Background The reduced pressure in the aircraft cabin may cause significant hypoxaemia and respiratory distress in patients with chronic obstructive pulmonary disease (COPD). Simple and reliable methods for predicting the need for supplemental oxygen during air travel have been requested.
Objective To construct a pre-flight evaluation algorithm for patients with COPD.
Methods In this prospective, cross-sectional study of 100 patients with COPD referred to hypoxia-altitude simulation test (HAST), sea level pulse oximetry at rest (SpO2 SL) and exercise desaturation (SpO2 6MWT) were used to evaluate whether the patient is fit to fly without further assessment, needs further evaluation with HAST or should receive in-flight supplemental oxygen without further evaluation. HAST was used as the reference method.
Results An algorithm was constructed using a combination of SpO2 SL and SpO2 6MWT. Categories for SpO2 SL were >95%, 92–95% and <92%, the cut-off value for SpO2 6MWT was calculated as 84%. Arterial oxygen pressure (PaO2 HAST) <6.6 kPa was the criterion for recommending supplemental oxygen. This algorithm had a sensitivity of 100% and a specificity of 80% when tested prospectively on an independent sample of patients with COPD (n=50). Patients with SpO2 SL >95% combined with SpO2 6MWT ≥84% may travel by air without further assessment. In-flight supplemental oxygen is recommended if SpO2 SL=92–95% combined with SpO2 6MWT <84% or if SpO2 SL <92%. Otherwise, HAST should be performed.
Conclusions The presented algorithm is simple and appears to be a reliable tool for pre-flight evaluation of patients with COPD.
- Pulmonary disease
- chronic obstructive
- air travel
- ambulatory oxygen therapy
- COPD epidemiology
- COPD mechanisms
- long-term oxygen therapy (LTOT)
- pulmonary rehabilitation
- respiratory measurement
Statistics from Altmetric.com
- Pulmonary disease
- chronic obstructive
- air travel
- ambulatory oxygen therapy
- COPD epidemiology
- COPD mechanisms
- long-term oxygen therapy (LTOT)
- pulmonary rehabilitation
- respiratory measurement
What is the key question?
Can pulse oximetry at rest and during exercise be used to decide whether patients with chronic obstructive pulmonary disease (COPD) should use supplemental oxygen during air travel?
What is the bottom line?
A simple and reliable pre-flight evaluation algorithm for patients with COPD based on sea-level pulse oximetry at rest and during a 6 min walk test is presented.
Why read on?
The pre-flight evaluation algorithm might be useful for physicians treating patients with COPD.
The reduced atmospheric pressure in the aircraft cabin may cause severe in-flight hypoxaemia and respiratory symptoms in patients with lung disease, for example, chronic obstructive pulmonary disease (COPD).1–9 With the growing prevalence of COPD10 and a large proportion of patients with COPD travelling by air,1 ,9 simple and practical methods for pre-flight evaluation of the patients' fitness to air travel have been requested.1 ,3 ,11
Current air travel statements1 ,2 ,12 recommend supplemental oxygen when the arterial oxygen pressure (PaO2) is expected to fall below 6.6 or 7.3 kPa (50 or 55 mm Hg). Various lung function variables, prediction equations and algorithms have been proposed to estimate in-flight PaO2, the need for in-flight supplemental oxygen, and to select patients needing more advanced pre-flight testing, such as the hypoxia-altitude simulation test (HAST).1 ,3 ,6 ,13–20 HAST is considered to be the clinical ‘gold standard’,3 ,21 but is time consuming and not widely available. Thus, it is important to minimise the number of patients needing referral to HAST. Prediction equations, sea level PaO2 and spirometric values alone have proven not to be reliable tools for estimating the risk of severe in-flight hypoxaemia.1 ,3–5 ,16 ,22 In an algorithm published by the British Thoracic Society (BTS), sea-level oxygen saturation by pulse oximetry (SpO2 SL) was used as a discriminating variable,13 and it was recently confirmed that a SpO2 SL<92% seems to be an appropriate cut-off value for recommending in-flight supplemental oxygen without further pre-flight evaluation.23 For SpO2 SL≥92%, however, the predictive properties for detecting in-flight hypoxaemia were lower.23 It has been shown that both exercise desaturation15 ,24 ,25 and aerobic capacity4 ,5 correlate significantly with in-flight PaO2. Thus, it would be of interest to study if a combination of SpO2 SL and standardised exercise testing could be used to minimise the number of patients needing more cumbersome pre-flight testing. Our hypothesis was that a combination of SpO2 SL and oxygen desaturation during a 6 min walk test (6MWT) can be used to differentiate between patients with COPD needing or not needing supplemental oxygen during air travel, and patients who need further pre-flight evaluation with HAST.
