Journal Logo

APPLIED SCIENCES

Predicting V˙O2max with an Objectively Measured Physical Activity in Japanese Women

CAO, ZHEN-BO1; MIYATAKE, NOBUYUKI2; HIGUCHI, MITSURU3; MIYACHI, MOTOHIKO1; ISHIKAWA-TAKATA, KAZUKO1; TABATA, IZUMI1

Author Information
Medicine & Science in Sports & Exercise: January 2010 - Volume 42 - Issue 1 - p 179-186
doi: 10.1249/MSS.0b013e3181af238d
  • Free

Abstract

Cardiorespiratory fitness is known to be an objective, reproducible measure that reflects the functional consequences of physical training (36) and recent physical activity (PA) habits, and is a powerful predictor of chronic disease morbidity and mortality. Prospective observational studies have shown that low cardiorespiratory fitness is strongly associated with the risk for developing coronary heart disease (13), hypertension (1), type 2 diabetes mellitus (32), and metabolic syndrome (20) as well as mortality from cardiovascular disease (13,17), cancer (33), and all causes of mortality (17). Hence, cardiorespiratory fitness has been suggested to be included in the European Health Monitoring System for the adult population (35), National Health and Nutrition Examination Survey in the United States (4), and the Exercise and Physical Activity Reference for Health Promotion 2006 in Japan (26) in which the recommended reference value for maximal oxygen uptake (V˙O2max) to prevent lifestyle-related disease was reported. Although cardiorespiratory fitness is an important health indicator, cardiorespiratory fitness assessment is usually not performed in many health care settings because of the absence of feasible and practical assessment methods.

The impracticality of maximal and submaximal exercise tests for the assessment of cardiorespiratory fitness in the general public is well recognized. Therefore, a variety of nonexercise prediction models (2,7,12,15,21,22,30,38) have been developed as alternative approaches to fitness assessment and the estimation of V˙O2max. Those nonexercise models are effective for use in large epidemiological cohorts in which exercise tests to predict or measure V˙O2max would be impractical. Those previous reports, however, have relied on a subjective self-reported PA measure, which, when compared with objectively measured PA, have been shown to have low correlations in the range of 0.14-0.53 and to suffer from social desirability and recall biases (31). In addition, perhaps the greatest limitation of subjective self-reported PA measures is their inability to accurately assess unstructured and incidental ambulatory PA, which may account for a greater proportion of total PA in sedentary people. To overcome the deficiencies of existing nonexercise prediction models, Cao et al. (3) developed a nonexercise prediction model for estimating V˙O2max using an objectively measured PA variable, pedometer-determined daily step counts (SC). They demonstrated that SC was useful in predicting V˙O2max variance and helped their nonexercise V˙O2max prediction model generate relatively accurate estimations of V˙O2max in Japanese women. However, SC alone does not discriminate the intensity of movement or reflect the amount of time spent in specific intensity categories of PA, which may weaken the accuracy of that prediction model. The need for further study to investigate the ability of PA intensity variables to improve their prediction accuracy has been emphasized.

Accelerometers are widely accepted as valid objective measurement tools that allow researchers to estimate how much energy individuals are expending, as well as to quantify the SC and the amount of time spent in light, moderate, and vigorous PA (VPA), which correspond to <3 metabolic equivalents (METs), 3 to 6 METs and >6 METs, respectively. Recently, accelerometry has been used in the monitoring of the levels of PA in populations (9,25). In 2005, Plasqui and Westerterp (27,28) developed a nonexercise model for estimating V˙O2max using a fitness index based on accelerometer counts and HR, and they cross-validated this model in 2006. They reported that this fitness index contributed significantly to the explained variation in V˙O2max, and the total explained variation of their nonexercise prediction model was 71%, with a SEE of 409 mL·min−1, or 13.7% of the average V˙O2max. However, this prediction model may be less feasible for use in certain clinical applications because of the cost and technical requirements for its use and because it uses a noncommercially available accelerometer. There are several accelerometry-based PA monitors that are commercially available. The Kenz Lifecorder (LC; SUZUKEN Co Ltd, Nagoya, Japan) is a recent addition to the growing number of uniaxial accelerometer options; it offers comparable instrument outputs with several potentially attractive features for researchers and practitioners. The LC has displayed reasonable estimates of PA intensity and energy expenditures under controlled conditions on a treadmill (19), over 24 h of typical daily activities undertaken in a respiratory chamber (19) and in a free-living environment using doubly labeled water as the criterion method (39). Furthermore, compared with many other accelerometers, the LC is somewhat more affordable and can potentially simplify the data interpretation process by reducing the time spent and the need for advanced technical expertise or software programs (24). However, to our knowledge, there is no information on the prediction of V˙O2max using the accelerometer-determined time spent in moderate to vigorous PA (MVPA) or VPA as the objective PA variables.

