Regular treatment with formoterol for chronic asthma: serious adverse events

Christopher J Cates; Matthew J Cates

doi:10.1002/14651858.CD006923.pub3

. Author manuscript; available in PMC: 2014 May 12.

Published in final edited form as: Cochrane Database Syst Rev. 2012 Apr 18;4:CD006923. doi: 10.1002/14651858.CD006923.pub3

Regular treatment with formoterol for chronic asthma: serious adverse events

Christopher J Cates ¹, Matthew J Cates ²

PMCID: PMC4017186 EMSID: EMS57813 PMID: 22513944

Abstract

Background

Epidemiological evidence has suggested a link between beta₂-agonists and increases in asthma mortality. There has been much debate about possible causal links for this association, and whether regular (daily) long-acting beta₂-agonists are safe.

Objectives

The aim of this review is to assess the risk of fatal and non-fatal serious adverse events in trials that randomised patients with chronic asthma to regular formoterol versus placebo or regular short-acting beta₂-agonists.

Search methods

We identified trials using the Cochrane Airways Group Specialised Register of trials. We checked websites of clinical trial registers for unpublished trial data and Food and Drug Administration (FDA) submissions in relation to formoterol. The date of the most recent search was January 2012.

Selection criteria

We included controlled, parallel design clinical trials on patients of any age and severity of asthma if they randomised patients to treatment with regular formoterol and were of at least 12 weeks’ duration. Concomitant use of inhaled corticosteroids was allowed, as long as this was not part of the randomised treatment regimen.

Data collection and analysis

Two authors independently selected trials for inclusion in the review. One author extracted outcome data and the second author checked them. We sought unpublished data on mortality and serious adverse events.

Main results

The review includes 22 studies (8032 participants) comparing regular formoterol to placebo and salbutamol. Non-fatal serious adverse event data could be obtained for all participants from published studies comparing formoterol and placebo but only 80% of those comparing formoterol with salbutamol or terbutaline.

Three deaths occurred on regular formoterol and none on placebo; this difference was not statistically significant. It was not possible to assess disease-specific mortality in view of the small number of deaths. Non-fatal serious adverse events were significantly increased when regular formoterol was compared with placebo (Peto odds ratio (OR) 1.57; 95% CI 1.06 to 2.31). One extra serious adverse event occurred over 16 weeks for every 149 people treated with regular formoterol (95% CI 66 to 1407 people). The increase was larger in children than in adults, but the impact of age was not statistically significant. Data submitted to the FDA indicate that the increase in asthma-related serious adverse events remained significant in patients taking regular formoterol who were also on inhaled corticosteroids.

No significant increase in fatal or non-fatal serious adverse events was found when regular formoterol was compared with regular salbutamol or terbutaline.

Authors’ conclusions

In comparison with placebo, we have found an increased risk of serious adverse events with regular formoterol, and this does not appear to be abolished in patients taking inhaled corticosteroids. The effect on serious adverse events of regular formoterol in children was greater than the effect in adults, but the difference between age groups was not significant.

Data on all-cause serious adverse events should be more fully reported in journal articles, and not combined with all severities of adverse events or limited to those events that are thought by the investigator to be drug-related.

BACKGROUND

Description of the condition

There is currently no universally accepted definition of the term ‘asthma’. This is in part due to an overlap of asthmatic symptoms with those of other diseases such as chronic obstructive pulmonary disease (COPD), but is also due to the probable existence of more than one underlying pathophysiological process. There are, for example, wide variations in the age of onset, symptoms, triggers, associations with allergic disease and the type of inflammatory cell infiltrate seen in patients diagnosed with severe asthma (Miranda 2004). Patients with all forms and severity of disease will typically have intermittent symptoms of cough, wheeze and/or breathlessness. Underlying these symptoms there is a process of variable, at least partially reversible airway obstruction, airway hyper responsiveness and, in most cases, chronic inflammation.

Airway obstruction

Patients with a history of asthma demonstrate chronic changes within the airways including goblet cell hyperplasia, airway smooth muscle (ASM) hyperplasia and hypertrophy (Ebina 1993; Ordonez 2001; Woodruff 2004) and excess myofibroblasts with increased subepithelial collagen deposition (Brewster 1990). In the acute setting, in patients who have died of status asthmaticus, airway obstruction is evident from air-trapping and lung hyperinflation with mucus plugging of the small and large airways (Dunnill 1960; Kuyper 2003). There is also shedding of ciliated bronchial mucosal cells, inflammatory cell infiltrates and submucosal oedema with transudation of fluid into the bronchial lumen (Carroll 1993). It is more difficult to measure the degree of ASM contraction (bronchoconstriction) in post-mortem studies although evidence for a role of bronchoconstriction in airway narrowing comes from other sources.

Airway hyper responsiveness

Patients with asthma typically display a degree of ‘airway hyper responsiveness’ to inhaled allergens (Cockcroft 2006), and to a variety of chemical stimuli including histamine, serotonin, bradykinin, prostaglandins, methacholine and acetylcholine as well as other triggers such as exercise, deep inhalation and inhalation of cold air (Boushey 1980). Bronchoconstriction is implicated as the primary effector mechanism of airway narrowing in these responses. This is because of both the short time frame of the response and because many of these stimuli typically either cause bronchoconstriction directly in vitro or promote bronchoconstriction through interference with the autonomic control of ASM. Further evidence comes from findings that this response can be abolished or diminished by bronchodilator medications such as atropine and beta₂-agonists (Phillips 1990; Simonsson 1967); although beta₂-agonists in particular may have additional mechanisms of action. Whether airway hyper responsiveness relates primarily to an abnormality of ASM, to increased ASM bulk (Wiggs 1990), to aberrant autonomic control or reflex pathways, or to physical damage to the airway epithelium remains to be established. Regular use of salbutamol has, however, been shown to increase airway hyper responsiveness to allergen exposure and produce tolerance to the protective effect of salbutamol against bronchoconstriction induced by both methacholine and allergens (Cockcroft 1993).

Inflammation

It has long been thought that the histological changes described above and the phenomenon of airway hyper responsiveness are due to a combined acute and chronic inflammatory response (Bousquet 2000). Patients with status asthmaticus have increased numbers of inflammatory cells including eosinophils and neutrophils, as well as a variety of pro-inflammatory cytokines and chemokines found in bronchial alveolar lavage (Tonnel 2001). In patients with chronic asthma there is also evidence of increased eosinophil numbers (Bousquet 1990), inflammatory cell adhesion molecules (Vignola 1993) and some evidence of an association between the extent of inflammation, disease severity and hyperreactivity. This association has however been questioned on the background of a number of negative results (Brusasco 1998), although it is made difficult to prove by the lack of a consistent marker of a sequential and variable inflammatory response (Haley 1998).

Description of the intervention

Beta₂-agonists and mortality: an historical perspective

Time trend data and case-control studies

Adrenaline was successfully used in the symptomatic treatment of asthma as far back as 1903 (Tattersfield 2006). Initially given subcutaneously, the inhaled route was tried in 1929 to reduce adverse effects but these remained a problem and in 1940 details of a new agent, isoprenaline (isoproterenol), were published in Germany (Konzett 1940). Although isoprenaline was more selective for beta- as opposed to alpha-adrenoreceptors, adverse effects including palpitations were still a major problem, particularly with oral administration (Gay 1949) and it first became available as atomiser spray for use in the UK in 1948 (Pearce 2001).

