Article Text

Download PDFPDF
Safety of formoterol in children and adolescents: experience from asthma clinical trials
  1. J F Price1,
  2. F Radner2,
  3. W Lenney3,
  4. B Lindberg2,4
  1. 1King's College Hospital, London, UK
  2. 2AstraZeneca R&D, Lund, Sweden
  3. 3Academic Department of Child Health, University Hospital of North Staffordshire, Stoke on Trent, UK
  4. 4Department of Paediatrics, University Hospital, Malmö, Sweden
  1. Correspondence to J F Price, Variety Club Children's Hospital, King's College Hospital, Denmark Hill, London SE5 9RS, UK; john.price{at}


Background The safety of long-acting β2 agonist (LABA) therapy in asthma remains controversial but no large scale analyses have been published of LABA safety in children.

Methods The frequency of asthma-related deaths and hospitalisations following formoterol use in children (4–11 years) and adolescents (12–17 years), compared with non-LABA treatment, was assessed in all AstraZeneca-sponsored, randomised, controlled, parallel-group trials (≥3 months) where formoterol was used as maintenance and/or as reliever therapy.

Results 11 849 children and adolescents under the age of 18 years from 41 trials were identified, 82% of whom used an inhaled corticosteroid (ICS) as concomitant medication. The number of asthma-related deaths (one 13-year-old boy among 7796 formoterol-treated patients, and none among 4053 non-LABA-treated patients) was too low to allow any between-group comparison. The frequency of patients with asthma-related hospitalisations was not different in formoterol-treated versus non-LABA-treated patients, either in children (1.16% (38/3263) vs 1.11% (24/2165)) or in adolescents (0.51% (23/4533) vs 0.85% (16/1888)). Asthma-related hospitalisations based on daily dose of formoterol were: (A) 4.5 or 9 µg: 1.9% (18/980); (B) 18 µg: 0.5% (14/2870); (C) 36 µg: 0% (0/67); and (D) variable dosing: 0.75% (29/3879). There was no difference between formoterol-treated and non-LABA-treated patients as regards ethnicity.

Conclusions Formoterol use in children and adolescents (4–17 years) with asthma in this large study where the majority are prescribed concomitant ICS is not associated with any increased risk of asthma-related hospitalisations. The results are not influenced by dose or ethnicity.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Randomised controlled trials in adults and children with asthma have shown that long-acting β2 agonists (LABA) added to inhaled corticosteroid (ICS) therapy, improve symptoms and lung function.1 2 However, studies of treatment with LABA in adults with asthma without mandatory ICS as concomitant medication have indicated there may be an association between the use of LABA and an increased frequency of severe asthma-related adverse events, including deaths.3,,8

The concern raised in adults with asthma was enhanced by a meta-analysis of studies of 3 months' duration or longer in which either salmeterol or formoterol was compared with placebo.9 This concluded there was significantly increased respiratory-related mortality and morbidity associated with both these LABA. The meta-analysis was dominated by patients from a single study (the Salmeterol Multicenter Asthma Research Trial, SMART) which accounted for 78% of the patients included.10 Current clinical guidelines strongly recommend that LABA should be used in combination with ICS and not as monotherapy.11 Neither of the trials included in the meta-analysis nor the meta-analysis itself were designed to assess the potential for ICS to prevent the possible adverse effects of LABA. A subgroup analysis of the SMART study suggested that the increased mortality observed was confined to patients not prescribed ICS at baseline.12 It should also be noted that the meta-analysis excluded trials with no placebo group. Thus pivotal studies evaluating the benefit of adding LABA to ICS, for example, the Formoterol and Corticosteroids Establishing Therapy study,13 were not included and this raises the question whether the results were truly based on a systematic review of the literature.10 There has been an implication of dose dependency for the possible negative effects, since a higher dose of formoterol (Foradil 24 mg metered dose twice daily) tended to be associated with more serious asthma exacerbations than a lower dose (Foradil 12 mg twice daily).4

What is already known on this topic

  • Current clinical guidelines strongly recommend that long-acting β2 agonist (LABA) be used in combination with an inhaled corticosteroid.