In the present study, we hypothesized that the accelerometer-determined PA intensity variables including the time spent in MVPA and VPA are potential predictors of cardiorespiratory fitness in Japanese women. To verify this hypothesis, the relationships between PA intensity variables and V˙O2max were investigated. More specifically, the purpose of this study was to develop new nonexercise V˙O2max prediction models using SC and the time spent in MVPA or VPA as the objective PA variables as well as additional covariates including age and body composition in Japanese women.

METHODS

The present study consists of three parts. First, the relationships between PA intensity variables and V˙O2max were investigated. Second, validation procedures were used to develop new nonexercise prediction models that included body composition and objectively measured PA variables as predictor variables. Finally, the accuracy of the new nonexercise prediction models was assessed using two separate cross-validation procedures.

Subjects.

One hundred and forty-eight Japanese women aged 20 to 69 yr participated in the present study. None of the subjects had any chronic diseases or were taking any medications that could affect the study variables. Seventy-six healthy women were tested in two independent institutions supervised by two of the coauthors N.M. and M.H., and 72 healthy women were tested in another institution by C.Z., M.M., and I.T. All subjects provided written informed consent according to local institute policy before the measurement of physical fitness. The research project was approved by the Ethical Committee of the National Institute of Health and Nutrition. The subjects' characteristics are described in Table 1.

TABLE 1
TABLE 1:
Physical characteristics of the study subjects.

Anthropometrics.

Body mass was measured using an electronic scale (Inner Scan BC-600; Tanita Co., Tokyo, Japan) and was determined to the nearest 0.1 kg. Height was measured to the nearest 0.1 cm using a stadiometer (YL-65; Yagami Inc., Nagoya, Japan). Body mass and height were measured with the subjects wearing light clothing and no shoes. Body mass index (BMI) was calculated by dividing the body mass in kilograms by the square of height in meters (kg·m−2). Waist circumference (WC) was measured at the umbilical level with an inelastic measuring tape at the end of normal expiration to the nearest 0.1 cm.

Maximal aerobic power.

V˙O2max was measured using a maximal graded exercise test (GXT) with bicycle ergometers [Lode Excalibur (N.M.); Lode BV, Groningen, The Netherlands; Monark Ergomedic 828E (M.H., C.Z.), Varberg, Sweden]. The initial workload was 30-60 W, and the work rate was increased thereafter by 15 W·min−1 until the subject could not maintain the required pedaling frequency (60 rpm). HR (WEP-7404; NIHON KOHDEN Corp., Tokyo, Japan) and a rating of perceived exertion were monitored throughout the exercise. During the progressive exercise test, the expired gas of subjects who were tested by two of the coauthors, NM and MH, was collected, and the rates of oxygen consumption (V˙O2) and carbon dioxide production (V˙CO2) were measured and averaged over 30-s intervals using an automated breath-by-breath gas analyzing system [Aeromonitor AE-280S (M.H.); Minato Medical Science, Tokyo, Japan; Oxycon Alpha (N.M.), Mijnhardt b.v., The Netherlands]. The Aeromonitor AE-280S consists of a microcomputer, a hot-wire flow meter, and oxygen and carbon dioxide gas analyzers (a zirconium element-based oxygen analyzer and an infrared carbon dioxide analyzer). Gas was sampled at the rate of 220 mL·min−1 through a filter by a suction pump through the analyzers. The Oxycon Alpha consists of a microcomputer, a capillaryline, and oxygen and carbon dioxide gas analyzers (O2, differential paramagnetic; CO2, infrared absorption). Expiratory volumes were determined using a Triple V turbine volume sensor which was calibrated before each test according to the manufacturer's instructions. The systems were calibrated before each test with gases of known concentration. The expired air of subjects who were tested in the institution by C.Z., M.M., and I.T. (CZ) was collected over 30-s intervals in Douglas bags. An oxygen and carbon dioxide mass spectrometer (Arco-1000; Arco System, Ogaki, Japan) was used to analyze oxygen and carbon dioxide concentrations. The volume of expired air was determined using a dry gas volume meter (DC-5; Shinagawa Seisakusho, Tokyo, Japan) and converted to STPD. During the latter stages of the test, each subject was verbally encouraged by the test operators to give a maximal effort. Achievement of V˙O2max was accepted if two of the following conditions were met: subject's maximal HR was > 95% the age-predicted maximal HR (220 − age), and the V˙O2 curve showed a leveling off.

