Share this article on:

Multiple variables explain the variability in the decrement in V˙O2max during acute hypobaric hypoxia


Medicine & Science in Sports & Exercise: June 1998 - Volume 30 - Issue 6 - p 869-879
Basic Sciences: Original Investigations

Multiple variables explain the variability in the decrement in V˙O2max during acute hypobaric hypoxia. Med. Sci. Sports Exerc., Vol. 30, No. 6, pp. 869-879, 1998.

Purpose: We used multiple regression analyses to determine the relationships between the decrement in sea level (SL, 760 Torr) V˙O2max during hypobaric hypoxia (HH) and variables that could alter or be related to the decrement in V˙O2max.

Methods: HH conditions consisted of 682 Torr, 632 Torr, and 566 Torr, and the measured independent variables were SL-V˙O2max, SL lactate threshold (SL-LT), the change in hemoglobin saturation at V˙O2max between 760 and 566 Torr (ΔSaO2max), lean body mass (LBM), and gender. Male (N = 14) and female (N = 14) subjects of varied fitness, training status, and residential altitude (1,640-2,460 m) completed cycle ergometry tests of V˙O2max at each HH condition under randomized and single-blinded conditions.

Results: V˙O2max decreased significantly from 760 Torr after 682 Torr (∼915 m) (3.5 ± 0.9 to 3.4 ± 0.8 L·min−1, P = 0.0003). Across all HH conditions, the slope of the relative decrement in V˙O2max (%V˙O2max) during HH was −9.2%/100 mm Hg (−8.1%/1000 m) with an initial decrease from 100% estimated to occur below 705 Torr (610 m). Step-wise multiple regression revealed that SL-V˙O2max, SL-LT, ΔSaO2max, LBM, and gender each significantly combined to account for 89.03% of the variance in the decrement in V˙O2max (760-566 Torr) (P < 0.001).

Conclusions: Individuals who have a combination of a large SL-V˙O2max, a small SL-LT (V˙O2, L·min−1), greater reductions in ΔSaO2max, a large LBM, and are male have the greatest decrement in V˙O2max during HH. The unique variance explanation afforded by SL-LT, LBM, and gender suggests that issues pertaining to oxygen diffusion within skeletal muscle may add to the explanation of between subjects variability in the decrement in V˙O2max during HH.

Submitted for publication October 1997.

Accepted for publication January 1998.

We acknowledge the generous support provided by David James, M.D. (Director, Cardiopulmonary Laboratory, Pulmonary Division, Veterans Administration Hospital, Albuquerque, NM) for allowing us to use the Poet pulse oximeter and Corval electronically braked cycle ergometer. Gratitude is also expressed to Paul Montner, M.D., for providing validation data of the Poet pulse oximeter from arterial blood during conditions at rest and during incremental exercise to V˙O2max.

Address for correspondence: R. A. Robergs, Ph.D., Johnson Center, B143, The University of New Mexico, Albuquerque, NM 87131. E-mail:

In 1967, Buskirk et al. (7) added to data from previous research (2,3,8,10,35) of the decrease in V˙O2max during hypoxia and revealed a 10.5% decrement in V˙O2max for every 1,000-m increase in altitude above 1,524 m above sea level. In 1997, Robergs and Roberts (38) recompiled research of the decrement in V˙O2max through to 1996 and showed an 8.7% decrement in V˙O2max for every 1,000-m increase above 1,050 m. Despite these attempts to model the decrement in V˙O2max during hypoxia by a single curve, research has shown that this decrement is highly variable between subjects. Paterson et al. (33) reported that compared with males, females had a significantly blunted decrease in V˙O2max during changes in F1O2 from 20.93 to 11.81%. Furthermore, it has been demonstrated that the decrement in V˙O2max with increasing altitude is exacerbated in individuals with high cardiorespiratory endurance (14,26,31,43,48,51), a high lactate threshold (LT) (26), and more severe hypoxemia (13,14).

Based on previous research (14,25-27,31,43,47,50), sea level V˙O2max (SL-V˙O2max) can account for approximately 12-88% of the variance (r2) in the decrement in V˙O2max depending on the gender and range of fitness of the subject population. In addition, Ferretti et al. (13) reported that the decrease in SaO2 with increasing hypoxia accounts for approximately 86% of the variance in the decrement in V˙O2max in trained individuals. As the majority of studies that measured both V˙O2max decrement and hypoxemia during exposure to hypoxia revealed significant correlations among SL-V˙O2max, SaO2 at V˙O2max, and V˙O2max decrement (13,14,26,27,30,31,43,47), there are obviously components of colinearity among these variables. Consequently, the unique contributions of SL-V˙O2max or hypoxemia to the explanation of the between subjects variability in the decrement in V˙O2max during hypoxia remain uncertain.

As SL-V˙O2max and reductions in oxygen supply or arterial blood oxygen content during hypoxia explain approximately 86% of the variability in the decrement in V˙O2max (13), other factors unrelated to oxygen delivery must also influence or be associated with changes in V˙O2max during hypoxia. It has been demonstrated (39) and theorized (49) that V˙O2max during both normoxia and hypoxia is dependent on both oxygen supply (blood flow × CaO2) and peripheral oxygen diffusion. In addition, individual differences in peripheral oxygen diffusion during hypoxia explain why some individuals either decrement more or less than expected based upon reductions in oxygen supply to the contracting muscle (42,48).

As no single study has evaluated multiple determinants for V˙O2max during hypoxia in a large number of subjects, it is not known whether sea level fitness, hypoxemia, gender, or additional factors associated with peripheral oxygen diffusion limitation exert independent effects on the magnitude of decrement in V˙O2max. As it is extremely difficult to control for multiple variables using human subjects in between group experimental research, we decided to study a relatively large number of subjects so that we could control for multiple variables using appropriate statistics. Consequently, it was our intent to study the decrement in V˙O2max during HH using multiple regression statistics with a sufficient sample size of men and women to evaluate variables associated with central and peripheral components of oxygen delivery to contracting muscle. Although multiple regression is based on correlational analyses, results can be interpreted more toward cause-effect associations owing to the unique variance explanation of the dependent variable by specific independent variables (34). We studied changes in V˙O2max during HH conditions of 682, 632, and 566 Torr because the majority of active individuals recreate, train, and/or compete within this range of hypoxia, and recent research has demonstrated V˙O2max decrements during acute exposure to low hypoxia equivalent to 580 m above sea level (14).