  • A literature review found there may be an increased rate of asthma exacerbations in children treated with a LABA.

  • No statistically significant differences in asthma-related and cardiac-related deaths were found between LABA and non-LABA-treated patients.

What this study adds

  • Formoterol use in children and adolescents with asthma receiving a concomitant inhaled corticosteroid is not associated with increased risk of asthma-related mortality or hospitalisations, or increased risk of cardiac-related events.

  • The results are not influenced by formoterol dose or by ethnicity.

Data on the safety of formoterol derived from all AstraZeneca randomised controlled clinical trials have recently been published.14 All-cause mortality and asthma-related and cardiac mortality and morbidity were evaluated. The vast majority of patients were treated with ICS. There were few asthma-related or cardiac-related deaths among patients randomised to formoterol, and there were no statistically significant differences between LABA and non-LABA-treated patients. Asthma-related serious adverse events (SAEs) (>90% of which were hospitalisations) were significantly lower among formoterol-treated patients compared to non-LABA-treated patients.

A literature review in 2003 concluded there may be an increased rate of asthma exacerbations, particularly those leading to hospitalisation, in children treated with a LABA.15 One trial recorded more asthma-related hospitalisation in children with asthma taking a LABA,16 although in this study ICS medication was not mandatory. No single study has been carried out in children on the scale of the SMART study.3 AstraZeneca-sponsored trials with formoterol in children and adolescents (aged 4–11 years and 12–17 years, respectively) provide a total exposure of over 12 000 patients. The aim of this analysis was to utilise the large database derived from AstraZeneca-sponsored studies to assess the safety of formoterol in children and adolescents. The primary safety outcome measure analysed was hospitalisation due to asthma-related events. Deaths, intubations, cardiac-related events and the impact of dose, ethnicity and gender were also evaluated.


Data source

All AstraZeneca randomised parallel-group clinical trials completed by January 2007 in patients with asthma, involving the use of formoterol either alone or in combination with budesonide, were identified through the company clinical safety database. From this database, 11 849 children were identified who had participated in 41 clinical trials that were of 3–12 months' duration, contained at least five children aged 4–17, and did not involve the use of salmeterol-containing products (figure 1). Of these 41 trials, 30 have been published as peer review papers, six have been published as abstracts in international meetings and one is accessible on the AstraZeneca clinical trials website (, while the remaining four are held as data on file (see tables E1 and E2 in the online supplement).

Figure 1

Flow-chart showing paediatric patients in AstraZeneca clinical trials in asthma.

Twenty-four trials involving 9407 patients that directly compared by randomisation formoterol and non-LABA treated patients were subjected to further analysis.

Ethics approval had been obtained for all studies included at the time they were conducted and no additional approval was required for this analysis.

Outcome events

The primary end point was SAEs defined using the International Conference on Harmonisation recommendations (see appendix A in the online supplement) reported with the seriousness criterion hospitalisation in patients aged 4–17 years. Subanalyses were performed in 4–11-year old children and in 12–17-year old adolescents. The influence of daily dose of formoterol, gender and ethnicity on asthma-related hospitalisations was also examined.

Secondary outcomes examined were: all cause mortality, asthma-related intubations, all-cause SAEs, asthma-related discontinuations due to adverse events (DAEs), and cardiac-related SAEs and DAEs. Deaths, SAEs and DAEs were evaluated at the time of the studies by the original investigators and prior to unblinding in blinded trials. Asthma-related deaths, SAEs and DAEs were defined as any such events coded to the preferred terms asthma, status asthmaticus or bronchospasm with reference to the Medical Dictionary for Regulatory Activities (MedDRA) version 9.0 ( Any death coded to respiratory failure was considered to be asthma related. The criteria defining a cardiac-related SAE or DAE are listed in appendix B in the online supplement.

To display all relevant data on adverse events, the adverse event data (ie, combined non-serious adverse events and SAEs) for all reported MedDRA preferred terms in the dataset of 24 trials with formoterol versus non-LABA treatment were aggregated.

