Study Design, Setting and period
A community-based cross-sectional study was conducted in the Bench-Sheko, Kafa, and West Omo zones from May 1 to 31, 2020. These three zones are found at 585 km, 460 km and 705 km from Addis Ababa, Capital of Ethiopia, respectively, to the southwest direction. The Bench-Sheko zone is administratively divided into six woredas (districts) and two town administrations. The Kafa zone is also divided into eleven woredas and two town administrations. Likewise, the West-Omo Zone has seven woredas. The main agro-ecology of these three zones includes dega (cool and humid high lands, which account for 56.7% of the land size), kolla (warm, semi-arid lowlands, which account for 28% of the land size) and weinadega (temperate, cool and sub-humid high lands, which account for 15.3%). Their annual mean temperature ranges from 15.1°C to 27°C, and the annual mean rainfall ranges between 400 mm and 2000 mm. In these zones, there are 7 hospitals, 97 health centres and 565 health posts that provide health services for residents. According to the population projection of Ethiopia figured for 2014 to 2017, the Bench-Maji zone (the former collective name for Bench Sheko Zone and West Omo Zone before their disunion) had a total population of 847,168 (417,751 males and 429,417 females). Similarly, the Kafa Zone had a population size of 1,102,278 (541,682 males and 560,596 females) (16).
Population and Sampling Techniques
The adult population who were 18 years old and above were included in this study. Individuals who were unable to respond due to different medical problems were excluded. The sample size was determined using a single population proportion formula , where “n” stands for sample size, “zα/2” stands for the reliability coefficient of standard error at the 5% level of significance, which equals 1.96, “p” stands for the proportion of good COVID-19 prevention practice, which was considered as 50% (since there was no previous study at comparable area), and “d” stands for the level of standard error tolerated, taken as 5%. The calculated sample was 384.16. After using the design effect of 2 and adding a non-response rate of 10%, the final sample size became 845.
A multistage sampling technique was employed to select the study participants. First, ten woredas and town administrations (three from Bench Sheko, three from West Omo and four from Kafa) were selected randomly and included in the study. Likewise, thirty percent of Kebeles (smallest administrative units) were selected from each of the selected woredas and town administrations. Then, the calculated samples were allocated to each of the selected Kebeles proportional to the size of households in the Kebeles. Finally, the households were selected using a systematic random sampling technique, and from each of the selected households, one eligible participant was selected by the lottery method.
Data Collection Technique and Data Quality Control
A pretested interviewer-administered structured questionnaire (see Additional file 1) adapted from WHO recommendations regarding COVID-19 prevention practices and a previous similar study was used to collect the data (13,17,18). The tool includes the sociodemographic characteristics of the study participants (age, sex, religion, residence, ethnicity, marital status, occupational status, educational status, family size), knowledge questions about COVID-19 (which includes about clinical symptoms, transmission, risk factors, treatment, and vulnerable groups), attitude questions (mainly about the success of the control at an individual and national level, the effectiveness of local treatments such as hot drinks, and the probability of the occurrence of the virus in the locality), practice questions (about risky and positive behaviors practiced recently) and others. The questionnaire was prepared in English and translated to the local language (Amharic) by a language expert. It was also back-translated to English by another expert to ensure its consistency. The Amharic version questionnaire was used to collect the data. To assure the quality of the data, two days of training was given before data collection for data collectors and supervisors about the objective of the study, techniques of data collection, different ethical issues, and care to be taken regarding COVID-19 during data collection. A pretest was performed on 10% of the total sample size in the comparable Kebeles that were not included in the actual study, and some modifications were added to the tool. Data were collected by thirty diploma nurses, and the overall collection process was supervised by ten BSc nurses. The supervisors checked each filled questionnaire for completeness during the data collection.
Variables and measurement
The outcome variable of this study was the practice of COVID-19 prevention methods. The independent variables were sociodemographic characteristics (age, sex, residence, ethnicity, marital status, occupational status, educational status, family size), knowledge about COVID-19, attitude toward COVID-19 prevention methods, intention to seek care and perceived mortality.
The participants were asked 14 knowledge questions, 9 attitude questions, and 8 practice questions. Participants who scored at least the mean score of the above questions for each category were considered to have good knowledge, a positive attitude, and good practice.
Data entry, processing and analysis
Data were entered into Epi data manager version 184.108.40.206 and exported to Statistical Packages for Social Science (SPSS) version 20 for analysis. Data were cleaned to check for outliers and miscoded variables. Furthermore, negatively worded items were reverse scored. Data are presented as tables and proportions (percentages). Binary logistic regression was used to assess the association between the independent variables and an outcome variable. The odds ratio (OR) with its respective 95% confidence interval (CI) and the p-value were used to measure the strength of the association. In the bivariable analysis, variables with p-values <0.25 were considered candidates for multivariable analysis. The final significance was declared at a p-value <0.05.