Cardio-Respiratory Reference Data in 4631 Healthy Men and Women 20-90 Years: The HUNT 3 Fitness Study

Purpose To provide a large reference material on key cardio-respiratory variables in a healthy population of Norwegian men and women aged 20–90 years. Methods Sub maximal and peak levels of cardio-respiratory variables were measured using cardiopulmonary exercise testing during treadmill running. Results The highest peak ventilation among men (141.9±24.5 L·min−1) and women (92.0±16.5 L·min−1) was observed in the youngest age group (20–29 years, sex differences p<0.001) with an average 7% reduction per decade. The highest tidal volumes were observed in the 30–39 and 40–49 year age groups among men (2.94±0.46 L) and women (2.06±0.32 L) (sex differences p<0.001), with a subsequent average 6% reduction per decade. Ventilatory threshold and respiratory compensation point were observed at approximately 77% and 87% of peak oxygen uptake (VO2peak) among men and women, respectively. The best ventilatory efficiency (EqVCO2Than) was observed in the youngest age group (20–29 years) in both men (26.2±2.8) and woman (27.5±2.7) (sex differences p<0.001) with an average 3% deterioration in ventilatory efficiency per decade. Conclusion This is the largest European reference material of cardio-respiratory variables in healthy men and women aged 20–90 years, establishing normal values for, and associations between key cardio-respiratory parameters. This will be useful in clinical decision making when evaluating cardiopulmonary health in similar populations.


Participants
The HUNT 3 fitness study is the third wave of the Nord-Trøndelag Health Studies (ntnu.edu/hunt). Data were collected between October 2006 and June 2008. The entire population .20 years of age (n594194) were invited to participate, 54% (n550821) accepted. Eligible candidates had to be free from cardiovascular disease, respiratory symptoms, cancer, and the use of blood pressure medication. 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. Subsequent the primary inclusion the medical interview excluded an additional 390 candidates not meeting medical inclusion criteria, leaving 5243 candidates. 4631candidates completed a VO 2peak test. These 3 locations were chosen due to geographical location to minimize travel distance for participants. We experienced technical difficulties with Cortex MetaMax P, and during service at Cortex data was lost, thus, total sample sizes on tidal volume (V T ) and breathing frequency (f B ) was n53667.

Ethics statement
The study was approved by the Regional committee for medical research ethics (2012/1672/REK nord), 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.

Cardio Pulmonary Exercise Test (CPET)
An individualized graded protocol [26] was used for measuring cardio-respiratory variables (Cortex MetaMax P, Cortex, Leipzig, Germany). Before starting the testing procedure several MetaMax II apparatus were tested against Douglas-bag and iron lung (Cortex, Leipzig, Germany) and those finally used found reliable and valid [27]. Speed and angle 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. The calibration included 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 performed every third test and the two-point gas calibration every fifth. Before each test the ambient room air was 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). Participants had a treadmill familiarization phase of 8-10 minutes during warm-up. They were instructed to avoid grabbing the handrails if not absolutely necessary. The individualized warm-up workload determined the initial speed/angle on the subsequent treadmill test. Candidates used a face mask (Hans Rudolph, Germany) of appropriate size linked to the MetaMax P. When participant maintained a stable oxygen uptake .30seconds, velocity (0.5-1.0 kmh-1) or inclination (1-2%) was increased. Increased workload was if possible obtained with increased speed and keeping a fixed slope angle of the treadmill. If a participant was unable to increase speed, the angle was increased. Tests were ended when candidates reached volitional exhaustion (e.g. shortness of breath and leg fatigue). VO 2max was considered achieved if subjects reached a VO 2 plateau that remained stable even with increased work load [28], i.e. VO 2 did not increase more than 2 mL?kg 21 ?min 21 despite increased work load, and R>1.05. Since 12.6% of the subjects failed to reach VO 2max we used the expression VO 2peak . Measurements were done at 3 different workloads, 2 submaximal and peak. Level 1: The individual initial workload was determined during warm-up, and stable VO 2 and heart rate were reached after 3 minutes. Level 2: Treadmill gradient was increased by 2% or speed increased 1 kmNh 21 , with steady state obtained after 2-3 minutes. Peak workload is described above.
Ventilatory anaerobic threshold (V Than ) and respiratory compensation point (RCP) relative to VCO 2 [31], hence, onset of hyperventilation [32]. Both V THan and RCP were established by the V-slope method [31].