Physical activity.

PA was measured by activity monitors using a uniaxial acceleration sensor (LC) and a triaxial acceleration sensor (AM; Panasonic Electric Work Co., Ltd, Osaka, Japan). Subjects were instructed in how to use the instrument and were told to wear it on their belt or waistband at the right midline of the thigh from the moment they got up until they went to bed except while bathing or swimming for seven consecutive days. The activity monitor was firmly attached to their clothes at the waist with the aid of a clip. The technical and estimation equation details of the LC and AM have been described in previous studies (10,19,39). LC and AM have the same measurement range of acceleration, are similar in size, use simple regression models to convert acceleration into PA intensity, do not need any individual calibrations, and easily export the data from the software into an Excel spreadsheet. Previous studies have shown that both LC and AM have displayed reasonable estimates of energy expenditures under controlled conditions and in free-living conditions (19,39). Yamada et al. (39) found that the correlation between the total energy expenditure (TEE) measured by LC and the TEE measured by AM was high (r = 0.94, P < 0.001). They also found that no significant differences between the two activity monitors were observed in the time spent in MVPA and VPA. Furthermore, unpublished data from our institute showed that no significant differences between the two activity monitors were observed in SC (N = 39, average age 40.4 ± 9.6 yr, LC: 8557 ± 2573, AM: 8690 ± 2859, P = 0.69). The accelerometer-determined PA variables used in the present study for analyses included SC, MVPA, and VPA, but not TEE. Thus, it was reasonable to use two activity monitors in the present study, although Yamada et al. reported that the LC did not correlate with doubly labeled water (DLW)-derived TEE, and the AM did, after statistically controlling for the influence of age, weight, height, and %fat.

Statistical analyses.

Measured and calculated values are presented as means ± SD. Pearson's product correlations were calculated between the independent variables (age, BMI, WC, SC, MVPA, and VPA) and V˙O2max. Hierarchical linear regression analysis was used to generate prediction equations for V˙O2max. We entered the age, a different body composition measure (i.e., BMI or WC), and SC into the first block and PA intensity variables into the second block. Because the outcome measurements were performed at different institutions and different activity monitors were used, the effects of institution and activity monitors were assessed by adding a dummy-coded institution variable and an activity monitor variable and then applying a multiple regression to determine whether the institution variable and activity monitor variable provided a significant increase in the explained variance of V˙O2max over the independent variable. The goodness of fit and precision of the regression equations were evaluated using multiple coefficient of determination (R2) and the absolute SEE and relative SEE (%SEE). The new nonexercise prediction models were assessed using two separate cross-validation procedures, using the predicted residual sum of squares (PRESS) method (14) and various subsamples of the entire sample. The PRESS method of cross-validation is based on the error in prediction for each case when only that case is deleted from the model-generating process. The PRESS adjusted R2 (R2p) can be calculated as 1 − (PRESS/SStotal). The PRESS SEE (SEEp) can be calculated using the following equation:

The models were further examined for accuracy by dividing the entire sample into subgroupings of age, SC, and V˙O2max, and then by comparing the constant errors among these subgroupings (CE). All analyses were done with SPSS Advanced Models 16.0J for Windows (SPSS Japan Inc., Tokyo, Japan). The statistical significance level was set at P < 0.05.

RESULTS

Results from cardiorespiratory fitness testing for V˙O2max, anthropometric variables, and PA variables are presented in Table 1. The highly varied nature of the sample is reflected by the respective physical characteristics data ranges.

