Physiological Risk Factors for Severe High-Altitude IllnessA Prospective Cohort Study

Jean-Paul Richalet1,2, Philippe Larmignat1,3, Eric Poitrine1, Murielle Letournel2, andFlorence Canouı-Poitrine4,5

1Universite Paris 13, EA 2363 “Reponses Cellulaires et Fonctionnelles a l’Hypoxie,” Association pour la Recherche en Physiologie de l’Environnement,

Unite de Formation et de Recherche Sante Medecine Biologie Humaine, Bobigny; 2Assistance Publique-Hopitaux de Paris (AP-HP), Hopital Avicenne,

Service de Physiologie, Explorations Fonctionnelles et Medecine du Sport, Bobigny; 3AP-HP, Hopital Avicenne, Service d’Anesthesie et Reanimation,Bobigny; 4Universite Paris Est, Faculte de Medecine, Laboratoire d’Investigation Clinique, EA 4393, Creteil; and 5AP-HP, Groupe Henri Mondor-

Albert Chenevier, Service de Sante Publique, Creteil, France

Rationale: An increasing number of persons, exposed to high alti-tude for leisure, sport, or work,may suffer from severe high-altitudeillness.Objectives: To assess, in a large cohort of subjects, the associationbetweenphysiological parameters and the risk of altitude illness andtheir discrimination ability in a risk prediction model.Methods: A total of 1,326 persons went through a hypoxic exercisetestbeforea sojournabove4,000m.Theywere thenmonitoredupathigh altitude and classified as suffering from severe high-altitudeillness (SHAI) or not. Analysis was stratified according to acetazol-amide use.Measurements and Main Results: Severe acute mountain sicknessoccurred in 314 (23.7%), high-altitude pulmonary edema in 22(1.7%), and high-altitude cerebral edema in 13 (0.98%) patients.Among nonacetazolamide users (n ¼ 917), main factors indepen-dently associatedwith SHAI were previous history of SHAI (adjustedodds ratios [aOR], 12.82; 95% confidence interval [CI], 6.95–23.66;P, 0.001), ascent greater than 400m/day (aOR, 5.89; 95%CI, 3.78–9.16; P, 0.001), history of migraine (aOR, 2.28; 95% CI, 1.28–4.07;P¼ 0.005), ventilatory response to hypoxia at exercise less than 0.78L/minute/kg (aOR, 6.68; 95% CI, 3.83–11.63; P , 0.001), and desa-turation at exercise in hypoxia equal to or greater than 22% (aOR,2.50; 95% CI, 1.52–4.11; P , 0.001). The last two parameters im-proved substantially the discrimination ability of the multivariatepredictionmodel (C-statistic rose from 0.81 to 0.88; P, 0.001). Pre-ventive use of acetazolamide reduced the relative risk of SHAI by44%.Conclusions: In a large population of altitude visitors, chemosensi-tivity parameters (highdesaturationand lowventilatory response tohypoxia at exercise) were independent predictors of severe high-altitude illness. They improved the discrimination ability of a riskprediction model.

Keywords: acute mountain sickness; high-altitude pulmonary edema;

high-altitude cerebral edema; ventilatory response to hypoxia

An increasing number of sea-level residents visit areas above 4,000m of altitude for leisure or work. For example, in 2010, 70,218

trekkers and mountaineers visited Nepal (1), and 25,000 trekkersclimbed Mount Kilimanjaro (5,895 m). Unacclimatized subjectsmay suffer from acute mountain sickness (AMS), high-altitudepulmonary edema (HAPE), or high-altitude cerebral edema(HACE) (2). Several risk factors that have been identified, inparticular speed of ascent greater than 300 to 500 m/day in theacclimatization period (2–4), previous history of AMS or HAPE(2, 3, 5, 6), obesity (7, 8), migraine (9, 10), persistence of a patentforamen ovale (11), Down syndrome (12), congenital pulmonaryabnormalities (13), perinatal pulmonary vascular insult (14), andHolmes-Adie syndrome (15), are clinical conditions associatedwith susceptibility to AMS or HAPE. However, in most previousstudies, the concept of individual susceptibility did not rely onobjective physiological measurements.

Considering that the risk is linked to the severity of the hyp-oxic exposure and/or the efficiency of the physiological adaptiveresponses, the ventilatory response to hypoxia and the decreasein arterial oxygen saturation, at rest or exercise, have been pre-viously suggested as physiological risk factors for AMS (10, 16–23). Pulmonary vascular reactivity to hypoxia and/or exercisehas been evaluated to assess the individual susceptibility toHAPE (21, 24, 25), as well as alveolar transepithelial sodiumtransport (26). Subjects with less compliant cerebrospinal fluidsystems showed a greater increase in intracranial pressure fora given increase in brain volume as a result of hypoxic cerebraledema and could be therefore more prone to develop AMS orHACE (“tight-fit” hypothesis) (2, 27), but no epidemiologicalstudy is available. A genetic basis of high-altitude intolerancehas been looked for in more recent studies, exploring factorsinvolved in physiological responses to hypoxia (28, 29). Alto-gether, although ventilatory response to hypoxia has been

(Received in original form August 4, 2011; accepted in final form October 16, 2011)

Author Contributions: J.-P.R. designed the study, performed the clinical evaluation

and the physiological measurements and the calculation of parameters, and

wrote the manuscript. P.L. performed the clinical evaluation and the physiolog-

ical measurements and revised the manuscript. E.P. participated in the manage-

ment and analysis of the data and revised the manuscript. M.L. participated in the

physiological measurements. F.C.-P. designed the epidemiological aspect of the

study, performed the statistical analysis, and wrote part of the manuscript.

Correspondence and requests for reprints should be addressed to J.-P. Richalet,

M.D., Ph.D., UFR SMBH, 74 rue Marcel Cachin, 93017 Bobigny Cedex, France.

