A community-based cross-sectional study design was used in three rural districts (Dallo Mena, Sewena, and Meda Welabu) in the Bale Zone, Oromia regional state, Ethiopia. The districts were specifically chosen for having high malaria transmission rates. The climatic conditions in the Bale Zone include 1.8% frosty highland, 7.7% highland, 25.5% midland, 63% lowland and 2% desert. Of the 18 districts in the Bale Zone, nine were pastoralist and the rest were agrarian; 14 districts are at risk of malaria transmission (Figure 1). The zone had a total population of 1 794 673 (909 164 men and 885 509 women), six hospitals, 84 health centres, 372 health posts, 100 private clinics, one NGO clinic, four other public clinics and 95 pharmacies and drug stores.

The target populations were all household (HH) members in the randomly selected Kebeles (the smallest administrative unit in Ethiopia’s context) of the three districts. Those who were permanent residents (resided there at least for 6 months) of the study area and aged 18 years and above were included in the study, while those who had a severe illness during the data-collection period or a known mental disorder that affects the accuracy of the given data were excluded. The total sample size was 634. It was calculated using a single population proportion formula based on the following assumptions:

Therefore, n₀ = (Z_α/2)² * (p*q)/d² = (1.96*1.96)*(0.5*0.5)/(0.05*.05) = 384

Considering a designing effect (k = 1.5), n₁ = n₀*k = 384*1.5 = 576.

Finally, considering the non-response rate of 10%, n_f = (n₁) + (n₁*0.1) = 576 + 58 = 634.

Where, n₀ was the initial sample size,

n₁ was the sample size calculated in consideration of the design effect (k).

n_f was the final sample size, including the non-response rate.

Initially, three districts were purposively selected out of 14 malarious districts in the Bale Zone. This was followed by a random selection (lottery method) of five Kebeles from each selected district. Finally, eligible households were selected by a simple random sampling technique using the lottery method from a sampling frame (Figure 2).

The heads of the selected households were then interviewed. If the head of the family was not at home during the visit, their partner or a family member aged 18 or older, was interviewed instead.

The data were collected using a structured interviewer-administered questionnaire, which was developed by reviewing the literature from previous studies in consideration of the objectives and research questions. The questionnaire was initially developed in English, translated to Afan Oromo (the local language) and then back translated to English in order to ensure consistency between the two versions. It covered contents such as socio-demographic characteristics, economic condition, history of illness and knowledge about malaria.

The questionnaire was designed with simple instructions that improved its clarity, conciseness and logical flow while reducing ambiguity, repetition and the inclusion of extraneous items. To reduce coding errors, all possible responses to questions were pre-coded with ‘other specify’ as an option, except for continuous variables such as age, monthly income and family size, where open-ended questions were used.

Prior to data collection, the questionnaire was pretested on 5% (n = 32) of the sample size from the non-sampled kebele. The pre-test was useful in determining the feasibility of the actual study, the clarity of the questions and the timing of questionnaire completion. The questionnaire was also submitted to relevant specialists in the subject, and it was finalised after required changes were made to incorporate all opinions received during the process.

To ensure high-quality results, the questionnaire’s validity and reliability were assessed. We used exploratory factor analysis (EFA) with the assistance of a statistician to confirm the construct validity of each item. To investigate the relationships between different questionnaire items, the related items were aggregated into a factor. The Keiser–Meyer–Olkin (KMO) Measure of Sampling Adequacy (MSA) criterion was used to determine the sampling adequacy of the malaria knowledge assessment questions.

The overall KMO-MSA was 0.767 for 17 items (Table 1). This value was higher than the minimum acceptable level (0.600) and ensured that the sample of items was adequate. The value for the determinants was 0.007, and that of the Bartlett’s test of sphericity (BTS) was χ² = 3089.541.

The Cronbach’s alpha coefficient was calculated to assess the extent to which research respondents could deliver identical responses to the same set of questions asked several times under the same conditions (reliability). The predetermined cutoff for the Cronbach’s alpha coefficient value was 0.70. In the current study, the Cronbach’s alpha coefficient value for 17 items used to measure the knowledge level was 0.779. Of the 17 items, 12 items that explained 89.4% of the variance in the principal component analysis (PCA) were used for composite index construction (Table 2).

