Our results clearly show that children in a Japanese elementary school had lower step counts (P < 0.001), TEE (P = 0.011), AEE, (P < 0.001), and PAL (P < 0.001) in 2009 than 1 decade earlier and spent less time playing outdoors (P < 0.001) and walking to school (P = 0.043). Although the prevalence of overweight and obesity was not significantly higher in 2009 than in 1999, the average BMI was significantly higher in 2009 (P = 0.024).

In 2004, Tudor-Locke et al. [23] recommended 15,000 and 12,000 steps per day for boys and girls, respectively. In the present study, 90.3 % of boys and 82.5 % of girls in 1999 reached these recommended step counts. In 2009, the percentages of boys and girls who achieved the recommended step counts were only 15.7 % and 40.8 %, respectively—significantly lower than the 1999 values (P < 0.001). The step counts in 2009 in the present study are quite similar to previous results by McCormack et al. [2] in 2007 and Nemoto et al. [4] in 2009 examination. In contrast, the step counts in 1999 in the present study were 20,832 ± 4694 and 16,087 ± 4840 for boys and girls, respectively. These values are relatively higher than previous reports by Rowlands et al. [1] in late 1990s and Kito et al. [3] in early 1980s. A possible reason for this is that the elementary school examined in the present study is located at the boundary of an urban area and serves a geographically larger school district than typical Japanese urban schools. The children had been instructed to walk to school even their house is over 2 km apart from school. However, the increased perception in parents for risks of accidents during walk to school only children, some parents drive their children partway recently. The decrease in the amount of time spent walking to school over the 10-year study period was mostly due to an increase in the proportion of children being driven partway by parents in the morning. Previous studies have also found a decrease in active transportation to school in Switzerland [24] and the USA [25]. Reasons that have been suggested for this trend include increased distances to school and, where distances have not increased, reduced availability of bikes and increased numbers of cars per household. Additionally, increased awareness of, and anxiety about, motor vehicle accidents that may occur while walking to school in Japan may also play a role in the tendency of parents to drive their children to school.

Cooper et al. [26] reported that children who walked to school had 43 % higher mean accelerometer counts per minute before school than those traveling by car. Loucaides and Jago [27] reported that children who walked to school had higher step counts during the before- and after-school period, and the whole day, in comparison to children who used motorized transport. It has been reported that rural children in Japan perform less PA, and especially spend less time walking to school, than urban children [5]. Thus, walking to school is an important factor in determining daily physical activity, and increasing the number of children who walk to school could be a major strategy for promoting childhood health.

Tudor-Locke et al. [28] reported in 2006 that boys took significantly more total steps per day than girls, including more steps during recess, lunchtime, and after-school. Girls took more steps than boys before school, but boys and girls took the same number of steps during structured physical education classes. Lunchtime PA represented the most important source of daily PA (15–16 %) obtained during school hours for both boys and girls, whereas recess accounted for 8–9 % and physical education class accounted for 8–11 % of the students’ total steps per day. However, almost half of the daily steps taken by schoolchildren are attributable to after-school activities. Nemoto et al. [4] reported that boys had higher step counts and PA than girls, and that this was associated with the amount of time spent playing outdoors. In the present study, we found that boys took significantly more total steps per day, and had greater TEE and PAEE, than girls. Our findings indicate that in 2009, children spent less time playing outdoors and walking to school than did the children in 1999, and the time spent playing outdoors was significantly correlated with the PA measured by the accelerometer; these findings are similar to those of previous studies. Haug et al. [29] reported that students with access to more outdoor facilities were three times more likely to be active during recess than students with access to fewer facilities. Thus, physical environment may be a factor that determines the PA of schoolchildren, and altering the environment may be one way to increase PA on a population level.

It is important to note that body composition, such as percent body fat or skeletal and muscle development, should be examined not only BMI. Scheffler, Rietsch, and their colleagues examined secular trends of height, BMI, and external skeletal robustness (Frame-Index or pelvic breadth) and the relationship between PA and BMI, percent body fat, and Frame-Index [30–34]. That means that two variables correlate with BMI—a fat content and a skeletal part. This indicates that BMI is not merely high due to a high body fat content but rather due to high skeletal part. Low physical activity can cause the reduction of external skeletal robustness, the increase of BMI and body fat. They demonstrate an association between external skeletal robustness and PA, which is not captured in BMI measurements. We found significantly but weak correlations between BMI and step counts, duration of sleeping, and bedtime. We did not examine any body composition in this study, and this is a limitation of this study.

Sigmundova et al. [35] reported that prevalence of overweight and obesity in Czech adolescents increased from 5.5 % in 1998–2000 to 10.4 % in 2008–2010, step count in weekdays, but not in weekends, significantly decreased, and self-reported walking and aerobic exercise duration decreased between those two cohorts. Sigmund et al. [36] reported similar trend of increased obesity prevalence and decreased moderate-to-vigorous intensity PA between 2002 and 2014 in Czech adolescents. Booth et al. [37] reviewed previous researches and found mixed results, and inconsistent magnitudes of change were identified when exploring trends in different contexts of physical activity for children and adolescents over the last few decades. Taken overall, organized sport participation was increased: consistent declines in active transport were found in all of seven studies investigating active transport. Because few studies investigated the secular trend of PA or step count using objective methods, they stated that need more evidence to conclude. Our finding about the secular trends of decreasing objectively measured step counts and PA in Japanese children supports the importance of PA to their healthy development. As Booth et al. [37] found from the review, consistent declines in active transport highlight this context as a suitable intervention target.

The information technology environment has changed markedly in recent years, and these changes have transformed children’s play in daily life. Nelson et al. [38] reported that secular trends further indicated dramatic increases in mid-adolescent weekly computer use from 1999 to 2004; from 8.8 to 11.1 h, and 10.4 to 15.2 h in girls and boys, respectively. We could not examine the effect of computer or smartphone use on children’s PA in the present study, and further research is needed in this area. Another limitation of the present study was that all of the participants were from just one elementary school, and therefore, the results may not be applicable to all populations in Japan. However, the decrease in step counts that we observed is similar to that reported in adults in a recent government report [39]. In addition, we did not examine activity in weekends, organized sport participation, as well as body composition, maturational status, and height and weight growth velocity [40]. As another potential confounders, we did not examine the effect of climate and weather, or SES and education status of parents on children’s PA. Further, large-scale researches are needed to examine these issues in more depth using random sampling.