To test this hypothesis, SpO2 was measured in a group of patients with COPD at rest and during a 6MWT, and the results were compared with oxygen tension and saturation obtained during HAST. The primary aim of the study was to develop a simple and reliable algorithm for pre-flight evaluation of patients with COPD based on these variables. The secondary aim of the study was to evaluate if HAST can be performed with SpO2 as a substitute for PaO2, since use of a non-invasive HAST could make the test simpler to perform and thereby more available.
This prospective cross-sectional study was performed at a pulmonary rehabilitation hospital in Norway. The Regional Committee for Medical Research Ethics approved the study (S-08640b), and written informed consent was obtained from the participants. The study was recorded in ClinicalTrials.gov (NCT00896584).
Construction of the algorithm
The construction of the algorithm was based on sea-level measurements of lung function, blood gases, pulse oximetry and 6MWT. For recommending in-flight oxygen, PaO2 HAST<6.6 kPa was chosen.1 ,14 ,21 ,26 To make the algorithm practical and clinically useful, the non-invasive variables with highest correlation to PaO2 HAST were analysed with receiver operating characteristics (ROC) analysis, first including all participants and second with subjects grouped according to SpO2 >95%, 92–95% and <92%.13 Results from the ROC analyses served as a basis for the construction of the algorithm, and thereafter all subjects were individually tested for calculation of the sensitivity and specificity of the new algorithm. Finally, the algorithm was prospectively validated on an independent sample of patients with COPD.
One hundred and thirty-nine consecutive patients with COPD who were referred from chest physicians in southern Norway to pre-flight evaluation were invited to participate in the study. The referral criteria were moderate to very severe COPD according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD),27 previous air travel intolerance, or SpO2 SL ≤95%.1 ,13 The inclusion criteria were a diagnosis of moderate to very severe COPD27 and the ability to perform a 6MWT. Exclusion criteria were unstable angina, uncontrolled hypertension, uncontrolled arrhythmia and long-term oxygen treatment. Thirty-nine patients were excluded, resulting in a study population of 100 subjects (figure 1). Sixty-nine of the participants had known comorbidities, the most frequent being systemic arterial hypertension, ischaemic heart disease and muscular-skeletal disorders. All patients used their daily medication.
Fifty additional patients with COPD who were referred to HAST with equal criteria as the study population were used to validate the algorithm.
Sea-level measurements and HAST
Lung function tests were performed according to standard criteria. SpO2 SL was measured with pulse oximetry (Nonin 3100 Wristox or Nonin PalmSat 2500, Nonin Medical Inc, North Plymouth, Massachusetts, USA), and simultaneously an arterial blood sample was drawn from a radial artery catheter and immediately analysed (ABL800 Flex, Radiometer, Copenhagen, Denmark). Exercise-related dyspnoea was measured with the modified Medical Research Council Dyspnoea Scale (mMRC).28
HAST was used to simulate a cabin pressure corresponding to an altitude of 2438 m above sea level (8000 ft).6 The subjects breathed 15.1% oxygen (15.1% O2, 84.9% N2, Yara Praxair, Norway) from a non-diffusing gas collection bag (170 litre Douglas-bag, Hans Rudolph Inc, Shawnee, USA) through a facemask (Mirage Full Face Mask, ResMed Corp, Poway, California, USA), and arterial blood samples were taken after 15 min hypoxic exposure.6 ,18 The SpO2 should be stable for 5 min before arterial blood sampling, otherwise the test was prolonged to 20 min. Electrocardiogram, SpO2 and dyspnoea were continuously monitored. The patients were recommended in-flight supplemental oxygen if PaO2 HAST was <6.6 kPa.1
To calculate sample size, we assumed that sensitivity and specificity would be approximately 80% in the planned study. It was then shown that 100 patients were needed to construct a new algorithm in which sensitivity and specificity should have CI length <16%. Patient characteristics are presented as mean and SD, unless otherwise specified. Relations between PaO2 HAST and patient characteristics were assessed from Pearson's correlation coefficient and one-way repeated measures analysis of variance. ROC analyses were performed with sea-level SpO2, SpO2 during 6MWT (SpO2 6MWT) and walking distance against PaO2 HAST <6.6 kPa as the discriminating variables. Statistical analyses were performed with PASW software (V.18.0; Chicago, Illinois, USA). Differences were considered significant if p<0.05.