We hypothesized that the variables SL-V˙O2max, SL-LT, blood hemoglobin and hematocrit, lean body mass (LBM), hemoglobin saturation (SaO2), and gender would all be significantly related to the decrement in V˙O2max between 760 and 566 mm Hg (ΔV˙O2max). We also hypothesized that a multiple regression model would add to the variance explanation of SL-V˙O2max alone owing to individual differences in the decrease in SaO2 at V˙O2max during HH. Finally, we hypothesized that LBM and the LT would add additional variance explanation to the model, thereby indicating the potential for the unique importance of peripheral factors, possibly reflecting peripheral oxygen diffusion limitation to V˙O2max during hypoxia.

Back to Top | Article Outline


Subjects. Subjects were recruited from the faculty, undergraduate and graduate student populations of the university, as well as members of the surrounding communities. Seventeen males and 14 females read and signed a written informed consent approved by the specific College Human Subjects Research Committee. All subjects resided at and were acclimatized to altitudes between 1,525 and 2,225 m above sea level. Subjects provided information of their training status based on the duration and weekly frequency of exercise. Females reported information of their menstrual cycle, including the days since the end of their last menstruation, the number of cycles experienced each year, and use of oral contraceptives. All females completed the study, whereas three males did not, owing to illness and the inability to complete all tests within a 2-wk period, resulting in equal subject numbers of 14 for each gender.

Test procedures. Testing was conducted at simulated conditions of SL (760 Torr) and during HH pressures of 682, 632, and 566 Torr (equivalent to 915, 1,524, and 2,439 m above sea level, respectively). The SL and HH conditions were systematically ordered using a Latin Squares design (24) and assigned to subjects based on subject number. The order of exposures remained blind to the subjects and all researchers except those operating the hyper/hypobarometric chamber and calibrating the gas analyzers and pneumotach. Subjects were required to complete no more than two V˙O2max tests on a given day and to complete all four tests within 1 wk.

Subjects were first familiarized with the equipment used in the research. Exercise was performed on an electronically braked cycle ergometer (Corval, Lode bv, Goningen, The Netherlands), and indirect expired gas analysis was performed using an automated Ergooxyscreen computerized system (Eric Jaeger, Wurtzburg, Germany), with data printed every 30 s. The Ergooxyscreen measures expired ventilation by a heated pneumotach and continuously pumps samples of expired gas from a mixing chamber through calcium carbonate drying canisters to oxygen (zirconian cell) and carbon dioxide (infrared) analyzers. The pneumotach was calibrated with seven samples of 1.0-LATPS volumes from a calibration syringe before each test. The gas analyzers were also calibrated before each test to room air and commercially prepared and medically certified calibration gases (15% O2 and 5% CO2, and 100% N2). Heart rate was measured using telemetry (5-s intervals) (Polar USA, Inc., Stamford, CT). Oxy-hemoglobin saturation (SaO2) was indirectly measured using finger pulse oximetry (Poet, Criticare Systems Inc., OH), and the analog signal of percent SaO2 was acquired at 0.1 Hz by a computerized data acquisition system (BIOPAC Systems Inc., Santa Barbara, CA). Details of the validation of finger pulse oximetry using the Poet are presented in "Appendix." Data of the changes in SaO2 with increasing exercise intensity during each HH condition are to be presented in another manuscript.

Exercise was performed within a hyper/hypobarometric chamber, and barometric pressure was measured using a commercially calibrated flight altimeter. The rate of pressure change was constant for all trials and individuals at 27 Torr·min−1 (∼300 m·min−1). During the transition to the predetermined HH condition, a catheter was placed in an antecubital vein, connected to a four-way stopcock, and infused with 2 mL of sterile saline every 5 min to prevent clotting (total saline infusion <10 mL). The subject was prepared for heart rate measurement, a pulse oximeter finger sensor was positioned on the index finger, and the hand was then placed in or on a heating pad (35-41°C) to promote increased peripheral cutaneous circulation. Once at the specific HH condition, the Ergooxyscreen system was calibrated, and resting data were acquired for 1 min. During this time, a 3-mL blood/saline sample was taken and discarded and a 1-mL blood sample was immediately obtained. Exercise was then started, with subjects completing a protocol requiring 15-, 20-, or 25-W increments each minute until exhaustion (21). The protocol was individualized to ensure a total test duration between 12 and 16 min, and for a given subject the same protocol was used for each HH condition. During exercise, computerized automated collection of expired gas analysis and oxy-hemoglobin saturation were performed as previously described, and a 1-mL sample of blood was obtained between 50 and 60 s of each stage. No saline was infused between blood samples. As blood was not able to be obtained from one subject during exercise testing, data of the lactate threshold are for 27 rather than 28 subjects.

V˙O2max was defined as the peak V˙O2 when associated with an RER greater than 1.1 or occurred with less than a 0.1 L·min−1 increase in V˙O2 from the previous value. These criteria were not met for six tests, and these tests were repeated successfully on another day within each subject's weekly test schedule.

After subjects completed the HH exercise testing, they reported to the laboratory on a separate day for body composition and blood hematology analyses. Body composition was assessed by two highly trained technicians using skinfolds. For females, skinfolds were measured at the triceps, suprailiac, and thigh sites, and the sum of three skinfolds was converted to body density using the equation of Jackson et al. (23). Body density was converted to percent body fat using the gender-specific formula of Lohman (28). For males, skinfolds were measured at the chest, abdominal, and thigh sites, and the sum of three skinfolds was converted to body density using the equation of Jackson et al. (22). The body density of males was converted to percent body fat using the Siri equation (44). Lean body mass (LBM) was calculated by adjusting total body mass by the body fat fraction.

