Self-reported physical activity compared with maximal oxygen uptake in adults

Abstract

Background

Physical activity level and cardiorespiratory fitness are both inversely associated with the risk of cardiovascular diseases and with all-cause mortality. Physical activity questionnaires are often validated against objectively measured maximal oxygen uptake (Vo  _2max).

Aim

To validate a self-report physical activity questionnaire against Vo  _2max and furthermore to establish whether a simple question on self-rated physical fitness could predict objectively measured Vo  _2max.

Methods

A total of 102 men and women aged between 35 and 65 years were recruited from an ongoing population-based intervention study, the Inter99 Study. Participants reported their self-rated fitness and daily physical activity using a new questionnaire based on metabolic equivalents (MET). Vo  _2max (ml/kg per min) was determined using a graded bicycle test with increasing workload until exhaustion and with simultaneous measurement of breath-to-breath oxygen uptake in 15-s periods. Statistical analyses were performed by linear regression analyses using the self-reported physical activity level as an independent variable and Vo  _2max (l/min) as an dependent variable, and with covariates sex, age and bodyweight.

Results

Data from 53 men and 47 women were analysed. The amount of daily vigorous activity (>6 MET) was significantly positively associated with Vo  _2max (P=0.0001, R  ² = 0.76), whereas the total amount of physical activity was not significantly associated with Vo  _2max (P=0.098, R  ² = 0.69). A significant trend across three groups of self-rated fitness in relation to Vo  _2max (ml/kg per min) was found (P for trend <0.0001).

Conclusion

The physical activity questionnaire has acceptable validity when compared with Vo  _2max in adult men and women. Furthermore, a simple question on self-rated fitness seems to reflect measured Vo  _2max objectively.

maximal oxygen uptake, physical activity, questionnaire, validation

Introduction

Physical activity questionnaires remain the most practical approach for assessing activity levels in large population studies investigating the health consequences of activity status. Both self-reported physical activity and objectively measured cardiorespiratory fitness are inversely associated with the risk of cardiovascular diseases and all-cause mortality [1–4]. Physical activity is the principal determinant of cardiorespiratory fitness, although there is a genetic component [5, 6], and the beneficial effects of physical activity on health and survival may be mediated via fitness status. Information from physical activity questionnaires may result in greater misclassification than information from objective fitness tests, but the questionnaires are easier and less expensive to administer in large study populations [1].

Maximal oxygen uptake (Vo  _2max) as a measure of cardiorespiratory fitness has been used as an indirect validation criterion in several validation studies on physical activity questionnaires [7–17]. The precision of Vo  _2max determination depends strongly on the exercise protocol and the Vo  _2max determination method [18]. Maximal tests are more precise, but also require more advanced equipment and pose a greater risk for the participants than estimates of Vo  _2max derived from submaximal tests. Most authors find modest correlations between the total volume of physical activity and Vo  _2max in both sexes and across different age groups, whereas the amount of vigorous physical activity, corresponding to activities greater than 6 metabolic equivalents (MET), is more highly correlated with Vo  _2max [9, 10, 14, 15].

The aim of the present study was to validate a new self-report physical activity questionnaire [19] against Vo  _2max, by exploring the relationship between the total amount of physical activity and vigorous physical activity, respectively, and objectively measured Vo  _2max in adult men and women. We also aimed to establish whether a simple question on self-reported physical fitness could predict objectively measured Vo  _2max.

Methods

Study population

From the 5-year follow-up of a population-based intervention study, Inter99, a volunteer sample of 102 men and women aged 35-65 years was recruited. Exclusion criteria were pregnancy or physical disability or disease that made riding a bicycle difficult or impossible. The Inter99 study is a population-based intervention study, initiated in 1999 [20]. The initial study population consisted of 61 301 men and women between 30 and 60 years, living in Copenhagen County and obtained through the Danish Civil Registration System. The population was pre-randomised into three groups: a high intensity intervention group (A), a low intensity intervention group (B) and a control group (C). Groups A and B (N = 13 016) were invited for an initial health examination and an assessment of the risk of ischaemic heart disease (IHD) at baseline. The risk of IHD was assessed using the computer programme, the Copenhagen Risk Score, which calculates the 10-year absolute risk of IHD by mean of sex, age, heredity, former IHD, diabetes, height, weight, smoking habits, cholesterol and blood pressure [21]. High risk was defined as the upper quintile of the absolute risk of IHD within 10 years calculated for each age and sex group, or at least one of the following characteristics: daily smoking, systolic blood pressure 160 mmHg or greater, total cholesterol 7.5 mmol/l or greater, body mass index 30 or greater, a history of diabetes or either diabetes or impaired glucose tolerance as evaluated by an oral glucose tolerance test [20].

