Physical activity, screen time and obesity status in a nationally representative sample of Maltese youth with international comparisons

Participants completed a questionnaire with pre-coded answer categories to assess the time spent watching television (TV), using a computer for chatting, internet, emails or homework, and playing games on a computer or games console. Separate questions were asked for use on weekdays and weekends and categories included none, 1 minute to 30 minutes, 31 minutes – 1 hour, 1–2 hours, 2–3 hours, 3–4 hours, and more than 4 hours. Questions were adapted from the Health Behaviour in School Children study (HBSC) [12]. A previous study [15] reported satisfactory test-retest reliability for these questions. Data for TV viewing were recoded into less than 2 hours, and 2 hours or more (i.e. exceeding American Academy of Pediatrics guidelines [16]) while other ST was recoded into less than 1 hour and 1 hour or more. To facilitate comparison with data from the Health Behaviour of School Children study (HBSC) the data for ‘playing games on a computer and games console’ for over two hours was also analysed.

Body height and weight were measured with the child in light clothing and without shoes. Height was measured using a SECA 213 Leicester Stadiometer (SECA, Hamburg, Germany) and recorded to the nearest 0.1 cm. Weight was recorded using a SECA 813 Digital Scale to the nearest 0.1 kg. Body mass index was calculated (kg/m²) and children were classified into normal weight, overweight and obese, using the age-related International Obesity Task Force criteria [17].

Physical activity was assessed using Actigraph GT3X accelerometers (Actigraph, Pensacola, USA) set at 10 second epochs to capture children’s intermittent physical activity [18]. Participants were instructed to wear the accelerometer on an elastic belt over the right hip for five consecutive days including three weekdays and two weekend days. The accelerometer data were then processed using Kinesoft version 3.3.62 with a valid day based on at least 600 minutes of monitor wear. Periods of at least 60 minutes of continuous zeros were considered to be non-wear time and were removed [19]. Participants who had at least three days of valid data were included in this study [20]. The following accelerometer variables were then derived: total physical activity per day (in counts per minute), mean minutes of sedentary time per day, and mean minutes of moderate to vigorous physical activity (MVPA) per valid day. Time spent in these thresholds was classified based on the Evenson criteria of <100 CPM for sedentary and >2296 for MVPA [21]. In a rigorous study by Trost these cutpoints were found to provide the best classification accuracy [22].

Data were filtered by different time periods to enable an analysis of activity patterns during weekdays and weekend days. The morning period before schools was from 5.30 am to 8.29 am; the period at school from 8.30 am to 1.59 pm; the afternoon period from 2.00 pm to 6.59 pm and the evening time from 7.00 pm to 11.59 pm.

A search of the Biosis Citation Index, Derwent Innovations Index, Medline and Scielo Citation Index online databases was conducted in July 2013 using Web of Science. This identified articles that had cited Evenson’s 2008 [21] study. We manually screened the articles yielded from the search to identify articles which reported studies where a) accelerometry had been used to assess physical activity, b) comparable cutpoints for levels of intensity of activity had been used, and c) 10–11 year olds were the subjects of the study. While this was not a systematic review, it provided an opportunity to compare the results in the current study, with similar studies from other countries.

Group means and standard deviations (SDs) were calculated for physical activity and sedentary time, and percentages for screen time. Independent sample t-tests were used to explore gender differences in activity and Chi-square tests were used for gender differences in screen time. Distributions were checked and as all variables approximated normality with relatively small deviations, parametric tests were used. Differences in physical activity by weight categories were analysed using analyses of covariance (ANCOVA) with the model adjusted for socioeconomic status. Chi-square tests were used to compare ST by weight status groups. To facilitate the development of targeted intervention strategies, separate analyses were performed for boys and girls, for weekdays and weekends, and for different periods of the day. Follow-up Bonferroni pairwise comparisons for significant main effects were calculated. All analyses were carried out using the IBM Statistical Package for Social Sciences (SPSS) version 21 and a p value of 0.05 was used for statistical significance.

The means and standard deviations of boys and girls PA are presented separately for weekday and weekend days, in Table 1. The mean total PA for boys was 477.8 CPM (Confidence Interval [CI]–463.6-492.3) and for girls 385.0 CPM (CI–374.8-394.7) (p<0.001). Boys were engaged in 58.5 minutes of daily MVPA (CI–56.2-60.7), while girls’ MVPA was 42.2 minutes (CI–40.7-43.5) (p<0.001). Only a quarter of the children in this study (24.7%) met the daily recommendation of over 60 minutes of MVPA, and the percentage was higher for boys (39%) than girls (10%).

