Validation of Maximal,Submaximal,and Nonexercise Indirect V ˙ O 2 max Estimations at 2600 m Altitude

Introduction

Cardiorespiratory fitness (CRF), the measure of an individual's functional ability, has been fully recognized as an independent predictor of all-cause and cardiovascular mortality (Blair et al., 1989, 1996; Kodama et al., 2009), even more powerful than other traditional cardiovascular risk factors (Myers et al., 2002; Katzmarzyk et al., 2005; Blair, 2009). Moreover, CRF has been inversely associated with the risk of developing type 2 diabetes mellitus (Sui et al., 2008; Lee et al., 2009) and dying from cancer (Schmid and Leitzmann, 2015).

Accordingly, the assessment of CRF at both patient (Ross et al., 2016) and population level (Sanders and Duncan, 2006; Kaminsky et al., 2013, 2017; Laukkanen et al., 2016) is increasingly recognized as a part of worldwide efforts to understand, characterize, classify, follow up, and mitigate health risks. Maximal oxygen uptake ( V ˙ O₂max) is the accepted criterion measure of CRF (Howley et al., 1995; Bassett and Howley, 1997). Although direct measurement of V ˙ O₂max by respiratory, breath-by-breath, gas exchange analysis is the most accurate method, it is impractical for large-scale epidemiological studies (Matthews et al., 1999; Castro-Pinero et al., 2010; Artero et al., 2014). Therefore, indirect estimations of V ˙ O₂max are frequently used to assess large populations or when access to direct measurement is limited.

A number of studies have validated indirect V ˙ O₂max tests in a variety of populations, most of them living at sea level (Leger et al., 1988; Oja et al., 1991; Sproule et al., 1993; Grant et al., 1995; George et al., 1997; Enright et al., 2003; Stickland et al., 2003; Aandstad et al., 2011; Beatty et al., 2012). However, very few studies have addressed the criterion validity of indirect V ˙ O₂max estimation methods when applied at higher altitudes (Gonzalez Camarena and Carrasco Sosa, 1989; Ahmaidi et al., 1993; Falgairette et al., 1994). Previous research has shown that V ˙ O₂max can be affected by altitude (Fulco et al., 1998; Boning et al., 2001; Schmidt et al., 2002); it is not clear to what extent validations, conducted at sea level, can be extrapolated to populations living at moderate altitude (2000 to 3000 m above sea level) (Bartsch et al., 2008). Most V ˙ O₂max estimation approaches employ a variable related to heart rate (HR) in their calculations, and the chronotropic response to exercise can be markedly affected by altitude (Richalet et al., 1992, 2015; Hernandez Lopez et al., 2000; Mourot, 2018), possibly biasing results to an unknown extent. Moreover, several physiological adaptations to moderate altitude can potentially alter V ˙ O₂max estimation under laboratory or field conditions. Perception of maximal exertion, central and peripheral fatigue, cognitive performance (Ando et al., 2019; Heinrich et al., 2019), effects of altitude-induced pH changes and chronic hypoxia on voluntary muscle contractility (Siegler and Marshall, 2015; Ruggiero et al., 2020), changes in hemoglobin mass, blood rheology, and pulmonary hemodynamics after exposure to moderate altitude (Saunders et al., 2013; Raberin et al., 2019), as well as changes in circulating metabolites and hormones, substrate utilization, and respiratory exchange ratio (RER) during exercise can all influence perceptions of effort in laboratory or field tests (Katayama et al., 2010).

Important issues to consider when choosing a test for CRF estimation include accuracy, simplicity, and applicability in specific populations and environments. The 20-m shuttle-run test (20-MST) (Leger et al., 1988), the 2-km walking test (UKK) (Oja et al., 1991), and the perceived functional ability–physical activity rating questionnaire (PFA-PAR) (George et al., 1997) meet these criteria. However, these methods were developed at sea level/low altitude and for the most part, have not been validated at moderate altitude, with the exception of a study in children at high altitude (3700 m) (Falgairette et al., 1994).

Worldwide, an estimated 130 to 140 million people live at moderate altitude (Cohen and Small, 1998; Peñaloza, 2012). Most of them live in Andean cities such as Bogota (2600 m) and Quito (2850 m), on the Mexican plateau (Mexico City, 2250 m), and in the African and south-central Asian highlands. As these are mostly low-to-middle income nations, young and middle-aged adults account for the largest segment of these populations. Therefore, valid CRF assessment methods are needed for use in this context. The purpose of this study was to assess the criterion validity of three equations for V ˙ O₂max estimation based on two exercise-based methods: The 20-MST (a maximal test) and the UKK (a submaximal test); and one nonexercise method: The PFA-PAR questionnaire, in a group of young healthy adults living at 2600 m.

