Article Text

Randomised controlled crossover trial of the effect on Ptco2 of oxygen-driven versus air-driven nebulisers in severe chronic obstructive pulmonary disease
  1. Llifon Edwards1,2,
  2. Kyle Perrin1,2,
  3. Mathew Williams1,
  4. Mark Weatherall1,2,3,
  5. Richard Beasley1,2
  1. 1Medical Research Institute of New Zealand, Wellington, New Zealand
  2. 2Respiratory Medicine Department, Capital & Coast District Health Board, Wellington, New Zealand
  3. 3Department of Medicine, University of Otago Wellington, Wellington, New Zealand
  1. Correspondence to Dr Kyle Perrin, Medical Research Institute of New Zealand, Private Bag 7902, Wellington 6242, New Zealand; kyle.perrin{at}


Background The comparative safety of oxygen versus air-driven nebulised bronchodilators in patients with acute exacerbations of chronic obstructive pulmonary disease (COPD) is uncertain. A randomised controlled trial was performed to assess the effect on the arterial partial pressure of carbon dioxide of nebulised bronchodilator driven with oxygen versus air in stable severe COPD.

Methods In an open label randomised study, 18 subjects with stable severe COPD attended on 2 days to receive nebulised bronchodilator therapy driven by air or oxygen. Subjects received 5 mg salbutamol and 0.5 mg ipratropium bromide by nebulisation over 15 min, then, after 5 min, 5 mg salbutamol nebulised over 15 min, followed by 15 min of observation. Transcutaneous carbon dioxide tension (Ptco2) and oxygen saturations were recorded at 5 min intervals during the study. The primary outcome was the Ptco2 after the completion of the second bronchodilator treatment.

Results Ptco2 was higher with nebulised bronchodilator therapy delivered by oxygen, but decreased back to the level associated with air nebulisation 15 min after completion of the second nebulised dose. One subject experienced an increase in Ptco2 of 11 mm Hg after the first bronchodilator nebulisation driven by oxygen. The mean Ptco2 difference between the oxygen and air groups after the second nebulisation was 3.1 mm Hg (95% CI 1.6 to 4.5, p<0.001).

Conclusion Nebulisers driven with oxygen result in significantly higher levels of Ptco2 than those driven with air in patients with severe COPD.

Clinical trial registration number The study was registered on the Australian New Zealand Clinical Trials Registry (ACTRN12610000080022).

  • Carbon dioxide
  • oxygen
  • nebuliser
  • acute medicine-other
  • respiratory
  • asthma
  • pneumothorax
  • COPD

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


It is well recognised that administering high-concentration oxygen therapy to patients with acute exacerbations of chronic obstructive pulmonary disease (AECOPD) may lead to carbon dioxide retention.1 ,2 The clinical relevance of this physiological response in the prehospital setting has been demonstrated in a recent randomised controlled trial of oxygen therapy in AECOPD.3 In this study, high-concentration oxygen therapy during ambulance transfer to hospital was more likely to cause severe hypercapnia and respiratory acidosis than oxygen titrated to achieve arterial saturations between 88% and 92%, with a mean difference in the arterial partial pressure of CO2 (Paco2) of 34 mm Hg and pH of 0.12. Importantly, high-concentration oxygen therapy was associated with a 2.4-fold increased risk of death compared with controlled oxygen therapy. To reduce this risk, the British Thoracic Society (BTS) Oxygen Guidelines recommend that oxygen should only be administered to patients with AECOPD if oxygen saturations are <88%, and that oxygen therapy should be adjusted to maintain saturations between 88% and 92%.4

One of the potential difficulties in administering controlled oxygen during hospital transfer is the need to initiate treatment with bronchodilator drugs by nebuliser, which are usually oxygen driven. This inevitably results in the administration of high-concentration oxygen therapy during the period of nebulisation. The BTS COPD guidelines note that compressed air is rarely available in the majority of ambulances, and recommends that oxygen-driven nebulisation be limited to 6 min.4 Although this may limit the risk of hypercapnia to some extent, it does not overcome the risks associated with the need for frequent administration of nebulised drugs over long journey times, or the potential for nebuliser masks to be inadvertently left in place for longer.

The objective of this study was to compare the effect of bronchodilator nebulisers driven by oxygen versus air on the time course and severity of carbon dioxide retention in subjects with severe stable COPD. Our hypothesis was that the administration of bronchodilator drugs by oxygen-driven nebuliser would result in an increase in Paco2 compared with that driven by room air, and that this effect would be greater after the second nebulised bronchodilator administration.


