Aerobic Capacity Reference Data in 3816 Healthy Men and Women 20–90 Years

Purpose To provide a large reference material on aerobic fitness and exercise physiology data in a healthy population of Norwegian men and women aged 20–90 years. Methods Maximal and sub maximal levels of VO2, heart rate, oxygen pulse, and rating of perceived exertion (Borg scale: 6–20) were measured in 1929 men and 1881 women during treadmill running. Results The highest VO2max and maximal heart rate among men and women were observed in the youngest age group (20–29 years) and was 54.4±8.4 mL·kg−1·min−1 and 43.0±7.7 mL·kg−1·min−1 (sex differences, p<0.001) and 196±10 beats·min−1 and 194±9 beats·min−1 (sex differences, p<0.05), respectively, with a subsequent reduction of approximately 3.5 mL·kg−1·min−1 and 6 beats·min−1 per decade. The highest oxygen pulses were observed in the 3 youngest age groups (20–29 years, 30–39 years, 40–49 years) among men and women; 22.3 mL·beat−1±3.6 and 14.7 mL·beat−1±2.7 (sex differences, p<0.001), respectively, with no significant difference between these age groups. After the age of 50 we observed an 8% reduction per decade among both sexes. Borg scores appear to give a good estimate of the relative exercise intensity, although observing a slightly different relationship than reported in previous reference material from small populations. Conclusion This is the largest European reference material of objectively measured parameters of aerobic fitness and exercise-physiology in healthy men and women aged 20–90 years, forming the basis for an easily accessible, valid and understandable tool for improved training prescription in healthy men and women.


Introduction
Evidence supports a strong inverse association between cardiorespiratory fitness and all-cause mortality [1][2][3][4]. Therefore, in order to increase the individual's fitness level, different types of exercise training are used both in prevention and treatment of cardiovascular and life-style related disease [5]. In order to prescribe a proper prevention-or treatment program either, the more reliable, individually cardiopulmonary exercise testing is needed [6,7], or one can rely upon previously established reference material. Most previous studies measuring cardiorespiratory fitness tend to use peak oxygen uptake (VO 2peak ), indirect methods, estimation by equation, selected populations and/or small sample sizes [8][9][10][11][12][13][14][15]. Previously established variables suitable for exercise prescription such as oxygen uptake, heart rate, Watts and Borg scale scores, are mostly based upon small sample sizes or poorly described populations [6,[16][17][18][19][20][21][22][23]. The aim of this study was to establish a large reference material of empirical cardiorespiratory fitness and exercise-physiology data in healthy men and women aged 20-90 years in order to provide an easily accessible, valid and understandable tool for improved exercise training prescription.

Participants
The HUNT 3 fitness study is the third wave of the Nord-Trøndelag Health Studies (www.ntnu.edu/hunt). Data were collected between October 2006 and June 2008. The entire population .20 years of age (n = 94194) were invited to participate, 54% (n = 50821) accepted. The HUNT 3 Fitness Study was designed to acquire reference material for submaximal and maximal cardiorespiratory parameters in a healthy population. Exclusion criteria were cancer, cardiovascular disease, obstructive lung disease and use of blood pressure medication. Participants also had to be cleared in a brief medical interview. Based upon self-reported information, 30513 candidates presented as suitable for VO 2max testing. Out of these, 12609 candidates resided in the 3 municipals selected for VO 2max testing, and 5633 of them volunteered to participate. These 3 locations were chosen due to geographical location to minimize travel distance for participants. 4621 candidates completed a VO 2max test, whereas 3816 tests were considered to have reached the true VO 2max .

Ethics Statement
The study was approved by the Regional committee for medical research ethics (2012/1228/REK midt), the Norwegian Data Inspectorate and the National Directorate of Health, and is in compliance with the Helsinki declaration. Written informed consent was obtained from all participants.

