Caregivers’ perception of malaria and treatment-seeking behaviour for under five children in Mandura District, West Ethiopia: a cross-sectional study

A community based cross-sectional survey was conducted in December 2014 in Mandura District in Western Ethiopia. Mandura is one of the 20 districts in the Benishangul-Gumuz Region of Ethiopia. The region is one of the two regions of the country where the risk of malaria is highest in almost the entire region [4]. The district is situated at the distance of 530 km from Addis Ababa (the capital city of the country). It is administratively organized into 20 kebeles (the lowest administrative units in the country).

According to the 2007 Ethiopian national census, the population of Mandura District was 40,746, of which 8122 were under five children. The majority of the inhabitants practiced traditional beliefs, with 47.76% of the population reporting they observed this belief, while 39.26% of the population said they practiced Ethiopian Orthodox Christianity, and 7.59% were Muslim [20]. The district lies at an altitude between 1050 and 1400 m above sea level, and has an average temperature 31–37 °C and annual rainfall of 900–1200 mm.

The required sample size was calculated using a single population proportion formula considering the following assumptions: a 95% confidence level, 5% margin of error, an estimated proportion that 80.87% of caregivers are aware of the cause of malaria [21], considering 5% non-response rate and two design effect. The resulting estimated sample size for the study was 503 households.

A two-stage sampling technique was used. First, six kebeles were selected by using simple random sampling from 20 kebeles in the district. Then, the total sample size was allocated proportionally to each of the kebeles based on the number of households with children under the age of five. A sampling frame for list of households with children under the age of five from each selected kebele were prepared in consultation with health extension workers. Then the total sample size was allocated proportional to the size of the households with children under the age of five at each kebele. In each selected kebele, a sample of households with children under the age of five was selected through simple random sampling using lottery methods. The desired target population for the study were caregivers of under five children. Details of the youngest one were collected if there was more than one child who had had fever in a household in last 2 weeks.

The data were collected using structured interviewer administered questionnaire. Trained research assistants, with a diploma in nursing administered a questionnaire to participants individually in their homes. The questionnaire was designed in English, and translated into local languages (Amharic, Gumuz, Sinash and Agew) widely spoken in the area to facilitate the interviews. The questionnaire covered caregivers’ knowledge and practices related to malaria transmission, symptoms, treatment and preventive methods. In addition, caregivers were asked whether the child experienced an episode of fever within 2 weeks preceding the interview and related practices of treatment-seeking behaviour. The other variables considered to be associated with treatment-seeking included socio-demographic factors, perceived susceptibility, severity, barriers, and benefits as described in the HBM [22].

In order to reduce recall bias, households without fever in the last 2 weeks were just interviewed for socio-demographic characteristics and malaria knowledge but for treatment-seeking and other related factors. Two supervisors and eight data collectors participated in the data collection process. Three days training was given to the data collectors and supervisors. The supervisors checked each questionnaire and discussed with data collectors for any clarifications.

Varying methods of calculating composite scores and categorizing them from quantitative knowledge data have increasingly been used [19]. In this study, caregiver’s malaria knowledge was measured based on a cumulative response on ten questions comprised of sign and symptoms, causes, mode of transmission, prevention and treatment of the disease. Mean score for the ten items was calculated as 7.1 (±SD 2.2) and categorized as poor and good knowledge based on the mean score below and above, respectively.

According to the HBM, the likelihood that an individual engages in a given health (enhancing) behaviour is a function of the perceived susceptibility, perceived severity, perceived treatment benefits, perceived barriers to treatment, cue to action and self-efficacy. There were six questions for perceived susceptibility, seven questions for cues to action and five questions to measure caregivers’ perceived severity of malaria. Perceived benefits of malaria treatment, perceived barriers to treatment-seeking and self-efficacy was measured using five, eight and three, respectively. The responses were ranked using 4-point Likert scale (1-strongly agree, 2-disagree, 3-agree, and 4-strongly agree). The mean scores for all constructs were computed and dichotomized into high and low. Respondent scores below the mean were labeled as having low perception of severity, susceptibility, benefits and barriers and those respondents who score mean and above categorized into high perceived severity and susceptibility of malaria disease, high perceived benefits of malaria treatment and high perceived impediments to treatment-seeking. Similarly, respondents who score below mean to questions related to cue to action and self-efficacy were categorized to have low cues to action and those who score mean and above were categorized as having high self-efficacy and high cues to action. The reliability of items of the scales were evaluated using Cronbach’s alpha. All the Cronbach’s alpha were found to be above the acceptable internal consistency level of 0.7 and above [23].

The outcome variable was caregivers’ treatment-seeking behaviour, operationally defined as seeking treatment from trained personnel or at a health facility. Caregivers were considered as seeking treatment when they visited any health facilities (Hospital, Health Center, Health Post, Private Clinic) after perceiving the illness of their child within 2 weeks of data collection period.

