Article Text

Download PDFPDF

The risk of asthma mortality with inhaled long acting β-agonists
  1. M Wijesinghe1,
  2. K Perrin1,
  3. M Harwood1,
  4. M Weatherall2,
  5. R Beasley1,3
  1. 1
    Medical Research Institute of New Zealand, Wellington, New Zealand
  2. 2
    University of Otago Wellington, Wellington, New Zealand
  3. 3
    University of Southampton, Southampton, UK
  1. Professor Richard Beasley, Medical Research Institute of New Zealand, PO Box 10055, Wellington 6143, New Zealand; richard.beasley{at}


This article reviews the available evidence as to whether inhaled long acting β-agonists (LABA) increase the risk of asthma mortality and considers the implications for the use of this treatment in the management of asthma. Randomised controlled trials suggest that LABAs prescribed as monotherapy may increase the risk of asthma death in certain circumstances, such as the unsupervised “off-label” use without concomitant inhaled corticosteroid (ICS) treatment in patients with unstable asthma. However, there is also evidence that the use of LABAs in conjunction with ICS treatment in adult asthma as recommended in current guidelines is not associated with an increased risk of asthma mortality. The only way in which a prescriber can ensure that a patient with asthma takes LABA treatment in conjunction with ICS is through a combination ICS/LABA product, an approach which may have additional therapeutic advantages. We propose that in the management of asthma, a case can now be made to limit the availability of LABAs to combination LABA/ICS therapy.

  • inhaled corticosteroids
  • long acting β-agonists

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.

Key learning points

  • Long acting beta-agonists (LABAs) prescribed as monotherapy may increase the risk of asthma mortality in certain circumstances, such as the unsupervised “off-label” use without concomitant inhaled corticosteroid (ICS) treatment in unstable angina.

  • LABAs prescribed with concomitant ICS treatment are not associated with an increased risk of asthma mortality.

  • In the management of asthma, a case can be made to limit the availability of LABAs to combination LABA/ICS treatment.

Concerns about the safety of long acting β-agonist (LABA) drugs represent the latest chapter in a long running debate about the role of inhaled β-agonist treatment in asthma mortality.110 One of the features that has characterised the debate has been the apparent conflicting and inconsistent nature of the evidence which has led to polarised opinions. The recent chapter in this debate is no exception, with widely differing views on the safety of LABA treatment and its role in the management of asthma.1116 In this review the findings of the key studies are considered and an attempt made to bring together the evidence in a consistent manner, with recommendations made regarding the use of LABA treatment in the management of asthma.


Concerns about the possible risk of mortality associated with LABA treatment were first raised in 1993 with publication of the Salmeterol Nationwide Surveillance Study (SNSS).17 This UK based randomised double blind study compared salmeterol with salbutamol as regular treatment in over 25 000 patients over a 16 week period. There was a non-significant threefold increased risk of death in subjects treated with salmeterol (12 of 16 787 subjects vs 2 of 8393 subjects, p = 0.10), but no increase in hospital admissions or life threatening events. Although inhaled corticosteroid (ICS) treatment was prescribed as regular maintenance therapy in about two thirds of the subjects in this study, the use of concomitant ICS treatment in subjects who died was not reported.


The main evidence that LABAs may increase the risk of mortality is obtained from the Salmeterol Multicenter Asthma Research Trial (SMART), which was published in 2006.18 This randomised placebo controlled trial undertaken in the USA investigated the risk of salmeterol treatment in adult subjects with unstable asthma. An interim analysis, which was undertaken after 26 000 subjects had been enrolled, showed no significant difference between the two groups in the primary outcome variable, respiratory related deaths or life threatening experiences (relative risk (RR) 1.40, 95% confidence interval (CI) 0.91 to 2.14). However, there was a significant twofold increase in respiratory related deaths (RR 2.16, 95% CI 1.06 to 4.41) and a fourfold increase in asthma mortality (RR 4.37, 95% CI 1.25 to 15.34) with salmeterol. Importantly, in subgroup analyses, there was no increased risk in subjects taking concomitant ICS treatment, with the risk associated with salmeterol limited to patients not receiving ICS as regular maintenance therapy (table 1).