VO 2max and Heart Rate
An individualized graded protocol [24] was used for measuring VO 2max (Cortex MetaMax II, Cortex, Leipzig, Germany). Prior of starting the testing procedure several MetaMax II apparatus were tested against Douglas-bag and iron lung (Cortex, Leipzig, Germany). Two MetaMax II apparatus were returned to Cortex due to unstable recordings (both ventilation and carbon dioxide analysis) and replaced by two new apparatus that were tested and found both reliable and valid. Hence, all MetaMax apparatus used in the project were both valid and reliable. Test-retest correlation of oxygen-uptake for the tested-personnel in the project was 0.99, p,0.001 and coefficient of variation was 1.8%. Bland-Altman plot were constructed where differences in two tests of maximal oxygen uptake (and sub maximal) from each person (test-1 minus test-2) were plotted against the average of the two tests ((test-1+test-2)/2). Average difference was 20.04 ml?kg 21 ?min 21 and standard deviation of the difference was 1.0. Therefore one can expect that 95% of all observations are at the average ((test-1+test-2)/2) 62 standard deviations. Thus, we can expect values to vary between 20.04 22?1 to 20.04+2?1 if we test maximal oxygen uptake twice in the same person within a short time period. Velocity and inclination of the test treadmills were calibrated prior to testing.
The MetaMax II was calibrated prior to the first test each day using a standard two-point gas calibration procedure recommended by the manufacturer. The calibration includes measurements of ambient air and a gas mix of known content (15.03% O 2 and 4.98% CO 2 in N 2 ), a calibration of the Triple-V volume transducer with a calibrated 3 L syringe (Calibration syringe D, Sensormedics, CareFusion, San Diego, CA, USA), and barometric pressure control. Volume calibration was implemented every third test and the two-point gas calibration every fifth. Before each test the ambient room air was routinely checked. Heart rate was measured by radio telemetry (Polar S610i, Polar Electro Oy, Kempele, Finland). Body mass was measured using the weighing scale Model DS-102 (Arctic Heating AS, Nøtterøy, Norway). All participants had a treadmill familiarization period of 8-10 minutes during warm-up. They were instructed to avoid grabbing handrails if not necessary. The individualized warm-up workload determined the initial velocity/inclination on the subsequent treadmill test. Candidates wore a face mask (Hans Rudolph, Germany) linked to the MetaMax II. When participant maintained a stable oxygen uptake .30seconds, velocity (0.5-1.0 kmh-1) or inclination (1-2%) were increased. Increased workload was preferably obtained with increased velocity and keeping a fixed slope angle. If a participant was unable to increase velocity, inclination was increased. The average velocity and slope during test protocol were 6.862.2 km?h 21 (range 2-17 km?h 21 ) and 10.061.6% (range 2-16%), respectively. Tests were terminated when candidates were exhausted or reached a VO 2 plateau that remained stable despite increased work load [25], i.e. VO 2 did not increase more than 2 mL?kg 21 ?min 21 despite increased work load.

Ventilatory Equivalent
We calculated the ventilatory equivalent (V E ?VO 2

21
) at VO 2max . The ventilatory equivalent describes the fraction of minute ventilation (V E ) to oxygen uptake (VO 2 ), hence, the higher the value the more ineffective is the V E . At higher levels of increasingly harder submaximal workloads, a disproportionate increase in V E relative to VO 2 occurs. This heralds an increasingly more inefficient V E .

Questionnaire-based Information
Physical activity index score (PAI) was calculated from replies in a self-administered questionnaire that consisted of 3 questions. Question 1: ''How frequently do you exercise?'' with response alternatives ''Never'' (0); ''Less than once a week'' (0); ''Once a week'' (1); ''2-3 times a week'' (2.5); ''Almost every day'' (5). Question 2: ''If you exercise as frequently as once or more times a week: How hard do you push yourself?'' with response alternatives ''I take it easy without breaking a sweat or losing my breath'' (1); ''I push myself so hard that I lose my breath and break into sweat'' (2); ''I push myself near exhaustion'' (3). Question 3: ''How long does each session last?'' with the following response options ''Less than 15 minutes'' (0.1); ''15-29 minutes'' (0.38); ''30 minutes to 1 hour'' (0.75); ''More than 1 hour'' (1.0). The numbers in brackets, corresponding to each subject's response to the 3 questions above, were multiplied to calculate the physical activity index score. An index score in the range 0.05-1.50 was considered to signify low activity, an index score in the range 1.51-3.75 was interpreted as medium activity and a score in the range 3.76-15.0 signified high activity. The index score is previously established as valid and reliable [26].

