The effect of a micronutrient powder home fortification program on anemia and cognitive outcomes among young children in rural China: a cluster randomized trial

Table 4 Dose–response effects of multiple micronutrient powder sachet consumption on heamoglobin (Hb) concentration, anaemia prevalence, and Bayley mental and psychomotor development scores

	Hb concentration (g/L)¹		Anemia prevalence (Hb <110 g/L)¹		Bayley MDI (points)²		Bayley PDI (points)²
	Marginal Effect (95% CI)	P value	Marginal Effect (95% CI)	P value	Marginal Effect (95% CI)	P value	Marginal Effect (95% CI)	P value
Effect of MNP sachet consumption after 6 months (dose–response per sachet)	0.023 (0.001–0.045)	0.040*	− 0.001 (− 0.002–0.0001)	0.087	0.028 (0.002–0.054)	0.034*	− 0.001 (− 0.038–0.035)	0.943
Effect of MNP sachet consumption after 12 months	− 0.002 (− 0.026–0.022)	0.883	0.000 (− 0.001–0.001)	0.427	0.007 (− 0.022–0.035)	0.640	− 0.004 (− 0.031–0.023)	0.774
Effect of MNP sachet consumption after 18 months	0.002 (− 0.022–0.027)	0.861	− 0.000 (− 0.001–0.001)	0.977	0.015 (− 0.016–0.047)	0.346	0.003 (− 0.026–0.031)	0.843
R-squared	0.14		0.10		0.26		0.21
Observations	6003		6003		6113		6096

¹ Average Treatment on the Treated analysis, using the treatment assignment as an instrumental variable for the total number of sachets used in the preceding six months. Regression estimates from multiple linear models adjusted for gender, age, Hb test machine and county fixed effects. Standard errors are clustered at the village level. Hb data missing for 349 children (17 in baseline, 115, 144 and 73 in 6 months, 12 months and 18 months follow-up) and feeding behavior data missing for 206 children (74, 91 and 41 in 6 months, 12 months and 18 months follow-up)
² Average Treatment on the Treated analysis, using the treatment assignment as an instrumental variable for the total number of sachets used in the preceding six months. Regression estimates from multiple linear models adjusted for gender, age, parenting practice (whether caregiver used toys played with child the previous day, sang to child previous day), BSID tester and county fixed effects. Standard errors are clustered at the village level. MDI data missing for 214 children (5 in baseline, 101, 88 and 20 in 6 months, 12 months and 18 months follow-up), PDI data missing for 233 children (7 in baseline, 102, 98 and 26 in 6 months, 12 months and 18 months follow-up), and feeding behavior data missing for 206 children (74, 91 and 41 in 6 months, 12 months and 18 months follow-up)
*p-value < 0.05 significant

ISSN: 1471-2458