The amount of usual physical activity (PA) accumulated by an individual is strongly related with all-cause morbidity and mortality risk(Reference Blair, Cheng and Holder1). Healthy adults should accumulate 30 min or more of moderate-intensity aerobic PA on 5 d each week or a minimum of 20 min of vigorous-intensity aerobic activity on 3 d each week(Reference Haskell, Lee, Pate, Powell, Blair, Franklin, Macera, Heath, Thompson and Bauman2). The use of step count recommendations as useful behavioural PA targets has been suggested in the past(Reference Hatano3, Reference Hatano4). These step count thresholds are more practical than the 30 min MVPA/d guideline since they do not imply constant time tracking and summing of at least moderate-intensity activity during the day, which is impractical for PA assessment on individual and population levels. Additionally, more detailed pedometer-based indices for public health (sedentary: <5000 steps/d; low active: 5000–7499 steps/d; somewhat active: 7500–9999 steps/d; active: 10 000–12 499 steps/d; highly active: ≥12 500 steps/d) have been introduced by Tudor-Locke and Bassett(Reference Tudor-Locke and Bassett5).
Being able to accurately quantify the amount of PA accrued during daily life is necessary for researchers and health professionals to better understand PA behaviour and to develop successful programmes to increase activity levels in various populations. Traditionally, PA has been measured through questionnaires, the strengths and limitations of which are well known(Reference Montoye, Kemper, Saris and Washburn6). Recognized benefits are the possibility to assess different dimensions of PA, the ability to collect data from a large number of people at low cost, and the unobtrusiveness of the instrument(Reference Sallis and Saelens7). However, there are limitations in subjects’ recall ability and social desirability bias can cause over-reporting of PA(Reference Sallis and Saelens7). Recently, more attention has been given to objective instruments to assess PA, e.g. accelerometers and pedometers. Pedometers have been shown to provide a valid and accurate measure of ambulatory activities in free-living conditions(Reference Bassett, Ainsworth, Legget, Mathien, Main, Hunter and Duncan8, Reference Welk, Differding, Thompson, Blair, Dziura and Hart9). They are simple to use, more objective than surveys, and less expensive (approximately $US 20–50) compared with accelerometers (approximately $US 150–500). Despite their limitations such as variability between different brands, insensitivity to non-ambulatory activities (i.e. cycling, swimming) and increased errors at slow walking speeds (<60 m/min)(Reference Bassett, Ainsworth, Legget, Mathien, Main, Hunter and Duncan8), pedometers have become popular devices for public use and for researchers. Moreover, step counts have proved to be useful in PA studies in free-living populations(Reference Bassett, Ainsworth, Legget, Mathien, Main, Hunter and Duncan8). However, the question remains of which exactly dimensions of PA are measured by pedometers. Several researchers(Reference Bassett, Ainsworth, Legget, Mathien, Main, Hunter and Duncan8, Reference Shephard10) report that pedometer step counts are an appropriate measurement of the distance walked, while others believe that pedometers also enable the quantification of ambulatory behaviour during occupational, leisure-time, household and transportation activity(Reference Hornbuckle, Bassett and Thompson11, Reference Sequeira, Rickenbach, Wietlisbach, Tullen and Schutz12). In addition, one could wonder if step counts are simply a measure for walking or if they can also be associated with self-reports of PA that encompass intensity (other moderate and vigorous PA, besides walking). Therefore, the first purpose of the present study was to compare pedometer data with data from questionnaires, the most commonly used(Reference Sallis and Saelens7) and practical method for PA assessment in large-scale studies(Reference Bouchard, Shepard, Stephens, Sutton and McPherson13). The objective was to evaluate the associations between step counts and PA (walking, moderate PA, vigorous PA) reported in four different validated questionnaires: the interviewer-administered Minnesota Leisure Time Physical Activity Questionnaire(Reference Taylor, Jacobs, Schucker, Knudsen, Leon and Debacker14) (MLTPAQ), the self-administered long version and self-administered short version of the International Physical Activity Questionnaire(Reference Craig, Marshall and Sjöström15) (long-form IPAQ, short-form IPAQ), and the self-administered Baecke Questionnaire(Reference Baecke, Burema and Frijters16). Additionally, questionnaire-based PA was compared between the five ‘step count groups’ based on the pedometer health indices of Tudor-Locke and Bassett(Reference Tudor-Locke and Bassett5).
