Malaria threatens the lives of 3.2 billion people globally and leads to over one million deaths annually [1]. Malaria is hyper-endemic in Ghana, accounting for 44% of outpatient attendance, 13% of all hospital deaths, and 22% of mortality among children less than five years of age [1, 2]. Though malaria is responsible for 9% of overall mortality in Ghana, at least 40% of malaria deaths occur among infants and children under the age of five [3].

The Roll Back Malaria campaign emphasizes the use of insecticide-treated bednets (ITNs) which are effective in reducing human contact with mosquitoes [1]. Results of recent research suggest that ITNs can reduce malaria episodes by 48–50% [4] and, if universally used, could prevent an estimated seven percent of global under five mortality [5]. The use of ITNs could save the lives of more children than any other single intervention except breastfeeding and oral rehydration therapy [5]. At an estimated cost of US$1.2 per person protected per year, ITNs are considered to be one of the most cost-effective health interventions in Ghana [6].

Despite evidence demonstrating that the use of ITNs decreases malaria-related morbidity and mortality, it is difficult to encourage consistent and appropriate use of bednets. Estimates from Africa as a whole suggest that only three percent of children less than five years of age sleep under ITNs, while up to ten times as many are thought to sleep under any bednet [1]. Past research in a variety of countries reveals that children may fail to sleep under bednets for a number of reasons. For example, parents attribute malaria to causes other than mosquitoes and may not associate bednet usage with the prevention of malaria. Additionally, parents may consider it too costly to purchase a bednet or to treat it with insecticide and, despite increasing availability and improved distribution systems, many communities have limited access to bednets [1, 7]. Parents may consider using bednets difficult because sleeping under them can be hot and uncomfortable; or, they may believe that bednets resemble a burial shroud or that insecticides used to treat ITNs will harm their children [8]. On the other hand, community members often value bednets because they decrease the nuisance associated with mosquitoes. However, such beliefs often undermine the consistent use of bednets during the dry season when mosquitoes are less noticeable [9–11].

Many other factors also influence caregivers' decisions and malaria prevention behaviours [12, 13]. For example, community perceptions, beliefs and attitudes about malaria causation, prevention and cure influence efforts to address malaria and are often overlooked in control efforts [14, 15]. In the case of Ghana, some individuals maintain the notion that certain types of "fever" (local terms for fever and malaria often coincide) are caused by the heat of the sun and, therefore, cannot be prevented by bednet use, intermittent treatment or insecticides [16]. Cost may also discourage parents from practicing optimal health behaviours. With malaria, the cost of bednets and the cost of treatment (clinic fees and medications) for sick individuals play an important role in determining which behaviours parents practice. In addition, the incidence of malaria and immunity levels within a population can influence community perceptions of disease severity and may influence the perceived importance of engaging in behaviours such as bednet use [13, 17].

As policy makers and programmers in Ghana seek to improve malaria prevention behaviours, it is critical that they are fully informed of the determinants of such behaviours. One way of improving the effectiveness of malaria prevention efforts is to better understand what distinguishes individuals who practice optimal behaviours ("doers") from individuals who do not ("non-doers"). The purpose of this paper is to identify knowledge-related factors that affect bednet use among children under five years of age. This study uses baseline data from Ghana to test the hypothesis that mothers who practice optimal malaria prevention behaviours by having their child sleep under a bednet (doers) 1) have a better understanding of the cause of malaria as well as appropriate protective methods; 2) are more likely to know the signs and symptoms of malaria; 3) are able to identify those most vulnerable to malaria; 4) know the potential consequences of malaria during pregnancy; and 5) understand how to protect pregnant women from malaria.

