Using active modes of transportation (e.g., walking, bicycling) for commuting to school has been suggested as a way to increase physical activity in children and youth (²²). A number of organizations, including the U. S. Centers for Disease Control and Prevention, provide promotional materials to implement active commuting strategies in community settings (Partnership for a Walkable America. Available at: http://www.walkableamerica; org/checklist-walkability.pdf. Accessed on: November 2, 2004, Centers for Disease Control and Prevention. Kidswalk-to-School: A Guide to Promote Walking to School. Available at: http://www.cdc.gov/nccdphp/dnpa/kidswalk/pdf/kidswalk.pdf. Accessed on: November 2, 2004) (²). These programs are consistent with the recommendation that children should accumulate at least 60 min of moderate-to-vigorous physical activity (MVPA) each day (¹). Incorporating physical activity into a daily routine (e.g., through active commuting) may produce long-term maintenance of the activity (¹²) and provide the basis for an active adult lifestyle.

It is unclear how much additional physical activity, if any, can be expected from initiatives that promote walking to school in U.S. children. Children who walk to school would seemingly be more active during the morning and afternoon compared with those riding in a bus or car. Active commuters, however, may compensate for their commuting activity by being less active at other times of the day, resulting in no net increase in physical activity. Using pedometers, Tudor-Locke et al. (²³) have shown that Russian children who did not walk to school were less likely to meet physical activity recommendations, compared with those that did. Also, Cooper et al. (⁶) reported greater physical activity levels of British school children walking to school at least once per week compared with those who always rode in a car. Data are lacking on the possible physical activity differences between U.S. children who walk to school and those who get driven. This information is needed to understand the potential impact of walk to school programs on children's physical activity levels.

This study was conducted to determine if fifth-grade students who walked or bicycled to school on a regular basis were more physically active than those who did not. A secondary purpose was to determine if active commuters compensated for their increased commuting activity by being less active during school or in the evening.

DESIGN

This was a cross-sectional study of active commuting to school and physical activity in fifth-grade students from eight randomly selected elementary schools. For 7 d, students wore an ActiGraph physical activity monitor (Model 7164, Manufacturing Technologies, Inc., Fort Walton Beach, FL) and completed a daily self-report survey instrument to identify their mode of transportation for commuting to and from school. Based on these student reports, children were categorized as those who walked or bicycled on at least five commutes per week, one to four commutes per week, and those who rode a car or bus for every commute during the week of data collection.

METHODS

Schools and subjects.

A stratified random sampling scheme was used to recruit four schools each (total of eight schools) from the 16 urban and 16 suburban public elementary schools in Columbia, South Carolina. The level of urbanization (urban, suburban) for each school was identified from the National Center for Education Statistics (Institute of Education Sciences. National Center for Education Statistics. Available at: http://nces.ed.gov/. Accessed on: May 20, 2002). If a selected school refused to participate, a randomly selected replacement school from the same urbanization stratum was recruited. Total enrollment at the eight schools ranged from 229 to 723 students (mean ± SD, 489 ± 166). Compared with suburban schools, urban schools were slightly smaller (suburban: 545 ± 111 vs urban: 433 ± 212), had fewer minorities (suburban: 78.5% vs urban: 65.2%), and fewer students receiving free or reduced school lunch (suburban: 71.7% vs urban: 59.8%).

All fifth-grade students at these schools (N = 636) were invited to participate. The fifth-grades ranged in size from 35 to 137 (74 ± 29) students. In suburban schools, the size of the fifth-grade ranged from 57 to 111 students and from 35 to 137 students in urban schools. Compared with suburban schools, the size of the fifth-grade in urban schools was similar (suburban: 79 ± 34 vs urban: 77 ± 45), but had fewer minorities (suburban: 79.3% vs urban: 68.4%), and fewer students receiving free or reduced school lunch (suburban: 57.4% vs urban: 51.9%).