In both groups those at high risk of disease were offered lifestyle consultation at baseline and at 1 and 3-year follow-ups, whereas those at low risk were just asked to fill out a questionnaire. Those at high risk in group A were offered participation in a group intervention in addition to the lifestyle consultations. Of the 13 016 individuals sampled, 82 were non-eligible, as they had died or could not be traced. Of the remaining 12934 a total of 6906 (53.4%) turned up for the investigation. Of these, 122 were excluded either because of alcoholism or drug abuse (n = 23) or because of linguistic problems (n = 99), leaving 6784 participants (52.5%) in the study [20]. At the 5-year follow-up all in groups A and B were invited to the centre again for a health examination, starting on 15 March 2004.

When the Inter99 participants met at the Research Centre for Prevention and Health for the 5-year examination, a pamphlet was handed out, informing them about the present study and inviting them to participate. If interested, participants were requested to call a telephone number listed in the pamphlet. Participants who called were informed in detail about the Vo  _2max testing, and were asked about their Civil Registration number, whether they were physically disabled, suffered from any diseases or were on medication. Women were specifically asked about pregnancy. Eligible participants were then scheduled for testing at the University Hospital, Rigshospitalet, Copenhagen, and were sent further information about the study by mail.

We aimed at obtaining test results from 100 participants, and 102 participants were therefore included in the study, leaving room for the exclusion of a few test results, if necessary, for example, because of failure to measure peak oxygen uptake (VO  ₂).

Physical activity questionnaire

When participants met at the test laboratory they were informed again about the testing in detail. They filled in a physical activity questionnaire and the Vo  _2max testing was carried out immediately thereafter.

Information on physical activity level was obtained by a new, initially validated, self-report questionnaire measuring physical activity on an average weekday as a total 24-h MET score [19]. MET express intensity of activity in comparison to resting energy expenditure, with one MET equal to the standard for resting energy expenditure (approximately 3.5 ml oxygen consumed per kilogram of bodyweight) [22]. In the questionnaire, participants were asked to describe their habitual physical activity and inactivity on an average weekday, by filling out the amount of time spent on nine different intensity levels of physical activity, ranging from sleep (0.9 MET) to vigorous physical activity (> 6.0 MET). A 24-h MET score was calculated by multiplying the time spent on each activity level by the assigned MET value and adding the nine MET activity levels together. In addition, participants were also asked to rate their own physical fitness as excellent, very good, good, fair or poor.

Maximal oxygen uptake

Vo  _2max was determined during a standardized graded bicycle test with increasing workload until exhaustion, based on the levelling off criterion and following an individualized protocol [23]. The subjects were seated in a chair in an upright position 5 cm above the crank on an electrically braked Krogh ergometer bicycle. All participants wore bicycle shoes locked to the pedals, and were instructed to follow a cadence of 60 rpm, monitored by a metronome and an rpm display. The test consisted of two submaximal workloads maintained at steady state for 4-5 min (approximately 50 and 75% of the maximal capacity, individually predicted from sex, age and habitual physical activity level). The test was carried out until exhaustion, ensuring that an increase in work output did not result in any further rise in Vo  ₂ (Vo  ₂ plateau) and that the respiratory exchange ratio was greater than 1.1. Vo  ₂ was determined continuously by breath-to-breath measurement, summarized every 15 s (CPX Express; Medical Graphics, St Paul, Minnesota, USA). Before testing, height was measured without shoes, weight was measured with light clothing (scales; Seca, Hamburg, Germany), and blood pressure was measured while participants were sitting down resting, to make sure that no participant had excessively high blood pressure before being tested (Digital Blood Pressure Monitor; A&D Instruments, oxford, UK). Electrocardiogram (Satellite plus) was measured continuously during testing, and heart rate was simultaneously monitored (Polar Sports Tester Xtrainer plus; Polar Instruments, Helsinki, Finland). A trained exercise technician and a medical doctor supervised the testing.