Materials and Methods

The study population consisted of young, apparently healthy medical students who had resided at 2600 m above sea level altitude for at least 2 years. Volunteers were recruited by posters and fliers installed at Universidad del Rosario's Medical School in Bogotá, Colombia. All procedures were reviewed and approved by the institutional review board (Comité de Ética en Investigación, Universidad del Rosario). After providing written informed consent, subjects were screened for cardiovascular exercise contraindications, completed a brief Physical Activity Readiness Questionnaire (Thomas et al., 1992), and provided information on smoking and physical activity habits. Physical inactivity was defined as reporting less than three 30-minute bouts of moderate intensity physical activity per week. Body weight was measured to the nearest 0.1 kg by using a calibrated physician's beam scale (model 338; Detecto, Inc., Webb City, MO), and height was measured to the nearest 1.0 cm by using a stadiometer (Shorr Productions, Olney, MD). Body-mass index (BMI) was calculated as weight in kilograms divided by height in meters squared and rounded to one decimal place (kg/m²). Percent body fat was estimated by using a six-site skinfold method (Yuhasz, 1974) performed by a trained tester. In addition, fasting blood samples were drawn for complete blood count determination and basic clinical chemistry.

Results

We studied 64 adult participants (30 female, 34 male), aged 20.8 ± 1.7, most of them (78%) being physically inactive and 19% being smokers. Average BMI, percent body fat, hemoglobin, hematocrit, fasting plasma glucose, and plasma lipids were within normal ranges (Table 1). The prevalence of excess body weight (BMI ≥25 kg/m²) was 32% for men and 13% for women. As expected for the altitude, hemoglobin values were relatively high, and more markedly so in men than women. All smokers consumed fewer than 10 cigarettes per day.

Directly measured V ˙ O₂max was higher in men (41.2 ± 5.8 mL/kg/min) than in women (32.2 ± 3.6 mL/kg/min; p < 0.001). Similar gender differences were seen for the V ˙ O₂max estimates provided by the 20-MST, the UKK, and the PFA-PAR questionnaire (Table 2). The 20-MST elicited the highest average peak heart rate (HRpeak) among the exercise protocols (194.2 ± 8.1 bpm [97.5% of predicted]), followed by the direct VO₂ measurement (189.2 ± 8.5 bpm [95% of predicted] and 165.9 ± 18.3 bpm [83.3% of predicted]) for the UKK (Table 2).

We found the highest Spearman correlation coefficient between an indirect test and the direct V ˙ O₂max measurement for the 20-MST (r = 0.80; 95% CI: 0.70–0.90), followed by the PFA-PAR (r = 0.72; 95% CI: 0.60–0.84) and the UKK (r = 0.69; 95% CI: 0.57–0.82) (Table 3). The LCC showed considerable variation, with the highest value among indirect methods for the 20-MST, whereas UKK and PFA-PAR displayed lower coefficients (Table 3). Adjustment by altitude greatly improved the agreement between PFA-PAR and the direct method, from 0.57 to 0.71. The SEEs versus the direct method were similar for the four alternative methods, but lowest for 20-MST (3.91 mL/kg/min for 20-MST, 5.48 mL/kg/min for UKK, 4.75 mL/kg/min for PFA, and 4.12 mL/kg/min for altitude-adjusted PFA-PAR).

Bland-Altman analyses of agreement showed that the 20-MST tended to over-estimate V ˙ O₂max relative to the direct determination (mean bias 1.68 mL/kg/min; 95% CI: 0.66–2.69). Meanwhile, the UKK showed no significant mean bias (0.70 mL/kg/min; 95% CI: −0.72 to 2.13). The PFA-PAR also over-estimated V ˙ O₂max (mean bias 4.91 mL/kg/min; 95% CI: 3.57–6.26), but this effect disappeared when PFA-PAR values were adjusted by altitude (mean bias −0.53 mL/kg/min; 95% CI: −1.80 to 0.75) (Fig. 1). Bland-Altman limits of agreement were narrowest for the 20-MST (−6.46 to 9.81) and widest for the UKK (−10.68 to 12.09).

OPEN IN VIEWER

FIG. 1.

Bland-Altman plots of agreement between the direct V ˙ O₂max measurement and the indirect estimation methods. (a) 20-MST, (b) UKK: 2-km walk test, (c) perceived functional ability–physical activity rating questionnaire, (d) altitude-adjusted perceived functional ability–physical activity rating questionnaire. Dashed lines represent the 95% limits of agreement. 20-MST, 20-m shuttle-run test; HR, heart rate; PFA-PAR, perceived functional ability–physical activity rating questionnaire; V ˙ O₂max, maximal oxygen uptake.