Only fifth-grade students were recruited for this study because self-report surveys have not been recommended for use in children younger than 10 yr of age (¹⁶) and the fifth-grade was the highest grade level at these local elementary schools. Data related to only one child per household is included. A total of 237 fifth-graders returned a signed consent form. Because of subject withdrawal (N = 5), monitor failure (N = 2), lost monitors (N = 3), and students being absent on the day of monitor distribution (N = 8), a total of 219 subjects were retained for analysis. This study was approved by the University of South Carolina's institutional review board, the school districts involved, and by the administration at each participating school.

Student survey.

To obtain daily commuting patterns, students completed a one-page survey instrument each day for five consecutive school days. Students were asked, “How did you get to school today?” The response options were a follows: school bus, city bus, parent's car, other car (neighbor, relative), walk and take the bus on a single trip, walk and get driven on a single trip, bicycle, walk, and other. On every day except Monday, they were also asked to recall where they went directly after school the day before using the following response options: home, friend's house, relative's house, park, after school program, errands with parent or other adult, sports practice, stay at school, and other) and the mode of transportation used to get to that location (same options as for the commute to school). Information about the morning commute was obtained on all 5 d, whereas specifics about the afternoon commute were obtained for Monday through Thursday only. Friday afternoon information was not collected because the children would have had to recall this information on the following Monday morning and this was considered an unacceptable delay.

Physical activity.

The ActiGraph physical activity monitor, used as an objective measure of physical activity, has been validated for use with children in laboratory and field settings (^10,13,21). It is a small (5.1 × 3.8 × 1.5 cm), lightweight (42.6 g), single plane (vertical) accelerometer that collects and stores accelerations from 0.05 to 2.00 g with a frequency response of 0.25–2.50 Hz. These settings capture normal human motion but will filter out high frequency vibrations (e.g., operating a lawn mower) or from mechanical sources (⁵). The analog acceleration is filtered and converted to a digital signal and this value (count) is stored in user-specified time intervals: 1-min intervals were used for this study. ActiGraph monitors were initialized for a common time for all children at a particular school, although this could not be standardized across schools because of different school schedules and administrator decisions. After data collection, each monitor was downloaded to a computer for subsequent data reduction and analysis.

Procedures.

On the first day of the study, height and weight were measured to the nearest 0.1 cm using portable height boards (Shorr Productions; Olney, MD) and to the nearest 0.1 kg using calibrated digital scales (BeFour, Inc. Model PS6600; Saukville, WI). On the first day of survey administration, research assistants explained the survey and read each question aloud to the students to ensure comprehension. After the initial administration, homeroom teachers kept the surveys in their classroom and prompted students to complete the survey during homeroom on each subsequent day.

Subjects wore the ActiGraph monitor for seven consecutive days, beginning with the first day of survey administration. The first day of the study was not standardized across schools. Therefore, students at one school may have started the study on Monday, whereas students at another school started on Wednesday. The difference in start dates was not a concern because a previous study did not detect any reactivity to sealed pedometers in similarly aged children (²⁴). At monitor distribution, a research assistant fit an elastic belt with an attached monitor to each student. The children were told not to adjust the belt once it was fitted. Students were given written and verbal instructions on the use and care of the monitors. They were instructed to wear the monitor during all waking hours, except when swimming, bathing, or sleeping. Telephone calls were made to the students at the midpoint of the week and on the day before the monitor was due to be collected as reminders to wear the monitor and to answer any questions.

Survey data reduction.

Students were categorized based on their reports of how they traveled to and from school. Students reporting 5 or more, 1–4, or 0 walking or bicycling school commutes were categorized as 1) regular active commuters, 2) irregular active commuters, or 3) nonactive commuters, respectively. Students were also categorized as white or minority. The minority group was composed mostly of African-American students (95%), but also included Hispanic, Asian or Pacific Islander, multiracial, and other.

ActiGraph data reduction.

ActiGraph data were reduced using a custom-developed software program. All data contained within the time frame from when the monitor was initialized until the same time the following week (end time) were processed. For days 2–7, all data from 5:00 a.m. until midnight were reduced to summary variables. Day 1 and day 8 were combined to form a composite seventh day of data. Day 1 consisted of data from when the monitor was initialized until midnight of that day. Day 8 consisted of data from 5:00 a.m. until the end time.

