0
Commentary |

US Food and Drug Administration-Mandated Trials of Long-Acting β-Agonists Safety in AsthmaLong-acting β-Agonists Safety Trials in Asthma: Will We Know the Answer? FREE TO VIEW

Samy Suissa, PhD; Amnon Ariel, MD, FCCP
Author and Funding Information

From the Center for Clinical Epidemiology (Dr Suissa), Lady Davis Institute–Jewish General Hospital, and the Departments of Epidemiology and Biostatistics and Medicine, McGill University, Montreal, QC, Canada; and Lung Unit (Dr Ariel), Emek Medical Center, Afula, Israel.

Correspondence to: Prof Samy Suissa, Centre for Clinical Epidemiology, Jewish General Hospital, 3755 Cote Ste-Catherine, Montreal, QC, H3T 1E2, Canada; e-mail: samy.suissa@mcgill.ca


Reproduction of this article is prohibited without written permission from the American College of Chest Physicians. See online for more details.


Chest. 2013;143(5):1208-1213. doi:10.1378/chest.12-2881
Text Size: A A A
Published online

For 2 decades, long-acting β-agonists (LABAs) have been associated with increased asthma-related death risks in several randomized trials, even when added to inhaled corticosteroids (ICSs). In reaction, the US Food and Drug Administration (FDA) recently mandated that the manufacturers of LABAs conduct five large, noninferiority, randomized trials of the LABA+ICS combination in 53,000 patients with asthma. Three methodologic issues in these trials could lead to masking of or falsely detecting elevated risks. First, the effect of LABA discontinuation among the many patients already using these drugs at enrollment can result in an underestimation of the relative risk by a factor of around 20%. This effect will bias downward the upper bound of the resulting CI away from the preset noninferiority margin of 2.0 for the relative risk, artificially making it more difficult to detect a risk increase. Second, the composite asthma outcome will be dominated by asthma hospitalization, possibly dwarfing an increased risk of asthma-related death, with differences as wide as seven deaths under the LABA+ICS combination vs one death under ICS alone remaining statistically uncertain. Finally, because of the multiple identical trials being requested from the different manufacturers of LABAs, even if each trial is powered at 90%, there is a 41% likelihood that at least one of the trials will not rule out a risk increase when, in truth, there is no risk increase. In view of these impediments, the FDA should preempt such complexities by establishing decision rules regarding the interpretation of the results from these momentous safety trials before their completion, expected in 2017.

Figures in this Article

Recently updated, international asthma guidelines state that the addition of long-acting β-agonists (LABA) to a daily regimen of inhaled corticosteroids (ICS) improves several asthma end points and achieves clinical control of asthma.1 Paradoxically, despite its therapeutic efficacy, chronic LABA use has been implicated in asthma mortality for the last 20 years, with large, randomized controlled trials generating important signals of increased incidence of asthma death.2,3 While the evidence is now convincing regarding this elevated risk in the absence of concurrent ICSs, the question of a possible risk when LABAs are added to regular ICS therapy remains controversial.

The US Food and Drug Administration (FDA) performed a meta-analysis of randomized trials involving > 36,000 patients with asthma taking ICS and comparing LABA use with no-LABA use on the risk of the composite outcome of asthma-related death, intubation, or hospitalization.4,5 The incidence of the asthma composite outcome was significantly increased (rate difference: 6.1 events per 1,000 patient-years; 95% CI, 0.9-11.4) with LABA use when the concurrent ICS was not part of the randomized treatment, but in the trials where ICS use was mandatory at the same dose, the incidence of the asthma composite outcome was not increased with LABA use (rate difference: 0.4 events per 1,000 patient-years; 95% CI, −3.8-4.6).5 Nevertheless, asthma mortality remains a concern.

A pooled analysis of randomized trials of formoterol involving close to 60,000 patients with asthma taking ICS reported seven asthma-related deaths among 46,000 patients taking formoterol and one in 14,000 patients not randomized to formoterol (rate ratio [RR], 2.3; 95% CI, 0.3-105).6 A meta-analysis of 57 trials involving about 35,000 patients with asthma using ICS found three asthma-related deaths among those taking a LABA and none in patients not randomized to a LABA (RR, 3.0; 95% CI, 0.5-17.6).7 These two meta-analyses were limited by the unknown extent of the concurrent ICS being mandated or not as part of the trials and whether the ICS dose was different between the LABA users and nonusers. While the risk increases were not statistically significant, the consistency and trend of the effect make this issue a continuing concern for patients, clinicians, and regulators.8

In response, the FDA recently mandated that the manufacturers of LABAs conduct five large, randomized trials in patients with asthma.9 These trials, which will assess the effect of adding a LABA to ICS on major asthma outcomes, began in 2011, with the results expected in 2017. While the “FDA believes that these clinical trials will … clarify the safety risk associated with LABAs when used concurrently with inhaled corticosteroids,”9 a recent paper questioned their relevance.10

In this article, we show that three important methodologic aspects inherent to the study design of these trials will affect their results and interpretation. We describe these issues and quantify their effects on the expected results.