Prior to the 1940s, mortality rates from asthma in a number of countries were stable and low at less than 1 asthma death per 100,000 people per year (Pearce 2001; Figure 1). During the 1940s and 50s there was a slight rise in mortality rates and concerns about a possible link to inhaled adrenaline were raised at an early stage (Benson 1948). However, the rise was small and the cause unclear and sales continued to increase with the introduction of aerosol or metered-dose inhalers in the early 1960s. During this decade there was an epidemic of asthma deaths in at least six countries including England, Wales and New Zealand (Figure 1). In all six countries the epidemics coincided with the licensing of an aerosol called ‘Isoprenaline Forte’, which contained five times the dose of isoprenaline per administration than the standard preparation (Stolley 1972). In other countries including the Netherlands, where isoprenaline forte was introduced late and sales volumes low, and in the US, where isoprenaline forte was not licensed, no increase in asthma mortality occurred. This was despite an approximate trebling in per capita alternative bronchodilator sales between 1962 and 1968 in the US (Stolley 1972). A detailed review of the epidemic in England and Wales concluded it was not due to changes in death certification, disease classification or an increase in asthma prevalence, but instead was most likely due to new methods of treatment (Speizer 1968). In England and Wales mortality rates fell following health warnings about the overuse of inhalers and banning of over the counter sales in 1968. It was around this time that more selective beta₂-agonists such as terbutaline (Bergman 1969) and salbutamol (albuterol) (Cullum 1969) were being developed.

In the late 1970s a second epidemic of asthma deaths occurred in New Zealand (Figure 1). It was later shown that this epidemic coincided with the introduction and rising sales of fenoterol, a new short-acting beta₂-agonist (Crane 1989; Figure 2). A significant association between mortality and fenoterol use was demonstrated in three consecutive case-control studies, the latter studies addressing criticisms of the first (Crane 1989; Grainger 1991; Pearce 1990). Furthermore the relative risk of asthma death in patients prescribed fenoterol increased markedly when analysis was restricted to subgroups defined by markers of severity, including previous hospital admission and use of oral corticosteroids. Following the publication of the first case-control study, the fenoterol market share in New Zealand fell from 30% in 1988 to 3% in 1991 and by the early 1990s the mortality epidemic appeared to be over (Figure 2). During the gradual decline in mortality in New Zealand from its peak in 1979, total sales of alternative beta₂-agonists, including salbutamol, gradually rose and the use of inhaled corticosteroids also increased during the latter half of the 1980s (Pearce 2007).

The introduction of long-acting beta₂-agonists

Given the relatively short action of beta₂-agonists such as salbutamol, in the late 1980s efforts were made to develop longeracting compounds. Subsequently the long-acting beta₂-agonists (LABAs), salmeterol and formoterol were released by Glaxo-SmithKline (GSK) and Novartis, respectively. Both drugs cause bronchodilation that lasts for more than 12 hours although formoterol has a faster onset of action (Kemp 1993; Ringdal 1998). Given previous concerns about the safety profile of some of the short acting beta₂-agonists, salmeterol and formoterol were subject to randomised controlled trials on larger numbers of patients. Using these trials several Cochrane reviews have addressed the efficacy of LABAs in addition to inhaled corticosteroids (Ni Chroinin 2004; Ni Chroinin 2005), in comparison with placebo (Walters 2007), short-acting beta2-agonists (Walters 2002), leukotriene-receptor antagonists (Ducharme 2006), and increased doses of inhaled corticosteroids (Greenstone 2005). The beneficial effects of LABAs on lung function, symptoms, quality of life and exacerbations requiring oral steroids have been demonstrated. However, with some studies demonstrating an associated increase in mortality concerns about the safety profile of LABAs have heightened and there has been much debate about the potential protective role of inhaled corticosteroids.

How the intervention might work

We have outlined the pharmacology of beta₂-agonists in detail in Appendix 1. Since the early epidemics in asthma mortality, a number of potential mechanisms have been proposed to explain a relationship to the use of beta₂-agonists. We discuss these mechanisms in detail in Appendix 2; they include direct toxicity, tolerance, delay in seeking help and reduction in use of inhaled corticosteroids.

Why it is important to do this review

We have taken a different approach from Salpeter 2006, in that we have not assumed a class effect of long-acting beta₂-agonists, but we have considered trials comparing regular formoterol to placebo or regular salbutamol/terbutaline. We have chosen not to include results from trials on salmeterol in this review, as there are known differences in the pharmacological properties of salmeterol and formoterol (Ringdal 1998; Van Noord 1996). This review forms part of a set of reviews on the safety of regular salmeterol and formoterol that has now been published (Cates 2008; Cates 2009; Cates 2009a; Cates 2011). We have also excluded studies which randomised participants to formoterol and inhaled corticosteroids for this review, as these are considered in another review (Cates 2009a).

In view of the difficulty in ascertaining the causation of deaths and serious adverse events (SAEs), we have considered all-cause fatal and non-fatal SAEs as the main outcomes of this review, with asthma-related and cardiovascular events as secondary outcomes.

OBJECTIVES

To assess the risk of mortality and non-fatal serious adverse events in trials which randomise patients with chronic asthma to formoterol alone

METHODS

Criteria for considering studies for this review

Types of studies

We included randomised trials (RCTs) of parallel design, with or without blinding, in which formoterol alone was randomly assigned to patients with chronic asthma. We excluded studies on acute asthma and exercise-induced bronchospasm.

Types of participants

We included patients with a clinical diagnosis of asthma of any age group, unrestricted by disease severity, previous or current treatment.

Types of interventions

We included trials that randomised patients to receive inhaled formoterol twice daily for a period of at least 12 weeks, at any dose and delivered by any device (metered-dose inhalers (MDIs) with chlorofluorocarbons (CFCs) or hydrofluoroalkane (HFAs), or dry powder inhalers (DPIs)). We included studies that used comparison groups with placebo or short-acting beta₂-agonists, and co-intervention with leukotriene receptor antagonists, inhaled or oral corticosteroids or theophylline was allowed as long as they are not part of the randomised intervention. We excluded studies that randomised patients to formoterol for intermittent use as a reliever, and studies that compared different doses of formoterol, or different delivery devices or propellants (with no placebo arm). We also excluded studies in which formoterol was randomised together with an inhaled steroid (in separate inhalers or a combined inhaler) from this review; however these are considered in a separate review (Cates 2009a).

Types of outcome measures

Primary outcomes

All-cause mortality
All-cause non-fatal serious adverse events

Secondary outcomes

Asthma-related mortality
Asthma-related non-fatal serious adverse events
Respiratory-related mortality
Respiratory-related non-fatal serious adverse events
Cardiovascular-related mortality
Cardiovascular-related non-fatal serious adverse events
Asthma-related non-fatal life-threatening events (intubation or admission to intensive care)
Respiratory-related non-fatal life-threatening events (intubation or admission to intensive care)

We did not restrict outcomes to those that the trial investigators considered to be related to trial medication.

Search methods for identification of studies

Electronic searches

We identified trials using the Cochrane Airways Group Specialised Register of trials, which is derived from systematic searches of bibliographic databases including the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, CINAHL, AMED and PsycINFO, and handsearching of respiratory journals and meeting abstracts (see Appendix 3 for details). All records in the Specialised Register coded as ‘asthma’ were last searched in January 2012 using the following terms: (((beta* and agonist*) and (long-acting or “long acting”)) or ((beta* and adrenergic*) and (long-acting or “long acting”)) or (bronchodilat* and (long-acting or “long acting”)) or (salmeterol or formoterol or eformoterol or Advair or Symbicort or serevent or Seretide or Oxis)) AND (serious or safety or surveillance or mortality or death or intubat* or adverse or toxicity or complications or tolerability)

Searching other resources

We checked reference lists of all primary studies and review articles for additional references. We also checked websites of clinical trial registers for unpublished trial data and FDA submissions in relation to formoterol.

Data collection and analysis

Selection of studies

Both authors (CJC, MJC) independently assessed studies identified in the literature searches by examining titles, abstract and keywords fields. We obtained studies that potentially fulfilled the inclusion criteria in full text. We independently assessed these full-text trial reports for inclusion. We resolved disagreements by consensus. We kept a record of decisions.

Data extraction and management

We extracted data using a prepared checklist before entering into Rev Man 5.0. We entered data on characteristics of included studies (methods, participants, interventions, outcomes) and results of the included studies. We contacted authors or manufacturers if serious adverse event data were not included in the trial report, and searched manufacturers’ and FDA websites for further details of adverse events.

Assessment of risk of bias in included studies

Both authors assessed the included studies for bias protection (including sequence generation for randomisation, allocation concealment, blinding of participants and assessors, loss to follow-up, completeness of outcome assessment and other possible bias prevention). We resolved disagreements by consensus.