The study comprised patients with COPD (n=100), with demographic characteristics as presented in table 1. According to the GOLD classification,27 22%, 46% and 32% were in the GOLD categories II, III and IV, respectively.
Sea-level SpO2 was used as a grouping variable as follows: SpO2 SL >95% (12% of patients), SpO2 SL 92–95% (55% of patients) and SpO2 SL <92% (33% of patients).
All patients were tested with HAST. Mean HAST values for PaO2 and SpO2 were 6.3 kPa (SD 0.6 kPa) and 83% (SD 4%), respectively. Seventy-three per cent of patients had a PaO2 HAST <6.6 kPa, indicating that they, in accordance with current guidelines, should use in-flight supplemental oxygen. There was an increase of 0.8 (SD 1.0) in dyspnoea score (Borg CR10) (p<0.001). Eighteen per cent of patients reported moderate to strong dyspnoea (Borg score 3–6), but there was no significant correlation between dyspnoea score and PaO2 HAST (r=0.16, p=0.115). None of the patients experienced hypoxia-induced myocardial ischaemia or arrhythmias.
The patients covered a distance varying from 150 to 604 m during the 6MWT (table 1). The mean decrease in SpO2 during the 6MWT was 10% (SD 5%) (p<0.001), and the mean SpO2 6MWT was 83% (SD 6%) (table 1).
Associations between in-flight PaO2 and sea-level characteristics
Significant correlations between sea=level characteristics and in-flight PaO2 are presented in table 2. PaO2 SL, SpO2 6MWT and SpO2 SL showed the strongest correlation with PaO2 HAST PaO2 SL was not included in the further analyses since one aim of the study was to develop a non-invasive evaluation method. Diffusing capacity of the lung for carbon monoxide (DL,CO), DL,CO/alveolar volume, total lung capacity (TLC), residual volume (RV), RV/TLC and dyspnoea measured with mMRC showed no significant relationship with in-flight PaO2.
ROC analyses were used as the basis for developing the pre-flight evaluation algorithm. SpO2 SL and SpO2 6MWT showed good diagnostic properties (area under curve 0.78 and 0.79, respectively) for detection of in-flight PaO2 <6.6 kPa (figure 2). The patients were grouped and data analysed according to the BTS pulse oximetry categories, SpO2 SL >95%, 92–95% and <92% (figure 3). In the group with sea-level SpO2 SL <92%, 30 of 33 (91%) patients dropped below the recommended level for minimum in-flight PaO2 (6.6 kPa), and were thereby in need of supplemental oxygen during air travel. Regarding the 55 patients in the group with SpO2 SL from 92% to 95%, a ROC analysis with SpO2 6MWT showed good prognostic properties (area under curve 0.80) for detection of in-flight PaO2 <6.6 kPa. The suggested cut-off value was SpO2 6MWT <84% (sensitivity 88%, 95% CI 80% to 96%; specificity 69%, 95% CI 52% to 85%). With regard to patients with SpO2 SL >95%, 5 of 12 (42%) had an in-flight PaO2 <6.6 kPa. In this group, ROC analysis showed exercise desaturation as a good prognostic variable, with an optimal cut-off value for SpO2 6MWT <84% (area under curve 0.71; sensitivity 80%, 95% CI 40% to 100%; specificity 71%, 95% CI 29% to 100%).
Based on the above analyses a pre-flight evaluation algorithm was constructed (figure 4). The algorithm was based on sea-level resting pulse oximetry (SpO2 SL) and exercise desaturation during the 6MWT (SpO2 6MWT) as the primary and secondary discriminator for evaluating whether the patient was fit to fly without further assessment, in need of further evaluation with HAST or should receive in-flight supplemental oxygen without further evaluation.
