Oxygen versus air-driven nebulisers for exacerbations of chronic obstructive pulmonary disease: a randomised controlled trial

Background In exacerbations of chronic obstructive pulmonary disease, administration of high concentrations of oxygen may cause hypercapnia and increase mortality compared with oxygen titrated, if required, to achieve an oxygen saturation of 88–92%. Optimally titrated oxygen regimens require two components: titrated supplemental oxygen to achieve the target oxygen saturation and, if required, bronchodilators delivered by air-driven nebulisation. The effect of repeated air vs oxygen-driven bronchodilator nebulisation in acute exacerbations of chronic obstructive pulmonary disease is unknown. We aimed to compare the effects of air versus oxygen-driven bronchodilator nebulisation on arterial carbon dioxide tension in exacerbations of chronic obstructive pulmonary disease. Methods A parallel group double-blind randomised controlled trial in 90 hospital in-patients with an acute exacerbation of COPD. Participants were randomised to receive two 2.5 mg salbutamol nebulisers, both driven by air or oxygen at 8 L/min, each delivered over 15 min with a 5 min interval in-between. The primary outcome measure was the transcutaneous partial pressure of carbon dioxide at the end of the second nebulisation (35 min). The primary analysis used a mixed linear model with fixed effects of the baseline PtCO2, time, the randomised intervention, and a time by intervention interaction term; to estimate the difference between randomised treatments at 35 min. Analysis was by intention-to-treat. Results Oxygen-driven nebulisation was terminated in one participant after 27 min when the PtCO2 rose by > 10 mmHg, a predefined safety criterion. The mean (standard deviation) change in PtCO2 at 35 min was 3.4 (1.9) mmHg and 0.1 (1.4) mmHg in the oxygen and air groups respectively, difference (95% confidence interval) 3.3 mmHg (2.7 to 3.9), p < 0.001. The proportion of patients with a PtCO2 change ≥4 mmHg during the intervention was 18/45 (40%) and 0/44 (0%) for oxygen and air groups respectively. Conclusions Oxygen-driven nebulisation leads to an increase in PtCO2 in exacerbations of COPD. We propose that air-driven bronchodilator nebulisation is preferable to oxygen-driven nebulisation in exacerbations of COPD. Trial registration Australian New Zealand Clinical Trials Registry number ACTRN12615000389505. Registration confirmed on 28/4/15. Electronic supplementary material The online version of this article (10.1186/s12890-018-0720-7) contains supplementary material, which is available to authorized users.


Background
Acute exacerbations of chronic obstructive pulmonary disease (COPD) result in over 9,000 hospital admissions every year in New Zealand (NZ). 1 Nebulised medication is routinely administered to patients with an acute exacerbation and can be delivered via either high flow oxygen or air.
Oxygen driven nebulisers expose patients to high concentrations of inspired oxygen.
The risks of high concentration oxygen have been shown by the recent randomised controlled trial (RCT) in which high concentration oxygen therapy (supplementary oxygen 8-10 L/min and high flow oxygen-driven nebulisers for bronchodilator delivery) caused a 2.4-fold increased risk of death compared with titrated oxygen therapy (supplementary oxygen titrated as needed to achieve an oxygen saturation of 88-92% and air-driven nebulisers for bronchodilator delivery) in the management of severe exacerbations of COPD. 2 Hypercapnia, an elevated arterial carbon dioxide tension (PaCO2), is a known risk of high concentration oxygen therapy in patients with COPD, 3 4 and oxygen driven nebuliser use has been associated with marked increases in PaCO2, which resulted in stupor, seizures and death. [5][6][7] The British Thoracic Society (BTS) recommends titrated oxygen therapy to achieve oxygen saturations of 88-92% and air-driven nebuliser use for acute exacerbations of COPD. 8 It is recommended that if air-driven nebulisers are unavailable, use of oxygen driven nebulisers should be limited for up to six minutes. 8 However, it is likely that in clinical practice patients have oxygen delivered through nebulisers for longer than this, particularly if two nebulisers are being given sequentially. We have shown that air-driven nebulisers will prevent the increase in PaCO2 that results from use of oxygen-driven nebulisers in patients with stable COPD. 9 However, there are only two published RCTs in patients with acute exacerbations of COPD. 10 11 While acknowledging the important limitations of these studies, including lack of blinding, administration of single bronchodilator doses and low power, they did identify increased risk of an elevation in PaCO2 in COPD patients with hypercapnia. 10 11 The risks of oxygen delivered nebulisation in acute exacerbations of COPD need to be robustly defined to identify whether the widespread implementation of air driven nebulisers is required to ensure the safe delivery of bronchodilators by nebulisation to this at-risk patient group.

A. DESIGN
A parallel group double blinded RCT comparing the effect of air versus oxygen driven bronchodilator nebulisation on PaCO2, pH and oxygen saturation (SpO2) in patients admitted to hospital with an acute exacerbation of COPD. Ninety patients will be recruited and randomised to receive two 15 minute administrations of salbutamol by nebulisation, delivered by air or oxygen at a flow rate of 8L/min.