Table 1 Adverse outcomes in SMART study18 in relation to inhaled corticosteroid (ICS) use

Consideration of the methodology of this trial is essential to interpret these findings. Subjects were given a full 6 month supply of salmeterol or placebo at enrolment and then received no scheduled study based medical review for the entire 6 month period of the study. It is likely that some of the patients recruited by public advertisement did not receive any long term medical care for the duration of the clinical trial. Most subjects had poorly controlled asthma with significant morbidity, yet just under half were receiving ICS treatment. As a result the SMART findings relate to the use of salmeterol monotherapy, contrary to current consensus guidelines, in which it is recommended that LABAs are added to pre-existing ICS treatment and that patients receive regular follow-up by their doctor.19 20

It was evident from both the SNSS and SMART studies that a randomised placebo controlled trial may not be the preferred method to investigate the risk of a rare outcome such as asthma death. The difficulties in designing and conducting clinical trials, in which over 25 000 subjects participated, to study 14 and 16 deaths respectively are apparent, resulting in the case of SMART that the findings are not necessarily generalisable to clinical practice. A complementary epidemiological approach to investigate the possible role of drug treatment in asthma mortality is to undertake a cohort or case–control study.21


The most important case–control study of LABA treatment and asthma mortality was undertaken in the UK.22 This epidemiological study reported the relative risk of death with asthma for all major groups of asthma medications. The study covered around one quarter of the entire population of the UK and included all deaths from asthma in patients under 65 years of age over a 5 year period (1994–1998), with controls being patients with asthma who had a hospital admission for asthma, matched for period, age and area. This study had considerably greater power than the SNSS17 and SMART18 studies (532 vs 14 and 16 asthma deaths, respectively), obtained more detailed clinical information, and most importantly, examined the risk of mortality in patients under standard medical care thereby making the findings more “generalisable”.23

In this case–control study,22 there was no evidence of any positive association with salmeterol and death in any period, and the upper 95% confidence intervals excluded the possibility of a clinically significant adverse effect (table 2). Furthermore, the risk of mortality did not increase when the analysis was restricted to subjects with chronic severe asthma, identified by a hospital admission in the previous 12 months (table 3). This pattern contrasted with that observed in the New Zealand case–control studies of similar design, in which the risk of death associated with the high dose preparation of the poorly selective β2-agonist fenoterol increased markedly in the severe subgroup analyses, indicating that the increased risk of mortality with fenoterol was not due to “confounding by severity” (fig 1).7 9 10

Figure 1 Comparison of the long acting β-agonist (LABA) findings from the UK case–control study22 with those relating to fenoterol from the similar New Zealand (NZ) studies of deaths from asthma.7 9 10 For each study, the left hand axis shows the overall relative risk of death for those prescribed fenoterol or LABA; the right hand axis represents the same analysis in the subgroup of those patients with a hospital admission in the previous 12 months (the most valid marker of chronic asthma severity).3234
Table 2 Odds ratio (95% confidence interval) for death associated with long acting β-agonist (LABA) treatment in the 3 months and 4–12 months before the index date*: UK case–control study22
Table 3 Adjusted odds ratio (95% CI) for death associated with long acting β-agonist (LABA) treatment overall and in severity subgroup defined by recent hospital admission*: UK case–control study22

The proportion of patients on LABA treatment prescribed concomitant ICS treatment was not reported in this case–control study. However, from the UK General Practice Research Database it has been reported that during this period around 95% of asthma patients prescribed LABA treatment were co-prescribed ICS treatment.24 As a result, one can conclude that the findings from the case–control study provide evidence that the use of LABAs in accordance with standard practice, which includes concomitant ICS treatment, does not increase the risk of asthma mortality. This interpretation is consistent with the findings from the SMART study, in which the use of LABA with ICS treatment also did not increase the risk of mortality.