Measures of treatment effect

We recorded the number of participants with a serious adverse event of any cause (fatal and non-fatal), and in view of the difficulty in deciding whether events are asthma-related, we noted details of the cause of death and serious adverse events where they were available. We noted the definition of serious adverse events, and sought further information if this was not clear (particularly in relation to hospital admissions and serious adverse events).

Unit of analysis issues

We extracted data using the number of participants who suffered one or more serious adverse events, in order to avoid double-counting events from the same participant.

Assessment of heterogeneity

We assessed heterogeneity in the pooled odds ratio using the I² statistic in RevMan 5.0.

Assessment of reporting biases

We assessed reporting bias by comparing the published and unpublished serious adverse events, and by comparing all serious adverse events with those that were thought to be drug-related. We also compared serious events with all adverse events (whether serious or not).

Data synthesis

The outcomes of this review were dichotomous and we recorded the number of participants with each outcome event, by allocated treated group. We planned to analyse mortality using risk difference, as many studies did not have any deaths in either arm. However, the recently revised Cochrane Handbook for Systematic Reviews of Interventions advises against this approach, so we only used the risk differences to estimate the absolute impact of treatment (Higgins 2008). Although meta-analysis with Peto odds ratio has advantages when events are rare (Bradburn 2007), it performs less well with unbalanced treatment arms and large effect sizes, and therefore we calculated the results for serious adverse events and mortality in RevMan 5.0 as pooled odds ratios using the Mantel-Haenszel (MH) fixed-effect model. We compared the results of this model to the Peto method and MH random-effects models (as sensitivity analysis), although Bradburn 2007 cautions against the random-effects model as the variance calculations are based on large sample assumptions. We explored heterogeneity on the basis of the subgroup and sensitivity analyses outlined below. We inspected funnel plots to assess publication bias.

Subgroup analysis and investigation of heterogeneity

We conducted subgroup analyses on the primary outcomes on the basis of dose of formoterol (usual dose versus high dose), age (adults versus children) and comparator used. We planned to carry out subgroup analysis based on reported corticosteroid use, but none of the studies reported whether the patients who actually suffered serious adverse events were taking inhaled corticosteroids or not. We compared subgroups using tests for interaction (Altman 2003).

The definition of serious adverse events was rarely reported in the trials, but there is a standard definition used by industry in clinical study reports (ICHE2a 1995) and this is listed in Appendix 4.

Sensitivity analysis

We checked the overall results for the primary outcomes to assess the impact of removing the results from unblinded studies, and also those participants that were randomised to high-dose formoterol (24 μg twice daily). We also checked the impact of using fixed or random-effects models for meta-analysis.

RESULTS

Description of studies

Results of the search

See Figure 3 for the study flow diagram. We found 512 abstracts from the initial search in October 2007 (reduced to 504 after removing duplicates). We identified 128 relevant abstracts that related to formoterol alone. We included 22 studies (32 references) in the review and excluded 95 others (see Excluded studies). Of those excluded 53 were less than 12 weeks in duration, (including 29 single-dose studies), three were not RCTs, eight were dose or device comparison studies, nine were on exercise-induced bronchospasm or acute asthma, nine used formoterol as reliever medication, eight were cross-over in design, three randomised to formoterol and budesonide and five were excluded for other reasons. One study that was included in Salpeter 2006 was excluded from this review because formoterol was randomised with budesonide (Price 2002) and therefore did not fit the inclusion criteria for this review. Both authors reached consensus on all the included studies after inspection of the full text from papers and websites.

We repeated the search in July 2008, when we found 48 further abstracts. Six were relevant to this review but we excluded five as they were short-term studies, and one was on exercise-induced bronchospasm; we found no new studies meeting the inclusion criteria. Similarly a further search in January 2012 (cumulative total of 748 citations) identified three potentially relevant studies (Happonen 2009; Kamenov 2007; Price 2008). On checking the full text, none of these yielded new trials that were suitable for inclusion (see Excluded studies).

We identified one additional included study from other reviews; this was a trial on the AstraZeneca controlled trials register (SD-037-0344) which is otherwise unpublished. Additionally two documents on the FDA website provided additional serious adverse event information for four of the included studies (Bensch 2001; Bensch 2002; Pleskow 2003; Wolfe 2006). Additionally 22 studies of at least four weeks duration are listed in the appendix to a Novartis FDA submission (NovartisNDA20-831 2005), but it is not clear how these relate to the 10 included studies that we found which were supported by Novartis.

Included studies

We included 22 studies on 8032 participants (6693 adults and 1339 children); the characteristics of these studies are fully described in Characteristics of included studies. Sponsorship of the studies is also listed in Table 1.

Eight studies enrolled adults over 18 years of age (Ekstrom 1998; Ekstrom 1998a; FitzGerald 1999; Hekking 1990; Kesten 1991; Molimard 2001; Steffensen 1995; van der Molen 1997), one study adults over 16 years of age (van Schayck 2002), eight studies adults and adolescents over 12 years of age (Bensch 2001; Busse 2004; Corren 2007; LaForce 2005; Noonan 2006; Pleskow 2003; SD-037-0344; Wolfe 2006) and five enrolled children from five up to 12 or 16 years of age (Bensch 2002; Kozlik-Feldmann 1996; Levy 2005; Von Berg 2003; Zimmerman 2004).

All of the studies were of 12 weeks duration with the exception of Bensch 2002 (52 weeks), FitzGerald 1999 (24 weeks), van der Molen 1997 (24 weeks) and Wolfe 2006 (16 weeks). This gives a weighted mean duration of 16 weeks for the 19 studies with a placebo arm.

Since many of the studies randomised patients to more than two treatment arms, we have reported the overall number of patients randomised to treatment categories under investigation. Adults and adolescents are combined, as separate data for adolescents were not provided in any of the studies. In total 8032 participants were randomised (6693 adults and 1339 children); of these 2146 adults and 483 children received placebo, 2483 adults and 568 children received formoterol up to 12 μg twice daily, 921 adults and 277 children received formoterol 24 μg twice daily, and 1143 adults and 11 children received regular salbutamol or terbutaline. Although Molimard 2001 was an open study that did not use a placebo group, we included it in the placebo comparisons as on-demand salbutamol was used by patients in both the formoterol and the comparison arm.

Doses of formoterol ranged from 4.5 μg to 24 μg twice daily in children and 6 μg to 24 μg twice daily in adults (but different delivery devices mean that the delivered dose may not be identical between studies, even if the nominal dose is the same). We have considered high-dose formoterol to be 24 μg twice daily.

Concurrent use of inhaled corticosteroids varied in the included studies from zero to 100%; Table 2 lists the inhaled steroid use by study. In almost all the studies at least half of the patients were taking inhaled corticosteroids at baseline.

Risk of bias in included studies

Figure 4 show a graphical representation of the domains of risk of bias across each study.

Allocation

Allocation concealment and sequence generation were often poorly reported in the included studies. However, the majority of the studies are supported by manufacturers of formoterol (see Table 1) and are therefore likely to have had appropriate protection against selection bias.

Blinding

All studies were double-blind except for Kozlik-Feldmann 1996, Molimard 2001 and van Schayck 2002, and of these only Molimard 2001 has contributed data to the primary outcomes.

Incomplete outcome data

The included studies generally had withdrawal rates of less than 20% with the exception of Corren 2007 (51% dropout in the placebo group), and Noonan 2006 (60% withdrawals in placebo arm and 51.2% in formoterol arm).

Selective reporting

Although paper reports of studies often did not include usable information on all-cause serious adverse events, it has proved possible to obtain serious adverse events information from 19 of the 22 included studies. This represents 100% of participants randomised to formoterol or placebo, but only 80% of participants compared to regular salbutamol or terbutaline. There are clear guidelines for the reporting of serious adverse events for industry (ICHE3 2007). These all have to be listed in detail for the regulatory authorities, but there is clearly no similar expectation from medical journals. Twenty-two studies of at least four weeks duration are listed in the appendix to a Novartis FDA submission (NovartisNDA20-831 2005), but it is not clear how many of these would fulfil the criteria for inclusion in this review, or how they related to the 10 Novartis studies included in this review (Table 1). It has not been possible to obtain further information from the sponsors.