Six data collectors were recruited to conduct quantitative data collection. They were master’s degree holders in public health and nursing professionals working in public universities. They had four to seven years of experience conducting research activities and data collection in similar settings.

After the final version of the questionnaire was formatted and all logistics were made ready, the actual data collection took place from 05 March 2018 to 25 March 2018, using an interviewer-administered questionnaire.

Each complete questionnaire was checked item by item to ensure that the data were error-free and complete. The data were entered in the Statistical Package for Social Science (SPSS) version 28. Each item used to assess knowledge was dichotomised and coded ‘0’ for incorrect and ‘1’ for correct responses. To compute the overall knowledge score, a composite index was constructed based on the PCA. Descriptive statistics such as mean, proportions and percentages were calculated. Inferential statistics including crude and adjusted odds ratios with 95% confidence intervals were computed using binary logistic regression to investigate a relationship between the dependent variable (knowledge level) and independent variables including sociodemographic characteristics (residence, age, gender, marital status, ethnicity, religion, educational status and occupation), history of malaria infection (current illness and ever contracting malaria) and family information (family size, presence of children younger than five years-old and pregnant women).

The model fit was checked using the Hosmer–Lemshow goodness-of-fit test, and multicollinearity was checked using the variance inflation factor (VIF) and tolerance. Nine variables that showed a significant association with knowledge level at a p-value of less than 0.05 during bivariable logistic regression analysis were included in the multivariable logistic regression model. The presence of an association was declared at a p-value of less than 0.05.

Ethical approval and an ethical clearance certificate (REC-012714-039) were obtained from the higher degree committee of the University of South Africa (Unisa). A letter of permission was obtained from the Oromia regional health bureau and communicated to the zonal health department and district health offices. The respondents were informed about the study’s purpose, benefits, potential effects (if any), their freedom to participate or not and privacy and confidentiality concerns, and written consent was obtained. To ensure anonymity, study respondents were interviewed at their residences. In terms of information confidentiality, no identifiers of the respondents’ identities were recorded; also, all data were securely stored to prevent unauthorised access.

The study had a total of 634 respondents. The study respondents’ mean age was 35.3 years, with an 8-year standard deviation (s.d.). As shown in Table 3, most (40.7%; n = 258) of the study respondents were in the age group of 31–40 years; more than half were females (55.8%; n = 354) and about three-quarters (76.3%; n = 484) were currently married. Slightly more than half (52.7%, n = 334) did not attend formal education. About 59.6% (n = 378) and 65.9% (n = 418) of the households had children aged below 5 years and pregnant women, respectively. Nearly four-fifths (79.2%; n = 502) were Oromo by ethnicity, 76.3% (n = 484) were Muslims, 45.1% (n = 286) were farmers and 13.23% (n = 60) earn a monthly income of only 150–1000 Ethiopian birr (Table 3).

The study respondents demonstrated good knowledge of the cause of malaria, with the exception of a few misunderstandings. All study respondents attribute malaria transmission to mosquito bites. However, in addition to the mosquito bite, about 5.7% (n = 36), 2.8% (n = 18) and 0.9% (n = 6) attributed malaria transmission to staying long in the sun, lack of rest or not sleeping enough and drinking too much alcohol, respectively (Table 4).

Unlike the cause of malaria, most study respondents lacked knowledge of most of the signs and symptoms of malaria. Of the eight common signs and symptoms of malaria presented to the study respondents, only three were known by more than 50% of the respondents, which indicates a lower level of understanding in this regard (Table 4).

Malaria prevention methods relatively known by the majority (more than 50% of the respondents) include destroying malaria breeding sites 86.8% (n = 550), sleeping under insecticide-treated bed nets (ITNs) 88.6% (n = 562) and using indoor residual sprays (IRS) 58.7% (n = 372), while the lesser-known prevention methods include the use of prophylactic medications (29% n = 184), mosquito repellent cream (16.7% n = 108) and immediate health seeking for suspected malaria cases 46.1% (n = 292) (see Figure 3).

