TABLE 2.
Newborn (cord blood plasma) micronutrient biomarker concentrations and their differences by maternal antenatal intervention in the JiVitA-3 trial in rural Bangladesh1
| IFA | MM | Difference between MM and IFA | ||||||
|---|---|---|---|---|---|---|---|---|
| Nutrient | n | Mean ± SD or geometric mean (−1 SD, +1 SD) | n | Mean ± SD or geometric mean (-1 SD, +1 SD) | Biomarker units, 95% CI | %, 95% CI | P value | R2 |
| Iron | ||||||||
| Ferritin, µg/L | 153 | 142 (67, 301) | 167 | 160 (85, 299) | +18 (2, 35) | +12.4 (1.3, 24.6) | 0.03 | 0.0071 |
| Folate | ||||||||
| Total plasma folate, nmol/L | 156 | 61 (39, 96) | 176 | 56 (35, 90) | −5 (−9, 0) | −7.5 (−14.9, 0.6) | 0.07 | 0.0072 |
| Vitamin B-12 | ||||||||
| Cobalamin, pmol/L | 155 | 286 (168, 489) | 176 | 309 (182, 527) | +23 (−10, 60) | +8.0 (−3.5, 21.0) | 0.17 | 0.0052 |
| Vitamin A | ||||||||
| Retinol, µmol/L | 134 | 0.63 ± 0.20 | 148 | 0.65 ± 0.25 | +0.03 (−0.04, 0.09) | +4.1 (−6.3, 14.4) | 0.43 | 0.0032 |
| Vitamin D | ||||||||
| 25(OH)D, nmol/L | 157 | 38 (27, 54) | 176 | 44 (31, 62) | +6 (2, 10) | +14.7 (4.8, 25.6) | 0.004 | 0.0381 |
| Vitamin E | ||||||||
| α-Tocopherol, µmol/L | 134 | 5.7 ± 2.0 | 148 | 6.2 ± 2.0 | +0.4 (−0.2, 1.1) | +7.7 (−3.3, 18.8) | 0.16 | 0.0119 |
| Zinc | ||||||||
| Total plasma zinc, µmol/L | 155 | 14.3 (11.7, 17.5) | 176 | 15.2 (12.6, 18.4) | +0.8 (0.1, 1.6) | +5.8 (1.0, 10.8) | 0.02 | 0.0214 |
| Iodine | ||||||||
| Thyroglobulin, µg/L | 149 | 35.3 (16.2, 77.2) | 170 | 35.2 (18.6, 66.5) | −0.2 (−5.2, 5.8) | −0.4 (−14.7, 16.3) | 0.96 | 0.0000 |
| fT4, pmol/L | 149 | 15.3 ± 1.7 | 171 | 15.2 ± 1.6 | −0.2 (−0.6, 0.2) | −1.1 (−3.9, 1.6) | 0.40 | 0.0028 |
Geometric mean (−1 SD, +1 SD) was calculated by back-transforming from the log10 scale: mean ± SD for log10-transformed ferritin in the IFA group is 2.153 ± 0.325, so 102.153 (102.153–0.325, 102.153+0.325) = 142 (67, 301). Differences in biomarker values between MM and IFA groups were assessed by linear regression with intervention (0 = IFA, 1 = MM) as the independent variable with robust SEs to account for cluster randomization. For biomarkers on the arithmetic scale the MM effect = β1 (95% CI) from the regression model. For log10-transformed variables, the MM effect = 10β0+β1–10β0, where β0 is the regression constant for the IFA group and β1 the MM effect or the upper or lower bound of its 95% CI: for ferritin, β0 = 2.1528, β1 (95% CI) = 0.0506 (0.0057, 0.0955), so 102.1528+0.0506–100.0506 = 18, then substituting upper or lower bounds of 95% CI for β1. On the arithmetic scale, 100 × β1(95% CI)/β0 was used to calculate percentage difference in values between the MM and IFA groups: for retinol, β0 = 0.6289, β1 (95% CI) = 0.0257 (−0.0394, 0.0908), so 100 × 0.0257/0.6289 = 4.1%, then substituting upper and lower bounds of 95% CI for β1. For log10-transformed biomarkers, percentage change = 10 β1(95% CI): for ferritin, 100.0506 (100.0057, 100.0955) = 1.124 (1.013, 1.246), indicating a 12.4 (95% CI: 1.3, 24.6)% increase in ferritin in the MM compared with IFA group after subtracting 1 and multiplying by 100 (16).
Conversion of biomarker units can be done as follows: folate nmol/L × 0.4413 to ng/mL, cobalamin pmol/L × 1.355 to pg/mL, retinol µmol/L × 28.65 to µg/dL, 25(OH)D pmol/L × 0.4006 to ng/mL, α-tocopherol µmol/L × 0.423 to µg/mL, zinc µmol/L × 6.534 to µg/dL, fT4 pmol/L × 0.0777 to ng/dL. fT4, free thyroxine; IFA, iron folic acid; MM, multiple micronutrients; 25(OH)D, 25-hydroxyvitamin D.