A third objective of the present study was to compare step count thresholds (7500, 10 000 and 12 500 steps/d) with the guideline of 30 min of moderate to vigorous PA (MVPA) per day. Wilde et al.(Reference Wilde, Sidman and Corbin17) found that sedentary women who added a 30 min walk to their daily habits accumulated approximately 10 000 steps. Another study(Reference Hultquist, Albright and Thompson18) also revealed that on days when women took a 30 min walk, their average step count was close to 10 000. Other researchers(Reference Le Masurier, Sidman and Corbin19) found that women who took 10 000 steps/d were more likely to meet the current MVPA guideline, compared with those not accumulating as many steps. Miller and Brown(Reference Miller and Brown20) found only a moderate level of agreement between meeting 10 000 steps/d and 150 min or more of PA over five or more sessions per week. However, no research could be located in either gender or in a European sample assessing whether step count thresholds are corresponding to the 30 min MVPA/d guideline. Therefore, the present study explored whether adults reaching 7500, 10 000 and 12 500 steps/d also reached 30 min MVPA/d, based on self-reports.
Methods
Participants and procedures
A convenience sample of 310 volunteers (146 men) living in Flanders, Belgium participated in the present study. Participant characteristics are shown in Table 1. No gender differences were found for the proportions of employed participants, mean ages and average daily step counts.
Table 1 Characteristics of the study participants: healthy adult volunteers, Flanders, Belgium
NS, P > 0·05.
All participants were visited at home by research assistants for the completion of the MLTPAQ. After this interview, participants were asked to complete the self-administered long-form IPAQ, the self-administered short-form IPAQ and the Baecke questionnaire. Then, participants were asked to register pedometer-based step counts for seven consecutive days. They were instructed to wear the pedometer on their waistband or belt during waking hours. All participants were asked to carry on their usual activities, to remove the pedometer only while swimming, bathing or showering, and to complete an activity log at the end of each day. All participants provided written informed consent and the study was approved by the Ethical Committee of Ghent University. Research assistants (masters-level graduates, native speakers) were given 2 h of training and interview practice.
Instruments
Pedometers
The Yamax Digiwalker SW-200 (Yamax, Tokyo, Japan) was used in the present study as it is known to be a valid, reliable and accurate instrument for counting steps in adult populations(Reference Crouter, Schneider, Karabulut and Bassett21).
Activity log
Based on procedures of Tudor-Locke et al.(Reference Tudor-Locke, Lind, Reis, Ainsworth and Macera22), participants were asked to keep daily activity records in an activity log during the seven consecutive days of pedometer registration. They were requested to record the date, the number of steps recorded at the end of the day, and the type and duration of non-ambulatory activities (e.g. 20 min of biking/swimming). Following established guidelines(Reference De Cocker, Cardon and De Bourdeaudhuij23, Reference Miller, Brown and Tudor-Locke24), 150 steps were added to the daily total for every minute of reported biking and/or swimming. The average daily step count was 9981 (sd 3455) without adding the equivalent for biking/swimming and 12 087 (sd 4888) with the added steps for biking/swimming. Of the total sample, 152 people (49·0 %) reported biking/swimming during the week of pedometer registration (mean: 7·6 (sd 15·1) min/d).
MLTPAQ
A structured interview was used to solicit detailed information on leisure-time PA over a 1-year period. A Dutch version of the validated MLTPAQ, developed for the Belgian Physical Fitness Study, was used(Reference Montoye, Kemper, Saris and Washburn6, Reference Taylor, Jacobs, Schucker, Knudsen, Leon and Debacker14, Reference De Backer, Kornitzer, Sobolski, Dramaix, Degré, de Marneffe and Denolin25). Participants’ engagement in different activities was queried, together with the number of months per year, the monthly frequency and the duration of the activity. Activities were classified as walking (structured walking for transport or in leisure time), moderate-intensity PA (MPA) or vigorous-intensity PA (VPA), based on their energy requirements (MPA: 3–6 MET; VPA: >6 MET; MET = metabolic energy equivalent task)(Reference Ainsworth, Haskell and Whitt26).
Long-form IPAQ
The self-administered long-form IPAQ was used to assess PA at work, transport-related PA, domestic and gardening activities, and PA during leisure time in a usual week. Total times engaged in walking, MPA and VPA, all expressed in min/week, were computed according to the guidelines (www.ipaq.ki.se). The IPAQ has been shown to be a valid and reliable instrument for measuring PA in Europe(Reference Craig, Marshall and Sjöström15) and in Flanders, Belgium(Reference Vandelanotte, De Bourdeaudhuij, Sallis, Philippaerts and Sjöström27).
Short-form IPAQ
PA was assessed using the self-administered short-form IPAQ, which provides information on the time spent walking, in MPA and in VPA (min/week) in a usual week. This version of the IPAQ has been found to be valid and reliable(Reference Craig, Marshall and Sjöström15).