The overall knowledge was measured based on the composite index that was constructed from 17 items used to assess the respondents’ knowledge on the symptoms of malaria, the causes of malaria and the method of malaria prevention after performing PCA. Slightly less than half (44.2%, n = 280) had good knowledge of malaria; the rest, 54.8% (n = 354), had poor knowledge.

In the current study, all respondents attributed malaria transmission to mosquito bites, while some study respondents additionally attributed it to staying long in the sun, lack of rest or not sleeping enough and drinking too much alcohol. In agreement with this study, studies conducted in Gabon, India and Cameroon revealed that the majority attributed the cause of malaria to mosquito bites.^19,20,25 Studies conducted in southern Africa, Gabon and India also disclosed the presence of individuals who think malaria spreads by eating stale or bad food, fresh fruit, maize or sugarcane, or by drinking contaminated water.^19,20,26

The current study revealed that water-containing pits, the bushes in the home environment, poor environmental sanitation and unsanitary human actions, including poor liquid waste and latrine management, were sources of malaria transmission. Other studies from Cameroon and India also reported that stagnant or dirty water, gutters, muddy sludge and puddles were sources of malaria transmission.^20,25 This means that the community is knowledgeable about malaria causes; however, their knowledge about the situations that facilitate malaria transmission lacks consistency. This may negatively affect the prevention practices; thus, a comprehensive community-based knowledge transfer intervention is very crucial.²⁶

In contrast to the current study, another study conducted in rural Zambia reported drinking dirty water and touching a malaria patient as routes of malaria transmission.²¹ The possible explanation for the observed difference may be because of the difference between the two studies, including setting (Ethiopia vs. Zambia), the sample size (634 vs. 75) and the scope of the study (knowledge assessment vs. knowledge, attitude and practice [KAP] survey). Unlike other studies, the current study did not report the misconception that malaria can be transmitted by walking in the rain or through sexual contact.^19,20 The possible explanations include differences in study respondents, setting and scope of the studies. The current study was conducted in three rural districts, focusing on the assessment of knowledge of signs and symptoms, causes and prevention methods among the general population. However, the study in India was conducted in an urban setting, focusing on KAP.²⁰ Furthermore, the respondents of the Gabon study were reproductive-age women, and the focus of the study was KAP on prevention.¹⁹

These findings clarified the presence of misconceptions about malaria that should be considered, which calls for the design of a contextual intervention strategy and effective implementation for a better understanding of the community at the grassroots level.

The current study revealed that the community had limited knowledge of the signs and symptoms of malaria. Only three signs and symptoms were known by more than 50% of the respondents, which include fever (95.3%), headache (79.2%) and talking nonsense (53.9%). The studies that agreed with the current study also revealed that fever, headache, vomiting and appetite loss are the most common malaria symptoms known by the majority of the respondents.^20,26,27 On the other hand, symptoms like pain, backache, fatigue, vomiting and anorexia that were reported by other studies^26,27 were not reported by the current study. The difference may be because the current study differs from others in terms of study setting (community-based) and the sample size (634 households). The study conducted in southern Africa was multinational (Zimbabwe and Zambia)²⁶ and had a large sample size (2066 households), while that of Gonder (Ethiopia) was facility-based with a smaller sample size (390).²⁷

In general, even though the community is aware of a few very common symptoms, their knowledge of malaria symptoms is limited. Thus, detailed gap identification and appropriate intervention are necessary to overcome the associated consequences.

The majority of the survey respondents mentioned destroying malaria-breeding sites, sleeping under ITNs and using IRS as malaria prevention methods. However, only a few respondents mentioned the use of prophylactic medications, mosquito repellent cream and immediate health-seeking for suspected malaria cases. The study conducted in South Gondor also revealed that the majority of the respondents mentioned using ITNs and IRS, while a few individuals mentioned fumigation.²⁸ Furthermore, the study conducted in Tanzania revealed that the majority of the study respondents mentioned ITN use, IRS and destroying breeding sites as measures of malaria prevention.²⁹ These findings suggest a gap in awareness or access to other effective prevention strategies, such as prophylactic medications and timely health-seeking behaviours. Public health initiatives must address these gaps, ensuring that communities are aware of all available techniques for efficiently combating malaria.

