Inhaled steroids for episodic viral wheeze of childhood

Criteria for considering studies for this review

Search methods for identification of studies

RCTs were identified using the Cochrane Airways Group database. The database was searched using the following terms : asthma and wheez* AND budesonide, OR pulmicort, OR beclomethasone, OR becotide, OR fluticasone, OR flixotide, OR triamcinolone, OR steroid, OR corticosteroid, OR glucocorticoid AND inhaled, OR aerosolised, OR nebulised AND child, OR infan*, OR pediatric, OR paediatric, OR adolescen* .

The bibliography of each identified RCT was searched for relevant papers. Authors of relevant RCTs were contacted to identify other published and unpublished relevant studies.

Data collection and analysis

Two reviewers (MM and JG) independently selected trials for inclusion. In addition to the inclusion criteria, their assessment of each study took note of the degree of allocation and concealment (double blind, single blind, open) using the Cochrane approach and of overall methodological quality using the Jadad Score. Disagreement between the reviewers was resolved by consensus or by the input of a third reviewer.

Lung function data were entered as negative values to conform with the Cochrane convention whereby effects that favour the treatment under review move to the left.

Results are presented with 95% confidence intervals. Summary estimates for dichotomous outcomes were reported as relative risks, and for continuous outcomes, as weighted mean differences or standardised mean differences, as indicated. We used the fixed effect models and, in case of heterogeneity, the random effect models. Group differences between rates were also summarised as a ratio of rates.

Description of studies

The search was completed in April 1999 and identified 829 titles and abstracts which were screened to identify 43 potentially relevant studies of corticosteroid treatment for 'wheeze' or 'asthma'. Full text versions of these papers were obtained, and assessed independently by two reviewers who agreed that five trials met the entry criteria for this review. Reasons for excluding the other 37 papers included: treatment of post bronchiolitis symptoms (5 papers); treatment of classical, chronic asthma with persistent symptoms (12 papers); treatment of recurrent pre‐school wheezing without a clear description of presence/absence of chronic symptoms or containing a heterogeneous group of subjects (some with episodic wheeze + some with chronic symptoms or some with bronchiolitis) (13 papers); treatment of virus triggers of classical asthma (1 paper ‐ Svedmyr 1995, see discussion); studies of drug delivery that did not measure clinical efficacy ( 2 papers); a study of side effects without estimates of clinical efficacy (1 paper); a study of children with low birth weight (1 paper); a study of episodic viral wheeze comparing corticosteroid to cromoglycate (1 paper ‐ De Baets 1998, see discussion); a study of episodic viral wheeze using oral steroid, but not an RCT (1 paper ‐ Brunette 1988, see discussion).

This review is derived from five RCTs which pertain exclusively to children with mild viral induced wheeze, with no or minimal symptoms in between exacerbations.

There was a poor response to a letter sent to the authors of trials whose description of their subject group was unclear. Of thirteen letters sent, there were four replies, three confirming their subjects as chronic asthmatics and one suggesting a further name to write to which also confirmed their subjects as having chronic asthma.

INCLUDED STUDIES

The five included studies consisted of two studies looking at low dose maintenance inhaled corticosteroid (Wilson 1995, Duoll 1997) and three studies looking at episodic high dose inhaled steroid (Wilson 1990; Connett 1993; Svedmyr 1999).

Maintenance therapy. 
 Both Wilson 1995 and Duoll 1997 used a parallel design, studying children who had reportedly wheezed during URTIs on at least 2 and 5 previous occasions respectively. Both trials studied children with mild viral wheeze, not requiring oral corticosteroids They had minimal or no symptoms in‐between exacerbations. Duoll's patients were aged 7‐9 years with 50% and 42% being atopic by skin prick test in the corticosteroid and placebo groups, respectively. Some had a prior diagnosis of asthma but all had minimal or no chronic symptoms. Wilson's subjects were aged 0.7 ‐ 6 years and atopy was not assessed, but a family history of atopy was reported in 67% and 42% of those in the corticosteroid and placebo groups respectively. Again none had symptoms to suggest chronic asthma. Duoll's trial tested beclomethasone dipropionate 400 micrograms daily by diskhaler for 6 months and Wilson used budesonide 400 micrograms daily for 4 months, administered by metered dose inhaler (MDI) with a spacer (either a Nebuhaler (Astra) or aerochamber with facemask (Trudell Medical)). Both had run‐in periods of one month.

