Discussion Points

• 1.
  

  Rehrer et al performed a randomized clinical trial to compare two corticosteroid regimens for treating adult patients in the emergency department (ED) presenting with acute asthma exacerbations.

  
  
  
  
  • A.
    

    What was the primary outcome in this study? Why was this outcome chosen by the authors? Would you have chosen the same outcome to investigate?

    
    
  • B.
    

    The Consolidated Standards of Reporting Trials (CONSORT) statement provides guidelines for clearly reporting trial design, conduct, and results in the scientific literature. Discuss how well this study adheres to the CONSORT checklist for describing randomization and blinding techniques (items 8 through 11 on the checklist, which can be found at http://www.consort-statement.org/ ).

    
    
  • C.
    

    Imagine conducting this study in your ED. Develop a plan for blinding and randomization that would work given the resources available at your institution.

    
    
  • D.
    

    More patients in the prednisone arm were given inhaled corticosteroids at discharge. Discuss possible reasons for this imbalance. How could this affect the study outcome?

  
  
  
  
• 2.
  

  In the rationale behind their study comparing a single, long-acting dose versus a 5-day course of corticosteroids, Rehrer et al hypothesized that “reducing the role of patient adherence to the care plan would improve outcomes.”

  
  
  
  
  • A.
    

    What percentage of patients included in the final analysis reported adherence with taking the study medications after discharge? How does this rate compare to national estimates of outpatient medication adherence?

    
    
  • B.
    

    Discuss how patients who were lost to follow-up may have affected the outcome of this study. Did the investigators include any interventions to minimize the number of patients lost to follow-up?

  
  
  
  
• 3.
  

  A 34-year-old female with known asthma presents to the ED with dyspnea, diffuse expiratory wheezing, accessory muscle use and the following vital signs: blood pressure 130/85 mm Hg, pulse 110 beats/minute, respiratory rate 35 breaths/minute, oxygen saturation 88% on room air, and temperature 36°C (98.6°F).

  
  
  
  
  • A.
    

    What is your initial management plan for this patient? When is a chest radiograph indicated? What criteria might one use to determine if the patient is stable for discharge? If she clinically improves and is stable for discharge, what medications would you prescribe?

    
    
  • B.
    

    Explain how corticosteroids and short-acting β-2 agonists address the pathophysiology of an acute asthma exacerbation. How do corticosteroids affect the patient response to short-acting β-2 agonist therapy?

    
    
  • C.
    

    What are the treatment goals for systemic corticosteroids in acute asthma management? What literature supports the use of systemic corticosteroids for acute asthma in the ED? What is the number needed to treat to reach these goals?

    
    
  • D.
    

    On reexamination, your patient appears tired and disoriented. Weingart described 2 basic strategies, lung protective and obstructive, for mechanical ventilation in the ED. Discuss why it is preferable to avoid intubating this patient if possible. Discuss your initial mechanical ventilation settings if intubation is required. Discuss your strategies for addressing elevated peak pressure and elevated plateau pressure in this patient.

  
  
  
  
• 4.
  

  Rehrer et al used a noninferiority trial design to compare 2 corticosteroid regimens for the treatment of acute asthma.

  
  
  
  
  • A.
    

    Discuss possible reasons for using a noninferiority trial design rather than a superiority trial to compare these 2 treatment strategies.

    
    
  • B.
    

    What are the null and alternative hypotheses in a noninferiority trial? How was the noninferiority margin, or allowable difference in outcomes between the 2 treatments, decided in this trial? Do you agree with this approach?

    
    
  • C.
    

    Why would someone interpreting a traditional frequentist hypothesis of the study data conclude that a single dose of oral dexamethasone failed to demonstrate noninferiority to 5 days of prednisone? What are other options for interpreting this study’s data (hint: think Bayesian)?

    
    
  • D.
    

    Will this study affect your management of adults with asthma? Will this study affect your practice for patients at high risk for noncompliance?

  
  

Answer 1

Q1. Rehrer et al performed a randomized clinical trial to compare two corticosteroid regimens for treating adult patients in the emergency department (ED) presenting with acute asthma exacerbations.

Q1.a What was the primary outcome in this study? Why was this outcome chosen by the authors? Would you have chosen the same outcome to investigate?