Four of the randomized trials, identically designed, each involve 11,700 adult and adolescent patients with asthma per study, for a total of 46,800 patients, while a smaller trial involves 6,200 children.9 Each trial compares the combination of a LABA with an ICS (LABA+ICS) vs monotherapy with the same ICS at the same dose. The primary end point is a composite of serious asthma outcomes, including asthma-related death, intubation, or hospitalization, observed over a 6-month follow-up. These trials are designed as noninferiority trials, which test whether the patients in the LABA+ICS group are “no worse” than those in the ICS-only group with respect to the risk of the serious asthma outcomes.11,12 For the four adult/adolescent trials, “no worse” is defined in terms of the relative risk for serious asthma outcomes comparing LABA+ICS vs ICS being ≤ 2, which, in practical terms to account for random variability, implies that the upper bound of the 95% CI for the relative risk must be < 2 to conclude that LABA+ICS is no worse than ICS alone. The sample size of 11,700 patients per trial was computed by fixing the power at 90%. That is, the trial is designed to rule out a relative risk of 2 for the composite outcome with a probability of 90% when, in fact, there is no increase in the risk. Under the hypothesized rate of 15 serious asthma events per 1,000 patients per year, the upper limit of 2 for the CI corresponds to a point estimate of 1.32 for the relative risk, so that only if the estimated relative risk is < 1.32 can LABA+ICS be established as noninferior to ICS. The fifth trial, involving 6,200 children, was designed to have 90% power to rule out a relative risk of 2.7, that is, for the upper bound of the 95% CI for the relative risk to be < 2.7.

The first methodologic issue relates to the patients already treated with a LABA+ICS combination and who are eligible to enter these trials. Upon randomization, these patients are allocated to either continuing LABA+ICS treatment, albeit possibly with another agent, depending on the study they have been recruited to, or discontinuing their LABA+ICS treatment and replacing it by a treatment of an ICS only, which can impact the outcome. A meta-analysis of five randomized controlled trials provides evidence that discontinuation of the LABA component of a LABA+ICS combination is not inconsequential on asthma outcomes, though it had insufficient numbers of patients to provide data on asthma deaths or hospitalizations.13 It reported that such LABA discontinuation is associated with an increase in ED or unscheduled visits for asthma (RR, 2.2; 95% CI, 0.8-6.4), in the use of systemic steroids (RR, 1.7; 95% CI, 0.8-3.4) and in the incidence of withdrawal due to lack of efficacy or loss of asthma control (RR, 3.3; 95% CI, 2.2-5.0). It is conceivable that these consistent effects can translate to at least a 50% increase (RR = 1.5) in the rate of the composite outcome under study in the FDA-mandated trials, particularly asthma hospitalization. Such effects of treatment discontinuation at randomization have been noted with ICSs in trials of patients with COPD.14

In view of the current asthma-treatment trends, we anticipate that at least half of the patients enrolled in the FDA-mandated trials will already be treated with LABA+ICS. Consequently, we can compute the expected effect of such LABA discontinuation in these trials. Table 1 shows that with the planned sample size and 6-month follow-up for any given trial, the projected composite outcome rate of 15 events per 1,000 patients per year in the ICS-only group and the maximal relative risk of 1.32 lead to an expected 58 and 44 events for the LABA and no-LABA groups, respectively, if we assume no effects of LABA discontinuation (RR, 1.32; 95% CI, 0.89-1.95). Among the patients who had no LABA+ICS use prior to randomization, the RR remains at 1.32, with 29 and 22 events expected in the two treatment groups, respectively.

Table Graphic Jump Location
Table 1 —Expected Rates Ratios of Composite Asthma Outcome in Any One of the Four Adult/Adolescent Trials Resulting From the Effect of LABA Discontinuation at Randomization, Assuming That 50% of Patients Will Already Be on a LABA at the Time of Randomization

Data given as No., unless otherwise indicated. ICS = inhaled corticosteroid; LABA = long-acting β-agonist.