B. OBJECTIVES
i. To

A. RANDOMISATION
Randomisation will be 1:1 via a block randomised (block size 6) computer generated sequence, provided by the study statistician independent of recruitment and assessment of participants. The allocated intervention will be stored in an opaque sealed envelope and opened at the time of randomisation by the un-blinded investigator.

B. BLINDING i. Blinding of the blinded investigator during the study visit
The blinded investigator will be sat behind a portable screen so that they cannot see the participant or pulse oximeter screen. The participant will be asked not to comment out loud whether they are wearing their mask or nasal prongs. The SpO2 on the transcutaneous monitor will be covered.

ii. Blinding of the laboratory technician
The laboratory technician analysing the Capillary Blood Gas (CBG) sample will be masked to the randomised treatment.

iii. Blinding of the participant
The oxygen and air cylinders will not be distinguishable from each other to maintain blinding of the participant as to which gas their nebuliser is driven by. Participants will not be advised which treatment regimen they are randomised to or of the detail that one regimen involves removal of nasal prongs, if worn, (oxygen-driven) and the other does not (air driven). They will be informed, however, that their oxygen saturations will be monitored closely and if there are any concerns the regarding low oxygen saturation the un-blinded investigator will provide appropriate oxygen therapy immediately. Participants will not be able to see the screen of the pulse oximeter.

Nebuliser equipment a. Hudson RCI Micro Mist Nebuliser Masks (Hudson RCI, Durham, North
Carolina, USA) will be used to deliver the air and oxygen driven nebulised bronchodilator.
b. Salbutamol 2.5 mg will be used for each 15 minute nebulisation c. Oxygen and air to drive the nebulisation will be supplied in compressed portable cylinders (size D).

Study visit conduct A. RECRUITMENT, TIMING AND CONSENT
Potentially eligible participants will be identified on the wards and invited to take part in the study. Patients can be recruited at any time during their medical admission; however the study visit should be timed to deliver the nebuliser regimen as close as possible to the time of the next prescribed bronchodilator dose. Written Informed consent will take place prior to any study specific procedures. place. This will be followed by measurements of PtCO2 every 1 minute until consecutive measurements are within 1mmHg.

ii. Pulse oximeter application
The finger oximeter will then be placed on the participant's index finger. The StO2 screen will be uncovered on the transcutaneous monitor and the un-blinded investigator will record the time, StO2 and SpO2. The transcutaneous monitor screen displaying StO2 will be covered again and until t=80.

iii. Blinded investigator set up
From this point on and until t=80 minutes the blinded investigator and transcutaneous monitor will go behind a screen to avoid visualisation of the participant and pulse oximeter screen. b. Immediately after the CBG sampling, the following baseline (t=0 min) measurements will be made: i. The un-blinded investigator will record the time, SpO2 and nasal cannulae oxygen therapy flow (if any).
ii. The blinded investigator will record the time, PtCO2 and heart rate ii. Randomisation The randomisation envelope will be opened by the un-blinded investigator and regimens applied as below.

iii. Regimens
Air Oxygen therapy is to be continued to be titrated by the un-blinded investigator using nasal cannulae (worn under the nebuliser mask at 0-15 and 20-35 min). Note that if the patient was not receiving oxygen before the intervention (was breathing room air) this will continue unless the oxygen saturations drop to <85%. If this occurs the nebuliser mask will be removed, nasal prongs attached and oxygen given before reattaching the nebuliser mask. If this does happen it will be recorded in the case report form. Any nasal cannulae oxygen therapy is to be discontinued and nasal prongs removed, during t=0-15 min and t=20-35 min. Oxygen must be restarted at the same flow given prior to nebuliser delivery (i.e. at t=15 the oxygen must be started at the flow delivered at t=0, and at t=35 it must be started again at the flow delivered at t=20 minutes). Note that if the patient was not receiving oxygen before the intervention (was breathing room air) this will continue unless the oxygen saturations drop to <85%. If this occurs the nebuliser mask will be removed, nasal prongs attached and oxygen given. If this does happen it will be recorded in the case report form.

i. Intervention
The un-blinded investigator is to document the nasal prong oxygen flow instituted at the end of the second nebuliser.* Until t=80 oxygen will be delivered at this flow rate via nasal cannulae, except if the SpO2 falls to <85%.
In this case the flow is to be titrated until the 88-92% target saturation range is met.
* If participant has been on the air driven regimen, this will be the titrated flow at t=35. If participant has been on the oxygen driven nebuliser this will be the flow delivered via nasal prongs prior to the last nebulisation (t=20 min). HR: Heart rate, PcapCO2: Capillary partial pressure of carbon dioxide, PtCO2:

ii. Measurements and monitoring by the un-blinded investigator
Transcutaneous partial pressure of carbon dioxide, SpO2: oxygen saturations

Outcome variables A. COMPARATOR GROUPS
Outcome variables will be compared between the air versus oxygen driven nebuliser regimens.

B. OUTCOME MEASUREMENTS
Capillary blood gas samples (CBG) will measure pCO2 (PcapCO2) and pH. The samples will be collected in capillary tubes and analysed at the Wellington Regional Hospital or Hutt Valley Hospital laboratory. The transcutaneous monitor will be placed on the participant's earlobe (or backup sites as listed above) to measure transcutaneous PaCO2 (PtCO2), oxygen saturation (StO2) and heart rate. Finger pulse oximetry will measure oxygen saturation (SpO2).