Data were entered and cleaned using EPi-Info version 3.5.2 (Centers for Disease Control and Prevention, Atlanta, GA) and exported to SPSS version 20 program for analysis. Descriptive statistics were used to summarize all variables of interest in the study population. The statistical association between seeking treatment and each of the explanatory factors were first checked using binary logistic regression. Multivariate logistic regression was used to control confounding effects. P value ≤0.20 was taken as a cut-off point for selecting variables for the logistic regression model. Both backwards and forward logistic regression was performed and gave similar result. P values ≤0.05 were considered statistical significant in the final model. The results of all logistic regression analyses are reported as odds ratios (OR) with 95% confidence intervals (CIs). The final multivariate model was tested for goodness of fit with the Hosmer–Lemeshow test.

A total of 491 caregivers were interviewed making the response rate 97.6%. The socio-demographic characteristics of the respondents are shown in Table 1. Most of the caregivers were female (96.9%), married (96.1%) and residing in rural villages (65%). The median age was 29 years (IQR 25:32) and 61.5% were in the age group 25–34 years. In terms of educational background, 75.6% of the caregivers had no formal education, and 63.1% were not able to read and write. Most of the participants were from Gumuz (42.4%) and Agew (42.2%) ethnic origin. Over 56% of the participants were Orthodox Christians, while 24% were followers of traditional religion called ‘Gumuza Muses’. Regarding their income, about 194 (39.5%) caregivers had monthly income less than 500 Ethiopian Birr (ETB), followed by those earning 501–1000 ETB per month.

A series of questions regarding risk factors, symptoms, prevention, early detection and treatment options were asked to determine caregivers’ level of knowledge about malaria. Majority (70.9%) of the participants associated malaria with mosquito bite. However, some associated malaria transmission with cold wet weather (21.8%) and consumption of contaminated food and drinking (1.2%). Mosquito breeding site was reported by majority (84.5%) as stagnant water and 4.9% reported running water as common breeding site.

Most respondents were aware of the common signs and symptoms of malaria in both adults and children. The most common frequently perceived signs and symptoms of malaria included hot body/fever (94.1%), followed by headache (82%). The vast majority (96.7) of the study participants knew that malaria is preventable. Using bed net (91%), eliminating mosquito breeding sites (57.2%), DDT spray (51.5%) and using personal protection (35.2%) were the most frequently mentioned prevention methods. Using the sum of all knowledge items, 45% of the participants were found to have sufficient knowledge about malaria. About 88.4% of caregivers reported modern health institution as the best option to treat malaria in under five children, while traditional medicine was mentioned by 11.4% of the participants.

Overall, 56.2% of caregivers had low perception about susceptibility to malaria infection and 51.1% had low perception about severity of the disease. Nearly 63.7% of caregivers had low perceived barrier to seek treatment for under-five children. However, 51.3% of caregivers had low perception of the benefits of seeking treatment. More than 60% of respondents had low cue to action and 58.7% had low self-efficacy. Details about the constructs of the HBM are shown in Table 2.

From the total number of children under-five (197) with fever in the past 2 weeks, 173 (87.8%) were brought to treatment. Of those children brought to treatment, 129 (74.6%) received treatment in the public health institution and 44 (25.4%) received treatment from local private providers like drug vendors. Some of the children (18.5%) received treatment at home before they were taken to health facility. Majority (61.3%) of the children sought treatment at least 24 h after the onset fever. The mean hours the child brought to the health institution was 5 days after the onset of fever. Moreover, some of the caretakers (18.5%) first consulted traditional healers and used home remedies for malaria treatment in their children.

In the bivariate analysis, among the socio-demographic characteristics of respondents, age, residence, ethnicity, occupation, and religion were significantly associated with malaria treatment-seeking behaviour. Of the malaria perception variables, perceived susceptibility for malaria, perceived barriers to treatment-seeking and cues to action were found to be significantly associated with malaria treatment-seeking behaviour.

The results of multivariate logistic regression analysis showed that five explanatory variables were independently associated with malaria treatment-seeking for under-five children suspected of having malaria (Table 3). Age of caregivers appears to be a significant determinant for seeking treatment. Younger caregivers (age <30 years) were 3.40 more likely (95% CI 1.27–9.10) to seek treatment as compared to those >30 years of age, adjusted for other variables. As compared with urban residents, caregivers from rural area were more likely to seek treatment for under five children, adjusting for other variables (AOR 2.80, 95% CI 1.01, 7.70).

Caregivers’ malaria knowledge was also a significant predictor of treatment-seeking behaviour. Children whose caregivers had good malaria knowledge were more likely to be taken for medical care (AOR 4.65, 95% CI 1.38–15.64). Caregivers who had high perception that their child could be susceptible to malaria were more likely to seek treatment (AOR 3.63, 95% CI 1.21–10.88). Moreover, caregivers who had high perceived barriers to seek treatment were less likely to bring their children for medical attention (AOR 0.18, 95% CI 0.06–0.52).