Data analysis

For the number of patients reporting asthma-related hospitalisations, SAEs and DAEs, 95% CIs for the estimated ORs for two binomial proportions were computed by the exact method as given by StatXact-8 (Cytel, Cambridge, Massachusetts, USA). For each patient, the person-time of participation in the trial was measured and cumulated to obtain person-years of exposure. The rates for all cause mortality were expressed per 1000 person-years computed for each treatment group. Confidence intervals (CIs) for the estimated common Poisson relative risk were computed by the exact method as given by StatXact-8. These risk ratios relate the number of deaths to the actual exposure time.

A stratified Mantel–Haenszel approach was applied to asthma-related hospitalisations and to asthma-related DAEs in the 24 trials where there was a direct comparison between formoterol and non-LABA treatment, utilising descriptive 95% credibility intervals for the RR using the method of Barker and Cadwell17 as described in appendix C in the online supplement. The reporting of asthma-related hospitalisations over time was also analysed in the 24 trial subset using a Cox regression model, stratified by study, and displayed using a Kaplan–Meier survival curve.

For RRs and ORs, differences were considered statistically significant when the 95% CI excluded 1.00.

Effects of dose, race and gender were presented descriptively, as were adverse event data.

Formoterol doses are expressed as delivered doses. Formoterol delivered doses of 9 and 18 μg correspond to metered doses of 12 and 24 μg, respectively.


Patient demographics and exposure

The data represent 3500 patient-years of exposure to formoterol and 1900 to non-LABA treatment. Of the 11 849 patients included, 7796 were randomised to formoterol and 4053 to non-LABA treatments. Just over 45% of the patients were children under 12 years of age. There were more males than females. One quarter of the subjects were non-Caucasian. Of the 7796 patients randomised to treatment with formoterol, 18% were not taking an inhaled steroid at trial entry (table 1). In four of the trials involving 4389 patients, the use of ICS was determined by severity of disease at baseline not by randomisation. In four of the trials involving 701 patients, the use of ICS was mandatory as concomitant medication by protocol. In 33 of the trials involving 7305 patients, the use of ICS or not was assigned by randomisation.

Table 1

Baseline demographics, overall exposure and outcome for the primary variable (patients with asthma-related hospitalisations)

Primary outcome

There was no difference in asthma-related hospital admission rates or relative risk of hospitalisation in children and adolescents between patients receiving treatment with formoterol (0.78%) and those receiving non-LABA treatments (0.99%) (RR 0.79, 95% CI 0.52 to 1.21) (table 1).

There was no increase in asthma-related hospitalisations with increasing doses of formoterol. Hospital admission rates were slightly lower in males than in females. There was no difference in hospital admission rates between formoterol-treated and non-LABA-treated patients relating to ethnicity. Children and adolescents of black and oriental origin tended to have higher hospital admission rates than those of Caucasian origin in both the formoterol and the non-LABA treatment groups (table 2).

Table 2

Patients with asthma-related hospitalisations by gender, race and daily formoterol dose

Secondary outcomes

All cause mortality was two in 7796 in formoterol-treated patients (rate=0.57 per 1000 patient-years) and none among 4053 non-LABA-treated patients. One death was due to an accident (carbon monoxide intoxication) in a 13-year-old female patient. The second death was classified as asthma-related (a 13-year-old boy who died due to respiratory failure, originally classified as subarachnoidal haemorrhage). The number of deaths was too low to allow any meaningful between-group comparisons.

No case of intubation was reported among the paediatric patients in these studies.

The reporting of SAEs due to any cause was similar between formoterol-treated and non-LABA-treated patients (2.1% vs 2.0%; for a list of the most commonly reported SAEs, see table E3 in the online supplement).

For asthma-related DAEs there was a decreased risk in the formoterol group (RR 0.67, 95% CI 0.46 to 0.98) and no effect of formoterol dose, gender or race.