Effects of interventions

See: Summary of findings for the main comparison Summary of findings - SAE all ages; Summary of findings 2 Summary of findings - mortality (all-cause); Summary of findings 3 Summary of findings - SAE children

Primary outcomes

All-cause mortality

Events were sparse in the trials and the presence or absence of mortality was not always reported in the paper publications.

Formoterol versus placebo

Data were available from 14 studies comparing formoterol (N = 3413) with placebo (N = 2050); this represents 80% of the randomised patients for this comparison. Three deaths occurred in these trials (two adults and one child), and overall there were three deaths on formoterol and none on placebo. The cause of death for one adult was not reported in SD-037-0344, there was one adult death from asthma in Pleskow 2003, and one child died of a subarachnoid haemorrhage in Von Berg 2003. The pooled odds ratio (OR) was not statistically significant (Peto OR 4.50; 95% confidence interval (CI) 0.41 to 49.49), see Figure 5. The confidence interval and point estimate are somewhat different for a Mantel-Haenszel OR 1.52 (95% CI 0.24 to 9.71) using fixed or random-effects models. This discrepancy is due to the continuity correction required for zero cells in the Mantel-Haenszel method. The point estimate of the pooled risk difference (RD) was an increase of 4 deaths per 10,000 treated with regular formoterol over 16 weeks, with a confidence interval from 20 fewer deaths to 28 more deaths per 10,000. There was no statistical heterogeneity in this outcome.

Formoterol versus salbutamol

Data were available from six studies comparing formoterol (N = 847) to salbutamol/albuterol (N = 571), representing 49% of the randomised patients for this comparison. Only two deaths occurred, both in Pleskow 2003; one in the formoterol arm (from asthma as reported above) and one in the salbutamol arm (from pancreatitis), and again the difference was not statistically significant. As only one study contributed to this outcome, we calculated no pooled OR.

Subgroup analyses

No subgroup analyses were possible for all-cause mortality as the data were too sparse.

Serious adverse events (SAEs) (non-fatal all-cause)

An example of the definition of a serious adverse event (SAE) is given in Pleskow 2003: “A serious adverse event was defined as any experience that was fatal or life-threatening, permanently disabling, requiring in-patient or prolonged hospitalisation, or was a congenital abnormality, cancer or drug overdose.” This is in line with the definition in Appendix 4, and we have assumed that this definition was used in the other trials (even though this was often not made explicit). In most cases the events were defined as serious because they were associated with hospital admission.

Formoterol versus placebo

Combined data from adults and children

Data were available on non-fatal SAEs for 19 studies comparing regular formoterol (N = 4017) to placebo (N = 2629); this represents all of the randomised patients from published studies for this comparison. The studies were largely on adults (N = 5311), but five trials were in children (N = 1335).

The overall result indicated an increased risk of SAEs with formoterol (Peto OR 1.57; 95% CI 1.06 to 2.31) with low heterogeneity (I² = 0%), Figure 6. SAEs were rare in the studies (occurring in 1.2% of patients on placebo over a weighted mean of 16 weeks), so the pooled risk difference is small at 0.007 (95% CI 0.0012 to 0.013) and using a weighted mean of the placebo arms, over a 16-week period there would need to be 149 patients treated (95% CI 66 to 1407) for one extra serious adverse event to occur; this number needed to treat was calculated by Visual Rx using the pooled odds ratio and baseline risk of 1%. This is illustrated in Figure 7 which shows that for every thousand patients treated over 16 weeks with formoterol there are an extra six patients who will suffer a serious adverse event, so that in comparison to 10 per thousand in the placebo group this rises to 16 per thousand on regular formoterol.

Sensitivity analysis by applying the Mantel-Haenszel method gave a very similar result, OR 1.57 (95% CI 1.05 to 2.37). Using a random-effects model (which is not recommended for rare events due to an increased risk of bias (Bradburn 2007)), the point estimate was similar but the confidence interval widened: OR 1.52 (95% CI 0.96 to 2.40). When we excluded the participants receiving high-dose formoterol (24 μg) arms from the analysis the confidence interval again widened (Peto OR 1.55; 95% CI 0.97 to 2.48)(Analysis 3.1).

Adults with non-fatal serious adverse events

When the adult data comparing regular formoterol (N = 3170) with placebo (N = 2137) were considered on their own, the results showed a smaller increase in risk which did not reach statistical significance (Peto OR 1.23; 95% CI 0.76 to 1.99)(Analysis 4.1). The Mantel-Haenszel method gave very similar results, OR 1.22 (95% CI 0.76 to 1.96).

Children with non-fatal serious adverse events

Although fewer children were studied, regular formoterol (N = 843) compared with placebo (N = 492), the separate results for children showed a larger increase in serious adverse events with regular formoterol (Peto OR 2.48; 95% CI 1.27 to 4.83)(Analysis 5.1). Again the Mantel-Haenszel method gave a very similar result, OR 2.92 (95% CI 1.26 to 6.74). There was low heterogeneity in this outcome (I² = 0%), and the result in children remained significant when the data on high-dose formoterol were excluded (Peto OR 2.52; 95% CI 1.15 to 5.51) and Mantel-Haenszel OR 2.59 (95% CI 1.08 to 6.19)(Analysis 6.1).

When the results in children are compared with adults using a test for interaction (Altman 2003), the increased risk in children relative to adults is a relative OR of 2.39 (95% CI 0.91 to 6.27) using the Mantel-Haenszel odds ratio, which is not statistically significant. The results are very similar when the Peto OR is compared between adults and children (relative OR 2.02; 95% CI 0.89 to 4.59).

High-dose formoterol versus lower doses

When the study arms using formoterol 24 μg twice daily were compared to those using lower doses (formoterol 12 μg twice daily), no significant difference was found in adults (Peto OR 1.35; 95% CI 0.64 to 2.85)(Analysis 7.1). The confidence interval for this result is too wide to rule out a difference in relation to dose, and although the three adult studies all used the same dry powder delivery device (Aerolizer), there is a high level of heterogeneity between studies (I² = 74%), which is unexplained.

Moreover the data from the Novartis database of published and unpublished placebo-controlled studies of at least four weeks duration has been included for comparison (Novartis 2005); the Novartis data show a significantly higher risk of asthma-related serious adverse events with formoterol 24 μg twice daily in comparison to 12 μg twice daily (Peto OR 2.16; 95% CI 1.13 to 4.11) and Mantel-Haenszel OR 2.08 (95% CI 1.11 to 3.89).

Formoterol versus salbutamol

Adults with non-fatal serious adverse events

In contrast, the results from nine studies in adults comparing regular formoterol (N = 1119) to salbutamol (N = 920), representing 80% of the randomised patients for this comparison, showed a non-significant reduction in the risk of SAEs (Peto OR 0.73; 95% CI 0.37 to 1.43)(Analysis 4.1). However, when the results from patients using high-dose formoterol were excluded, there was a significant reduction in risk for formoterol in comparison with salbutamol or terbutaline (Peto OR 0.41; 95% CI 0.19 to 0.90) and Mantel-Haenszel OR 0.40 (95% CI 0.17 to 0.94) (Analysis 3.1).

A test for interaction between the results comparing formoterol with placebo and salbutamol was not significant; the risk in trials against regular salbutamol compared to those against placebo using Mantel-Haenszel OR gives a relative OR of 0.46 (95% CI 0.21 to 1.01), whilst for the Peto method the relative OR is 0.59 (95% CI 0.26 to 1.36).

No studies were found comparing regular formoterol to regular salbutamol or terbutaline in children.

SAEs all-cause (fatal and non-fatal combined)

When fatal and non-fatal SAEs are considered together the findings are very similar to those for the non-fatal events, with a significant increase in risk with regular formoterol in comparison with placebo (Peto OR 1.61; 95% CI 1.09 to 2.37) and Mantel-Haenszel OR 1.63 (95% CI 1.08 to 2.44), but not in comparison with regular salbutamol (see Analysis 8.1).