Discussion

Our study confirms that V ˙ O₂max can be estimated through maximal (20-MST), submaximal (UKK), and nonexercise (PFA-PAR) methods in young adult populations living at moderate altitude. Several indirect tests for V ˙ O₂max estimation have been extensively validated at low altitude. However, their criterion validity at moderate altitudes is unknown. To the best of our knowledge, this is the first study that has attempted to validate several exercise and nonexercise methods for CRF estimation at moderate altitude. Here, we show that, although previously developed regressions for the 20-MST and the UKK can be applied without modification in our study population, PFA-PAR estimations require an altitude correction factor. These findings will empower researchers at moderate altitude with proven tools to assess and follow up CRF in these populations.

Out of the three indirect methods, the 20-MST showed the highest overall agreement with the criterion measure, an anticipated finding given its maximal nature compared with the other tests. Previous studies of the 20-MST have reported comparable correlation coefficients (Sproule et al., 1993; Grant et al., 1995; McNaughton et al., 1998; Aandstad et al., 2011), SEE values (Leger et al., 1988; Suminski et al., 2004; Ruiz et al., 2009), and Bland-Altman limits of agreement (Aandstad et al., 2011) with the criterion measurement. Some studies have shown no bias between the predicted and measured V ˙ O₂max using the 20-MST in children or adults (Suminski et al., 2004; Aandstad et al., 2011), whereas others have reported significant under-prediction of V ˙ O₂max (Grant et al., 1995; Stickland et al., 2003; Ruiz et al., 2009). A study of circumpubertal boys living at low (300 m) or high altitude (2700 m) assessed V ˙ O₂max by cycle ergometry and the 20-MST (Falgairette et al., 1994), finding significant overestimation by the 20-MST at either altitude. In our study, Bland-Altman analysis also showed an over-estimation of V ˙ O₂max by the 20-MST of 1.68 mL/kg/min at moderate altitude (about half a metabolic equivalent [MET], 1 MET ≈3.5 mL/kg/min), a difference that may bear clinical significance depending on the context of the individual being tested. However, measures of variability (limits of agreement and SEE) were narrowest for the 20-MST when compared with both the UKK and the questionnaire, indicating that the 20-MST was the most precise among the methods assessed.

The UKK walk test has been used to estimate CRF in healthy (Oja et al., 1991; Laukkanen et al., 1992b), overweight (Laukkanen et al., 1992a), and exercise-trained adults (Laukkanen et al., 2000), with an overall good performance (Oja et al., 1991). In our young adult sample, the UKK showed a moderate correlation and agreement with the direct V ˙ O₂max measurement. Given that Bland-Altman analyses indicated no significant bias, a lower agreement could be explained by the lower precision of this test. This is supported by the wider limits of agreement and SEE values. As a submaximal test relying on HRpeak to estimate V ˙ O₂max, the predictive accuracy of the UKK may be lower when compared with a test where physiological mechanisms are driven to a maximal stage. This limitation can be of special concern in healthy young adults when compared with middle-age and older adults, because in the latter group the UKK probably elicits higher relative workloads due to their lower fitness levels and biomechanical limitations (Laukkanen et al., 1992a, 2000). Another factor that may limit the UKK applicability in some settings is the need for accurate HRpeak measurement. HR response to exercise can be impaired by medications, cardiac conduction abnormalities, and, possibly, the effects of altitude. In this regard, prior research has shown that HRpeak can decrease with exposure to extreme altitude (Hernandez Lopez et al., 2000; Mourot, 2018). This effect is believed to be due to increased parasympathetic tone at strenuous exercise. In this study, the chronotropic responses to maximal and submaximal exercise, assessed by HRpeak, were similar to those described for populations at sea level (Oja et al., 1991; Kaminsky et al., 2015). This suggests that the lower validity in the prediction of V ˙ O₂max by the UKK was not related to altitude-impaired chronotropic responses to exercise.

The PFA-PAR questionnaire provides an acceptable estimation of directly measured V ˙ O₂max when a correction factor is applied to the original equation (George et al., 1997) to account for altitude. In our study, Bland-Altman analyses initially demonstrated significant V ˙ O₂max overestimation when the unadjusted-PFA-PAR estimation was compared with the direct measurement. However, a linear adjustment of 13% in the PFA-PAR estimation greatly improved the test's calibration. Analogous adjustments for altitude have also been applied to CRF cutoffs in large epidemiological studies (Ramirez-Velez et al., 2017). Altitude-adjusted PFA-PAR showed an intermediate degree of agreement when compared with the 20-MST and the UKK walk test, a remarkable finding given its nonexercise nature. The correlation of PFA-PAR with the criterion measurement and SEE were within the reported ranges for nonexercised methods (Ross et al., 2016). However, we found a lower correlation compared with the seminal PFA-PAR validation paper (George et al., 1997). In that study, the validity of the questionnaire prediction was limited when subjects were unfamiliar with prolonged exercise, which could affect the subjects' ability to accurately report their subjective exercise capacity. This could have been the case in our sample, which consisted mostly of physically inactive subjects.