E-mail: richalet@smbh.univ-paris13.fr

Am J Respir Crit Care Med Vol 185, Iss. 2, pp 192–198, Jan 15, 2012

Copyright ª 2012 by the American Thoracic Society

Originally Published in Press as DOI: 10.1164/rccm.201108-1396OC on October 27, 2011

Internet address: www.atsjournals.org

AT A GLANCE COMMENTARY

Scientific Knowledge on the Subject

Clinical risk factors for high altitude–related diseases suchas acute mountain sickness and high-altitude pulmonary orcerebral edema are known but no clear physiological pa-rameter has been associated with an increased risk forthese diseases.

What This Study Adds to the Field

This study performed in a large cohort of subjects (1,326)suggests an association between low ventilatory response tohypoxia at exercise and the risk of developing severe highaltitude–related illnesses. This knowledge may help in thedetection of high-risk subjects before a sojourn at high alti-tude and facilitate the prevention of life-threatening diseases.

frequently found altered in subjects with a history of high-altitude intolerance, results are contradictory, studies are basedon a small number of subjects, and no definitive clear answer isavailable (18, 25, 30).

The aim of this study was to assess the association betweenphysiological parameters exploring peripheral chemosensitivityand the risk of incapacitating severe high-altitude illness (SHAI),regrouping severe AMS, HAPE, and HACE, and their discrim-ination ability in a risk prediction model.

METHODS

Patients

During a 17-year period, from 1992 to 2008, 3,994 sea-level residentswho underwent, for the first time, an outpatient altitude medicine con-sultation before a sojourn of at least 3 days above 4,000 m of altitudewith overnight sleeping above 3,500 m were included in the study. In-dividual informed consent was obtained.

Baseline Examination

General characteristics. Subjects were questioned about their sociode-mographic characteristics, personal and familial medical history, treat-ments, objective of the sojourn (tourism, mountaineering, trekking, orwork), and usual physical and mountaineering activity. Tolerance tohigh altitude was evaluated on the basis of a questionnaire about pre-vious experience of sojourn at altitudes above 3,500 m, looking for theoccurrence of severe AMS, peripheral edema, HAPE, or HACE. Sub-jects were classified as susceptible if they had previously experienced ontwo or more occasions severe AMS, HAPE, or HACE during a sojournat high altitude (minimal altitude reached, 4,000 m; minimal sleepingaltitude, 3,500 m). Their level of susceptibility was considered undeter-mined if they never had been to high altitude before inclusion in thestudy. Regular physical activity was defined as at least 30 minutes, threetimes per week, of endurance-based physical activity. Regular moun-taineering was defined as at least 40 days/year.

Physiological measurements. Subjects went through a routine hypoxicexercise test (29, 31), modified from a previous version (10, 19). Thehypoxic exercise test consisted of four 4-minute phases: rest in nor-moxia, rest in hypoxia (fraction of inspired oxygen, 0.115), exercisein hypoxia, and exercise in normoxia (Figure 1). The operative poweroutput used during the two exercise phases was the same and fixed toapproximately 30% of maximal normoxic oxygen consumption, basedon the value of theoretical maximal heart rate (HR, beats/min). ECGwas monitored, and minute ventilation (V

:

E, L/min) was measured witha metabograph (Oxycon; Jaeger, Wuerzburg, Germany). Arterial oxygensaturation (Sa, %) was measured by transcutaneous oximetry at an earlobe prewarmed with a vasodilating cream. Desaturation and ventilatoryand cardiac responses to hypoxia were calculated, respectively, at rest (r)and exercise (e), using the mean values of HR, V

:

E, and Sa during the last30 seconds of each phase:

Desaturation at rest: DSar ¼ Sarn – Sarh (%)Desaturation at exercise: DSae ¼ Saen – Saeh (%)Ventilatory response at rest: HVRr ¼ (V

:

Erh – V:

Ern)/DSar/BW3 100 (L/min/kg)

Ventilatory response at exercise: HVRe ¼ (V:

Eeh – V:

Een)/DSae/BW 3100 (L/min/kg)

Cardiac response at rest: HCRr ¼ (HRrh – HRrn)/DSar (beats/min/%)Cardiac response during exercise: HCRe ¼ (HReh – HRen)/DSae (beats/

min/%),

where indices rh, rn, eh, and en stand, respectively, for rest in hypoxia,rest in normoxia, exercise in hypoxia, and exercise in normoxia; BWrepresents body weight (kg). In a fifth phase, power output in normoxiawas progressively incremented to reach the same HR as in the hypoxicexercise phase. This allowed calculating the decrease in power output(DPower, W) induced by hypoxia for the same level of HR. This de-crease in power output was related to the correspondent decrease insaturation induced by hypoxia at exercise, leading to a relative de-crease in power output, DPower/DSae (W/%).

Follow-up and Ascertainment of Cases

Outcome was the occurrence of severe AMS, HAPE, or HACE duringthe high-altitude sojourn. Each subject was asked to fill out a question-naire based on Hackett’s score of AMS and symptoms of HAPE orHACE during their future stay at high altitude and to send it back afterreturning home (3). Moderate AMS was defined by a Hackett’s scoreless than 6, without significant impact on planned activity. Severe AMSwas defined by a Hackett’s score of at least 6, with incapacitatingsymptoms leading to a stop or a significant reduction of planned activ-ity. HAPE was defined by the presence of clinical signs of respiratorydistress (dyspnea, cyanosis, rales), confirmed by thorax X-ray on de-scent to low altitude. HACE was defined by clinical signs of neurolog-ical deficit (ataxia, mental confusion). The diagnosis of HAPE orHACE was always confirmed by an expert, either on the spot, wherethe disorder occurred, or later on when hospitalized. Peripheral edemawas defined as the presence of subcutaneous edema in at least twolocations. Subjects were asked to note any medication, especially acet-azolamide (ACZ), taken during their sojourn. From the questionnaire,subjects were classified as having no or moderate AMS, severe AMS,HAPE, or HACE. Subjects with severe AMS, HAPE, and HACE werepooled as having severe high-altitude illness (SHAI) and consideredintolerant to high altitude. Some subjects may have suffered from twoor three of the above-mentioned conditions. The common feature ofthis group was the presence of incapacitating symptoms leading toa significant reduction of planned activity. Information about the as-cent characteristics (daily speed of ascent, altitude reached) was col-lected in the follow-up questionnaire.

