January 11, 2016

Review of a Randomised Control Trial

Non-invasive ventilation (NIV) has been increasingly used both as a preventive and as a curative measure for acute respiratory failure (ARF) in many clinical scenarios, cardiothoracic surgery being one of the most recently reviewed ones. In this setting, NIV can reduce the work of breathing, increase end-expiratory lung volume and improve oxygenation but also induce intrathoracic pressure changes which can affect preload, afterload, heart rate and myocardial contractility.1,2 However, approximately 20% of the patients on NIV after cardiothoracic surgery will require re-intubation3, whereas patient discomfort, gastric distension and skin necrosis are well documented complications.4

High flow oxygen therapy via nasal cannula (HFNC) is a technique whereby heated and humidified oxygen is delivered at high flow rates, resulting in the washout of nasopharyngeal dead space, the reduction in inspiratory resistance and metabolic work but also the provision of positive pressure for lung recruitment.5 Its’ use is increasing in the adult population and in different scenarios, with mixed results on oxygenation, desaturation episodes and re-intubation rates6-8 There is a limited number of studies that have focused on the effect of HFNC in patients post cardiothoracic surgery both in the preventive9 and the curative setting.10

In their publication in JAMA, Stéphan et al.11 present the results of a multicentred, randomised, non-inferiority trial with a margin of 9%, which set out to prove that HFNC was not inferior to NIV in preventing or resolving ARF after cardiothoracic surgery. Their primary composite outcome was treatment failure (re-intubation, switch to other treatment, treatment discontinuation), with several secondary outcomes including changes in respiratory variables, comfort, dyspnoea and skin breakdown score, complications and number of bronchoscopies. They reported an absolute difference of treatment failure of 0.9% (95%CI -4.9% to 6.6%, p=0.003) and concluded that among cardiothoracic surgery patients with or at risk of respiratory failure, the use of HFNC was not inferior to BiPAP.11

Given the unique nature of non-inferiority trials, a few issues warrant special mention: firstly, the specified non-inferiority margin of 9% was based on the study by Ferrer et al.12; this study however, quoted the smallest benefit for NIV vs. conventional oxygen therapy to be a little below 3%. The authors chose the arbitrary 9% margin ‘after discussion with consulting physicians representing the BiPOP study group’ – despite deviating from the accepted statistical approach to setting a non-inferiority threshold13, HFNC would have still been proven not inferior to NIV, were the threshold left at 3%. Secondly, the active control group in a non-inferiority design is normally an established treatment with which the new intervention is compared. In Stéphan et al., the standard group treatment (BiPAP at least 4 times a day) is far from established, as highlighted by the authors themselves11 and mentioned beforehand.3,9,10 Thirdly, the choice of statistical analysis being solely intention-to-treat (ITT) has serious limitations in the context of non-inferiority trials. If for whatever reason, a large number of patients from the control group do not adhere to treatment, this will appear less effective. In a superiority trial, this is accepted as a conservative method as analysis; in the non-inferiority design it has the risk of presenting both arms as equally ineffective and per-protocol analysis should be carried out as well. In the present study, even though all patients received treatment as randomised and the ones switching between the two groups were not statistically different (p=0.15), patients that discontinued treatment in BiPAP arm were more than in the HFNC group (p=0.04), potentially reducing the difference between the two.

Furthermore, and not exclusively for non-inferiority trials, the absence of blinding has the risk of introducing bias, a caveat that the authors have gone to lengths trying to overcome by setting clear criteria for the crossover and re-intubation situations. Two final comments on the choice of inclusion criteria and composite primary endpoint: Stéphan et al., included patients either at risk of (preventive) or already demonstrating (curative) signs of ARF. Although both situations are common in post cardiothoracic patients (with a very fine line separating the two), the severity of illness and hence the treatment required might differ significantly. Composite endpoints are widely preferred as they increase statistical power but not without creating difficulty in interpreting the trial results. Re-intubation, switch over to other treatment and treatment discontinuation weigh very differently as separate patient outcomes and ideally should be used interchangeably.

In conclusion, the very interesting study by Stéphan and colleagues adds to the uncertainty around the use of NIV and HFNC in the post-operative setting. Studies that separately focus on the prevention and the treatment of ARF are clearly needed, before recommendations on the effectiveness on HFNC can be attempted.   

This article review was prepared by ESICM Journal Review Club member Victoria Metaxa.


  1. Guarracino F, Ambrosino N. Non invasive ventilation in cardio-surgical patients. Minerva Anestesiol 2011;77:734-41
  2. Nava S, Hill N. Non-invasive ventilation in acute respiratory failure. Lancet 2009;374:250-9
  3. Zhu G-F, Wang DJ, Liu S et al. Efficacy and safety of noninvasive positive pressure ventilation in the treatment of acute respiratory failure after cardiac surgery. Chin Med J (Engl). 2013;126(23):4463-4469
  4. Carron M, Freo U, BaHmmam AS et al. Complications of non-invasive ventilation techniques: a comprehensive qualitative review of randomised trials. Br J Anaesth 2013; 110(6):896-914
  5. Dysart K, Miller TL, Wolfson MR et al. Research in high flow therapy: Mechanisms of action. Respiratory Medicine 2009; 103:1400-5
  6. Frat JP, Arnaud WT, Mercat A et al. for the FLORALI Study Group and the REVA Network. High-Flow Oxygen through Nasal Cannula in Acute Hypoxemic Respiratory Failure. NEJM 2015; 372:2185-96
  7. Lee JH, Rehder KJ, Williford L et al. Use of high flow nasal cannula in critically ill infants, children and adults: a critical review of the literature. Intensive Care Med 2013; 39:247-57
  8. Maggiore SM, Idone FA, Vaschetto R, et al. Nasal high-flow versus Venturi mask oxygen therapy after extubation: effects on oxygenation, comfort, and clinical outcome. Am J Respir Crit Care Med. 2014;190(3):282-288
  9. Lorut C, Lefebvre A, Planquette B et al. Early postoperative prophylactic noninvasive ventilation after major lung resection in COPD patients: a randomised controlled trial. Intensive Care Med 2014; 40:220-7
  10. Corley A, Caruana LR, Barnett AG et al. Oxygen delivery through high-flow nasal cannulae increase end-expiratory lung volume and reduce respiratory rate in post-cardiac surgery patients. Br J Anaesth 2011; 107:998-1004
  11. Stéphan F, Barrucand B, Petit P, et al; for the BiPOP Study Group. High-flow nasal oxygen vs noninvasive positive airway pressure in hypoxemic patients after cardiothoracic surgery: a randomized clinical trial. JAMA 2015; 313:2331-9
  12. Ferrer M, Valencia M, Nicolas JM, et al. Early noninvasive ventilationvaverts extubation failure in patients at risk: a randomized trial. Am J Respir Crit Care Med; 2006; 173(2):164-170
  13. Mulla SM, Scott IA, Jackevicious CA et al. How to use a noninferiority trial. JAMA 2012; 308:2605-11

Comment on this news