Baecke questionnaire
In the Baecke questionnaire(Reference Baecke, Burema and Frijters16), responses were scored on a five-point scale and resulted in three different indices reflecting PA during work (work index), PA during leisure time excluding sport (LT index) and PA during sport activities (sport index). The summation of the three indices was defined as the overall PA index (total PA index). Good validity of the Baecke questionnaire for the assessment of PA was found in the past(Reference Pols, Buenodemesquita, Ocke, Wentick, Kemper and Collette28, Reference Philippaerts, Westerdorp and Lefevre29).
Data analysis
Analyses were carried out using the SPSS for Windows statistical software package version 12·0 (SPSS Inc., Chicago, IL, USA). Average daily steps were calculated and values above 20 000 steps/d were recorded as 20 000 steps/d to limit unrealistically high average step counts(Reference Tudor-Locke, Ham, Macera, Ainsworth, Kirtland, Reis and Kimsey30). Because of the non-normal distributions in PA data, Spearman correlation coefficients (r s) were calculated to assess the relationship between step counts and questionnaire-based PA. The same technique was used to assess the correlations between PA data derived from the different questionnaires. Correlations were interpreted as low (<0·30), moderate (0·30–0·50) or high (>0·50). Differences in questionnaire-based PA between the five step count groups based on the pedometer indices(Reference Tudor-Locke and Bassett5) were evaluated with multivariate ANOVA. First, the skewed questionnaire outcomes were log-transformed to approximate normal distributions. Parametric analyses were performed on the log-transformed data and adjusted for age, gender and employment status. F values, P values and partial η 2, as a measure of the effect size, are reported. For reasons of clarity and comparability, the means and standard deviations of the non-transformed data are used in Table 3. Finally, cross-tabulations of the number of participants (not) reaching the step count thresholds of 7500, 10 000 and 12 500 steps/d and/or the guideline of 30 min MVPA/d were performed, and the percentages of agreement between the different PA recommendations were calculated based on the cross-tabulations. A P value ≤0·05 was considered as statistically significant.
Results
The correlations between step counts and questionnaire-based PA are shown in Table 2. Step counts correlated moderately with total PA (r s = 0·37) and MPA (r s = 0.31), and lowly with walking (r s = 0·19) and VPA (r s = 0·25), reported in the long-form IPAQ. Low correlations were found between step counts and total PA (r s = 0·28), walking (r s = 0·15) and VPA (r s = 0·20) reported in the short-form IPAQ, except for MPA (r s = 0·33). A low correlation was found between step counts and walking (r s = 0·10) and VPA (r s = 0·16) reported in the MLTPAQ; total PA (r s = 0·32) and MPA (r s = 0·32) reported in the MLTPAQ correlated moderately with step counts. Finally, step counts correlated moderately with the total PA index (r s = 0·44), LT index (r s = 0·44) and sport index (r s = 0·31) of the Baecke questionnaire, except for the work index (r s = 0·19). Correlations between PA reported in the different questionnaires are also presented in Table 2.
Table 2 Correlation matrix: Spearman correlation coefficient (95% CI)
IPAQ, International Physical Activity Questionnaire (short-form, short version of the questionnaire; long-form, long version of the questionnaire); MLTPAQ, Minnesota Leisure Time Physical Activity Questionnaire; PA, physical activity; MPA, moderate physical activity; VPA, vigorous physical activity LT, leisure time.
†In min.
Table 3 shows the mean (sd) amounts of questionnaire-based PA (min/week) for the different step count groups divided according to the pedometer indices of Tudor-Locke and Bassett(Reference Tudor-Locke and Bassett5). Significant differences between the five step count groups were found for total PA, MPA and VPA reported in the long-form IPAQ, short-form IPAQ and MLTPAQ. For walking, only data based on the short-form IPAQ differed significantly between the step count groups. The work, leisure-time, sport and total PA indices of the Baecke questionnaire differed significantly between the five step count groups.
Table 3 Differences in questionnaire-based physical activity (PA) across step count groups: healthy adult volunteers, Flanders, Belgium
IPAQ, International Physical Activity Questionnaire (long-form, long version of the questionnaire; short-form, short version of the questionnaire); MLTPAQ, Minnesota Leisure Time Physical Activity Questionnaire; MPA, moderate physical activity; VPA, vigorous physical activity.
Significance of differences between the five groups: *P ≤ 0·05, **P ≤ 0·01, ***P ≤ 0·001 (NS, P > 0·05).
