Gaming behaviour with Pokémon GO and physical activity: A preliminary study with medical students in Thailand

Pokémon GO becomes the most rapidly downloaded mobile application in history. This study aimed to determine the physical activity of medical students, who played Pokémon GO, and the change in their use of Pokémon GO and physical activity over time. An observational study was conducted. Physical activity was measured by using self-administered questionnaires at baseline (phase 0), 1 month (phase 1) and 3 months (phase 2) post-Pokémon GO download. The changes in physical activity (phase 0 to 1 and phase 1 to 2) were analysed using Wilcoxon Signed Ranked test. The trend (3-point analysis) of physical activity from phase 0, 1 to 2 were analysed using Friedman’s test. The relationship between physical activity and time spent gaming was analysed by using Spearman’s rank correlation. Twenty-six participants (mean age 22.04±1.70 years) participated in the study. There was no statistically significant change in physical activity during the three-month period (p = 0.45). Only 11 participants (42.3%) were still playing Pokémon GO 3 months after download. The key reasons for playing game were ‘have fun’ and ‘pass time/boredom’. The most common commuting mode to play the game was walking; some drove a car or motorcycle while playing the game. There was no correlation between physical activity and time spent gaming. This study highlights how the lack of sustainability of the game and the motivation behind using Pokémon GO as a game rather than a physical activity app may have undermined the potential of using the game to improve physical activity. Further studies need to explore the reasons for the lack of sustainability and how to combine fun with behavioural change.


Introduction
Physical inactivity is an important health risk that contributes to 3.2 million deaths and 69.3 million disability-adjusted life year (DALYs) globally each year. [1] Moreover, it is a risk factor for major non-communicable diseases (NCDs), including coronary heart disease, type 2 diabetes, breast cancer, and colon cancer. [2] a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 To effectively increase physical activity (PA) at a population level requires multi-sectoral, multi-disciplinary multi-strategy-based approaches. [3][4][5] The expansion of mobile devices and computational intelligence has led to the development of many approaches of motivating people to participate in and monitoring their PA. [6] Therefore, the use of digital technologies has been increasingly recognised as a potentially effective intervention, to increase PA, particularly among the younger people who are leading a sedentary lifestyle. [7][8][9] For example, playing active video games has been found to increase PA levels in both children and adults [10,11] and PA mobile phone games may be an important element effecting behavioural change in PA interventions. [12] Pokémon GO is an augmented reality game which involves virtual elements that build upon the existing environment. [13] The game requires players to physically move to collect Pokémon (game characters) and walk in the real world for 2, 5 or 10 km to hatch a Pokémon egg. [14,15] Pokémon GO was first released in July 2016 and launched in Asia in August 2016. [16,17] It becomes the most rapidly downloaded mobile application in history. [18,19] In view of its PArelated game patterns and its popularity, we hypothesised that playing Pokémon GO could increase and potentially maintain PA of game players. One study has already shown that Pokémon GO significantly increased PA of game players in the first 30 days of playing the game. [20] In Thailand, physical inactivity is on the rise, particularly among young adults. [21] This correlates with the rising incidence of obesity and diabetes in the younger population. [22][23][24][25] A recent study of a group of Thai medical students found that more than half (50.5%) of them with a mean age of 20.93±1.82 were physically inactive. [26] However, motivating young people to engage in regular PA remains challenging. [27,28] This study aimed to investigate the gaming behaviour of a group of medical students in Southern Thailand who were actively playing Pokémon GO, and the change in their PA. Moreover, the present study focused on the sustainability of the game over time.

Study design
We conducted an observational study. The baseline data (phase 0) was obtained from the PA survey conducted in 2015 with medical students of Walailak University. [26] Phase 1 data was collected in September 2016, 1 month after downloading and playing Pokémon GO. The data included: demographic data, patterns of and reasons for playing Pokémon GO, game-related injuries, PA and sedentary time. Phase 2 data was collected three months after downloading the game and they were: patterns of and reasons for playing Pokémon GO, game-related injuries, PA and sedentary time. The online questionnaires were used to collect all the data in phase 1 and phase 2.