If we incorporate the effect of discontinuation among the patients who used LABA+ICS prior to randomization, the group allocated to continue the LABA+ICS will not be affected and thus is expected to also have 29 events. However, the group allocated to ICS only, who thus had to discontinue the LABA component at study enrollment, should have 50% more events than expected (ie, 33 events). As a result, when the data are pooled, the expected, overall RR of the composite outcome comparing LABA+ICS with ICS only becomes 1.05 (95% CI, 0.73-1.53). Thus, unknowingly, the maximal RR of 1.32 determined to establish safety could possibly be underestimated to 1.05. Under the null hypothesis of no excess risk, the expected RR of 1.0 is reduced to 0.8 (e-Table 1). Table 2 displays the results of sensitivity analyses that varied the degree of LABA discontinuation at randomization; namely, assuming that 30%, 50%, and 70% of the recruited patients were on a LABA at randomization. It shows that with increasing discontinuation rates, the upper bound of the CI artificially decreases further from the targeted and desired value of 2 for the relative risk.

Table Graphic Jump Location
Table 2 —Expected Rate Ratios of Composite Asthma Outcome in Any One of the Four Adult/Adolescent Trials Resulting From the Effect of LABA Discontinuation at Randomization in 30%, 50%, and 70% of Patients

Data given as No., unless otherwise indicated. See Table 1 legend for expansion of abbreviations.

The second methodologic issue arises from the primary outcome definition in these trials, namely, a composite end point that includes asthma-related death, intubation, or hospitalization occurring during the 6-month follow-up period. While the FDA recognizes that the results will be dominated by the hospitalization component of the outcome, it is useful to anticipate the effect of this dominance.9 We used data from a pooled analysis of formoterol trials and a meta-analysis that provided data on person-years of exposure and the corresponding incidences of asthma deaths.6,7 There were seven and three asthma deaths under LABA+ICS and one and no deaths under ICS only, respectively, for the two analyses, corresponding to pooled incidences of 0.33 asthma deaths per 1,000 persons per year under LABA+ICS and 0.07 under ICS only. Applying these asthma-death rates, we can expect to observe a total of five asthma deaths among the 46,800 subjects across the four trials: four under LABA+ICS and one under ICS only (RR, 4.0; 95% CI, 0.5-35.8). Even a sevenfold increase in the incidence of asthma death (seven vs one asthma deaths) with LABA+ICS will not lead to a statistically significant effect (RR, 7.0; 95% CI, 0.9-56.9). Table 3 provides expected results under three plausible scenarios of 1.13, 1.00, and 0.75 for the RR of asthma hospitalization or intubation estimated from meta-analyses that compared LABA+ICS with ICS.15 With all three scenarios, Table 3 illustrates the dominance of asthma hospitalization on the composite outcome and the negligible effect of a fourfold increase in the rate of asthma death.

Table Graphic Jump Location
Table 3 —Expected Rates Ratios of Composite Asthma Outcome and Components, Combining All Four Trials, for Three Different Scenarios of the Effect of LABA+ICS on Asthma Hospitalization or Intubation Based on Data From Meta-Analyses of Randomized Trials15

Data given as No., unless otherwise indicated. See Table 1 legend for expansion of abbreviations.

Rate ratios of asthma hospitalization or intubation estimated from meta-analyses are 1.13, 1.00, and 0.75 for scenarios 1, 2, and 3, respectively.15

The third issue arises from the inference involving four trials of the same question. The 90%-power calculation to rule out a relative risk of 2.0 for the composite outcome when, in truth, there is no increase in the risk, arrived at a sample size of 11,700 patients with asthma per study. Thus, for each study, the 90% power implies that there is only a 10% probability that the study will produce a false result with the upper bound of the CI larger than 2.0 (which will deem the drug as “unsafe”), when in truth there is no increase in the risk. While this calculation applies appropriately to each single study, it remains that all four studies will be assessed together by the FDA, so that the study-specific, 90%-power calculation will not apply to all studies as a group. From this perspective, there is, in fact, a probability of 34% [= 1 − (0.90)4] that simply by chance, at least one study will not show a noninferior outcome with the upper bound of the CI > 2.0. Including the fifth trial in children, this probability increases to 41%. Consequently, because of this simultaneous replication of identical studies, it becomes quite likely that simply by chance, one of the study drugs will be deemed “unsafe,” when, in truth, there is no increase in the risk. Figure 1 illustrates a simulation of the four identical, 6-month trials of this sample size and event rates, showing that under the null hypothesis, the upper bound of the CI for Trial 3 crosses the safety limit of 2.0 purely by chance; its corresponding LABA would, thus, be falsely judged as “unsafe.”