Ventilatory efficiency
We calculated the ventilatory equivalent EqVO 2 (V E ?VO 2 21 ) and EqVCO 2 (V E ?VCO 2 21 ) at VO 2peak and V Than . The ventilatory equivalents describe the fraction of minute ventilation (V E ) to oxygen uptake (VO 2 ), or to expired carbon dioxide (VCO 2 ).

Statistical analysis
Parametric analysis was used and QQ-plots supported the assumption of normally distributed data. Data are presented as arithmetic mean ¡ standard deviation. Analysis of variance (Anova) was used to determine differences between age groups. If a significant F-ratio was achieved, post hoc evaluations were completed using Bonferroni tests. An Independent-Samples T-test was used for establishing level of significance between sexes. Linear regression and curve linear regression, with 95% confidence interval, were used to test associations between cardiorespiratory parameters. Multiple linear regressions were used to generate prediction models. All statistical tests were two-sided. SPSS 20.0 (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 peak as described above. A p-value of ,0.05 was considered statistically significant.

Results
Descriptive characteristics for men and women are presented in Table 1.
Women had approximately 32% (p,0.001) lower V Tpeak than men. For both sexes the highest V Tpeak was found among those aged 30-49 years, despite no differences in stature compared to the youngest age groups. In both sexes we observed an average 4% (p,0.05) and 11% (p,0.001) drop in V Tpeak per decade in age groups 40-69 years, and between the 2 most senior age groups, respectively ( Table 2). The highest breathing frequency (f B ) in both men and women, 50¡9 breaths?min 21 and 47¡7 breaths?min 21 , respectively, was found in the youngest age group (20-29 years), with an average 5% (p,0.05) decrease per subsequent decade ( Table 2).
Stratified by height an 11% (p,0.01) rise in V Epeak and V Tpeak was observed per 10 cm increased height, in both sexes (Table 3).

Carbon dioxide (VCO 2 ) elimination
Women displayed roughly 34% (p,0.001) lower VCO 2peak than men. Stratified by age the highest VCO 2peak was found in the youngest age groups. No significant differences in VCO 2peak were observed neither for men nor women between age groups 20-29 and 30-39 years, whereupon an approximate 6% (p,0.001) and 5% (p,0.001) decrease was observed between age groups 30-39 vs. 40-49 years, respectively. In subsequent age groups exponential reductions were observed, ending with an average 18% (p,0.001) lower peak VCO 2 in the most senior age group compared with men and women aged 60-69 years ( Table 2).

Ventilatory anaerobic threshold (V Than )
The highest V Than was observed in the youngest age groups. No statistical differences in V Than was observed between the 3 youngest age groups (20-49 years) among both sexes, whereupon we observed an approximate 10% (p,0.001) lower V Than per decade. V Than was obtained at 75.2¡10.7 and 76.7¡9.4% of VO 2peak for men and women (20-29 years), respectively, which corresponds to approximately 88¡7% of peak heart rate (f cpeak ), with no major differences between age groups (Table 4).

Respiratory compensation point (RCP)
The highest RCP was observed in the youngest age groups with roughly the same decline rate per decade as observed for V Than (Table 4).

