Objective. Results from previous investigations that examined the psychological side effects of theophylline have been inconsistent, and none have reported about inhaled corticosteroids. The objective of this study was to assess the relative psychological side effects of theophylline and beclomethasone in asthmatic children.
Methods. This was a multicenter, randomized, double-blind, parallel-groups study in which 102 asthmatic patients were assigned to one of two treatments: beclomethasone three times daily or theophylline twice daily. At baseline, 1 month, and 1 year, parents completed standardized behavioral questionnaires while the children received psychometric testing of attention and concentration, memory and learning, and problem-solving.
Results. Although power was sufficient to detect meaningful mean score changes, no consistent differential treatment effects were observed. Two significant treatment-by-period interactions were discordant, with one suggesting slightly better attention in the theophylline group, whereas the other indicated a small advantage in attention scores in the beclomethasone group. Numerous significant period effects revealed that behavior and cognitive test performance improved over the 1-year period, regardless of treatment, and confirmed a well established practice effect resulting from repeated administrations of such tests.
Conclusions. Neither theophylline nor beclomethasone should be avoided out of concern for significant psychological side effects. The possibility remains that a subset of asthmatic children may be susceptible to such medication-induced changes; investigators have suggested that preschool children may be at particular risk, although no controlled studies with this age group have been conducted. Parental perceptions of medication side effects can be influenced by temporary effects present at initiation of treatment or by erroneous attribution of the psychological effects of the chronic illness. Reports of psychological changes in response to asthma medications must be addressed respectfully but objectively, with due consideration of available evidence and an awareness of other potential explanations.
- WCST =
- Wisconsin Card Sorting Test •
- CBCL =
- Child Behavior Checklist •
- WISC–R =
- Wechsler Intelligence Scale for Children–Revised •
- W-J =
- Woodcock-Johnson Psychoeducational Battery
The choice of medications used to treat pediatric asthma is routinely based on a number of considerations, including the child's asthma history and symptoms and each medication's relative benefits and risks. The side-effect potential of any medication is a necessary consideration that can contribute to the choice of an alternative medication. Theophylline, for several decades the most frequently selected asthma treatment, has been widely replaced as the first-choice chronic asthma medication in favor of aerosolized antiinflammatory drugs such as cromolyn, nedocromil, and corticosteroids. Several factors, including its potential for medical and psychological side effects, have led physicians away from theophylline treatment of pediatric asthma. The list of theophylline's reported psychological side effects includes impaired attention,1 impulse control,2 motor steadiness,3 visual–spatial planning,4 memory,2 5 and school performance.6 Half of the parents surveyed stated that their asthmatic children became restless and overactive when treated with theophylline.7 In contrast, inhaled corticosteroids have been reported to cause relatively fewer side effects.8 9 Case reports of hyperactive, aggressive, and oppositional behavior after inhaled corticosteroid treatment have been recorded, but controlled studies of psychological side effects resulting from this group of medications are lacking.10-13
This report includes a component of a multicenter study of asthma treatment conducted by the American Academy of Allergy and Immunology. In that study, the relative effectiveness of beclomethasone dipropionate was compared with that of theophylline in improving asthma symptoms, sparing supplemental bronchodilator and systemic steroid use, preventing absence from school or work, and avoiding side effects such as headaches, insomnia, gastrointestinal distress, and impeded growth velocity.14 A battery of psychological tests also was administered to participants in each of the two blinded treatment groups at baseline, 1 month, and 1 year after treatment onset. These results have not been described previously but are presented here to provide the first data examining the long-term (1-year) psychological effects in asthmatic children treated with theophylline or inhaled corticosteroid.