Results from this study suggest that region of residence, food security and beliefs about malaria were associated with sleeping under a bednet. For example, doers were nearly five times more likely to come from the Afram Plains District than non-doers. Mothers from food secure households were 2.5 times more likely than their food insecure peers to have children sleep under bednets. Doers also had relatively high knowledge related to malaria; however, misconceptions about malaria and lack of knowledge also existed among doers. More than 90% of respondents identified mosquito bites as a cause of malaria, though mosquito bites were not commonly identified as the only cause. Doers were more likely than non-doers to believe that malaria is caused by overworking oneself and that keeping surroundings clean is protective against malaria during pregnancy. Likewise, both doers and non-doers identified children and pregnant women as the groups most vulnerable to malaria (85% and 60%, respectively). However, doers were more likely than non-doers to also believe that adult males were most vulnerable to malaria. Lastly, doers were 5.6 times more likely than non-doers to believe that dizziness is not a symptom of malaria. Though dizziness is not considered a traditional symptom of malaria it can occur as an atypical manifestation, particularly when severe and acute anaemia results from infection.

This study tested the hypothesis that mothers who practice optimal malaria prevention behaviours by having their child sleep under a bednet (doers) 1) have a better understanding of the cause of malaria as well as appropriate protective methods; 2) are more likely to know the signs and symptoms of malaria; 3) are able to identify those most vulnerable to malaria; 4) know the potential consequences of malaria during pregnancy; and 5) understand how to protect pregnant women from malaria. The results of this study do not support these hypotheses. Logistic regression analyses showed that doers did not express higher knowledge related to protective methods against malaria. Additionally, they were no more likely than non-doers to identify only the correct cause of malaria, the standard signs and symptoms of the disease, or those groups recognised as most vulnerable to the disease. No notable statistical differences were observed between doers and non-doers in relation to knowledge of the consequences of malaria during pregnancy. Though doers were more likely than non-doers to believe that keeping surroundings clean was protective against malaria during pregnancy (which offers some protection by reducing mosquito breeding near the home), they were not more likely to identify the better protective methods of both using ITNs and taking antimalarial drugs.

Greater knowledge about malaria causation, symptoms and prevention as well as groups most vulnerable to malaria does not necessarily translate into improved health behaviour, as evidenced by the fact that in some cases non-doers were more knowledgeable about malaria than doers. Similar to what other researchers have found in Ghana and elsewhere [15, 17–20], it appears that "knowledge and behaviour change have no simple and direct relationship, and other intervening and contradicting variables may lead to behaviour consistent and inconsistent with knowledge" (p.376) [9]. The gap between knowledge, attitudes and practices, however, is not a new phenomenon, nor one unique to malaria. Others have noted similar results for other health behaviours and products including the use of personal protective equipment during pesticide application [21], the use of condoms [22, 23] and family planning methods [24, 25], among others.

This study is subject to a number of limitations. As noted previously, the survey was designed to evaluate the impact of Freedom From Hunger's malaria module on key behaviours related to malaria prevention and control. Consequently, information about a broad array of behavioural determinants such as cost and availability of bednets, attitudes about bednet use, perceptions of malaria disease risk and so on was not collected. In addition, the Eastern and Central regions were substantially different with respect to socio-demographic characteristics, geography and accessibility and availability of health services. Interpreting differences between the two regions is therefore difficult because such differences may reflect variations in health care access, education, income and malaria prevalence. Exposure to educational messages and malaria interventions likely varies between the study areas as well. For example, according to the Demographic and Health Survey (DHS) conducted in Ghana in 2003, households in the Eastern Region reported higher rates of exposure to media messages on malaria compared to households in the Central Region. In addition, the Ministry of Health (MOH) sponsors a bednet programme to improve access for children less than five years of age in the Central Region. In Afram Plains, the United States Agency for International Development (USAID) and the United Kingdom's Department for International Development (DFID) offer vouchers for ITNs. The authors of this paper have not been able to identify which programme has been more successful (the voucher programme in the Eastern region or the MOH programme in the Central Region). However, based on the experience of FFH staff in Ghana, there is some indication that individuals in the Eastern Region (who live in close proximity to the Volta River) are more likely to use bednets as a measure to reduce nuisance from mosquitoes and other insects. Such regional differences in bednet usage among children less than five years of age have also been identified by the 2003 DHS. Results from the DHS show the Central Region as that having the lowest rates of bednet ownership and usage among children under age five in the country, with the Eastern Region having just slightly higher rates in both cases [26].

