The data presented in this manuscript are drawn from survey data for the Communication for Health (C4H) project- a five-year (2015–2020) project in Ethiopia funded by the United States Agency for International Development focusing on integrated social and behavior change communication (SBCC) in multiple health areas including malaria . The goal of the C4H project was to improve health practices through standardized SBCC interventions in four major regions of Ethiopia: Amhara, Oromia, Tigray, and SNNPR. The project conducted formative studies, including a baseline survey in 2016 targeting all woredas in the four study regions and a follow-up survey in 2019 assessing the exposure and effectiveness of the interventions among 160 woredas where the interventions had been implemented. The C4H project study population was women of reproductive ages (15–49 years) who were the primary beneficiaries of the project’s social and behavior change interventions. The C4H project implemented integrated social and behavior change activities including mass-media and interpersonal communication interventions targeting the general population. Health messages promoting timely care-seeking were included in radio spots, a weekly drama/education radio series; vans mounted with loudspeakers disseminate messages in intervention areas; offline videos focusing on postnatal care, essential newborn care, seeking care for newborn illnesses, immunization, and malaria prevention and treatment; a mobile application that contain key health messages including prevention and care seeking and complementary training of community health workers to conduct effective house to house education and outreach activities such as cue-cards and family health guide.
The study data are drawn from two cross sectional studies in 2016 and 2019. In each round, C4H applied probability proportionate to size (PPS) sampling to select project woredas and enumeration areas (EAs). In each EAs, all households were listed to identify eligible households (with women aged 15–49) from which 35 eligible households were randomly selected. The C4H data collectors interviewed all women aged 15–49 face to face from the eligible households using electronic data collection tools. A total of 2,770 women were interviewed at baseline and 1,773 at the follow-up survey, because the intervention was implemented in fewer communities than the potential number determined at baseline. Among all women surveyed, 2453 were caregivers of children under five years out of which 479 caregivers had a child or children with a fever in the two weeks prior to the survey. See the study flow chart in Supplemental Fig. 1 for more details.
Assuming a 5% type I error, the sample of 479 women caregivers of children under five years old who reported fever results in 86% power to estimate the percent of caregivers who sought care promptly with 7% precision (estimate ranging from 44 to 51%).
The study received ethical approval from the Ethiopian Public Health Institute (EPHI) ethical review committee, Addis Ababa, Ethiopia, and the institutional review board (IRB) of the Johns Hopkins Bloomberg School of Public Health (JHSPH), Baltimore, Maryland, USA (JHSPH IRB # 00,007,138 and EPHI-IRB-173–2019). All study procedures were performed in accordance with the Declaration of Helsinki. Informed consent was obtained from all study participants. Participants less than 18 years provided informed assent after parental consent was obtained.
The study interviewed a total of 4,543 women aged 15–49 (94% response rate). Of these, 2,770 (92% response rate) were interviewed at baseline; because the intervention was implemented in fewer communities than the potential number determined at baseline, 1,773 (97% response rate) were interviewed at follow-up surveys. The questionnaires aimed to collect information on common themes, including sociodemographic characteristics, malaria related behavior and psychosocial factors, gender equitable norms, and access to health education messages. The study administered questionnaires in the Amharic, Oromia, and Tigrigna languages.
The study measured participants’ care-seeking behaviors and psychosocial factors accounting for contextual factors, including demographic characteristics, vulnerability, and gender equitable norms. The survey asked all caregivers, “Has [child’s name] been ill with a fever at any time in the last two weeks?” Any care-seeking was defined using the question, “Did you seek advice or treatment for the fever from any source?” This study defined overall care-seeking as “prompt,” “delayed,” or “none,” based on the time elapsed before seeking care.
The key outcome variable was prompt care-seeking for fever; explored by the survey question, “How long after the onset of fever did you seek treatment?” This study defined prompt care-seeking as seeking treatment within 24 h or less and Delayed care-seeking as seeking treatment more than 24 h after the onset of fever.
Psychosocial factors explored in this study included ideational factors relevant to care-seeking as well as overarching contextual factors. Ideational factors include cognitive, social and emotional constructs, which are quantitatively assessed using a battery of Likert scale questions to capture these latent variables . A limited number of ideational psychosocial factors related to care-seeking (knowledge, self-efficacy, response efficacy, attitudes, involvement in decision-making, and social support in the household) were explored as follows:
Knowledge: The study assessed this by exploring participants’ awareness of malaria, signs and symptoms, cause, and prevention measures. Specific questions included the following:
◦ “Have you ever heard of an illness called malaria?”
◦ “What signs or symptoms would lead you to think a person has malaria?”
◦ “What do you think is the cause of malaria?”
◦ “How can someone protect themselves against malaria?”
