Published ahead of print on March 17, 2004, doi:10.1164/rccm.200308-1174OC
© 2004 American Thoracic Society
Progression of Asthma Measured by Lung Function in the Childhood Asthma Management ProgramIra J. and Jacqueline Neimark Laboratory of Clinical Pharmacology; Department of Pediatrics, Division of Allergy-Clinical Immunology; and Division of Biostatistics, National Jewish Medical and Research Center, Denver, Colorado Correspondence and requests for reprints should be addressed to Stanley J. Szefler, M.D., National Jewish Medical and Research Center, 1400 Jackson Street, Denver, CO 80206. E-mail: szeflers{at}njc.org
From the Childhood Asthma Management Program cohort, which was randomly assigned to receive budesonide, nedocromil, or placebo for 4–6 years, we determined the prevalence of and factors associated with at least 1% per year loss in postbronchodilator FEV1% predicted. Participants who had a significant reduction in postbronchodilator FEV1% predicted (SRP), comprised 25.7% of the cohort (n = 990). Using logistic regression, predictors of SRP at baseline were younger age (p = 0.0005), male sex (p < 0.0001), clinic (p = 0.02), and higher postbronchodilator FEV1% predicted (p = 0.02). Examination of the SRPs indicated that the effect of baseline lung function was such that the higher the lung function, the less steep the reduction in postbronchodilator FEV1% predicted (p < 0.0001). A similar proportion of SRPs was found in each treatment group. Among the SRPs, the rate of reduction in postbronchodilator FEV1% predicted was similar in all treatment groups. At a single site where biomarker assessment was performed, SRPs also had more prominent eosinophilic inflammation during the washout period. The course and mechanisms of lung function reduction or slow lung growth velocity in children with asthma must be defined.
Key Words: asthma progression • markers of inflammation • airway inflammation The concept of asthma as a progressive disease has been supported by recent findings of airway remodeling in biopsy samples from both adults and children with asthma (1, 2). In addition to pathologic evidence from the biopsy specimens, an accelerated loss of lung function in adults with asthma compared with those without asthma has been described in many cohort studies (3–6). A progressive reduction in lung function in children with mild to moderate asthma has been characterized to only a limited degree in cross-sectional and longitudinal studies (7, 8). Zeiger and colleagues (8), using the baseline data from the National Heart, Lung, and Blood Institute Childhood Asthma Management Program (CAMP), reported a change in prebronchodilator FEV1 of almost 1% per year of asthma duration in children with mild to moderate asthma. Similar findings were reported by Agertoft and Pedersen in a cohort of children with asthma who did not receive inhaled budesonide over a 1- to 7-year period. In this group of children, a decline in FEV1% predicted of 1.2 to 3.1% per year was noted (7). There are challenges involved around the issue of asthma progression in children. One is the lack of recent longitudinal data that describe normal lung growth in children, and hence, there is an inherent difficulty of tracking delayed growth in the context of the normal lung development. Second, the effects of standard interventions on the natural history of asthma are still unclear. CAMP, the largest controlled asthma study in children, was designed to evaluate in children with mild to moderate asthma the effects of three inhaled treatments: budesonide, nedocromil, and placebo (9). Although the longitudinal data from the CAMP cohort showed no significant reduction in either postbronchodilator or prebronchodilator FEV1% predicted from baseline to 4–6 years of treatment in all treatment groups, hidden in those means were patients who had a reduction in lung function over time. In this analysis, we examine asthma progression by evaluating reduction in postbronchodilator FEV1% predicted over time. We compare the characteristics of patients who had a significant reduction in postbronchodilator FEV1% predicted over 4 to 6 years of treatment to those of patients who did not. At the Denver CAMP site, an ancillary study that evaluated the clinical utility of inflammatory markers, particularly circulating eosinophil counts, serum eosinophil cationic protein, exhaled nitric oxide, and induced sputum analysis, was performed during the closeout visits. The relationship of these markers to this phenomenon of reduction in lung function was also analyzed. The results of our study have not been reported in an abstract format.