Borg Scale of Perceived Exertion and VO 2 at 2 Sub Maximal and Maximal Workload
Candidates were asked to state their subjective rating of perceived exertion (Borg scale) at the end of 3 different levels. Borg scale visualizes work load intensity, denoted by numbers 6-20 [22], with a proportional relation between increased rating of perceived exertion and the reported Borg scale number. Level 1: The individual initial workload for the test was determined during warm-up. All individuals reached a stable VO 2 and heart rate after 3 minutes at the first submaximal work load. Level 2: Treadmill gradient was elevated by 2% or velocity increased 1 km?h 21 . After 1-2 minutes at this sub maximal workload steady state was obtained. The maximum level is described previously.

Watts
Workload in watts was calculated at the 3 described workloads. Calculations were based on treadmill slope gradient, velocity and body mass input in Cortex MetaMax II (Cortex, Leipzig, Germany). Minimum slope gradient was 2% and mean slope gradient at maximum workload was 10.461.4%.

Statistical Analysis
Parametric analysis was used based on the large sample size. QQ-plots supported the assumption of normally distributed data. Descriptive data are presented as arithmetic mean 6 standard deviation. An Independent-Samples T test was used for establishing level of significance between sexes and age groups. Linear regression, with 95% confidence interval, was used to illustrate associations between physiological parameters. All statistical tests were two-sided. SPSS 16.0 (Statistical package for Social Sciences, Chicago; Illinois, USA) and GraphPad Prism 4.01 (GraphPad Software, San Diego, California, USA) were used to analyze data. Correlations were done using data from Level 1, Level 2 and Level 3 (maximum) as described above. A p-value of ,0.05 was considered statistically significant.  21 ). Maximal workload at VO 2max was 33% lower (p,0.001) in women compared with men (121624 W vs. 181636 W). There were no significant sex differences in maximal heart frequencies or Borg scores at termination of the test ( Table 1).

Overall
The highest VO 2max and maximal heart rate for both sexes were observed in the age group 20-29 years. Among men and women in this age group VO 2max were 54.468.4 mL?kg 21 ?min 21 and 43.067.7 mL?kg 21 ?min 21 (sex differences, p,0.001) with corresponding heart rates of 196610 beats?min 21 and 19469 beats?min 21 (sex differences, p,0.05). In both sexes VO 2max and maximal heart rates were approximately 8% (<3.5 mL?kg 21 ?min 21 ) and 3.5% (6 beats?min 21 ) lower per increased decade, respectively. The Physical Activity Index (PAI) scores were also highest for both sexes in this age group. For men and women PAI scores were 4.6464.04 and 3.9663.25, respectively (no significant sex differences), which are considered to indicate a high physical activity level. All other age groups, regardless of sex, had PAI scores in the range 2.67-3.70, which are considered to indicate medium activity [26] (Table 2).
We observed an EqVO 2max of 33.964.0 and 34.165.3 among men and women, aged 20-29 years, respectively. Generally no differences were found between sexes and age groups, except from a 3% (p,0.05) higher EqVO 2max in females aged 30-39 and 40-49 years compared to corresponding groups of males. Additionally, EqVO 2max increased by 3% (p,0.05) between the two most senior male groups ( Table 2).
The highest maximal oxygen pulse was observed in the 3 youngest age groups (20-49 years) for both sexes, with no significant difference between these age groups. Women in these age groups had 33% lower (p,0.001) oxygen pulse compared with men (14.762.7 mL?beat 21 vs. 22.363.6 mL?beat 21 ). In the subsequent age groups an approximately 8% reduction in oxygen pulse per decade was observed among both sexes ( Table 3).