Table 2 presents the Pearson correlations matrix of V˙O2max and all independent variables. These correlations between V˙O2max and all independent variables were statistically significant (P < 0.01) and ranged from a low of 0.43 for SC to a high of −0.62 for WC, indicating that each independent variable was related to V˙O2max. The correlation coefficients between V˙O2max and MVPA or VPA were significantly higher than the correlation coefficient between V˙O2max and SC. After statistically controlling for the influence of age using partial correlation analysis, the correlations between V˙O2max and SC, MVPA, and VPA significantly increased to 0.55, 0.54, and 0.60, respectively.

TABLE 2
TABLE 2:
Correlations matrix of V˙O2max and independent variables.

Table 3 shows the multiple regression analysis. All variables used in the model were independently related to V˙O2max. Among the BMI and WC prediction models in the current study, the WC modelVPA showed the highest multiple correlations and lowest SEE. When estimating V˙O2max with age, body composition, and SC, the addition of MVPA raised the R2 from 0.648 to 0.694 for the BMI modelMVPA and from 0.681 to 0.716 for the WC modelMVPA, representing increases of 7.1% for the BMI modelMVPA and 5.1% for the WC modelMVPA in the explained variance of V˙O2max. VPA significantly increased the explained variance in V˙O2max by an additional 12.3% in the BMI modelVPA and 9.4% in the WC modelVPA, and decreased the SEE by 0.408 mL·kg−1·min−1 in the BMI modelVPA and 0.334 mL·kg−1·min−1 in the WC modelVPA. When the institution variable or activity monitor variable as an independent variable was added to the multiple regressions, we found that the institution variable and activity monitor variable were not statistically significant (P > 0.05) and produced no appreciable difference in the accuracy of the models (R2Δ < 0.009, data not shown). The cross-validation results of the PRESS method are also shown in Table 3. The shrinkage of R2 and the increment of SEE for each prediction model were minor, particularly in modelVPA: R2 obtained from the BMI modelVPA and WC modelVPA, respectively, decreased by 0.014 and 0.011, and the SEE obtained from the BMI modelVPA and WC modelVPA increased by 0.092 mL·kg−1·min−1 and 0.073 mL·kg−1·min−1, respectively.

TABLE 3
TABLE 3:
Multiple regression non-exercise models estimating V˙O2max (mL·kg−1·min−1) in the entire sample.

The second stage of the cross-validation analysis was to examine the accuracy of the models by analyzing and comparing the CE and the SD of the CE for various subsamples of the sample. These results are provided in Tables 4 and 5. Table 4 suggests that modelMVPA is most accurate in predicting V˙O2max for individuals who are older (>50 yr), more active (SC ≥ 10,000 steps·d−1), and with average fitness (25-35 mL·kg−1·min−1). Subgroups of V˙O2max showed high absolute CE values (>2.6 mL·kg−1·min−1) in the high fitness subgroup. When prediction modelVPA was applied to the subgroups (Table 5), the CE values (<0.8 mL·kg−1·min−1) for all ages and SC groupings except for individuals who were less active (CE < 1.11 mL·kg−1·min−1) and for the average fitness subgroup (25-35 mL·kg−1·min−1) were small. At the extremes of fitness, the modelVPA systematically overestimated or underestimated V˙O2max by about 2.9 mL·kg−1·min−1. Figures 1 and 2 show the tendency for modelMVPA and modelVPA to consistently underestimate for individuals with high fitness. For most of those subjects whose V˙O2max was found to be ≥ 37 mL·kg−1·min−1, the prediction models systematically underestimated V˙O2max, and their average measured V˙O2max value was 40.3 ± 4.0 mL·kg−1·min−1 (N = 25). The average V˙O2max estimated by the models for these subjects, however, were, for the BMI and WC modelsMVPA, 37.3 ± 3.5 and 37.4 ± 3.7 mL·kg−1·min−1, respectively, and for the BMI and WC modelsVPA, 37.9 ± 4.8 and 38.0 ± 5.1 mL·kg−1·min−1, respectively.

TABLE 4
TABLE 4:
Constant error (CE) and SD for subgroups of the entire sample.
TABLE 5
TABLE 5:
Constant error (CE) and SD for subgroups of the entire sample.
FIGURE 1
FIGURE 1:
Relationships between the measured and predicted V˙O2max values for the multiple regression modelMVPA. The solid line is the line of equality (measured V˙O2max = predicted V˙O2max). The areas within the dashed lines show where the modelMVPA tends to underestimate V˙O2max. Data from the entire sample (N = 148) were used for the analysis.
FIGURE 2
FIGURE 2:
Relationships between the measured and predicted V˙O2max values for the multiple regression modelVPA. The solid line is the line of equality (measured V˙O2max = predicted V˙O2max). The areas within the dashed lines show where the modelVPA tends to underestimate V˙O2max. Data from the entire sample (N = 148) were used for the analysis.