Episodic therapy. 
 Wilson 1990 and Connett 1993 both carried out cross‐over studies of episodic inhaled corticosteroids and hence could be directly compared. Svedmyr 1999 used a parallel design: the outcomes obtained from both designs were analysed separately, then combined in absence of heterogeneity.

Connett 1993 studied children aged 1 ‐ 5 years with a clear history of wheeze in the presence of an upper respiratory tract infection (URTI) with no interval symptoms; none were on prophylactic treatment. The severity of episodes was not documented, except for their need for bronchodilators. Atopy was not reported. Budesonide 1600 micrograms daily was given to those who could use a Nebuhaler while 3200 micrograms daily was given to those using a Nebuhaler with facemask. The observation period lasted for six months. The intervention was started at the onset of URTI symptoms until the children were asymptomatic for 24 hours. Wilson 1990 studied children aged 1 ‐ 5 years with a history of wheeze in the presence of an URTI, and at least 2 episodes in the previous 3 months. Again, severity was not documented other than the requirement for bronchodilators. However, this study did allow co‐intervention with prophylactic treatments (6 had sodium cromoglycate and 7 had inhaled corticosteroids) raising the possibility that some of the subjects had more severe illness than others. Atopy was not reported. Beclomethasone dipropionate 2,250 micrograms daily was given by MDI via a Volumatic (Glaxo‐Welcome) at the onset of signs of "asthma" for 5 days during a two year trial period. Although this paper describes the attacks as "acute asthma", it goes on to describe how all but one episode was precipitated by an URTI. The authors confirmed that most of the study group had no symptoms between episodes ('interval symptoms'), although four did go on to start either inhaled prophylactic cromoglycate or corticosteroid subsequently. These two cross‐over studies did not report the time lapse between treatment and placebo although correspondence with authors confirmed a minimum of a four week washout period.

Svedmyr 1999 studied children aged 1 ‐ 3 years with at least three reported episodes of wheezing during an URTI and no interval symptoms; parents reported wheezing or noisy breathing during most URTIs. The severity of prior episodes was not defined other than by the requirement for occasional bronchodilators. Atopy (defined by a total IgE level > 2 SD) was present in 46% and 41% in the corticosteroid and placebo groups, respectively. The corticosteroid group received 1600 micrograms of budesonide by MDI via a Nebuhaler and facemask for the first three days of an URTI and then 800 micrograms for the following 7 days. The intervention started at the first signs of an URTI and the observation period lasted for 12 months.

SEVERITY. The majority of children in the included studies seem to have predominantly mild episodic viral wheeze (i.e. treated only with bronchodilators). The patients studies by Duoll 1997, Svedmyr 1999 and Wilson 1995 had no or low (1‐2 per year) numbers of prior hospital admissions. The authors of the other two trials confirmed, by correspondence, that subjects were in the mild category. 
 
 ATOPY. Duoll's subjects (Duoll 1997) were older than those in the other studies and approximately 50% were atopic. Atopy is a little more difficult to define in younger children, but similar numbers of atopic children were found in Svedmyr 1999 and 50% of children in Wilson 1995 had a family history of atopy. Atopy was not reported in the earlier studies of Wilson 1990 and Connett 1993.

Difficulties arose in the way outcome measures were reported. Meta‐analysis is dependent on mean + SD for continuous data and whilst some studies reported mean (SD), others used the mean of differences between groups with the standard error of the differences, or median and range. Correspondence with authors could not resolve these issues, hence meta‐analysis of several outcome measures was not possible. None of the studies reported the validity of their symptom diaries, making interpretation of meta‐analysis difficult.