Statistical Analysis

The baseline characteristics of subjects who developed SHAI and ofthose who remained free of SHAI during high-altitude sojourn werecompared by Student t test or Wilcoxon-Mann-Whitney as appropriateand Pearson x2 test or Fisher test as appropriate for quantitative andqualitative variables, respectively. Quantitative variables are reported asmean (SD) or median (interquartile range) as appropriate. Qualitativevariables are reported as n (%). In univariate analysis, odds ratios (ORs)for SHAI associated with each factor were estimated in logistic regres-sion; ORs of quantitative variables were calculated for a 1-SD increaseor decrease estimated in the overall cohort. Given the nonnormality

Figure 1. Hypoxia exercise test. RN, RH, EH, and EN represent, respec-

tively, rest in normoxia, rest in hypoxia, exercise in hypoxia, and exer-

cise in normoxia. Hypoxia was produced by breathing a normobaric

hypoxic gas mixture (fraction of inspired oxygen, 0.115). Exercise wasperformed at approximately 30% of maximal power output in phases

EH and EN. In phase EN1, power output was adjusted so that the heart

rate reached the same value as in EH. HVRe ¼ ventilatory response to

hypoxia during exercise; Sa, SaO2 ¼ arterial oxygen saturation; Sae ¼arterial oxygen saturation during exercise; Sar ¼ arterial oxygen satu-

ration at rest; V:

E ¼ pulmonary ventilation; V:

Ee ¼ pulmonary ventilation

during exercise. Dotted line, V:

E; solid line, Sa.

Richalet, Larmignat, Poitrine, et al.: Physiological Risk Factors of Altitude Illness 193

distribution, physiological parameters were dichotomized around themedian value estimated in non-SHAI and non-ACZ user subjects.

Two-by-two analyzes were used to assess first-order interactions andconfounding by fitting multiplicative models. Variables with P, 0.15 inthe univariate analysis were included in a multivariate logistic regres-sion model. To assess the clinical utility of physiological parameters, weevaluated whether their addition in a risk model with clinical riskfactors improved the ability of that model to discriminate betweenthose who will develop SHAI from those who will not. For that pur-pose, we compared C-statistics between nested models with and with-out physiological parameters (32).

A supplementary analysis was done to estimate the effect of preven-tive ACZ use on SHAI risk within the entire cohort. As it was an ob-servational study and not a randomized trial, we built a propensity score,defined as the conditional probability of being treated given the initialcharacteristics of subjects, to balance the characteristics in the twogroups (treated and nontreated).

All tests were two-tailed, and a P value of less than 0.05 was con-sidered statistically significant. Analyzes were performed with Statasoftware version 11.0 (Stata Inc., College Station, TX).

RESULTS

Among the 3,994 persons who underwent an outpatient altitudemedicine consultation before their high-altitude sojourn, 1,326returned the questionnaire (33.2%). There was no major differ-ence between respondents and nonrespondents. However,respondents were slightly older than nonrespondents, less fre-quently current smokers, had a more frequent history of SHAI,greater mountaineering experience and physical activity, anda lower body mass index. They showed higher systolic and dia-stolic blood pressure, slightly higher desaturation at exercise andcardiac and ventilatory responses to hypoxia at rest and exercise.No significant association was found between physiological var-iables and year of examination.

Frequency of Severe High-altitude Illness

Among the 1,326 respondents, 318 experienced severe altitudeillness during their high-altitude sojourn (24.0%; 95% confi-dence interval [CI], 21.1–26.7%), 213 among nonusers ofACZ (23.2%; 95% CI, 20.5–26.1%), and 105 among ACZusers (25.7%; 95% CI, 21.5–30.2%). The frequency of the var-ious forms of diseases related to high-altitude exposure is pre-sented in Table 1 in subjects with or without preventive use ofACZ.

Risk Factors for Severe High-altitude Illness

With univariate analysis (Table 2), younger age, female sex,history of migraine, regular mountaineering, regular physical ac-tivity, previous history of SHAI, and rapid ascent were signifi-cantly associated with SHAI. Aconcagua, the Alps, Kilimanjaro,

and Ladakh were associated with increased risk of SHAI com-pared with the Annapurna tour (reference). High DSar (.12%)and DSae (>22%) and low HCRr (,1.07 beats/min/%), HVRr

(,0.32 L/min/kg), HCRe (,0.84 beats/min/%), and HVRe

(,0.78 L/min/kg) were significantly associated with SHAI (Table2). A high hypoxia-induced decrease in power output (DPower/DSae . 2 W/%) was also associated with SHAI. The associationof DSae and SHAI was modified by the use of preventive ACZ(P interaction ¼ 0.067). Therefore, to take into account thisinteraction, we stratified the analysis on preventive acetazol-amide use.

The results of the multivariate analysis stratified according topreventive ACZ use are shown in Figure 2. Among non-ACZusers, factors remaining significantly associated with SHAI wereyoung age (,46 yr; P ¼ 0.045), female sex (P ¼ 0.036), history ofmigraine (P ¼ 0.005), regular physical activity (P ¼ 0.049), pre-vious history of severe altitude illness (P, 0.001), ascent greaterthan 400 m/day (P , 0.001), cardiac response to hypoxia duringexercise (HCRe , 0.84 beats/min/%; P ¼ 0.001), desaturationduring exercise in hypoxia (DSae > 22%; P , 0.001), and ven-tilatory response to hypoxia at exercise (HVRe , 0.78 L/min/kg;P , 0.001). Moreover, Ladakh (P ¼ 0.025) and the Alps (MontBlanc) (borderline significance; P ¼ 0.051) were associated withSHAI. There was a trend for Aconcagua (P ¼ 0.11).