The 6- to 17-year-old asthmatic children (N = 102) included in this investigation all demonstrated cough, dyspnea, and wheezing requiring intermittent or constant bronchodilator treatment; had a forced expiratory volume in 1 second (FEV1) >50% of predicted; and demonstrated at least a 15% reversibility of airflow obstruction after inhaled bronchodilator inhalation. Additionally, all participants did not have the following: acute respiratory infection within 3 weeks; systemic glucocorticoid treatment within the previous month or for >30 days in the previous 2 years; theophylline and aerosol steroid treatment together for >1 month in the previous year; regular aerosol cromolyn treatment within the past 60 days or topical nasal glucocorticoids in the past 30 days; adverse reaction to theophylline or glucocorticoids; serious emotional, behavioral, or cognitive impairment; or a history of any other illness that would contraindicate the use of study drugs.
Patients were screened and observed for a 1-month period during which those previously receiving theophylline or aerosol steroids discontinued the medication. Patients whose illness was of sufficient severity to produce symptoms on a majority of days but whose symptoms were maintained adequately with only bronchodilator use as needed during the observation phase were then randomized into the study. Patients whose symptoms were inadequately controlled during the observation period were not randomized.
Patients were randomized to one of two treatment conditions. Those in the beclomethasone dipropionate condition were instructed to inhale two puffs (42 μg per activation) from a metered-dose inhaler four times a day and to take two placebo tablets (that matched the active theophylline tablets) twice a day. Patients in the theophylline condition followed the same dosing instructions for the two medications while actually receiving placebo aerosolized medication and a daily dosage of theophylline adjusted primarily to maintain optimal symptom control and secondarily to achieve a target theophylline blood level of 8 to 15 μg/mL. Patients were followed for 1 year after randomization in a protocol that included, at scheduled intervals, spirometry, methacholine challenge tests, stadiometry, cortisol stimulation tests, ophthalmologic examinations, and physical examinations. In addition, all children used peak flow meters to monitor their pulmonary functions at home.
Psychometric testing was conducted at baseline, 1 month, and 1 year. This design afforded the opportunity to evaluate both short- and long-term behavioral/psychological changes related to study medications. Psychometric testing was conducted only at the study centers able to provide sufficient space and personnel. Thus, of the 195 pediatric subjects randomized to the treatment groups, 102 actually completed the psychometric protocol.
A battery of psychometric measures was chosen to provide an efficient evaluation of psychological functions shown or hypothesized to be influenced by theophylline. These included measures of attention and concentration (Stroop Color and Word Test, Wisconsin Card Sorting Test [WCST], Target Test), memory and learning (Benton Revised Visual Retention Test, Rey Auditory Verbal Learning Test, Serial Digit Learning Test, WCST), problem solving (WCST), and problem behavior (Child Behavior Checklist [CBCL]). To shorten the testing session length, the Rey Sum subscale was modified from the original to be the sum of three rather than five trials. Two additional tests (Wechsler Intelligence Scale for Children–Revised (WISC–R) and Woodcock-Johnson (W-J) Psychoeducational Battery were used, not as measures of specific cognitive functions, but as broader assessments of cognitive growth and achievement that help determine the larger impact of particular medication-mediated cognitive changes. In both cases, editions of these tests current at the time of the study's initiation were used. Because of potential practice effects, these two tests and the WCST were administered only at the baseline and 1-year visits. At the time of testing, parents of each child completed the CBCL, a questionnaire examining problem behavior and symptoms of anxiety, depression, attention, hyperactivity, sexuality, and social adaptation. Because evaluations were completed throughout the year, including during periods when children were not in school, parental but not teacher observations were obtained.
The experimental design is a randomized, double-blind, parallel-groups, type, with repeated measures on selected psychometric and behavioral scales at baseline, 1 month, and 1 year after randomization to beclomethasone or theophylline. To assess the differential effect of treatment on changes over time, the outcomes each were tested separately using repeated-measures analysis of variance (ANOVA), with treatment as the between-subject factor and period (time) as the within-subject factor. In this model, the test of significance of the interaction between treatment and period is a direct test of the differential effect attributable to treatment over time. The main effects attributable to treatment and period also were evaluated, but they were of less interest in the absence of a significant interaction, because it was expected that both treatments would enhance performance on some scales over time. A secondary objective was to determine whether selected baseline subject characteristics could be used to explain any differences found in changes in outcomes between treatments. This hypothesis was tested separately for each outcome by adding these characteristics as covariates into the repeated-measures ANOVA model and assessing their significance. Power to detect clinically relevant differences with the given sample sizes was calculated for each effect in the ANOVA model using standard methods based on the noncentral Fdistribution.