Risk of bias in included studies

All included studies stated that treatment allocation was randomised and the concealment was 'double‐blind'. The method used for generation of a random sequence was not described so the authors of four papers were contacted (Wilson 1990, Wilson 1995, Connett 1993, Duoll 1997). Each confirmed that the randomisation was carried out by either the drug company or pharmacy but only one was aware of the method used ‐ computer program Connett 1993 These four papers state that a 'double‐blind' method was used without indicating the details of how this blinding occurred. Correspondence confirmed that a drug company or pharmacy coded the inhalers with the code unknown to the subjects and the outcome assessors.

The two cross‐over studies of Wilson 1990 and Connett 1993 did not state the time lapse between treatment or did not comment on any possible carry‐over effect. It is essential to consider the possibility of a carry‐over effect. If the effective intervention is followed closely by placebo, the therapeutic effect could be carried over into the placebo period so minimising any possible difference between placebo and treatment. Correspondence with the authors of both trials confirmed a minimum washout period of four weeks, which should be adequate for inhaled corticosteroids.

All authors reported withdrawals and gave reasons for them. Wilson 1990 had 11 withdrawals, 6 did not have two treatment pairs in the cross‐over study and 5 for social reasons. Connett 1993 had 7 withdrawals, 3 did not suffer URTIs during the study, 2 were non‐compliant and 2 could not tolerate the Nebuhaler and facemask. Two of the parallel studies had similar number of withdrawals in each group, Wilson 1995 had 16 withdrawals (9 in the corticosteroid group, 7 in the placebo group) and Svedmyr 1999 had 12 withdrawals (6 in each group). In Duoll 1997 there were 10 withdrawals, 2 from the corticosteroid group and 8 from the placebo group. The main reason for these withdrawals was difficulty attending clinic and unrelated to the disease state.

Only one study measured compliance (Duoll 1997) by counting used blisters from the Diskhaler. There was no difference between treatment and placebo groups (being 74% and 75% respectively).

In general all trials were of high methodological quality, with grade A (double blind) allocation of concealment and with number of, and reasons for, withdrawals unlikely to affect outcomes.

Effects of interventions

Analysis of each outcome was broken down into three subgroups, based upon the available trials: 1. parallel trials (Wilson 1995 and Duoll 1997) of low dose maintenance inhaled corticosteroids; 2. cross‐over studies (Wilson 1990 and Connett 1993) of high dose episodic inhaled corticosteroids; 3. the parallel study (Svedmyr 1999) of high dose episodic inhaled corticosteroids. These three subgroups consider two treatment strategies: maintenance low dose inhaled corticosteroid and episodic high dose inhaled corticosteroid.

MAINTENANCE LOW DOSE INHALED CORTICOSTEROIDS

Episode severity.

1. Primary outcome: wheezing episodes requiring oral corticosteroids

There were very few episodes that required oral steroids. The two trials that examined maintenance low dose corticosteroid show the relative risk (RR) of needing a course of steroids lay close to the line of no effect [N=2 trials, RR=0.82 (95%CI: 0.23,2.90)].

2. Emergency department/doctor visits.

Only one of the two trials reported this outcome (Wilson 1995) in which the corticosteroid group had 4 subjects visit the emergency department compared to 3 in the control group. Analysis confirms no clear effect [N=1 trial, RR=1.4(95%CI: 0.36,5.49)].

3. Hospital admissions.

There were no admissions in Duoll 1997 and 2 children in the placebo group in Wilson 1995. The analysis for maintenance inhaled corticosteroid shows no clear effect [N=1 trial, RR=0.21 (95%CI: 0.01,4.11)]. The small numbers of children admitted and seeking emergency treatment underlines the low power of these trials to show any reduction in these rare outcomes.

4. Lower respiratory tract (LRT) symptom severity.

Due to heterogeneity of symptom diaries, meta‐analysis of this outcome was not possible. Duoll 1997 reported no significant difference in average severity of LRT symptoms between treatment and placebo groups [N=1 trial, WMD=‐0.50 (95%CI: ‐1.43,0.43)]. Wilson 1995 also showed no significant differences in the severity of day and night time cough and wheeze scores between those receiving maintenance low dose inhaled corticosteroid and placebo. These were presented as medians and are not entered in this analysis.