Ventilatory efficiency at VO 2peak and at V Than
EqVO 2peak were similar between sexes and age groups. EqVCO 2peak was on average 1.2% (p,0.05) higher in women than in men. In men aged 20-59 years no differences were observed between subsequent age groups, upon which we observed a 4.4% (p,0.001) higher EqVCO 2peak for men aged 60-69 years compared to those aged 50-59 years. No differences were shown between the two oldest male age groups. In women we observed no differences in EqVCO 2peak between subsequent age groups throughout all decades. Comparing the youngest and oldest age groups an 8% (p,0.001) higher EqVCO 2peak was observed in the oldest age group, in both men and women (Table 5). EqVO 2Than was on average 3% (p,0.001) higher in women than in men. EqVO 2Than was similar and lowest in the two youngest male age groups (20-29 and 30-39 years). We observed a 5.2% (p,0.05) higher EqVO 2Than for men aged 40-49 years compared to those aged 30-39 years, with no differences between the subsequently older male age groups (40-49 through +70 years). In women no differences were observed between subsequent age groups throughout all decades. When comparing the youngest and oldest age groups a 12% (p,0.05) higher EqVO 2Than was observed in the oldest age group, among both men and women (Table 5).
EqVCO 2Than was on average 2.5% (p,0.001) higher in women compared to men. EqVCO 2Than was alike and lowest in the two youngest male age groups (20-29 and 30-39 years), whereas a 4.9% (p,0.01) higher EqVCO 2Than was observed in men 40-49 years compared to the 30-39 years group. No differences were found between men aged 40-49 years and 50-59 years, whereupon an average 5.3% (p,0.05) higher EqVCO 2Than was observed per decade between the three oldest male age groups. In women aged 20-69 years no differences in EqVCO 2Than was observed between subsequent age groups, upon which a 5.4% (p,0.05) higher EqVCO 2Than was observed in the oldest age group compared to those 60-69 years. Comparing the youngest and the oldest age groups showed a 16% (p,0.001) higher EqVCO 2Than in the oldest group, in both men and women (Table 5).
EqVCO 2VThan , among the 3 oldest age groups, stratified by fitness quartiles In those aged 50-59 years there was a 7.2% (p,0.05) and 8.2% (p,0.001) increase between the most fit (Q 1 ) and least fit (Q 4 ), men and women, respectively. The middle (60-69 years) and the most senior groups (+70 years) had increases of 13.4% (p,0.001; men) vs. 13.1% (p,0.001; women), and 16.4% (p,0.01; men) with no significant differences among women, between the fittest (Q 1 ) and least fit (Q 4 ), respectively (Table 6). Estimating key cardio pulmonary parameters from non-exercise prediction models Prediction equations for V Epeak , VCO 2peak and V Than were derived from nonexercise variables, including weight, height, age and sex. Weight and age proved negligible in predicting V Tpeak , as did weight and height in predicting EqVCO 2VThan and EqVO 2VThan , hence these variables were excluded from the respective models. For all models gender should be substituted with 1 or 2 for men and women, respectively. The final regression models are presented in table 7. Non-exercise prediction models for VO 2peak are previously published from the HUNT 3 fitness material [33].
Association between EqVCO 2VThan and age Figure 1 displays the relationship between EqVCO 2VThan and age, with relative low, but statistical strong correlations r50.27 (p,0.0001) and r50.18 (p,0.001) among men and women, respectively. The HUNT 3 Fitness Study Associations between V E and VCO 2 Figure 2 show the relationship between V E and VCO 2 from start of test until V Than , with strong correlations (men: r50.94; women: r50.93).
Sex and age group differences in peak ventilation (V Epeak ) and tidal volume (V Tpeak ) In this study women had approximately 34% and 32% lower V Epeak and V Tpeak , respectively, and a 4% lower peak f Bpeak than men. This is in agreement with other population-based studies on V Epeak [39,[42][43][44], V Tpeak and f Bpeak [22,45], and as expected as women have smaller lung size and dynamic lung function volumes than men, also after adjusting for differences in stature [46]. We observed 6-30% higher V Epeak among men and women compared to that seen in Brazilian [42] (n53992), American [43] (n5988) and French [44] (n5150) populations, as well as in small sample size studies [22,45,47,48]. Yet, a Norwegian study [49] (n5759) displays V Epeaks fairly consistent with ours. Hence, there might be population differences, which highlight the need of reference data in different populations. Lower V Epeak with increasing age is consistent with findings in Brazilian [42] (n53992), Israeli [23] (male 51424), Canadian [39] (n5100), [25] (male 5816) and French [44] (n5150) studies, and in line with an age attenuation in dynamic lung function largely attributed to decreased elastic recoil [50,51].
In this study the highest V Tpeak was observed among the 30-49 year groups, in both men and women, with a decrease in subsequent age groups. These findings are unexpected, since the highest V T should be in the youngest age group, with deterioration between subsequently older age groups [50,51]. Our findings could be explained by the relative low sample size in the youngest age group. Contrary to us a Canadian [39] (n5100) and Israeli [23] (n51424) study presented their highest V Tpeak in the youngest age groups (15-25 yrs). Interestingly, while the Canadian study displayed the same male average V Tpeak as us, they have significantly lower V Epeak , signaling a necessarily lower f Bpeak (not displayed).