Daily inclusion criteria were established to determine days and times with acceptable accelerometer data. Blocks of time incorporating at least 20 continuous minutes of “0” output from the ActiGraph were considered to be times when the subject was not wearing the monitor. These data points were eliminated and not used in any calculations. Also, days with less than 6 h of data were eliminated from data reduction to account for unrepresentative days of activity.

The reduced data were placed into separate data sets (usual, weekdays, weekend). Previously, it has been shown that 4 d of activity monitoring are needed to provide a reliable estimate (R = 0.80) of usual physical activity in similarly aged children (¹⁹). Students with at least 4 of 7 d of ActiGraph data were retained for the usual data set. The weekday data set contained reduced data for all weekdays meeting inclusion criteria (3–5 d). For the weekend data set, students needed to have at least one weekend day that met the daily inclusion criteria (Table 1).

To investigate daily patterns of activity, each weekday was divided into four time blocks: before school (5:00 a.m.–8:00 am), during school (8:00 a.m.–dismissal), after school (dismissal–6:00 p.m.) and evening (6:00 p.m.–10:00 p.m.). Duration of during and after school time blocks varied because the duration of the school day varied by up to 30 min. The appropriate school schedule was applied to each student's ActiGraph data to account for these differences. Inclusion criteria were also needed for these time blocks, because a student may have met the daily inclusion criteria but was missing data from one of the specific time blocks. For example, a student may have had no data for the before school time block but acceptable data for the rest of the day. To be included in each time block analysis, students needed to have acceptable ActiGraph data (nonconsecutive zeros) for at least 1) 30 min before school, 2) half of the total minutes in the during and after school time blocks, and 3) 2 h for the evening time block.

After processing the data through the inclusion criteria, two outcome variables were calculated for each subject. Several ways exist to measure levels of MVPA: calculate 1) the total number of minutes spent in MVPA over the measurement time period, or 2) the average number of minutes per day spent in MVPA. Both of these approaches assume that every subject wore the monitor for the same number of minutes each day. Because this is typically not the case, both summary variables were calculated relative to the total number of minutes the monitor was worn. For the following descriptions, “included days or time blocks” refers to days or time blocks meeting the inclusion criteria for the time period being analyzed (usual, weekdays, weekends, and weekday time blocks). Average ActiGraph counts per minute was calculated as the total counts for all included days summed and divided by the total number of minutes the monitor was worn for all included days. The average percent of time spent in moderate-to-vigorous physical activity was calculated using age-specific count cutoffs (e.g., ≥ 3.0 METs; ≥ 1017 counts per minute for a 10-yr-old; (¹¹). For each day of data meeting the inclusion criteria, the number of minutes at or above the MVPA count cutoff was divided by the total number of minutes the ActiGraph was worn on that day to provide the percent of that day spent in MVPA. These daily percent values were then summed and divided by the number of included days to obtain the average percent of time spent in MVPA across the time period. Analogous variables were calculated for each of the four weekday time blocks (e.g., average counts per minute before school, average percent of time spent in MVPA before school).

Data analysis.

For each of the weekly time periods (usual, weekdays, weekend, weekday time blocks), mixed model ANOVA were used to identify group differences in ActiGraph summary variables while controlling for age, gender, race, and urbanization category of the school. School was also included as a random factor in all models to account for cluster sampling. A model with all two-way interactions that included group (regular, irregular, nonactive) and one of the controlling variables were analyzed first. No significant interactions were detected, so the model was reanalyzed with only main effects. Because of the different inclusion criteria applied to the data sets (all days, weekdays, weekend, weekday time blocks), the sample size for each analysis is slightly different, ranging from N = 161 to 209. Each analysis calculated group differences within each of these time frames. Comparisons of activity between different time periods (e.g., weekday vs weekend physical activity) are not presented in this study. Because several of the dependent variables were nonnormally distributed, they were log transformed for analysis, but nontransformed means and standard deviations are reported.

