Table 4.

Causal Mediation Analysis Results Evaluating FEV₁ and FVC as Potential Mediating Traits between African Ancestry and Nocturnal Asthma among African American Participants (n = 1,040)

Mediator	% African Ancestry*	Average Mediation Effect			Average Direct Effect			% Mediated
		Estimate	95% CI	P Value	Estimate	95% CI	P Value	%	P Value
FEV1	>20th to 40th	0.002	6.1 × 10−5 to 0.01	0.04	0.021	−4.5 × 10−4 to 0.04	0.06	9.0	0.08
	>40th to 60th	0.004	5.4 × 10−5 to 0.01	0.05	0.042	2.1 × 10−3 to 0.04	0.04	8.9	0.07
	>60th to 80th	0.007	−1.1 × 10−4 to 0.02	0.05	0.063	4.0 × 10−3 to 0.12	0.03	8.8	0.07
	>80th	0.009	−1.9 × 10−4 to 0.20	0.06	0.081	5.3 × 10−3 to 0.16	0.04	8.7	0.08
FVC	>20th to 40th	0.003	5.0 × 10−4 to 0.01	0.01	0.020	−1.2 × 10−4 to 0.04	0.05	12.4	0.04
	>40th to 60th	0.006	9.8 × 10−4 to 0.01	0.02	0.040	−2.5 × 10−3 to 0.08	0.06	12.7	0.05
	>60th to 80th	0.009	1.5 × 10−3 to 0.02	0.01	0.060	3.4 × 10−4 to 0.12	0.05	12.7	0.03
	>80th	0.012	2.1 × 10−3 to 0.01	0.02	0.079	−9.8 × 10−4 to 0.16	0.05	13.0	0.04

Definition of abbreviation: CI = confidence interval.

Nocturnal asthma was defined as having any night with disrupted sleep due to asthma in the preceding 4 weeks. Mediation effects were estimated using the subjects in the lowest quintile of percent African ancestry (i.e., ≤73.7%) as the “control” group and each successive higher quintile as the “treatment” group. The CIs for both the mediation and direct effect estimates were based on the nonparametric bootstrap with 1,500 samples. The mediation effect was estimated using a linear model of lung function (i.e., FEV₁ or FVC) and a logistic model for the binary outcome of nocturnal asthma.

^*The quintiles of % African ancestry were defined as follows: 20th percentile = 73.7%; 40th percentile = 79.2%; 60th percentile = 83.2%; and 80th percentile = 87.0%.