Table 3.

Mediation effect between SARS-CoV-2 infection and score of gross motor domain of ASQ-3.

Mediation path	β (95%CI)	p-value	Proportion mediated (a*b/c)
COVID-19 effect on mediator (Path a: X → M)	34.14 (26.90, 41.38)	<0.001	84.73%
Mediator effect on Gross motor (Path b: M → Y)	−0.09 (−0.16, −0.02)	0.center
Indirect effect (Path a*b: X → M)	−3.14 (−6.01, −0.20)	–
Direct effect (Path c′: X → Y adjusted M)	−0.57 (−4.21, 3.08)	0.759
Total effect (Path c: X → Y adjusted M)	−3.71 (−6.48, −0.93)	0.009

ASQ-3, Age and Stage Questionnaire Chinese version; CI, confidence interval. X, independent variable (SARS-CoV-2 infection). M, mediator (mother–infant separation days). Y, dependent variable (score of gross motor development). Path a, the effect of SARS-CoV-2 infection on mediator. Path b, the effect of mediator on gross motor development. Path a^*b (indirect effect), the mediated effect of SARS-CoV-2 infection on gross motor development by mother–infant separation days. Path c′ (direct effect), the remaining effect of SARS-CoV-2 infection on gross motor development extent not explained by the mediators included in the model. Path c (total effect), the total effect of SARS-CoV-2 infection on gross motor development. In the mediation model, we adjusted maternal education (bachelor or above vs. high school or below), low birthweight (yes vs. no), infant gender (boy vs. girl), preterm (yes vs. no), admitted to NICU (yes vs. no), and breastfeeding at 3 months after birth (yes vs. no).