Rating of Perceived Exertion, %VO 2max and %maximal Heart Rate
As can be seen from Table 4, rating of perceived exertion reported as Borg Score fairly well estimate the relative exercise intensity expressed as percent of maximal heart rate and percent of VO 2max . Data also show that there may be sex differences in these relationships in the lowest intensities corresponding to Borg below 16. For example the actual exercise intensity for men and women that report to exercise at Borg 6-9 corresponds to 75.8% (CI: 74.5-77.1) and 79.1% (CI: 77.6-80.5) of maximal heart rate, respectively (sex differences, p,0.05). Furthermore, Borg 13-15 corresponds to a heart rate of 84.7% (CI: 84.3-85.1) for men and 88.4% (CI: 88.0-88.7) for women (sex differences, p,0.001). The same discrepancies apply to %VO 2max in corresponding Borg range. At Borg Score above 16 there were no sex or age group differences, with the exception of an age group difference between the 50-59 years and the 60-69 years group in both percent of maximal heart rate (p,0.01) and percent of VO 2max (p,0.05). VO 2 , Heart Rate, Watt and Physical Activity Index Figure 1 displays the relationship between VO 2 and heart rate. The overall correlation for VO 2 (L?min 21 ) and heart rate was found to be moderate (r = 0.51, p,0.0001). Stratified by sex, this association became stronger; males r = 0.70 (p,0.0001), female r = 0.61 (p,0.0001). The association between percent VO 2 and percent maximal heart rate was strong; all r = 0.89 (p,0.0001), male r = 0.90 (p,0.0001), female r = 0.87 (p,0.0001). Figure 2 demonstrates the strong association between VO 2 (L?min 21 ) and treadmill workload (Watts), and the correlation between watts and heart rate; VO 2 vs. Watts: all r = 0.90 (p,0.0001), male r = 0.89 (p,0.0001), female r = 0.84 (p,0.0001). Heart rate vs. Watts: Viewing the full sample size a moderate correlation was observed between treadmill workload (Watts) and heart rate (r = 0.55, p,0.0001). Good correlations were observed when the data was stratified by sex; male r = 0.71 (p,0.0001), female r = 0.66 (p,0.0001). Figure 3 demonstrates a poor, but statistically strong association between Physical activity index and VO 2 (mL?kg 21 ?min 21 ); all r = 0.24 (p,0.0001), male r = 0.29 (p,0.0001) and female r = 0.27 (p,0.0001). Figure 4 exhibits a moderate association between VO 2 (both mL?kg 21 ?min 21 and L?min 21 ) and age groups, male: r = 0.54 and r = 0.54, respectively; female: r = 0.52 and r = 0.50, respectively. Discussion This is the largest European reference material of objectively measured aerobic capacity and exercise-physiology in healthy men and women aged 20-90 yrs. Our observations are an important supplement to previously published data that have mostly been either indirect or based on small, selected, or poorly-described populations [8][9][10][11][12][13][14].