We performed an open label, randomised, controlled, crossover study.


Eligible participants were aged over 18 years with a doctor diagnosis of COPD and a forced expiratory volume in 1 s (FEV1) ≤40% of predicted. Exclusion criteria were: sensitivity or other contraindications to salbutamol or ipratropium bromide; additional risk factors for hypercapnic respiratory failure (body mass index >40 kg/m2, severe musculoskeletal weakness, chest wall restriction); long-term oxygen therapy with >4 litres/min oxygen via nasal cannulae; current warfarin therapy. Participants were recruited from existing outpatient COPD databases, and the study was undertaken in the Clinical Measurement Unit of Wellington Regional Hospital.

Study intervention

The study comprised two visits 1 week apart. At the first visit, weight and height were recorded and spirometry was performed using a handheld spirometer (Micro Spirometer; Micro Medical Ltd, Rochester, UK), with the best of three attempts recorded. An arterial blood gas was performed with the subject breathing room air, or, if the subject was hypoxaemic (oxygen saturations <88%), on oxygen titrated to maintain arterial oxygen saturations between 88% and 92%. The arterial blood gas samples were obtained by radial puncture with a 22 gauge needle into a heparinised syringe and analysed immediately (Radiometer ABL800 FLEX, Copenhagen, Denmark).

The Paco2 was estimated with a transcutaneous carbon dioxide tension (Ptco2) monitor (TOSCA 500; Radiometer, Basel; Switzerland). Subjects who required nasal cannula oxygen at baseline because of oxygen saturations below 88% continued to receive this throughout the nebuliser treatment periods. An earlobe was cleaned with an alcohol swab and allowed to dry, and the Ptco2 probe was attached using an attachment clip and contact gel. A minimum of 10 min was allowed for arterialisation to occur and Ptco2 readings to stabilise, at which stage the first randomised treatment was started.

Subjects received the two treatments in random order at study visits 1 week apart. The study treatments were identical apart from the nebuliser delivery method. Salbutamol (5 mg) and ipratropium bromide (500 μg) were nebulised over 15 min, followed by a 5 min interval, and then a further 5 mg salbutamol was nebulised over 15 min. After the second nebulisation, monitoring was continued for a further 15 min. All nebulised drugs were administered with a Hudson RCI Micro Mist Nebuliser Mask (Hudson RCI, Durham, North Carolina, USA). Oxygen-driven nebulisation was delivered using a wall supply of oxygen at a flow rate of 8 litres/min. Air-driven nebulisation was delivered with a portable high-flow air-compression device (Portaneb; Respironics, Murrysville, Pennsylvania, USA). The Ptco2, oxygen saturation and heart rate were monitored continuously throughout the 50 min study period, and measurements were recorded at 5 min intervals. The FEV1 was measured at baseline and at the end of each study treatment.

A computer-generated randomisation schedule was provided by the statistician. Subjects were allocated to their treatment order by the study investigator. Blinding of the investigator and participants was not possible because of the use of the compressed air-driven device and wall-mounted oxygen. The protocol was terminated if an increase in Ptco2 >10 mm Hg from baseline was found at any stage during either of the treatment periods.

Statistical analysis

The primary outcome variable was the Ptco2 at the end of the second nebulisation period (t=35 min). Secondary outcomes included the time course of Ptco2 over the study period, the number of patients experiencing a rise in Ptco2 >10 mm Hg, the time course of heart rate and oxygen saturation responses during the study period, and change in FEV1 at the end of monitoring. The primary analysis was a mixed linear model, with the visit order and baseline measurement of the particular variable treated as a covariate. For the variables Ptco2, heart rate and oxygen saturation, the sandwich estimator of variance–covariance structure of repeated measurements was used, and the prespecified comparisons were at 15, 35 and 50 min. FEV1 was measured twice, at baseline before each treatment and at 50 min, and a simple unstructured variance–covariance matrix was modelled. Data description includes paired t tests at each measurement time, but the p values in these tables are not adjusted for multiple analyses.

Power calculations

For a difference of 4 mm Hg in the oxygen arm compared with the room air arm with a SD of 5.6,5 a sample size of 18 had 80% power to detect the nominated difference, with a type I error rate of 5%.