Secondary outcomes

Mortality by cause of death

Asthma mortality and cardiovascular mortality

Only one death related to asthma was reported using formoterol 24 μg in Pleskow 2003, and one death in Von Berg 2003 from sub-arachnoid haemorrhage on formoterol 9 μg, so there are insufficient data to assess the impact of formoterol on disease-specific mortality. The third death in SD-037-0344 has no reported cause.

SAEs related to asthma and the cardiovascular system

When formoterol is compared to placebo, there is a significant increase in asthma-related serious adverse events on regular formoterol (Peto OR 1.99; 95% CI 1.12 to 3.53) and Mantel-Haenszel OR 1.90 (95% CI 1.03 to 3.48)(Analysis 11.1). This finding is also apparent from the Novartis integrated database (Novartis 2005), which is concordant with the results of the 10 Novartis studies included in the review as shown in Analysis 12.1.

In comparison with regular salbutamol there is a decrease in asthma-related serious adverse events on regular formoterol but this is not significant (Peto OR 0.74; 95% CI 0.29 to 1.88) and Mantel-Haenszel OR 0.72 (95% CI 0.32 to 1.76)(Analysis 11.1). There is a significant increase in hospital admissions for asthma when regular formoterol is compared to placebo: Peto OR 3.28; 95% CI 1.65 to 6.52 and Mantel-Haenszel OR 4.28; 95% CI 1.60 to 11.46 (Analysis 13.1). For this outcome we assumed that all the patients with asthma-related SAE documented in the FDA submission were admitted to hospital; in Pleskow 2003 this is clearly stated to be the case for two patients on placebo and five on formoterol 24 μg, but is not clearly reported for the single patient on formoterol 12 μg (Mann 2003). No significant difference was seen when regular formoterol is compared to regular salbutamol. Very few events relating to the cardiovascular system were reported, so although the direction of effect is in favour of formoterol the confidence intervals in comparison to placebo and salbutamol are very wide (Analysis 14.1).

Impact of inhaled corticosteroids

It has not been possible to assess whether inhaled corticosteroids have an impact on SAEs with regular formoterol from the included studies, as this would require individual patient data relating inhaled corticosteroid usage to those patients who suffered the events, and these data are not available. However, data are presented in Novartis 2005 in which patients on inhaled corticosteroids are compared with those who are not, for asthma-related serious adverse events. Although the increased risk is larger without inhaled corticosteroids (Peto OR 3.22; 95% CI 1.17 to 8.89) and Mantel Haenszel OR 7.31 (95% CI 0.97 to 55.06), than with inhaled corticosteroids (Peto OR 2.58; 95% CI 1.21 to 5.49) and Mantel-Haenszel OR 3.47 (95% CI 1.21 to 9.97), the confidence intervals are wide and there is no statistically significant interaction between the increased risk and use of inhaled corticosteroids. Moreover there remains a significant three-fold increase in odds for patients who were taking inhaled corticosteroids, and this is maintained when the data on formoterol 24 μg twice daily are excluded (Peto OR 2.76; 95% CI 1.06 to 7.15) and Mantel-Haenszel OR 3.11 (95% CI 1.01 to 9.55)(Analysis 15.1).

ADDITIONAL SUMMARY OF FINDINGS

Regular formoterol versus placebo or salbutamol for chronic asthma.

Patient or population: patients with chronic asthma
Settings: community
Intervention: regular formoterol
Comparison: placebo or salbutamol
Outcomes			Illustrative comparative risks^* (95% CI)		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
			Assumed risk	Corresponding risk
			Placebo or salbutamol	Regular formoterol
Mortality (all-cause) - formoterol versus placebo			Medium-risk population		OR 1.52 (0.24 to 9.71)	5463 (14)	⊕⊕⊕○ moderate¹
(follow-up: weeks)	mean	16	0 per 1000	0 per 1000 (0 to 0)
Mortality (all-cause)-formoterol versus salbutamol			Medium-risk population		OR 0.5 (0.03 to 8.05)	1418 (6)	⊕⊕⊕○ moderate¹
(follow-up: weeks)	mean	13	0 per 1000	0 per 1000 (0 to 0)

Methods	A randomised, double-blind, double-dummy, placebo-controlled, multicentre parallel-group study over 12 weeks at 26 trial sites in the USA. 2-week, single-blind, placebo lead-in period
Participants	Population: 541 adolescents and adults (12 to 75) years with mild to moderate persistent asthma Baseline characteristics: mean age 35.5 years. FEV₁ 66% predicted. Concomitant inhaled corticosteroids used by 51% of participants. Inclusion criteria: mild to moderate persistent asthma requiring daily use of an inhaled B₂-selective adrenergic agent (short- or long-acting). FEV₁ between 40% and 80% predicted after an 8-hour period of abstinence from short-acting B₂-adrenergic agonist use. Bronchodilator reversibility at least a 15% increase in FEV₁ within 30 minutes after inhalation of 180 μg of albuterol delivered via MDI. Co-interventions permitted: ICS 51%, slow release theophylline 17% Exclusion criteria: URTI, hospitalisation/asthma exacerbation < 4 weeks, serious illness
Interventions	Formoterol 12/24 μg BD Albuterol 180 μg QDS Placebo QDS Formoterol delivered by Aerolizer (dry powder) and albuterol by MDI
Outcomes	Primary outcome: FEV₁. Safety was evaluated by adverse event reports. Paper report: “Adverse events, whether or not trial drug related, were reported by 68% of patients on formoterol 12 mcg, 76% on formoterol 24 mcg, 70% on albuterol and 71% on placebo. No deaths occurred during the study.” No SAE data reported but the numbers match FDA submission for study 040 (http://www.fda.gov/cder/foi/nda/2001/20831-ForadiLmedr_Pl.pdf). This lists all (asthma-related in brackets) SAEs as: 1 (0) on formoterol 12 μg, 5 (4) on formoterol 24 μg, 2 (2) on albuterol and 1 (0) on placebo
Notes	Supported by Novartis
*Risk of bias*
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not reported
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding (performance bias and detection bias) Adverse Events	Low risk	Double-blind, double-dummy. Matching capsules and devices
Incomplete outcome data (attrition bias) Adverse Events	Unclear risk	All patients included in the safety analysis
Selective reporting (reporting bias)	Low risk	No SAE data in paper publication, but FDA submission provided SAE information

Methods	A randomised, double-blind, placebo-controlled, multinational, multicentre, paediatric study over 52 weeks at 42 centres. Initial run-in period of 2 weeks
Participants	Population: 518 children (5 to 12) years with persistent asthma on preventer therapy Baseline characteristics: mean age 9 years. FEV₁ 71% predicted. At least 69% were taking ICS, with 26% on sodium cromoglycate and 5% nedocromil sodium. Inclusion criteria: persistent asthma diagnosed by ATS criteria. Sodium cromoglycate, nedocromil sodium and/or ICS for at least 4 weeks before entry in the study, but still required daily use of inhaled salbutamol (albutamol) to control symptoms. Baseline FEV₁ ranged from 50% to 85% of predicted and increased by 15% or more within 30 minutes after inhaling 200 μg salbutamol (180 μg albuterol in the United States) delivered by a metered-dose inhaler. Co-interventions: 70% on ICS, 30% on cromones (all stable dosage) Exclusion criteria: unstable asthma, URTI, OS course or exacerbation asthma within 4 weeks, QT interval on ECG > 0.46 sec
Interventions	Formoterol 12 μg or 24 μg BD Placebo BD delivery was Aerolizer
Outcomes	Primary outcome: FEV₁ (area under curve) Paper reports: “There were no deaths in this study.” Table 4 in the paper reports SAE data but not in a way that allows all-cause SAE data to be extracted. Full SAE data found on FDA website
Notes	Supported by Novartis
*Risk of bias*
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not reported
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding (performance bias and detection bias) Adverse Events	Low risk	Double-blind
Incomplete outcome data (attrition bias) Adverse Events	Low risk	407/518 (78%) completed the study
Selective reporting (reporting bias)	Low risk	All-cause SAE data not extractable from published paper, but present in FDA web report