Setting and participants
The study was conducted in the three campuses of School of Medicine, Walailak University, in Southern Thailand (Nakhon Si Thammarat, Trang and Phuket). Preclinical medical students (years 1-3) studied at Nakhon Si Thammarat campus while the clinical students (years 4-6) did their clinical rotations at Trang or Phuket campus. All medical students who participated in the previous PA survey were eligible to participate in this study, except for the year 1 and 4 students. [26] The year 1 students were excluded because we did not have their baseline data at Phase 0 while the year 4 students were excluded because there was a transition from preclinical to clinical year, which was a potential confounding factor. [26] A total of 189 medical students (years 2, 3, 5 and 6) met these inclusion criteria. We asked the medical students who had played Pokémon GO at least 1 month to participate in the study. Students with disabilities or medical conditions which limited their ability of PA, including movement disorder; muscular weakness; heart disease; uncontrolled pulmonary disease; and uncontrolled medical conditions, were excluded. All the participants from phase 1 were asked to complete the questionnaire in phase 2. We reminded them through Facebook or telephone.

Research instruments
The phase 1 questionnaire consisted of three parts: participant profiles (level of education, sex, birth date, body weight and height, student's allowance, place of living and underlying illnesses); patterns of playing Pokémon GO (time spent gaming, mode of commuting while playing game, reasons for playing Pokémon GO and game-related injuries) and information on PA and sedentary time using the Global Physical Activity Questionnaire (GPAQ) version 2 (Thai version). The phase 2 questionnaire excluded the participant profiles.
We developed the questionnaires in Thai language. All the items were initially drafted in Thai except the reasons for playing Pokémon GO, which were adapted from an existing questionnaire [29]. The items were translated and back-translated by two researchers (AW and ST). The questionnaires were assessed by two external PA experts for content validation. The finalised version of the questionnaires was created online via Google Forms. A pilot study was conducted with four students to determine the face validity of the questionnaires as well as to test the study flow and recruitment strategy.

Data analysis
Participants' demographic characteristics, reasons for and patterns of playing games, and injury pattern were analysed using percentages for categorical variables, and mean with standard deviation (SD) for continuous variables. PA participation and sedentary time were measured at baseline (phase 0, non-Pokémon GO players), 1 month (phase 1, Pokémon GO players) and 3 months (phase 2, both recent and former players) post-Pokémon GO download. The PA participation, including moderate-to vigorous-intensity PA, was presented as energy expenditure using the following formula: energy expenditure = [4 MET X time of moderate PA (minutes/week)] + [8 MET X time of vigorous PA (minutes/week)]. [30] The sedentary time was adjusted and presented as minutes/day.
The changes in PA and sedentary time from phase 0 to phase 1 as well as from phase 1 to phase 2 were analysed using Wilcoxon Signed Ranked test. The trend (3-point analysis) of PA and sedentary time from phase 0, 1 to 2 were analysed using Friedman's test. The relationships between PA, sedentary time and time spent on gaming were analysed using Spearman's rank correlation. The relationship was considered statistically significant when p<0.05.

Ethics approval
Ethical approval was obtained from the Human Research Ethics Committee Walailak University (protocol number 061/2016). All medical students were above 18 years old (legal age for consent in Thailand) and their participation was entirely voluntary. Information about the study was provided on the first screen of the online questionnaire and the informed consent was taken by asking the participants to indicate their agreement to participate by ticking a checkbox.

Physical activity participation and sedentary time
In phase 0, the median energy expenditure from moderate-to vigorous-intensity PA was 600 MET-min/week (IQR 0-2700). In phase 1 and phase 2, the median energy expenditure were 996 MET-min/week (IQR 490-1770) and 960 MET-min/week (IQR 420-1470), respectively. There was no statistically significant difference in energy expenditure between phase 0 and phase 1 (p = 0.93), as well as phase 1 and phase 2 (p = 0.80). The 3-point analysis of PA from phase 0, 1 to 2 did not show any significant change (p = 0.45) (Fig 1).