The pre-flight evaluation algorithm had a sensitivity of 99% (95% CI 96% to 100%) and a specificity of 82% (95% CI 67% to 96%) when all 100 subjects were individually tested. According to the algorithm, one-third (33%) of the patients would be advised to perform extended pre-flight testing with HAST. Six per cent of the patients were not correctly classified by the algorithm; of these, one patient was misclassified as fit to fly despite a PaO2 HAST <6.6 kPa (SpO2 SL 97% and SpO2 6MWT 87%, measured PaO2 HAST 6.3 kPa), and five patients would have been recommended to use in-flight oxygen without, in fact, having a PaO2 HAST <6.6 kPa (mean PaO2 HAST 7.2 kPa (SD 0.3 kPa)). The patients selected by the algorithm for further pre-flight evaluation with HAST had a mean PaO2 HAST of 6.6 kPa (SD 0.6 kPa).
After the algorithm was established, it was prospectively validated on an independent sample of 50 patients with COPD who were referred to HAST (table 3). Eight patients had SpO2 SL >95% (16%), 27 patients had SpO2 SL 92–95% (54%) and 15 patients had SpO2 SL <92% (30%). For all but four patients a correct choice was obtained with regard to use of in-flight supplemental oxygen. These four patients were recommended supplemental oxygen without having a PaO2 HAST <6.6 kPa. However, it should be noted that they all had PaO2 HAST values close to the recommended limit (mean PaO2 HAST 6.6 kPa (SD 0.1 kPa)). The sensitivity and specificity for the algorithm in this independent sample of patients were 100% (95% CI 90% to 100%) and 80% (95% CI 60% to 95%), respectively. The 20 patients which the algorithm selected for further pre-flight evaluation with HAST had a mean PaO2 HAST of 6.9 kPa (SD 0.5 kPa).
HAST: PaO2 versus SpO2
The secondary aim was to evaluate if HAST can be performed with SpO2 as a substitute for PaO2. There was a strong correlation between PaO2 HAST and SpO2 HAST (r=0.81, p<0.001) during HAST. The area under the ROC curve when using pulse oximetry to detect in-flight PaO2 <6.6 kPa was 0.93, indicating strong prognostic properties for the method (figure 5). The analysis suggested a cut-off value for SpO2 HAST ≤85% with a sensitivity of 89% (95% CI 81% to 96%) and a specificity of 81% (95% CI 67% to 96%) when SpO2 HAST was used as a substitute for PaO2 HAST <6.6 kPa (figure 5). When using SpO2 HAST, instead of PaO2 HAST an independent sample of 50 patients with COPD, we obtained a sensitivity of 90% (95% CI 77% to 100%) and a specificity of 85% (95% CI 70% to 100%). Three patients were misclassified as fit to fly despite having PaO2 HAST <6.6 kPa (mean PaO2 HAST 6.3 kPa), and three patients would have been recommended to use in-flight oxygen without, in fact, having a PaO2 HAST <6.6 kPa (mean PaO2 HAST 6.7 kPa).
A large number of patients with COPD travel by air, most of them without severe in-flight medical problems.9 ,13 However, some patients develop severe hypoxaemia.1 ,2 ,4–7 Thus, simple and consistent pre-flight assessment guidance regarding the need for in-flight supplemental oxygen has been requested.1 ,3 ,11 In the present study we have constructed and validated a simple and clinically feasible algorithm for pre-flight assessment of patients with COPD based on sea-level resting SpO2 and SpO2 values during a 6MWT.
Various proposed equations and single sea-level variables have proven not to predict in-flight hypoxaemia with a satisfactory precision.1 ,3–5 ,16 ,22 Several authors have suggested that exercise-related outcomes may be useful discriminators.4 ,5 ,14 ,24 ,25 Previous studies from our group show that aerobic capacity correlates with in-flight hypoxaemia.4 ,5 Chetta et al showed that desaturation during the 6MWT provides useful information in the pre-flight assessment.24 Since oxygen saturation measured with pulse oximetry, at rest and during a 6MWT, is frequently used in the medical care of patients with COPD,31 an algorithm employing a combination of these variables would be simple to implement in a busy clinical practice. An assessment algorithm that can discriminate between patients who will need supplemental oxygen during air travel and those who can travel without such equipment would be of considerable value in the evaluation of patients with COPD intending to travel by air, especially if the number of unequivocal findings needing more extensive pre-flight evaluation is reduced.