Among ACZ users, young age, female sex, history of mi-graine, regular physical activity, HCRe, DSae, and Alps wereno longer significant. Ladakh remained borderline significant(P ¼ 0.052). Other factors, that is, history of SHAI, rapid ascent,and HVRe, were still associated with SHAI but less stronglythan in non-ACZ users.

Contribution of Physiological Parameters (DSae and HVRe)

to SHAI Risk Prediction

The C-statistics of an SHAI prediction model that included sex,history of migraine, history of SHAI, regular physical activity,rapid ascent, and geographical location were 0.806 and 0.722in subjects without and with preventive use of ACZ,respectively (Table 3). The C-statistics reached 0.879 (P ,0.001) and 0.769 (P ¼ 0.005), respectively, when DSae andHVRe were simultaneously added to the model (M5). The ad-dition of HCRe to the model did not significantly improve therisk prediction (0.883; P ¼ 0.16).

Supplementary analysis. Preventive use of ACZ was associ-ated with a lower risk of SHAI (adjusted OR, 0.56; 95% CI,0.40–0.80; P ¼ 0.001) in multivariate analysis including adjust-ment for propensity score.

DISCUSSION

This is the largest epidemiological study to date of subjects ex-posed to high altitude–related illness who were previously

TABLE 1. FREQUENCY OF SEVERE HIGH-ALTITUDE ILLNESS ACCORDING TO PREVENTIVE USE OFACETAZOLAMIDE

Overall

(n ¼ 1,326)

No ACZ Use

(n ¼ 917)

ACZ Use

(n ¼ 409) P Value*

Severe acute mountain sickness 314 (23.7); 95% CI, 21.4–26.0 211 (23.0) 103 (25.2) 0.39

High-altitude pulmonary edema 22 (1.7); 95% CI, 1.0–2.5 16 (1.7) 6 (1.5) 0.72

High-altitude cerebral edema 13 (0.98); 95% CI, 0.5–1.7 2 (0.5) 11 (1.2) 0.37

Severe high-altitude illness† 318 (24); 95% CI, 21.1–26.7 213 (23.2) 105 (25.7) 0.34

Definition of abbreviations: ACZ ¼ acetazolamide; CI ¼ confidence interval.

Data represent number of subjects (%); 95% CI.

*No ACZ use versus ACZ use: Pearson x2 or Fisher test as appropriate.yMultiple diagnoses possible.

194 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 185 2012

TABLE 2. CHARACTERISTICS OF SUBJECTS ACCORDING TO SEVERE HIGH-ALTITUDE ILLNESS

No Severe High-altitude

Illness (n ¼ 1,008)

Severe High-altitude

Illness (n ¼ 318) P Value* Crude OR (95% CI)†

Baseline characteristics

Age, yr 45.3 (14.1) 42.6 (12.8) 0.003 0.82 (0.72–0.93)

Women 391 (38.8) 151 (47.5) 0.01 1.43 (1.11–1.84)

Current smoking 102 (10.1) 22 (6.9) 0.09 0.66 (0.41–1.07)

Height, cm 171.0 (8.7) 170.4 (8.7) 0.30 0.93 (0.82–1.06)

Weight, kg 66.9 (11.4) 65.4 (10.6) 0.04 0.86 (0.75–0.99)

Body mass index, kg/m2 22.8 (2.7) 22.5 (2.70) 0.06 0.97 (0.76–1.03)

Systolic blood pressure, mm Hg 130.7 (14.2) 130.2 (12.9) 0.55 0.96 (0.84–1.10)

Diastolic blood pressure, mm Hg 82.1 (9.6) 81.5 (9.6) 0.38 0.94 (0.83–1.08)

Medical history

Hypothyroidism 16 (1.59) 3 (0.94) 0.40 0.59 (0.17–2.04)

Depression 11 (1.09) 6 (1.9) 0.27 1.74 (0.64–4.75)

Cranial injury with loss of consciousness 13 (1.29) 6 (1.89) 0.42 1.47 (0.55–3.90)

Hypertension 58 (5.75) 11 (3.5) 0.11 0.59 (0.30–1.13)

Coronary heart disease 5 (0.5) 0 (0.0) 0.60 —

Rhythm disorders 25 (2.5) 6 (1.9) 0.54 0.76 (0.31–1.86)

Raynaud 16 (1.6) 7 (2.20) 0.47 1.39 (0.57–3.42)

Hypercholesterolemia 56 (5.6) 8 (2.5) 0.03 0.43 (0.21–0.93)

Asthma 40 (4.0) 13 (4.1) 0.92 1.03 (0.54–1.95)

Upper airway infections 18 (1.8) 5 (1.6) 0.80 0.88 (0.32–2.39)

Bronchopneumopathy 43 (4.3) 13 (4.1) 0.89 0.96 (0.51–1.80)

Pneumothorax 8 (0.8) 2 (0.6) 1.00 0.79 (0.17–3.74)

Thromboembolism 77 (7.6) 29 (9.1) 0.40 1.21 (0.78–1.90)

Migraine 111 (11.0) 61 (19.2) ,0.001 1.91 (1.36–2.69)

Premenstrual syndrome (in nonmenopausal women, n ¼ 774) 21 (13.0) 14 (19.2) 0.22 1.58 (0.75–3.32)

Snoring 203 (20.2) 62 (19.6) 0.82 0.96 (0.70–1.32)

Physical and mountaineering activity

Physical activity

None or occasional 706 (70.1) 194 (61.0)0.002

1.00 (ref)