In contrast, neither of the two studies reported other malaria-prevention measures such as the use of prophylactic medications, the use of a mosquito repellent cream or immediate health seeking for suspected malaria cases.^28,29 The difference may be because of the variation in the study setting, design and time frame. The findings imply that the community in the current study area has demonstrated moderately good knowledge on malaria prevention measures, even if the commonly known measures are few. Thus, additional efforts are required to improve.

Malaria awareness is low in the current study area. The majority of the survey respondents (54.8%) lacked knowledge. This finding is consistent with the findings of a study conducted in Cambodia,³⁰ which found that the community’s understanding of malaria epidemiology and vector ecology was limited. This could be attributed to the lower literacy levels seen in the two research populations.

On the other hand, the findings of the current study are lower than the results of other studies conducted in southern Africa and Cameroon.^26,31 The difference may be because of the variation in the characteristics of the population, like social structure, economic condition and level of literacy. The other possible explanation could be methodological differences between the current study and other studies. This implies that the overall knowledge in the study area is low, and this calls for improvements in information, education and communication to enhance the knowledge of the community. As pointed out by others, to prevent malaria infection and promote malaria-free zones, understanding the community’s knowledge of malaria control is essential.²⁹ It is crucial that educational initiatives focus primarily on best practices to increase local populations’ adherence to malaria-prevention initiatives.³¹

The present study shows that those persons who completed elementary school and secondary school or higher possessed knowledge levels 3.25 and 9.55 times more, respectively, than those who were illiterate. This finding aligns with the study conducted in South Gonder, Ethiopia, which indicated that the highest knowledge scores were observed among persons with college education and above (84.3%) and those with high school education (82.9%) compared to uneducated individuals.²⁸ Another institutional-based cross-sectional study conducted at Adis Zemen Hospital, Northwestern Ethiopia, also revealed that those who attained primary education were more knowledgeable than those who were unable to read and write.³⁰ This implies that awareness creation and knowledge transfer activities should primarily target people with no formal education and those with lower educational levels to improve their knowledge of malaria and its prevention.

Orthodox Christians were 64% less likely to have good knowledge of malaria compared to their Muslim counterparts, while no statistically significant difference was observed between Muslims and Protestants. Divorced individuals were 5.91 times more likely to be knowledgeable compared to widowed individuals, while no significant difference was observed between single, married and divorced individuals. Contrary to the current study, a cross-sectional household survey conducted in Nyabondo, Western Kenya, reported that marital status and religion were not significantly associated with knowledge related to malaria.³² The possible explanation for the observed variation could be because of the differences between the two studies in terms of the sample size and the type of data. The current study utilised quantitative data from 634 households, while the other study utilised both qualitative and quantitative data from 80 households.

People from households with children younger than five years-old were 3.32 times more knowledgeable compared to their counterparts; people from households with pregnant women were 2.84 times more knowledgeable compared to their counterparts. Those from households sizes six to 10 family members, and 11–15 family members were 2.03 and 19.82 times more knowledgeable compared to those with one to five family members. This finding suggests that larger families and those with young children or pregnant women are more engaged in acquiring knowledge, possibly because of the heightened need for information related to child-rearing and prenatal care. Consequently, targeted educational resources may be beneficial in further enhancing awareness and understanding among smaller families.

The current study contradicts previous research that found statistically significant relationships between malaria knowledge and other factors such as age, gender, family monthly income, place of residence and occupation.²⁸ The disparities may be because of the differences in the study population and the scope of the studies. The respondents of the current study were all adults, while those of the other study were adult men only. Furthermore, the current study specifically focused on knowledge assessment, unlike other studies that assessed knowledge, attitude and practice of malaria control.

The existing literature was used to inform the construction of the data-collection tool, which ensured its reliability and statistical validity. During the data-collection period, repeated visits to households ensured the highest response rate. Furthermore, the study produced very informative findings on malaria signs, symptoms and preventive knowledge in rural areas.

Because of the cross-sectional character of this study, determining a cause–effect relationship is difficult. Furthermore, the knowledge assessment primarily focused on three aspects: knowledge of malaria causes, signs and symptoms, and prevention methods, with little emphasis placed on sources of information and malaria definition. However, these elements were considered for further qualitative investigation.