Discussion
The findings from this study contradict those found in some Pokémon GO studies. The number of students who played Pokémon GO dropped significantly over three months and the game did not seem to change the participants' level and duration of physical activities. In addition, there were no correlations between time spent gaming and energy expenditure. This challenged the hypothesis that Pokémon GO was effective in improving PA among those playing the game. However, there was a significant decrease in sedentary time over three months, although it was not correlated to gaming time.
A USA study investigated the effect of Pokémon GO on PA via wearable PA trackers among 792 Pokémon GO players. The study found that Pokémon GO led to significantly increased walking by 1473 steps a day on average, over a period of 30 days. There was also a dose-response relationship between Pokémon GO and PA. [20] The same trend was found in the study by Howe et al. where Pokémon GO users recorded 955 additional steps per day during the first week of installation of the game. However, the number of daily steps dropped significantly by the sixth week after installation. [32] Studies among college students also found Pokémon GO might be an effective tool to increase PA. [33][34][35] In contrast, a cross-sectional study conducted among university students in Hong Kong did not find any significant difference in PA levels between Pokémon GO players and non-players at 4 weeks after the releasing of the game. [36] Most previous studies focused on the influence of Pokémon GO on PA over the first 4-6 weeks of playing the game. [20,32,36] While some studies found positive effects on PA within the first few weeks after playing the game, [20,32] others found the game to be ineffective in young and older adults. [36,37] Our study extended the follow up period to three months to look at the sustainability of the game as well as its effect on PA. The findings found that the game might not change the participants' PA levels among a group of university students, and the number of game players decreased significantly. This finding is important because it provides preliminary evidence on the limitation of using a game app to improve PA and that more studies need to look into the factors that influence the sustainable use of the game and how the mechanisms of behavioural change are effected via games. This observation is supported by studies that propose gaming alone might not effective; combining gaming with motivation and social support might work better. [20,36] Among the participants in this study, the main reason for playing Pokémon GO was not to exercise; only a quarter of the participants played the game for exercise which dropped even lower at the end of phase 2. At 3 months after downloading the game, some participants became the former players, meaning that they were not interested in playing Pokémon GO. Among the active players, all of them still walked to catch the Pokémon, however, they spent less time and frequency to play the game. Therefore, the increases in PA from the previous studies [20,32] might be a side effect of the game. [38] Although the players were required to travel to public places to catch Pokémon and incubate Pokémon eggs, [14,15] some players in our study travelled by taking slow-moving vehicles instead of walking. This could explain the lack of correlation between PA and time spent playing the game. In addition, although we found a significant decrease in sedentary time among the Pokémon GO players, it was not directly correlated to time spent gaming. It might be an indirect effect of the game that causing the players to stay outdoor. [36] The strengths of this study were 1) the study provided longer term evaluations of the game usage pattern and its effect on PA from baseline to three months after downloading and playing the game, and 2) it presented patterns of and reasons for playing Pokémon GO over time, which could explain the sustainability of using the game to seek PA.
There were some limitations of the present study. First, a key limitation of the study was a small sample size. We needed 130 medical students to achieve the sample size, but we had only managed to recruit 13.8% (n = 26) of the target population. At 3 months after downloading the game, less than half of the participants were still playing the game. This limits the power of our study to detect statistical significances and associations between variables. For example, there was a big change in the PA participation between phase 0 and phase 1 (from 600 METmin/week to 996 MET-min/week) despite not being statistically significant. While the change in sedentary time between phase 0 and phase 1 was significant. This interpretation should be aware. Reasons that can explain the cause of the small sample size and high dropout rate probably because of 1) the decreased in popularity of the Pokémon GO game during the study period, 2) the recruitment of participants which included only Pokémon GO players who played the game more than a month, and 3) the lack of Pokéstops (where players could get essential items to play the game) in rural areas, where this study was conducted. This might limit the interest and dissemination of the game.
Second, using GPAQ, a self-reporting questionnaire, to collect the PA participation may lead to the biased data collection due to the small change in PA and potential recall bias of the participants. [39,40] More precise measurements of PA using accelerometers or performing shorter reporting spans (ecological momentary assessment) may be better suited for measuring PA in such study. Third, the lack generalisability because the study was conducted among medical students who might have a different PA patterns compared to young people in the community. Nevertheless, the findings from this study shed light on the PA and gaming behaviour of a group of Thai university students. Fourth, only Pokémon GO players were invited to participate in this study. Therefore, the difference between PA patterns of Pokémon GO players and non-Pokémon GO players could not be identified.
Safety issue has been a major concern of Pokémon GO and future augmented reality games. This is due to the risk of games-related injuries such as repetitive strain injuries [36,41] as well as serious events due to distractions from the real world and criminal incidences. [42][43][44] However, our study did not report any injuries or adverse events from using Pokémon GO.

Conclusions
Despite its potential to change PA behaviour, this study found that Pokémon GO may lack sustainability, which in turns might have affected its usefulness as a tool to improve PA. The motivation behind using Pokémon GO as a game rather than a PA app might have undermined the effectiveness of the game to improve PA. Further studies with a bigger sample size need to be conducted to explore how to combine fun with behavioural change and make it sustainable.