Analytical procedures. Blood samples were dispensed into borosillicate tubes, and 0.5 mL of whole blood was then pipetted into 1 mL of 7% perchloric acid. Samples were immediately mixed, and after the test they were refrigerated at 4°C. At the end of each testing day, samples were centrifuged at 4°C, and the supernatant was removed and stored at −47°C for subsequent analysis of lactate. Lactate was assayed using an enzymatic spectrophotometric procedure (29). Hematocrit was determined from venous blood by microcentrifugation and corrected for trapped plasma by multiplying by 0.96 (11). Blood hemoglobin concentration was determined using the methemoglobin method (Sigma, Reagent 525).

The LT was detected by graphical presentation using log lactate versus V˙O2 (L·min−1) plots. Bisegmental linear regression was performed on two portions of the data that gave the least residual error, and the point of intersection of the two lines was denoted as the LT (L·min−1) (4). Complete data of the changes in the LT during increasing HH are to be presented in another manuscript.

Statistics. All statistical procedures were run using Biomedical Statistical Program software (BMDP Statistical Software, Inc., Los Angeles, CA) (6). Descriptive data were computed (BMDP-1D), and stepwise multiple regression (BMDP 2R) was performed (N = 28) to explain ΔV˙O2max (L·min−1) from 760 to 566 Torr using independent variables of SL-V˙O2max, SL-LT (L·min−1), the decrement in SaO2max from 760 to 566 Torr (ΔSaO2max), blood hemoglobin concentration, LBM, and gender. To provide prediction equations for V˙O2max at each barometric pressure, separate multiple regression analyses were performed for each HH condition using SL-V˙O2max, SL-LT, LBM, hemoglobin, hematocrit, and SL-SaO2max as independent variables. Prediction equations were not cross-validated.

The analysis of slope for the SaO2 data during increasing HH was evaluated using ANOVA assessment of linear and quadratic components (BMDP 2V). Differences between genders and across HH conditions were assessed using two-way repeated measures ANOVA (BMDP 4V). The accepted statistical significance for all analyses was P < 0.05. All data are presented as mean±SD, and all gas volumes are expressed STPD.

Back to Top | Article Outline


Descriptive data. The descriptive data of the subjects, as well as subjects grouped by gender, are provided in Table 1. Of the females studied, 6 were taking oral contraceptives. Four females reported to miss 1, 2, 4, and 6 cycles per year, respectively. Testing of females not on oral contraceptives occurred within 10 d after the start of menstruation.



Subjects ranged in activity status from recreationally active to nationally competitive triathletes and cyclists. Mean data of both genders indicated below average body fat. On average, males were taller, heavier and leaner than the females, with a higher V˙O2max when expressed absolute (L·min−1) (Table 1). When V˙O2max was expressed relative to lean body mass, values were similar between genders (Table 5).



V˙O 2max. V˙O2max test duration was not different between HH conditions, varying between 14.3 ± 2.1 and 13.1 ± 2.2 min for tests at 760 and 566 Torr, respectively. Data of variables measured at V˙O2max are presented in Table 2 and Fig. 1. Each of V˙O2max expressed absolute (L·min−1) (Fig. 1A) and relative to LBM (mL·kg−1 LBM·min−1) (Table 2) significantly decreased with increasing HH (both P < 0.0001). Compared with 760 Torr, V˙O2max expressed as absolute or relative decreased significantly by 682 Torr (P = 0.0002 and P = 0.0003, respectively).



Figure 1

Figure 1

The % change in V˙O2max from 760 mm Hg (% V˙O2max) with increasing HH is presented in Fig. 1B. % V˙O2max significantly decreased with increasing HH (P < 0.0001). There was a significant decrement in %V˙O2max by 682 Torr (P = 0.0121). The decrement in %V˙O2max between 682 and 566 Torr was linear (r = 0.99, P < 0.001). Linear extrapolation from 566 Torr to the pressure at 100% V˙O2max revealed an estimated initial decrease in V˙O2max at 705 Torr (∼ 610 m) (Fig. 1B). However, the between subject variability for the absolute and relative decrements in V˙O2max were high, as indicated by the large standard deviation error bars (Fig. 1).

Maximal heart rate. HRmax significantly decreased with increasing HH (P = 0.012), becoming significantly lower at 632 than at 760 Torr (P = 0.0082) (Table 2).

Ventilation. Expired ventilation at V˙O2max (V˙Emax) increased significantly with increasing HH (P = 0.0041), being significantly greater at 566 than 760 Torr (Table 2). The ventilatory equivalent for oxygen at V˙O2max (V˙E/V˙O2max) also significantly increased with increasing HH (P < 0.0001), being significantly greater at 632 than 760 Torr (Table 2).

Hemoglobin saturation (SaO 2 ). There was a significant decrease in SaO2 rest and at V˙O2max with increasing HH (both P < 0.0001), with significance from 760 Torr beginning at 682 Torr (P < 0.0001) for rest and V˙O2max conditions (P < 0.0001) (Fig. 2). SaO2 at V˙O2max (SaO2max) was significantly lower than rest SaO2 at 760 Torr and all HH conditions (P < 0.0001). There was a significant rest-exercise interaction (P < 0.0001), indicating that SaO2 decreased more at V˙O2max during HH than at rest. The decrease in SaO2max during HH was curvilinear (P = 0.0014).

Figure 2

Figure 2

Multiple regression analyses. Multiple linear regression analyses were performed to determine the variables that significantly and uniquely accounted for the variance in the decrement in V˙O2max with increasing HH (760 to 566 Torr) (ΔV˙O2max) and for the absolute V˙O2max for a given HH condition between 760 and 566 Torr. The correlation matrix of the variables used in the main multiple regression analyses is presented in Table 3. The variables SL-V˙O2max, SL-LT (L·min−1), ΔSaO2max (760 to 566 Torr), LBM, and gender each significantly contributed to an equation accounting for 89.03% of the total variance in ΔV˙O2max (Table 4). The data points of select relationships between variables pertinent to the multiple regression findings are presented in Fig. 3 to illustrate the spread of the data and slope of each regression line of best fit. Significance values for the univariate regressions are not reported, owing to the inappropriateness of simple linear regression between the dependent variable and multiple independent variables used in a multiple regression model. To illustrate the association between multiple variables and ΔV˙O2max, the data from a subset of the subjects are presented in Fig. 4. The partial correlations of Table 4 are presented to reveal the alterations to the univariate simple correlation coefficients that occurred during stepwise multiple regression.