Figure Jump LinkFigure 1. A single simulation of the expected data for the four adult/adolescent trials under the null hypothesis of no effect (rate ratio = 1) with corresponding estimated rate ratios and 95% CIs for the composite asthma outcome. The CIs for trials 1, 2, and 4 do not cross the threshold value of 2, so that the corresponding long-acting β-agonists (LABAs) would be judged as noninferior, but the CI for trial 3 does cross the threshold value of 2, so the corresponding LABA could not be judged as noninferior.Grahic Jump Location

We discussed three methodologic issues that will inherently affect the risk estimates produced by the FDA-mandated trials and their interpretation. First and foremost, in view of the effect of LABA discontinuation from patients already using these drugs at enrollment and existing data from meta-analyses, the relative risk for each study could be underestimated by a factor of at least 20%. While 20% does not, at first glance, appear to be a major reduction, it is noteworthy that the highest value of 1.32 for the estimated relative risk to deem LABAs as “safe,” corresponding to the upper limit of the CI just below the threshold value of 2, will be underestimated to 1.05 with the discontinuation effect. This underestimation will artificially make it more difficult to detect a real risk.

Second, in view of the nature of the composite asthma outcome dominated by asthma hospitalization and the existing data on hundreds of randomized trials involving tens of thousands of patients, the studies will likely show no increased risk of the composite outcome. However, asthma deaths, dwarfed in the composite outcome, may show an increased risk with the LABA+ICS combination, but only a result of nine deaths to one will result in a statistically significant increase in asthma deaths; even a numerically worrisome spread of seven deaths to one will be interpreted as statistically uncertain.

Finally, in view of the four identical trials currently being conducted in adults and adolescents, the role of chance arising from such replication has not been considered in the decision process. Indeed, each trial is designed with a 10% probability that the study LABA is found “unsafe” when, in truth, it is safe. However, this becomes a 34% probability that one of the four trials will suggest that the study LABA is “unsafe” when, in truth, it is safe; this probability rises to over 40% if we include the fifth trial in children. If one of the trials passes the predetermined threshold risk limit, it will be impossible to conclude whether this represents a real effect of the specific LABA studied in the trial or simply bad luck.

In all, inherent features of the study design of the FDA-mandated trials to assess the safety of a LABA added to ICS therapy can lead to inaccurate results, possibly concealing a genuine risk of increase in asthma-related deaths. The identical replicated design is also a threat to one or more individual LABAs, which could be incriminated simply as a result of ill luck. Thus, it is quite likely that the clinical trials will not, in the end, “clarify the safety risk associated with LABAs,”9 but rather bring more uncertainty.

These multiple, identical, FDA-mandated trials represent, for the regulatory agencies, a new paradigm that calls for new decision tools. In view of the potential impediments of this approach, the FDA may wish to preempt such complexities by convening a group of scientists, including experts in the design and analysis of randomized trials, that could arrive at preset decision rules regarding the interpretation of the data resulting from these critical safety trials.

Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Suissa has participated in advisory boards and as speaker for all manufacturers of the long acting β-agonists under study, namely, AstraZeneca plc, GlaxoSmithKline plc, Merck & Co Inc, and Novartis AG; and is on the data monitoring committee of one of the trials on the safety of formoterol fumarate (Novartis AG). Dr Ariel has participated in clinical trials, advisory boards, and as speaker for all manufacturers of the long acting β-agonists under study, namely AstraZeneca plc; GlaxoSmithKline plc; Merck & Co, Inc; and Novartis AG.

Other contributions: This paper is the outcome of a session at the conference of the Israeli Society of Pulmonology, held in December 2011 in Eilat, Israel.

Additional information: The e-Table can be found in the “Supplemental Materials” area of the online article.

FDA

US Food and Drug Administration

ICS

inhaled corticosteroid

LABA

long-acting β-agonist

RR

rate ratio

Global strategy for asthma management and prevention. Global Initiative for Asthma website.http://www.ginasthma.org/uploads/users/files/GINA_Report_2012.pdf. Updated_2012. Accessed February 21, 2013.
 
