Individual and community-level characteristics of study participants
A total of 15683 women in Ethiopia were asked about their fertility desire among these a weighted sample of 12019 rural women was selected for this study. Those women who are sterilized and infecund were not included. The majority 4,675(38.9%) of respondents were in the age group of 15-24 and the rest were 4,063(33.8%) in the age group of 25-34 and 3,281(27.3%), in the age group of 35-49. Regarding marital status, more than half 8,259 (68.72%) of respondents were married, 2,663 (22.15%) were not married, and 1,097 (9.13%). Around 4,619 (38.43%) women were orthodox, 4,160(34.62%) Muslim, 2,941(24.47%) protestant, and the remaining were others. Near to half 5,269(43.84%) of women were poor, 2,898(24.11%) were middle and 3,852(35.02%) were rich. More than half 6,773 (56.36%) of women were not educated and the rest 35.8% and 7.82% had primary and secondary education respectively. Majority of women 10,258 (85.35%) had no media exposure such as newspaper, TV and radio (Table1).
Table 1
Weighted individual and community-level characteristics of study participants, EDHS 2016 (N = 12,019)
Variables
|
Categories
|
Frequency (%)
|
Educational status
|
No education
|
6,773(56.36)
|
|
Primary
|
4,306(35.8)
|
|
Secondary and above
|
940(7.82)
|
Working status
|
Not working
|
8,663 (72.08)
|
|
Working
|
3,356 (27.92)
|
Region
|
Larger central
|
11,373 (94.62)
|
|
Small peripherals
|
605 (5.04)
|
|
Metropolis
|
41 (0.34)
|
Media exposure
|
No
|
10,258 (85.35)
|
|
Yes
|
1,761 (14.65)
|
Number of living children
|
No
|
3,444 (28.65)
|
|
1-3
|
4,053 (33.72)
|
|
4 and above
|
4522 (37.62)
|
Knowledge of family planning method
|
No
|
237 (1.97)
|
|
Yes
|
11,782 (98.03)
|
Family planning use
|
No
|
6,980 (58.07)
|
|
Yes
|
5,039 (41.93)
|
Family planning a professional visit
|
No
|
8,998 (74.86)
|
|
Yes
|
3,021 (25.14)
|
Family planning message
|
No
|
9,565 (79.58)
|
|
Yes
|
2,454 (20.42)
|
Fertility preference
|
No
|
8,048 (66.96)
|
|
Yes
|
3,971 (33.04)
|
Community media exposure
|
Low
|
6,247 (51.97)
|
|
High
|
5,772 (48.03)
|
Community illiteracy proportion
|
Low
|
6,678 (55.56)
|
|
High
|
5,341 (44.44)
|
Community-level poverty
|
Low
|
7,859 (65.39)
|
|
High
|
4,160 (34. 61)
|
Desire to limit childbearing
The proportion of women who desired to limit childbearing in rural Ethiopia was 3,971(33.04%), (95% CI=32.2%, 33.9%)
Factor associated with women’s desire to limit childbearing
Among those variables entered in multivariable analysis, age of women, number of children, marital status, and region and community level wealth index were significantly associated with women’s desire to limit childbearing in rural Ethiopia.
Women in the age group of 25- 34 years of age are 1.61 times (61%) more likely to have a desire to limit childbearing (AOR=1.61,95% CI=1.28,2.13 ) as compared to their counterparts. Similarly in the age range 35- 49 years, they are 4.92 times (AOR=4.96, 95% CI = 3.64, 6.65) more likely to limit childbearing. The other important significant factor which contributed to women’s desire to limit childbearing was the number of children. A woman who had no children was 94% less likely to limit childbearing (AOR=0.06, 95% CI =0.04, 0.09), and having children 1-3 also decreased women's desire to limit childbearing by 71% (AOR=0.29, 95% CI = 0.23, 0.36) as compared to those who have 4 and more children. Regarding marital status, married women were less likely to limit child bearing and decreased desire by 57% (AOR=0.45,95% CI= 0.27, 0.75).
The region of the country was a significant predictor of limiting childbearing. Women who were living in small peripherals regions of Ethiopia such as Afar, Somali, Gambella, and Beninshangul Gumz were less likely or decreases the desire to limit childbearing by 0.33 times or 67% (AOR=0.33,95% CI=0.24, 0.45) as compared to those in large central regions of Ethiopia. Lastly, community level poverty proportion was significantly associated with limiting child bearing. Those community with high poverty proportion were less likely to limit child bearing that was decreased by 28 % (AOR=0.72, 95% CI= 0.57, 0.89) as compared to those with low proportion (Table2).