The study population comes from the main CAMP study. Subject enrollment began in December 1993 and ended in September 1995 (10). Children with mild to moderate asthma, 5 to 12 years of age at screening, were randomized to inhaled budesonide (200 µg twice daily), nedocromil (8 mg twice daily), or placebo. All patients received as needed albuterol treatment. After a mean (± SD) of 4.3 ± 0.7 years of treatment, the subjects began a 2- to 4-month washout period beginning in March 1999. All subjects completed washout by October 1999. The CAMP study was approved by the Institutional Review Board of all eight study sites, including the National Jewish Medical and Research Center, and the CAMP Steering Committee. Informed consent and assent were obtained from all participants and their guardians. Participant data relating to demographics, asthma history and severity, atopy, and healthcare use were obtained from the CAMP dataset. In the course of the CAMP study, spirometric measures were obtained three times per year, and methacholine challenge was performed once per year. Spirometry was performed with a Collins-Stead Wells dry-seal Survey III device interfaced to a computer, according to American Thoracic Society standards (10), and values were expressed as percentage predicted of normal values with adjustments made for age, sex, height, and ethnicity (11, 12). Short- and long-acting bronchodilators were withheld for 4 and 24 hours, respectively. Prebronchodilator and postbronchodilator (using two inhalations of albuterol, 90 µg/inhalation, by metered dose inhaler) FEV1 and FVC were obtained. Bronchodilator reversibility was calculated as follows: (postbronchodilator FEV1 – prebronchodilator FEV1) x 100/prebronchodilator FEV1). Methacholine challenges were performed by administering doubling concentrations of methacholine using the Wright nebulizer-tidal breathing technique (10). The provocative concentration that induced a 20% fall from baseline FEV1 (PC20) was obtained by linear interpolation of the logarithmic dose–response curve. The change in postbronchodilator FEV1% predicted based on serial measures per individual patient from randomization through the treatment phase of CAMP was determined (i.e., each subject had an individual regression FEV1% predicted on observation time). Patients who completed at least 4 years in CAMP were included in the analysis. These patients had to have at least four data points, one of which was during or after the fourth year of treatment. Of the 1,041 participants in CAMP, 990 met these requirements. A slope definition of –1% or less per year based on available literature (7, 8), at a level of significance of p < 0.1 (as epidemiologic literature supports that a p value of up to 0.35 may be appropriate for an exploratory attempt) (13), with an R2 of at least 0.14 to ensure strength in our observations, was used to define a significant reduction. (The lowest R2 in this group was actually 0.25.) Subjects who had a significant reduction in postbronchodilator FEV1% predicted per year based on the previously mentioned criteria were termed SRPs; otherwise, they were grouped as NSRPs. Figure E1 in the online supplement depicts the regression slope of a child in the CAMP study who had a progressive reduction in postbronchodilator FEV1% predicted.
