Table 2.

Multivariate logistic regression models predicting (a) past year substance use and (b) substance misuse (DAST > 3) using multiple imputation by chained equation (n = 1,282)

(a) Past year substance use
	Main effects		Moderation model
Predictor	AOR	95% CI	AOR	95% CI
Cumulative ACE	1.43	1.34–1.54	1.61	1.47–1.77
Mental illness	0.89	0.60–1.33	2.87	1.46–5.60
ACE*mental illness			0.78	0.69–0.89
(b) Substance misuse (DAST > 3)
	Main effects		Moderation model
Predictor	AOR	95% CI	AOR	95% CI
Cumulative ACE	1.21	1.11, 1.32	1.21	1.08–1.36
Mental illness	2.07	1.20, 3.55	2.14	0.81–4.86
ACE*mental illness			0.99	0.85–1.16

Note: Significant results in bold. ACE = adverse childhood experiences; AOR = adjusted odds ratio; 95% CI = 95% confidence interval. All models adjust for race/ethnicity, sexual orientation, gender identity, annual income, education level, and free lunch status as a child. Past year substance use included as a covariate for model predicting substance misuse