DISCUSSION

The results of the present study showed that the PA variables of the time spent in MVPA and VPA were significantly related to V˙O2max, thus supporting our first hypothesis that the accelerometer-determined PA intensity variables including the time spent in MVPA and VPA are potential predictors of cardiorespiratory fitness in Japanese women. Furthermore, this study indicates that the nonexercise model for the prediction of V˙O2max can be substantially improved by the inclusion of both of these objectively measured PA variables of the time spent in MVPA and VPA, which are easily and reliably measured using an accelerometer.

A positive relationship between the objectively measured intensity of PA and cardiorespiratory fitness has been established in youth (5,6,8). However, few studies have reported this relationship in adults. Hebestreit et al. (11) conducted a study of 71 patients with cystic fibrosis (aged 12-40 yr) and found MVPA, assessed by accelerometer, to be significantly associated (r = 0.55, P < 0.001) with V˙O2max, as assessed by a GXT with a bicycle ergometer. However, they did not assess the relationship between VPA and V˙O2max. The present study is the first to examine the relationships between accelerometer-determined MVPA and VPA and V˙O2max in healthy adults. The results of the present study demonstrated that each of the independent variables used in this study was independently related to V˙O2max. The correlation coefficient of 0.52 between accelerometer-determined MVPA and V˙O2max found in this study was similar to that found in patients with cystic fibrosis (11) and was higher than the correlation coefficient (r = 0.25) in children reported by Dencker et al. (6). Furthermore, the results of the present study also showed stronger relationship (r = 0.58) between VPA and V˙O2max compared with those in children (r = 0.30-0.38) (5,6) and adolescents (r = 0.45) (8). The relationships tended to be stronger for VPA than that for MVPA, which is consistent with other findings reported in children (6). These findings suggest that the proportion of the variance in cardiorespiratory fitness explained by VPA increases with aging and that VPA make a bigger contribution to the variance in cardiorespiratory fitness compared with MVPA. Those prior studies, in conjunction with the present study, document the value of using accelerometer-determined PA intensity variables including the time spent in MVPA and VPA when estimating V˙O2max.

In our previous study, an equation was developed to predict V˙O2max from age, BMI, and SC in Japanese women. We found that SC was useful in predicting V˙O2max variance and helped the nonexercise V˙O2max prediction model generate relatively accurate estimations of V˙O2max in Japanese women. However, SC alone does not discriminate the intensity of movement or reflect the amount of time spent in specific intensity categories of PA, which may weaken the accuracy of that prediction model. To further increase the accuracy of a nonexercise prediction model, a new equation using additional accelerometer-determined PA intensity variables including the time spent in MVPA and VPA was developed. The new nonexercise equations in the present study resulted in a validity coefficient of R ranging from 0.83 to 0.85 for modelMVPA and 0.85 to 0.86 for modelVPA, and a value of SEE ranging from 3.14 to 3.29 mL·kg−1·min−1 for modelMVPA and 2.98 to 3.11 mL·kg−1·min−1 for modelVPA (Table 3). Previously published nonexercise test prediction models reported varying success in predicting a measure of cardiorespiratory fitness, with SEE and R values ranging from 3.44 to 8.63 mL·kg−1·min−1 and 0.46 to 0.88, respectively (2,7,12,15,16,18,21,22,27-30,34,37,38). In addition, the SEE for the nonexercise prediction modelVPAs in this study were lower than the 10-20% values reported for most submaximal exercise methods used to estimate V˙O2max (23). Therefore, the R values determined by the regression model in the present study were within the range of those associated with previous nonexercise methods for estimating V˙O2max, and the SEE values were lower than those previous values. To examine the unique contribution of accelerometer-determined PA intensity variables in addition to the previous objectively measured PA variable, we used hierarchical linear regression to develop the model. In the present study, the regression equation yielded R2 values ranging from 0.65 to 0.68 when using age, body composition (BMI or WC), and SC. The coefficient of determination increased to R2 values ranging from 0.69 to 0.72 when MVPA was added to the equations. When VPA was added to the equations as a surrogate for MVPA, the regression model resulted in R2 values ranging from 0.73 to 0.75. Those results indicated that the accelerometer-determined PA intensity variables, which have not been used in previously developed equations, substantially improved the accuracy of the estimation of V˙O2max in adult women when compared with the use of age, BMI, and SC alone. Furthermore, the results of the present study also showed that modelVPA was more accurate than modelMVPA.