5. Bronchodilator use.

Bronchodilator use was not reported in Duoll 1997. Wilson 1995 found no differences in median puffs of bronchodilator for those receiving maintenance low dose corticosteroid compared to placebo.

6. Lung physiology.

Duoll 1997studied children aged 7‐9 years and examined daily PEF measurements, and monthly FEV1 and methacholine sensitivities (PD20). There was no difference between the placebo and beclomethasone groups in frequency, severity and duration of episodes of reduced PEF [N = 1 trial, WMD(95% CI) = 0.1 (‐0.6, 0.8), 12 (‐8,32), and ‐0.5 (‐3.4, 2.4), respectively]. FEV1 was consistently and significantly higher in the beclomethasone group [N = 1 trial, WMD (95% CI) = ‐0.10 (‐0.21, 0.01)] despite being similar to the placebo group before and after the trial. Similarly, the beclomethasone group had a higher PD20 during the treatment (data not entered as neither CI or SD given).

Wilson 1995 studied younger children, aged 6 months to 6 years, and lung function was not measured.

Indicators of episode frequency.

1. Frequency of wheezing episodes.

Reduction in frequency of wheezy episodes would be an important outcome for prophylactic inhaled corticosteroids. The duration of maintenance low dose corticosteroids was different in the two trials, Duoll 1997 being 6 months and Wilson 1995 3 months. Both papers report hardly any difference in the frequency of wheezing episodes between corticosteroid and placebo groups. The analysis only depicts the Duoll 1997 data [N=1 trial, WMD=‐0.40 (95%CI: ‐2.40,1.60)]. Wilson 1995 reported 95% CI of the mean differences instead of standard deviations and is not entered in this analysis.

2. Frequency of wheeze per URTI

No trials reported this outcome.

Indicators of episode duration.

1. Duration of episodes with LRT symptoms.

There was no benefit in terms of duration of symptoms (cough, wheeze) [N=1 trial, WMD=‐0.50 (95%CI: ‐1.82,2.82)] . Duoll 1997 reported a mean duration of LRT symptoms of 8.4 days in those given beclomethasone compared to 7.9 days in those given placebo. Wilson 1995 reported a mean duration of symptoms of 8 days in those given budesonide compared to 8.6 days in those given placebo. Wilson 1995 reported means and 95% CI of the mean differences and again standard deviations could not be derived, hence the analysis reflects only the Duoll 1997 data.

2. Duration of bronchodilator use.

This outcome was not reported.

In summary, only two outcomes could be pooled across trials. Compared to placebo, children receiving maintenance low dose of inhaled corticosteroids showed an improvement in FEV1 but did not experience any reduction in the number of wheezing episodes requiring oral corticosteroids or requiring hospital admission.

EPISODIC HIGH DOSE INHALED CORTICOSTEROID

Indicators of severity.

1. Number of wheezing episodes requiring oral corticosteroids (the primary outcome).

The meta‐analysis of the number of episodes requiring oral corticosteroids during the two cross‐over trials of episodic treatment with high dose inhaled corticosteroid shows a trend favouring treatment [N=2 trials, RR=0.53 (95%CI: 0.27,1.04)]. The effect was a 50% reduction in requirement for oral corticosteroids. This clinically important effect is approaching statistical significance and deserves further confirmation by additional trials. The parallel study of acute high dose corticosteroid, Svedmyr 1999 also reported a slightly lower number of episodes requiring oral corticosteroid in the inhaled corticosteroid group but this was also not significant [N=1 trial, RR=0.82 (95%CI: 0.52,1.29)].

2. Emergency department/doctor visits.

Only one study reported this outcome Svedmyr 1999 in which 10 subjects in the corticosteroid group visited the emergency department 16 times compared to 12 subjects visiting 23 times in the control group. The analysis represents visits rather than individuals and shows an effect favouring corticosteroids [N=1 trial, RR=‐0.70 (95%CI: 0.50,0.97)].