Markers of Inflammation during the Closeout Visits
Statistical Analysis and Considerations A mixed-model analysis using SAS (SAS Institute, Cary, NC) Proc Mixed was used to evaluate the effect of the following variables: age, duration of asthma, baseline value of the outcome measure, clinic site, sex, treatment assignment, and atopy on the development of reduced lung function in the SRP group alone. Mixed models, also called random effects or random coefficient models, are used for data that are hierarchic or clustered in nature. The main feature of these data is that the observations are not independent from one another and that has to be taken into account in the model. Repeated measures such as those in the CAMP study are an important source of such data and require the use of mixed models or other strategies that account for this lack of independence (16). We have used mixed models because they are the most flexible means of analyzing such data. The model contains a random intercept and the following fixed variables: baseline value of the outcome measure, age at randomization, race, sex, clinic, duration of asthma at baseline, severity of asthma at baseline, skin test reactivity at baseline, and the number of visits in the study. Inclusion of these variables was made to be consistent with the CAMP analysis of the primary outcome (9). Statistical analyses were performed using JMP versions 5.0 and SAS software (SAS Institute). When groups are defined in datasets, as we have done here, without a physiologic or clinical referent, some care needs to be taken that the defined group is not merely a statistical artifact. Because after defining our SRP group we found that 69% of these subjects were above the median value of FEV1% predicted at baseline and possibly subject to regression to the mean, we created four random and independent serial measurements of 990 observations each from the baseline distribution of FEV1%. When we applied our SRP definitions to this simulated dataset, only 5% of the simulated subjects would have been classified as SRPs. This is significantly less than the approximately 20% that we found in the real dataset. As an additional test on regression to the mean, we evaluated a linear regression analysis of a subsample of the NSRPs who met each of the following characteristics: younger than 9 years, having had asthma duration less than 5 years before the study, and having a 100 or more postbronchodilator FEV1% predicted at randomization. The slope in each subgroup was different from the SRP group (data not shown). Second, by looking at the pattern of reduction in postbronchodilator FEV1% predicted in the SRP group, on average, the SRP's postbronchodilator FEV1% predicted measurements actually increased from randomization until 2–4 months into the treatment phase before declining at more than 1.0% per year with a fairly good sample size at each visit (Figure 1) . Finally, using a different outcome measure such as the FEV1/FVC ratio in comparing SRPs and NSRPs, we found that although there was a drop in the FEV1/FVC ratio among the SRPs from randomization to the end of treatment to below normal values; no change in FEV1/FVC ratio was seen among the NSRPs (Tables 1 and 2) . The SRPs had a significantly greater decline than the NSRP group (p < 0.0001). Therefore, the phenomenon of decline can be seen not only using the FEV1% predicted but also using FEV1/FVC.
The baseline characteristics of the CAMP cohort have been previously reported (9). There were 51 patients who were not included in this analysis (n = 51) because of an observation period less than 4 years. The histogram of changes in postbronchodilator FEV1% predicted per year (i.e., slope) for the cohort is shown in the online supplement (Figure E2). The mean and median changes in postbronchodilator FEV1% predicted per year for the whole group were –0.20 and –0.26. There were 253 of 990 (25.6%) children evaluated who demonstrated a 1% or more reduction in postbronchodilator FEV1% predicted per year, based on the set definition. Using the mixed model with a random slope variable and no adjustment for other covariates, the slopes between SRPs and NSRPs are significantly different (–0.24 vs. 0.05 postbronchodilator FEV1% predicted per month, respectively, p < 0.0001) (Figure 1). The percentage of postbronchodilator SRPs was significantly different between study sites (p = 0.02) (Figure E3). There was no significant difference in the percentages of SRPs among treatment groups (i.e., 26, 26, and 24% in the budesonide, nedocromil, and placebo treated groups, respectively). Among the SRPs, there was no significant difference in the slopes between the three different treatments. At randomization, SRPs were younger and of lower height and weight, had an earlier asthma diagnosis, had a shorter duration of asthma, were predominantly male, had a lower percentage of any positive prick skin test, and had lower mean serum IgE compared with the NSRPs (Table 1). SRPs did not have evidence for more severe disease at randomization compared with NSRPs: (1) equal proportions of mild and moderate asthma that were seen; (2) higher prebronchodilator and postbronchodilator FEV1% predicted, FVC % predicted, and FEV1/FVC; and (3) lower percentage of hospitalization before treatment. At randomization, SRPs and NSRPs had similar bronchodilator reversibility, methacholine reactivity, and circulating eosinophil counts. There was a decrease in the percentages of SRPs in each age range with increase in age at baseline (Figure 2) .