Castle W, Fuller R, Hall J, Palmer J. Serevent nationwide surveillance study: comparison of salmeterol with salbutamol in asthmatic patients who require regular bronchodilator treatment. BMJ. 1993;306(6884):1034-1037. [CrossRef] [PubMed]
 
Nelson HS, Weiss ST, Bleecker ER, Yancey SW, Dorinsky PM; SMART Study Group. The Salmeterol Multicenter Asthma Research Trial: a comparison of usual pharmacotherapy for asthma or usual pharmacotherapy plus salmeterol. Chest. 2006;129(1):15-26. [CrossRef] [PubMed]
 
Levenson M. Long-acting beta-agonists and adverse asthma events meta-analysis. Statistical briefing package for joint meeting of the Pulmonary-Allergy Drugs Advisory Committee.http://www.fda.gov/ohrms/dockets/ac/08/briefing/2008-4398b1-01-FDA.pdf. Accessed March 8, 2013.
 
McMahon AW, Levenson MS, McEvoy BW, Mosholder AD, Murphy D. Age and risks of FDA-approved long-acting β2-adrenergic receptor agonists. Pediatrics. 2011;128(5):e1147-e1154. [CrossRef] [PubMed]
 
Sears MR, Ottosson A, Radner F, Suissa S. Long-acting beta-agonists: a review of formoterol safety data from asthma clinical trials. Eur Respir J. 2009;33(1):21-32. [CrossRef] [PubMed]
 
Rodrigo GJ, Moral VP, Marcos LG, Castro-Rodriguez JA. Safety of regular use of long-acting beta agonists as monotherapy or added to inhaled corticosteroids in asthma: a systematic review. Pulm Pharmacol Ther. 2009;22(1):9-19. [CrossRef] [PubMed]
 
Chowdhury BA, Dal Pan G. The FDA and safe use of long-acting beta-agonists in the treatment of asthma. N Engl J Med. 2010;362(13):1169-1171. [CrossRef] [PubMed]
 
Chowdhury BA, Seymour SM, Levenson MS. Assessing the safety of adding LABAs to inhaled corticosteroids for treating asthma. N Engl J Med. 2011;364(26):2473-2475. [CrossRef] [PubMed]
 
Sears MR. The FDA-mandated trial of safety of long-acting beta-agonists in asthma: finality or futility?. Thorax. 2013;68(2):195-198. [CrossRef] [PubMed]
 
Schumi J, Wittes JT. Through the looking glass: understanding non-inferiority. Trials. 2011;12:106. [CrossRef] [PubMed]
 
Fleming TR. Current issues in non-inferiority trials. Stat Med. 2008;27(3):317-332. [CrossRef] [PubMed]
 
Brozek JL, Kraft M, Krishnan JA, et al. Long-acting β2-agonist step-off in patients with controlled asthma: systematic review with meta-analysis. Arch Intern Med. 2012;172(18):1365-1375. [PubMed]
 
Suissa S, Ernst P, Vandemheen KL, Aaron SD. Methodological issues in therapeutic trials of COPD. Eur Respir J. 2008;31(5):927-933. [CrossRef] [PubMed]
 
Rodrigo GJ, Castro-Rodríguez JA. Safety of long-acting β agonists for the treatment of asthma: clearing the air. Thorax. 2012;67(4):342-349. [CrossRef] [PubMed]
 

Figures

Figure Jump LinkFigure 1. A single simulation of the expected data for the four adult/adolescent trials under the null hypothesis of no effect (rate ratio = 1) with corresponding estimated rate ratios and 95% CIs for the composite asthma outcome. The CIs for trials 1, 2, and 4 do not cross the threshold value of 2, so that the corresponding long-acting β-agonists (LABAs) would be judged as noninferior, but the CI for trial 3 does cross the threshold value of 2, so the corresponding LABA could not be judged as noninferior.Grahic Jump Location

Tables

Table Graphic Jump Location
Table 1 —Expected Rates Ratios of Composite Asthma Outcome in Any One of the Four Adult/Adolescent Trials Resulting From the Effect of LABA Discontinuation at Randomization, Assuming That 50% of Patients Will Already Be on a LABA at the Time of Randomization

Data given as No., unless otherwise indicated. ICS = inhaled corticosteroid; LABA = long-acting β-agonist.