†Some post hoc pairwise comparisons were not significantly different: a, sedentary–low active NS; b, sedentary–somewhat active NS; c, low active–somewhat active NS; d, low active–active NS; e, low active–highly active NS; f, somewhat active–active NS; g, somewhat active–highly active NS; h, active–highly active NS.
In total, 80·6 % of participants reached 7500 steps/d, 45·0 % reached 10 000 steps/d and 39·4 % reached 12 500 steps/d, whereas 30 min MVPA/d was reached by 85·4 % according to the long-form IPAQ, 84·8 % according to the short-form IPAQ and 68·1 % according to the MLTPAQ (Fig. 1).
Fig. 1 Percentage of participants reaching the step count of 7500 steps/d, 10 000 steps/d and 12 500 steps/d, and the standard of 30 min of moderate to vigorous physical activity (MVPA) per day, as assessed by three different questionnaires: the long version and short version of the International Physical Activity Questionnaire (long-form IPAQ, short-form IPAQ) and the Minnesota Leisure Time Physical Activity Questionnaire (MLTPAQ)
Using the long-form IPAQ, the step count thresholds of 7500, 10 000 and 12 500 steps/d and the 30 min MVPA/d recommendation were in agreement in respectively 77·7 %, 54·4 % and 48·7 % of the participants (see Table 4). Based on the short-form IPAQ, respectively 76·5 %, 51·8 % and 48·1 % of participants were well placed. Using the MLTPAQ, the recommendations were in agreement in respectively 71·9 %, 53·1 % and 51·9 % of the participants. Some 89·2 % of the participants reaching 7500 steps/d, 94·2 % of those reaching 10 000 steps/d and 93·4 % of those reaching 12 500 steps/d reached 30 min MVPA/d, according to the long-form IPAQ. With the short-form IPAQ these figures were respectively 88·0 %, 90·6 % and 91·8 %. Of the participants, 74·8 % reaching 7500 steps/d, 73·4 % reaching 10 000 steps/d and 75·4 % reaching 12 500 steps/d reached 30 min MVPA/d based on the MLTPAQ.
Table 4 Number of participants reaching both, one or neither of two types of physical activity recommendations: 7500 steps/d, 10 000 steps/d, 12 500 steps/d and the 30 min of moderate to vigorous physical activity (MVPA) per day standard, as assessed by three different questionnaires
IPAQ, International Physical Activity Questionnaire (long-form, long version of the questionnaire; short-form, short version of the questionnaire); MLTPAQ, Minnesota Leisure Time Physical Activity Questionnaire.
Discussion
The first aim of the present study was to evaluate the associations between step counts and questionnaire-based PA. Objectively measured step counts correlated positively with the subjectively reported PA levels. Significant, positive associations between step counts and questionnaire-based PA were also found in previous studies(Reference Welk, Differding, Thompson, Blair, Dziura and Hart9, Reference Bassett, Schneider and Huntington31) conducted in the USA, where walking is one of the most commonly reported forms of PA(Reference Crespo, Keteyian, Heath and Sempos32). A median correlation of r = 0·33 (range: 0·02–0·94) was found between step counts and self-reported PA in a review of mostly non-European studies(Reference Tudor-Locke, Williams, Reis and Pluto33). Surprisingly, in the present study, a low correlation was found between step counts and walking reported in the long-form IPAQ, short-form IPAQ and MLTPAQ. An explanation for this remarkable finding could be the lack of sensitivity of questionnaires(Reference Ainsworth, Leon, Richardson, Jacobs and Paffenbarger34) to detect walking, resulting in under-reporting of walking behaviour. Bassett et al.(Reference Bassett, Cureton and Ainsworth35) also found that subjects underestimated their daily walking distance in the College Alumnus Questionnaire, compared with pedometer-based values. Furthermore, different types of walking were assessed through the various questionnaires. The long-form IPAQ asked about the combination of walking in various domains (i.e. work, transport, home, leisure time) while the MLTPAQ assessed leisure/pleasure walking and walking to work only. The pedometer measured all ambulatory activity throughout the day.
Furthermore, differences in questionnaire-based PA between groups based on the pedometer indices of Tudor-Locke and Bassett(Reference Tudor-Locke and Bassett5) were evaluated. Results showed that the classification of activity levels based on step counts also highlighted differences in self-reported levels of PA (low to medium effect sizes). The more active participants were, based on step counts, the higher the levels of total, MPA and VPA reported in the long-form IPAQ, short-form IPAQ and MLTPAQ. The application of these findings is that individuals who engaged more in MPA and VPA accumulated more steps per day. Again, unexpectedly, this was not the case for walking reported in the long-form IPAQ and MLTPAQ. Only walking in the short-form IPAQ differed significantly between step count groups. All indices (work, leisure-time, sport and total PA) based on the Baecke questionnaire also increased gradually when step counts augmented. Concluding, step counts are capable of discriminating between total, MPA and VPA reported in the different questionnaires.