Frequent 301 (29.9) 124 (39.0) 1.50 (1.15–1.95)

Mountaineering activity

None or occasional 922 (91.7) 287 (88.3)0.20

1.00 (ref)

Frequent 84 (8.3) 34 (10.7) 1.31 (0.86–2.00)

Previous maximal altitude reached by daytime, m 4,079 (1,196) 4,163 (1,082) 0.27 1.07 (0.95–1.21)

Previous maximal altitude reached for sleeping, m 3,190 (1,257) 3,296 (1,154) 0.18 1.09 (0.96–1.23)

Previous history of severe high-altitude illness

No 420 (41.7) 74 (23.3) 1.00 (ref)

Yes 103 (10.2) 132 (41.5) ,0.001 7.27 (5.09–10.4)

Undetermined (no experience of high altitude) 485 (48.1) 112 (35.2) 1.31 (0.95–1.81)

Altitude sojourn

Type of activity

Mountaineering 128 (12.7) 53 (16.7) 1.00 (ref)

Tourism 109 (10.8) 25 (7.9),0.001

0.55 (0.32–0.95)

Work 166 (16.5) 25 (7.9) 0.36 (0.21–0.62)

Trekking 605 (60.0) 215 (67.6) 0.86 (0.60–1.23)

Location of objective

Aconcagua 27 (2.7) 15 (4.7) 2.54 (1.17–5.51)

Andes (other than Aconcagua) 239 (23.8) 60 (18.9) 1.15 (0.67–1.95)

Alps (Mont Blanc) 52 (5.2) 33 (10.4) 2.90 (1.55–5.43)

Kilimanjaro 119 (11.8) 41 (12.9) 1.57 (0.89–2.79)

Ladakh 63 (6.3) 35 (11.0),0.001

2.54 (1.38–4.68)

Annapurna tour 105 (10.4) 23 (7.2) 1.00 (ref)

Everest BC 30 (3.0) 12 (3.8) 1.82 (0.81–4.09)

Nepal (other than Annapurna tour or Everest BC) 124 (12.3) 47 (14.8) 1.73 (0.99–3.04)

Tibet 76 (7.6) 22 (6.9) 1.32 (0.69–2.54)

Others 171 (17.0) 30 (9.4) 0.80 (0.44–1.45)

Events during altitude sojourn

Maximal altitude reached, m 5,075 (1,056) 5,092 (982) 0.80 1.02 (0.90–1.15)

Rapid ascent (.400 m/d) 256 (26.0) 154 (49.4) ,0.001 2.78 (2.13–3.61)

Hypoxic test parameters

Desaturation at rest, DSar > 12% 415 (41.0) 189 (58.9) ,0.001 2.05 (1.59–2.66))

Desaturation at exercise, DSae > 22% 534 (53.4) 268 (84.3) ,0.001 4.68 (3.37–6.48)

Cardiac response at rest, HCRr , 1.07 beats/min/% 517 (52.7) 205 (65.9) ,0.001 1.74 (1.33–2.26)

Cardiac response at exercise, HCRe , 0.84 beats/min/% 492 (49.4) 193 (60.9) ,0.001 1.59 (1.23–2.06)

Ventilatory response at rest, HVRr , 0.32 L/min/kg 499 (50.5) 201 (64.0) ,0.001 1.74 (1.34–2.26)

Ventilatory response at exercise, HVRe , 0.78 L/min/kg 503 (50.7) 278 (87.4) ,0.001 6.76 (4.74–9.63)

Decrease in power output (n ¼ 787) . 2 W/% 301 (51.6) 122 (63.5) 0.005 1.63 (1.17–2.28)

Definition of abbreviations: BC ¼ base camp; CI ¼ confidence interval; HCRe ¼ cardiac response to hypoxia during exercise; HCRr ¼ cardiac response to hypoxia during

rest; HVRe ¼ ventilatory response to hypoxia during exercise; HVRr ¼ ventilatory response to hypoxia during rest; OR ¼ odds ratio; ref ¼ reference value; DSae ¼ change

in arterial oxygen saturation during exercise; DSar ¼ change in arterial oxygen saturation during rest.

Data represent number of patients (n, %), mean (SD), and median and interquartile range (for hypoxic test parameters) as appropriate.

* Pearson x2 test or Fisher test as appropriate for qualitative variables; Student t test or Wilcoxon-Mann-Whitney test for quantitative variables.yUnivariate logistic regression.

g

g

g

g

�

Richalet, Larmignat, Poitrine, et al.: Physiological Risk Factors of Altitude Illness 195

studied for their response to hypoxia. These results suggest thathigh DSae (>22%) and low HVRe (,0.78 L/min/kg) are inde-pendent risk factors for SHAI, in addition to young age, femalesex, regular physical activity, rapid ascent, history of severealtitude illness, history of migraine, and geographical location(Ladakh). Two physiological parameters (DSae and HVRe) sub-stantially increased the discrimination accuracy of an SHAI riskprediction model. Among ACZ users, young age, female sex,history of migraine, regular physical activity, HCRe, DSae, andAlps were no longer significant. In previous studies, the uncer-tainty of the clinical diagnosis led to possible misclassification ofthe subjects because the symptoms of AMS are unspecific andmainly self-related. We therefore considered only the severe

manifestations of AMS, HAPE, and HACE, with impairmentof the individual’s autonomy, leaving the light and moderatecases of AMS as normal manifestations of incomplete acclima-tization to hypoxia.