Association between V T and V E below and above V Than
On the initial sub maximal workloads the V T vs. V E slope displays a steeper gradient than towards test termination. This is to be expected, since it is well established that V T increases steeper than f B below V Than , whereas f B mostly accounts for the increase in V E at workloads above V Than [47,52].

Sex and age group differences in V Than and RCP
We observed V Than at an average 77% of VO 2peak , in both sexes, and in line with other studies (n5204-3992) [24,40,42,53,54] minor sex differences. Previous studies report V Than at significantly lower fractions (49-70% of VO 2peak ) [23,24,40,42,53,54], or more consistent to that observed by us [25,55,56]. Differences are most likely caused by use of different methods and analyzing Table 6. Ventilatory efficiency and oxygen uptake presented in fitness quartiles: The HUNT 3 fitness study. The HUNT 3 Fitness Study approaches applied in the different studies, which makes direct comparisons difficult [23,25,42,55,56]. In this study V Than was observed at <75% in the youngest age groups and at <80% of VO 2peak in the oldest age groups, in both men and women, with significant differences (p,0.05) between the youngest and the 60-69 year age group. Age related increase in V Than (as percent of VO 2peak ) is reported in previous studies as well [23-25, 42, 44, 53, 54, 57]. This is to be expected since V Than (L?min 21 ) declines at a slower rate than VO 2peak (L?min 21 ) with increasing age [58,59], and consequently occurs at a higher percent of VO 2max/peak [60]. This is suggested to be, at least partly, due to changes in skeletal muscle composition associated with increasing age, with the selective loss of type II fibers and therefore a relative increase in type I fibers [61]. Contrary to this Lenti and colleagues [34] report a decrease in percent V Than in a trained senior group (n516), compared to their young trained, whereas the untrained groups are consistent with our findings (n516). The small sample size taken into account, their data must be interpreted with caution. RCP was observed at 86% and 90% of VO 2peak , among men and women, respectively. This is consistent with the findings of several other studies [18,35,37]. However, direct comparisons are difficult due to their small sample sizes (n59-22) and the use of different measuring methods.

Ventilatory efficiency stratified by sex
In line with previous studies [17,22] we observed similar EqVO 2peak in men and women. Wasserman [52] suggests that EqVCO 2 should be determined at V Than , or between V Than and RCP as V E is least variable in this range. Our submaximal level 2 measurements are close to V Than and we observed slightly higher (p,0.001) EqVCO 2VThan in women than men, hence indicating less efficient ventilation in women. These observations are in agreement with previous studies [19,20,62]. Women's lower ventilatory efficiency might be explained by differences in ventilatory stimuli (e.g. [H + ], [K + ]), metaboreceptors, and central command [63,64].

Ventilatory efficiency at V Than stratified by age groups
We observed deterioration in ventilatory efficiency, both in EqVO 2Than (p,0.05) and EqVCO 2Than (p,0.001), between the youngest and oldest age groups. This is in agreement with the findings of other studies (n569-474) [19,20,57,62]. It is presently uncertain which factors are responsible for diminished ventilatory efficiency during exercise with increasing age [65]. Clearly, increased dead space might be a major contributing factor, as well as the lung's mechanical limitation to airflow, which deteriorate as the lung loses elastic recoil with increasing age [66]. In women evidence points to decreased leg muscle strength as a contributing factor [65]. Other suggestions are factors linked to muscle afferent excitability as a result of fiber type shifts [67], and neuromuscular alterations with growing age [68].