Baseline characteristics and number of subjects responding were compared between the groups using t tests or χ2 tests, as appropriate. Data in the study groups were summarized using means and SE values for continuous variables and percentages for categoric variables. PC20 was analyzed after taking a common logarithmic transformation and is summarized using the geometric mean and ∼95% confidence intervals. All analyses were performed using two-tailed tests of significance at the 0.05 level, and test results are reported using individual Pvalues.
The comparison of baseline subject characteristics between the two randomized treatment arms is summarized in Table1. None of the characteristics were significantly different between the groups at the 0.10 level. There were 52 subjects in the beclomethasone arm and 50 subjects in the theophylline arm. They averaged 11 to 12 years of age and generally had moderately severe asthma for 7 to 8 years.
The longitudinal analysis of CBCL scores is summarized in Table2. There were no significant differences between the two treatment groups for any CBCL subscale or summary scores at the 0.10 level when averaged over the three repeated measures. Significant differences were found between the three time periods (baseline, 1 month follow-up, and 1 year follow-up) when averaged across treatment groups for the social problems subscale (P = .0035) and the total internalizing (P = .0023) and total problems summary scales (P = .025), with all three indicating an improvement in behavior from baseline to 1 month. There was a significant treatment-by-period interaction for the attention problems subscale (P = .035). However, the magnitudes of these statistically significant differences were well within the range considered to be of no clinical importance. All effects for the remaining subscales and summary scores were not statistically significant. These findings do not reflect a systematic longitudinal differential in behavior between subjects treated with beclomethasone and those treated with theophylline.
Alternatively, drugs can affect the behavior of a subset of subjects in ways that are not reflected in comparisons of the group averages. Therefore, the numbers of subjects increasing by 10 or more points on the CBCL summary scales from baseline to 1 month were compared between the two treatment groups. With either 0 or 2 subjects in each group for each summary scale that changed that much in 1 month, the differences between groups were not significant.
The longitudinal analysis of cognitive scale scores is summarized in Table 3. The Rey, Serial Digits, Stroop Color and Word, and Target Letter and Number instruments were given on the baseline, 1 month, and 1 year visits. The WCST, W-J, and WISC–R tests were administered only on the baseline and 1-year follow-up visits. There were no significant differences between the treatment groups when averaged over periods at the 0.10 level for all scales, except for the Rey Sum (P = .032), for which the theophylline group was higher. There were significant differences between the time periods when averaged across groups for Rey Delay (P = .0001), Serial Digits (P = .0087), Stroop Color and Word (P = .0001), Stroop Color (P = .011), Target Letter and Number (P = .0001), WCST Errors (P = .0001), W-J Math, Reading, and Writing (P = .0001), and WISC–R Full Scale IQ (P = .001) and Performance IQ (P = .0001). In most cases, the subjects improved over the 1-year study period, and all changes were of little clinical significance. With respect to the interaction between treatment group and time period, only the interaction for Rey Sum was significant at the 0.05 level (P = .022). Although there are some changes over time regardless of treatment, these findings do not reflect a consistent longitudinal differential in cognitive functioning between subjects treated with beclomethasone compared with those treated with theophylline.
For both the behavioral and the cognitive outcomes, the baseline subject characteristics summarized in Table 1 were included as covariates in additional analyses to determine whether any differences in outcomes found previously could be explained by differences in the covariates. These included standard demographic characteristics, physical and illness characteristics, and previous use of theophylline. In every case, the covariate effect was not significant at the 0.10 level, and conclusions from these analyses were the same as those from the unadjusted analyses with respect to the tests for main effects and interactions.
