Only 1.1% of the school children could be considered pathological Internet or game users at baseline. In general, heavy Internet users spent significantly more hours on Internet, showed more heavy game use, had more psychosocial problems and were more often unhappy at school and at home than non-heavy Internet users. Likewise, heavy game users spent more hours on gaming, showed more heavy Internet use, had more psychosocial problems and were less physically active than non-heavy game users. Heavy Internet and game use was not related to alcohol use or BMI. Thus heavy Internet and game use is not an isolated problem but is accompanied by other health issues. This means that school health education programmes should focus on more than one health issue at a time, as did the programme we tested.
A year after the education programme had started, we found the number of hours spent daily on Internet had increased, as had the number of pathological Internet users. While the number of game users had decreased, the heaviness of game use had slightly increased.
The differences between the various levels of education were minor. Second grade school children in 2008 spent more time on Internet (hours/day), were more heavy Internet users and were more often pathological Internet users compared with second grade school children in 2007. In contrast, there was a marked decrease in heavy Internet use and a non-significant decrease in CGUS scores among third grade school children in 2008 compared with 2007. Heavy Internet (or game) use is associated with more hours spent on game (or Internet) use, more heavy game use (or Internet) and more problematic behaviour.
School children following a lower general secondary education were more likely to be heavy Internet users than school children following other educational levels. This relationship between heavy Internet use and education level has been reported earlier for 13- to 15-year-old school children . We did not find heavy Internet or game use to be associated with alcohol consumption among the school children, which is in contrast with the results of Ko et al. (using other questionnaires) [19, 20]. Moreover, in contrast with other investigators , we found no significant relationship between heavy Internet use and physical activity; however, we did find physical activity to be negatively associated with more heavy game use, but we did not find literature for comparison. Heavy Internet and game users reported more psychosocial problems, as reported by Cao and Su  for Internet addicts. We found no relationship between heavy Internet/game use and BMI, in contrary to many studies that did [21–23].
The increased time spent on Internet (hours/day) by the school children in 2008 compared with the same school children in 2007 is consistent with national trends of increased frequency and duration of Internet use with increasing age among secondary school children [38, 39]. We also found that second grade children spent more time on Internet in 2008 than in 2007 which is consistent with the increased use of Internet reported among school children in recent years . However, we did not find an increase in Internet use among third or fourth grade school children. The number of pathological Internet users increased over time, which is contrary to a previous Dutch report which reported a decrease (4.2% in 2006, 3.6% in 2007, and 3.2% in 2008) . The proportion of children who played computer games (64%) was comparable to that reported in previous studies . We found a lower proportion of pathological game users than reported previously (1.6% vs 3.2%) but results are not quite comparable because different measurement scales were used .
Some limitations of this study should be mentioned. The school at which the intervention took place is not an average Dutch secondary school. In this school, most school children are of Dutch ethnicity, come from a high socioeconomic background and follow higher levels of education. Therefore, the amount of problematic behaviour is almost certainly less than the national average and implementation of the education programme was probably easier than in an average secondary school. It can be hypothesized that effects of the prevention program might be greater at schools with more problematic behaviour. Unfortunately, we did not use a control school at this stage of the study and thus changes in Internet and game behaviour over time must be interpreted with caution. It is not clear in what way the education programme contributed to the changes in Internet and game use and how use would have changed over time without the education programme, as a result of government initiatives, parenting or television programmes on the subject. Moreover, the school children might have become more aware of their behaviour when they completed the questionnaire for the first time, which led to changes when completing it the second time. The questionnaire measuring game use and the cut-off score for heavy game use is based on the validated questionnaire measuring internet use, but is not yet validated itself. Therefore, results of this study regarding game use and its generalizability have to be interpreted carefully. It is unlikely that the computer based follow-up questionnaire (in comparison to the paper-and-pencil questionnaire) caused measurement bias as several studies have shown results between these different methods are comparable . Finally, all computed scores of behavioral factors were based on self report and could therefore have resulted in some reporting bias. But even taking these limitations into account, we can draw some interesting conclusions from our data.
As mentioned before, heavy Internet and game use was associated with more psychosocial problems, less physical activity, and greater unhappiness. The question is which comes first - do heavy Internet and game users have more psychosocial problems because of their lack of interactions with family and friends in real life, or do school children lacking social interactions choose Internet and game use as an easier way to communicate with other people. Other studies have found that "chat" users are socially fearful and may use the Internet as a form of low-risk social approach and an opportunity to rehearse social behaviour . The decrease in physical activity suggests that time spent on gaming goes at the cost of time spent on physical activity. However, other studies have reported that Internet and game use does not increase leisure time but replaces time spent on other leisure activities . The observation that heavy users were more often unhappy at school and at home is important and highlights the importance of detecting heavy Internet and game use and of providing support, treatment and preventive measures at school.
The fairly large increase in time spent on Internet probably reflects the combined influence of an increase in use over the last years and an increase in use with increasing age. Despite the increased time spent on Internet, CIUS scores did not change, which suggests that spending more time on the Internet does not lead to more heavy Internet use. It may also be possible that the CIUS is not sensitive enough to detect such changes or that hours of Internet use and heaviness of internet use are not linearly associated. The decrease in the number of game users with time was noticeable and might be due to the education programme but more evaluations over time are needed to see whether this trend continues. The fact that Internet use increased among second grade children but not among third or fourth grade children raises the question whether second grade school children are less amenable to the influence of the educational programme.