Intended for healthcare professionals


Predictions of hypoxaemia at high altitude in children with cystic fibrosis

BMJ 1994; 308 doi: (Published 01 January 1994) Cite this as: BMJ 1994;308:15
  1. P J Oades,
  2. R M Buchdhal,
  3. A Bush
  1. Department of Paediatrics, Royal Brompton National Heart and Lung Hospital, Sydney Street, London SW3 6NP
  1. Correspondence to: Dr Oades.
  • Accepted 1 October 1993


Objective: To assess the usefulness of a hypoxic challenge in a laboratory at sea level in predicting acute desaturation at altitude in children with lung disease.

Design: Comparison of responses to hypoxic challenge in different settings.

Subjects: 22 Children (12 boys) aged 11 to 16 years with cystic fibrosis in whom the mean forced expiratory volume in one second was 64% (range 24-100%).

Setting: Lung function laboratory, the Alps, and aboard commercial jet aircraft. Main outcome measures - Spirometric lung function at sea level and finger probe oximetry with air and 15% oxygen. Oximetry during high altitude flight and on a mountain at altitude of 1800 m.

Results: Significant desaturation (range 0 to 12%) occurred with all hypoxic challenges (P<0.002). The best predictor of hypoxic response from a single reading was the laboratory test (r(sup2)=76% for flight and r(sup2)=47% for mountain altitude), but the mean errors of prediction were not clinically significantly different. In six children who showed the greatest desaturation the laboratory test overestimated desaturation, but other predictors underestimated desaturation in three by up to 5%.

Conclusions: The laboratory hypoxic challenge directly predicted the worst case of desaturation during flight and at equivalent high altitude. Spirometry and baseline oxygen saturations may underestimate individual hypoxic response. The test may have wider applications to other patients with stable chronic lung diseases, particularly in determining who needs supplementary oxygen during air travel and who should be advised against holidays at high altitude.

Clinical implications

  • Clinical implications

  • Children with cystic fibrosis may show desaturation during flight or at altitude

  • This study found that hypoxic challenge at sea level predicts the worse case of desaturation

  • Predictions based on spirometry or sea level saturations may underestimate hypoxic desaturation in some people

  • If desaturation occurs to 83% then humidified oxygen should be available during flights and advice be given against holidays at high altitude

  • The test may be applicable to other patients with stable chronic lung disease, although not in unstable conditions such as asthma


As altitude increases the partial pressures of inspired gases fall. This results in lower alveolar partial pressures of oxygen and a risk of hypoxaemia. Thus healthy children living at high altitude have lower oxygen saturations than those at sea level.1 The healthy children living at high altitude have lower hazards to patients with cystic fibrosis with hypoxic lung disease of holidaying at high altitude have been described. 2 Additionally, although cabins of commercial aircraft are pressurised during flight at high altitude, the cabin altitude is still between 1525 and 2134 m,3 equivalent to breathing down to 15% oxygen at sea level. Passengers with obstructive lung disease may already have low alveolar partial pressures of oxygen and are therefore more prone to serious desaturation during flight.4 We examined whether desaturation at altitude and hence a requirement for additional inspired oxygen could be predicted from spirometry, oxygen saturations, and a hypoxic challenge at sea level.

Patients and methods

We studied 22 children (12 boys) aged 11 to 16 years with cystic fibrosis - 14 (seven boys) before a holiday in the French Alps and eight before going to America (no mountain measurements).

Two weeks before the holidays the children underwent spirometry (Compact Spirometer; Vitalograph, Buckingham) and a hypoxic test in a laboratory at sea level. We recorded oxygen saturations initially while the child was breathing air and then, when wearing a nose clip, while breathing 15% oxygen in nitrogen (British Oxygen Corporation) from a Douglas bag via a Rudolph valve. The composition of the inspired gas was checked with a Hudson oxygen monitor (Viamed, Keighley). A second saturation was recorded at steady state, indicated at 5 to 10 minutes by no further change in saturations. Further control measurements were made in the departure lounges in the airport before each outward and return take off and at low altitude in the mountain resort (250 m). Inflight hypoxia was assessed by measuring saturations one hour after take off for outward and return flights when at high altitude (11 000-12 000 m, verified by aircrew).

Oxygen saturations on the mountain were measured on two days at 1800 m; the spot was reached by cable car and its altitude verified on a local ordnance survey map. All saturations were measured with the children sitting at rest after morning physiotherapy and routine drug treatments by using the same Ohmeda Bio X3740 pulse oximeter (Hatfield) and index finger probe. All measurements were recorded at room temperature except those on the mountain (−3 degree C to 5 degree C), where the children wore gloves before the measurement. Readings were allowed to stabilise for 2 to 3 minutes, when the oximeter pulse rate was equal to the opposite radial pulse.


Statistical analysis was by the Wilcoxon matched pairs signed rank test and Pearson correlation with derivation of the mean errors of prediction and coefficients of determination (adjusted for degrees of freedom; with the Minitab computer statistics package.5 The biases in the laboratory test in predicting flight or mountain saturations were determined by the method of Bland and Altman.6 Informed consent was obtained from children and their parents, and the study was approved by the hospital ethics committee.


Lung function varied within the group (the mean forced expiratory volume in one second was 64% of predicted; range 24-100%). There were no significant differences between the control oxygen saturations (mean 97.1%, range 93-99%; mean maximum intra-individual variation 1.3%, range 0.4%).