Because the results of this study are largely negative with respect to the primary hypothesis of differential treatment effects, it is important to know whether the experimental design and sample sizes were adequate to detect clinically significant differences in the primary outcome variables. In Table 4, the results of such a retrospective power analysis are summarized for the CBCL, Rey, and Serial Digits instruments using the observed variances and sample sizes, two-tailed tests, and a 0.05 α level. This analysis indicates that there is ≥ 98% power to detect clinically significant differences for the period effect and for the treatment-by-period interaction, but that the power to detect treatment effects drops to 86% for Serial Digits, to 88% to 100% for Rey Sum and Delay, and 72% to 76% for the CBCL summary scales. These results indicate adequate power in this study as implemented to detect clinically significant main effects and interactions if, in fact, they were present.
This study represents the most comprehensive investigation conducted to date with a pediatric sample of sufficient size to detect psychological side effects of asthma medications. The preponderance of evidence, which includes scores from a large battery of standardized psychological measures and parental observations, indicates that neither inhaled beclomethasone nor theophylline causes major changes in behavior or cognition. The two significant treatment-by-period interactions appear isolated and discordant, with one suggesting slightly better attention in the theophylline group, whereas the other indicates a small advantage in attention scores in the beclomethasone group. The numerous significant period effects consistently reveal improvement in behavior or cognitive performance over the 1-year period, regardless of treatment. This finding confirms a well established practice effect resulting from repeated administrations of cognitive tests and the tendency for parents to rate their children's behavior as less problematic over time.15 Whatever its source, score improvement over the 1-year period is incompatible with any detrimental effect from theophylline or beclomethasone. Neither medication was found to interfere with attention, memory, discrimination, problem-solving, academic achievement, intelligence level, or emotional or behavioral adjustment.
Other studies of psychological change induced by theophylline have produced inconsistent findings; on balance, they are primarily negative. Theophylline has been credited with producing a variety of behavioral, cognitive, and psychological problems in children.4-6 16 On close inspection, however, results from such studies are flawed by methodologic problems that include absence of blinding, small sample size resulting in inadequate statistical power, absence of standardized psychological measures, questionable statistical procedures, and overemphasis on isolated findings that were statistically but not clinically significant.17 18 The picture emerging from the most recent controlled studies reveals that theophylline produces no untoward changes in groups of asthmatic children and may even produce a small, beneficial change in the form of improved attention and memory.19-21 This conclusion was confirmed by a recent metaanalysis involving studies of theophylline and caffeine, another central-nervous-system-stimulating methylxanthine. This study found that contrary to popular belief, neither drug caused deleterious effects in children and may at times produce slightly better focused attention and behavior.18Although most of these studies were fairly small, the sample size in the present study was not, and statistical power was sufficient to detect group differences of meaningful proportion. Another study of relatively large sample size found that academic achievement in 255 Iowa school children with asthma, and in particular the subgroup receiving theophylline, was equivalent to those of other children as reflected by scores on standardized achievement tests.22This outcome established further that despite previous assertions to the contrary,6 learning and school progress were not impeded measurably in children taking theophylline for extended periods of time. It is unclear whether negative public media attention about theophylline before the initiation of this study may have created a reluctance among some parents to participate in theophylline studies and hence altered study results. Consistent findings between this and other studies suggest that this is not the case.
Oral corticosteroids can produce psychological changes that include psychosis in 3% of adults23 and mood and memory changes in children,15 particularly when doses exceed the equivalent of 40 mg/day of prednisone. Largely because of the assumption that minimal systemic absorption of these medications greatly reduces the possibility of such adverse outcomes, few studies have addressed the possibility of behavioral side effects of inhaled corticosteroids. However, more recent appreciation for the systemic side effects of inhaled corticosteroids9 and individual case reports of dramatic behavioral changes in asthmatic patients treated with inhaled corticosteroids10-13 have raised the possibility of more frequent and subtle psychological effects similar to those that can accompany the use of oral corticosteroids. No evidence of small or dramatic behavior changes associated with inhaled corticosteroids was seen in this study.