A logistic regression procedure also identifies age (p = 0.0005), clinic (p = 0.02), sex (p < 0.0001), and postbronchodilator FEV1% predicted (p = 0.02) as the variables that significantly discriminate between SRPs and NSRPs (Table 2). In other words, being of younger age, male, and having a higher lung function predict the SRP group. Among the SRPs, the influence of age, duration of asthma, sex, ethnicity, treatment assignment, clinic site, atopy, initial lung function measurement, and severity of asthma at baseline on the reduction in postbronchodilator FEV1% predicted was evaluated in the mixed-model analysis. Examined individually, initial lung function, age, duration of asthma, sex, and clinic were significant. Using a step-down procedure, only lung function (prebronchodilator and postbronchodilator FEV1% predicted) at randomization had a significant effect on the pattern of reduction in postbronchodilator FEV1% predicted across the visits (p < 0.0001). In the SRP group, the effect of baseline lung function was such that the higher the lung function, the less steep the reduction in postbronchodilator FEV1% predicted. At the end of treatment, lower spirometric measurements were found among SRPs compared with NSRPs, whereas similar methacholine reactivity, circulating eosinophil counts, and serum IgE levels at the fourth year of treatment were found between SRPs and NSRPs (Table 3) . SRPs had fewer hospitalizations yet a similar number of emergency care visits, days on inhaled corticosteroid, total inhaled corticosteroid dose, prednisone courses, and prednisone days from randomization to the end of the treatment period compared with NSRPs.
A regression slope or change in prebronchodilator FEV1% predicted per individual patient was also determined. A slope definition of –1% or less per year at a level of significance of p of less than 0.1 with an R2 of at least 0.14 was also used to determine the prevalence of reduction in prebronchodilator FEV1% predicted. There were 189 (19.0%) of children in this cohort who had a reduction in prebronchodilator FEV1% predicted from randomization through the treatment phase. In addition, 62% of postbronchodilator SRPs also had a reduction in percentage predicted prebronchodilator FEV1 and 82.5% of those who had a loss in percentage predicted prebronchodilator FEV1 were also postbronchodilator SRPs. A  statistic = 0.62 (SE, 0.03; p < 0.0001) indicated that this agreement was greater than chance.
Markers of Inflammation
Based on a predefined set of criteria, a reduction in postbronchodilator FEV1% predicted was observed in 25.7% of children with mild to moderate asthma enrolled in the CAMP study. Factors associated with a reduction in postbronchodilator FEV1% predicted in this CAMP study population included younger age at entry, male sex, study site, and higher postbronchodilator FEV1% predicted at baseline. There was a higher proportion of SRP, the younger the participants. This suggests that the process starts early and that the phenomenon does not proceed uniformly over time. That reduction in lung function occurs early has been demonstrated in a large cohort from the Tucson Children's Respiratory Study where 1,200 newborns were enrolled to evaluate the natural history of wheezing in the first 6 years of life (17, 18). Children with persistent wheezing displayed a decline in lung function from infancy to 6 years compared with children who were "transient," "late onset," and "never" wheezers. The relationship between sex and asthma is somewhat complex. Our study found a higher proportion of males in the SRP group compared with the NSRP group. This finding is also supported by the logistic regression procedure when other covariates were included in the analysis. Being female is recognized to be a risk factor for persistence of asthma and bronchial hyperresponsiveness into adulthood (19, 20). From these epidemiologic data, adults with asthma are more likely to be female, in addition to the fact that there is greater incidence of asthma among females from puberty to adulthood. Male sex, on the other hand, is a known risk factor for childhood asthma (21), but more males undergo remission in later years, resulting in fewer males with asthma as adults (22). However, recent studies have shown that male sex is a risk factor for asthma progression from childhood to early adulthood (20, 23). In the study by Rasmussen and colleagues (23), males with asthma were twice more likely to develop airway remodeling (using postbronchodilator FEV1/FVC ratio as a surrogate) by 18 years compared with females with asthma (23 vs. 10.3%, p = 0.009). In their follow-up study just recently published by Sears and colleagues, males with persistent asthma had a lower FEV1/FVC ratio at age 26 years compared with females (20). Despite the use of two different