The primary outcome was relapse, defined by the authors as “an unscheduled return visit to a health care provider for additional treatment for persistent or worsening asthma within 14 days.” A primary outcome should be evaluated from both technical and conceptual perspectives. An ideal outcome from a technical perspective is precise, reliable, and valid. An ideal outcome from a conceptual perspective measures an aspect of health that is important to patients and society. These 2 perspectives often conflict with each other. For example, a 15% improvement in peak expiratory flow rate would be a reproducible and precise outcome, but this is an intermediate outcome, one that is not experienced by patients. The patient would care more about return to normal activity. Days until return to baseline activity would be a patient-centered outcome; however, this measurement is harder to measure precisely, as well as to reproduce.

Outcomes used in previous asthma studies include peak expiratory flow rates, forced exhalation volumes, hospital admission, and asthma severity scores such as the pulmonary index score. Use of an accepted outcome measurement allows easier comparisons to existing data, and the primary outcome chosen by Rehrer et al is concordant with the primary objective from a recent Cochrane review. Conducting follow-up via a telephone call is easier and less expensive than a repeated examination by a health care provider as required by outcomes such as peak expiratory flow rates or asthma severity scores.

A 2011 Annals of Emergency Medicine Journal Club reviewed a similar article comparing 2 days of dexamethasone with 5 days of prednisone in the treatment of adult asthma. The primary outcome in that study was return to regular activity within 72 hours. The design was critiqued because of the potential for recall bias as the data were collected by telephone 2 weeks after the index ED visit. Although patients may not be able to accurately recall their symptoms from 11 days ago, they should remember whether they consulted a health care provider for asthma in the previous 2 weeks. Rehrer et al studied an outcome that is important to patients and providers, accepted as valid by the literature, unlikely to have recall bias, and more objective and reproducible than a subjective reporting of symptoms.

Q1.b The Consolidated Standards of Reporting Trials (CONSORT) statement provides guidelines for clearly reporting trial design, conduct, and results in the scientific literature. Discuss how well this study adheres to the CONSORT checklist for describing randomization and blinding techniques (items 8 through 11 on the checklist, which can be found at http://www.consort-statement.org/ ).

Rehrer et al reported that a randomization table was maintained by the pharmacy, and patients were allocated to one of the treatment arms according to the results of that table. The article did not specify the type of randomization sequence or the method used to generate it. This leaves us to infer that a simple randomization sequence was used. No concealment mechanism was reported for the pharmacy to be unaware of the allocation for the upcoming subjects; however, treating providers enrolled patients in the study, whereas pharmacists maintained the randomization sequence and study medications. Thus, we can be reasonably confident that the providers enrolling patients were unaware of the randomization sequence. This reduces the potential for selection bias, in which patients are preferentially enrolled or excluded according to the next allocation in the randomization sequence.

Advantage Pharmaceuticals produced the medications for the study, with identical capsules used for the dexamethasone, prednisone, and placebo medications. Providers, nurses, research assistants, and study participants were blinded about whether subjects received capsules of prednisone or of dexamethasone followed by placebo.

The article did not explicitly address items 8 and 9 on the CONSORT checklist in regard to randomization sequence generation and allocation concealment. The authors clearly described how the enrolling providers were unaware of the randomization sequence and that all parties involved in the study, other than pharmacists, were blinded to the treatment arm of each patient. In summary, this was a double-blind, randomized, controlled trial conducted in the ED, with adequate procedures for randomization and blinding, though the authors could have better explicated the exact process.

Q1.c Imagine conducting this study in your ED. Develop a plan for blinding and randomization that would work given the resources available at your institution.

Thinking through the methods as if you were conducting the study is a helpful tool for deciding whether randomization and blinding were adequate. In this study, pharmacists were essential for concealing the randomization sequence from enrolling providers, and the use of identical capsules for each treatment arm enabled adequate blinding. Many EDs lack the systems and resources used in this study, such as research assistants or cooperation with pharmacy and pharmaceutical companies. With the resources available at your ED, consider how you could conduct a double-blind, randomized, controlled trial assessing prednisone versus dexamethasone in acute asthma.

A possible method for conducting this study in an ED without a clinical pharmacist would be to stock identical study medications in the ED Pyxis; however, it may lack space for study medications. Alternatively, a research assistant may go to the main hospital pharmacy to acquire study medications, but this introduces a delay between randomization and medication administration. This is only one example of the decisions facing researchers attempting to optimize study conditions in the setting of limited resources.

Q1.d More patients in the prednisone arm were given inhaled corticosteroids at discharge. Discuss possible reasons for this imbalance. How could this affect the study outcome?