History of migraine and previous episodes of SHAI are riskfactors already mentioned in several studies (2, 3, 5, 9, 10).Although the mechanistic link remains to be established, thisrelationship between migraine and altitude illness might be sug-gestive of the “tight-fit” hypothesis (2, 27). Interestingly, therisk of SHAI for subjects with migraine disappeared in subjectstaking ACZ (Figure 2). However, it should be pointed out thathypoxia itself could trigger episodes of migraine, independentlyof AMS. Moreover, as it was not a randomized study, the

Figure 2. Clinical and physiological risk factors for severe high-altitude illness (SHAI). Adjusted odds ratios (ORs) and 95% confidence intervals (CIs)

for SHAI are stratified according to preventive use of acetazolamide. Adjusted ORs were assessed by multivariate logistic regression with adjustmentfor all variables listed in the figure plus year of inclusion. Threshold for quantitative parameters are medians calculated on the nonacetazolamide user

group. yAll other locations taken as reference. HCRe ¼ cardiac response to hypoxia during exercise; HVRe ¼ ventilatory response to hypoxia during

exercise; DSae ¼ change in arterial oxygen saturation during exercise.

TABLE 3. COMPARISON BETWEEN C-STATISTICS OF MULTIVARIABLE MODELS PREDICTING SEVERE HIGH-ALTITUDE ILLNESS

No ACZ Use (n ¼ 917) ACZ Use (n ¼ 409)

C-statistic* P Value† C-statistic* P Value†

Model 1 (M1): Multivariable model adjusted for previous

history of altitude illness, rapid ascent, physical activity,

age, sex, history of migraine, and location

0.806 Reference 0.722 Reference

M2 ¼ M1 1 HCRe 0.824 0.01 0.718 0.43

M3 ¼ M1 1 DSae 0.851 ,0.001 0.749 0.02

M4 ¼ M1 1 HVRe 0.865 ,0.001 0.766 0.008

M5 ¼ M1 1 HVRe and DSae 0.879 ,0.001 0.769 0.005; M5 vs. M4, 0.55

M6 ¼ M1 1 HVRe, DSae, and HCRe 0.883 ,0.001; M6 vs. M5, 0.16 0.770 0.004; M6 vs. M4, 0.53

Definition of abbreviations: ACZ ¼ acetazolamide; HCRe ¼ cardiac response to hypoxia during exercise; HVRe ¼ ventilatory response to hypoxia during exercise; DSae ¼change in arterial oxygen desaturation during exercise.

* C-statistic indexes were derived from logistic regression model.yComparison of C-statistics of M2, M3, M4, and M5 versus M1 was done by Delong test.

196 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 185 2012

causality link must be interpreted cautiously as subjects whotook ACZ had different characteristics (age, sex, physiologicalparameters) than subjects who did not. This study suggests thatfrequent physical activity is a risk factor for SHAI, in line witha common thought in the mountaineering world that a highlevel of maximal oxygen uptake is not a predictor of success inhigh-altitude expeditions (33). In fact, trained subjects presentmore severe desaturation induced by exercise in hypoxia, lead-ing to a more pronounced decrease in aerobic performancecompared with sedentary subjects (34). Similarly, we foundthat the hypoxia-induced decrease in performance (DPower/DSae) was associated with a higher risk for SHAI in univariateanalysis (it was not introduced in the multivariate analysisbecause this parameter was not measured in the whole popu-lation). High-altitude visitors obviously should not stop train-ing before an expedition, but should realize that intenseaerobic training is not a protective factor against altitude-related disorders. More frequent AMS in female than in malesubjects has been reported in some previous studies (8, 10) butnot in others (5). In the present study, sex (female) was inde-pendently associated with SHAI only in non-ACZ users. Bodyweight and body mass index were not found to be independentrisk factors, in contrast with others, who found an associationbetween obesity and AMS (7, 8, 10). However, no obese per-son was included in the present study. As previously suggested,a protective role of advanced age (>46) was observed in non-ACZ users (35). Severe desaturation at exercise in hypoxia(DSae > 22%) and low cardiac (HCRe , 0.84 beats/min/%)and ventilatory response to hypoxia (HVRe , 0.78 L/min/kg)at exercise were independent physiological risk factorswhereas all resting physiological parameters disappeared inmultivariate analysis, probably because of high interindividualvariability. Only DSae and HVRe improved the discriminationability of the risk prediction model. The rationale of these twoexercise parameters being risk factors for SHAI, especiallyamong non-ACZ users, has already been mentioned: Subjectspresenting a low response to hypoxia and low arterial satura-tion during exercise will be exposed to more severe hypoxemiaduring physical activity at altitude and will be more prone todevelop severe AMS, HAPE, and HACE because the level ofhypoxia is directly linked to the incidence of high altitude–related disorders (36).

This is the first study suggesting an independent associationbetween the geographical location of ascent and SHAI. Whenadjusted for all other risk factors, especially rate of ascent,one location, Ladakh, remained a higher risk for ACZ andnon-ACZ users. No clear explanation, linked to the climate orthe difficulty of the terrain, is available although many informalreports mention the higher risk of this location.

Although it was not a randomized controlled trial, this studyconfirms in a large number of subjects the efficacy of the preven-tive use of ACZ in high altitude–related illness, with a relativereduction of SHAI risk of about 44% (3, 37, 38). The fact thatthe frequency of the various forms of high-altitude illnesses wasnot different between ACZ users and nonusers (Table 1) wasdue to the difference between characteristics of treated andnontreated subjects. The preventive use of ACZ may reducethe risk of SHAI in susceptible subjects to the same level ofnonsusceptible subjects.

The multivariate analysis showed that in subjects takingACZ, several factors (age, sex, history of migraine, regular phys-ical activity, HCRe, DSae, Alps, and Aconcagua) were no longerassociated with SHAI. However, it seems important to note thatexcessive speed of ascent, history of SHAI, and low HVRe

remain risk factors even in subjects taking ACZ, although therisk is substantially reduced.

Adding physiological parameters to the well-known clinicalrisk factors is useful in the prevention of SHAI, justifying theuse of hypoxic exercise testing in subjects intending to undergophysical activity at high altitude. This testing should be proposedto two categories of subjects: first, subjects with no previous ex-perience of high altitude, and therefore lacking major informa-tion among risk factors; second, subjects who experienced severesymptoms in previous exposures to high altitude, to determinewhether these episodes of SHAI were due partly or mainly tophysiological characteristics that will make them prone to SHAIwhen reexposed.