To estimate the prediction model's performance, we conducted two cross-validation analyses based on PRESS and various subsamples of the sample. For the PRESS procedure, the shrinkage of the coefficient of determination (< 0.014) and the increment of SEE (<0.092 mL·kg−1·min−1) for each prediction model were minor (Table 3). In the second stage of the cross-validation analysis, the CE values of both modelMVPA and modelVPA were small except for individuals at extremes of fitness. The results of two cross-validation analyses provide evidence for supporting the validity of the prediction model used in present study. Our finding of a significant underestimation of V˙O2max among individuals with high fitness (Figs. 1 and 2) has been consistently observed in the previous studies (15,38). The present study drew on a smaller estimation bias (<3 mL·kg−1·min−1) compared with the study by Wier et al. (38) (<8 mL·kg−1·min−1). Wier et al. (38) pointed out that estimating V˙O2max for highly fit individuals is not a pressing problem for the typical work force because no negative consequences are seen because of high fitness. Furthermore, they suggested that the estimation bias can be corrected by modifying the intercept using the CE value.

Compared with the Plasqui and Westerterp (28) study and our previous study (3), the present study drew on a similar or larger sample and achieved a better prediction accuracy and model stability, as evidenced by the larger R2, smaller SEE values, absence of systematic bias, and minor shrinkage of the R2 and increment of SEE in the PRESS procedure. Various tests should be evaluated not only for their accuracy and validity but also for their applicability in a varied study population, their cost, and the ease and convenience of the protocol. The wide age range of the highly varied women who obtained measurements of V˙O2max in our study helps support the generalizability of the prediction model. In addition, because each of these predictor variables is easily obtained, it is believed that the nonexercise V˙O2max prediction model using SC and MVPA or VPA as a surrogate for the PA variable can be a routine component of primary health care examinations for women in large epidemiological cohorts.

This study has several limitations. First, the prediction model developed in this study may have limited generalizability because it was developed in a group of relatively healthy Japanese women 20 yr and older. The stability of the predicted V˙O2max values using the present model is unknown in groups of individuals whose characteristics vary substantially from the range of characteristics in our study samples (e.g., men, children and adolescents, individuals with metabolic syndrome, and other racial groups) because the relationship between objectively measured PA and V˙O2max in such groups may have different characteristics than that in our study. Further investigation is required to validate our prediction models in these groups. Second, accelerometers do not capture all types of PA, such as cycling or swimming, which may weaken the accuracy of our prediction model when our prediction models are applied in individuals who regularly exercise by riding a bike or swimming. Therefore, further study is needed to investigate that possibility.

To our knowledge, this study marks the first attempt to develop new nonexercise V˙O2max prediction models using accelerometer-determined PA intensity including the time spent in MVPA or VPA as the objective PA variables that can be used in large epidemiological cohorts. This study demonstrated that MVPA and VPA were useful in predicting V˙O2max variance and improved the ability of the regression models to predict V˙O2max accurately. The new nonexercise prediction equations derived in this study are applicable to estimating V˙O2max in Japanese women.

This research was supported by a Research Grant for Comprehensive Research on Cardiovascular and Life-Style Related Diseases from the Ministry of Health, Labour and Welfare, Japan (19160101).

The results of the present study do not constitute endorsement by ACSM.

The authors thank Dr Shigeho Tanaka (Health Promotion and Exercise Program, National Institute of Health and Nutrition) for helpful discussion of the activity monitors.

Disclosures: This research was supported by a Research Grant for Comprehensive Research on Cardiovascular and Life-Style Related Diseases from the Ministry of Health, Labour and Welfare, Japan (19160101).

