3 shows the results of comparing the slope of the methacholine dose-response curve, end FEV₁ change, height, and baseline % predicted FVC and FEF_25-75% among children with and without asthma, wheezing, or methacholine responsiveness. Highly significant differences were found for both indices of response to methacholine among children with and without current asthma, physician-diagnosed asthma, possible asthma, and with a history of wheezing. As anticipated, the slope of the methacholine dose-response curve and the end FEV₁ change were significantly different in children considered methacholine responders (PD₂₀ methacholine < 10 mg/ml) and those not so considered. Children with a PD₂₀ methacholine < 10 mg/ml also had lower average FVC and FEF_25-75% values. There were no significant differences between boys and girls in terms of methacholine response indices. Height did not vary significantly in relationship to asthma, wheezing, sex, or methacholine response. Baseline % predicted FVC and FEF_25-75% were significantly lower in children with current asthma, diagnosed asthma, and wheezing. FEF_25-75% also varied significantly by possible asthma. Interestingly, girls tended to have lower baseline FBC values, whereas boys had significantly lower baseline FEF_25-75% values.

Table 4 shows that the correlations between age, height, and weight with either the slope of the methacholine dose- response curve or end FEV₁ change were not significant. However, the correlations between these two methacholine response indices and baseline FVC and FEF_25-75% were highly significant, explaining from 3.6% to 8.9% of the variance in methacholine response.

Table 5 presents the results of four stepwise multiple variable regression models for predicting either the slope of the methacholine dose-response curve or end FEV₁ change. Because of the strong interrelationships between the spirometric variables that served as the bases for the three different definitions of asthma and a history of wheezing, only one of these variables at a time could be considered in a model. The values shown in the table are the p values for the variables entering the model (values of p < 0.05) and for the remaining variables (values of p > 0.05). Not height, weight, age, or sex are significant predictors of either the slope of the methacholine dose- response curve or of the end FEV₁ change in any of the five models. For example, in Model 1 for predicting the slope of the methacholine dose-response curve, a variable indicating a diagnosis of current asthma, if entered into the model, becomes the only significant variable. Adding the variable female sex to the model does not add significant predictive information, as indicated by the p value for female of 0.175. Similarly, height, if added to the same model as a variable, is not significant (p = 0.981). Female sex approaches significance (p = 0.058) in Model 3 when possible asthma is considered.

Table 6 is similar to Table 5, but baseline % predicted FVC, FEF_25-75%, FEV₁/FVC, and atopy are now considered in addition to sex, weight, and age. Again, variables representing asthma and a history of wheezing are each individually significant except for predicting end FEV₁ change, in which case diagnosed asthma and wheezing only approach significance. Baseline FEF_25-75% and FVC, as % predicted values, are always significant. FEV₁/FVC is significant in these models if the variable FEF_25-75% is removed (data not shown). Even with FEF_25-75% in the models, atopy is significant for the slope of the methacholine dose-response curve. When a term representing atopy is added to the models predicting end FEV₁ change, the p values are larger than those found in models predicting slope of the methacholine dose-response curve. Interestingly, female sex follows a pattern similar to atopy, being significant in models of slope of the dose-response curve but losing significance when predicting end FEV₁ change. Since atopy was significant in most of the models presented in Table 6, we reran the models in Table 5 with atopy included as a variable. The results of these analyses were similar to those presented in Table 6. Atopy was a significant predictor of methacholine dose-response (data not shown), whereas sex, height, weight, and age remained insignificant.

Table 7 shows the actual regression equations for the models presented in the upper half of Table 6, predicting the slope of the methacholine dose-response curve. Even though the models are highly statistically significant, the amount of variability explained by the models is relatively low, as shown by the values for r². The similarities of the intercepts and coefficients in the equations are striking.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study evaluated various combinations of variables proposed by others in order to determine the combinations most predictive of methacholine responsiveness (2, 5, 9, 12, 18, 19). After considering several combinations of variables, we found that the % predicted FEF_25-75%, % predicted FVC, atopic status, and sex were all independently significantly related to methacholine responsiveness in the children we studied. If FEF_25-75% was removed from a model, FEV₁/FVC became significant. These findings are in general agreement with those in previous studies, even though many previous studies considered only one or two of these variables (2, 5, 12, 18).