Desaturation occurred in all hypoxic states (P<0.002). There was no significant difference between oxygen saturations measured on the inward and outward flights (mean (range) 91.7% (87-96%) V 92.1% (88-96%)) or between repeated mountain measurements (92.8% (89-96%) v 92.5% (89-94%)).

In most cases there was significant correlation between the possible predictors and oxygen saturations on the outward flight and at the first measurement on the mountain (fig 1). In these cases the laboratory test had the highest coefficients of determination (r(sup2)=76% and 47% respectively; table I), and, although for the return flights a better prediction was obtained by using the mean of control saturations (laboratory, outward and return airport, and low altitude mountain resort), the best predictor from a single reading was still the laboratory test. For the group as a whole, however, the mean differences between actual and predicted percentage oxygen saturation for all predictors were clinically small and not significantly different (P>0.05).

Fig 1
Fig 1

Regression lines between oxygen saturations on outward flight and possible predictors (above) and between first oxygen saturations at altitude and predictors (below). Forced expiratory volume in one second is expressed as percentage of predicted normal values for sex and height


Mean error for the different variables in predicting hypoxic saturation at altitude in children with cystic fibrosis

View this table:

The laboratory test directly identified the six children who showed desaturation to less than 90%; it did not underestimate desaturation in any of them (table II). Spirometry and laboratory control saturations both identified only three out of six (four out of six when combined), with a maximum underestimation of desaturation of over 5%. The mean of control oxygen saturations identified three out of six; although the margin of error was narrower, desaturation was still underestimated by as much as 4%.


Recorded variables and their predicted hypoxic saturations in six children showing greatest desaturation. *Laboratory hypoxic values are used as direct indication of hypoxic response

View this table:

The laboratory test overpredicted desaturation during outward flight and on the mountain (first reading) by a mean of 1% (95% confidence interval 0.31 to 1.69%) and 2% (0.65 to 3.35%) respectively (fig 2). Overestimation of mountain desaturation was significantly greater at lower than at higher saturations (P=0.011; r=0.656) with a similar trend in overestimating values in flight (P=0.067; r=0.397).


Bland and Altman plots 66 showing that oxygen saturations measured were higher during flight and on mountain than was predicted by laboratory test. 95% Confidence intervals for mean hypoxic oxygen saturation 0.31 to 1.69 (laboratory plus in flight) and 0.73 to 3.41 (laboratory and mountain). Superimposed points have been separated symmetrically along x axis to help interpretation


The ability of a person to tolerate acute hypoxia depends on many factors other than lung disease, and the clinical importance of temporary hypoxaemia at altitude is uncertain. Early symptoms may be confused with travel sickness and may contribute to morbidity when combined with other travel stresses such as reduced humidity 7.

Current consensus for the minimum desirable arterial partial pressure of oxygen during a flight is an arbitrary value of 50 mm Hg (6.7 kPa).8 This corresponds to an oxygen saturation of about 83%. A higher value may be applicable for long flights or when patients have multiple medical problems. Hypoxaemia may be safely corrected by supplying oxygen through a Venturi mask or nasal cannula,9 and with advanced warning most airlines will provide supplementary oxygen.

It is important to identify which patients are at risk of serious desaturation at altitude, and a hypoxic challenge would seem a direct way of assessing individual response. In this group of children lung function and control oxygen saturations at sea level, particularly when a mean of repeated measurements was used, showed significant correlation with hypoxic desaturation; overall, they had small errors of prediction that were not significantly different from those of the laboratory test. A few children, however, showed more desaturation than expected and would not have been identified without the laboratory test. This may be explained by individual variation in bronchoconstriction,10 pulmonary vasoconstriction,11 and hyperventilation induced by hypoxia.

Danger of underestimation

The laboratory test may have overestimated desaturation because we used the maximum likely cabin altitude that would give the lowest likely inspired partial pressure of oxygen. In addition, reduced air density at altitude 12 and cold mountain air13 may have had a beneficial effect on lung function and hence altitude oxygen saturation. Overestimation would merely lead to unnecessary prescription of oxygen in flight. Unlike the results of the laboratory test, the other predictors made the more dangerous mistake of underestimating desaturation in some of the children who showed the greatest desaturation. The maximum underestimation of 4% to 5% oxygen saturation borders on clinical relevance, and the magnitude of this error would be greater if cabin altitude was slightly higher or if baseline saturations and results of spirometry were lower. This hypothesis is supported by the observation that the laboratory test overpredicted desaturation more at lower saturations. This is not unexpected because of the change in slope of the oxygen dissociation curve; at lower saturations the fall is greater per unit fall in alveolar partial pressure of oxygen.

In this small group the test was 100% sensitive and highly specific (91.75-100%) in identifying those at risk of showing desaturation below 90%. Although desaturation is a prerequisite for clinical problems, however, individual tolerance of hypoxaemia is extremely variable,14 so the test can identify only a group among whom clinical problems are more likely to arise. No child reached the current criteria for requiring additional inspired oxygen. Clearly children with more severe lung disease need to be studied to determine the

precise cut off in spirometric variables and control saturations below which a hypoxic stress test is desirable. If oxygen is prescribed then retention of carbon dioxide should be excluded.

The laboratory hypoxic test is a cheap and simple non-invasive test and is of practical value in determining fitness to fly in children with cystic fibrosis.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.