measures of asthma progression, the CAMP and the New Zealand cohorts consistently demonstrated the relationship between male sex and risk for asthma progression. The reason for the effect of study clinic on reduction in lung function is unclear. Zeiger and colleagues found the clinical center in this eight-site study to be a significant predictor in the regression models of bronchial hyperresponsiveness, lung function, symptoms, and other measures of asthma severity scores at baseline (i.e., before randomization) (8). The significant differences among the different centers in the regression models in regard to these measures were attributed to diversities in their respective cohorts, their environments, and recruitment strategies. We found a higher lung function to be a predictor of reduction in lung function for the whole group. In the mixed-model analysis of the SRP group alone, only lung function (prebronchodilator and postbronchodilator FEV1% predicted) at randomization had a significant effect on the pattern of reduction in lung function across the visits. The effect of baseline lung function was such that the higher the lung function, the less steep the decline. Although the regression to the mean phenomenon could not be entirely excluded, it does not completely account for the observations as we have done some analysis to confirm the results using other variables such as FEV1/FVC ratio (please refer to the STATISTICAL ANALYSIS AND CONSIDERATIONS section). Alternatively, the use of prediction equations or reference values could partly explain the high proportion of children who displayed a reduction in lung function, especially among the younger children. Rosenfeld and colleagues compared three popular prediction equations used for interpretation of lung function (i.e., Knudson, Polgar, and Dockery) in patients with cystic fibrosis (24). They found significant variability in lung function among 6- to 11-year-old white children who were very short (i.e., < 116 cm for males and < 113 cm for females) or very tall (i.e., > 155 cm for males and > 148 cm for females) using the different prediction equations. Such variability in predicted FEV1 values plotted against height (at its extremes) was less pronounced in older cystic fibrosis patients studied. When we looked at the distribution of the CAMP white cohort aged 6–11 years, the subjects whose height measurements were at those extreme ends in the Rosenfeld study, hence, who may have been predisposed to the wide variations in the predicted lung function, comprised a minority. Among the 326 6- to 11-year-old white boys, only 16 were very short, and 8 were very tall. Among the 229 6- to 11-year old white girls, 5 were very short, and 26 were very tall. Indeed, a high percentage of SRPs was found among the very short young children (69% in boys and 80% in girls) and a low percentage of SRPs among the very tall children (12.5% in boys and 11.5% in girls). However, only 10% (55 of 555) white children aged 6 to 11 years old had a very short or very tall height that would place their predicted lung function within the more variable or dubious range. Hence, the effect of the variability at the extreme of the height distributions would have a minimal effect on our findings. There appears to be a group who is at risk to develop a reduction in lung function over time, possibly because of a more aggressive remodeling process. The CAMP Continuation Study and its extension will allow us to determine over an additional 8 years of observation whether patterns of reduced lung function persist in children as they get older. It is possible that we observed a critical rapid decline phase, which will eventually settle in a subgroup and will continue in a smaller group. In nonsmoking adults with severe symptomatic asthma despite therapy with high-dose inhaled corticosteroids and long-acting bronchodilators, in contrast to what we found, Brinke and colleagues reported that significantly older age, a longer duration of asthma, and a higher degree of bronchial reactivity to histamine were seen in those patients with persistent airflow limitation defined as postbronchodilator FEV1% predicted or FEV1/FVC of less than 75% predicted (25). However, after adjustment for age, sex, and asthma duration was made, persistent airflow limitation occurred almost nine times more often in severely asthmatic patients with sputum eosinophilia. The authors suggested that airway inflammation underlies the development of persistent airflow limitation. However, it remains to be seen whether eosinophilia is a marker or the factor causing fibrotic changes in the airway that lead to persistent airflow limitation. Perhaps monitoring of inflammatory markers in addition to lung function and symptoms would be helpful in differentiating whether this is secondary