The purpose of study was to investigate status of complementary feeding and its associated factors among mothers who had children 6-23 months in the Yiya Gulele District from June 15 – June 30/2019. Community based cross-sectional study design was conducted. The source population was all mothers who had child aged 6-23 months living in Yaya Gulale district. Study population was all mothers who had child aged 6-23 months living in selected kebeles. Study subjects were those mother who child had aged 6-23 months and give full information for research team. Inclusion criteria was all mothers who had child aged 6-23 months and resident of district more than 1 year and exclusion criteria was mothers who were seriously sick, mental health problem or difficult to give an information to research team.
Sampling Procedure: All Kebeles were stratified into rural and urban because of the presence of difference among study participants based on their residence (1 urban and 17 rural kebeles). In the first stage, one urban kebeles and six kebeles (30%) from the rural kebeles was selected by using simple random sampling (lottery) methods. The sample sizes were distributed to each selected kebeles proportional to household size of the kebeles. Sampling frame was prepared depending on health post family folder and the interval was determined by the division of total household with mother who had child aged 6-23 months to required sample, which results in an interval of three household. Then systematic sampling was used to select study subject from the random start and mothers with child aged 6-23 months in the selected household was selected and interviewed. For eligible participant who was not found at home, the interviewers revisited the household two times at different time intervals and when interviewers failed to get the eligible participant, the household was registered as non-response.
Schematic presentation of sampling procedure, maternal knowledge on complementary feeding and associated Factors in Yaya Gulale district, Oromia, Ethiopia, 2019:
Sample Size Determination:
The sample size is determined by using single population proportion formula.
n = (Z/2)2p(1-p)
d2
Based on the following assumptions: 95% confidence level, prevalence of expected knowledge on CF was 87% from study conducted in Abyi-Adi town, 5% margin of error, 10% non-respondent rate.
n= sample size
z = critical value = 1.96 for 95% CI
p = prevalence of expected knowledge on CF d = precision (marginal of error) b/n the sample and the proportion 4% =0.04
n=((1.96 × 1.96) × 0.87 ( 0.13)/(0.04 × 0.04), n =273
Considering non-response rate 10%
As the target population is less than 10,000 (i.e. 3253, we need to apply correction formula)
[Please see the supplementary files section to view the formula.]
Where, nf = required sample size when target population is < 10,000
N = Population target (3253)
Therefore, n = 273
1+273
3253
= 252
For possible non- response during the survey the final sample size is increase by 10% to n = 252+252*10% which was 252+25 = 277, using design effect= 2, nf= 277*2=554
Sample size calculation by using stat calc for second objective for the study that will be conducted in Yaya Gulale district, Ethiopia from June 15 – June 30/2019:
Objective
|
Assumption
|
Reference
|
Variable
|
Exposed
|
Unexposed
|
Power
|
AOR
|
Sample size
|
1
|
Maternal educational status
|
83.6
|
67.1
|
80
|
2.5
|
488
|
Y.Mulugeta,2017
|
2
|
Urban settlement
|
44.3
|
27.4
|
80
|
2.11
|
568
|
Duna Ayana et al,2017
|
3
|
History of PNC
|
30.2
|
13.4
|
80
|
2.8
|
432
|
Ergib Mekbib et al,2014
|
4
|
Maternal educational status
|
95
|
83.2
|
80
|
3.84
|
506
|
Ergib Mekbib et al,2014
|
5
|
History of ANC
|
80.1
|
62.6
|
80
|
2.4
|
418
|
Y.Mulugeta,2017
|
Generally, sample sizes were calculated for the first and the second objectives and the largest sample size was found to be 568 from the second objective.
Data Collection Procedure: Data was collected using structured questionnaire using face to face interview. Pretested was conducted before main study commence taking 5% of sample to validate questionnaire. The questionnaire was developed from previous literatures and then modified to the study objectives. Questionnaires were translated in to local language (Afan Oromo).
Data Quality Control: Data collectors and supervisors were taken training for one day on how to select study participants and other technical procedures. Pretesting was conducted to ensure questionnaires were ethical acceptable by taking 5% of from the total sample size prior to data collection. The collected data was rechecked for completeness and consistency by supervisors and principal investigators before transferring in to computer software. Principal investigator was supervise daily to check the completeness of the questionnaire and consistency of related data as well as double data entry was employed to verify whether the data was properly entered.
Study Variables: Dependent Variable was complementary feeding. Independent Variables were socio-demographic characteristics of the mothers such as mothers’ age, educational status, occupation, husband educational status, husband occupation and family size. Health Care service such as ANC, Place of delivery and PNC and Source of Information
Data Analysis and Management:
Data was checked for incompleteness, inconsistency, edited, coded. Then data was entered by using EP Info software version 7 for cleanness and analyzed by using SPSS software version 20. Model goodness of fit was checked by Hosmer-Lemeshow goodness of fit. Descriptive statistics were used to describe status of complementary feeding by measuring mothers knowledge using Likert scale questionnaire. Mothers’ knowledge on Complementary Feeding was measured based on a two scales (‘1’, ‘o’). A score of ‘1’ was good knowledge while ‘0’ score was poor knowledge on subject matter. Then, when the mothers were answer more than 50% above, it decided that she has good knowledge. When she answered less than 50% she had poor knowledge (19). Binary logistic regression was used to compute association of the dependent variable with independent variables using 95%CI taking adjusted odd ratio and p-value <0.05 as statistically significant. Variables with p-value less than 0.25 in the binary logistic regression analysis were entered to multiple logistic regressions to control confounders.