Description of participant characteristics
Children were born to 4990, 4,717, and 6,273 mothers for the TDHS 2004/5, 2010, and 2015 respectively. During TDHS 2004/5; the majority were males 3385 (50%), born to mothers with primary education 1872 (26.1%), lived in rural areas 5621 (81%), and lived in the poorest wealth quintile households 1504 (22.8%). During TDHS 2010; the majority were males 3203 (49.7%), born to mothers with primary education 4051 (68.2%), lived in rural areas 5255 (80.5%), and lived in poorer wealth quintile households 1462 (23.8%). During TDHS 2015/6; the majority were males 4230 (50.6%), born to mothers with primary education 5068 (64.5%), lived in rural areas 6540 (74.2%), and lived in the poorest wealth quintile households 1931 (24.4%).
The mean child age was 29.7 (16.4), 29.2 (16.4), and 29.2 (16.4) months in the years 2004, 2010, and 2016 respectively. The average household size was 7 (4.2), 7.2 (3.8), 7.4 (4.2) for survey years 2004/5, 2010, and 2015/6 respectively. The overall mean HAZ was -180.213 (134.4), -168.6 (140.8), -149.6 (135.2) in the years 2004, 2010, and 2015/6 respectively. (Table 1).
Trends of child stunting by socioeconomic status
Over the years, stunting declined from 45.5% in 2004 to 42.8 in 2010 and then to 35.6% in 2016. The prevalence of stunting significantly declined from 45.5% to 42.8% at p<0.001) for 2004/5 to 2010 respectively and then from 42.8% to 35.6% at p<0.001 for 2010 and 2015/6 respectively. The overall decline from 2004/5 to 2015/6 was also statistically significant at p<0.001 (all p-values computed according to Chi-square test for trend). Basing on socioeconomic status as measured by household wealth index, the larger decline was among the richer wealth quintile households while a small decline was among the richest wealth index. (Table 2).
Trends of socioeconomic inequalities in stunting
Over the years, the levels of socioeconomic inequalities in stunting were significantly concentrated among the poor as the CIX had negative values in all surveys. The CIX was -0.019 (p<0.001), 0.0178 (p <0.001) and -0.0096 (p <0.001) in 2004, 2010 and 2016 respectively. Socioeconomic inequalities in stunting insignificantly declined, the difference in CIX was 0.0015 (p = 0.7658) from 2004 to 2010 and 0.0081 (p=0.1145) from 2010 to 2015. (Table 3).
Table 1: Percentage distribution of stunting among under-five children for TDHS 2004/5, 2010 and 2015/6
Characteristics
|
2004/5 (N=6,761) n (%)
|
2010 (N= 6415) n (%)
|
2015/6 (N= 8,443) n (%)
|
Area of residence
Urban
|
1159 (19)
|
1177 (19.5)
|
1903 (25.8)
|
Rural
|
5621 (81)
|
5255 (80.5)
|
6540 (74.2)
|
Maternal education
No education
|
1872 (26.1)
|
1645 (25.6)
|
1855 (21.6)
|
Primary
|
4337 (69.4)
|
4051 (68.2)
|
5068 (64.4)
|
Secondary
|
571 (4.5)
|
736 (6.1)
|
1520 (14)
|
Maternal BMI
|
|
|
|
<18.5
|
270 (48)
|
313 (50.2)
|
250 (40.3)
|
≥18.5
|
2714 (45.2)
|
2326 (42.1)
|
2694 (35.3)
|
Childbirth weight
|
|
|
|
Small
|
400 (40.2)
|
246 (51.8)
|
404 (48.9)
|
Average
|
2026 (46.8)
|
1844 (43)
|
2080 (35.4)
|
Large
|
553 (45.4)
|
439 (37.9)
|
438 (29.1)
|
Wealth quintiles
Poorest
|
1504 (22.8)
|
1326 (21.7)
|
1931 (24.4)
|
Poorer
|
1360 (20.8)
|
1462 (23.8)
|
1770 (21.8 )
|
Middle
|
1364 (21.5)
|
1378 (22.5)
|
1665 (19.5)
|
Richer
|
1501 (20)
|
1314 (18.6)
|
1736 (18.2)
|
Richest
|
1051 (14.9)
|
952 (13.4)
|
1341 (16)
|
Table 2: Percentage of stunting among children aged 3 - 59 months by socioeconomic status
Survey year
|
|