Figure 3

Figure 3

Figure 4-Da

Figure 4-Da

The decrement in our subjects' V˙O2max from 760 Torr during HH of 566 Torr can be estimated from the data presented in Table 4. For variables measured at sea level, prediction of the decrement in V˙O2max at 566 Torr can be estimated by the following equation: Equation [1]

For estimation of the absolute V˙O2max of our male and female subjects for a given HH of 682, 632, or 566 Torr, the following equations using sea level data can be used: Equation [2-4]

Gender differences. Pertinent data grouped by gender are presented in Table 1 and 5. These results are presented owing to the significant inclusion of gender in the multiple regression model. On average, males were taller, heavier, more lean, and of higher cardiorespiratory endurance (V˙O2max, L·min−1) than the female subjects (Table 1). There was no gender main effect difference in V˙O2max when expressed as mL·kg−1 LBM·min−1 (P = 0.1934) (Table 5). V˙O2max (mL·kg−1 LBM·min−1) significantly decreased with increasing HH (P < 0.0001) and occurred with a significant gender-altitude interaction (P = 0.0026). However, no gender difference was evident at any altitude. V˙O2max (mL·kg−1 LBM·min−1) significantly decreased from 760 at 682 Torr (P = 0.0001) for males and from 760 at 632 Torr (P = 0.0048) for females.

Data of the change in V˙O2max (L·min−1) with increasing HH revealed significant main effects for gender (P < 0.0001) and HH (P < 0.0001) and a significant gender-HH interaction (P = 0.0001) (Table 5). Males had a significantly larger V˙O2max at all HH conditions than did the females (P < 0.003). V˙O2max significantly decreased from SL in males at 682 Torr (P = 0.0028) and for females at 632 Torr (P = 0.0001).

Results for V˙E at V˙O2max (V˙Emax) revealed significant main effects for gender (P < 0.0001) and HH (P = 0.0029) and a nonsignificant gender-HH interaction (P = 0.0995) (Table 5). Males had significantly larger V˙Emax for each HH condition than females; however, both males and females had similar increases in maximal ventilation with increasing HH. Results for the ventilatory equivalent for oxygen (V˙E/V˙O2) at V˙O2max (V˙E/V˙O2max) revealed no gender difference (P = 0.5025), a significant HH main effect (P < 0.0001), and a nonsignificant gender-HH interaction (P = 0.151) (Table 5). Males had a larger V˙E/V˙O2max than females, and both males and females increased V˙E/V˙O2 similarly with increasing HH.

Results for maximal heart rate revealed a nonsignificant main effect for gender (P = 0.857), a significant main effect for HH (P = 0.008), and a nonsignificant gender-HH interaction (P = 0.1261) (Table 5). Maximal heart rate decreased from SL for both genders at 632 Torr (P = 0.0061).

Results for SaO2 revealed no gender difference (P = 0.4227), a significant main effect for HH (P < 0.0001), and a nonsignificant interaction (P = 0.57) (Table 5). For both males and females, SaO2 at V˙O2max decreased significantly from sea level by 682 Torr (P < 0.0001).

Back to Top | Article Outline


The results confirmed each of our hypotheses. We demonstrated that the males and females of this study had a detectable reduction in V˙O2max at 682 Torr. Linear extrapolation of mean data indicated that the decrement in V˙O2max commenced at 705 Torr. However, the high between subjects variability for the decrement in V˙O2max during HH and the findings from multiple regression indicated that it is inappropriate to characterize this relationship by a single curve. A large portion (89%) of the between subjects variability in the decrement in V˙O2max could be explained using multiple (N = 5) variables. In addition to sea level V˙O2max, the ΔV˙O2max during HH is further increased in individuals who have a low SL-LT, greater reduction in ΔSaO2max, large LBM, and are male. Conversely, the multiple determinants of the V˙O2max decrement indicate that individuals with a large V˙O2max, but who have a large SL-LT, have a small LBM, and do not exhibit a large hypoxemia, will not decrease their V˙O2max during HH to the extent anticipated based on SL-V˙O2max alone.

Discussion of these findings will be structured by the type of variable and by the type of statistics performed.

Back to Top | Article Outline


Univariate findings. Our study investigated changes in V˙O2max from sea level during three low to moderate HH conditions. Using mean data, the decrement in V˙O2max with decreasing barometric pressure was significant at 682 Torr and decreased linearly thereafter to 566 Torr. In comparison, only two other studies have evaluated changes in V˙O2max that included more than three hypoxic conditions within pressures ranging from 760 to 440 Torr (1,43). Andersen et al. (1) quantified a curvilinear decrement in V˙O2max in seven male subjects from sea level (753 Torr) to each of 559, 477, 440, 422, and 404 Torr. The occurrence of statistical significance for the decreases in V˙O2max at each altitude were not reported. Conversely, Squires and Buskirk (45) reported a linear decrease in V˙O2max during barometric pressure conditions between approximately 680 and 580 Torr, with significance occurring at 656 Torr (1219 m).

The concept of a critical altitude for inducing a reduction in V˙O2max may be inappropriate. For example, Gore et al. (15) documented a reduction in V˙O2max in elite trained endurance athletes during acute HH equivalent to an increased altitude of 580 m (695 Torr). The findings of Gore et al. (14,15) and similar findings of a greater V˙O2max decrement during hypoxia in more endurance-trained individuals (26,27,31,43,50) have led to the interpretation that more fit individuals will experience greater decrements in V˙O2max during hypoxia. Our data indicate that such an interpretation is not true for all individuals, even when of different endurance training status and cardiorespiratory endurance fitness. For example, data of V˙O2max at each HH condition are presented in Fig. 4 for a subset of the male and female subjects used in the study. The two most endurance-trained male subjects of this study had relatively minor reductions in V˙O2max. A larger but less endurance-trained subject had a more dramatic decrement in V˙O2max. This larger decrement was also associated with a relatively large decrease in SaO2. The same trends were evident for the female subjects (Fig. 4). Clearly, the decrement in V˙O2max during hypoxia is caused and related to multiple factors, and no one variable can be stated as the principle determinant of the decrement in all individuals.

