Thys 2002.

Methods	Country: Belgium Design: Single‐centre randomised controlled parallel‐group trial conducted between 1999 and 2000 Study site: One urban university teaching hospital in Brussels, Belgium Method of analysis: t‐tests, ANOVA Aim: To determine whether (1) benefits of NIV on need for intubation, length of stay, and mortality are noted when NIV is performed in an emergency department (ED) very early after admission; and (2) early NIV has a real, rather than placebo, effect on objectively measured parameters and clinical status
Participants	Eligible for study: 187 (of either COPD or APO diagnosis) patients admitted to emergency department with acute respiratory failure Recruited: 20 adult patients (10 in NIV group; 10 in control group) with acute respiratory failure secondary to AECOPD (n = 12) or APO (n = 8). Of those with AECOPD, 7 were in NIV group; 5 in control group Completed: 20 participants (10 in each group) Age: Unable to ascertain for COPD‐only participants. Total (combined diagnoses) for NIV group: mean (SD) age = 71 (9) years; usual care group: mean (SD) age = 76 (7) years Gender: Unable to ascertain for COPD‐only participants. Total (combined diagnoses) gender (M:F) = 7:3 (NIV group); 4:6 (control group) Criteria used to define COPD: AECOPD defined as acute respiratory distress in a cigarette smoker with known history of long‐lasting dyspnoea on exertion with frequent exacerbations and cough, and mucus hyperproduction, without symptoms or signs of other specific causes (absence of pneumothorax, pneumonia, pleural effusion, no reason to suspect an episode of pulmonary embolism) Inclusion criteria: Aged 18 years or older with evidence of ARF (3 of the following criteria: acute onset of moderate to severe dyspnoea; respiratory rate > 30 (or < 10) breaths/min; hypoxaemia (PaO2 < 7.3 kPa (55 mmHg) on room air) or need for oxygen supplementation; respiratory acidosis (pH < 7.33) Exclusion criteria: immediate indication for endotracheal intubation (respiratory and/or cardiac arrest); major unrest; haemodynamic instability despite a fluid challenge; facial or thoracic trauma; lack of co‐operation; difficult adaptation of a facial mask to a patient’s facial anatomy; clinical suspicion of pulmonary embolism; retrosternal pain suggestive of a myocardial ischaemia even with normal admission electrocardiogram (ECG)
Interventions	Intervention description: Usual care (no sham NIV) plus bi‐level NIV (BiPAP ST/D 30; Respironics, Inc., Murrysville, PA, USA). BiPAP was used with inspiratory PS initially set at 10 cmH2O and EPAP at 4 cmH2O, used in assist‐control mode with a backup frequency of 10 breaths/min, via face mask (Bird, Bird Corp., Riverside, CA, USA). IPAP was increased by 2 cmH2O steps, until signs of discomfort (increasing sensation of dyspnoea), observation of air leaks, or pressure of 20 cmH2O was reached. EPAP was similarly increased until discomfort. Supplemental oxygen included via nasal catheter as required to maintain oxygen saturation > 90% Control description: Supplemental oxygen, bronchodilator aerosol therapy (fenoterol 1500 mg and ipratropium bromide 0.4 mg) repeated every 20 minutes and IV glucocorticoids (methylprednisolone 80 mg). Sham NIV employed, involving same setup as NIV group but a modified T‐connector piece (several holes created) between mask and tubing to enable IPAP = EPAP = 0 cmH2O. Oxygen was added with a nasal catheter inside the mask as needed to obtain saturation of 90% Treatment failure and success were defined in advance. Treatment success led to study end, treatment failure led to intubation in the active NIV group. Failure in the placebo NIV group led first to active NIV (rescue protocol) Duration of intervention: Until treatment success or failure Intervention delivery by: An attending physician delivered initial care and referred eligible patients. NIV or sham NIV was delivered by 2 study investigators. The attending physician remained present for the duration of the study (to decide treatment success or failure at any time point) Setting: Emergency department
Outcomes	Prespecified primary outcomes: Treatment failure (defined by all as worsened dyspnoea, respiratory and/or heart frequency, sweating and agitation, or deterioration in blood gases and/or haemodynamic status). For NIV group, this represented intubation; for control group, this represented cross‐over to NIV intervention (before potential subsequent intubation) Prespecified secondary outcomes: Hospital mortality, ICU admission, LOS (of ED, ICU, and hospital), NIV complications (skin damage, gastric dilatation, vomiting) Effect of NIV on: ‐ Dyspnoea (VAS), ABGs (at baseline, 20 minutes after Rx and at end of Rx) ‐ Continuous trace ECG, HR, SpO2, BP, EMG (of SCM muscle), respiratory inductive plethysmography (thoracoabdominal movements) Follow‐up period: Until hospital discharge
Notes	Study included patients with a primary diagnosis of COPD or APO; however, only data from participants with COPD are described (kindly supplied by study author following email request) Funder: Partly supported by a grant to F. Thys from the "Fondation Saint‐Luc"
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Insufficient information available.
Allocation concealment (selection bias)	Low risk	Randomisation was performed with opaque, sealed envelopes in batches of 20 that were opened at the time of inclusion
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Sham/placebo NIV used
Blinding of outcome assessment (detection bias) All outcomes	High risk	Attending physician was not blinded to group allocation, which may have affected decisions re failure/success, although a priori criteria were defined. Potential for secondary outcomes to be biased owing to knowledge of intervention, except length of stay. Study authors report the study as 'single‐blind' on page 546
Incomplete outcome data (attrition bias) All outcomes	Low risk	No evidence of data loss or attrition
Selective reporting (reporting bias)	High risk	No study protocol available for cross‐referencing. Primary study outcomes appear to be reported, but data for several secondary outcomes were not reported (e.g. SpO2, BP, HR, complications)
Imbalance of outcome measures at baseline All outcomes	Unclear risk	Appears balanced at baseline, but participants with COPD not discernible from those with APO
Comparability of intervention and control group characteristics at baseline	Unclear risk	Groups appear evenly matched at baseline, but participants with COPD not discernible from those with APO. No adjustments evident in analyses
Protection against contamination All outcomes	Low risk	All participants in control group received cross‐over rescue NIV, but data gathered before cross‐over were reported
Selective recruitment of participants	High risk	A large number of potentially eligible patients were not referred by local emergency medical teams "A total of 187 of these patients had a diagnosis of APO or acute exacerbation of COPD. The investigators were contacted for this study in 65 cases (37 acute exacerbations of COPD and 28 APO)”
Other bias	Low risk	No other sources of bias identified