to a variable growth pattern or inflammation driven. Of interest, those who had a progressive reduction in lung function did not have clinical evidence for more severe disease, either at randomization or by the end of treatment. The SRPs even had fewer hospitalizations during the trial than the NSRPs. This is largely due to the relatively narrow range of severity and no significant lung function abnormalities at entry. In addition, the SRPs were younger, and their pulmonary function remained within acceptable limits. Also, this suggests that there is no absolute association between symptoms and lung function. Nevertheless, the link between the reduction in lung function in relationship to the development of severe asthma will be monitored in the CAMP Continuation Study and its extension phase. A child who continues to lose 1–2% per year lung function could have severe airflow obstruction by midadult life. Again, an analysis of this cohort in the CAMP Continuation Study and its extension phase will allow further evaluation of the phenomenon of reduction in lung function over an additional 8 years, giving a total of 12–14 years of observation. The concern in childhood asthma is that the disease adversely impacts the growth of a child's airways such that maximal lung growth is not achieved. Lower lung function in young adults with diagnosed or undiagnosed asthma compared with healthy control subjects is seen in various studies (3, 26). In addition, childhood FEV1% predicted predicts adult lung function level (27, 28). Hence, interventions to preserve or even improve lung function in childhood may have long-term implications. The recent National Asthma Education and Prevention Program Executive Summary report now recommends inhaled corticosteroids as the first-line therapy for all patients with persistent asthma based on numerous studies which have demonstrated that inhaled corticosteroids afford improvement in lung function and patients' quality of life, significant asthma symptom control, and reduction in morbidity and mortality (29). These are the parameters that clinicians and researchers alike routinely monitor to determine treatment response. Although the main CAMP study was not designed to evaluate the differences in the percentage of treatment responders nor was it originally intended to particularly assess the phenomenon of lung function reduction or disease progression, the definition used to assess lung function reduction over time in this study could be considered solid. This analysis indicates that a similar proportion of subjects on budesonide and placebo had a reduction in lung function over 4–6 years of treatment. Furthermore, among the SRPs, there was no difference in the rate of reduction in postbronchodilator FEV1% predicted between treatment groups. Similar total doses of inhaled and oral steroids during the treatment phase in CAMP were found between SRPs and NSRPs. Insights related to this observation were demonstrated in the first report of CAMP study outcomes (9). Pulmonary function FEV1 PC20 methacholine values were similar for all treatment groups 4 months after discontinuing active study medication. In another placebo-controlled, long-term study, Waalkens and colleagues evaluated the effectiveness and safety of budesonide in children with moderately severe asthma. In a subgroup of patients originally on budesonide who were rerandomized to placebo at the end of the study, much of the gains in lung function and bronchial hyperresponsiveness noted in the inhaled corticosteroid group were lost by the end of the 6-month follow-up (30). Whether introduction of controller therapy very early in the disease process or perhaps use of higher corticosteroid dose or combination therapy or enhanced medication delivery would prevent asthma progression is not known. The inhaled Steroid Treatment as Regular Therapy in early asthma (START) study is a very large, randomized, double-blind, 3-year study that evaluated whether early intervention using low-dose inhaled corticosteroid would prevent severe asthma-related events and accelerated decline in lung function in over 7,000 patients with less than 2 years of mild persistent asthma (31). Although their findings showed that patients on inhaled corticosteroid compared with placebo had significantly less risk of a severe asthma exacerbation, fewer courses of systemic corticosteroids, required less additional medications, and more symptom-free days, both budesonide and placebo treated showed a reduction in postbronchodilator FEV1% predicted (–1.79 vs. –2.68, respectively) over a 3-year period. Although the inhaled corticosteroid group incurred less decline in lung function compared with the placebo group (mean difference of 0.88, p = 0.0005), the differences in the 3-year change in postbronchodilator FEV1 