The strength of the current study was the prospective design,the use of a validated criterion to assess SHAI, and the large pop-ulation with measurement of physiological response to hypoxia.However, the reported results should be interpreted with somelimitations in mind. The response rate was low, leading to a pos-sible selection bias. Classification of SHAI was based on a self-evaluation without immediatemedical control, leading to possibleclassification bias; however, an expert using a validated scoremade the adjudication. Similarly, information concerning speedof ascent and use of medication was self-declarative and shouldbe subject to caution. We used the Hackett’s score because whenthe study was started, in 1991, the Lake Louise score was not yet inuse for epidemiological studies of AMS. However, the Hackett’sscore has been highly correlated to the Lake Louise score andEnvironmental Symptoms Questionnaire for the clinical evalua-tion of AMS (39). Last, this study was conducted on the basis ofa single outpatient altitude medical consultation and could bereplicated in other populations, possibly leading to a predictionequation.

In conclusion, in a large population undergoing outpatient al-titude medicine consultation with hypoxic exercise testing, highvalues of DSae (>22%) and low values of HCRe (,0.84 beats/min/%) and HVRe (,0.78 L/min/kg) were independent predic-tors of SHAI, in addition to young age, female sex, regularphysical activity, rapid ascent (.400 m/day), history of SHAI,history of migraine, and geographical location (Ladakh). Nev-ertheless, given the low response rate, these results must beconfirmed in further studies. In subjects using ACZ for preven-tive purposes, some factors were no longer a risk. Moreover,DSae and HVRe substantially improved the discrimination ac-curacy of an SHAI risk prediction model.

Author disclosures are available with the text of this article at www.atsjournals.org.

Acknowledgment: The authors thank the physicians and nurses of the “Service dePhysiologie, Explorations Fonctionnelles et Medecine du Sport” of Avicenne Hos-pital in Bobigny for help in managing the patients and performing the tests since1992: Dr. Christian Rathat, Dr. Jean-Pierre Fouillot, Dr. Jean-Charles Poupelin,Catherine Godia, and Pascale Leloup. The authors also thank Prof. SylvieBastuji-Garin for helpful advice and Kalaivani Veerabudun, M.Sc., for help in datamanagement.

References

1. Ministry of Tourism, Nepal [Internet] [accessed 2011 Dec 12]. Available

from: http://www.tourism.gov.np/uploaded/statistics2010.pdf

2. Hackett PH, Roach RC. High-altitude illness. N Engl J Med 2001;345:

107–114.

3. Hackett PH, Rennie D, Levine HD. The incidence, importance, and

prophylaxis of acute mountain sickness. Lancet 1976;2:1149–1155.

4. Bloch KE, Turk AJ, Maggiorini M, Hess T, Merz T, BoschMM, Barthelmes

D, Hefti U, Pichler J, Senn O, et al. Effect of ascent protocol on acute

mountain sickness and success at Muztagh Ata, 7546 m. High Alt Med

Biol 2009;10:25–32.

5. Schneider M, Bernasch D, Weymann J, Holle R, Bartsch P. Acute

mountain sickness: influence of susceptibility, preexposure, and ascent

rate. Med Sci Sports Exerc 2002;34:1886–1891.

6. Pesce C, Leal C, Pinto H, Gonzalez G, Maggiorini M, Schneider M,

Bartsch P. Determinants of acute mountain sickness and success on

Mount Aconcagua (6962 m). High Alt Med Biol 2005;6:158–166.

Richalet, Larmignat, Poitrine, et al.: Physiological Risk Factors of Altitude Illness 197

7. Ri-Li G, Chase PJ, Witkowski S, Wyrick BL, Stone JA, Levine BD,

Babb TG. Obesity: associations with acute mountain sickness. Ann

Intern Med 2003;19:253–257.

8. Honigman B, Theis MK, Koziol-McLain J, Roach R, Yip R, Houston C,

Moore LG, Pearce P. Acute mountain sickness in a general tourist

population at moderate altitudes. Ann Intern Med 1993;118:587–592.

9. Mairer K, Wille M, Bucher T, Burtscher M. Prevalence of acute mountain

sickness in the eastern Alps. High Alt Med Biol 2009;10:239–245.

10. Richalet JP, Keromes A, Dersch B, Corizzi F, Mehdioui H, Pophillat B,

Chardonnet H, Tassery F, Herry JP, Rathat C, et al. Caracteristiques

physiologiques des alpinistes de haute altitude. Sci Sports 1988;3:89–108.

11. Allemann Y, Hutter D, Lipp E, Sartori C, Duplain H, Egli M, Cook S,

Scherrer U, Seiler C. Patent foramen ovale and high-altitude pul-

monary edema. JAMA 2006;296:2954–2958.

12. Richalet JP, Chenivesse C, Larmignat P, Meille L. High altitude pul-

monary edema, Down syndrome, and obstructive sleep apneas. High

Alt Med Biol 2008;9:179–181.

13. Hackett PH, Creagh CE, Grover RF, Honigman B, Houston CS, Reeves

JT, Sophocles AM, Van Hardenbroek M. High-altitude pulmonary

edema in persons without the right pulmonary artery. N Engl J Med

1980;302:1070–1073.

14. Sartori C, Allemann Y, Trueb L, Delabays A, Nicod P, Scherrer U.

Augmented vasoreactivity in adult life associated with perinatal vas-

cular insult. Lancet 1999;353:2205–2207.

15. Richalet JP, Letournel M, Salama J. Holmes-Adie syndrome associated

with high altitude pulmonary edema and low chemoresponsiveness to

hypoxia. Clin Auton Res 2011;21:55–56.