REFERENCES

1. Barlow CE, LaMonte MJ, Fitzgerald SJ, Kampert JB, Perrin JL, Blair SN. Cardiorespiratory fitness is an independent predictor of hypertension incidence among initially normotensive healthy women. Am J Epidemiol. 2006;163(2):142-50.
2. Blair SN, Kannel WB, Kohl HW, Goodyear N, Wilson PW. Surrogate measures of physical activity and physical fitness. Evidence for sedentary traits of resting tachycardia, obesity, and low vital capacity. Am J Epidemiol. 1989;129(6):1145-56.
3. Cao ZB, Miyatake N, Higuchi M, Ishikawa-Takata K, Miyachi M, Tabata I. Prediction of VO2max with daily step counts for Japanese adult women. Eur J Appl Physiol. 2009;105(2):289-96.
4. Centers for Disease Control and Prevention, National Center for Health Statistics (2004) National Health and Nutrition Examination Survey 2003-2004. LABORATORY PROCEDURES MANUAL. Hyattsville, MD: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention. p. 1-14 Available from: http://www.cdc.gov/nchs/data/nhanes/nhanes_03_04/lab_pm.pdf. Cited 3 May 2008
5. Dencker M, Thorsson O, Karlsson MK, et al. Daily physical activity and its relation to aerobic fitness in children aged 8-11 years. Eur J Appl Physiol. 2006;96:587-92.
6. Dencker M, Thorsson O, Karlsson MK, Lindén C, Wollmer P, Andersen LB. Daily physical activity related to aerobic fitness and body fat in an urban sample of children. Scand J Med Sci Sports. 2008;18:728-35.
7. George JD, Stone WJ, Burkett LN. Non-exercise VO2max estimation for physically active college students. Med Sci Sports Exerc. 1997;29(3):415-23.
8. Gutin B, Yin Z, Humphries MC, Barbeau P. Relations of moderate and vigorous physical activity to fitness and fatness in adolescents. Am J Clin Nutr. 2005;81(4):746-50.
9. Hagströmer M, Oja P, Sjöström M. Physical activity and inactivity in an adult population assessed by accelerometry. Med Sci Sports Exerc. 2007;39(9):1502-8.
10. Hara T, Matsumura Y, Yamamoto M, et al. The relationship between body weight reduction and intensity of daily physical activities assessed with 3-dimension accelerometer. Jpn J Phys Fit Sports Med. 2006;55:385-91.
11. Hebestreit H, Kieser S, Rüdiger S, et al. Physical activity is independently related to aerobic capacity in cystic fibrosis. Eur Respir J. 2006;28(4):734-9.
12. Heil DP, Freedson PS, Ahlquist LE, Price J, Rippe JM. Nonexercise regression models to estimate peak oxygen consumption. Med Sci Sports Exerc. 1995;27(4):599-606.
13. Hooker SP, Sui X, Colabianchi N, et al. Cardiorespiratory fitness as a predictor of fatal and nonfatal stroke in asymptomatic women and men. Stroke. 2008;39(11):2950-7.
14. Holiday DB, Ballard JE, McKeown BC. PRESS-related statistics: regression tools for cross-validation and case diagnostics. Med Sci Sports Exerc. 1995;27(4):612-20.
15. Jackson AS, Blair SN, Mahar MT, Weir LT, Rossand RM, Stuteville JE. Prediction of functional aerobic capacity without exercise testing. Med Sci Sports Exerc. 1990;22(6):863-70.
16. Jurca R, Jackson AS, LaMonte MJ, et al. Assessing cardiorespiratory fitness without performing exercise testing. Am J Prev Med. 2005;29(3):185-93.
17. Katzmarzyk PT, Church TS, Blair SN. Cardiorespiratory fitness attenuates the effects of the metabolic syndrome on all-cause and cardiovascular disease mortality in men. Arch Intern Med. 2004;164(10):1092-7.
18. Kohl HW, Blair SN, Paffenbarger RS Jr, Macera CA, Kronenfeld JJ. A mail survey of physical activity habits as related to measured physical fitness. Am J Epidemiol. 1988;127(6):1228-39.
19. Kumahara H, Schutz Y, Ayabe M, et al. The use of uniaxial accelerometry for the assessment of physical-activity-related energy expenditure: a validation study against whole-body indirect calorimetry. Br J Nutr. 2004;91:235-43.