The results from our group have previously indicated that the BTS algorithm with only sea-level SpO2 cannot be used with confidence to predict in-flight hypoxaemia.23 A large number of patients at risk of developing severe hypoxaemia was not detected, and a considerable number of patients needed more advanced pre-flight evaluation.23 By adding a 6MWT, including measurement of SpO2, the current study shows that the number of patients needing referral to HAST was markedly reduced. The 6MWT is a widely used test to assess exercise performance in patients with COPD, and is much more available than HAST. However, it is important that the 6MWT is performed according to guidelines, and it must be stressed that shortcuts must be avoided.29 The suggested algorithm has a high sensitivity, which was reproduced when applying it prospectively on a separate group of patients with COPD. None of these study subjects were misclassified as fit to fly without supplemental oxygen. Due to somewhat lower specificity, the algorithm overestimated the risk of in-flight hypoxaemia, resulting in unnecessary use of supplemental oxygen in 8% of the patients.
According to our results, patients with COPD who have SpO2 SL >95% and without severe exertional desaturation (SpO2 ≥84%) can travel safely by air without further pre-flight assessment. In addition, further pre-flight assessment is not necessary in patients with SpO2 SL <92%13 ,23 or in patients with SpO2 SL 92–95% and SpO2 6MWT <84%. These patients should, according to our results, be equipped with supplemental oxygen during the flight. Thus, extended pre-flight assessment with HAST might be limited to patients with either the combination of resting SpO2 SL >95% and severe exercise desaturation (SpO2 <84%) and to patients with SpO2 SL between 92% and 95% without severe exercise desaturation (≥84%). In these two groups of patients, the level of in-flight hypoxaemia was difficult to predict, underlining the need for pre-flight testing with HAST.
Even though HAST is increasingly used in pre-flight assessment,32 ,33 it is not widely available.11 ,21 HAST has been shown to be a good predictor of in-flight PaO218 ,32 ,34 ,35 and the results obtained are reproducible.32 However, one might find it cumbersome to take repeated arterial blood samples or insert a radial artery catheter, and substitution of arterial blood gas measurement with pulse oximetry would simplify the HAST procedure considerably. To our knowledge, comparison of arterial blood gases and pulse oximetry during HAST has not previously been published. As expected, a strong correlation between PaO2 HAST and SpO2 HAST was observed, and when using a cut-off value for SpO2 HAST ≤85% as a substitute for a PaO2 HAST <6.6 kPa, acceptable values for sensitivity and specificity of the test were obtained. Our results show that use of pulse oximetry during HAST may underestimate the need for in-flight oxygen. Thus, the authors would recommend arterial blood gas measurement during HAST as the method of choice.
To our knowledge, this is the first prospective study to use a set of common baseline characteristics for the construction of a pre-flight evaluation algorithm. The good prognostic properties of the algorithm were confirmed by a prospective validation on a separate group of subjects with COPD. However, it should be noted that a selected group of patients was studied; they were all referred for pre-flight evaluation. These are the patients for whom the algorithm is intended. One must also keep in mind that the present study only comprised patients with moderate to very severe COPD27 and that the algorithm may not be applicable to patients with other lung diseases.
In conclusion, an algorithm for pre-flight evaluation of patients with COPD is presented, employing simple non-invasive oximetry at rest and during walking. By using the algorithm, the majority of a population consisting of patients with moderate to very severe COPD could be classified as fit to fly or in need of supplemental oxygen without more advanced pre-flight assessment.
Leiv Sandvik, Professor in Biostatistics, Oslo University Hospital, is gratefully acknowledged.
Funding The study was funded by grants from The Norwegian Heart and Lung Patient Organisation and The Norwegian Foundation for Health and Rehabilitation.
Competing interests None.
Ethics approval Ethics approval was provided by The Regional Norwegian Committee for Medical Research Ethics.
Provenance and peer review Not commissioned; externally peer reviewed.