Sex Differences in VO 2max and Maximal Heart Frequencies
Despite being generally more physical active than men, women had a 34% and 18.5% lower absolute (L?min 21 ) and relative (mL?kg 21 ?min 21 ) VO 2max , respectively, than men. When applying   [27] where differences in body mass are taken into consideration for a more accurate comparison [27][28][29], women had 22.7% lower VO 2max (mL?kg 20.75 ?min 21 ) than men. The higher VO 2max in men is in accordance with former studies [5,[10][11][12][13][14]30]. We observed higher VO 2max compared to that reported in American [9,11,13,14], Japanese populations [8] and a large Brazilian study [31], especially in the younger age groups. Earlier Scandinavian research supports our VO 2max levels [25,32,33]. A study of Nomadic Lapps [34], who were physically active in taking care of their reindeers, observed VO 2max values close to our findings, and it is indicated that Roman legionnaires [35] had a VO 2max in the range of 50 mL?kg 21 ?min 21 . Also contemporary hunter-gatherer societies display VO 2max in the range 50-65 mL?kg 21 ?min 21 in young male populations [36]. This supports the assumption that by living an active life, VO 2max in the range we display could be expected. When applying a scaled VO 2max [27], where differences in body weight is considered for a more precise comparison [27][28][29], the HUNT 3 fitness study still displays a considerable higher VO 2max (mL?kg 20.75 ?min 21 ) than North American, German and Asian studies. Scaled VO 2max in HUNT 3 was approximate 20% higher, considering both sexes and all age groups, than North American [9,10,37] and a German study [38]. However, a North American study by Jackson and colleagues [14] display only an estimated 10% lower scaled VO 2max , compared to our findings, among both sexes in the 20-29 year age group, with diminishing differences per subsequent decade. A Japanese study [8] displays an average 14% lower scaled VO 2max than us, among both sexes and all age groups. The dissimilarities between our and other findings may be explained by that we measured VO 2max whereas most others [8,9,11,14] use VO 2peak or estimated VO 2max [13,15], and that most other populations might also lead a more sedate lifestyle than that of Scandinavians. Despite our relatively high mean VO 2max , 25 men and 9 women (.50 years) displayed values below 8 METs (28 mL?kg 21 ?min 21 ) and 6 METs (21 mL?kg 21 ?min 21 ), respectively. This is associated with higher all-cause mortality and cardiovascular events among healthy men and women [4]. 1% of men 50-59 years displayed METs associated with increased risk, with 4% and 11% prevalence of ''increased risk MET'' with each subsequent decade. Women displayed approximately half the prevalence over the same age groups. We report a somewhat higher maximal heart rate than previous studies, which could be explained by that others [10][11][12] measure peak heart rate. In line with former studies [10][11][12] there were no significant sex differences in maximal heart rate.

Differences in VO 2max and Physical Activity Level Stratified by Age Group and Sex
The highest VO 2max , in both men and women, were observed in the age groups 20-29 years. This fits with that both sexes in this age group had the highest level of physical activity compared to all other age groups. Between the two youngest age groups of both men and women (20-29 years and the 30-39 years) there was no difference in absolute VO 2max (L?min 21 ). However, among the 30-39 years old, body mass was 8.4% and 6.4% higher, and physical activity level 32.1% and 15.2% lower for men and women, respectively. A lower physical activity level did not influence the absolute VO 2max (L?min 21 ) but probably contribute to the higher body mass in those aged 30-39 years. Relative to body mass VO 2max (mL?kg 21 ?min 21 ) was 10% and a 7% lower among men and women aged 30-39 years compared to those in the 20-29 year age group. Thus, lower relative VO 2max in those aged 30-39 years old was caused by a higher body mass in our study population.
Despite similar body weights and physical activity level, a highly significant lower absolute (4.4%) and relative (3.8%) VO 2max was observed among men and women aged 40-49 years compared to those aged 30-39 years. In line with our findings, Sanada and colleagues [8] observed a similar drop in VO 2max between these age groups among healthy Japanese men. However, they observed a considerably larger ''drop'' in VO 2max with no change in body mass among women compared to our observations. Their findings are in agreement with another study of women by Jackson and colleagues [9]. A likely explanation for the lower VO 2max in those aged 40-49 vs. 30-39 could be a reduced level of physical activity in the oldest age group. However, our data does not support this. The Sanada [8] study shows a decrease in skeletal muscle mass with simultaneous increase in percent body fat. This would deteriorate demand properties and decrease VO 2max . We do not know if this is the case in our study, but the drop in absolute VO 2max yields a reduction in supply properties, hence a reduction in VO 2max . The reason for different ''drop'' in VO 2max among women in our study, the Sanada [8] and Jackson [9] studies is not known and warrant further studies.
Over the 3 next decades (40-69 years) the decrease in absolute and relative VO 2max had more than doubled (<10% per decade), Table 4. Relationships between perceived exertion, VO 2max and f cmax in the HUNT 3 fitness study. in both sexes, compared to that observed between age groups 30-39 years and 40-49 years. There were no significant differences in body mass between these age groups, with the exception of a <2% reduction in males between 50-59 and 60-69 years. The reduction in VO 2max per decade is in line with previous studies [8,11,14]. A significant reduction in physical activity level with increased age among women was observed in our study, with no changes in the male group. Thus, reduced physical activity level may explain reduced VO 2max with increased age among women but not among men. Whether this is really an age-related decline or due to that men have a tendency to over-report physical activity [39] is not known, and future investigations should aim to obtain objectively measured physical activity levels.
Between the 60-69 years and the +70 year age groups we observed the largest ''drop'' in absolute VO 2max (L?min 21 ) in both sexes (<14.5%). Relatively VO 2max (mL?kg 21 ?min 21 ) were 10% and 11% lower in males and females aged +70 compared to those aged 60-69 years. The large drop in relative VO 2max occurred despite no significant drops in body mass and physical activity level. This suggests that the drop may be due to ''age-related'' adaptations in the organism. Our findings for these age groups are in line with previous studies reporting reductions in relative and absolute VO 2max in the range 13-25% and 12-29% [8,10,12,40].
Ventilatory efficiency, EqVO 2max , remains generally unchanged throughout the age groups; hence, it is not a factor in explaining the diminishing VO 2max with increasing age. However, subsequent 50 years of age O 2 -pulse displayed a steady decrease, which indirectly indicate a reducing stroke volume, hence this yield a reduction in VO 2max .