Methods	Randomised, double-blind, double-dummy, placebo-controlled, parallel-group study over 12 weeks in outpatient clinics at 18 US centres. Run-in 2 weeks single-blind placebo
Participants	239 adolescents and adults (13 to 85) years with persistent asthma. (9 patients were aged less than 18 years old) Baseline characteristics: mean age 38 years. FEV₁ 65% predicted. Concomitant inhaled corticosteroids used by 64% of participants. Inclusion criteria: persistent asthma, requiring regular or on-demand bronchodilators, FEV₁ % predicted at least 40%, bronchodilator reversibility by either an increase of at least 15% in FEV₁ over baseline or increases of at least 12% and 200 mL in FEV₁ over baseline within 30 minutes of inhaling albuterol 180 to 360 μg (2 to 4 puffs) via pMDI. Co-interventions permitted: ICS, cromolyn, montelukast, theophyllines, intra nasal ICS Exclusion criteria: RT infection, use of OCS, parenteral CS or hospitalisation for asthma due to exacerbation < 1 month, use of unstable anti-inflammatory regime, corrected QT interval > 460 ms, current smoker, ex smoker > 10 Pack years, serious medical condition, pregnancy, lactation, use oforal beta blocker, non-potassium sparing diuretic, anti-arrhythmic, tricyclic antidepressant, MAOI, NSAIDS
Interventions	Formoterol 10μg BD Placebo BD Albuterol 180 μg QDS Delivery was DPI for formoterol and MDI for albuterol (salbutamol)
Outcomes	The primary efficacy variable was the 12-hour AUC of FEV₁ after 12 weeks’ treatment No deaths in the study. Non-fatal SAEs: formoterol 2 (1 asthma exacerbation and 1 basal cell carcinoma). Albuterol 3 (1 eating disorder, 1 renal calculi and 1 pneumonia). Placebo 0. No events were considered to be related to study medication Any AE: formoterol 43, albuterol 47 and placebo 48
Notes	Funding Novartis
*Risk of bias*
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	1:1:1 ratio using computer-generated list of random numbers
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding (performance bias and detection bias) Adverse Events	Low risk	Double-dummy matched devices
Incomplete outcome data (attrition bias) Adverse Events	Low risk	209/239 (87%) completed the study; SAE reported for both those who withdrew and those who continued in the study
Selective reporting (reporting bias)	Low risk	Serious adverse events (fatal and non-fatal) documented in the paper publication

Methods	A randomised, parallel-group, single-centre study over 12 weeks in Germany
Participants	Population: 22 children with clinically stable asthma Baseline characteristics: mean age 10 years. Before treatment, all values for FEV₁ were found beyond 100% of that predicted. Inclusion criteria: mild asthma, requiring no preventive therapy. The patients needed no further therapy, i.e. with corticosteroids or theophyllines Exclusion criteria: requiring additional therapy for asthma
Interventions	Formoterol 24 μg BD Salbutamol 200 μg QDS Delivery was not reported
Outcomes	FEV₁, FVC, histamine provocation, beta-receptor binding sites on mononuclear leucocytes, adverse events “No side effects due to beta₂-agonist therapy were seen”.
Notes	No indication of sponsorship for this study
*Risk of bias*
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information
Allocation concealment (selection bias)	Unclear risk	No information
Blinding (performance bias and detection bias) Adverse Events	High risk	Open
Incomplete outcome data (attrition bias) Adverse Events	Low risk	All patients completed the study
Selective reporting (reporting bias)	High risk	No serious adverse event data documented

Methods	Randomised, double-blind, multicentre, parallel-group study over 12 weeks at 22 sites in the US. Run-in 2 weeks
Participants	Population: 265 adolescents and adults (13 to 81) years with persistent asthma Baseline characteristics: mean age 37 years. FEV₁ 68% predicted. Concomitant inhaled corticosteroids used by 67% of participants. Inclusion criteria: persistent asthma. FEV₁ % predicted at least 40%, bronchodilator reversibility by an increase of at least 15% in FEV₁ over baseline within 30 minutes of inhaling albuterol (180 to 320 μg) (still eligible if the increase in FEV₁ was 12% or greater and at least 0.2 L over baseline) Exclusion criteria: pregnant/lactating women; smoking history > 10 pack-years; history of malignancy; RTI or hospitalisation with exacerbation in previous month; treatment with systemic steroids
Interventions	Formoterol 10 μg BID Placebo Salbutamol 180 μg QID Delivery was DPI (pMDI for salbutamol)
Outcomes	The primary efficacy variable was 12-hour AUC of FEV₁ after 12 weeks’ treatment. “There were no deaths in the study. A total of three patients experienced serious adverse events and were withdrawn from the study.” Formoterol 1 small cell lung cancer, albuterol 1 increase in heart rate and 1 coronary artery stenosis
Notes	Sponsored by Novartis
*Risk of bias*
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	1:1:1 using a computer-generated sequence
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding (performance bias and detection bias) Adverse Events	Low risk	Double-blind, double-dummy
Incomplete outcome data (attrition bias) Adverse Events	Low risk	235/265 (89%) completed the study
Selective reporting (reporting bias)	Low risk	Deaths and serious adverse events reported in the paper

Methods	Randomised, double-blind, placebo-controlled, multicentre, parallel-group study over 12 weeks in 22 US centres. Run-in 2 weeks single-blind placebo
Participants	Population: 249 children (5 to 13) years with mild to moderate persistent asthma Baseline characteristics: mean age 9 years. FEV₁ 75% predicted. Concomitant inhaled corticosteroids used by 72% of participants. Inclusion criteria: mild to moderate persistent asthma. FEV₁ % predicted at least 50%, bronchodilator reversibility by an increase of at least 15% in FEV₁ over baseline within 30 min of inhaling albuterol 180 μg via MDI, regular use of an on-demand bronchodilator Exclusion criteria: history of respiratory tract infection, hospitalisation due to an asthma exacerbation in the month prior to visit 1, clinically significant medical condition, history of allergy to any inhaled medications, QTc interval > 0.46 seconds, or used parenteral or oral corticosteroids in the month prior to visit 1
Interventions	Formoterol 10 μg BD Placebo BD Delivery was DPI
Outcomes	The primary efficacy variable was the 12-hour AUC of FEV₁ at 12 weeks “One formoterol-treated patient experienced a serious adverse event (asthma exacerbation), which led to hospitalisation and discontinuation from the study. The event was not suspected as study-drug related by the investigator.” No web report found
Notes	Sponsored by Novartis
*Risk of bias*
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information
Allocation concealment (selection bias)	Unclear risk	No information
Blinding (performance bias and detection bias) Adverse Events	Low risk	Double-blind, matched placebo
Incomplete outcome data (attrition bias) Adverse Events	Low risk	227/249 (91%) completed the study
Selective reporting (reporting bias)	Low risk	Serious adverse events reported in the paper

Methods	Randomised, outpatients, open, parallel-group study over 3 months from February 1998 to March 1999 in France. Run-in 2 to 3 weeks
Participants	Population: 259 adults (18+) years with moderate persistent asthma Baseline characteristics: mean age 39 years. FEV₁ 73% predicted. Concomitant inhaled corticosteroids used by 100% of participants. Inclusion criteria: moderate persistent asthma, daily treatment with an inhaled corticosteroid (the same product at a stable dose for at least 1 month prior to the first visit) and daily treatment with inhaled bronchodilators (taken regularly or on demand). FEV₁ % predicted at least 60%, bronchodilator reversibility by an increase of at least 10% in FEV₁ over baseline to be documented at the first visit or 3 months prior. Exclusion criteria: known hypersensitivity to sympathic amines or to lactose, pregnancy or breast-feeding, women of childbearing potential who did not use a reliable contraceptive method, significant change in the regular asthma medication, asthma exacerbation or respiratory tract infection in the month prior to the first visit, incapacity to use a metered-dose inhaler correctly or to complete the patient diary
Interventions	Formoterol dry-powder capsule 12 μg BD Salbutamol on-demand via MDI
Outcomes	The primary efficacy variable was the mean change in morning predose PEF for the entire treatment period “No drug-related serious AE was reported”. No report in paper of mortality or all-cause SAE, but authors have provided additional information: “If we consider all-cause SAE, no SAE was reported in the on-demand salbutamol group and two SAEs were reported in the formoterol group (not suspected to be drug-related) : one case of hospitalisation due to a malignant tumour of breast and one case of hospitalisation due to an uterine haemorrhage. No death was reported in this study.”
Notes	Supported by Novartis
*Risk of bias*
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information
Allocation concealment (selection bias)	Low risk	Centralised phone randomisation was used to avoid inclusion bias
Blinding (performance bias and detection bias) Adverse Events	High risk	Open
Incomplete outcome data (attrition bias) Adverse Events	Low risk	229/259 (88%) completed the study
Selective reporting (reporting bias)	Low risk	All-cause SAE information provided by authors