RESULTS

Subject characteristics for all students who provided total or partial physical activity data are provided in Table 2. Most (84%) were nonactive commuters and only 5% of the students reported walking or bicycling to school on a regular basis. Only three students selected the multimode travel option, “walk and get driven on a single trip.” All three of these students were excluded from the analysis because of incomplete questionnaire, ActiGraph data, or both. The percent of males was slightly lower in the regular group compared with the others and 73% of the regular active commuters went to schools categorized as urban compared with 46 and 40% for the irregular and nonactive groups, respectively (χ²_{df = 1} = 3.85, P = 0.05). No significant group differences were detected for body mass index (BMI) or the percent classified as overweight based on the 85th percentile from the age- and gender-specific BMI charts from the Centers for Disease Control and Prevention.

Across all days and weekdays, average ActiGraph counts per minute was approximately 15% greater (94 counts per minute) for regular active commuters, compared with those in the irregular and nonactive commuting groups (P ≤ 0.04; Fig. 1). Because sample sizes varied for each analysis, group sizes are provided in the appropriate columns in Figure 1 and all subsequent figures. For the weekend, no significant difference among the groups was detected. For the weekday time blocks, average counts per minute for regular active commuters was approximately 33% greater before school and 27% greater after school, compared with irregular and nonactive commuting groups (P ≤ 0.01; Fig. 2). No group differences for average counts per minute were detected during school or in the evening.

Similar results were obtained using the percent of time spent in moderate-to-vigorous physical activity as the dependent variable. Across all days and weekdays, average time spent in MVPA was approximately 3% greater for regular active commuters, compared with those in the irregular and nonactive commuting groups (P ≤ 0.04; Fig. 3). For the weekend, no significant difference among the groups was detected. For the weekday time blocks, average time spent in MVPA for regular active commuters was approximately 8.5% greater before school and 8.5% greater after school, compared with irregular and nonactive commuting groups (P ≤ 0.01; Fig. 4). No group differences were detected during school or in the evening.

DISCUSSION

This study examined the association between levels of active commuting to school and objectively measured physical activity in fifth-grade students. Those who actively commuted to or from school on a regular basis (at least five times per week) were more physically active on all days and weekdays, compared with irregular and nonactive commuters. The 3% difference in the percent of time spent in MVPA translates to approximately 24 additional minutes of MVPA per day for the regular active commuting students (on average, students wore the ActiGraph monitor for 800 min across all days and weekdays). The greater daily physical activity level of the regular active commuters was a result of their greater activity levels before and after school. Performing similar calculations as above, regular active commuters accumulated, approximately, an additional 7 min of MVPA before school (8.5% of 85 min) and 18 min of MVPA after school (8.5% of 215 min).

The values calculated above are approximations based on the average number of minutes the monitor was worn and the average difference between the regular group and the others. At the individual level, the additional physical activity associated with active commuting would be proportional to the distance and duration of those commutes. The additional activity, therefore, associated with regular active commuting may be minimal for a child who lives across the street from the school but substantial for those living further away. On the student survey, children were asked how many minutes each commute required. The response options (in minutes) for commute duration were 1–5, 6–10, 11–15, 16–20, 21–30, or more than 30. These responses were used to estimate the proportion of the regular group's additional minutes of MVPA that could be attributed to active commuting behavior. Nine regular group participants had complete data for this post hoc analysis; five students most frequently reported 1–5 min, three students reported 6–10 min, and one student reported 10–15 min of active commuting. The distribution of responses was the same for the commutes to and from school. Based on these responses, we chose 5 min as an estimated duration for all active commuting trips. Assuming that each active commute was performed at a moderate or greater intensity, this would account for 71% (5 of 7 min) and 28% (5 of 18 min) of the additional MVPA detected before and after school, respectively. A child who walked to and from school would accumulate about 10 additional minutes of MVPA per day directly from their commuting behavior. Although 10 additional minutes of physical activity may seem rather modest, in light of our increasingly sedentary lifestyles and high levels of pediatric obesity, any opportunity to increase children's physical activity should be utilized. Incorporating more walking into routine daily events is seen as an important goal in long-term maintenance of initial increases in physical activity (^9,15). Also, regularly walking to school may have a positive impact on other health outcomes and behaviors (e.g., social and academic performance, air quality, greater independence, and improved community health) and may promote long-term use of walking as a mode of transportation.