Similarly, an earlier case–control study based on the General Practice Research Database reported no significant difference in risk of asthma death associated with LABA treatment after adjusting for confounding variables, although the confidence intervals were wide.25 In a smaller case–control study of intensive care unit (ICU) admissions for asthma, also undertaken in the UK, those prescribed salmeterol were at no greater risk of a near fatal asthma attack, after controlling for the selective prescribing of salmeterol to the more severe asthma subjects.26 Two further studies have reported no significantly increased risk in severe non-fatal asthma27 or asthma death28 in patients prescribed salmeterol in comparison with theophylline use, although the interpretation of these findings depends on the risk associated with theophylline treatment and the confidence intervals were again wide.


Another informative approach is the systematic review and meta-analysis of randomised controlled trials of LABA treatment. A number of such meta-analyses have now been undertaken2931 and have reported the effects of LABA treatment on asthma mortality, and markers of risk of asthma mortality such as hospital or ICU admission.

LABAs as a class

The meta-analysis which has received greatest publicity is that undertaken by Salpeter and colleagues.29 This meta-analysis included all randomised placebo controlled trials of the LABAs salmeterol and formoterol and reported the findings in relation to risk of hospital admission, life threatening attacks and mortality. For hospital admissions, a recognised marker of risk of asthma mortality,3234 there was a 2.6-fold (95% CI 1.6 to 4.3) increased risk associated with LABA treatment. For life threatening episodes requiring intubation and ventilation, there was a non-significant 1.5-fold (95% CI 0.9 to 2.5) increase in risk, when deaths were excluded from the analysis. The findings relating to mortality were dominated by the SMART study, with only two deaths observed in the other 18 trials included in the meta-analysis. The use of ICS treatment was inconsistent across the studies, and it was not possible to assess its effect on the risk associated with LABAs.

However, concomitant ICS treatment was a requirement of studies included in the Cochrane meta-analysis of the risk of hospital admissions with LABA treatment, prescribed in addition to ICS treatment.30 This identified no increased risk, contrasting with the findings of the Salpeter et al meta-analysis,29 suggesting that the increased risk may relate to LABA monotherapy (table 4). Three additional meta-analyses have been undertaken in which the effect of LABA treatment in patients taking ICS has been compared with at least doubling the dose of ICS treatment.31 35 36 These studies have identified a similar or reduced risk of severe attacks resulting in withdrawal or hospital admissions with LABA treatment.

Table 4 Risk of asthma hospitalisations with long acting β-agonist (LABA) treatment, according to concurrent use of inhaled corticosteroids (ICS) derived by meta-analysis from systematic reviews†


To investigate further the risk of severe asthma related events with salmeterol administered with ICS, a meta-analysis of all eligible GlaxoSmithKline studies has recently been reported in abstract form.37 This meta-analysis of 19 850 subjects included 52 trials comparing ICS and salmeterol (administered via a single device or concomitant administration of ICS and salmeterol separately), with ICS alone. There was no increase in risk of asthma related hospitalisations with an odds ratio of 1.02 (95% CI 0.64 to 1.63), comparing ICS and salmeterol versus ICS alone. The odds ratio was 0.83 (95% CI 0.47 to 1.49) for fluticasone and salmeterol via a single device versus fluticasone alone. For fluticasone and salmeterol in separate devices versus fluticasone alone, the odds ratio was 1.29 (95% CI 0.41 to 4.10). One asthma death and one asthma related intubation occurred among subjects receiving ICS and salmeterol in separate inhalers. There were no asthma related deaths or intubations in the subjects with moderate to severe asthma receiving fluticasone and salmeterol in a single combination device.


Clinical data of the safety of formoterol is available from both the Novartis and AstraZeneca clinical trial databases. The major clinical data of formoterol from Novartis is the analysis of the three prospective, randomised, placebo controlled, double blind clinical studies which Novartis presented to the US Food and Drug Administration (FDA).38 This analysis included studies of both 12 μg and 24 μg twice daily dose regimens, allowing an assessment of the dose–response relationship for serious asthma exacerbations with formoterol. Although insufficiently powered to examine mortality, this showed that there was a dose dependent increase in the risk of serious asthma exacerbations associated with formoterol treatment, compared with both placebo and regular salbutamol treatment (table 5). In the high dose formoterol group one patient died of cardiorespiratory arrest and two patients required intubation and mechanical ventilation. However, in a subsequent large clinical trial, treatment with formoterol 24 μg twice daily was not associated with an increase in serious asthma exacerbations compared with lower dose formoterol regimens.39