EqVCO 2VThan stratified by fitness quartiles
Individuals in the three oldest age groups (50-59, 60-69, +70 years) in the present study are more likely to be referred to clinical exercise testing than younger age groups, and have been studied in more depth than other age groups when it comes to EqVCO 2VThan . Interestingly, among individuals in these age groups approximately 25% had EqVCO 2VThan higher than 30. EqVCO 2VThan ,30 is considered normal with a possible increase among older age groups [1]. In these age groups we observed minor differences in EqVCO 2VThan between those that were among the three first quartiles of fitness (VO 2peak quartiles), whereupon we observed a significant drop in ventilatory efficiency (hence, an increase in EqVCO 2VThan ) in those categorized as being least fit (Q 4 ) in both sexes and all three age groups. All fitness quartiles in the oldest age group (+70 years), and the least fit quartiles in the two younger age groups (50-59, 60-69 years) had EqVCO 2VThan <.30, which could be caused by high dead space ventilation due to diminished alveoli perfusion [19]. More importantly, diminished ventilatory efficiency can reflect disease severity and prognosis in several patient groups including chronic obstructive pulmonary disease, pulmonary arterial hypertension, hypertrophic cardiomyopathy and interstitial lung disease [1,15], and are displayed in the range 41-60 in more severe cases of congestive heart failure [15]. Although we cannot totally exclude the possibility of unknown diseases in some of our participants, the self-report and medical interview, adhering to our inclusion criteria, provides a healthy population. Therefore seen in context of our sample size our findings may represent normal ventilatory efficiency values for the oldest age groups and least fit population.

Non-exercise prediction models for key cardio-respiratory variables
There is a plethora of VO 2peak prediction models. However, models on other key cardio-respiratory variables are less abundant. The accuracy of previously published models on VO 2peak from the HUNT 3 fitness data (Men: 12.8%, Women: 14.3%) [33] is in fair agreement with previous large sample studies (Jurca 2005). Smaller sample studies [69][70][71] with uniform populations [72,73] show accuracies in the range ¡7-17%. Also V Than accuracy (¡19.9%) is approximately the same compared to 7 previous small-scale studies [40]. Our precision of V Epeak prediction (¡17.2%) is in agreement with a large study of males [23], contrasted by a small study [39] showing 28% accuracy. However, it was hard to compare our V Tpeak accuracy (¡15.5%) with others [23], since key data was not presented. The 12.1% accuracy in predicting EqVCO 2VThan was similar to a small sample study [62]. Our prediction models will provide a rough estimation of these key variables, regardless of gender and age. Moreover, the models use non-exercise variables that are easy to measure, thus making these models easy to use in both clinical settings and for recreational athletes.

Association between ventilatory efficiency and workload
The slope of the ventilatory equivalent for oxygen (EqVO 2 ?W 21 ) increases with rising workload (w), which demonstrates reduced ventilatory efficiency as the workload increases. This is supported by two former case studies [30,52]. More interestingly the EqVCO 2 ?W 21 displays a gradient close to zero, which indicates a constant ventilatory efficiency throughout the incremental workload, and is in fair agreement with Wasserman and colleagues [52], yet, contradicted by another study [30] that presents an increase in EqVCO 2 as the workload approaches peak. However, it is noteworthy that these studies are based on single case observations, and thus difficult to compare with our findings.

Strengths and limitations
The large sample size, inclusion of men and women, wide age distribution and cardio-respiratory measurements up to the true VO 2max makes this study robust. The lack of spirometry data limits the assessment of ventilatory parameters. Also this study may be subject to bias due to self-selection caused by the low participation rate. However, almost all those 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, some 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 [26].