Two cardiac-related SAEs were reported among the paediatric patients, one among formoterol-treated patients (tachycardia) and one among non-LABA-treated patients (endocarditis). Eight formoterol-treated paediatric patients, compared with none in the other groups, discontinued treatment due to a cardiac-related adverse event. Two of the patients reported Wolff–Parkinson–White syndrome which was present on the ECGs at the screening visit. Neither of these patients experienced a tachyarrhythmia during the study. Of the remaining six patients aged 8–13 years, two reported tachycardia, three reported palpitations and one reported supraventricular extrasystoles. The DAE frequency of tachyarrhythmia in formoterol-treated children (six out of 7796) was less than 0.1% (table 3).

Table 3

Secondary outcomes

Two dose-ranging studies have been conducted over 3-month periods in children with asthma aged 6–17 years. There was no detectable difference in change in heart rate from baseline between patients taking placebo, 9 and 18 µg formoterol per day (see table E4 in the online supplement).

Subset analysis of 24 trials

The overall Mantel–Haenszel RR for formoterol-treated patients was 1.10 (95% CI 0.74 to 1.67) (figure 2A). There were no hospital admissions in either treatment arm in nine of the studies.

Figure 2

Forest-plot of Mantel–Haenszel (MH) exposure-adjusted risk ratios for asthma-related (A) hospitalisations (MH dataset) and (B) discontinuations due to adverse events (DAEs) (MH dataset).

The Kaplan–Meier probability curve for time to first asthma-related hospitalisation shows no evidence that treatment with formoterol increases the risk of asthma-related hospital admission over time (figure 3A). In a Cox regression model, also stratified by study, the HR was 1.10 (95% CI 0.73 to 1.66). Thus no differences in hospitalisation rates were seen with either of the subset analyses.

Figure 3

Kaplan–Meier survival curves describing (A) time to first hospitalisation and (B) time to onset of event leading to asthma-related discontinuation.

There was no significant difference in rate of asthma-related DAEs between formoterol treatment and non-LABA treatment (Mantel–Haenszel RR 0.72, 95% CI 0.50 to 1.04) (figure 2B). There were no asthma-related DAEs in either treatment arm in eight studies. The Kaplan–Meier probability curve for asthma-related events leading to discontinuation shows no evidence that treatment with formoterol increases the risk of asthma DAEs over time (figure 3B). In a Cox regression model, also stratified by study, the HR was estimated to be 0.71 (95% CI 0.49 to 1.05), thus not formally indicating any risk difference for DAEs, at a particular time point during the trials.

The frequencies of adverse events (non-serious and SAEs combined) were similar between formoterol-treated and non-LABA-treated patients (see table E5 in the online supplement). The reporting of cardiac- and vascular-related adverse events was less than 0.15% for any preferred term, and the only cardiac-related adverse event that appeared more common in formoterol-treated patients was ‘palpitations’, as was the musculoskeletal adverse event ‘tremor’ (table 4). Both these events are expected side-effects from treatment with β2 agonists. Sleep disturbances were very rarely reported with no discernable difference between formoterol-treated and non-LABA-treated patients (see table E6 in the online supplement).

Table 4

Incidence of adverse events (serious and non-serious) belonging to the system organ classes ‘cardiac disorders’ and ‘vascular disorders’ and of the preferred term ‘tremor’ in trials with LABA as well as non-LABA treatment arms (Mantel–Haenszel dataset)


Our analysis combines multicentre studies from different countries and involves subjective and cultural differences. This variation was considered to be lessened by using the end point hospitalisation rather than other surrogate measures of a severe asthma attack, such as short courses of oral corticosteroids. Hospital admission rates for children with asthma will vary from country to country, but the rates are generally low. Consequently, when comparing different treatment regimes, studies to determine the relative risk of hospitalisation need to involve a very large number of patients. By utilising the AstraZeneca database of clinical trials, it was possible to make a safety analysis of many more children with asthma than have been included in previously published reports. Our data indicate that treatment with formoterol was not associated with an increase in asthma-related hospital admission. One asthma-related death was reported in 12 395 children participating in 41 trials involving 5410 treatment-years. Apart from stating that asthma-related deaths were extremely rare in these clinical trials, no other conclusion can be drawn.

