Statistical models of acute mountain sickness

Richard D. Vann, Neal W. Pollock, Carl F. Pieper, David R. Murdoch, Stephen R. Muza, Michael J. Natoli, Luke Y. Wang

Research output: Contribution to journalArticlepeer-review

7 Scopus citations


Acute mountain sickness (AMS) is caused by exposure to altitudes exceeding 2500 m and often resolves by acclimatization without further ascent. Statistical models of AMS score and the probability of an AMS diagnosis were developed to allow the combination of dissimilar exposures for simultaneous analysis. The study population was 302 trekkers from a previous investigation who provided self-reported symptoms upon arrival at 3840 m during hikes through altitudes of 1500 to 6200 m. AMS score (Hackett scale) was estimated by linear regression and the probability of an AMS diagnosis (Lake Louise criteria) by logistic regression. AMS score or probability was significantly associated with exposure day and altitude. Increased altitude over the prior 3 days resulted in higher estimated AMS score or probability and decreased altitude in lower score or probability. The odds ratio (OR) of AMS was 3.6 if not on acetazolamide. Females appeared slightly more susceptible than males (1.5 OR). The approach offers the advantages of (1) improved statistical power by combining exposures, (2) insight into the dose-response relationship of altitude exposure and AMS risk, (3) quantitative tests for the significance of factors that might affect AMS susceptibility, and (4) practical tools to track individual climbers and plan operational ascents.

Original languageEnglish (US)
Pages (from-to)32-42
Number of pages11
JournalHigh Altitude Medicine and Biology
Issue number1
StatePublished - Mar 2005
Externally publishedYes


  • Acclimatization
  • Acetazolamide
  • Altitude
  • Lake Louise
  • Probability

ASJC Scopus subject areas

  • Physiology
  • Public Health, Environmental and Occupational Health


Dive into the research topics of 'Statistical models of acute mountain sickness'. Together they form a unique fingerprint.

Cite this