Isolated or Temporary Behavioral Changes
Three circumstances are identified under which psychological side effects of these medications could occur. First, a limited amount of evidence suggests that preschool children may be at greatest risk for behavioral change resulting from the use of theophylline24 or inhaled corticosteroids.12 13 Almost all controlled investigations of this question have included children ≥ 8 years of age. Because the present study included only children ≥ 6 years of age, the possibility of marked psychological effects in preschool children remains to be assessed in controlled, prospective studies.
Second, a subgroup of asthmatic children may be particularly susceptible to medication side effects, whereas most asthmatic children are not. Previous studies of theophylline side effects, however, have failed to find any relationship between medication-related change and age, sex, socioeconomic status, intelligence, or psychiatric history.21 Although these were relatively small studies, the appreciably larger present investigation similarly found no demographic, psychological, or disease-related correlate that might identify an at-risk subgroup in either treatment group. History of theophylline use, which characterized half of the children in this study, did not increase or decrease susceptibility to medication-induced changes.
Finally, medication-mediated behavioral changes may occur as temporary reactions to initiation of the treatment regimen. Stein and Lerner25 found theophylline–placebo differences on standardized psychological measures at treatment onset but not 2 weeks later, leading them to conclude that the children had habituated to the central nervous system stimulation much as individuals adjust to the effects of coffee. In contrast, another study found that behavioral changes identified 1 week into theophylline treatment did not return to baseline at 1 month, 3 months, or 6 months.15Unfortunately, the 1-month interval between treatment randomization and the first follow-up visit may have missed temporary behavioral changes that occurred in the interim.
Perceived benefits and liabilities accompanying any medication are subjective and susceptible to distortion. These perceptions are shaped by personal experiences, attitudes toward the illness and the medical profession, information provided by the physician, and information obtained through public media. Additionally, the temptation to credit to a medication any concomitant change in illness or psychological state frequently is irresistible. In the case of an ill child, behavioral changes resulting from the illness itself may be more easily ascribed to a medication used to treat the illness. In some instances, temporary changes in sleep, mood, or behavior evident on initiation of a new medication may be perceived by the parent well after tolerance is established. In a study of theophylline side effects, asthmatic children whose parents stated that theophylline made their child more restless and hyperactive received either active medication or placebo for 1 week, crossing over to the remaining condition for a second week. Laboratory testing revealed no untoward changes in activity level or cognition, and parents were unable to detect behavioral differences when blinded to treatment condition.21 In short, reports of psychological changes in response to asthma medications must be addressed respectfully but objectively, with due consideration of the evidence available and an awareness of other potential explanations. Careful discussion with parents and child may lead to a decision to try another medication. However, in some cases, particularly in which more serious changes in mood or behavior are noted, referral to a mental health professional may prove more effective.
- Received June 6, 1997.
- Accepted August 18, 1997.
Reprint requests to (B.G.B.) National Jewish Medical and Research Center, 1400 Jackson St, Denver, CO 80206.
- Rachalefsky G,
- Wo J,
- Adelson J,
- et al.
- ↵American Asthma Report. New York, NY: Research and Forcasts, Inc; 1989
- ↵National Asthma Education Program, Executive Summary: Guidelines for the Diagnosis and Management of Asthma. Washington, DC: US Department of Health and Human Services; June 1991
- Meysboom RH,
- de Graaf-Breederveld N
- Tinkelman DG,
- Reed CE,
- Nelson HS,
- Offord KP
- Furukawa CT,
- Shapiro GG,
- Bierman CW,
- Kraemer MJ,
- Ward DJ,
- Pierson WE
- Copyright © 1998 American Academy of Pediatrics