Multiple regression findings. Because of the multiple variables associated with the decrement in V˙O2max during hypoxia, we performed step-wise multiple regression analyses to explain the decrement in V˙O2max using several independent variables (Table 3 and 4). The results showed that SL-V˙O2max explained the greatest variability in the decrement in V˙O2max, with further significant contributions provided by SL-LT, ΔSaO2max, LBM, and gender. The combination of these variables explained approximately 89% of the variance (r2) in the decrement in V˙O2max. Although our subject number was large relative to other studies on this topic, it remains small for application of multiple regression analyses. Because of our sample size, it is likely that the multiple correlations we report are inflated, as is typical for multiple regression using a subject to independent variable number ratio lower than 10:1 (24,34). Nevertheless, based on the originality of our methodology and findings, our data contribute significantly to the current knowledge on the decrement in V˙O2max during hypoxia.

Our results clearly indicate that the concept of an average decrement in V˙O2max with increasing HH is inappropriate. Cardiorespiratory and muscular endurance (SL-V˙O2max and SL-LT), ΔSaO2max, LBM, gender, and perhaps other variables all combine to determine the absolute decrement and therefore need to be considered when estimating the influence of hypoxia on a person's V˙O2max.

The fact that gender significantly contributed to the explanation of total variance in V˙O2max is important. There is a component of the oxygen transport and utilization cascade that is different between genders and unrelated to the remaining independent variables we measured. This is a meaningful finding, for as indicated in the data of Tables 1 and 5, the females of the study were on average less fit (V˙O2max, L·min−1), of smaller LBM, yet of similar relative fitness (V˙O2max-mL·kg−1 LBM·min−1), hematology, and ΔSaO2max to the male subjects.

It is now well accepted that tissue oxygen diffusion limitation is an important component of oxygen transfer to contracting skeletal muscle and has an independent role in decreasing V˙O2max during hypoxia (37,39,49). Wagner (49) has theorized that as hypoxia becomes more extreme, the importance of peripheral oxygen diffusion to V˙O2max increases. This fact may be revealed in how the slope (coefficient) for SL-V˙O2max for each HH condition (see eq. 2-4) decreased with increasing HH. However, less is known of the extent of individual differences in peripheral oxygen diffusion during normoxia or hypoxia and what peripheral factors are associated with these between-subject differences. Nevertheless, Shephard et al. (42) revealed that increases in the muscle mass exercised are associated with more severe reductions in V˙O2max during hypoxia, thus indicating the potential for the size of the active muscle mass to influence peripheral oxygen diffusion. In addition, Wagner (48) has estimated peripheral oxygen diffusion and identified larger values for trained than untrained individuals.

We propose that the unique gender, LT and LBM contributions to the decline in V˙O2max with increasing HH may be related to the role of peripheral oxygen diffusion in the measurement of V˙O2max, especially during HH. Both male and female subjects with a larger LBM and sea level V˙O2max experienced the greatest absolute decrements in V˙O2max with increasing HH. This finding is consistent with the muscle atrophy associated with chronic exposure to severe hypoxia (16,17). Presumably, the adaptation of muscle atrophy during chronic hypoxia decreases muscle fiber dimensions and decreases the distance for peripheral oxygen diffusion, hence increasing the transfer of oxygen from blood to within skeletal muscle fibers.

The unique importance of SL-LT is revealed by the altered correlation direction of V˙O2max decrement and SL-LT between the univariate and multiple regression analyses. The univariate correlation between V˙O2max decrement and SL-LT was 0.517. However, the partial correlation after SL-V˙O2max was entered in to the multiple regression model changed to −0.4567. This can be interpreted to indicate that the univariate correlation between SL-LT and a greater V˙O2max decrement during HH was predominantly owing to colinearity caused by a high correlation between SL-LT and SL-V˙O2max (Table 3). When the variability in SL-LT owing to SL-V˙O2max was removed, individuals with a high SL-LT actually had a smaller decrement in V˙O2max during HH. Thus, the unique benefit of a high LT to exercise during HH may be to decrease the V˙O2max decrement. This alteration in interpretation of the LT to V˙O2max decrement findings would have been overlooked using simple univariate correlation, as in the results of Koistinen et al. (26).

The interpretation of the between subjects differences in SL-LT to reflect differences in peripheral skeletal muscle characteristics are supported by the results of Green et al. (18,19). Green et al. demonstrated that during 5-7 d of endurance training at 67% V˙O2max, alterations in muscle metabolism occurred before increases in muscle mitochondrial enzyme activities and V˙O2max (18). Furthermore, additional research of endurance training through 8 wk revealed decreases in muscle lactate accumulation and increased glycogen sparing during the first week (3-7 d) (19). No additional improvement in these measures occurred during the remaining 7 wk. Conversely, increases in V˙O2max were not maximized until 4 wk of training (19). The authors concluded that the majority of metabolic adaptations to endurance training occur independently of increases in the oxidative potential of the trained skeletal muscle and whole body V˙O2max. Clearly, there are yet to be clarified functions of peripheral metabolism that are independent of oxygen supply and utilization to contracting skeletal muscle.

Back to Top | Article Outline

Maximal Heart Rate

Previous research on the maximal heart rate (HRmax) response to exercise during acute hypoxia has been equivocal. Elliot and Atterbom (12) reported that HRmax did not decrease in men and women at V˙O2max from 1,576 m to 2,743 and 3,962 m. Similar results for different levels of hypoxia have also been reported (9,18,46). However, Koistinen et al. (26) reported a significant decrease in HRmax by 6 b·min−1 from sea level to 3000 m, and Andersen et al. (1) reported a significant decrease in HRmax by 10 b·min−1 between sea level and 5,000 m. Other researchers have also shown a decrease in HRmax during acute hypoxia (10,31,33,40), with larger decrements in HRmax during severe hypoxia as studied in Operation Everest II research (9,36,51).