Statistical analyses

Statistical analyses were performed by means of descriptive statistics, using means, standard deviations and frequencies. The association between the self-reported physical activity level and aerobic capacity was analysed by linear regression analysis with Vo  _2max (l/min) as a dependent variable and the daily total physical activity MET score or daily vigorous activity MET score as independent variables. Sex, age group and weight in kilograms were included as covariates. R  ² described to what extent a variation in the response variable was explained by the included explanatory variables. The continuous independent variables, total physical activity MET score, vigorous activity MET score and weight in kilograms were tested for linearity in the regression models. The vigorous physical activity MET score was not linear but polynomial, and a squared form of the variable was therefore added in the final adjusted regression model. Interaction between total physical activity∗sex and total physical activity∗age group, as well as vigorous physical activity∗sex and vigorous physical activity∗age group was tested for in the full adjusted regression models. P values less than 0.05 were considered significant. Analyses were performed using the SAS 9.1 statistical package (SAS Institute, Cary, North Carolina, USA).

All participants gave written informed consent and the study was approved by the local ethics committee (KA04070m).

Table 1.

Distribution of study population: age groups, physical activity level in leisure time and at work and self-rated fitness (N = 100)

	Men (n = 53) no. (%)	Women (n = 47) no. (%)
Age group
Young (35, 40 years)	11 (21)	12 (26)
Middle age (45, 50-55 years)	29 (55)	25 (53)
Older (60-65 years)	13 (24)	10 (21)
Leisure time physical activity (n = 99)1
Sedentary	1 (2)	3 (7)
Light activity	26 (49)	24 (52)
Moderate activity	23 (43)	17 (37)
Vigorous activity	3 (6)	2 (4)
Work physical activity (n = 99)1
Sedentary	20 (38)	17 (37)
Light activity	16 (30)	17 (37)
Moderate activity	8 (15)	6 (13)
Vigorous activity	1 (2)	−
Not working	8 (15)	6 (13)
Self-rated fitness (n = 99)1
Excellent	1 (2)	−
Very good	16 (30)	8 (17)
Good	32 (60)	30 (65)
Fair	3 (6)	7 (15)
Poor	1 (2)	1 (2)

a Missing questionnaire response from one female participant. Answers from only n = 99 are reported.

Table 2.

Characteristics of study population (N=100)

	Men (n=53) Mean (SD)	Women (n =47) Mean (SD)
Body weight (kg)	83.05 (12.97)	66.63 (13.37)
Body mass index (kg/m2)	25.71 (3.53)	24.12 (4.53)
Vo  2max, (ml/kg per min)	37 (7)	31 (5)
Vo  2max, (l/min)	3.044 (0.557)	2.024 (0.439)
Daily total physical activity (MET)	49.7 (10.2)	47.2 (8.3)
Daily vigorous physical activity (MET)	5.0 (4.7)	3.9 (4.8)

MET, Metabolic equivalent; Vo  _2max, maximal oxygen uptake.

Results

Fifty-three men and 49 women were included in the study. Data from two participants were excluded, one participant terminated the test before peak oxygen consumption was measured and one participant had given very unlikely answers in the questionnaire. This left data from 100 participants for analysis, 53 were men and 47 were women. Age distribution and other study population characteristics are presented in Tables 1 and 2.

Six men and two women were on antihypertensive medication, two women and one man were on lipid-lowering medication, and one male participant had type II diabetes.

Vo  _2max (ml/kg per min) in relation to self-rated fitness is presented in Fig. 1. Only one participant rated his fitness as ‘excellent’ and only two rated their fitness as ‘poor', so excellent/very good were combined and fair/poor were combined. Across the remaining three groups of self-rated fitness, a significant trend with Vo  _2max was seen (P for trend < 0.0001). In Table 3, the linear regression model of the relationship between self-reported, daily total physical activity MET score and Vo  _2max (l/min) is presented. When weight, sex and age were included as covariates, 69% of the variation in Vo  _2max (l/min) was explained. The effect of total physical activity on Vo  _2max (l/min) was not significant, although the linear relationship was positive, as illustrated in Fig. 2. In Table 4 the effect of self-reported daily vigorous physical activity (> 6 MET) on Vo  _2max (l/min) is presented. Seventy-six per cent of the variation in Vo  _2max (l/min) is explained by the included variables, and the effect of vigorous physical activity is significant, also when total physical activity is adjusted for in the model.

This is also illustrated in Fig. 3, in which the slope of the ascending line is more pronounced than the slope of the line in Fig. 2, indicating a stronger relationship between vigorous activity and Vo  _2max (l/min) than between the total amount of physical activity and Vo  _2max (l/min). As seen in Fig. 3, the relationship between self-reported vigorous activity and Vo  _2max (l/min) was polynomial rather than linear.