A third objective of the present study was to compare currently used PA guidelines, to show differences in the percentages of participants reaching the step count thresholds. Of the participants, 80·6 % reached 7500 steps/d, 45·0 % reached 10 000 steps/d and 39·4 % reached 12 500 steps/d. The guideline of 30 min MVPA/d, on the other hand, was reached by 85·4 % according to the long-form IPAQ, followed by 84·8 % according to the short-form IPAQ and by 68·1 % according to the MLTPAQ. There are various possible explanations for this discrepancy. It is known that over-reporting of PA may occur when using self-reported questionnaires, including the IPAQ(Reference Sallis and Saelens7). The lower percentage according to the MLTPAQ could be explained by the 1-year time frame of the questionnaire, which could cause recall biases. Also the fact that the MLTPAQ was interviewer-administered may have had an influence: research assistants were trained to detect over-reporting of PA(Reference Montoye, Kemper, Saris and Washburn6).
The current study also explored whether reaching different step count thresholds was sufficient to reach 30 min MVPA/d. When participants reached 7500 steps/d, the MVPA recommendation was reached in 89·2 % (by long-form IPAQ; 88·0 % by short-form IPAQ). Of participants, 94·2 % (by long-form IPAQ; 90·6 % by short-form IPAQ) of those reaching 10 000 steps/d reached 30 min MVPA/d. Participants reaching 12 500 steps/d reached 30 min MVPA/d in 93·4 % (by long-form IPAQ; 91·8 % by short-form IPAQ) of cases. All figures based on the MLTPAQ were lower (respectively 74·8 %, 73·4 % and 75·4 %). However, these results indicated that step count thresholds are stringent enough. Le Masurier et al.(Reference Le Masurier, Sidman and Corbin19) found that 91 % of women reaching the 10 000 steps/d standard accumulated more than 30 min MVPA/d based on accelerometer data. In the present study, it was easier for participants reaching the step count thresholds to reach 30 min MVPA/d, rather than the other way around. No more than half of the participants (respectively 49·6 %, 48·1 % and 48·6 % based on the long-form IPAQ, short-form IPAQ and MLTPAQ) reaching 30 min MVPA/d reached 10 000 steps/d, for example. Welk et al.(Reference Welk, Differding, Thompson, Blair, Dziura and Hart9) reported that participants who performed more than 30 min of PA daily reached at least 10 000 steps/d some 73 % of the time. Tudor-Locke et al.(Reference Tudor-Locke, Sisson, Collova, Lee and Swan36) found that approximately 3000 steps are expected in 30 min of moderate-intensity ambulatory activity.
The present average step count level (12 100 steps/d) is clearly higher than that of a representative sample of Belgian adults (9700 steps/d)(Reference De Cocker, Cardon and De Bourdeaudhuij23). A possible explanation could be the fact that the present study population is a convenience sample of mostly working volunteers, implying cautious interpretations and limited generalizability of the results.
The present results reflect that caution is needed when assessing PA standards, especially when using different methods. It is remarkable that despite moderate correlations between step counts and questionnaire-based PA, the percentages reaching the standards differ notably. The subjective nature of questionnaires, possibly causing over-reporting, can be an explanation of the discrepancy. To avoid this problem, a gold standard, such as doubly labelled water or accelerometers, could be used to determine if the 30 min MVPA/d guideline is effectively reached.
In summary, there is a modest relationship between step counts and questionnaire-based PA in the present European sample. Objective pedometers not only provide a measurement of walking but also give an indication of total, MPA and VPA. Even though pedometers alone cannot discriminate between the intensity of activities nor reflect the amount of time spent in specific intensity PA categories, they provide sufficient information to be valuable in PA assessment in large, free-living populations. Less agreement was found between the currently used PA guidelines (30 min MVPA/d and step count thresholds), suggesting that caution is needed when comparing active quantities in different samples based on different methods.
Acknowledgements
The authors declare that they have no competing interests. The study was not funded.
K.A.D.C. participated in the data collection, analysis and interpretation, and led the writing of the paper. G.M.C. and I.M.D.B. participated in the data collection, conceptualization of the study and provided substantive feedback on the manuscript. All authors have read and approved the final manuscript.
The authors would like to acknowledge all subjects participating in the study.
