Study design, period, and area
A community-based triangulated cross-sectional (embedded design) study was conducted from March 1 to March 31, 2020, in South Gondar Zone, which is one of the Zones in Amhara Region of Ethiopia and bordered on the South by East Gojjam, on the Southwest by West Gojjam and Bahir Dar, on the West by Lake Tana, on the North by Central Gondar, on the Northeast by Wag Hemra, on the East by North Wollo, and on the Southeast by South Wollo. The Zone is comprised of 13 woredas (Andabet, Dera, Ebenate, Estie, Farta, Fogera, Gunabegemider, Libokemkem, Lay Gayenet, Mena Meketewa, Sedie Muja, Simada, and Tachegayenet) and two administrative towns (Debre Tabor and Woreta), the capital of which is Debre Tabor, 666 km Southwest of Addis Ababa, the capital city of Ethiopia and 103 km Southeast of Bahir Dar, the capital city of Amhara Region. The total area of the Zone is 14,095.19 square kilometers and according to the 2007 census, the total population was 2,051,738, from which 1,041,061 were males and 1,010,677 females . From the total population of 277, 887 were under-five children. Among the total population, 96.14 percent of the population was Orthodox religious and 3.68 percent were Muslim. The zone has one general hospital, seven district hospitals, and 93 health centers.
All caregivers who had children less than 5 years old in South Gondar Zone were the source population of this study. However, caregivers who had less than five years of old children live in selected kebeles were the study population of this study.
Inclusion and exclusion criteria
Caregivers with children under five years of age who provided information and were available at the time of data collection were included in the study. Whereas, caregivers who were seriously ill at the time of data collection were excluded from the study.
Sample size determination
The sample size was calculated by using the assumption of a 95% confidence interval, 5% degree of precision, 80% power, unexposed to the exposed ratio of 1:1, and 10% non-response rate. The sample size was determined for each specific objective and finally, to have a possible maximum sample size, a large number from those computed was taken as sample size. Since a multi-stage sampling technique was applied; to correct the design effect correction was made accordingly by multiplying the initial sample size.
For the first objective, the sample size was determined by using a single population proportion formula
Where, n= Initial sample size
P = prevalence of traditional uvulectomy
d = Marginal error
Za/2 = the value under standard normal table
For the practice, 84.7% of the proportion of traditional uvulectomy has been taken from a study of "harmful traditional practices; an analysis of its prevalence and associated factors among children in Ethiopia" conducted by Abdela Kufa .
To correct the design effect; Multiplied the value by 2 and the total sample size was 199 * 2 = 398 with 10% for non- response rate = 438.
For associated factors; even though there were many associated factors to practice uvulectomy, occupation (farmer) was the most significant explanatory variable . So, based on this variable and the above assumptions, the sample size was calculated by using software (EPI info version 18.104.22.168). To have a possible maximum sample size, a large number from those computed was taken as sample size. Therefore, the final sample size was taken 634 (table 1).
For quantitative; first, by using the lottery method six woredas were selected, and based on the total number of caregivers having children less than five years in each district the study sample proportional allocation was made. After that, of the six districts, the recruited kebeles (small administrative units) were selected by the lottery method and proportional allocation of sample for the selected kebeles was made. In addition, “Gotes” from the recruited kebeles were selected with simple random sampling. Finally, the study subject or participant caregivers were selected in each chosen “Gotes” by using a sampling frame which was obtained from the local health extension workers (figure – 2).
For qualitative; to get the right persons for answering the questions of the study a purposive sampling method was used and 10 in-depth interviews were conducted.
Knowledge; those caregivers who scored the mean or above for knowledge assessment questions were grouped as good knowledge and those caregivers who scored below the mean for knowledge assessment questions were grouped as poor knowledge.
Practice; this was measured from the report of caregivers and was coded 1 if caregivers reported that had performed traditional uvulectomy on their children; otherwise, it was coded 0.
Attitude; related issues of the respondent were assessed by five yes or no questions focusing on the perception about uvula, harmfulness of traditional uvulectomy, future intention, encouraging others to perform uvulectomy on their children, and the eradication of traditional uvulectomy. It was measured separately from the report of caregivers and coded 1 if the caregivers reported yes; otherwise, it was coded 0.
Traditional uvulectomy; surgical removal of the uvula by traditional practitioners.
Data collection tools and procedures
For a quantitative approach
Data collection tool; a structured questionnaire containing socio-demographic background, questions about traditional uvulectomy knowledge, practice, and attitude as well as associated factors which were adapted from various works of literature previously done on similar topics [13, 15-17, 19, 22, 27, 30]. The questionnaire was first prepared in English and then translated to the local language (Amharic).
Data collection procedure; the data was collected by two trained health extension workers from caregivers who have children less than 5 years old using an interviewer-administered questionnaire. Study participants having two under-five children were included with their younger children. And also, the data collection process was supervised by one clinical nurse. Instead of personal identifier code for each study participants were given. Data were checked and cleaned daily for completeness and consistency during data collection.
For the qualitative approach;
In-depth interviews by using semi-structured guiding questions were used. Before data collection, the study participants were fully informed about the purpose, advantages of the study, with the right to refuse at any stage of the in-depth interview. And also, they were assured the information would not be disclosed. After taking permission, an in-depth interview was conducted. All the in-depth interviews were recorded by the audio recorder.
Data Quality Control
To ensure the quality of data, intensive training was given for 1 day before the data collection procedure starts to data collectors and supervisors about data collection processes and techniques. Local language-translated tools were used. A pre-test was conducted on 32 eligible caregivers (5% of sample size) at Andabet woreda Jaragedo and Semete kebele and that was not included in the study and tool modification was made by adding a variable health personnel hospitality problem. Moreover, the data collection process was supervised strictly.
Data processing and analysis
For quantitative data;
Data were entered and edited using Epi-Data by the principal investigator and then exported into SPSS version 23 for analysis. Descriptive measures of statistics like the frequency with percentage, mean with standard deviation, and median with interquartile range was used to describe the socio-demographic characteristics and the practice of traditional uvulectomy. Logistic regression (both bivariable and multivariable) was performed to identify the factors associated with traditional uvula cutting. Odds Ratio (OR), 95% CI, and P-value were used to assess the strength of association and statistical significance (p < 0.05). Variables with a p-value < 0.2 in the binary regression analysis were entered for multivariable analysis. To check the model goodness of fit Hosmer and Lemeshow test was used.
For qualitative data;
The recorded interviews were transcribed to written version (text) word by word and translated to the English language. The translated data was entered into open code, with the thorough reading code based on meaning were given and analyzed using thematic analysis approach. Finally, triangulated with the quantitative result were made.