16. Hackett PH, Roach RC, Schoene RB, Harrison GL, Mills WJ Jr. Ab-

normal control of ventilation in high-altitude pulmonary edema.

J Appl Physiol 1988;64:1268–1272.

17. Matsuzawa Y, Fujimoto K, Kobayashi T, Namushi NR, Harada K,

Kohno H, Fukushima M, Kusama S. Blunted hypoxic ventilatory

drive in subjects susceptible to high-altitude pulmonary edema.

J Appl Physiol 1989;66:1152–1157.

18. Milledge JS, Beeley JM, Broome J, Luff N, Pelling M, Smith D. Acute

mountain sickness susceptibility, fitness and hypoxic ventilatory re-

sponse. Eur Respir J 1991;4:1000–1003.

19. Rathat C, Richalet JP, Herry JP, Larmignat P. Detection of high-risk

subjects for high altitude diseases. Int J Sports Med 1992;13:S76–S78.

20. Selland MA, Stelzner TJ, Stevens T, Mazzeo RS, McCullough RE,

Reeves JT. Pulmonary function and hypoxic ventilatory response in

subjects susceptible to high-altitude pulmonary edema. Chest 1993;

103:111–116.

21. Hohenhaus E, Paul A, McCullough RE, Kucherer H, Bartsch P. Ven-

tilatory and pulmonary vascular response to hypoxia and suscepti-

bility to high altitude pulmonary oedema. Eur Respir J 1995;8:1825–

1833.

22. Bartsch P, Swenson ER, Paul A, Julg B, Hohenhaus E. Hypoxic venti-

latory response, ventilation, gas exchange, and fluid balance in acute

mountain sickness. High Alt Med Biol 2002;3:361–376.

23. Burtscher M, Szubski C, Faulhaber M. Prediction of the susceptibility to

AMS in simulated altitude. Sleep Breath 2008;12:103–108.

24. Grunig E, Mereles D, Hildebrandt W, Swenson ER, Kubler W,

Kuecherer H, Bartsch P. Stress Doppler echocardiography for

identification of susceptibility to high altitude pulmonary edema. J

Am Coll Cardiol 2000;35:980–987.

25. Dehnert C, Grunig E, Mereles D, von Lennep N, Bartsch P. Identifi-

cation of individuals susceptible to high-altitude pulmonary oedema

at low altitude. Eur Respir J 2005;25:545–551.

26. Mairbaurl H. Role of alveolar epithelial sodium transport in high alti-

tude pulmonary edema (HAPE). Respir Physiol Neurobiol 2006;151:

178–191.

27. Wilson MH, Milledge J. Direct measurement of intracranial pressure at

high altitude and correlation of ventricular size with acute mountain

sickness: Brian Cummins’ results from the 1985 Kishtwar expedition.

Neurosurgery 2008;63:970–974.

28. MacInnis MJ, Koehle MS, Rupert JL. Evidence for a genetic basis for

altitude illness: 2010 update. High Alt Med Biol 2010;11:349–368.

29. Richalet JP, Gimenez-Roqueplo AP, Peyrard S, Venisse A, Marelle L,

Burnichon N, Bouzamondo A, Jeunemaitre X, Azizi M, Elghozi JL.

A role for succinate dehydrogenase genes in low chemo-

responsiveness to hypoxia? Clin Auton Res 2009;19:335–342.

30. Savourey G, Launay JC, Besnard Y, Guinet-Lebreton A, Alonso A,

Sauvet F, Bourrilhon C. Normo- or hypobaric hypoxic tests: propo-

sitions for the determination of the individual susceptibility to altitude

illnesses. Eur J Appl Physiol 2007;100:193–205.

31. Brugniaux JV, Schmitt L, Robach P, Jeanvoine H, Zimmermann H,

Nicolet G, Duvallet A, Fouillot JP, Richalet JP. Living high-training

low: tolerance and acclimatization in elite endurance athletes. Eur

J Appl Physiol 2006;96:66–77.

32. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas

under two or more correlated receiver operating characteristic curves:

a nonparametric approach. Biometrics 1988;44:837–845.

33. Oelz O, Howald H, Di Prampero PE, Hoppeler H, Claassen H, Jenni R,

Buhlmann A, Ferretti G, Bruckner JC, Veicsteinas A, et al. Physio-

logical profile of world-class high-altitude climbers. J Appl Physiol

1986;60:1734–1742.

34. Mollard P, Woorons X, Letournel M, Lamberto C, Favret F, Pichon A,

Beaudry M, Richalet JP. Determinant factors of the decrease in

aerobic performance in moderate acute hypoxia in women endurance

athletes. Respir Physiol Neurobiol 2007;159:178–186.

35. Gaillard S, Dellasanta P, Loutan L, Kayser B. Awareness, prevalence,

medication use, and risk factors of acute mountain sickness in tourists

trekking around the Annapurnas in Nepal: a 12-year follow-up. High

Alt Med Biol 2004;5:410–419.

36. Maggiorini M, Buhler B, Walter M, Oelz O. Prevalence of acute

mountain sickness in the Swiss Alps. BMJ 1990;301:853–855.

37. Grissom CK, Roach RC, Sarnquist FH, Hackett PH. Acetazolamide in

the treatment of acute mountain sickness: clinical efficacy and effect

on gas exchange. Ann Intern Med 1992;116:461–465.

38. van Patot MC, Leadbetter G III, Keyes LE, Maakestad KM, Olson S,

Hackett PH. Prophylactic low-dose acetazolamide reduces the inci-

dence and severity of acute mountain sickness. High Alt Med Biol

2008;9:289–293.

39. Savourey G, Guinet A, Besnard Y, Garcia N, Hanniquet AM, Bittel J.

Evaluation of the Lake Louise acute mountain sickness scoring sys-

tem in a hypobaric chamber. Aviat Space Environ Med 1995;66:963–

967.

198 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 185 2012