Our reporting of a significant decrease in HRmax from 760 to 632 Torr, even though such a decrease was of a mean difference of approximately 2 b·min−1, indicates that HRmax is affected by hypoxia at even low altitudes above sea level. Nevertheless, the physiological implications of such a small difference may be negligible.

Back to Top | Article Outline


We report an exponential decrease in SaO2max with increasing HH. An exponential decrease in resting SaO2 during increasing HH was reported by Schoene et al. (41) and Ferretti et al. (13). Conversely, Squires and Buskirk (45) reported a linear decrease in SaO2max in 12 men during HH equivalent to 712, 680, 656, 632, and 575 Torr. SaO2 ranged from 90.7 to 79.1% for 712 and 575 Torr conditions, respectively. The hypoxemia at 575 Torr from Squires and Buskirk (45) was more extreme than our average of approximately 85% for both males and females at 566 Torr (Fig. 2). The most logical explanations for these discrepancies are based on the moderate altitude acclimatization of the subjects used in our study, the moderate hypoxic exposures, and the lower likelihood for exercise-induced hypoxemia in many of the subjects we studied. As previously stated, our subjects were of varied fitness and resided between 1,640 and 2,460 m, which would have induced a chronic hyperventilation, a raised alveolar PO2, and a higher SaO2 at rest and during exercise. As Schoene et al. (41) and Ferretti et al. (13) studied more extreme ranges of hypoxia (to PB = 240 and 430 Torr, respectively), there would have been a greater hemoglobin desaturation and a greater likelihood for results to mirror the oxyhemoglobin dissociation curve. As our results compare favorably with those of Ferretti et al. (13) and Schoene et al. (41), it seems that the decrement in SaO2 during mild to moderate hypoxia closely mirrors reductions based on the sigmoidal oxyhemoglobin dissociation curve. However, it is likely that reductions will be larger for more endurance-trained subjects owing to a greater exercise-induced hypoxemia (14,15,45).

Back to Top | Article Outline


Our research is based on correlation analyses, and the associations between variables cannot be interpreted as cause and effect. However, the multiple regression approach to explain the statistical variability in the decrement in V˙O2max during HH produced results that are both supported by experimental research and add to the findings of such research. Furthermore, multiple regression analyses are interpreted to extend simple correlation analyses toward cause-effect associations (32). Consequently, the results from multiple regression analyses provide direction for future experimental cause-effect research (24,34).

We documented gender-specific differences in the decrement in V˙O2max during increasing HH and that SL-V˙O2max, SL-LT, LBM, hemoglobin desaturation, and gender all significantly contribute to explain the ΔV˙O2max. An accurate estimation of the decrease in V˙O2max with increasing altitude cannot be made without accounting for the aforementioned variables. In addition, the association between V˙O2max decrement during hypoxia and SL-V˙O2max may decrease with increasing hypoxia, indicating that other factors may be more influential in influencing the decrement. Consequently, we propose that a lower LT and larger LBM are associated with a limitation in peripheral oxygen diffusion during exercise at V˙O2max, especially during HH. These findings indicate that for both men and women, for a given V˙O2max individuals who have a large LT, have less hypoxemia, and have a small LBM will retain more of their SL exercise potential during acute HH.

There are many important applied interpretations of our results. As it is clear that certain individuals demonstrate a superior retention of SL-V˙O2max during hypoxia, these individuals should perform better at altitude. Whether these individuals may also benefit more from altitude training is unknown. Based on our findings, coaches and athletes can better understand the tolerance to hypoxia and thereby improve selection of athletes who may be more likely to compete better at altitude.

Clearly, more research of peripheral oxygen diffusion and muscle fiber and capillary morphology need to be conducted to elucidate their influence on oxygen transport and utilization by skeletal muscle during HH. In addition, further research of more elite athletes needs to be conducted to develop similar prediction equations for V˙O2max and performance decrement during hypoxia. Finally, care should be taken in the interpretation of results from univariate analyses used to compare mean differences between two or more groups in physiological research. Our data reveal that what might seem as a group difference may in fact be the result of differences on other functional and anatomical measurements that influence the dependent variable.