Guidelines and recommendations for the management of asthma state that ‘Regular use with short- and long-acting β2 agonists is not advised unless accompanied by regular use of an inhaled glucocorticoid’.7 11 Regular β2 agonist use without anti-inflammatory therapy may partly control symptoms but mask inflammation, allowing an increase in airway reactivity and therefore increasing the risk of unexpected asthma attacks.18 Despite this many published trials in adults and children of the use of LABA have included a proportion of patients given a LABA without concomitant inhaled steroid therapy. The vast majority (82%) of the children receiving formoterol in this analysis were also being treated with an ICS. In a Cochrane review19 it was concluded that in adults, there was a decrease in asthma-related SAEs in patients receiving regular formoterol with ICS compared with patients only treated with ICS. In a more recent Cochrane review (‘Addition of long acting beta agonists to inhaled corticosteroids for chronic asthma in children’), it was concluded that although the addition of LABA did not significantly reduce the risk of asthma exacerbations requiring rescue systemic steroids, it improved lung function compared with the same or double the dose of ICS.20

The analysis described in this paper does not carry the same weight as a randomised controlled trial of a comparable size. Analyses of multiple trial datasets such as this can not entirely overcome the difficulties of variation in study design, entry criteria, asthma severity and treatment regimes. Three double-blind randomised controlled trials in children of treatment duration greater than 8 weeks have been published comparing inhaled steroid (budesonide) plus formoterol with inhaled steroid alone.21,,23 However, the results for the frequency of severe asthma exacerbations were contradictory between the different studies.

Our primary analysis could be criticised for including trials in which there was no non-LABA treatment arm. Therefore, an additional analysis was made of hospitalisation in the 24 trials in which children were randomly allocated to either LABA or non-LABA treatment arms. The results of the additional analysis were in full agreement with the results from the full dataset.

We found no difference in hospital admission rates between formoterol-treated and non-LABA-treated children from either Caucasian or non-white backgrounds. The data suggest, however, that overall hospitalisation admission rates in black and oriental children may be greater than in Caucasian children. Several studies have reported variation in asthma management and increased use of hospital facilities in lower socioeconomic groups and ethnic minorities.24 Suboptimal asthma control and underuse of controller medication are also more frequently seen in Hispanic and black ethnic populations.25

The evidence of a dose-dependent relationship between formoterol and the risk for developing asthma-related hospitalisations in children is inconsistent. In one paediatric study there appeared to be an increasing rate of asthma-related hospitalisations with increasing dose of formoterol.16 On the other hand, a large (n=2085) postmarketing trial in patients ≥12 years of age did not show any increased rate of asthma-related hospitalisations with increasing formoterol dose.5 The absence of an increasing risk with increasing dose found in our study is an argument against a true risk connected with formoterol treatment in children and adolescents.

The risk of asthma-related DAEs was slightly lower in the formoterol-treated children, while discontinuations due to cardiac-related adverse events were more common. Most were tachyarrhythmias, a side effect well recognised with β2 agonist treatment.

There are powerful theoretical reasons why LABA should not be used alone for the treatment of asthma and this is reflected in national and international guidelines.11 26 It has been hypothesised that LABA treatment could, by effective relief of symptoms, mask an increase in airway inflammation and delay awareness of worsening asthma.27 Most of the children in this analysis were taking formoterol in combination with an ICS. Given that the use of LABA in combination with ICS for the management of children and adolescents with asthma is increasing,28 29 ongoing pharmacovigilance is warranted. Our data provide strong evidence that formoterol use in children and adolescents with asthma, who are prescribed concomitant ICS, is not associated with any increased risk of asthma-related hospitalisations. The results are not influenced by dose or ethnicity.


The authors acknowledge Stefan Peterson (AstraZeneca, Lund, Sweden) for his statistical expertise and assistance, and Ian Wright, of Wright Medical Communications, UK, who provided editing assistance funded by AstraZeneca.



  • Competing interests At the time of preparation and submission of this manuscript, FR and BL were full-time employees at AstraZeneca R&D, Lund, Sweden. AstraZeneca (Lund Sweden) clinical trials are included and the analysis is made from the company clinical safety database. This is stated in the paper.

  • Provenance and peer review Not commissioned; externally peer reviewed.