Between April and May 2010, 41 subjects were approached for inclusion in the study. Of these, 17 declined to participate, three were not eligible as they were currently taking warfarin, and three were not eligible because of an FEV1 >40% of predicted (figure 1). The characteristics of the 18 randomised subjects are shown in table 1. The subjects had a mean age of 73, were predominantly male, and had severe airflow obstruction with a mean FEV1 of 27% of predicted. The mean Paco2 was 47.8 mm Hg (range 38–56 mm Hg), and mean oxygen saturation was 92.7%.

Figure 1

Flow of subjects in the study. COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume in 1 s.

Table 1

Baseline characteristics of subjects

In one subject, the protocol was stopped during the oxygen treatment arm because the Ptco2 had increased by 11 mm Hg after 15 min. At the end of the final nebulised treatment (t=35 min), the mean (SD) Ptco2 was 53.0 (6.9) mm Hg in the oxygen-driven arm and 49.9 (7.1) mm Hg in the air-driven arm. In the mixed linear model incorporating baseline Ptco2 and accounting for repeated measures, the mean Ptco2 difference between the oxygen and air treatment arms was 3.1 mm Hg (95% CI 1.6 to 4.5, p<0.001) at 35 min. At 15 min the difference was 3.0 mm Hg (95% CI 0.08 to 5.2, p=0.001).

The change in Ptco2 over the 50 min study period is shown in table 2 and figure 2. Over the 50 min study period, the oxygen treatment arm showed a progressive rise in Ptco2 over the 15 min duration of the first nebulisation, followed by a decrease towards baseline during the 5 min interval, and a further rise over the 15 min of the second nebulisation. At the end of the study period, 15 min after the second nebulisation, the Ptco2 had returned to baseline in the oxygen treatment arm. In the mixed linear model, the mean Ptco2 difference between the oxygen and air treatment arms was −0.1 mm Hg (95% CI −0.6 to 0.4, p=0.69) at 50 min.

Table 2

Time course of Ptco2 changes over the study period

Figure 2

Transcutaneous carbon dioxide tension versus time for oxygen-driven (solid line) and air-driven (dashed line) nebulisers over the study period. Vertical bars are ±1 SD. One subject was withdrawn at time 15 min during oxygen treatment because the Ptco2 increased 11 mm Hg.

There was no significant difference between the two treatments in the change in heart rate, FEV1 or FEV1 percentage of predicted over the duration of the study period (see online supplement). There was a significant increase in oxygen saturation in the oxygen treatment arm throughout the study period (table 3, figure 3). In the mixed linear model incorporating baseline oxygen saturation and accounting for repeated measures, the difference in oxygen saturation was maximal at the end of the second nebulisation (6.8%, 95% CI 5.5 to 8.1, p<0.001) and remained significantly greater at the end of the study period (1.5%, 95% CI 0.8 to 2.2, p<0.001).

Table 3

Time course of oxygen saturation changes over the study period

Figure 3

Oxygen saturation versus time for oxygen-driven (solid line) and air-driven (dashed line) nebulisers over the study period. Vertical bars are ±1 SD.


This randomised controlled trial has shown that two oxygen-driven bronchodilator nebulisations resulted in a significant rise in Ptco2 compared with air-driven nebulisation in subjects with stable severe COPD. The mean difference in Ptco2 after oxygen and air nebulisation was 3.0 mm Hg in the first 15 min period, and, although the difference decreased during the 5 min interval period, there was a similar difference of 3.1 mm Hg during the second nebulisation, returning to baseline level during the subsequent 15 min observation period. Of concern is the finding that, in one of the subjects who had chronic respiratory failure, the Ptco2 increased by 11 mm Hg after 15 min of the first nebulisation, illustrating the potential risk of worsening hypercapnia if bronchodilator nebulisation is driven by oxygen.

These findings complement those of Gunawardena et al,6 who investigated the effects of a single oxygen-driven bronchodilator nebulisation on Paco2 in three groups of inpatients: normocapnic subjects with an acute exacerbation of asthma; normocapnic subjects with AECOPD; and hypercapnic subjects with AECOPD. They showed a significant rise in Paco2 of 7.7 mm Hg in the hypercapnic group after 15 min of nebulised salbutamol treatment driven by oxygen at a flow rate of 8 litres/min. There was no significant difference in Paco2 in the other two groups. O'Donnell et al obtained similar findings in a crossover study of 10 subjects with AECOPD, who were administered a single dose of salbutamol via air and then high-concentration oxygen.7 Nebulisers driven by compressed air had no effect on Paco2; however, oxygen-driven nebulisers significantly increased Paco2 in those with baseline hypercapnia. The similarly rapid increase in Paco2 has been demonstrated in patients with COPD with respiratory failure given high-concentration oxygen without bronchodilator drugs.8 Our study extends these findings by showing that, when oxygen-driven bronchodilator nebulisation is administered on a second occasion after a short interval of 5 min, the Ptco2 again increases, but not to a higher level than after the first nebulisation. Similar to the study of Gunawardena et al,6 we observed that the Ptco2 fell to baseline levels in the 15 min period after nebulisation.