Back to Top | Article Outline


1. Andersen, H. T., E. B. Smeland, J. O. Owe, and K. Myre. Analyses of maximum cardiopulmonary performance during exposure to acute hypoxia at simulated altitude-sea level to 5000 meters (760-404 mm Hg). Aviat. Space Environ. Med. 56:1192-1197, 1985.
2. Astrand, P. O., and I. Astrand. Heart rate during muscular work in man exposed to hypoxia. J. Appl. Physiol. 13:75-80, 1958.
3. Balke, B., F. J. Nagle, and J. Daniels. Altitude and maximum performance in work and sports activity. JAMA 194:646-649, 1965.
4. Beaver, W. L., K. Wasserman, and B. J. Whipp. Improved detection of lactate threshold during exercise using a log-log transformation. J. Appl. Physiol. 59:1936-1940, 1985.
5. Bland, J. M., and D. G. Altman. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet February 8:307-310, 1986.
    6. Brown, W. J., M. B. Brown, L. Engelman, and R. I. Jennrich. BMDP Statistical Software Manual, Vol. 1a, 1, and 2. Berkeley: University of California Press, 1990, pp. 1-1354.
    7. Buskirk, E. R., J. Kollias, R. F. Akers, E. K. Prolop, and E. P. Reategui. Maximal performance at altitude and on return from altitude in conditioned runners. J. Appl. Physiol. 23:259-266, 1967.
    8. Consolazio, C. F., L. O. Matoush, and R. A. Nelson. Energy metabolism in maximum and submaximum performance at high altitudes. Fed. Proc. 25:1380-1385, 1966.
    9. Cymerman, A., J. T. Reeves, J. R. Sutton, P. B. Rock, B. M. Groves, M. K. Malconian, P. M. Young, P. D. Wagner, and C. S. Houston. Operation Everest II: maximal oxygen uptake at extreme altitude. J. Appl. Physiol. 66:2446-2453, 1989.
    10. Dill, D. B., and W. C. Adams. Maximal oxygen uptake at sea level and at 3,090-m altitude in high school champion runners. J. Appl. Physiol. 30:854-859, 1971.
    11. Dill, D. B., and D. L. Costill. Calculation of percentage changes in volumes of blood, plasma, and red cells in dehydration. J. Appl. Physiol. 37:247-248, 1974.
    12. Elliot, P. R., and H. A. Atterbom. Comparison of exercise responses of males and females during acute exposure to hypobaria. Aviat. Space Environ. Med. 49:415-418, 1978.
    13. Ferretti, G., C. Moia, J. M. Thomet, and B. Kayser. The decrease of maximal oxygen consumption during hypoxia in man: a mirror image of the oxygen equilibrium curve. J. Physiol. (Lond.) 498:231-237, 1997.
    14. Gore, C. J., S. C. Little, A. G. Hahn, G. C. Scroop, D. B. Watson, K. I. Norton, R. J. Wood, D. P. Campbell, and D. L. Emonson. Increased arterial desaturation in trained cyclists during maximal exercise at 580 m altitude. J. Appl. Physiol. 80:2204-2210, 1996.
    15. Gore, C. J., A. G. Hahn, G. C. Scroop, K. I. Norton, P. C. Bourdon, S. M. Woolford, J. D. Buckley, T. Sanef, D. P. Campbell, D. B. Watson, and D. L. Emonson. Reduced performance of male and female athletes at 580 m altitude. Eur. J. Appl. Physiol. 75:136-143, 1997.
    16. Green, H. J., J. R. Sutton, A. Cymerman, P. M. Young, and C. S. Houston. Operation Everest II: adaptations in human skeletal muscle. J. Appl. Physiol. 66:2454-2461, 1989.
    17. Green, H. J., J. R. Sutton, E. E. Wolfel, J. T. Reeves, G. E. Butterfield, and G. A. Brooks. Altitude acclimitization and energy metabolic adaptations in skeletal muscle during exercise. J. Appl. Physiol. 73:2701-2708, 1992.
    18. Green, H. J., R. Helyar, M. Ball-Burnett, N. Kowalchuk, S. Symon, and B. Farrance. Metabolic adaptations to training precede changes in muscle mitochondrial capacity. J. Appl. Physiol. 72:484-491, 1992.
    19. Green, H. J., S. Jones, M. Ball-Burnett, B. Farrance, and R. Ranney. Adaptations in muscle metabolism to prolonged voluntary exercise training. J. Appl. Physiol. 78:138-145, 1995.
    20. Hannhart, B., J. P. Haberer, C. Saunier, and M. C. Laxenaire. Accuracy and precision of fourteen pulse oximeters. Eur. Respir. J. 4:115-119, 1991.
    21. Howley, E. T., D. R. Bassett, and H. G. Welch. Criteria for maximal oxygen uptake: review and commentary. Med. Sci. Sports Exerc. 27:1292-1301, 1995.
    22. Jackson, A. S., and M. L. Pollock. Generalized equations for predicting body density of men. Br. J. Nutr. 4:497-504, 1978.
    23. Jackson, A. S., M. L. Pollack, and A. Ward. Generalized equations for predicting body density of women. Med. Sci. Sports Exerc. 12:175-182, 1980.
    24. Keppel, G. Design and Analysis: A Researchers Handbook. London: Prentice Hall, 155-160, 1983.
    25. Klausen, K., D. B. Dill, and S. M. Horvath. Exercise at ambient and high oxygen pressure at high altitude and at sea level. J. Appl. Physiol. 24:336-347, 1970.
    26. Koistinen, P., T. Takala, V. Martikkala, and J. Leppaluoto. Aerobic fitness influences the response of maximal oxygen uptake and lactate threshold in acute hypobaric hypoxia. Int. J. Sports Med. 26:78-81, 1995.
    27. Lawler, J., S. K. Powers, and D. Thompson. Linear relationship between V˙O2max and V˙O2max decrement during exposure to acute hypoxia. J. Appl. Physiol. 64:1486-1492, 1988.
    28. Lohman, T. G. Applicability of body composition techniques and constants for children and youth. Exerc. Sport Sci. Rev. 14:325-357, 1986.
    29. Lowry, O. H., and J. V. Passonneau. A Flexible System of Enzymatic Analysis. New York: Academic Press, 1976, 199-200.
    30. Maresh, C. M., B. J. Noble, K. L. Robertson, and W. E. Sime. Maximal exercise during hypobaric hypoxia (447 Torr) in moderate-altitude natives. Med. Sci. Sports Exerc. 15:360-365, 1983.
    31. Martin, D., and J. K'roy. Effects of acute hypoxia on the V˙O2max of trained and untrained subjects. J. Sports Sci. 11:37-42, 1993.
    32. Norton, L. H., B. Squires, N. P. Craig, G. McLeary, P. McGrath, and K. I. Norton. Accuracy of pulse oximetry during exercise stress testing. Int. J. Sports Med. 13:523-527, 1992.
    33. Paterson, D. J., H. Pinnington, A. R. Pearce, and A. L. Morton. Maximal exercise cardiorespiratory responses of men and women during acute exposure to hypoxia. Aviat. Space Environ. Med. 58:243-247, 1987.
    34. Pedhazur, E. J. Multiple Regression in Behavioral Research, 3rd Ed. Fort Worth: Harcourt Brace, 1997, pp. 170-174.
    35. Pugh, L. G. C. E., M. B. Gill, S. Lahirim, J. S. Milldege, M. P. Ward, and J. B. West. Muscular exercise at great altitudes. J. Appl. Physiol. 19:431-440, 1964.
    36. Reeves, J. T., B. M. Groves, J. R. Sutton, P. D. Wagner, A. Cymerman, M. K. Malconian, P. B. Brock, P. M. Young, and C. S. Houston. Operation Everest II: preservation of cardiac function at extreme altitude. J. Appl. Physiol. 63:531-539, 1987.
    37. Richardson, R. S., D. R. Knight, D. C. Poole, S. Sadi Kurdak, M. C. Hogan, B. Grassi, and P. D. Wagner. Determinants of maximal exercise V˙O2 during single leg knee-extensor exercise in humans. Am. J. Physiol. 268:H1453-H1461, 1995.
    38. Robergs, R. A., and S. Roberts. Exercise Physiology: Sports, Performance and Clinical Applications. St. Louis: Mosby Year-Book, 1997, pp. 647-649.
    39. Roca, J., M. C. Hogan, D. Story, D. E. Bebout, P. Haab, R. Gonzalez, O. Euno, and P. D. Wagner. Evidence for tissue diffusion limitation of V˙O2max in normal humans. J. Appl. Physiol. 67:291-299, 1989.
    40. Saltin, B., R. F. Grover, C. G. Blomqvist, L. H. Harley, and R. L. Johnson, Jr. Maximal oxygen uptake and cardiac output after 2 weeks at 4,300 m. J. Appl. Physiol. 45:400-409, 1968.
    41. Schoene, R. B., R. C. Roach, P. H. Hackett, J. R. Sutton, A. Cymerman, and C. S. Houston. Operation Everest II: ventilatory adaptation during gradual decompression to extreme altitude. Med. Sci. Sports Exerc. 22:804-810, 1990.
    42. Shephard, R. J., E. Bouhlel, H. Vandewalle, and H. Monod. Muscle mass as a factor limiting physical work. J. Appl. Physiol. 64:1472-1479, 1988.
    43. Shephard, R. J., E. Bouhlel, H. Vandewalle, and H. Monod. Peak oxygen intake and hypoxia: influence of physical fitness. Int. J. Sports Med. 9:279-283, 1988.
    44. Siri, W. E. Body composition from fluid spaces and density: analysis of methods. In Techniques for Measuring Body Composition. J. Brozek and A. Henschel (Eds.). Washington DC: National Academy of Sciences, 223-224, 1961.
    45. Squires, R. W., and E. R. Buskirk. Aerobic capacity during acute exposure to simulated altitude, 914-2,286 meters. Med. Sci. Sports Exerc. 14:36-40, 1982.
    46. Sutton, J. R., J. T. Reeves, P. D. Wagner, B. M. Groves, A. Cymerman, M. K. Malconian, P. B. Brock, P. M. Young, S. D. Walter, and C. S. Houston. Operation Everest II: oxygen transport during exercise at extreme simulated altitude. J. Appl. Physiol. 64:1309-1321, 1988.
    47. Terrados, N., M. Mizuno, and H. Anderson. Reduction in maximal oxygen uptake at low altitudes: role of training status and lung function. Clin. Physiol. 5(Suppl. 3):75-79, 1985.
    48. Wagner, P. D. Muscle O2 transport and O2 dependent control of metabolism. Med. Sci. Sports Exerc. 27:47-53, 1995.
    49. Wagner, P. D. A theoretical analysis of the factors determining VO2max at sea level and altitude. Respir. Physiol. 106:329-343, 1996.
    50. Young, A. J., A. Cymerman, and R. L. Burse. The influence of cardiorespiratory fitness on the decrement in maximal aerobic power at high altitude. Eur. J. Appl. Physiol. 54:12-15, 1985.
    51. Young, P. M., J. R. Sutton, H. J. Green, J. T. Reeves, P. B. Rock, C. S. Houston, and A. Cymerman. Operation Everest II: metabolic and hormonal responses to incremental exercise to exhaustion. J. Appl. Physiol. 73:2574-2579, 1992.
    Back to Top | Article Outline