Differences in Maximum Heart Rate Stratified by Age Group and Sex
Maximum heart rate has regularly been estimated by an equation subtracting age from 220 beats?min 21 , which have limited scientific merit [41]. The highest heart frequencies were found in the youngest age groups, regardless of sex. Maximal heart frequencies in men and women were 196610 beats?min 21 and 19469 beats?min 21 , respectively, with a decline of approximately 3.5% (6 beats?min 21 ) per decade. Our maximal heart rate decline  . Decline of oxygen uptake (VO 2 ) relative to age. Mean and SD for each age group is presented in Table 2. doi:10.1371/journal.pone.0064319.g004 gradient is approximately half that compared to using the 220 beats?min 21 minus age equation, which is consistent with former research [10,42,43].

Differences in Oxygen Pulse Stratified by Gender and Age Groups
The highest maximal oxygen pulse was observed in the 3 youngest age groups among men (22.363.6 mL?beat 21 ) and women (14.762.7 mL?beat 21 ), with no significant difference between these age groups. Previous reference material on oxygen pulse in healthy populations is based upon case reports [7] or small studies in athletes [44] [45] making comparison with our data complicated.

Association between VO 2 and Heart Rate
The benchmark studies of the relationship between submaximal VO 2 and heart frequencies from the 509s and 609s have small sample size. When comparing our findings with that of Å strand [6] there are obvious discrepancies. Compared with our observations, Å strand reports lower heart frequencies at workloads corresponding to VO 2 lower than 3 L?min 21 , and higher heart frequencies at VO 2 higher than 3 L?min 21 . Since there are no references to sample size or gender in Å strands data it is difficult to interpret the discrepancies between the studies. Another study by Å strand [17] with 86 relatively well trained male and female students, agrees with our data, for the age group 20-29 years, on VO 2 less than 3 L?min 21 . When VO 2 exceed 3.0 L?min 21 for females and 4.0 L?min 21 for males the present study displays lower heart frequencies than reported in the Å strand study. A study [18] with 44 females, aged 20-65 years, displayed lower heart frequencies than we observed for VO 2 ,1.5 L?min 21 , but higher heart frequencies for VO 2 .1.5 L?min 21 . The observed association between percent VO 2max and percent maximal heart rate is in agreement with previous studies [16,18,19] on intensities .70% of VO 2max . At exercise intensities ,70% of VO 2max we observed higher percent heart rate than previous studies. In our study 30% of VO 2max corresponded to 60% of maximal heart rate where others [16,18,19] have reported that this is equal to 50% of maximal heart rate. However, we use treadmill testing while the others use bicycle ergometer, which could explain the discrepancies.