C. PRIMARY OUTCOME VARIABLE
i. PtCO2 at completion of the second nebulisation (at t=35min*), adjusted for baseline.
*or the last recorded measurement should the t=35 measurement not be obtained (e.g. study terminated early due to rise in PtCO2 ≥10mmHg).

D. SECONDARY OUTCOME VARIABLES
i. PcapCO2 at completion of the second nebulisation (immediately prior to t=35min), adjusted for baseline PcapCO2.
ii. pH at completion of the second nebulisation (immediately prior to t=35 min), adjusted for baseline.
iii. PtCO2 at six minutes after the initiation of the first and second nebulisers (t=6 min and t=26 min) iv. PtCO2, SpO2 and heart rate at 5 minute intervals from t=0 to t=80 min v. Greatest PtCO2 change from baseline at any of the recorded time points between t=0 and t=35 minutes when the nebuliser is in place.

NOTE:
PtCO2 measurements recorded and analysed will be the actual (raw) data from the Transcutaneous Monitoring System, but adjustments for 'drift' will also be reported upon.

Methodology notes A. WASH IN, REGIMEN AND OBSERVATION PERIODS
Delivery of two 15 minute nebulisers with a 5 minute gap with a 2.5mg salbutamol dose were selected to represent real-life nebuliser delivery in the acute setting.
Titration of oxygen before and during the regimens represents recommended evidence based on best practice. 2 8 Maintenance of any nasal cannulae oxygen therapy at a constant flow during the observation period is designed to assess the risk of rebound hypoxia (a reduction in SpO2 following abrupt cessation of oxygen therapy to below the baseline level prior to instituting oxygen therapy). To detect any rebound hypoxemia a constant fraction of inspired oxygen is required. The results would represent the risk of rebound hypoxia in the situation that oxygen saturation monitoring does not occur following the abrupt cessation of high concentration oxygen therapy.

C. CLINICALLY SIGNIFICANT OUTCOMES
A rise in PtCO2 from baseline of >4mmHg is considered a physiologically significant change and >8mmHg a clinically significant change, based on previous definitions. 12 A change in pH >0.06 is based on the magnitude of the difference in pH that is expected to result from an increase in PaCO2 and >8mmHg and has clinical relevance as a marker of requirement for NIV in an exacerbation of COPD. 23

D. CAPILLARY BLOOD GAS SAMPLING
CBG allows accurate measurement of pCO2 24 and pH, 24 25 and is a less invasive alternative to ABG measurement. Whilst the earlobe is the preferred site for pO2 measurements, sampling blood from the fingertip or earlobe accurately reflects arterial pCO2 and pH over a wide range of values. 24 Fingertip sampling by retractable lancets as used commonly for blood sugar measurements is more familiar to participants and tends to yield better blood flow than earlobes. pH is a valuable outcome measure as it is an independent predictor of death in exacerbations of COPD. [26][27][28] Fingertips will be used in the first instance. However, in the event of unsuccessful sampling from a fingertip (e.g. sample too small) or contraindication (e.g. if a participant has particularly thick/tough skin on examination), then the earlobe can be used as a back-up site.

E. TRANSCUTANEOUS MONITORING
The transcutaneous monitor provides continuous and non-invasive PtCO2 monitoring. The accuracy of PtCO2 monitoring has been shown in a variety of settings including in healthy subjects, 29 acute exacerbations of COPD, 30 sleep disorders, 31 critical illness, 32 and other patients. 33 34

Safety Monitoring
There will be a Data Monitoring Safety Board comprising independent physicians, which will review data from any participants in which an SAE or rise in PtCO2 of >10mmHg occurred. This meeting will occur after 50% of participants have been recruited, or will be arranged sooner if 5 patients experience a PtCO2 rise of >10mmHg. The meeting will also be brought forward should any investigator have safety concernssuch as several patients in succession experiencing PtCO2 elevation, but not meeting the number of 5. This meeting will include experts independent from the study.

Power and Statistical Methods
A difference in PaCO2 of 4 mmHg represents a physiologically significant change. 12 13 In our controlled study of oxygen versus air driven nebulisers in stable COPD the standard deviation of PtCO2 was 5.5. 9 With 90% power and alpha of 5% this requires a total of 82 patients to detect a 4 mmHg difference. We anticipate a drop-out rate of <10% so our target recruitment is 90 patients. Analysis will be by intention to treat.
Participants recruited will be stratified in to groups that were receiving oxygen immediately prior to randomisation or those that were not. Our primary analysis is ANCOVA with PtCO2 as the response variable, and randomised treatment and baseline PtCO2 as co-variates. For other continuous outcome variables we will also use similar ANCOVA. Exploratory analyses for PtCO2, heart rate and SpO2 taken at five minute intervals will be presented graphically and analysed by mixed linear models.

Note on changes to protocol since previous version
This is Version 2.0, which was not submitted in the initial application of ethics approval.