    Validation of finger pulse oximetry. Use of the Poet oximeter during exercise had previously been validated to arterial blood SaO2 during exercise to V˙O2max on six highly trained competitive cyclists. Before the exercise test, 1% lidocaine was injected superficial to the distal radial artery, a 20-gauge catheter was placed percutaneously, and a three-way stopcock was attached to the catheter. Approximately 2 mL of dead space blood was sampled and discarded before each 3-mL specimen collection. Exercise was characterized by 45 min of steady state cycling, followed immediately by a 2-min stage duration incremental protocol to peak V˙O2. A blood specimen was collected at rest, after 30 and 45 min of exercise, and at peak V˙O2. Simultaneous recordings of hemoglobin saturation were recorded using finger pulse oximetry. The arterial blood samples were used for determination of PaO2, PaCO2, pH, and calculated hemoglobin saturation at a core temperature of 37°C by automated blood gas analysis (GEM Premier, Mallinckrodt Sensor Systems). Quality control calibrations of the blood gas analyzer were performed twice daily resulting in an accuracy in the measurement of PaO2, PaCO2, and pH of ± 1 mm Hg, ± 1 mm Hg, and ± 1%, respectively.

    Data from six males performing steady state and incremental cycle ergometry exercise to peak V˙O2 was used to compare SaO2 measurements obtained from finger pulse oximetry to those calculated from measurement of blood partial pressures of oxygen sampled from the radial artery (N = 24 paired data points). Values for arterial blood SaO2 ranged from 98 to 89%, whereas those estimated by pulse oximetry ranged from 99 to 87%. Assessment of the accuracy of the Poet finger pulse oximeter during exercise was conducted by quantifying agreement as proposed by Bland and Altman (5). Differences between pulse oximetry and the calculated intra-arterial SaO2 from measurements of arterial PO2 were determined, and the mean ± SD of these differences was computed. These data were supported by correlation analysis, and reporting of the slope and y intercept of the line of best fit. SaO2 determined from arterial blood and pulse oximetry had close and acceptable agreement (Fig. 5) (r = 0.798, slope = 0.918, y intercept = 6.615%, mean ± SD of differences = −1.01 ± 1.84%) (5), indicating the validity of finger pulse oximetry during exercise to V˙O2max (32).

    Figure 5

    Figure 5



    © Williams & Wilkins 1998. All Rights Reserved.