Fig. 1.

Box and whiskers plot of the relationship between groups of self-rated fitness and maximum oxygen uptake (ml/kg per min). The bottom and top of box edges are located at the 25th and 75th percentiles. The dot represents the mean value and the whiskers extend from the box to a distance of at most 1.5 interquartile ranges. 1, Excellent and very good (n = 25); 2, good (n = 62); 3, fair and poor self-rated fitness (n = 12). P for trend <0.0001 across groups of self-rated fitness (N=99). Vo  _2max, Maximum oxygen uptake.

Table 3.

Effects of total physical activity: metabolic equivalent score on maximal oxygen uptake adjusted for sex, age and weight, linear regression (N = 100)

	Parameter estimates β	SE	P value
Total activity MET score	7.56	4.52	0.0980
Weight	15.54	3.14	<0.0001
Sex			<0.0001
Women (reference)	0
Men	772.72	97.05
Age group (years)			<0.0001
35-40	685.14	122.63
45-55	343.06	104.60
60-65 (reference)	0

MET, Metabolic equivalent. R  ² = 0.69.

Fig. 2.

Linear relationship between metabolic equivalent score for total volume of self-reported daily physical activity and maximum oxygen uptake (l/min). Based on the parameter estimates from the linear regression analyses, the model describes the linear relationship between the total volume of self-reported daily physical activity and maximum oxygen uptake (Vo  _2max; l/min) for a woman 60-65 years of age and weighing 70 kg. MET, Metabolic equivalent.

Discussion

In this study we compared Vo  _2max with the self-reported physical activity level and found that the amount of daily vigorous activity (> 6 MET) was strongly associated with Vo  _2max (l/min), whereas the total volume of daily physical activity showed a slight, but not statistically significant, association with Vo  _2max (l/min) in this volunteer sample of adult men and women. Furthermore, we found that self-rated fitness was associated with objectively measured Vo  _2max (ml/kg per min), meaning that participants could assess their own aerobic capacity fairly accurately.

Table 4.

Effects of vigorous physical activity: metabolic equivalent score on maximal oxygen uptake adjusted for sex, age, weight and total physical activity, linear regression (N=100)

	Parameter estimates β	SE	P value
Vigorous activity MET score	82.55	20.52	0.0001
Vigorous activity MET score2	−3.12	1.32	0.0206
Total activity MET score	−1.64	4.61	0.7236
Weight	15.54	2.83	<0.0001
Sex			<0.0001
Women (reference)	0
Men	733.01	87.73
Age group (years)	<0.0001
35-40	673.44	110.51
45-55	366.72	95.94
60-65 (reference)	0

MET, Metabolic equivalent. R  ² = 0.76.

Fig. 3.

Linear relationship between metabolic equivalent score for self-reported daily vigorous physical activity and maximum oxygen uptake (l/min). Based on the parameter estimates from the linear regression analyses, the model describes the polynomial relationship between self-reported daily vigorous physical activity and maximum oxygen uptake (Vo  _2max; l/min) for a woman 60-65 years of age and weighing 70 kg. MET, Metabolic equivalent.

The findings indicate that the questionnaire gives valid estimates of vigorous activity and that the amount of time spent on vigorous physical activity is associated with cardiorespiratory fitness. our questionnaire was validated against accelerometry (MTI actigraph) and physical activity diaries in a previous study [19], and a high correlation was found between total daily MET score as measured by the questionnaire and physical activity diaries, respectively (r = 0.74), and a rather poor correlation was seen between the total daily MET score as measured by the questionnaire and average total accelerometry counts (r = 0.20). These correlations correspond to findings in similar validation studies, supporting the conclusion that the questionnaire is a valid alternative to measuring physical activity by diary.

Participants in the present study seemed capable of rating their physical fitness in correspondence with objectively measured Vo  _2max. Likewise, a recent study by Mikkelsson et al. [24] found moderate correlations between self-estimated physical fitness and Vo  _2max estimated from submaximal ergometer tests in a group of 40-year-old men. Mikkelsson et al. [24] recommended a direct Vo  _2max test, like the one applied in the present study, for determining how well self-estimated fitness corresponds to measured fitness in future studies.

