Study design and setting
A community based cross sectional study was conducted among women in 15 rural kebele (the smallest administrative units of Ethiopia) of Central Gondar zone, ANRS, Northwest Ethiopia, from September to December, 2019. Central Gondar zone is found in ANRS and its capital city, Gondar is located 727Kms away from Addis Ababa, the capital city of Ethiopia. According to the 2007 census, the Zone has a population of 2,288,442 inhabitants of whom 462,952 were women of reproductive age. According to the 2019 Central Gondar zone health department report, there were 14 districts (2 urban and 12 rural), 75 health centers and 9 hospitals in the zone.
Sample size
The sample size was calculated for both objectives and then we have taken the largest samples. Accordingly, sample size calculation for the first objective i.e. to determine the Prevalence of institutional delivery service utilization was calculated using single population proportion formula;
n= z (a/2)2p (1-p)/d2
n= Sample size
Za/2= 1.96 standard score corresponding to 95% CI
d=0.05
p = 0.34 (Proportion of institutional delivery service utilization) [14]
n= z (a/2)2p (1-p)/d2
n= (1.96)20.34(1-0.34)/ (0.05)2=345
n = 345, since there is design effect we multiply by 2 then it will be 690 with 10% non-response rate, it will give 759.
Whereas the sample size calculation for the second objective by using significant factors like ANC visit, knowledge of pregnancy complications, Educational status and Attitude were considered and calculated from previous study conducted on institutional delivery service utilization among women of rural Ethiopia [15, 16] using EPI INFO STAT CALC with, 95% confidence interval, 80% power,5% margin of error, design effect of 2 and 10% non-response rate was 1,394 which is larger than samples from single population proportion formula. Hence, the total final sample size was 1,394 women who gave birth in the last 12 months.
Sampling procedure
Whenever two or more eligible women were found within the same household, just one of them were selected randomly and included in the study. All women who gave birth in the past 12 months within the randomly selected rural kebele were included in the study. Women who were seriously ill or unable to speak were excluded from the study. The list of eligible women was obtained from registration books of the respective kebele’ health extension workers.
Data collection tool, procedure, and quality control
A structured questionnaire was used to collect data which was developed by reviewing literatures [17-25] and conducting pilot test (or field test) to make the questionnaire valid and clear. The questionnaire was initially prepared in English and translated to the local language Amharic and back translated to English by language experts to see the consistency.
The questionnaire was pretested on 70 women (5% of the entire sample size) in the rural communities of West Gojjam zone, Bahir Dar Zuria districts which has similar context with the study area.
The data were collected by ten diploma graduate nurses and supervised by three BSc graduate public health officers. Two days training was given for the data collectors and supervisors about the objectives and data collection process by the Principal investigator. The data were checked for accuracy and consistency daily.
Study variables
The outcome variable of the study was institutional delivery service utilization which was defined as coded as “Yes” if women reported that they gave their most recent birth (within the last one year) at health institution, and “No” if otherwise.
The independent variables of the study includes basic socio-demographic information like (age, marital status, educational status, and occupation of mothers; educational status, and occupation of the husband; family size, and wealth index of households, family discussion, and community organization discussion), Health related information, and supervision system, source of information, predisposing and enabling factors which impede institutional delivery service utilization behavior of women(distance from health facilities, supervision by the health care providers, decision power about place of birth), need related factors(status of pregnancy i.e. wanted or unwanted pregnancy, ANC follow-up, visiting health facilities during complications),characteristics of health delivery system and Courtesy of health service providers.
Wealth index was constructed using a principal component analysis (PCA) after having data on household assets, which included a durable asset list, recording the land and animals owned and observing housing materials. Finally, different important family asset factors were summed up to categorize individuals into wealth tertiles (poor, medium, rich).
Data processing and analysis
The collected data were checked for completeness, consistency, and missing values; coded and entered, using Epi Data version 3.1 and cleaned and analyzed using STATA software version14.1.
Using descriptive methods, the data was summarized, prevalence of institutional delivery service utilization was determined and odds ratios (OR) were obtained using logistic regression. The data obtained from individuals in each household are pooled to create a single large data set then the studies use the number of individuals institutional delivery service utilization analyzed as the statistical n value, which is we assume the data gathered at each kebele to be an independent measurement so that we can use simple logistic regression by ignoring clustering [26]. The effect of different variables on institutional delivery service utilization was explored using crude and adjusted odds ratios. After checking the correlation of independent variables, significance was determined using crude and adjusted odds ratios with 95% confidence intervals. To determine the association between the different predictor variables with the dependent variable, first bi-variable analysis between each independent variable and outcome variable was investigated using a binary logistic regression model and then all variables having p-value < 0.2 in the bi-variable analysis were suggested as a criterion for variable selection for inclusion into a multivariable model. So that all variables with a p-value of < 0.2 in the bi-variable were analyzed for multi-variable logistic regression.
Adjusted Odds Ratios (AOR) with 95% Confidence Intervals (CI) was calculated to show the presence and strength of associations. A variable with p-value of less than 0.05 in the multivariable logistic regression model was considered as statistically significant.
Ethical consideration
Ethical approval for the study was obtained from the institutionalized review board, university of Gondar with a ref.no O/V/P/RCS/05/1048/2019 on a date of March 4 2019. Official letter that explains the objectives of the study was written to the respected Amhara public Health Institute.
The Amhara public Health institute wrote a letter to zonal health departement.The zonal health department was wrote a letter for each selected district health office. The selected district health office in turn wrote a letter for the study kebeles for cooperation. The objectives and the benefits of the study were explained for the study subjects. Written consent was obtained from each participant. The right of the participants to withdraw from the study whenever they want to do so was respected. Anonymous questioner was used to protect the identity and confidentiality of the information obtains from individual participants.