. 2021 Jan 4;38(5):528–544. doi: 10.1002/da.23129

Table 4.

Final multivariable models predicting 30‐day suicidal thoughts and behaviors (STB; n = 5164)

			% (SE) or Med (SE) (IQR)	Any STB (n = 395)	Passive ideation only (n = 243)	Active ideation without plan or attempt (n = 41)	Active ideation with plan or attempt (n = 111)
		n	% (SE) or Med (SE) (IQR)	OR (95% CI)^a	OR (95% CI)^a	OR (95% CI)^a, ^b	OR (95% CI)^a
Individual characteristics	Age
	Spline 18‐29 years			0.91 (0.82–1.01)	0.94 (0.84–1.04)	0.81 (0.71–0.92)^*	0.94 (0.82–1.07)
	Spline 30‐49 years			1.04 (1.01–1.07)^*	1.03 (0.99–1.08)	1.02 (0.96–1.08)	1.06 (1.03–1.10)^*
	Spline 50+ years			0.93 (0.89–0.98)^*	0.95 (0.88–1.03)	0.76 (0.61–0.96)^*	0.92 (0.86–0.98)^*
	Gender—female (vs. male)	4167	77.3 (1.3)	0.82 (0.69–0.98)^*	1.52 (1.04–2.22)^*	0.45 (0.24–0.83)^*	0.38 (0.29–0.51)^*
	Marital status—married (vs. single, divorced, or legally separated, or widowed)	2411	47.9 (2.0)	0.73 (0.55–0.96)^*	0.78 (0.48–1.25)	0.71 (0.29–1.74)	0.65 (0.44–0.96)^*
	Children in care (vs. no children in care)	1974	38.7 (0.8)	0.61 (0.42–0.89)^*	0.60 (0.41–0.87)^*	0.56 (0.23–1.38)	0.61 (0.35–1.06)
	Lifetime mood disorder before onset COVID‐19 outbreak	568	12.1 (0.7)	2.92 (2.40–3.54)^*	2.09 (1.46–2.99)^*	5.05 (2.68–9.52)^*	4.21 (2.70–6.58)^*
	Lifetime anxiety disorder before onset COVID‐19 outbreak	1893	37.9 (1.1)	1.90 (1.40–2.58)^*	1.63 (1.24–2.14)^*	2.24 (1.17–4.29)^*	2.39 (1.30–4.40)^*
	Frequency of direct exposure to COVID‐19 patients (scaled 0–4)		1.9 (0.1) (0.9–3.1)	1.05 (0.95–1.16)	1.10 (1.02–1.19)^*	1.07 (0.82–1.39)	0.95 (0.71–1.25)
	Frequency of working at home (scaled 0–5)		0.0 (0.2) (0.0–0.1)	0.84 (0.73–0.96)^*	0.84 (0.75–0.95)^*	0.94 (0.69–1.26)	0.78 (0.60–1.00)^*
Potentially modifiable contextual factors	Perceived lack of communication, coordination, personnel, or supervision (scaled 0.0–4.0)		1.9 (0.1) (1.0–2.5)	1.44 (1.27–1.63)^*	1.35 (1.13–1.61)^*	1.85 (1.34–2.56)^*	1.59 (1.34–1.90)^*
	Perceived inefficiency of protective equipment (scaled 0–3)		0.3 (0.0) (0.0–1.1)	0.89 (0.73–1.08)	0.90 (0.71–1.15)	0.86 (0.64–1.15)	0.83 (0.62–1.11)
	Financial stress (scaled 0.0–4.0)^c		0.7 (0.0) (0.0–1.7)	1.34 (1.26–1.43)^*	1.28 (1.14–1.44)^*	1.10 (0.86–1.41)	1.50 (1.24–1.83)^*

Note: Individual characteristics and potentially modifiable contextual factors included in the final multivariable models are selected using the lasso shrinkage method, optimizing the Bayesian Information Criterion.

Abbreviations: COVID‐19, coronavirus disease 2019; IQR, interquartile range.

Statistically significant (α = .05).

^{^a}

All analyses adjust time of survey (weeks), hospital membership, and all predictors shown in the rows.

^{^b}

Due to data sparseness, the model was estimated using Penalized Maximum Likelihood Estimation (Firth‐type Estimation).

^{^c}

Likelihood ratio tests for linearity of effect were significant, and models including polynomials suggest a better model fit when including a quadratic term. For ease of interpretability, we present here the models including the linear term only; models including the quadratic term are presented in Table S6.