Study design and area
A community based cross-sectional study was conducted in Aneded district from February to March 2016. Aneded district located 283 km from Addis Ababa, the capital city of Ethiopia, and 282 km from Bahir Dar, the capital city of the Amhara National Regional State. The 2007 Ethiopian census reported the population to be 104,053 (50,991 males and 53,062 females) and the total number of children under-five at 12,351. It has 19 rural Kebeles and one urban kebele (the smallest administrative unit) and each Kebele is divided into Ketenas. There are 5 health centers, 20 health posts, 3 private clinics and 2 drug stores/pharmacies in the district. Health centers are public institutions that provide diagnostic and therapeutic services, staffed by mid-level health professionals (Diploma and Degree level health professionals). Ideally, five health posts within each health center. Health posts are staffed by health extension workers and low level health professionals. Staff working at health centers and health posts can assess and treat a sick children using integrated management of neonatal and childhood illnesses (IMNCI) protocols. The clinics in the district provide diagnostic and therapeutic services while the drug stores/pharmacies are established to sale drugs/medicines.
Sample size determination and sampling procedures
The sample size was determined using two population proportion formula by considering different factors that affected mothers’ health care seeking behavior during child illness from previous study with Open Epi software version 2.3. To find adequate sample size, we took factor that gave largest sample size among the factors. Hence, the final sample size was calculated by considering (P1 = 69%) the proportion of media exposed mothers who sought health care and (P2 = 47.7%) the proportion of non-media exposed mothers who sought health care for their children illnesses [13], 5% level significance, 80% power, design effect of 2 and 10% non response rate. Therefore, the total calculated sample size was 410 mothers.
Mothers or caregivers who were living in rural district and who had a child or children under 5 years of age with history of any common childhood illness like diarrhea, fever, and/or ARI three months preceding the survey were included. Then, to select study participants, multistage cluster sampling technique was used. Among the 19 rural Kebeles in the district, five Kebeles (25% of the study area) namely Gudalem, Amberzura, Daget, Yewobie, and Nefasam were selected first using simple random sampling technique (lottery method) and then a minimum of 50 % of Ketenas from the selected Kebeles were selected using a simple random sampling method. The sample size was proportionally allocated to each kebele by considering the total population in each kebele. Since cluster sampling technique was used, households were visited to assess the presence of under five children who were sick within the past three months. Each households were visited until the sample size was reached. Mothers or guardians who had had sick child in the past three months were requested to participate in this study after providing adequate information and obtaining informed verbal consent. Data were collected using interviewer administered pre-tested structured questionnaire. The questionnaire draws from Andersen’s Behavioral Model [14] and a review of relevant published literature. In Andersen’s Behavioral Model access to and use of health services is considered a function of three characteristics: 1) Predisposing factors: the socio-cultural characteristics of individuals that exist prior to their illness, 2) Enabling factors: the logistical aspects of obtaining care for personal /family, 3) Need factors: the most immediate cause of health service use, perceived need and refers to how people view their general health, functional state and judgment to seek professional help (Additional file 1).
Five data collectors and two supervisors (nurses) were involved in the data collection. To assure data quality, a one day training was given to the data collectors and supervisors on the study objectives and data collection techniques. The overall data collection activity was supervised by study investigators.
Operational definitions/measurements
The outcome variable was mothers’/caregivers’ health care seeking behaviors during childhood illness. The response was dichotomized as “yes” when women had appropriate health care seeking behavior or “no” when they did not have appropriate health care seeking behavior.
Appropriate health care seeking behavior was defined as situations when women visited any health facility/institution (governmental or private or both) during common childhood illnesses. Conversely, inappropriate refers to situations when a women did not visit any of formal health sectors.
Common childhood illnesses: include diarrhea, acute respiratory infection (ARI) and fever.
Awareness of childhood illness: refers to when a mother/care giver recognizes one or more symptom questions.
Mothers’ awareness on sign of severity of illnesses: refers to when a mother/care giver mentioned one or more of the signs of a sever childhood illnesses.
Perceived illness severity: refers to when a women/care giver thought that her sick child was severely ill.
In this study, the authors dichotomized the wealth of the participants that the term ‘rich’ was used to describe those who were in the fourth or fifth quintile where as the term ‘poor’ was used to explain those who were the first three quintiles.
The informal sector refers to institutions that are not legally able to diagnosis and treat childhood illnesses and includes local/traditional healers and Holy water. The formal health sector (health care system) are public health institutions and licensed private clinics able to diagnosis and treat childhood illnesses.
Data processing and analysis
During data collection, supervisors and investigators manually checked the questionnaires daily for completeness. The collected data were entered into Epi- Data version 3.1 and exported to SPSS version 20 for data cleaning and analysis. Descriptive statistics was computed to summarize the descriptive results and presented in texts, graphs and charts. Principal component analysis (PCA) was employed to measure the level of wealth of the household. Multivariable logistic regression modeling used to identify factors associated with mother/caregiver health care seeking behavior. David W. Hosmer and Stanley Lemeshow in their second edition book entitled “Applied Logistic Regression” recommended to use a P-value of less than 0.25 as a screening criteria for variable selection for the multivariable analysis [15]. There are also other published articles in BMC journals which used a p-value of 0.2 as a cut-off point to select variables for the multivariable analysis [16,17,18]. Therefore, in this study, variables having P-value ≤0.2 in the bivariate analysis were considered for multivariable analysis. Odds Ratio with 95% confidence interval (CI) was used to determine the presence of an association.
Ethical considerations
Ethical approval was obtained from the Ethical Review Board (IRB) of University of Gondar, Institute of Public Health. Support letters were obtained from Amhara Regional State Health Bureau, East Gojjam Zonal Health Department and Aneded District Health Office. Verbal consent was obtained from each study participant.