Send to

Choose Destination
PLoS One. 2014 Jul 28;9(7):e100642. doi: 10.1371/journal.pone.0100642. eCollection 2014.

Risk prediction score for severe high altitude illness: a cohort study.

Author information

Université Paris Est, Faculté de Médecine, EA 4393, Laboratoire d'Investigation Clinique, Créteil, France; AP-HP, Groupe Henri Mondor-Albert Chenevier, Service de Santé Publique, Créteil, France.
AP-HP, Hôpital Avicenne, Service d'Anesthésie et Réanimation, Bobigny, France.
AP-HP, Hôpital Avicenne, Service de Physiologie, Explorations Fonctionnelles et Médecine du Sport, Bobigny, France.
Université Paris 13, Sorbonne Paris Cité, EA2363 « Réponses cellulaires et fonctionnelles à l'hypoxie », U.F.R. SMBH, Bobigny, France; AP-HP, Hôpital Avicenne, Service de Physiologie, Explorations Fonctionnelles et Médecine du Sport, Bobigny, France.



Risk prediction of acute mountain sickness, high altitude (HA) pulmonary or cerebral edema is currently based on clinical assessment. Our objective was to develop a risk prediction score of Severe High Altitude Illness (SHAI) combining clinical and physiological factors. Study population was 1017 sea-level subjects who performed a hypoxia exercise test before a stay at HA. The outcome was the occurrence of SHAI during HA exposure. Two scores were built, according to the presence (PRE, n = 537) or absence (ABS, n = 480) of previous experience at HA, using multivariate logistic regression. Calibration was evaluated by Hosmer-Lemeshow chisquare test and discrimination by Area Under ROC Curve (AUC) and Net Reclassification Index (NRI).


The score was a linear combination of history of SHAI, ventilatory and cardiac response to hypoxia at exercise, speed of ascent, desaturation during hypoxic exercise, history of migraine, geographical location, female sex, age under 46 and regular physical activity. In the PRE/ABS groups, the score ranged from 0 to 12/10, a cut-off of 5/5.5 gave a sensitivity of 87%/87% and a specificity of 82%/73%. Adding physiological variables via the hypoxic exercise test improved the discrimination ability of the models: AUC increased by 7% to 0.91 (95%CI: 0.87-0.93) and 17% to 0.89 (95%CI: 0.85-0.91), NRI was 30% and 54% in the PRE and ABS groups respectively. A score computed with ten clinical, environmental and physiological factors accurately predicted the risk of SHAI in a large cohort of sea-level residents visiting HA regions.

[Indexed for MEDLINE]
Free PMC Article

Supplemental Content

Full text links

Icon for Public Library of Science Icon for PubMed Central
Loading ...
Support Center