Baseline survey showed that 85.0 % of the students had moderate (50.7 %) and high (34.3 %) levels of activity. There was no statistical significant difference in physical activity levels among the three groups at week 1, week 8, and week 15 (p = 0.473, p = 0.893, and p = 0.077, respectively). Moreover, there was no significant increase in reported physical activity within each group at week 1, 8, and 15 (p = 0.469, p = 0.222, and p = 0.289, respectively).
Fitbit Flex Data
There was no significant difference in the baseline daily step counts between students in group 1 and those in group 2 (p = 0.823). Mixed model showed that students in group 2 (Fitbit and game) walked significantly more than those in group 1 (Fitbit alone) by 17.3 % (p < 0.0001). Daily step counts in week 9 were compared among the three groups, and the difference was statistically significant (p = 0.001). Students in group 1 walked significantly less compared with those in the control group by 15.9 % (p = 0.003); however, there was no significant difference between group 2 and the control group (p = 0.456). Similarly, there was a statistically significant difference of daily step counts among the three groups from week 10 to week 14 (p = 0.006). Students in group 1 walked significantly less than those in the control group by 7.6 % (p = 0.0104), while there was no significant difference between group 2 and the control group (p = 0.427) (Table 1).
Students walked significantly more on weekdays compared with weekends in the three periods (p < 0.0001) by 27.5, 40.0, and 31.7 %, respectively. When weekdays and weekend step counts were analyzed separately from week 1 to week 8, group 1 and group 2 had a statistically significant drop in daily step counts in week 6 and 7 (the recess and exam week) compared to weeks 1 to 5. Group 1 showed a 25 % decrease in daily step counts while group 2 decreased by 16 %.
Sleep Time from the Survey
Students reported daily sleep time from 400.8 ±63.8 min (6.68 h) to 436.1 ±68.1 min (7.26 h). No statistically significant differences were found among the three groups at week 1, week 8, and week 15, and no significant differences were found in each group at week 1, week 8, and week 15 between weekdays and weekends.
Sleep Quality from the Survey
Although no statistically significant differences among the three groups were found for week 1 and week 8 (p = 0.569 and p = 0.170, respectively), week 15 data showed a significant difference in sleep quality among the three groups (p = 0.008), with 81.11 % in group 1 and 85.88 % in group 2 reporting a good quality of sleep, compared to 67.05 % in the control group.
The mean daily sleep time was summarized in Table 2. There was no significant difference in sleep time at baseline between group 1 and group 2 (p = 0.204). Mixed model showed that there was no significant difference of daily sleep time from week 2 to week 8 between group 1 and group 2 (p = 0.788). The daily sleep time at week 9 was compared among the three groups, and the result showed that there was no statistical difference either (p = 0.130). However, we observed that students in group 1 (Fitbit alone) slept 21.7 min longer than those in the control group (p = 0.073); students in group 2 (Fitbit plus game) slept 18.4 min longer than the control group (p = 0.148). During week 10 to week 14, there was significant difference of daily sleep time among the three groups (p = 0.034). Students in group 1 slept significantly longer than those in the control group (p = 0.017) by 17.5 min. Students in group 2 slept 14.4 min longer than those in the control group; however, it was just boundary significant (p = 0.064). It was observed that students always slept significantly longer during weekends compared to weekdays in the three periods (p < 0.0001) by 63.4, 59.2, and 51.2 min, respectively. By week 14, all students’ (average of all groups) sleep time reached on average 7.90 h during weekdays and 8.04 h during weekends.
Subgroup analysis showed that being able to track one’s daily sleep hours increased significantly sleep time in all three groups (p < 0.000 for all three groups). Compared to week 1, group 1 (FitBit alone) and group 2 (FitBit + game) slept significantly longer from week 5 onwards; by week 14, group 1 slept 69.85 min longer and group 2 slept 28.9 min longer (p < 0.000).
From week 9 to week 14, all students including the control group had access to FitBit Flex and the game; no significant differences in sleep time were found between group 1 and group 2 (p = 0.655). At week 9, group 1 average sleep time during the weekday was 449.23 min (7.48 h) ±SD 60.70, weekend 486.05 min (8.10 h) ±SD 87.64 while group 2 average sleep time was 429.42 min (7.15 h) ±SD 91.44 for weekdays and 471.00 min (7.81 h) ±90.35 for weekends. By week 14, group 1 increased sleep time by 15 min during weekdays, 17 min during weekends and group 2 by 13 min during the weekend and 49 min during weekdays. From weeks 9 to 14, group 1 slept 17.4 min more (p = 0.015) and group 2 13.8 min (p = 0.070) more than the control group. For the control group, between week 9 and week 14 during weekdays, sleep time increased significantly from week 12 onwards with an overall 61-min increase.
There was no statistical significant difference in stress scores among the three groups in the three data collection points (p = 0.073, p = 0.259, and p = 0.357, respectively). The survey score lower than 13 is deemed as low stress, 14–19 as medium, and above 20 as high. Table 3 summarizes the results.