20. LaMonte MJ, Barlow CE, Jurca R, Kampert JB, Church TS, Blair SN. Cardiorespiratory fitness is inversely associated with the incidence of metabolic syndrome: a prospective study of men and women. Circulation. 2005;112(4):505-12.
21. Malek MH, Housh TJ, Berger DE, Coburn JW, Beck TW. A new non-exercise based VO2max equation for aerobically trained females. Med Sci Sports Exerc. 2004;36(10):1804-10.
22. Malek MH, Housh TJ, Berger DE, Coburn JW, Beck TW. A new non-exercise-based Vo2max prediction equation for aerobically trained men. J Strength Cond Res. 2005; 19(3):559-65.
23. McArdle WD, Katch FI, Katch VL. Exercise Physiology: Energy, Nutrition and Human Performance. (5th edition) Lippincott Williams and Wilkins; 2001. pp. 222-48.
24. McClain JJ, Sisson SB, Washington TL, Craig CL, Tudor-Locke C. Comparison of Kenz Lifecorder EX and ActiGraph accelerometers in 10-yr-old children. Med Sci Sports Exerc. 2007;39(4):630-8.
25. Metzger JS, Catellier DJ, Evenson KR, Treuth MS, Rosamond WD, Siega-Riz AM. Patterns of objectively measured physical activity in the United States. Med Sci Sports Exerc. 2008;40(4):630-8.
26. Ministry of Health Labour and Welfare of Japan. Exercise and Physical Activity Reference for Health Promotion 2006. National Institute of Health and Nutrition Web site [Internet]. April 2007. p. 9-10. [cited 2008 Mar 3]. Available from: http://www.nih.go.jp/eiken/programs/pdf/epar2006.pdf
27. Plasqui G, Westerterp KR. Accelerometry and heart rate as a measure of physical fitness: proof of concept. Med Sci Sports Exerc. 2005;37(5):872-6.
28. Plasqui G, Westerterp KR. Accelerometry and heart rate as a measure of physical fitness: cross-validation. Med Sci Sports Exerc. 2006;38(8):1510-4.
29. Rankin SL, Briffa TG, Morton AR, Hung J. A specific activity questionnaire to measure the functional capacity of cardiac patients. Am J Cardiol. 1996;77(14):1220-3.
30. Sanada K, Midorikawa T, Yasuda T, Kearns CF, Abe T. Development of nonexercise prediction models of maximal oxygen uptake in healthy Japanese young men. Eur J Appl Physiol. 2007;99(2):143-8.
31. Sallis JF, Saelens BE. Assessment of physical activity by self-report: status, limitations, and future directions. Res Q Exerc Sport. 2000;71(suppl 2):S1-14.
32. Sawada SS, Lee IM, Muto T, Matuszaki K, Blair SN. Cardiorespiratory fitness and the incidence of type 2 diabetes: prospective study of Japanese men. Diabetes Care. 2003;26(10):2918-22.
33. Sawada SS, Muto T, Tanaka H, et al. Cardiorespiratory fitness and cancer mortality in Japanese men: a prospective study. Med Sci Sports Exerc. 2003;35(9):1546-50.
34. Siconolfi SF, Lasater TM, Snow RC, Carleton RA. Self-reported physical activity compared with maximal oxygen uptake. Am J Epidemiol. 1985;122(1):101-5.
35. Sjöström M, Poortvliet E, Nelson M. Monitoring public health nutrition in Europe: nutritional indicators and determinants of health status. J Public Health. 2005;13:74-83.
36. Tabata I, Nishimura K, Kouzaki M, et al. Effects of moderate intensity-endurance and high intensity-intermittent training on anaerobic capacity and VO2max. Med Sci Sports Exerc. 1996;28(10):1327-30.
37. Whaley MH, Kaminsky LA, Dwyer GB, Getchell LH. Failure of predicted VO2peak to discriminate physical fitness in epidemiological studies. Med Sci Sports Exerc. 1995;27(1):85-91.
38. Wier LT, Jackson AS, Ayers GW, Arenare B. Nonexercise models for estimating VO2max with waist girth, percent fat, or BMI. Med Sci Sports Exerc. 2006;38(3):555-61.
39. Yamada Y, Yokoyama K, Noriyasu R, et al. Light-intensity activities are important for estimating physical activity energy expenditure using uniaxial and triaxial accelerometers. Eur J Appl Physiol. 2009;105(1):141-52.
Keywords:

CARDIORESPIRATORY FITNESS; MAXIMAL OXYGEN UPTAKE; ACCELEROMETER; PREDICTION MODELS; INTENSITY

©2010The American College of Sports Medicine