Table 2.
Bivariable analysis |
Multivariable analysis |
|||||
---|---|---|---|---|---|---|
Variable
|
Crude OR
|
95% CI
|
P-value
|
aOR
|
95% CI
|
P-value
|
Study group |
|
|
|
|
|
|
Control
|
1.0 |
ref |
|
1.0 |
ref |
|
Intervention
|
0.08 |
0.03, 0.20 |
<0.001 |
0.10 |
0.04, 0.28 |
<0.001 |
Child sex |
|
|
|
|
- |
|
Male
|
1.0 |
ref |
|
|
- |
|
Female
|
0.92 |
0.50, 1.70 |
0.78 |
|
- |
|
Age in months |
1.15 |
0.79, 1.66 |
0.47 |
|
- |
|
Exclusive breastfeeding |
|
|
|
|
|
|
No
|
1.0 |
ref |
|
1.0 |
ref |
|
Yes
|
2.15 |
1.12, 4.13 |
0.02 |
1.86 |
0.91, 3.80 |
0.09 |
Height-for-age z-scores |
0.84 |
0.65, 1.09 |
0.19 |
|
- |
|
Weight-for-age z-scores |
0.67 |
0.50, 0.89 |
0.007 |
0.74 |
0.53, 0.98 |
0.04 |
Weight-for-length z-scores |
0.78 |
0.60, 1.01 |
0.06 |
|
- |
|
Cognitive composite score |
0.97 |
0.95, 0.99 |
0.02 |
|
- |
|
Language composite score |
0.98 |
0.95, 0.99 |
0.03 |
0.98 |
0.95, 1.01 |
0.06 |
Motor composite score |
0.99 |
0.97, 1.02 |
0.65 |
|
- |
|
Maternal age in years |
1.01 |
0.96, 1.07 |
0.65 |
|
- |
|
Maternal education |
|
|
|
|
|
|
None/primary level
|
1.0 |
ref |
|
1.0 |
ref |
|
Lower secondary
|
0.49 |
0.25, 0.94 |
0.03 |
0.47 |
0.21, 0.97 |
0.03 |
Tertiary
|
0.39 |
0.14, 1.09 |
0.07 |
0.52 |
0.15, 1.79 |
0.30 |
Maternal parity |
|
|
|
|
|
|
Less than five children
|
1.0 |
ref |
|
|
- |
|
Five or more children
|
0.99 |
0.51, 1.94 |
0.99 |
|
- |
|
Household head age in years |
1.01 |
0.96, 1.04 |
0.97 |
|
- |
|
Household head number of years in school |
0.86 |
0.77, 0.95 |
0.005 |
0.86 |
0.78, 0.96 |
0.03 |
Household size |
|
|
|
|
|
|
3 to 5
|
1.0 |
ref |
|
|
- |
|
6 to 13
|
0.94 |
0.51, 1.74 |
0.85 |
|
- |
|
Poverty score |
0.97 |
0.94, 0.99 |
0.04 |
|
- |
|
Child dietary diversity score |
0.85 |
0.71, 1.02 |
0.08 |
|
- |
|
Multivariable model performance |
|
|
|
|
|
|
Bootstrapped AUC (95% CI) | 0.82 | 0.76, 0.88 |
aOR – adjusted odds ratio, AUC – area under receiver operating characteristic curve, CI – confidence interval, OR – odds ratio, ref – reference
*Odds ratios are from a mixed-effect multilevel regression model. The outcome variable was binary KABC-II scores categorised into average to upper extreme vs. below average to lower extreme.