The average V ˙ O₂max values for our moderate altitude-acclimatized population were lower than those previously reported for young populations living at low altitude (Kaminsky et al., 2015). According to a study in healthy runners in which altitude was simulated in a hypobaric chamber (Squires and Buskirk, 1982) and a pooled analysis of several studies analyzing changes in V ˙ O₂max with altitude (Fulco et al., 1998), V ˙ O₂max for a 2600 m altitude would be reduced by ∼13%. Our study findings are in agreement with previous reports that show a reduction of about 10%–15% in the peak oxygen consumption in subjects living at 2600 m (oxygen partial pressure ∼65 mm Hg), despite adequate erythropoietic adaptations (Perry et al., 1992; Padilla et al., 1998; Boning et al., 2001, 2004; Schmidt et al., 2002). In fact, hemoglobin and hematocrit values in our sample were comparable to those reported for young untrained populations acclimatized to moderate altitude (Schmidt et al., 2002; Boning et al., 2004), supporting the notion that normal erythropoietic adaptation had already taken place. These findings are consistent with the reported effects of living at altitude, but the effects of lower aerobic power and higher hemoglobin values did not seem to affect the estimation of V ˙ O₂max via indirect field tests. In line with this result, a study in asthmatic adolescents living at moderate altitude (Ahmaidi et al., 1993) found valid V ˙ O₂max estimates using the 20-MST.

It should be noted that, although SEEs and Bland-Altman limits of agreement were within the reported ranges, indirect V ˙ O₂max estimations display a relatively low precision. In our study, the SEEs were equivalent to 1.17 METs for the 20-MST and the altitude-adjusted PFA-PAR questionnaire, and 1.54 METs for the UKK. Given the significant decrease in cardiovascular events and all-cause mortality (13% and 15%, respectively) that has been associated with having a 1-MET higher level of CRF (Kodama et al., 2009; Kaminsky et al., 2013), such a marked variation cannot be accepted at the individual level. On the other hand, the use of indirect methods in epidemiological studies is justified on the basis of their simplicity, lower cost, and the possibility of applying them to a large number of people at the same time (Castro-Pinero et al., 2010). The wide limits of agreement of all indirect methods indicate that none of them can be interchangeably used with the direct method (Bland and Altman, 1986), but for most practical purposes they are intended as a good approximation to V ˙ O₂max, and not as a replacement of the gold standard. Interestingly, it has been demonstrated that exercise (Beatty et al., 2012) and nonexercise V ˙ O₂max indirect methods (Stamatakis et al., 2013; Artero et al., 2014; Nes et al., 2014) are able to predict hard outcomes such as cardiovascular and all-cause mortality on their own, regardless of their relatively lower accuracy versus the direct measurement.

Strengths of our study include the assessment of different aspects of criterion-related validity, not just simple correlation. We employed the Lin concordance correlation coefficient, which performed far better than linear correlation coefficients in assessing overall agreement between the direct and indirect methods. Some studies rely on correlation coefficients to study the validity of an indirect V ˙ O₂max method (Weiglein et al., 2011; Mayorga-Vega et al., 2015), but this approach does not account for the spread of the data or changes in the measurement scale, and it has been extensively discouraged (Bland and Altman, 1986). A significant limitation of our study was the relatively narrow age range of participants (young medical students), which restricts its generalizability to populations with similar demographics. Another relevant limitation could be the nonrandomized order in which we performed the field tests. Still, this is unlikely to have impacted our results because the UKK walk test did not represent a strenuous effort for our young study population, and because the washout period between the two assessments was long enough to allow physiological parameters to return to baseline. Another limitation is the lack of information regarding birthplace of the study participants and their ancestors. It has been shown that adaptations to chronic hypoxia are transmitted between generations (Marconi et al., 2004). However, all subjects had lived at 2600 m for at least 2 years, a period in which acute and chronic adaptations to moderate altitude had already taken place. Further, we acknowledge that our results are applicable only to similar populations of young healthy adults who are not endurance trained and live at moderate altitude. Future studies in which paired estimations can be performed in the same individuals at moderate altitude and at sea level, allowing for an appropriate acclimation time, will confirm and provide greater robustness to our findings.

Conclusion

Adequate estimates of CRF for epidemiological studies can be obtained through maximal, submaximal, and nonexercise methods for populations living at moderate altitude. Among the methods we tested, the highest agreement with the criterion measurement was found for the 20-MST. Given the significant number of young adults living at moderate altitude throughout the world, validation of simple tools for estimating their aerobic power should encourage epidemiological CRF determination in these populations.