Self-efficacy: This refers to an individual's belief in one’s capacity to execute behaviors necessary to produce specific performance attainments . This study assessed participants' self-efficacy for care-seeking and malaria prevention using the following statement: “I can take my child to treatment within 24 h of onset fevers, and I am able to have children under five years sleep under an ITN each night.”
Response efficacy: This refers to a person's attitudes as to whether the recommended action step will avoid the threat. The study sought to understand if participants understood the benefits gained if they engaged in prompt care-seeking or malaria prevention behavior. This study proposed statements to participants like, “Having my children sleep under an ITN each night will prevent malaria,” and, “Seeking treatment for my under five children within 24 h of onset fever improves chances of recovery and survival.”
Attitudes towards care-seeking: This study assessed this using the statement, “I should seek treatment for children under five years within 24 h of onset of fever.” Responses for self-efficacy, response efficacy, and attitudinal questions were recorded on a four-point Likert scale of “Strongly agree,” “Agree,” “Disagree,” and “Strongly disagree.”
Involvement in decision-making: this study explored this using the question, “Who usually makes decisions about health care for yourself: you, your (husband/partner), or you and your husband/partner jointly?” Respondents who noted they made decisions by themselves or jointly with their partners were considered involved in decision making.
Social support: This study assessed this by asking about the level of spousal support on household chores using the question, “Does your husband help you with household chores like looking after the children, cooking, cleaning the house, and doing other work around the house?” This question served as a proxy for actual spousal support on caregiving.
Community norms: This study measured caregivers’ perceptions regarding gender equitable norms using a 21-item Gender Equitable Men scale validated in Ethiopia . The scale measure attitudes toward gender norms in intimate relationships or differing social expectations for men and women. It includes four sub scales related to physical violence, sexual relationships, reproductive health and disease prevention, and domestic chores and daily life. Responses to all 21 questions were condensed into a composite score, split into tertiles, and then subdivided into two categories of perceived gender equity: high perceptions of equitable gender norms (first tertiles) and moderate or low perceptions of equitable gender norms (second and third tertile).
This study condensed all ideational psychosocial variables into a composite score, referred to as care-seeking ideation, ranging from 0–12 (Cronbach's alpha of 0.7). Caregivers with a care-seeking ideation score of 9 or more (greater than the median) were categorized as having a high care-seeking ideation, while those with a score of 8 or less were categorized as having low care-seeking ideation. Supplemental Table 1 summarizes the psychosocial variables used in generating the malaria care-seeking ideation.
The study also explored the following overarching contextual psychosocial variables: caregivers’ region (Amhara, Oromia, SNNPR, Tigray), residence (urban versus rural), malaria transmission zone (high versus low/no), age (15–24, 25–34, and 35–49 years old), education (none versus primary or higher), religion (Christian versus non-Christian), marital status (married or cohabiting versus not), and wealth quintile based on ownership of household assets (richest, richer, middle, poorer, poorest).
Access to ITNs within the household was defined as the number of bed nets per household member. Given the assumption that a bed net can be used by up to two people , a value of ≥ 0.5 bed nets per household member was defined as adequate ITN supply, while < 0.5 was classified as inadequate. Exposure to health communication messages was assessed by asking whether caregivers had the Family Health Guide (FHG), a national health communication tool given to all families. FHG comprises more than 79 messages with cues to action and illustrations focusing on multiple health areas including malaria.
Additionally, a vulnerability index was assessed in the survey using four states the participant said they experienced in the last 12 months: (1) lacked adequate nutrition, (2) lacked shelter/house to stay in, (3) was unable to afford to send children to school, and (4) lacked sufficient money to buy medicines/medical treatment. Responses included never (coded as 1), rarely (coded as 2), sometimes (coded as 3), and often (coded as 4). We computed a composite score from the sum of the four responses. The vulnerability index was divided into two categories: non-vulnerable (score ≤ 7) and vulnerable (8–16), with high scores indicating a greater level of poverty. Finally, the study included cross-sectional survey timing (2016 baseline versus 2019 follow-up) as a contextual variable to account for temporal changes that might explain inter relationships between psychosocial factors and care-seeking.
Since we performed analysis on children under five years with a fever in the two weeks preceding the survey, we explored any clustering effect that might occur with multiple children under five years in the same household/caregiver. No evidence of clustering in the data (intraclass correlation = 0.015) appeared. The study used chi-square, t-tests, Analyses of Variance for bivariable tests of association and exploratory analysis, to compare psychosocial factors across the spectrum of care-seeking. We employed multivariable logistic regressions to explore psychosocial factors associated with both care-seeking and prompt care-seeking. Data management and analysis was performed using Stata software, version 16 (Stata Corporation, College Station, TX, USA) and Excel 2016 (Microsoft Corp, Seattle, WA, USA). We weighted the data using the svyset command in Stata created from the probabilities of woreda, EA and household selection to ensure the data was representative of the study population.