between the budesonide and placebo-treated groups were narrower in the 5- to 10-year age group (–1.84 vs. –2.31, mean difference = 0.47) and even reversed in favor of placebo in the 11- to 17-year-old age group (–0.91 vs. –0.39, mean difference = –0.52). These data and our observations suggest that low-dose inhaled corticosteroid therapy has limited effects compared with placebo in preventing reduction of postbronchodilator FEV1% in children with mild, persistent asthma even at an early onset. One potential reason for the inability of inhaled corticosteroids to prevent reduction in lung function could be related to the presence of a corticosteroid-independent mechanism such as the presence of neurogenic inflammation (32) or structural changes in the airways such as subepithelial fibrosis and increased deposition of elastin (33). Although we can only speculate on potential reasons why this phenomenon is occurring, certainly adherence is one of the major reasons that we have considered. However, data initially from the CAMP main article and more recently, from a CAMP ancillary study that addresses medication adherence involving three CAMP sites (Denver, San Diego, and Baltimore) argue against nonadherence with study medications as the cause of the reduction in lung function. From the main CAMP article, compliance with treatment (defined as the percentage of days on which the child was reported to have taken the study drug) was similar in those randomized to budesonide compared with placebo (9). Adherence in this study was considered above average as compared with clinical care and thus as good an assessment of treatment effect as possible. From the CAMP adherence ancillary study (data on file), there was no difference in mean adherence values (based on residual "click" counts and canister weight change on return for the Turbuhaler and metered dose inhaler, respectively) measured for all three treatments between the SRP (70%) and NSRP (66%). These findings support the lack of significant association between adherence and reduction in lung function. In summary, although this phenomenon of reduction in lung function in children with mild to moderate asthma occurs in only a minority yet substantial number, it seems to start early in the course of the disease and in younger patients with high lung function at the start of these measurements. Along with tracking height and weight velocity, children with asthma warrant serial measurement of pulmonary function over time along with detailed documentation of healthcare use and escalation in medication requirements. The investigation of the pathophysiologic mechanisms (such as loss of elastic recoil and reduced closing volume), immunologic processes, and genetic markers of airway remodeling should be considered in identifying mechanisms and risk factors for altered lung growth.
The authors thank Dr. Bruce Bender and Lening Zhang for providing data on adherence and Ms. Jan Manzanares for assistance in the preparation of this article.
Supported in part by contracts from the National Heart, Lung, and Blood Institute (NO1-HR-16044, NO1-HR-16045, NO1-HR-16046, NO1-HR-16047, NO1-HR-16048, NO1-HR-16049, NO1-HR-16050, NO1-HR-16051, and NO1-HR-16052); General Clinical Research Center grants from the National Center for Research Resources (M01RR00051, M01RR0099718-24, M01RR02719-14, and RR00036); a National Institutes of Health grant (HL-36577); and Astra Zeneca Pharmaceuticals. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Conflict of Interest Statement: R.A.C. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; J.D.S. received over $1,000 in honoraria from Glaxo Smith Kline in 2002 and received grant support from Merck ($200,000) for an investigator-initiated study and $50,000 from Glaxo Smith Kline for participating in a phase III study; J.R.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this article; S.J.S. has been reimbursed by AstraZeneca for serving as a consultant ($6,000 from 2002–2003) and as a member of an Advisory Board ($3,000) and has received funding support to conduct a clinical study on marker of inflammation at the completion of the CAMP study ($80,000). Received in original form August 25, 2003; accepted in final form March 15, 2004
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
9 vs. < 9 years; duration of asthma: 7–12 vs. 0–3 years and 4–6 vs. 0–3 years; sex; study site; number of positive skin tests: greater than or equal to 5, 3 to 4, and 1 to 2 vs. none; and prior use of cromolyn or inhaled corticosteroid in the last 6 months) and continuous variables (e.g., log IgE, log circulating eosinophil count, log FEV1 PC20 methacholine, and baseline spirometric measurements). Regression analyses are reported as odds ratios with 95% confidence intervals for a 1 SD change in the continuous variable. 
