Skip to main content

Table 2 Mediation analysis of the effect of log cord blood lead concentrations on cognitive scores when mediated by stunting status at both study sites

From: Stunting and lead: using causal mediation analysis to better understand how environmental lead exposure affects cognitive outcomes in children

Mediation Analysis Effects

Coefficient

95% confidence interval

p value

Lower bound

Upper bound

Pabna

 Total effect

− 0.160

− 0.787

0.467

0.617

 Controlled direct effect

0.560

− 0.169

1.289

0.132

 Reference interaction

− 0.779

− 1.248

− 0.310

0.001

 Mediation interaction

0.128

− 0.033

0.289

0.120

 Pure indirect effect

− 0.069

− 0.182

0.044

0.230

 Proportion mediated

− 0.368

− 1.922

1.186

0.642

 Proportion attributable to interaction

4.067

− 11.814

19.947

0.616

Sirajdikhan

 Total effect

0.001

− 0.460

0.463

0.996

 Controlled direct effect

0.205

− 0.507

0.918

0.573

 Reference interaction

− 0.209

− 0.792

0.374

0.483

 Mediation interaction

0.002

− 0.023

0.028

0.850

 Pure indirect effect

0.002

− 0.023

0.027

0.849

 Proportion mediated

4.622

− 2022.850

2032.095

0.996

 Proportion attributable to interaction

− 196.420

− 86513.200

86120.360

0.996

  1. Analysis conducted using med4way command in Stata version 16; for Sirajdikhan, a(0) was set to 1.381, the 25th percentile of log cord blood lead μg/dL; a(1) was set to 2.268, the 75th percentile of log cord blood lead μg/dL; and m was set to 0; for Pabna, a(0) was set to 0.149, the 25th percentile of log cord blood lead μg/dL; and a(1) was set to 0.895, the 75th percentile of log cord blood lead μg/dL. Other included covariates were sex of child, child’s age in months, mother’s approximate age in years, education category, IQ, HOME score, environmental smoke exposure, protein intake, child’s birth length, child’s birth weight, mother’s weight, mother’s height, log cord blood arsenic and manganese, log blood lead, arsenic, and manganese collected at 20–40 months