Methods	This represents results from placebo-controlled trials of at least 4 weeks duration from the Integrated Database of Novartis Clinical Trials
Participants
Interventions	Formoterol 24 μg bd via Aerolizer or Certihaler Formoterol 12 μg bd via Aerolizer or Certihaler Placebo Albuterol
Outcomes	Asthma-related SAE data
Notes	These data have been entered as a separate subgroup and not combined with other studies as 5 of the 22 studies are already included in this review, but no separate information has yet been received in relation to the other 17 studies
*Risk of bias*
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No individual study data available
Allocation concealment (selection bias)	Unclear risk	No individual study data available
Blinding (performance bias and detection bias) Adverse Events	Unclear risk	No individual study data available
Incomplete outcome data (attrition bias) Adverse Events	Unclear risk	No individual study data available
Selective reporting (reporting bias)	Unclear risk	No individual study data available

Methods	Randomised study over 12 weeks at a lung function laboratory in The Netherlands. Run-in 8 weeks
Participants	Population: 162 adolescents and adults (16 to 60) years Baseline characteristics: mean age 35 years, FEV₁ 86% predicted. Concomitant inhaled corticosteroids used by 95% of participants Inclusion criteria: history of bronchial symptoms or a clinical diagnosis of asthma, FEV₁ % predicted at least 50%, and either PC₂₀ on histamine < 8 mg/mL⁻¹ or bronchodilator reversibility by an increase of at least 15% in FEV₁ over baseline after inhalation of 800 μg salbutamol Exclusion criteria: not reported At the start of the 8-week washout period, patients to cease all their pulmonary medication (inhaled corticosteroids, cromoglycates, bronchodilators) and to use only rescue medication on demand
Interventions	Salbutamol 100 μg BD Formoterol 12 μg BD Placebo Delivery was MDI
Outcomes	The effects of bronchodilator use on the perception of airway obstruction There is no indication that data on adverse events were collected. Author has confirmed that no serious adverse events were reported in this study
Notes	No support declared
*Risk of bias*
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No information
Allocation concealment (selection bias)	Unclear risk	No information
Blinding (performance bias and detection bias) Adverse Events	High risk	Open?
Incomplete outcome data (attrition bias) Adverse Events	Low risk	128/162 (79%) completed the study
Selective reporting (reporting bias)	Low risk	No information on SAEs in paper but data provided by author

Study	Reason for exclusion
Adler 2006	2-week study
Akpinarli 1999	6-week treatment periods
Ankerst 2003	Short-term high-dose tolerance study
Arvidsson 1989	2-week study
Aziz 1998	Volunteers (not asthmatics)
Bauer 1995	Single dose
Behling 1999	3-week study
Boner 1994	Single dose
Bousquet 2005	Single dose
Bousquet 2005a	Single dose
Brambilla 2003	4-week study
Bronsky 2002	Single dose
Bronsky 2004	Single dose
Brusasco 2002	Single dose
Burgess 1998	Single dose
Cazzola 2002	Single dose
Ceylan 2004	Formoterol versus montelukast
Chetta 1993	Single dose
Cheung 2006	3-week study
Chuchalin 2002	Formoterol randomised with budesonide
Chuchalin 2005	1-week study
Chuchalin 2005a	Formoterol randomised with budesonide
Chung 1995	Cross-over design
Dahl 2004	Device comparison
Daugbjerg 1996	Single dose
Dey 2005	Cross-over design
Dietrich 2006	Device comparison
Dubakiene 2006	Device comparison
Dusser 2005	HFA versus DPI delivery comparison
Ericsson 2006	Randomised with budesonide
Eryonucu 2005	Single dose
Ferrari 2000	Single dose
Fitoussi 2002	3-day study
Gessner 2003	8-week study
Graff-Lonnevig 1990	Cross-over design
Green 2006	Cross-over design
Happonen 2009	Cross-over study comparing formoterol Easyhaler with Foradil Aerolizer
Hedenstrom 1992	Cross-over design
Hermansen 2006	EIB single doses
Houghton 2004	Single dose
Ind 2002	Formoterol used as reliever
Jain 2004	Formoterol used as reliever
Jenkins 2005	6-week treatment periods
Kamenov 2007	Formoterol - HFA metered-dose inhaler compared to formoterol dry powder inhaler
Kesten 1992	Not RCT
Kohler 2003	Acute asthma
Kruse 2005	Cross-over design
Lebecque 1994	Single dose
Lebecque 1994a	Single dose
Lee-Wong 2008	Acute asthma
Lemaigre 2006	Single dose
Lotvall 1997	Single dose
Lotvall 2005	Formoterol as reliever
Lotvall 2008	Single dose
Lyseng-Williamson 2003	Not RCT
Maesen 1992	Short-term study over 3 days
Maesen 1992a	Delivery device and dose
Malo 1990	Single dose
Malolepszy 2001	Formoterol versus theophylline
Malolepszy 2002	Acute asthma
Mann 2003	Overview of 3 included studies: Bensch 2001; Bensch 2002; Pleskow 2003
Marzo 2000	Healthy volunteers
Matthys 2004	3-week study
Midgren 1992	4-week study
Molimard 2005	Cumulative dose study
Najafizadeh 2007	Acute asthma
Nandeuil 2006	8-day study
Newnham 1994	4-week study
Newnham 1995	4-week study
Novartis 2005b	Both arms received formoterol maintenance
Otto-Knapp 2008	4-week study
Patessio 1991	EIB
Pauwels 2003	Formoterol as reliever
Pearlman 2002	Overview of other studies
Pohunek 2004	Single dose
Price 2002	Randomised together with budesonide
Price 2008	Overview of safety data from AstraZeneca-sponsored trials of duration of at least 4 weeks
Randell 2005	Device comparison
Richter 2007	Dosage comparison
Rico-Mendez 1999	Study included patients with chronic bronchitis
Ringdal 1998	Single dose
Rosenborg 2000	Single dose
Rosenborg 2002	Short-term dose-response study
Rosenborg 2002a	Short-term dose-response study
Rubinfeld 2006	Acute asthma
Schlimmer 2002	Acute asthma
Sprogoe 1992	6-week study
Stahl 2003	Formoterol as reliever
Stelmach 2008	Study on exercise-induced bronchospasm
Tattersfield 1999	Description of exacerbations in FACET study
Tattersfield 2001	As-needed formoterol
Totterman 1998	3-day study
van den Berg 1995	8-day study
van der Woude 2004	Single dose
van Veen 2003	2-week cross-over
Verini 1998	5-day study
Villa 2002	As-needed formoterol
Villa 2002a	As-needed formoterol
Vilsvik 2001	EIB
Wallin 1999	8-week study
Wegener 1992	Single dose
Wong 1992	Single dose
Yasuo 1984	4-week study
Yasuo 1984a	Single dose
Yasuo 1984b	Single dose
Yates 1995	2-week cross-over
Yurdakul 2002	No placebo or SABA control