Table 5 Occurrence of serious adverse exacerbations* with formoterol†38

The major clinical data of formoterol from AstraZeneca is a meta-analysis based on their clinical trials database which has recently been reported in abstract form.40 This included 18 studies of at least 12 weeks duration comparing ICS and formoterol with ICS alone in over 12 000 patients with about 3500 patient years of exposure. Formoterol treatment resulted in fewer asthma related hospitalisations with an odds ratio of 0.59 (95% CI 0.38 to 0.92) and fewer asthma related serious adverse events with an odds ratio of 0.57 (95% CI 0.37 to 0.87). There were two asthma related deaths both in patients prescribed formoterol.


One of the striking features of the epidemics of asthma mortality associated with isoprenaline forte and high dose fenoterol in the 1960s and 1970s, respectively, was the strong temporal association between increasing sales of these drugs and increasing rates of mortality, and the sudden and pronounced reduction in mortality with their regulatory restriction.5 6 9 10 While such ecological analyses represent weak evidence of a causal association, they do provide a worthwhile check of the generalisability of the findings from the epidemiological and clinical studies.

Trends in asthma mortality in the USA have been of particular interest due to the suggestion by Salpeter et al that around 4000 deaths in the USA could be attributed to salmeterol treatment,29 based on the SMART data.18 However, asthma mortality was steady and then decreased rather than increased with the introduction of salmeterol, initially as single inhaler treatment in 1993 and then as fluticasone/salmeterol combination therapy in 2001 (fig 2).41 During the 10 year period after the introduction of salmeterol, mortality fell by about one third, such that in 2005 there were 3884 deaths from asthma in the USA,42 fewer than the 4000 deaths predicted by Salpeter et al to be solely attributable to salmeterol.29 These observations further suggest that the SMART study results are not generalisable to the usual use of LABAs in the USA. Indeed, it has been calculated that there would be a total of 10 589 asthma deaths in the USA based on the SMART data, around 2.5 times the actual number of deaths recorded in the US vital statistics.43

Figure 2 The time trends in asthma mortality, inhaled corticosteroid (ICS) and long acting β-agonist (LABA) sales in the USA since 1993. For asthma mortality, the 1993 to 1998 rates reflect the International Classification Of Diseases, ninth revision (ICD-9), code 493, and the 1999 to 2005 rates reflect the ICD-10 codes. (A) Sales of inhaled LABA drugs as single inhaler, and combination ICS/LABA treatment, in units sold (millions) in the USA between 1993 and 2006. The asthma mortality rate (per 100 000 population) in the USA between 1993 and 2005 is shown for all ages and for the 5–34 year age group. (B) Sales of ICS as a single inhaler, and combination ICS/LABA treatment, in units sold (millions) in the USA between 1993 and 2006. The asthma mortality rate (per 100 000 population) in the USA between 1993 and 2005 is shown for all ages and for the 5–34 year age group.

It is important to acknowledge that the 10th revision to the International Classification of Diseases (ICD-10) resulted in an estimated 10% reduction in the number of deaths classified as asthma in the USA,44 accounting in part for the fall in the asthma death rate when the ICD-10 was introduced in 1999. However, this classification change is unlikely to have influenced the asthma mortality rates in the 5–34 year age group which progressively declined from 0.56 per 100 000 in 1995 to 0.35 per 100 000 in 2005.42 The major cause of the reduction in asthma mortality in the USA over the last decade is likely to be the progressive and pronounced increase in ICS use. In the 10 years since 1995 when asthma mortality first levelled off and began to decline, the sales of ICS treatment have increased about threefold (fig 2). The most recent increase in ICS treatment has been driven by the progressively greater use of combination ICS/LABA products, which now represent just over half of ICS use in the USA. More importantly, in the USA over 85% of LABA use is now in the form of combination ICS/LABA treatment. It could be argued that combination LABA/ICS treatment can be considered as a vehicle for delivering and improving compliance with ICS treatment in the management of asthma.45 46 Combination LABA/ICS treatment also represents the only way in which a prescriber can ensure that LABAs are not used as monotherapy.


