Prediction of acute mountain sicknessBMJ 1995; 311 doi: https://doi.org/10.1136/bmj.311.7011.989 (Published 14 October 1995) Cite this as: BMJ 1995;311:989
- Correspondence to: Dr Austin
Acute mountain sickness is a common problem when ascending above 3000 m.1 2 3 Attempts to predict its development have so far been unsuccessful, and the side effects of current prophylaxis limit its widespread use. We tried to develop a simple clinical method of prediction.
Subjects, methods, and results
Two groups of New Zealanders (23 in 1992 and 17 in 1993) went to Nepal, ascending to a maximum of 5640 m. Before they left they completed a questionnaire on illnesses related to mountain sickness and had their breath holding times and responses to hyperventilation and pharyngeal stimulation measured. Breath holding was measured in seconds with subjects seated after a maximal inspiration. The gag reflex was assessed by touching the posterior pharynx with a wooden spatula and scoring the response as 0 if there was no response, 1 if the response was mild, 2 if it was moderate, and 3 if it was severe. The subjects also hyperventilated maximally for one minute and assessed their own response on this four point scale.
Twice daily during the trek we completed a variant of the general high altitude questionnaire4 and measured breath holding times. Measurements were taken at the same time and while subjects were seated at each attempt. Subjects scoring 15 and under were classed as being well, those scoring 16-30 as having mild acute mountain sickness, and those scoring over 30 as having severe sickness, 15 being a traditional cut off point.2
The figure shows the relation between worst score on the questionnaire during the ascent and scores for the gag reflex and hyperventilation before the ascent. Nineteen people developed acute mountain sickness, nine of them severely, and 21 were well. One subject was extremely unwell (score 69) and was evacuated to Kathmandu. All 10 of those with a severe gag reflex (score of 2 or 3) developed acute mountain sickness (four mildly, six severely), but only nine of the 30 subjects with scores of 0 or 1 for the gag reflex developed the sickness. All six subjects with severe dizziness on hyperventilation (score of 3) developed acute mountain sickness (three mildly, three severely) compared with only nine of the 26 who did not become dizzy on hyperventilation or only mildly so.
With Fisher's exact test for independence the highest predictors of acute mountain sickness were gag reflex (P=0.002), breath holding (P=0.041), and hyperventilation (P=0.083). Using these univariate predictors we were able correctly to predict 78% of cases by the gag reflex ((10+21)/40), 63% by hyperventilation ((9+16)/40), 63% by breath holding ((6+19)/40), and 85% by combining these in a multivariate model ((15+19)/40).
Although the groups were small, we found that those who developed acute mountain sickness were obviously different, and we believe that the results have clinical applications.
The fact that acute mountain sickness seems to be associated with the gag reflex, breath holding, and hyperventilation suggests that differences between people in the control of the respiratory pattern are important factors in the pathogenesis of acute mountain sickness. Guz et al noted that blocking the vagus and glossopharyngeal nerves was associated with prolonged breath holding,5 and the strength of the gag reflex may be an indicator of this modulation of respiratory pattern.
Response to hyperventilation has been studied extensively in elite mountaineers, and they have been found to tolerate hypocapnia. Though hard to standardise, hyperventilation for one minute at low altitude seems to be a useful estimate of tolerance of hypocapnia and the subsequent development of acute mountain sickness.
It does appear possible to distinguish those at risk of developing acute mountain sickness. If a hypersensitive gag reflex, extreme dizziness on hyperventilation, or short breath holding time is found, these people should be observed closely. In view of this study the role of predictive tests deserves further evaluation.
Conflict of interest None.