In addition to the duration of their school commutes, children were also asked how far they traveled to and from school. The responses to these distance questions were inconsistent and of questionable validity. Although not possible for the current study, a more sophisticated and objective approach would be the use of geographic information systems that can plot the family residence in relation to the school to objectively determine the distance from the home to the school.

The secondary aim of this study was to determine if active commuting students compensated for this additional activity by decreasing their activity at other times of the day. A previous study (⁷) indicated that students deprived of school recess did not increase their activity after school and, thus, were less active than those participating in recess. In this study, nothing indicated that physical activity decreased, either during school or in the evening, for the regular active commuters. Similar to the Dale et al. study (⁷), nothing indicated that the nonactive commuters increased their activity during school or in the evening to compensate for being driven to and from school.

The regular active commuters appeared to be more active on the weekends as well, although no statistically significant group differences were detected (ActiGraph counts per minute; P = 0.06, %MVPA; P = 0.24). The relatively small sample size (N = 121) available for the weekend analyses limited our ability to detect statistical significance, although the pattern was similar to the weekly and weekday findings. Because subjects did not commute to school on the weekends would suggest the presence of some unidentified variable(s) responsible for, at least part of, the group physical activity differences during the weekend and for the additional minutes of MVPA recorded for the regular group during weekday afternoons.

One such variable is the effect of sports team participation. In a subset of students with parent survey data (N = 147), the percent of students participating on a sports team during the previous year was similar for regular (43%) and nonactive (57%) active commuters, but greater for the irregular active commuters (81%). High sports participation in the irregular group might have been responsible for their sporadic commuting behavior because of sports practices directly after school. From the student surveys, however, none of the irregular active commuters reported going to sports practice immediately after school. This response option was rarely chosen across all subjects and likely reflects the standard routine of having sports practices and games later in the afternoon for this age group of children. Further, even though previous year sports participation was greater in the irregular active commuters, their physical activity levels were more similar to the nonactive commuters. It seems unlikely, therefore, that sports participation is responsible for the greater afternoon physical activity levels of the regular active commuters.

Another potential confounding variable is the difference in the physical environments of the regular active commuters compared with the irregular and nonactive commuters. Increasing evidence indicates that urban environments categorized by mixed land use (retail and residential), sidewalks, short block lengths, and a gridlike pattern of streets are associated with greater amounts of walking in adults (¹⁴). It remains unclear if the same associations apply to school-aged children and, more specifically, the commute to school. A large percent (73%) of the regular active commuters were from schools classified as “urban.” Urbanization level of the school was controlled for in all analyses, indicating that the group differences in physical activity were independent of this gross measure of the physical environment. More specific environmental factors and related community- and school-based policies related to the physical environment are beginning to be explored, but more research is needed to understand the effect of the environment on the mode of travel used to get to and from school.

Our results are similar to a study of British school children (⁶) that found increased physical activity (using the ActiGraph monitor) for students who reported walking to school at least once per week compared with those who always rode in a car or bus. They found that both male and female walkers accumulated more physical activity before school but only the boys accumulated more activity in the afternoon and evening. In the current study, the interaction between gender and commuting group, for either dependent variable, was not significant. Gender main effects for the differences in physical activity did exist for most analyses, with boys being more active than girls (data not presented).

Only 5% of this sample was classified as regular active commuters. In a previous study of observed school commuting behavior in these same schools (¹⁷), the prevalence of active commuting to school was also 5%. This would indicate that we may have captured a relatively large percentage of the active commuting fifth-graders at these elementary schools. Why the active commuting rates were so low in this sample is unclear. Subjectively, six of the schools were within or adjacent to residential neighborhoods with moderate to high potential for walking or bicycling to school. National (Oakridge National Laboratories. Our Nation's Travel: 1995 NPTS Early Results Report. Available at: http://npts.ornl.gov/npts/1995/doc/NPTSBooklet.pdf. Accessed on: November 2, 2004) and regional survey data (^4,18) have reported higher rates of active commuting. Such comparisons, however, must be made cautiously because the national (Oakridge National Laboratories. Our Nation's Travel: 1995 NPTS Early Results Report. Available at: http://npts.ornl.gov/npts/1995/doc/NPTS_Booklet.pdf. Accessed on: November 2, 2004) and Georgia surveys (⁴) classify children as active commuters if they report even one active commute during the previous week or month. Baseline data from the Marin County Safe Routes to School Program (¹⁸) reported approximately 13% walking and 5% bicycling to school, based on 3 d of student self-reports. Comparisons between Marin County and Columbia, SC are difficult possibly because of different cultural beliefs, education and income levels, and different data collection methodologies. The psychological, social, and economic forces that determine whether or not a child will be allowed to walk or bicycle to school is the focus of ongoing research, but beyond the scope of this paper.