Although the focus of this review is primarily on the use of LABAs and the risk of mortality in adults, their use in paediatric asthma warrants consideration. Many of the pathological, physiological and clinical characteristics of asthma in children differ to those in adults, as do the pharmacokinetics of medications used.47 Current guidelines for the management of paediatric asthma are largely based on extrapolation from studies in adults or existing guidelines rather than evidence derived from the paediatric literature.48 A meta-analysis of the effects of LABAs on exacerbation rates of asthma in children included studies exclusively conducted in under 18-year-olds.49 This showed that the relative risk of hospitalisation for asthma in patients treated with LABAs, compared with placebo or a short acting β-agonist, ranged from 3.3 to 21.6 in the three studies that reported this outcome. When these data are considered together with the lack of major clinical benefit from LABA treatment in children, serious concerns can be expressed regarding the use of LABA treatment in childhood asthma. The safety of LABAs and the risk of mortality in children clearly needs to be urgently addressed.


Finally, it is relevant to consider the risk of mortality associated with LABA treatment in chronic obstructive pulmonary disease (COPD). In a large randomised, double blind, placebo controlled trial there was no significant risk of mortality associated with the use of salmeterol either as a separate inhaler or as combination salmeterol/fluticasone inhaler treatment in adults with COPD.50 This finding was reassuring due to the presence of concomitant COPD in a significant proportion of patients with chronic severe asthma and incompletely reversible airflow obstruction.


We conclude that the balance of evidence from clinical studies suggests that LABAs prescribed as monotherapy may increase the risk of asthma mortality in certain circumstances, such as the unsupervised “off label” use, without concomitant ICS in patients with unstable asthma. However, the balance of evidence also suggests that the use of LABAs in conjunction with ICS treatment in adults as recommended in current guidelines may not be associated with an increased risk of mortality.

The only way in which a prescriber can ensure that a patient with asthma takes LABA in conjunction with ICS treatment is through an ICS/LABA combination product. This also has the potential advantages of improving compliance with ICS treatment,45 46 and allowing a lower dose of ICS to be taken.51 52 We propose that on the basis of currently available evidence, a case can now be made to limit the availability of LABAs to combination LABA/ICS treatment in the management of asthma. In contrast to asthma, it is recommended that in COPD, LABAs may be prescribed without concomitant ICS treatment,53 and for this reason it is not suggested that the availability of LABAs is also limited to combination LABA/ICS treatment in COPD.

Key references

  • Crane J, Pearce N, Burgess C, et al. Asthma and the b agonist debate. Thorax 1995;50(suppl 1):S5–10.

  • Nelson HS, Weiss ST, Bleecker ER, et al. The Salmeterol Multicenter Asthma Research Trial: a comparison of usual pharmacotherapy for asthma or usual pharmacotherapy plus salmeterol. Chest 2006;129:15–26.

  • Anderson HR, Ayres JG, Sturdy PM, et al for the Mortality and Severe Morbidity Group of the National Asthma Task Force. Bronchodilator treatment and deaths from asthma: case-control study. BMJ 2005;330:117–23.

  • Bisgaard H, Szefler S. Long-acting beta2 agonists and paediatric asthma. Lancet 2006;367:286–8.

  • Calverley PM, Anderson JA, Celli B, et al for the TORCH Investigators. Salmeterol and fluticasone propionate and survival in chronic obstructive pulmonary disease. N Engl J Med 2007;356:775–89.



  • Competing interests: The Medical Research Institute of New Zealand holds a current research grant from MedSafe, New Zealand Ministry of Health, to investigate the risk of death with long acting β-agonist therapy. The Medical Research Institute of New Zealand and the University of Otago Wellington have received research grants from AstraZeneca, GlaxoSmithKline and Novartis. RB has received fees for consulting and speaking and/or reimbursement for attending symposia from AstraZeneca, Boehringer Ingelheim, GlaxoSmithKline and Novartis. RB is a member of the NHLBI/WHO Global Initiative on Asthma Assembly. MW, KP, MH and MW have no conflict of interest to declare.