After the intervention, the malaria prevalence was significantly decreased in school children in the intervention school, although no decrease was observed in children in the control schools. Previous studies conducted in Ghana reported that falciparum prevalence was higher in the rainy season than in the dry season, and a little higher in the middle of the rainy season than at the beginning or end [21, 22]. In the present study, pre- and post-intervention surveys were conducted at the end and middle of the rainy season, respectively. As is consistent with the trend reported from the previous studies, there was a slight increase in prevalence in the control schools at the post-intervention survey. Therefore, in the intervention school the effect of seasonal variation is unlikely to be the principal reason for the decrease. Although a marked difference in prevalence between intervention and control schools was observed at the baseline, intervention impacts can be considered as one of the main factors influencing prevalence reduction.
Previous studies conducted in sub-Saharan African countries including Ghana showed that people believed that not only mosquito bites, but also eating mango, drinking dirty water, and being exposed to hot sun were causes of malaria [17, 19, 20, 23, 24]. The same finding was observed in this study at pre-intervention. As Table 4 shows, 95.2% of the community adults correctly answered that mosquito bites can cause malaria. However, only a small proportion of the respondents disagreed with the incorrect statements that "heat from the sun", "eating mango", and "drinking dirty water" cause malaria. After the intervention, this knowledge was significantly improved in the intervention area. The improvement is important because a lack of understanding of the linkage between malaria and mosquito bites is associated with poor adherence to vector control interventions [20, 25].
In this study, both community adults and school children showed significantly increased knowledge in the item "Paracetamol alone cannot cure malaria". In Ghana, most malaria cases have been managed at the household level . However, during the fever episode, nearly 40% of children under 5 years old were not treated with any anti-malaria medicine , and in the absence of anti-malarials, paracetamol alone was commonly administered as treatment . Knowledge improvement regarding treatment could be beneficial even to children. In Kenya, Geissler et al  reported that a considerable number of children self-treated their febrile illness without help from their caregivers.
In Ghana, most children under five years of age have yet to be protected by ITNs . It has been emphasised that untreated conventional nets should be treated with insecticide to increase coverage of ITNs . The observation survey found that 45% of the nets that respondents showed us were conventional nets. In response to the intervention, community adults who treated nets with insecticide increased from 21.5% to 50.0%, compared with the 25.3% to 30.5% in the control area. This fact suggests that the intervention was effective in increasing the coverage of ITNs.
Although the extra opportunity to treat their nets was provided to community people during the one-day campaign, providing an opportunity alone is unlikely to increase the net treatment rate. According to the local health authority, community people rarely participate in free net treatment services which health workers offer regularly in the study villages. Previous studies reported that barriers of insecticide-treatment were not only cost and access to treatment place, but also fear about insecticide, and poor linkage between malaria and ITNs [27–29]. Thus, community awareness raised by the children about the malaria likely had a substantial impact on increasing net treatment practices.
The results also showed that the frequencies of talking with children and guardians/neighbours about malaria unexpectedly decreased at the post-intervention survey. This trend was seen both in intervention and control groups. There is a possibility that community adults were busier at post-intervention period than pre-intervention period because of seasonally related changes to farming labour intensity. Unexpectedly, at the baseline, the frequency of talking between children and guardians/neighbours in intervention groups was much higher than those in control groups. This might be due to a higher burden of malaria, suggested by the higher prevalence at pre-intervention among school children in the intervention school.
This study has five major limitations. First, although conventional ITNs and long-lasting insecticide-treated nets (LLITNs) co-existed in the study site, no attempt was made to teach the respondents to recognize the two major differences to avoid confusion which might arise among study participants; conventional ITNs should be regularly treated and not be washed frequently. In contrast, LLITNs have no need of treatment until nets are washed many times and should be washed to activate insecticide agents. This limitation might be reflected in the result showing that knowledge "Necessity of ITN re-treatment" was not improved in the intervention area. Additionally, almost all of the community respondents reported that they washed their nets "very often" or "sometimes" in the past six months, although some of the nets must have been conventional ITNs. Second, although we used the local word "Asra" to define malaria, "Asra" does not necessarily correspond to "malaria" as defined by modern medicine. As shown in the previous study in Ghana, "Asra" was used interchangeably to define both malaria and fever . Third, in data analysis, clustering of individuals within the same school was not taken into account, resulting in failure in addressing the cluster effects including class differences within the school. Fourth, results of the parasitological survey might be biased because of selection bias. Although no age and sex difference was observed between children who participated in the survey and those who did not, sex difference was observed in the children in the control schools. Finally, a randomised controlled design was not employed. However, no baseline difference was observed in demographic characteristics of the study participants and bed net related characteristics.
For the intervention, schoolteachers successfully adopted education activities using the PLA approach, such as role-playing, poetry recitals, slogan chanting, song composition and dramatization. These activities could be socially and culturally acceptable, because teachers themselves designed these activities. The results showed that most of the community adults were exposed to these activities. Moreover, participants in the intervention area were more likely to talk about malaria than those in the control area. Thus, the application of this strategy in other malaria endemic areas is recommended.
Scaling up school-based health education interventions should be easy if a well-established school health system is available. The study in Thailand utilized a school health system based on the Health Promoting School concept . In recent years, WACIPAC has introduced the strategy for the setting up and establishment of school health management systems as a national programme in 10 West African countries. Moreover, WACIPAC has recommended the necessity of coordination among donors interested in school health. Partnership has been promoted between governments in target countries and donors for the establishment of school health management systems. Thus, school-based intervention has the potential to be scaled up on the basis of the systems in each country in West Africa.