Analysing the data by time-period, we found that boys were significantly more active than girls during all measured times. Weekday MVPA was highest during the period after school (2.00-6.59 pm), when boys were active for 24.3 minutes (CI–22.9-25.5), seven minutes more than girls (p<0.001). During the school day (8.30 am −1.59 pm), boys were engaged in 20.8 minutes of MVPA (CI–19.9-21.6), 5.6 minutes more than girls (p<0.001). On weekends, equal amounts of MVPA were obtained for the morning and afternoon periods for boys (25.3 and 23.1 minutes respectively), and for girls (16.7 and 16.4 minutes) (p<0.001).

The number and proportion of boys and girls in each ST category are presented separately for weekday and weekend days in Table 2. A high prevalence of over one hour of playing games on a computer or games console was reported by boys on weekdays (44.8%) and on weekends (51.6%) and these values are considerably higher than those reported for girls (28.1% and 35.0%) (p<0.001). Almost a third of boys (29.3%) watched over two hours of TV on weekends, compared to 20.6% of girls (p<0.001) and this prevalence is double that on weekdays.

Table 3 provides a comparison of the results of the current study and previously published studies. Using the same Evenson cut points [21], mean MVPA of Maltese boys and girls were marginally higher than those of children in Philadelphia, USA [23], while compared to data from Canberra, Australia [24], mean MVPA of Maltese boys and girls were considerably higher. In contrast, means were much lower than those in the PEACH study carried out in Bristol, England [25]. Comparing median MVPA values to those of children in a longitudinal study of the National Institute of Child Health and Human Development (NICHD) in ten geographical locations in the USA, Maltese children were again more active [26]. The percentage of Maltese children meeting recommendations for PA (60 minutes a day) was slightly lower than that of children in USA, while compared to Australian data, percentages were higher for Maltese boys and lower for Maltese girls.

Comparing the percentages of children exceeding two hours of screen time on weekdays in this study to other countries in Europe, Canada and USA, using the Health Behaviour in School-aged children (HBSC) study [3] as a comparator, we found substantially lower rates in Maltese children. TV watching percentages for Maltese children were lower than children in all other countries, while for playing games on computers and games consoles, they were only higher than the lowest HBSC values observed in Switzerland (and boys in Luxembourg).

Table 4 provides a presentation of PA by weight status. For weekdays the overall pattern is that MVPA and total PA differ by weight group. Follow-up tests indicated differences between the normal weight and obese and for boys on a weekday there were also differences between the overweight and obese groups. There were similar patterns when the analyses were repeated for the separate time periods. In all instances PA levels were lower in the obese group than the normal weight group. There were significant differences between the TV viewing times of boys in different weight categories on weekends only. Almost half of obese boys (45.5%) watched over two hours of TV compared to 30.8% of overweight and 24.7% of normal weight boys (p=0.004) (see Table 5). Obese boys (45.5%) were also more engaged in using a computer for chatting, emails or homework than overweight (28.0%) and normal weight boys (25.6%) (p=0.006), however this was significant only on weekdays. No significant differences were observed in the screen time of girls in different weight categories. In contrast, objectively measured overall sedentary time of overweight boys was significantly lower than that of normal weight boys both on weekdays and on weekends (see Table 4). Differences in girls were not statistically significant.

The strength of this study is its use of objective measurement of physical activity and weight status with a nationally representative sample of Maltese 10–11 year olds. A large percentage of the sample selected (80%) accepted to participate in this study, however a limitation might be that we do not know the weight status of those who did not provide consent. It is usually parents of obese children who refuse consent in studies on obesity [41]. Therefore if this is the case, the sample may be under representative of overweight and obese children. Although accelerometry has been used to provide measures of intensity and duration of physical activity, current data cannot describe what the participants are doing while they are sedentary or active and cannot assess some activities like swimming and cycling. This study has used Evenson’s cutpoints and other higher accelerometer cutpoints would have produced lower prevalences of children meeting PA guidelines. A further limitation was that we were not able to calculate energy expenditure and the results provide simple estimates of movement. This underestimates the energy expenditure during activity of children who are heavier. Screen time was self-reported, and considering that children engage in many different types of screen time, the measure is likely to lack precision and accuracy. It is important to recognise that other factors including socio-economic position could also have affected the associations found in this study. We relied on evidence of the reliability and validity of sections of the questionnaire from other studies and cannot confirm how robust the measures were in our sample. Children were classified into different weight categories using the IOTF standards, and although these standards are used in many studies for international comparison, another study in Malta [14] has shown that these standards produce lower prevalence of overweight and obesity compared to WHO standards.