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Overall results	14		Peto Odds Ratio (Peto, Fixed, 95% CI)	Subtotals only
1.1 Formoterol versus placebo	14	5463	Peto Odds Ratio (Peto, Fixed, 95% CI)	4.50 [0.41, 49.49]
1.2 Formoterol versus salbutamol	6	1418	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.47 [0.02, 8.98]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Formoterol versus placebo or salbutamol	19		Peto Odds Ratio (Peto, Fixed, 95% CI)	Subtotals only
1.1 Formoterol versus placebo	19	6646	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.57 [1.06, 2.31]
1.2 Formoterol versus salbutamol	9	2119	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.73 [0.37, 1.43]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Formoterol versus placebo or salbutamol	18		Peto Odds Ratio (Peto, Fixed, 95% CI)	Subtotals only
1.1 Formoterol versus placebo	18	5438	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.55 [0.97, 2.48]
1.2 Formoterol versus salbutamol	9	1848	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.41 [0.19, 0.90]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Formoterol versus placebo or salbutamol	15		Peto Odds Ratio (Peto, Fixed, 95% CI)	Subtotals only
1.1 Formoterol versus placebo	15	5311	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.23 [0.76, 1.99]
1.2 Formoterol versus salbutamol	9	2119	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.73 [0.37, 1.43]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Serious adverse events	5		Peto Odds Ratio (Peto, Fixed, 95% CI)	Subtotals only
1.1 Adults (all-cause)	3	1598	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.35 [0.64, 2.85]
1.2 Children (all-cause)	1	342	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.20 [0.52, 2.74]
1.3 Asthma SAE	1	3104	Peto Odds Ratio (Peto, Fixed, 95% CI)	2.16 [1.13, 4.11]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Patients with at least one asthma-related SAE	1		Peto Odds Ratio (Peto, Fixed, 95% CI)	Subtotals only
1.1 Patients taking inhaled corticosteroids (all doses of formoterol))	1	3807	Peto Odds Ratio (Peto, Fixed, 95% CI)	2.58 [1.21, 5.49]
1.2 Patients not taking inhaled corticosteroids (all doses of formoterol)	1	1824	Peto Odds Ratio (Peto, Fixed, 95% CI)	3.22 [1.17, 8.89]
1.3 Patients taking inhaled corticosteroids (formoterol 12 mcg twice daily)	1	2708	Peto Odds Ratio (Peto, Fixed, 95% CI)	2.76 [1.06, 7.15]
1.4 Patients not taking inhaled corticosteroids (formoterol 12 mcg twice daily)	1	1103	Peto Odds Ratio (Peto, Fixed, 95% CI)	3.71 [0.75, 18.48]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Formoterol versus placebo or salbutamol	8		Odds Ratio (M-H, Fixed, 95% CI)	Subtotals only
1.1 Formoterol versus placebo	8	3743	Odds Ratio (M-H, Fixed, 95% CI)	1.09 [0.94, 1.26]
1.2 Formoterol versus salbutamol	4	1151	Odds Ratio (M-H, Fixed, 95% CI)	0.98 [0.76, 1.26]

Study ID	Sponsor
Bensch 2001	Novartis
Bensch 2002	Novartis
Busse 2004	Novartis
Corren 2007	AstraZeneca
Ekstrom 1998	AstraZeneca
Ekstrom 1998a	AstraZeneca
FitzGerald 1999	Novartis
Hekking 1990	Not reported
Kesten 1991	Novartis
Kozlik-Feldmann 1996	Not reported
LaForce 2005	Novartis
Levy 2005	Novartis
Molimard 2001	Novartis
Noonan 2006	AstraZeneca
Pleskow 2003	Novartis
SD-037-0344	AstraZeneca
Steffensen 1995	Novartis
van der Molen 1997	AstraZeneca
van Schayck 2002	Not reported
Von Berg 2003	AstraZeneca
Wolfe 2006	Novartis
Zimmerman 2004	AstraZeneca

Drug	Intrinsic efficacy (%)
Isoprenaline, adrenaline	100
Fenoterol	42
Formoterol	20
Salbutamol	4.9
Salmeterol	<2

Database	Frequency of search
MEDLINE (Ovid)	Weekly
EMBASE (Ovid)	Weekly
CENTRAL (The Cochrane Library)	Quarterly
PSYCINFO (Ovid)	Monthly
CINAHL (EBSCO)	Monthly
AMED (EBSCO)	Monthly

Conference	Years searched
American Academy of Allergy, Asthma and Immunology (AAAAI)	2001 onwards
American Thoracic Society (ATS)	2001 onwards
Asia Pacific Society of Respirology (APSR)	2004 onwards
British Thoracic Society Winter Meeting (BTS)	2000 onwards
Chest Meeting	2003 onwards
European Respiratory Society (ERS)	1992, 1994, 2000 onwards
International Primary Care Respiratory Group Congress (IPCRG)	2002 onwards
Thoracic Society of Australia and New Zealand (TSANZ)	1999 onwards

Date	Event	Description
5 January 2012	New search has been performed	No new studies found. Minor edits made and plain language summary revised
5 January 2012	New citation required but conclusions have not changed	New search carried out in January 2012 but no new studies included
10 November 2008	Amended	The ‘Summary of findings’ tables have been reordered. Contents are unchanged. An additional reference has also been added for Corren 2007. The Primary Analysis has been changed to Peto Odds Ratio.

Patient or population: patients with chronic asthma
Settings: community
Intervention: regular formoterol versus placebo or salbutamol
Outcomes	Illustrative comparative risks^* (95% CI)		Relative effect (95% CI)	No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Control	Regular formoterol versus placebo or salbutamol
SAEs - formoterol versus placebo	Medium-risk population		OR 1.57 (1.05 to 2.37)	6646 (19)	⊕⊕⊕⊕ high
(follow-up: mean 16 weeks)	10 per 1000	16 per 1000 (10 to 23)	OR 1.57 (1.05 to 2.37)	6646 (19)	⊕⊕⊕⊕ high
SAEs - formoterol versus salbutamol	Medium-risk population		OR 0.72 (0.37 to 1.43)	2119 (9)	⊕⊕⊕○ moderate¹
(follow-up: mean 13 weeks)	23 per 1000	17 per 1000 (9 to 33)	OR 0.72 (0.37 to 1.43)	2119 (9)	⊕⊕⊕○ moderate¹

PERMALINK

Regular treatment with formoterol for chronic asthma: serious adverse events

Christopher J Cates

Matthew J Cates

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors’ conclusions

BACKGROUND

Description of the condition

Airway obstruction

Airway hyper responsiveness

Inflammation

Description of the intervention

Beta2-agonists and mortality: an historical perspective

Time trend data and case-control studies

Figure 1. Changes in asthma mortality (5 to 34 age group) in three countries in relation to the introduction of isoprenaline forte in the UK and New Zealand and of fenoterol in New Zealand. (From Blauw 1995. With permission from the Lancet).

Figure 2. Inhaled fenoterol market share and annual asthma mortality in New Zealand in persons aged 5 to 34.

The introduction of long-acting beta2-agonists

How the intervention might work

Why it is important to do this review

OBJECTIVES

METHODS

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

RESULTS

Description of studies

Results of the search

Figure 3. Study flow diagram.

Included studies

Risk of bias in included studies

Figure 4. Methodological quality summary: review authors’ judgements about each methodological quality item for each included study.

Allocation

Blinding

Incomplete outcome data

Selective reporting

Effects of interventions

Primary outcomes

All-cause mortality

Formoterol versus placebo

Figure 5. Forest plot of comparison: 1 All-cause mortality, outcome: 1.1 Overall results.

Formoterol versus salbutamol

Subgroup analyses

Serious adverse events (SAEs) (non-fatal all-cause)

Formoterol versus placebo

Combined data from adults and children

Figure 6. Forest plot of comparison: 2 Adults and children non-fatal serious adverse events, outcome: 2.1 Formoterol versus placebo or salbutamol.

Figure 7. Serious adverse events with regular formoterol compared to placebo. In the control group 12 people out of 1000 had serious adverse events over 16 weeks, compared to 19 (95% CI 13 to 27) out of 1000 for the active treatment group.

Adults with non-fatal serious adverse events

Children with non-fatal serious adverse events

High-dose formoterol versus lower doses

Formoterol versus salbutamol

Adults with non-fatal serious adverse events

SAEs all-cause (fatal and non-fatal combined)

Secondary outcomes

Mortality by cause of death

Asthma mortality and cardiovascular mortality

Beta₂-agonists and mortality: an historical perspective

The introduction of long-acting beta₂-agonists

Appendix 1. Pharmacology of beta₂-agonists