Wealth quintiles
|
|
|
Poorest
|
Poorer
|
Middle
|
Richer
|
Richest
|
TDHS 2004/5
|
792 (52.3)
|
691 (50.2)
|
653 (48.8)
|
625 (46)
|
235 (23)
|
TDHS 2010
|
651 (49.3)
|
670 (46.1)
|
600 (45.5)
|
485 (39.5)
|
238 (26.6)
|
Difference (% point)
|
-3
|
-4.1
|
-3.3
|
-6.5
|
3.6
|
TDHS 2010
|
651 (49.3)
|
670 (46.1)
|
600 (45.5)
|
485 (39.5)
|
238 (26.6)
|
TDHS 2015/6
|
777 (41.5)
|
713 (40.6)
|
664 (40.2)
|
506 (30.2)
|
287 (20)
|
Difference (% point)
|
-7.8
|
-5.5
|
-5.3
|
-9.3
|
-6.6
|
TDHS 2004/5
|
792 (52.3)
|
691 (50.2)
|
653 (48.8)
|
625 (46)
|
235 (23)
|
TDHS 2015/6
|
777 (41.5)
|
713 (40.6)
|
664 (40.2)
|
506 (30.2)
|
287 (20)
|
Difference (% point)
|
-10.8
|
-9.6
|
-8.6
|
-15.8
|
-3
|
Table 3: Trend of inequalities in stunting among children aged 3 - 59 months from 2004 to 2016
Year
|
CIX
|
P-value
|
TDHS 2004/5
|
-0.0193
|
<0.001
|
TDHS 2009/10
|
-0.0178
|
<0.001
|
Difference
|
0.0015
|
0.7658
|
TDHS 10
|
-0.0178
|
<0.001
|
TDHS 2015/16
|
-0.0096
|
0.0049
|
Difference
|
0.0081
|
0.1145
|
TDHS 2004/5
|
-0.0193
|
<0.001
|
TDHS 2015/6
|
-0.0096
|
0.0049
|
Difference
|
0.00965
|
0.0538
|
CIX: Concentration index
Socioeconomic determinants of child stunting
Table 4 shows the results from the multivariable pooled Poisson regression models for the factors associated with stunting across two survey phases. We found that for phase I; children living in rural areas had 10% (APR=0.9, 95% CI: 0.81, 0.99) lower prevalence of getting stunted. Children born to mothers who had at least secondary education had 20% (APR=0.8, 95% CI: 0.64, 0.9), 30% (OR=0.7, 95% CI: 0.64, 0.86) lower prevalence of being stunted for phase I and II respectively. The prevalence of getting stunted decreased as the levels of household wealth increased; there were 50% (APR= 0.5 95% CI: 0.44, 0.6), 40% (APR= 0.6 (95% CI: 0.48, 0.7) and 50% (APR= 0.5 (95% CI: 0.43, 0.59) lower prevalence of getting stunted among children living in the richest wealth index households for phase I, II and III respectively. (Table 4).
Decomposition of inequality in child stunting
Household wealth index had a larger contribution by increasing inequalities in stunting in all years although the contribution declined from 93% in 2004/5 to 72% in 2015/6. Maternal years of schooling was the second contributor by increasing the levels of inequalities in stunting over time, the contribution increased across surveys from 9% in 2004/5 to 37% in 2015/6. Area of residence reduced inequalities in stunting in all surveys although the contribution changed over time by increasing then decreasing. The declined trend of the contribution of the household wealth index to the CIX is consistent with the declining inequalities in stunting measured by the CIX. While CIX declined over time, the contribution of mother’s years of schooling increased over time. The insignificant decline of the CIX maybe explaining by the competing effect of household wealth and maternal years of schooling. The variables in this model explained inequalities in stunting more than 80%, the contribution of other factors not explained by the model varied over time as depicted by the residual. (Table 5).
Decomposition of change in socioeconomic inequality in child
We revealed that inequalities in stunting among children aged 3 – 59 months didn’t change significantly. Due to the observed insignificant changes of inequalities in stunting, we will not get useful information by decomposing the observed changes in the concentration index (Table 2).