3. Hospital admissions.

For the cross‐over studies, Wilson 1990 reported 4 children admitted from each group and Connett 1993 reported 1 child admitted from each group. The parallel study of Svedmyr 1999 reported 5 children admitted in the corticosteroid group and 2 in the placebo group. The meta‐analyses for both study designs show no clear effect [N=2 trials, RR=1.00 (95%CI: 0.31,3.24) and N=1 trial, RR=3.00 (95%CI: 0.67,13.51) respectively]. No heterogeneity between trial design was observed.

4. Symptom scores.

Due to heterogeneity of symptom diaries, a meta‐analysis of this outcome was not possible. Connett 1993 reported significantly lower daytime wheezing severity in the group treated with budesonide versus placebo. Wilson 1990 did not report LRT symptoms separately but the total night and daytime symptom scores were significantly lower in the beclomethasone group. Both these studies reported these data as means of differences between treatment and placebo periods with confidence intervals of the differences. Standard deviations could not be derived, hence these data are not entered in the analysis.

Svedmyr 1999 reported significantly lower LRT symptom scores in the group given budesonide [N=1 trial, WMD=‐0.17 (95%CI: ‐0.34,‐0.003)]. Svedmyr 1999 also reported significantly lower scores of cough and noisy breathing during in those receiving budesonide. That study also reported significantly less sleep disturbance (i.e. episodes of awakening due to symptoms) in the corticosteroid group [N=1 trial, WMD=‐0.17 (95%CI: ‐0.20,‐0.14)].

Although a meta‐analysis was not possible, all three studies using episodic high dose steroid showed significant reductions in symptom scores.

5. Bronchodilator use.

The cross‐over study of Connett 1993 showed no difference in doses per week (budesonide 20.2, placebo 22.2, p>0.05) and, in the parallel study of Svedmyr 1999 there was no significant difference in puffs during episodes. Again, these studies reported these data as means of differences between treatment and placebo periods with confidence intervals of the differences, hence these data are not entered in the analysis.

6. Lung physiology.

The three studies of high dose episodic inhaled corticosteroid involved children less that 5 years old and lung physiology was not performed.

Indicators of episode frequency.

1. Frequency of wheezing episodes.

This outcome was not reported.

Indicators of episode duration.

1. Duration of episodes with LRT symptoms.

Both cross‐over trials looking at the acute treatment of episodes reported means of the differences with confidence intervals of the differences between the two treatment periods. Standard deviations could not be derived and these data are not entered in the analysis. Wilson 1990 demonstrated a shorter duration of LRT symptoms with beclomethasone compared to placebo (mean duration = 6.11 and 8.38 days respectively). Connett 1993 also found a similar reduction with budesonide compared to placebo (mean duration 4.12 and 5.47 days respectively). Neither of these differences were significant, but as both show means favouring corticosteroids it could be that the studies were not powered adequately to find significance. Svedmyr 1999 did not report this outcome.

2. Duration of bronchodilator use.

Connett 1993 showed no significant difference in duration of days that bronchodilator was required when episodic inhaled corticosteroids were given compared to placebo (mean days ‐ beclomethasone = 7, placebo = 8, p<0.5). None of the other studies reported this outcome.

Parental preference for treatment.

This outcome was reported only in the two cross‐over trials of episodic high dose inhaled corticosteroids. The question was asked after treatment pairs were completed. The order in which treatments were given was randomised in both studies so that there should have been no bias in favour of either corticosteroid or placebo. The meta‐analysis shows a clear effect favouring inhaled corticosteroid [N=2 trials, RR =0.64 (95%CI: 0.48,0.87)].

In summary, there are some benefits from the use of episodic high doses of inhaled corticosteroids. There is evidence to suggest a reduction in the number of episodes requiring oral corticosteroids, which was the primary outcome of this review, although this is not statistically significant. In these doubled blinded studies, there was a clear parental preference for episodic inhaled corticosteroids as compared to placebo. There was no difference in admissions to hospital possibly due to insufficient power of the studies to demonstrate a group difference in children with mild exacerbations.

Characteristics of included studies [ordered by study ID]

Characteristics of excluded studies [ordered by study ID]

Internal sources

• NHS Research and Development, UK.

External sources

• No sources of support supplied