Evaluating whether each arm received the same treatment other than the experimental treatment being studied is part of critical analysis of a randomized controlled trial. Differences in treatment other than the experimental therapy may confound the study. Items 5 and 13 of the CONSORT checklist instruct authors to describe the study protocol for interventions and what patients actually received for treatment. An imbalance between treatment arms is shown in Table 1 of the article by Rehrer et al, in which 27% of patients in the prednisone arm were prescribed inhaled corticosteroids at discharge compared with 17% of patients in the dexamethasone arm.

Possible reasons for this imbalance include random chance, a difference in inhaled corticosteroid use before arrival, imperfect blinding affecting physician behavior, or sicker clinical appearance of patients in the prednisone arm at discharge. As explored in a previous journal club, clinical trials with fewer than 500 subjects per arm will often have a 10% or greater difference in some patient characteristics because of random chance alone. Patients receiving inhaled corticosteroids before arrival would be more likely to receive a discharge prescription for inhaled corticosteroids, but home-inhaled steroid use before arrival was roughly equivalent between the 2 arms. Imperfect blinding seems unlikely, given that the study capsules were identical, and physicians would logically be more apt to prescribe inhaled corticosteroids to patients administered placebo capsules. Patients in the prednisone arm may have appeared more sick at discharge, prompting physicians to prescribe a stronger outpatient medication regimen; however, the discharge peak flow was slightly higher in the prednisone arm. Given that we are unable to identify a systematic reason for the differences in prescription of inhaled corticosteroids between the 2 treatment arms, this difference may be due to random variation in prescribed treatments.

Inhaled corticosteroids in addition to oral corticosteroids have been shown to reduce relapse in patients with acute asthma who are discharged from the ED. The likely effect, if any, of adding inhaled corticosteroids to the discharge medications would be to decrease the risk of relapse in those patients. Greater use of inhaled corticosteroids is a potential confounder in favor of the prednisone arm, so this imbalance between the 2 arms would not weaken a conclusion that a single dose of dexamethasone was noninferior to 5 days of prednisone in reducing relapse in acute asthma.

Answer 1

Q1. Rehrer et al performed a randomized clinical trial to compare two corticosteroid regimens for treating adult patients in the emergency department (ED) presenting with acute asthma exacerbations.

Q1.a What was the primary outcome in this study? Why was this outcome chosen by the authors? Would you have chosen the same outcome to investigate?

The primary outcome was relapse, defined by the authors as “an unscheduled return visit to a health care provider for additional treatment for persistent or worsening asthma within 14 days.” A primary outcome should be evaluated from both technical and conceptual perspectives. An ideal outcome from a technical perspective is precise, reliable, and valid. An ideal outcome from a conceptual perspective measures an aspect of health that is important to patients and society. These 2 perspectives often conflict with each other. For example, a 15% improvement in peak expiratory flow rate would be a reproducible and precise outcome, but this is an intermediate outcome, one that is not experienced by patients. The patient would care more about return to normal activity. Days until return to baseline activity would be a patient-centered outcome; however, this measurement is harder to measure precisely, as well as to reproduce.

Outcomes used in previous asthma studies include peak expiratory flow rates, forced exhalation volumes, hospital admission, and asthma severity scores such as the pulmonary index score. Use of an accepted outcome measurement allows easier comparisons to existing data, and the primary outcome chosen by Rehrer et al is concordant with the primary objective from a recent Cochrane review. Conducting follow-up via a telephone call is easier and less expensive than a repeated examination by a health care provider as required by outcomes such as peak expiratory flow rates or asthma severity scores.

A 2011 Annals of Emergency Medicine Journal Club reviewed a similar article comparing 2 days of dexamethasone with 5 days of prednisone in the treatment of adult asthma. The primary outcome in that study was return to regular activity within 72 hours. The design was critiqued because of the potential for recall bias as the data were collected by telephone 2 weeks after the index ED visit. Although patients may not be able to accurately recall their symptoms from 11 days ago, they should remember whether they consulted a health care provider for asthma in the previous 2 weeks. Rehrer et al studied an outcome that is important to patients and providers, accepted as valid by the literature, unlikely to have recall bias, and more objective and reproducible than a subjective reporting of symptoms.

Q1.b The Consolidated Standards of Reporting Trials (CONSORT) statement provides guidelines for clearly reporting trial design, conduct, and results in the scientific literature. Discuss how well this study adheres to the CONSORT checklist for describing randomization and blinding techniques (items 8 through 11 on the checklist, which can be found at http://www.consort-statement.org/ ).