Table Graphic Jump Location
Table 2 —Expected Rate Ratios of Composite Asthma Outcome in Any One of the Four Adult/Adolescent Trials Resulting From the Effect of LABA Discontinuation at Randomization in 30%, 50%, and 70% of Patients

Data given as No., unless otherwise indicated. See Table 1 legend for expansion of abbreviations.

Table Graphic Jump Location
Table 3 —Expected Rates Ratios of Composite Asthma Outcome and Components, Combining All Four Trials, for Three Different Scenarios of the Effect of LABA+ICS on Asthma Hospitalization or Intubation Based on Data From Meta-Analyses of Randomized Trials15

Data given as No., unless otherwise indicated. See Table 1 legend for expansion of abbreviations.

Rate ratios of asthma hospitalization or intubation estimated from meta-analyses are 1.13, 1.00, and 0.75 for scenarios 1, 2, and 3, respectively.15

References

Global strategy for asthma management and prevention. Global Initiative for Asthma website.http://www.ginasthma.org/uploads/users/files/GINA_Report_2012.pdf. Updated_2012. Accessed February 21, 2013.
 
Castle W, Fuller R, Hall J, Palmer J. Serevent nationwide surveillance study: comparison of salmeterol with salbutamol in asthmatic patients who require regular bronchodilator treatment. BMJ. 1993;306(6884):1034-1037. [CrossRef] [PubMed]
 
Nelson HS, Weiss ST, Bleecker ER, Yancey SW, Dorinsky PM; SMART Study Group. The Salmeterol Multicenter Asthma Research Trial: a comparison of usual pharmacotherapy for asthma or usual pharmacotherapy plus salmeterol. Chest. 2006;129(1):15-26. [CrossRef] [PubMed]
 
Levenson M. Long-acting beta-agonists and adverse asthma events meta-analysis. Statistical briefing package for joint meeting of the Pulmonary-Allergy Drugs Advisory Committee.http://www.fda.gov/ohrms/dockets/ac/08/briefing/2008-4398b1-01-FDA.pdf. Accessed March 8, 2013.
 
McMahon AW, Levenson MS, McEvoy BW, Mosholder AD, Murphy D. Age and risks of FDA-approved long-acting β2-adrenergic receptor agonists. Pediatrics. 2011;128(5):e1147-e1154. [CrossRef] [PubMed]
 
Sears MR, Ottosson A, Radner F, Suissa S. Long-acting beta-agonists: a review of formoterol safety data from asthma clinical trials. Eur Respir J. 2009;33(1):21-32. [CrossRef] [PubMed]
 
Rodrigo GJ, Moral VP, Marcos LG, Castro-Rodriguez JA. Safety of regular use of long-acting beta agonists as monotherapy or added to inhaled corticosteroids in asthma: a systematic review. Pulm Pharmacol Ther. 2009;22(1):9-19. [CrossRef] [PubMed]
 
Chowdhury BA, Dal Pan G. The FDA and safe use of long-acting beta-agonists in the treatment of asthma. N Engl J Med. 2010;362(13):1169-1171. [CrossRef] [PubMed]
 
Chowdhury BA, Seymour SM, Levenson MS. Assessing the safety of adding LABAs to inhaled corticosteroids for treating asthma. N Engl J Med. 2011;364(26):2473-2475. [CrossRef] [PubMed]
 
Sears MR. The FDA-mandated trial of safety of long-acting beta-agonists in asthma: finality or futility?. Thorax. 2013;68(2):195-198. [CrossRef] [PubMed]
 
Schumi J, Wittes JT. Through the looking glass: understanding non-inferiority. Trials. 2011;12:106. [CrossRef] [PubMed]
 
Fleming TR. Current issues in non-inferiority trials. Stat Med. 2008;27(3):317-332. [CrossRef] [PubMed]
 
Brozek JL, Kraft M, Krishnan JA, et al. Long-acting β2-agonist step-off in patients with controlled asthma: systematic review with meta-analysis. Arch Intern Med. 2012;172(18):1365-1375. [PubMed]
 
Suissa S, Ernst P, Vandemheen KL, Aaron SD. Methodological issues in therapeutic trials of COPD. Eur Respir J. 2008;31(5):927-933. [CrossRef] [PubMed]
 
Rodrigo GJ, Castro-Rodríguez JA. Safety of long-acting β agonists for the treatment of asthma: clearing the air. Thorax. 2012;67(4):342-349. [CrossRef] [PubMed]
 
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).
Supporting Data

Online Supplement

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Find Similar Articles
CHEST Journal Articles
PubMed Articles
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543