The study was conducted in Mwea Division, Kirinyaga District in Central Kenya located approximately 100 km North East of Nairobi in a riverine plain at an altitude of about 1159 m above sea level. It has a population density of 246 persons per km² in a total area of 581 Km². The main economic activity is rice growing and horticultural farming. Indigenous cattle are kept mainly for beef and draught power. The study area has two rainfall seasons with the long rains occurring from March to May and the short rains from October to November 1415. A previous study in the area recorded a malaria parasite prevalence rate of 24% among children aged below nine years of age and a 30–300 fold increase in malaria vectors 16. ITNs use and distribution in the area is promoted by the Ministry of Health and other partners like PSI and UNICEF at subsidized prices particularly to pregnant women and children under five years. This is one of the key strategic approach of the National Malaria Control Strategy by the Kenyan Government. Refer to an earlier publications for a further description of the study area 1617.

The study used a combination of qualitative and quantitative ethnographic cross-sectional household survey conducted in April 2005. Mwea Division was conveniently selected amongst other divisions because of its biggest burden of malaria in Kirinyaga District. The four study villages were purposefully selected and a standard sample size of 100 households was randomly selected from each village.

Interviews using structured questionnaires were conducted with household heads or spouses from the selected households in the four study villages. The questions focused on various sub-themes like, socio-demographic characteristics of the respondent and issues concerning the role of mosquitoes in malaria transmission. Other issues focused were perceived benefits of bed net use, reasons for treating bed nets and preferred colour and shape of a treated bed net. Questionnaires were prepared in English and verbally translated into the local language (Kikuyu) during interview time. Each question in the questionnaire was interpreted into the local language and corrections made accordingly before administration. Pre-testing was done in a non-study village and adjustments made accordingly. Two Focus Group Discussions (Men and Women) were held in each village with an aim of gathering descriptive information on bed net ownership and use. Clearance for the study was obtained from Ministry of Health. Questionnaires were administered after explaining the purpose of the study and criteria used to select each respondent. Informed verbal and written consents were obtained from the focus group participants and the household heads. Confidentiality of information was maintained during the whole study.

Data was recorded, entered and processed using the Statistical Package for Social Science (SPSS) version 11.5 for windows, MS Access and MS Excel. Association between dependent and independent variables were measured by use of the Chi-square test.

At the end of the study, 127 (34.5%) males and 241 (65.5%) females were successfully interviewed. Most of the respondents were protestants (54.9%) followed by catholic (43.8%). On marital status, 68.5% were married, 17.1% windowed, 8.2% single and 6.3% were separated at the time of interview. On occupation, 74.5% of the respondents were farmers, 7.3% were in self business, 4.9% in formal employment and 13.3% in other minor occupations like casual labour. Out of the total respondents, 40.5% had only completed primary education, 18.2% had dropped out at primary school, 13.9% managed to complete secondary level, 3.8% dropped in secondary schools, 19.8% were informally educated and 3.8% with university/college education only.

Malaria was perceived to be a major public health concern and one of the most frequently occurring diseases by 93% of respondents in the four study villages. Typhoid was rated second and bilharzia third by 38% and 10.3% respondents, respectively. There was an observed significant difference in rating the third most frequent occurring disease between the irrigated and non-irrigated villages in the study area. Most respondents in the irrigated villages reported bilharzia as the third most frequently occurring disease, while common cold was perceived to be the third most frequently occurring disease in the non-irrigated villages. The three most frequently mentioned diseases in irrigated areas were all waterborne and this was attributed to presence of large pools of stagnant water in the rice paddies.

Mosquito bite was mentioned to be the main 'cause' of malaria by 95% of the respondents. Other mentioned causes were: long rains or being rained on (12.5%), stagnant water (16%), dirty domestic surroundings (4.6%), wet and cold conditions (10.6%), eating raw food/mangoes (5.2%) and taking of dirty or polluted water (4.1%). Significantly, more males (10.5%) compared to females (2.5%) believed that malaria could also be caused by eating of raw food/mangoes (χ² = 10.19; df; P = 0.001) (Table 1). Participants' perceptions on the link between malaria and non-biomedical causes were expressed during FGDs in all the four villages:

'Fruits like mangoes and tomatoes are bitten by mosquitoes and when people eat them they get malaria' (Kagio, Women FGD).