Rehrer et al reported that a randomization table was maintained by the pharmacy, and patients were allocated to one of the treatment arms according to the results of that table. The article did not specify the type of randomization sequence or the method used to generate it. This leaves us to infer that a simple randomization sequence was used. No concealment mechanism was reported for the pharmacy to be unaware of the allocation for the upcoming subjects; however, treating providers enrolled patients in the study, whereas pharmacists maintained the randomization sequence and study medications. Thus, we can be reasonably confident that the providers enrolling patients were unaware of the randomization sequence. This reduces the potential for selection bias, in which patients are preferentially enrolled or excluded according to the next allocation in the randomization sequence.

Advantage Pharmaceuticals produced the medications for the study, with identical capsules used for the dexamethasone, prednisone, and placebo medications. Providers, nurses, research assistants, and study participants were blinded about whether subjects received capsules of prednisone or of dexamethasone followed by placebo.

The article did not explicitly address items 8 and 9 on the CONSORT checklist in regard to randomization sequence generation and allocation concealment. The authors clearly described how the enrolling providers were unaware of the randomization sequence and that all parties involved in the study, other than pharmacists, were blinded to the treatment arm of each patient. In summary, this was a double-blind, randomized, controlled trial conducted in the ED, with adequate procedures for randomization and blinding, though the authors could have better explicated the exact process.

Q1.c Imagine conducting this study in your ED. Develop a plan for blinding and randomization that would work given the resources available at your institution.

Thinking through the methods as if you were conducting the study is a helpful tool for deciding whether randomization and blinding were adequate. In this study, pharmacists were essential for concealing the randomization sequence from enrolling providers, and the use of identical capsules for each treatment arm enabled adequate blinding. Many EDs lack the systems and resources used in this study, such as research assistants or cooperation with pharmacy and pharmaceutical companies. With the resources available at your ED, consider how you could conduct a double-blind, randomized, controlled trial assessing prednisone versus dexamethasone in acute asthma.

A possible method for conducting this study in an ED without a clinical pharmacist would be to stock identical study medications in the ED Pyxis; however, it may lack space for study medications. Alternatively, a research assistant may go to the main hospital pharmacy to acquire study medications, but this introduces a delay between randomization and medication administration. This is only one example of the decisions facing researchers attempting to optimize study conditions in the setting of limited resources.

Q1.d More patients in the prednisone arm were given inhaled corticosteroids at discharge. Discuss possible reasons for this imbalance. How could this affect the study outcome?

Evaluating whether each arm received the same treatment other than the experimental treatment being studied is part of critical analysis of a randomized controlled trial. Differences in treatment other than the experimental therapy may confound the study. Items 5 and 13 of the CONSORT checklist instruct authors to describe the study protocol for interventions and what patients actually received for treatment. An imbalance between treatment arms is shown in Table 1 of the article by Rehrer et al, in which 27% of patients in the prednisone arm were prescribed inhaled corticosteroids at discharge compared with 17% of patients in the dexamethasone arm.

Possible reasons for this imbalance include random chance, a difference in inhaled corticosteroid use before arrival, imperfect blinding affecting physician behavior, or sicker clinical appearance of patients in the prednisone arm at discharge. As explored in a previous journal club, clinical trials with fewer than 500 subjects per arm will often have a 10% or greater difference in some patient characteristics because of random chance alone. Patients receiving inhaled corticosteroids before arrival would be more likely to receive a discharge prescription for inhaled corticosteroids, but home-inhaled steroid use before arrival was roughly equivalent between the 2 arms. Imperfect blinding seems unlikely, given that the study capsules were identical, and physicians would logically be more apt to prescribe inhaled corticosteroids to patients administered placebo capsules. Patients in the prednisone arm may have appeared more sick at discharge, prompting physicians to prescribe a stronger outpatient medication regimen; however, the discharge peak flow was slightly higher in the prednisone arm. Given that we are unable to identify a systematic reason for the differences in prescription of inhaled corticosteroids between the 2 treatment arms, this difference may be due to random variation in prescribed treatments.

Inhaled corticosteroids in addition to oral corticosteroids have been shown to reduce relapse in patients with acute asthma who are discharged from the ED. The likely effect, if any, of adding inhaled corticosteroids to the discharge medications would be to decrease the risk of relapse in those patients. Greater use of inhaled corticosteroids is a potential confounder in favor of the prednisone arm, so this imbalance between the 2 arms would not weaken a conclusion that a single dose of dexamethasone was noninferior to 5 days of prednisone in reducing relapse in acute asthma.