There are some differences between our findings and those of others. We did not find a significant relationship between height and slope of the methacholine dose-response curve, either when height was used individually or when it was used in combination with other variables, as Le Souëf and colleagues had used it (9). A possible explanation for this difference is that Le Souëf and colleagues evaluated height but apparently not measures of pulmonary function in relationship to AR. Given the relationship between height and measures of lung function, a relationship between height and AR in a large population is probable. Peat and coworkers found that height, FVC, FEV₁/FVC, and sex were each related to AR, but that height lost significance when combined in models with FVC and FEV₁/FVC (2). Our findings with FVC and FEV₁/FVC were consistent with those of Peat and coworkers until we added FEF_25-75% into predictive models, which resulted in FEV₁/FVC losing significance. Peat and coworkers apparently did not evaluate the relationship of FEF_25-75% to AR. The study by Ulrik also found that baseline FEV₁ was related to AR, but did not examine other measures, such as FVC and FEV₁/FVC (12).

We had thought that an important difference between our study and that of Le Souëf and colleagues would be their use of the tidal free-breathing technique in contrast to our use of a dosimeter for delivery of methacholine. In theory, free breathing should allow the inspiratory flow to exceed the output of the nebulizer, meaning that for a variable portion of inspiration, the concentration of inspired agonist will be below the intended concentration (9, 10). The taller the child (the greater the inspiratory flow), the larger will be the proportion of inspiration with a low agonist concentration. This apparently is not an important difference, since both Peat and coworkers and Forastiere and associates used the free-breathing technique, and both groups found stronger relationships between baseline measures of lung function and AR than between height and AR (2, 5).

Other differences in study methods could have influenced the results of these different studies. Peat and coworkers used histamine rather than methacholine as an agonist, whereas Forastiere and associates used methacholine (2, 5). Others have shown the comparability of these two agonists (21). The mathematical expression used to determine the slopes of agonist dose-response curves appears to have been the same in these other studies as in our study (2, 5, 9). It is also possible that differences in the predictive equations for lung volumes could influence results, but the source of predictive equations was not given in these other studies (5, 9, 12). Some sets of predictive equations for children include sex differences and others do not. These differences may or may not be important, because the extent to which sex plays a role in bronchial hyperresponsiveness among children is unclear. Some investigators have observed sex differences while others have not (5, 8, 9, 12). Studies also vary in the schedules they use to increase the dose of agonist. It is possible that a slower rate of increase in methacholine concentration would reveal more subtle differences in AR, but there are few data to support this possibility, and the more spirometries done during a challenge test, the greater the concern that fatigue- or exercise-induced effects may become important.

Two potentially important differences between our study and those of others are sample size and age range (2, 5, 9). Although our sample size of 474 children was smaller than the samples of Le Souëf and colleagues, Peat and coworkers, and Forastiere and associates, we were able to detect highly statistically significant associations between several variables and AR. The studies by both Peat and coworkers and Le Souëf and colleagues had a broader age range of children, and this would be expected to have produced a broader range of heights. Yet Le Souëf and colleagues found that height, and not age, was the more predictive variable even when they restricted their analysis to single age groups (9). Le Souëf and colleagues also had longitudinal data from following the children in their study over a period of 6 yr, which allowed them to carefully study the effects of age on AR. Since we studied our children only on one occasion, we were unable to evaluate age over a broad enough range to allow meaningful observations. It is possible, although we think unlikely, that racial, social, or geographic differences are related to the differences in outcomes of these studies.

In accord with previous studies, we did not find any indication that calculating the end FEV₁ change provided important information beyond that provided by the slope of the methacholine dose-response curve (16, 22). The minor differences between the two indices of methacholine response that we examined and other variables do not appear to be related to important physiologic differences. These minor differences also suggest that differences in calculating the relationship between methacholine dose and lung response are unlikely to result in major differences between studies.

The relatively consistent coefficients and r² values for Equations 1 to 4 in Table 7 show that the relationships between the variables we examined and AR were similar in children characterized as having current asthma, physician-diagnosed asthma, possible asthma, and wheezing. These findings have at least three possible interpretations. The first interpretation is that all four categories represent the same disease process. It is also possible that AR, at least as measured with methacholine, is present at similar levels in all four of these diagnostic groups, and that other factors, beyond AR, produce the differences observed in respiratory symptoms and in the resulting classifications. The third possibility is that there are many errors in assigning individual children to diagnostic groups, obscuring true relationships between AR and the diagnostic groups. The importance of these possibilities is that studies based on comparisons of children with and without asthma may differ because of errors in classification. Unfortunately, our data did not allow us to evaluate the reproducibility or accuracy of the initial classifications of our children.

In conclusion, we found that methacholine responsiveness in children is significantly influenced by the % predicted FEF_25-75%, % predicted FVC, atopy, and sex. Given the consistency of our findings and those of others (2, 5), investigators should consider correcting methacholine responsiveness for baseline % predicted FEF_25-75% before analyzing data for age-related changes in such responsiveness.