Study design and period
A community-based cross-sectional study was employed from January to February 2020.
Study setting
This study was conducted from January to February 2020 in the Nifas Silk Lafto sub-city, Addis Ababa. The estimated area of Addis Ababa city is 174.4 km2 and has an estimated density of 5535.8 people per square kilometer [10]. The health service area coverage of the city is 90%. The rate of health service use is 1.71 per capita per year [11]. There are 13 districts, 465,428 population, 2 private hospitals, and 10 health facilities in the Nifas Silk Lafto Sub-city.
Population
The study population was mothers with under-five children who had an illness during the previous 2 weeks of data collection. In the absence of a mother, another child caregiver in the household was interviewed to reduce the non-response rate.
Sample size determination
To estimate the required sample size, single proportion population formula was used with the following assumption, 30% proportion of health care seeking to the ill child [12], 5% level of significance, 95% confidence interval (Z α/2 = 1.96), 5% absolute precision or margin of error, 2 design effect and 10% non-response rate. Finally, the sample size required for the study was 875.
Sampling methods
A multi-stage sampling technique has been employed to recruit study participants. Six woredas were selected by a random sampling method from a thirteen woreda (the smallest administrative area in the city that contains five to six sub-districts or Ketena). The total sample size is proportionally allocated to each woreda. To get study participants from selected woreda cluster methods were employed. Three Ketenas were considered to take mothers with sick children. The clusters were known to be Ketenas (the smallest administrative unit in the Woreda) with a population of at least 4000, although often it depends on the geographical location. All mothers with under-five children who were ill in the previous 2 weeks were interviewed until the sample size per woreda was obtained.
Operational definition
Healthcare-seeking behavior
defined as caregiver’s response for signs and symptoms of illnesses to reduce the severity, complication, or even death after recognized child’s illness and if caregiver reported visiting any health institutions considered as having healthcare-seeking behavior.
Fever
If mother perceived as fever or hot body for any child 2 weeks preceding the survey.
Diarrhea
If the caregivers described their sick children had three or more loose or watery stools per day at any time within the 2 weeks before the survey.
Cough
all cases who had cough less than 2 weeks preceding the survey as perceived by mothers or caretakers.
Caregiver
is a mother or equivalent family member who is primarily responsible for caring for a child and who could explain enough about the sickness-related behaviors of a child.
Common childhood illness
common childhood illness includes diarrheal, fever, and cough.
Knowledge score
respondents who scored above or equal to the mean were labeled as having optimal knowledge and those who scored below the mean were labeled as having poor knowledge about childhood illnesses [13].
Data collection tool and procedure
Pre-tested and standardized questionnaires were used to interview child caregivers. The questionnaires were developed through reviewing different literature [14,15,16]. To check its consistency, the questionnaire was initially prepared in English and later translated into the local language (Amharic) and back to English. The questionnaire contains three-section, socio-demographic characteristics, child’s illness-related characteristics, and healthcare-seeking behavior of caregiver. Six diploma nurse data collectors were recruited and trained for 2 days on the objectives of the study, and the approaches to data collection. Field supervision and follow-up were made by the principal investigator and two Bsc nurses to monitor the progress and quality of the data collection process.
To ensure data quality, different methods of data quality assurance were employed. Two days of intensive training were given to data collectors and supervisors. A pre-test was performed in a 5% of sample size in Arada- sub city woreda 01. The collected data was checked daily base on the field to avoid incompleteness and inconsistency by principal investigators and supervisors.
Data management and analysis
Each questionnaire was manually checked, coded, and entered into EPI info version 7 statistical software and exported to a statistical package of social science (SPSS) Version 20.0 software for analysis. Univariable, bivariable and multivariable analyses were performed. Percentage, frequency, and mean were calculated for the socio-demographic characteristics of caregivers. Finally, binary logistic regression models were fitted as a multivariable analysis tool to obtain odds ratios, 95 percent confidence intervals, and control confounders. Variables with less than 0.05 p-values in the multivariable logistic regression analysis with a 95% confidence interval were considered significant. The analysis results were presented using tables, figures, and text as appropriate.