'Mosquitoes bite mangoes and injects the germs and when one eats the mangoes s/he gets malaria' (Murinduko Women FGD).

'Some people become sick after eating raw food like tomatoes, mangoes and fermented porridge. I think it is the sourness, which makes the germs to become active in the body' (Murinduko and Mbui Njeru Men and Women FGDs).

'Most people in the village don't have good health due to poor nutrition and when they get exposed to cold weather the body become weak and they get sick with malaria' (Murinduko Women FGD).

The most common cited signs and symptoms of malaria were headache (70%), feeling cold (65%), (with the tendency to bask in the sun), fever (57%), general body weakness (57%), body/joint pains and vomiting. The level of knowledge on common signs and symptoms of malaria was average in all the villages and there was no significant difference in the score between the villages.

Overall, most respondents reported that the months of July (57%), April (53%), August (46%) and May (44%) were the four months of the year with high cases of illness episodes perceived to be malaria in the area (Multiple responses). However, all the other months of the year were mentioned but by relatively less number of respondents indicating that the disease was perceived to be endemic in the study area. The same months associated with high illness episodes were reported to experience high numbers of indoor and outdoor nuisance adult biting mosquitoes. This corresponded well with the peak rice growing seasons in the rice irrigated areas and also with long rain season in the non-irrigated areas (Figure 1). Most focus group discussion participants in irrigated areas attributed this to the presence of flooded rice paddies while non-irrigated areas attributed it to the presence of vegetations around houses. The expansion of informal "Jua Kali" small scale rice farming outside the scheme was also blamed for the current increases in mosquito numbers and malaria cases in the previously non-irrigated areas.

Seventy five per cent (75%) of the households reported to own at least one bed net. However, only 48.6% of the total population (1,776) had access to bed nets and 46.7% reported to have slept under bed net in the previous night. This translated into 96% use among the people who had access to bed nets. Among the reported nets, 62% were insectide-treated and the remaining 38% were untreated. There was significant variations between Age of the occupant (χ² = 60.103; df 7; P = 0.000), village type (χ² = 143.327; df 3; P = 0.000), relationship to the household head (χ² = 104.541; df 7; P = 0.000) and access to bed net among the population. The same variables caused significantly variations on both the conditions of bed nets present and the number of people who slept under bed nets in the previous night. There was a significant variation in the number of bed nets owned between the four study villages (χ² = 29.4; df 3; P = 0.000), with both Mbui-Njeru (91%) and Kagio (77%) reporting the highest number of households owning at least a bed net during the interview time. There were significance differences in reported use of treated mosquito nets between different socio-demographic profiles of the respondents like occupation (χ² = 7.955; df 3; P = 0.047), gender (χ² = 4.254; df 1; P = 0.039), education levels (χ² = 33.622; df 6; P = 0.000).

Among the households owning bed nets, 95% cited protection against mosquito bite as the main reason for using bed nets. Use of bed nets for protection against malaria was mentioned second by 54% (n = 300) of the respondents. Other reported benefits of using bed net were protection from other nuisance insects (25%) while 11% of the respondents acknowledged that the nets offered them warmth at night (Figure 2). A small percentage (13%) cited hotness as one of the major problems associated with sleeping under bed nets especially during the hot months of the year. Other mentioned problems were lack of enough air circulation (3%), difficulty in tacking in the nets each night (3%) and difficulty in getting up at night (2%). Main reason given for treating bed nets with insecticides was to repel the nuisance adult biting mosquitoes at night (83%, n = 350). Other mentioned reasons were to kill mosquitoes (42%) and to make the nets stronger (4%; Figure 3).

Overall, despite the general feeling from most respondents that treated nets were very important in protecting their households from mosquito bites, 55% of the respondent in the sample, felt that they could only afford to pay for 50 Ksh (0.67 US$) for a medium sized ITN. This was the subsidized price for a medium sized bed net during this survey in a nearby government dispensary (meant for pregnant women and children less than five years of age). Rectangular-shaped bed nets were preferred by 63% of the respondents. However, in Murinduko village, 52% preferred the conical shaped nets, while in Kagio, the differences in preference between rectangular and round/conical-shaped bed nets were not significant. Green-coloured bed nets were reported to be the most preferred colour by most respondents (51%) in the whole study area. Other preferred colours reported were navy blue (21%), white (16%) and light blue (11%).