There are a number of methodological issues relevant to the interpretation of our study results. First, we elected to enrol subjects with stable COPD rather than an acute severe exacerbation, which enabled us to study the subjects on two separate days and thereby conduct a randomised crossover trial. As a result, we are likely to have underestimated the magnitude of the increase in Ptco2 with oxygen-driven nebuliser use in AECOPD, as patients with stable COPD are less likely to develop oxygen-induced hypercapnia.1 However, we recruited patients with COPD who had severe airflow obstruction, with a mean FEV1 of 27% and mean baseline Paco2 of 47.8 mm Hg. As a result, our subjects were representative of patients likely to experience severe exacerbations of COPD requiring ambulance transfer to hospital. To increase the internal validity of the study, we excluded patients who had other risk factors for oxygen-induced hypercapnia. This is likely to have further underestimated the magnitude of Ptco2 increase demonstrated, as an unselected group of patients with COPD in clinical practice are more likely to have comorbidities such as morbid obesity or musculoskeletal weakness.

The study was designed to replicate the initial ambulance management approach to AECOPD. In our recent audit of prehospital management of AECOPD, the average duration of ambulance transfer was 49 min,9 and similar ambulance transfer times of 33 and 45 min have been reported from the UK10 and Australia.3 Thus our findings are likely to underestimate the risk associated with longer ambulance transfers, with a greater number of, or continuous, nebulised bronchodilator treatments. Although the study design was based on prehospital management, the findings also apply to in-hospital care in the emergency department, medical ward or high-dependency unit, in which bronchodilator nebulisers driven by oxygen may be used frequently and/or continuously. To ensure further generalisability to current practice, the bronchodilator regimen was an initial nebulisation with salbutamol and ipratropium bromide, followed by a second nebulisation with salbutamol.

We used a measurement of Ptco2 non-invasive assessment of Paco2 to minimise the risk of complications associated with the insertion of in-dwelling arterial catheters on two separate visits. Although Ptco2 only provides an estimate of Paco2, the device we used has been validated in a previous study of patients with acute asthma and pneumonia, and has minimal bias and acceptable limits of agreement.11 It has also been validated in a study of patients with AECOPD.12

The potential for oxygen-driven nebulisers to result in a rapid and marked increase in Ptco2 was demonstrated by the subject who experienced an increase in Ptco2 of 11 mm Hg during the first nebulisation. This physiological response poses a significant risk to patients transferred by ambulance, especially in the context of longer trip times and frequent dosing. Although the BTS COPD guidelines suggest limiting the length of oxygen-driven treatments to no longer than 6 min,4 this poses practical problems and compliance uncertainties. A safer approach would be to use alternative methods of bronchodilator administration, such as multiple metered dose inhaler actuations through a spacer, a technique that has shown efficacy in COPD.13 A second option would involve ambulance units carrying portable air jet compressor nebulisers for the administration of bronchodilators to patients, which has been demonstrated to be effective in a recent randomised controlled trial.3 Oxygen could then be continuously titrated as required by the use of nasal prongs, with the nebuliser mask applied over the prongs for drug delivery.

In conclusion we have shown that the administration of bronchodilator via oxygen compared with air-driven nebulisers results in worse hypercapnia in patients with severe COPD. Given the weight of evidence demonstrating harm with high-concentration oxygen in AECOPD, we propose that it is critical that health professionals in both the community and hospital settings prioritise the implementation of alternative methods of drug delivery in this high-risk group.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • An additional material is published online only. To view this file please visit the journal online (

  • Funding Funding was received from the Health Research Council of New Zealand, the Wellington Hospitals and Health Foundation, the Asthma and Respiratory Foundation of New Zealand, the Clyde Graham Charitable Trust, and the Royal Australasian College of Physicians.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was approved by Central Regional Ethics Committee (CEN/09/12/093), Wellington, New Zealand.

  • Provenance and peer review Not commissioned; internally peer reviewed.

Linked Articles

  • Primary survey
    Darren Walter