Africa is particularly vulnerable to malaria for several reasons, including being exposed to the most severe form of the disease, having inadequate resources to bear the economic burden of the consequences and having to cope with the lack of proper infrastructure to effectively treat cases 1. Hence, prevention of the disease becomes paramount. The use of individual methods of protection are particularly important, especially in areas lacking any formal mosquito control programmes, like Burkina Faso. Bed nets, window screens, house sprays, ceilings, closed eaves and in some cases, zooprophylaxis can reduce the risk of malaria 23456. The evaluation of the different types of mosquito control methods used by households may be used to aid in the development and promotion of preventive measures that can be more readily integrated into communities. Thus, the aim of this study was to assess the reported use of different mosquito control methods among residents and their effects on the risk of malaria in the semi-urban area of Nouna, Burkina Faso.

The demographic characteristics of the cases and controls were similar (e.g. religion, household size, relationship of the respondent to the household head). An exception was the highest reported level of education. More cases than controls reported having secondary education or higher (OR = 2.35, 95%CI: 1.18-4.70), therefore this variable was included as a potential confounder in multivariate models.

Most cases (99.2%) and controls (98.6%) reported using at least one mosquito preventive measure during the wet season. These included mainly sleeping under bed nets (both insecticide-treated and untreated) and burning mosquito coils (Table 1). Few used smoke from burned plant material regularly (≥ once per week) as a means to repel mosquitoes. Of those households that did, plants were burned mostly in the bedroom (11.2%) and the living areas (6.5%).

Several different types of animals were also owned or kept by the households. Animals were frequently kept within household compounds by both the cases (89.6%) and controls (87.2%) at night. Of the different types of livestock, cattle were most often found within the compound (Table 2). Donkeys were found to be significantly and positively correlated with female An. gambiae abundance in sleeping rooms (Pearson's r = 0.21, P = 0.0002). Separate analyses by unfed (Pearson's r = 0.16, P = 0.0045), fed (Pearson's r = 0.22, P < 0.0001) and sub-gravid (Pearson's r = 0.16, P = 0.02) status also produced significant correlations. No associations were found with any other types of animals.

Reported use of insecticide-treated and untreated bed nets as well as mosquito coils did not significantly affect the risk of malaria in univariate analyses (Table 3). Conversely, when animals such as donkeys, rabbits and pigs were kept within compounds, a statistically significant protective effect was observed. An effect was not found with other animals such as cows, sheep, goats and poultry. Nonetheless, no variables included in multivariate models remained significant (P < 0.05), after adjusting for reported bed net use and level of education.

Most of the respondents reported using some protective measures against mosquitoes during the wet season, a finding also echoed by Samuelsen et al 9. Sleeping under bed nets and burning mosquito coils were the most popular forms of control during the rainy season. In fact, reported bed net use was very high. This could have been the result of promotions in the region encouraging the use of insecticide-treated bed nets 3. However, actual bed net use was not investigated in this study and could have been over-reported by the respondents, particularly since knowledge of the protective effects of bed nets does not necessarily translate into actual net use 10. Insecticide house sprays were not a popular method of mosquito control in Nouna, likely because of the costs associated with such measures (Ouédraogo pers com. 2008). Similarly, smoke from the burning of plant material was also not as extensively used as in other regions like The Gambia 11.

The evidence regarding the effectiveness of zooprophylaxis is generally not clear. Keeping animals near houses can shift the biting behaviour of mosquitoes away from people and, therefore, mitigate malaria risk 12. Conversely, sharing the house with livestock could also increase the risk of malaria. Greater numbers of mosquitoes may be initially attracted to households because of the presence of animals but then end up preferentially feeding on humans 1314. In The Gambia, it was found that animals conferred little or no protection against An. gambiae and that the presence of cattle did not result in greater numbers of zoophilic species or a change in species composition. Moreover, the human blood index (HBI), or proportion of human-fed mosquitoes, was not altered among indoor resting mosquitoes between households with and without cattle 15.

In this study, most domestic animals sleeping within the compound or house were not associated with either an increased or decreased risk of malaria. Donkeys, rabbits and pigs, however, were associated with some zooprophylactic effect in univariate analyses. One possible explanation for the protective effect observed in this study could be that if vector breeding sites are closer to animal enclosures than to houses, mosquitoes may choose to feed on livestock as opposed to humans 1617. We observed significant, positive correlations between donkeys and the abundance of female An. gambiae mosquitoes captured in the sleeping rooms of the participants. This could suggest that donkeys, in particular, attract mosquitoes. Likewise, Pålsson et al 18 found that pigs, but not goats, were associated with increased numbers of mosquitoes collected from bedrooms in Guinea Bissau. Nonetheless, when bed net use or other behaviours, such as staying indoors are considered, animals may be a more accessible blood meal source 1920. It is also possible that certain types of animals, in particular equines such as horses and donkeys, exert stronger zooprophylactic effects. Bøgh et al 21 observed that the preference of An. gambiae for horses was strong in The Gambia, although since horse ownership was also very high, accessibility may have also influenced host choice. Another study in The Gambia also found that horses located near houses were correlated with lower numbers of mosquitoes in rooms, whereas no significant association was found with cattle 22. Additional evidence suggests that the An. gambiae mosquitoes found in certain areas of West Africa may be less anthropophilic compared to those found in other areas, which may also explain some of the zooprophylactic effects observed in the study 2324.

Zooprophylaxis may lower the risk of malaria in this region. However, before this method of mosquito control could be recommended, other factors need be considered such as the potential for increased mosquito survival from greater feeding success or the creation of additional vector breeding sites from hoof prints, animal enclosures and water tanks 2526. It would also be necessary to further investigate separately the impact, in particular, of donkey, rabbit and pig proximity on malaria outcome and acquire more knowledge about mosquito anthropophilic biting behaviour in the area.

A study limitation was that mosquito captures were only conducted for one night. As a result, there was no way to determine if the species composition and counts obtained were representative of the conditions typically found in households. Nonetheless, cases and controls were largely sampled on the same day to avoid confounding from weather conditions. Also, sampling was not carried out on public holidays or other days that might have affected collections. Therefore, cases and controls were assumed to be equally affected by environmental factors and any bias non-differential. An additional limitation of this study was that active testing for malaria among the controls did not occur before recruitment, which may have biased the results towards the null. However, no controls reported having malaria during the recruitment period and none from the second field season had a fever (≥37.5°C) at the time of data collection.

Given the infrequent use of smoke from the burning of plants and likely respiratory side effects as well as the interesting but unconfirmed results of zooprophylaxis, other measures to protect against mosquitoes may be of greater benefit in this region. Bed nets and mosquito coils have already gained wide acceptance and should be further encouraged. Community-based house spraying programmes may also be an effective tool.

The authors declare that they have no competing interests.

SY, AS, RS were responsible for designing the study, data collection was conducted by SY and AS, SY and VRL examined, analysed and interpreted the data, SY, VRL and RS drafted the paper. All authors read, edited and approved the final manuscript.