Vo  _2max is considered the most valid measure of cardiorespiratory fitness. The procedures necessary for estimating Vo  _2max are, however, costly and involve a small but finite risk for participants [25], and it is therefore not a feasible method in large study populations. A number of studies have used it as a validation criterion, when assessing the validity of the self-reported measurement of physical activity [8–16, 26, 27]. In line with the findings in this study, most of the validation studies found that vigorous physical activity is more closely correlated with Vo  ₂ than the total amount of physical activity. Siconolfi et al. [15] found that the self-reported frequency of activity sufficient to generate sweating related more closely to Vo  _2max than a more complex self-reported physical activity index that included stair-walking, blocks walked and the frequency of sports and recreational activity. The study was carried out among 36 men and 32 women with a mean age of 41 and 42 years, respectively, a study population very similar to this one. Bonnefoy et al. [10] validated 10 different physical activity questionnaires against Vo  _2max testing in elderly men (treadmill), and found a low to moderate correlation with the total amount of physical activity and a high correlation with vigorous activity, corresponding to the findings in this study. Richardson et al. [14] validated the Stanford 7-day questionnaire against peak Vo  ₂, measured by graded treadmill exercise, and found significant correlations between peak Vo  ₂ and both total and vigorous physical activity, but more pronounced for vigorous activity. Wareham et al. [16] found moderate, but significant, correlations between the time spent on vigorous activity and cardiorespiratory fitness measured as submaximal Vo  ₂ in a validation study of the EPIC-Norfolk physical activity questionnaire.

Although applied in many studies, it remains debatable whether cardiorespiratory fitness is an appropriate standard for the validation of physical activity questionnaires. Physical fitness and physical activity essentially represent two different dimensions, but they are strongly linked and they display very similar beneficial effects on health and survival [1]. Physical activity is a complex, multidimensional exposure, which is difficult to measure and an ideal ‘all purpose’ validation criterion does not exist [28]. The fact that higher correlations with Vo  _2max are found for vigorous activity than for total daily physical activity may be caused by the fact that vigorous activities are more reliably recalled and reported in questionnaires [28]. Individuals with high energy expenditure may not necessarily be fit, as measured by cardiorespiratory fitness, but may engage in prolonged moderate activity, rather than bursts of vigorous activity. Likewise, individuals who do engage in vigorous activity may compensate by spending less time on moderate activity, and therefore have a relatively low total energy expenditure. By relating Vo  _2max to self-reported physical activity, we can only study whether the physical activity questionnaire appears to measure the component of physical activity that relates to cardiorespiratory fitness, but we cannot establish whether it is an all-round valid measure of physical activity.

The present study has some strengths and limitations that should be further addressed. The precision of Vo  _2max determination depends strongly on the exercise protocol and the Vo  _2max determination method. In this study direct assessment was utilized with a maximal exercise test, and the measure of cardiorespiratory fitness is therefore considered rather accurate. The validation study has been conducted in the same population as the questionnaire was intended for (the Inter99 study), and the sex and age distribution of the validation sample corresponds to the distribution in the Inter99 study [20]. Finally, the relationship between vigorous physical activity and Vo  _2max was best described by a polynomial curve and not by a straight line in the regression analysis. This phenomenon may have been caused by severe overestimation of the time spent on vigorous physical activity by a few of the respondents. Self-reported physical activity is known to produce greater misclassification than objectively measured cardiorespiratory fitness, and the ‘true’ relationship between vigorous physical activity and Vo  _2max may have been linear in a larger study sample.

Altogether, we suggest that the physical activity questionnaire is a useful tool for assessing the daily physical activity level in large epidemiological studies, but we do not claim to have explored all aspects of physical activity measurement. The measurement of physical activity and the validation of activity questionnaires remain both a challenge and a controversial issue.

In conclusion, we found that self-reported physical activity, as measured by a questionnaire, developed for and tested in a population-based intervention study, reflects cardiorespiratory fitness when compared with the measurement of Vo  _2max in volunteer men and women from the same study population. The amount of time spent on daily vigorous activity is strongly associated with Vo  _2max, as is the self-reported fitness category.

Acknowledgements

The Inter99 Steering Commitee consisted of Torben Jørgensen (principal investigator), Knut Borch-Johnsen (co-principal investigator), Hans Ibsen, Charlotta Pisinger, Charlotte Glümer and Troels F. Thomsen. This study was supported by research grants from the Copenhagen Hospital Corporation and the Danish Physical Therapy Association.

There is no conflict of interest.