Results from this study are comparable to findings from other studies conducted in various regions of Ghana. For example, the NetMark project reports that 82% of surveyed Ghanaians identified mosquitoes as a cause of malaria and 65% correctly identified fever as a symptom of malaria [27]. Likewise, Adongo and colleagues found that 79% of respondents identified mosquitoes as the cause of malaria; however, alternative cultural explanations – including exposure to heat – were also common. In addition, 92% of respondents believed that bednets could prevent malaria [9]. Agyepong and Manderson found that the most commonly identified signs and symptoms of malaria among children included hot body (fever), yellow eyes and urine, vomiting and refusal to feed [13].

Findings from this study and research carried out by others suggest that many Ghanaians fail to use bednets and insecticide treated nets in particular. While a majority of individuals in this study possess at least some level of correct knowledge, alternative beliefs and attitudes related to disease causation, prevention and treatment persist and appear to affect the adoption of ideal health behaviours. In the words of Agyepong and Manderson, "people's ability to comply with interventions and to treat sickness is affected by their acceptance of the interventions, their understanding of the nature of the illness and the relationship between vector and infection, and other social, economic and cultural factors" (p.79) [10].

When developing interventions for populations with traditional beliefs and culturally defined behaviours, it is useful to classify such beliefs and behaviours as positive, harmful, or neutral. Programme planners and implementers should reinforce positive beliefs and behaviours and identify culturally appropriate means for modifying harmful beliefs and behaviours. Neutral beliefs and behaviours can be left alone or, if needed, incorporated into health messages that promote ideal health behaviours. Though such culturally based ideas and behaviours may vary somewhat between communities and efforts to identify and include them in health education messages can add to the initial cost of programmes and interventions, the benefit of the added effectiveness over time cannot be discounted.

In Ghana, caregivers maintain a number of positive beliefs and practice behaviours conducive to the prevention of malaria. The majority of individuals in this study were able to recognize that mosquitoes play a role in malaria transmission and that bednets can protect against disease. Results presented here did not highlight any beliefs about malaria that were outright harmful; however, participants did possess some neutral beliefs that could be used to support positive beliefs and practices. For example, the notion that malaria is caused by exposure to heat or overworking oneself can be classified as a neutral belief – one that has traditionally served to explain the effects of malaria on the blood as well as the visible symptoms of the disease [14]. Combining alternative, culturally based ideas about malaria causation – including exposure to heat – with biomedical explanations of disease transmission may improve the uptake of educational messages and encourage the adoption of appropriate preventive behaviours [9]. Programme planners may wish to develop messages that suggest that mosquitoes carrying the disease prefer to feed on blood that has been heated by long hours working or playing in the sun. Bednets could then be promoted as a way of keeping mosquitoes from detecting and feeding on hot blood at night, particularly during the dry season when net use typically declines and complaints about nets being too hot increase. It is important that programme planners use local terminology for diseases that correspond closely to the clinical definition of malaria. Additionally, while this particular study did not assess issues of cost and access, there is evidence to suggest that programmes and policies should continue to improve the availability of ITNs for the poorest income groups and those most affected by malaria [7].

In Ghana, malaria remains an important cause of childhood morbidity and mortality. The use of bednets – and in particular ITNs – is an effective method for reducing the incidence and disease burden of malaria. Despite high levels of knowledge about malaria, bednet use remains inadequate if the lives of vulnerable Ghanaian children are to be protected. Efforts to improve the use of bednets must extend beyond sharing information about the benefits of ITN use. Such efforts must address sociocultural, economic and other determinants of behaviour. Educational messages must be culturally sensitive and capitalize on the positive beliefs and behaviours that already exist in local communities. Likewise, programmes that mobilize communities can play a critical role in the adoption of preventive behaviours and improved rates of child survival.