Table 2
Mixed effect binary Logistic regression analysis showing factors associated with women's desire to limit childbearing, EDHS 2016
Variables
|
Categories
|
Desire
|
|
Model II(community level)
|
Model III(individual level)
|
Model IV (full model)
|
|
|
No
|
Yes
|
-
|
|
|
Age group
|
15-24
|
4,205
|
470
|
-
|
Ref
|
Ref
|
|
25-34
|
2,744
|
1,319
|
-
|
1.65(1.28,2.13)*
|
1.61(1.25,2.09)*
|
|
35-49
|
1,100
|
2,181
|
-
|
5.08(3.76, 6.86)*
|
4.92(3.64,6.65)*
|
Education level
|
no education
|
3,826
|
2,947
|
-
|
Ref
|
Ref
|
|
Primary
|
3,379
|
927
|
-
|
0.98(0.79,1.20)
|
0.96(0.78,1.18)
|
|
Secondary and above
|
843
|
97
|
-
|
0.71(0.45,1.12)
|
0.71(0.45,1.12)
|
Wealth index
|
Rich
|
2,572
|
1,280
|
-
|
Ref
|
Ref
|
|
Middle
|
3,492
|
1,777
|
-
|
0.82(0.67,0.99)*
|
0.83(0.68,1.01)
|
|
Poor
|
1,984
|
914
|
-
|
1.08(0.87,1.35)
|
1.20(0.96,1.51)
|
Religion
|
Orthodox
|
2,978
|
1,641
|
-
|
1.01(0.58,1.76)
|
1.004(0.58,1.74)
|
|
Protestant
|
1,968
|
973
|
-
|
0.95(0.55,1.64)
|
0.92(0.53,1.59)
|
|
Muslim
|
2,888
|
1,272
|
-
|
0.71(0.39,1.27)
|
0.86(0.48,1.52)
|
|
Other
|
214
|
85
|
-
|
Ref
|
Ref
|
Media exposure
|
Yes
|
1,276
|
485
|
-
|
Ref
|
Ref
|
|
No
|
6,772
|
3,486
|
-
|
1.04(0.78,1.40)
|
1.06 (0.79, 1.43)
|
Number of children
|
No
|
3,201
|
243
|
-
|
0.063(0.04, 0.10)*
|
0.06 (0.04, 0.09)*
|
|
1-3
|
3,040
|
1,013
|
-
|
0.30(0.24, 0.37)*
|
0.29(0.23, 0.36)*
|
|
4 and above
|
1,807
|
2,715
|
-
|
Ref
|
Ref
|
Knowledge of FP
|
Yes
|
7,856
|
3,926
|
-
|
Ref
|
Ref
|
|
No
|
192
|
45
|
-
|
0.62(0.31, 1.25)
|
0.91(0.45,1.86)
|
Marital status
|
Not married
|
2,449
|
213
|
-
|
Ref
|
Ref
|
|
Married
|
5,079
|
3,181
|
-
|
0.43(0.25,0.72)*
|
0.45(0.27,0.75)*
|
|
Widowed/divorced
|
520
|
577
|
-
|
1.47(0.91,2.39)
|
1.52(0.94,2.47)
|
Use of FP
|
Yes
|
2,976
|
2,063
|
-
|
Ref
|
Ref
|
|
No
|
5,072
|
1,908
|
-
|
0.91(0.76, 1.10)
|
0.96(0.79,1.15)
|
FP professional visit in the last 12 months
|
Yes
|
1,865
|
1,156
|
-
|
Ref
|
Ref
|
|
No
|
6,183
|
2,815
|
-
|
1.002(0.85,1.18)
|
1.01(0.85, 1.19)
|
Working status
|
Working
|
2,097
|
1,259
|
-
|
Ref
|
Ref
|
|
Not working
|
5,951
|
2,712
|
-
|
0.86(0.71, 1.04)
|
0.86(0.71, 1.04)
|
Region
|
larger central
|
7,511
|
3,862
|
Ref
|
-
|
Ref
|
|
Small peripherals
|
510
|
95
|
0.39(0.32, 0.49)*
|
-
|
0.33(0.24,0.45)*
|
|
Metropolis
|
27
|
13
|
0.99(0.79, 1.26)
|
-
|
1.21(0.89,1.64)
|
Community illiteracy proportion
|
Low
|
4,486
|
2,191
|
Ref
|
-
|
Ref
|
|
High
|
3,562
|
1,780
|
1.11(0.94,1.30)
|
-
|
0.89(0.71,1.11)
|
Community media exposure
|
Low
|
4,068
|
2,179
|
Ref
|
-
|
Ref
|
|
High
|
3,980
|
1,792
|
0.88(0.75,1.02)
|
-
|
0.91(0.73,1.12)
|
Community-level poverty
|
Low
|
5,187
|
2,672
|
Ref
|
-
|
Ref
|
|
High
|
2,861
|
1,299
|
0.91(0.77,1.08)
|
-
|
0.72(0.57,0.89)*
|
Ref= Reference, *= Statistically Significant Variables |
Community-level variance and model fitness
The community level variance of null and full model was 0.27 and 0.38 respectively. This indicated that the variance was greater than zero showing the areas were heteroginous. Regarding model comparision the best fitted model was full model or amodel with individual and community level variables.The fitted model had low AIC/BIC and high LLR as compared to previous model (Table 3).
Table 3
Community-level variance of multilevel (mixed effect) model predicting women’s desire to limit childbearing, EDHS 2016
Random effect
|
Null model
|
Full model
|
Community level variance
|
0.27
|
0.38
|
PCV
|
Reference
|
15.6
|
MOR)
|
1.34
|
1.59
|
ICC (%)
|
7.6
|
1.05
|
AIC
|
14895.38
|
11078.94
|
BIC
|
14909.83
|
11259.48
|
LLR
|
-7445.69
|
-5514.47
|