Several limitations to this study require mention. The low 36.6% recruitment rate across the eight participating schools and the small samples of regular and irregular active commuters limit the generalizability of the findings and we cannot rule out the possibility of selection bias. As with all physical activity studies, the possibility exists that only the physically active children were recruited into the study. This does not seem likely, however, because numerous other studies have also reported low activity levels (which were comparable to those found in this study). Selection bias could have occurred if the regular active commuters participating in our study were also more active than other regular active commuters who were not in this study. No information was available for nonparticipants, however. Alternatively, selection bias may have improved our recruitment of active commuters if they believed that this was a study specifically about them.

In the current sample, minority and students from low socioeconomic status families were under represented. Our sample, therefore, may not accurately represent the physical activity and commuting patterns of all fifth-grade students in the Columbia, SC area. The mean ActiGraph counts per day and minutes of MVPA per day accumulated by our subjects, however, are comparable to those from other studies (^19,20). Also, even if the sample were representative of the Columbia area, our findings could not be generalized to the U.S. population, which was not an aim of this study. Only fifth-grade students (ages 10–11 yr) are represented in the current study. Fifth-grade students were the focus of this study because 1) it is the highest grade at area elementary schools, and 2) these students live closer to their school compared with middle- or high-school students and, thus, have a greater opportunity to walk to school. Although parents may be more willing to allow an older child (6th–12th grade) to walk to school, two previous studies did not detect differences in the prevalence of active commuting to school between primary and secondary school children (^3,8). Significant declines in physical activity are seen as children progress through adolescence (²⁰) and the impact of active commuting on this decline is unknown. Lastly, the activity levels of bicycling students may have been underestimated because of the inability of the ActiGraph to detect nonvertical accelerations when placed on the individual's hip. The effect of this under representation is likely minor, because the prevalence of bicycling commutes was very small (<1%).

Community-based efforts to increase walking and bicycling to school are gaining support from a number of organizations (Partnership for a Walkable America. Available at: http://www.walkableamerica.org/checklistwalkability.pdf. Accessed on: November 2, 2004, Centers for Disease Control and Prevention. Kidswalk-to-School: A Guide to Promote Walking to School. Available at: http://www.cdc.gov/nccdphp/dnpa/kidswalk/pdf/kidswalk.pdf. Accessed on: November 2, 2004) (²), and one of the objectives of these programs is to increase the physical activity levels of children. It may seem obvious that incorporating active transportation into a child's daily routine would increase physical activity (²³). Because of the cross-sectional design, we cannot establish whether the active commuting to school caused an increase in physical activity or if the children who are more active chose to actively commute. Longitudinal studies are needed to determine the effect of active commuting on current and future physical activity levels, other health outcomes, and attitudes, beliefs, and enjoyment of overall physical activity. Experimental interventions are needed to determine the change in children's physical activity resulting from the incorporation of active commuting to school, alone or in combination with other school- and community-based programs. Future research in this area will require larger, more diverse samples and should include different grade levels to explore age-related differences in active commuting and physical activity. These future studies can also explore the effect of walking to school on other health, behavioral, and environmental outcomes (e.g., weight status, social or academic development, and air quality).

In conclusion, active commuting to and from school at least five times per week was associated with approximately 24 additional minutes of moderate-to-vigorous physical activity, compared with irregular or nonactive commuters. The higher physical activity in the regular active commuters was owing to greater physical activity before and after school with no decrease in activity recorded during school or in the evening. Additional observational and experimental research on the effect of active commuting to school on physical activity and other health outcomes is recommended.