On the use of the outcome variable “small for gestational age” when gestational age is a potential mediator: a maternal asthma perspective

Table 5 Basic scenarios without confounding with heterogeneity of errors

Scenario (model, interpretation of effect)	mean	se	null ICS effect?
Basic 1 (M₁, total)	−0.2350	0.2398	yes
Basic 1 (M₂, direct)	−0.0126	0.1931	yes
Basic 1 (M₃, total)	0.0029	0.0022	yes
Basic 1 (M₄, direct)	−0.0008	0.0015	yes
Basic 2 (M₁, total)	−366.75	0.2388	no
Basic 2 (M₂, direct)	−0.0374	0.2258	yes
Basic 2 (M₃, total)	1.1826	0.0018	no
Basic 2 (M₄, direct)	−0.0008	0.0015	yes
Basic 3 (M₁, total)	−100.24	0.2398	no
Basic 3 (M₂, direct)	−100.01	0.1931	no
Basic 3 (M₃, total)	0.3250	0.0020	no
Basic 3 (M₄, direct)	0.4236	0.0014	no
Basic 4 (M₁, total)	−466.75	0.2388	no
Basic 4 (M₂, direct)	−100.04	0.2258	no
Basic 4 (M₃, total)	1.5032	0.0018	no
Basic 4 (M₄, direct)	0.4316	0.0014	no

Legend. Mean of the estimated exposure (ICS) effect (mean difference or log odds ratio) on the outcome (BW, LBW or SGA) based on 1000 samples of size 20,000, with Monte Carlo standard error (se)
Basic Scenario 1: no ICS effect on BW; Basic Scenario 2: indirect ICS effect on BW;
Basic Scenario 3: direct ICS effect on BW; Basic Scenario 4: direct and indirect ICS effect on BW
M₁: BW ~ ICS; M₂: BW ~ ICS + GA; M₃: LBW ~ ICS; M₄: SGA ~ ICS;
null ICS effect = yes if 0 ∈ mean ± 1.96se
Abbreviations: BW birth weight, GA gestational age, ICS inhaled corticosteroids, LBW low birth weight, SGA small for gestational age

ISSN: 1471-2288