Table 4: The effect of socioeconomic factors on changes in stunting among children aged 3 - 59 months in the TDHS 2004/5, 2010 and 2015/6
Child stunting Phase I Phase II Phase III
|
2004/5 – 2010 2010 – 2015/6 2004/5 – 2015/6
|
Characteristics APR (95% CI) APR (95% CI) APR (95% CI)
|
Survey year 0.96 (0 .9, 1.0) 0.8 (0.76, 0.88) *** 0.8 (0.74, 0.86) ***
|
Area of residence
Urban 1 1 1
Rural 0.9 (0.8, 0.99)* 1.1 (0.93, 1.21) 0.97 (0.87, 1.08)
|
Mother’s level of education
|
No education 1 1 1
Primary 0.99 (0.92, 1.06) 0.9 (0.85, 0.97) ** 0.99 (0.93, 1.07)
Secondary 0.8 (0.64, 0.9) ** 0.7 (0.63, 0.85) *** 0.9 (0.77, 1.01)
|
Maternal BMI
|
<18.5 1 1 1
|
≥18.5 0.89 (0.82, 0.99)* 0.89 (0.8, 0.97)* 0.96 (0.86, 1.1)
|
Childbirth weight
|
Small 1 1 1
|
Average 0.85 (0.77, 0.9)*** 0.72 (0.66,0.79)*** 0.77 (0.7, 0.83)***
|
Large 0.72 (0.66, 0.79)*** 0.62 (0.8, 1.1)*** 0.63 (0.59, 0.69)***
|
Wealth quintiles
Poorest 1 1 1
Poorer 0.9 (0.85, 1)* 0.9 (0.83, 0.99)* 0.9 (0.87, 1.01)
Middle 0.88 (0.81, 0.96)* 0.9 (0.82, 0.98)* 0.9 (0.84, 0.99)*
Richer 0.8 (0.76, 0.9) *** 0.8 (0.71, 0.87) *** 0.8 (0.74, 0.89) ***
Richest 0.5 (0.44, 0.6) *** 0.6 (0.48, 0.7) *** 0.5 (0.43, 0.59) ***
|
Adjusted for child age, child sex, child-size at birth; diarrhea status, maternal age, zones, cesarean delivery, marital status, skilled birth attendant, number of ANC visits, and year of interview
* Significant at P < 0.05; ** Significant at P < 0.01; *** Significant at P < 0.001
Table 5: Decomposition of concentration indices for under-five stunting in the TDHS 2004/5, 2010 and 2015/6
Stunting (height for age < 2 SD)
TDHS 2004/5 TDHS 2010 TDHS 2015/6
Variables Contribution % Contribution % Contribution %
Residence area
|
-0.0004
|
-2.17
|
-0.0004
|
-23.8
|
0.001
|
-13.65
|
Child age in (months)
|
-0.0006
|
3.06
|
-0.0006
|
3.5
|
-0.0002
|
2.65
|
Child age squared
|
0.0008
|
-4.16
|
0.0008
|
-4.5
|
0.0003
|
-3.26
|
Child sex
|
0.000
|
0.014
|
0.000
|
-0.2
|
0.000
|
-0.4
|
Mother education (years)
|
-0.0017
|
8.95
|
-0.004
|
21.6
|
-0.004
|
36.6
|
Wealth index
|
-0.018
|
93
|
-0.016
|
89.5
|
-0.007
|
71.5
|
Household size
|
0.0004
|
-2.19
|
0
|
-0.08
|
0.0005
|
-5.4
|
Skilled birth attendant
|
-0.004
|
22.58
|
0
|
-0.25
|
-0.002
|
17.5
|
Place of delivery
|
0.0006
|
-3.4
|
-0.004
|
23.1
|
-0.0016
|
16.2
|
Residuals
|
-0.0531
|
-15.684
|
-0.0518
|
-8.87
|
-0.063
|
-21.74
|
Total
|
-0.0229
|
100.00
|
-0.0242
|
100.00
|
-0.013
|
100.00
|
Where %; represents the percentage contribution of the determinant's CIX to the overall CIX