Association between Heart Rate and Watts
Comparing our results to a study by Å strand [20] with 84 healthy males, good agreement was found for workloads .150W, whereas we observed progressively higher heart frequencies than Å strand at lower workloads. This is also the case when comparing our results to another study of males by Å strand [6]. Again, the inconsistency between our and Å strand's findings could be explain by treadmill vs. bicycle ergometer.

Association between VO 2 and Watts
We systematically display higher VO 2 (L?min 21 ) at any given watt than that observed in two studies from Å strand [17,21]. Discrepancies seem to be caused by higher initial VO 2 cost (L?min 21 ) in our results, while the slope gradient is in close proximity with Å strand [17,21]. The differences may be explained by that we applied treadmill work whereas Å strand used a cycle ergometer.

Association between Borg Scale Scores, % VO 2max and % Maximal Heart rRate
We observed a mismatch between the Borg study [22] and our findings. We observed considerably higher percent VO 2max and percent maximal heart rate for a given exertion interval than Borg [22], but differences vanished at the highest Borg-levels.
Relative to VO 2max and maximal heart rate Borg scale differed between sexes. In the range 6-15 on Borg scale males worked at a lower percent (4%) of both VO 2max and maximal heart rate than females, i.e. the relative rating of perceived exertion in males were higher at a given work load. There were no differences between sexes at Borg16-20. Our data clearly support the notion that Borgscale may be used as a robust tool to guide exercise intensity in healthy men and women, but that one should be aware of sex differences at the lowest Borg levels.

Association between Physical Activity Index Scores and VO 2max
There was a poor overall correlation (r = 0.24) between selfreported physical activity level and VO 2 , which indicates that only 5.7% of differences in VO 2max can be explained by the physical activity scores. This is in agreement with prior research [11,46,47]. Physical activity index score (PAI) is a weighted product between duration, frequency and intensity. Intensity might be weighted to low, thus it could explain the poor correlation between PAI and VO 2 .

Limitations
This study may be subject to bias due to self-selection caused by the low participation rate. However, almost all of those who were invited to the current Fitness study from the large HUNT study agreed to participate in the fitness test. Due to limited capacity at the test sites resulting in long waiting lines, many potential participants chose to withdraw their participation from the study. Those who finally participated in the study could thus be healthier than those who quit or declined participation. However, a comparison of the participants in the fitness study with a healthy sample of the total HUNT population (i.e. free from cardiovascular or pulmonary diseases, cancer, or sarcoidosis) confirmed that the fitness participants did not considerably differ from other healthy HUNT participants [48]. In future studies physical activity should be measured objectively rather than being self-reported, given the large inconsistencies between VO 2max and self-reported physical activity.

Conclusions
The discrepancies between this and previous studies highlighted the need of a large reference material as presented in this study. The HUNT 3 Fitness study presents the largest Europen reference material of objectively measured parameters of aerobic capacity and exercise-physiology in healthy men and women aged 20-90 years. Our data establishes normal values for the key physiological factors VO 2max and heart rate, as well as associations between commonly used exercise parameters. The data forms the basis for a user-friendly tool for exercise intensity control in healthy men and women.

Author Contributions
Conceived and designed the experiments: UW. Performed the experiments: UW OR. Analyzed the data: HL OR BS UW. Contributed reagents/materials/analysis tools: HL OR BS UW. Wrote the paper: HL OR BS UW. Conception and design of the work, acquisition of data, or analysis and interpretation of data: HL OR BS UW. Drafting the article or revising it critically for important intellectual content: HL OR BS UW. Final approval of the version to be published: HL OR BS UW.