Sleep disturbances predict active suicidal ideation the next day: an ecological momentary assessment study

BMC Psychiatry

Table 2 Parameter estimates for the multilevel models with sleep disturbances as the predictor variable

Models	Est.	SE	t(df)	p
Predictor	Est.	SE	t(df)	p
Model 1 (Active SI)
Intercept	3.19	0.18	17.76 (72)	<.001
Sleep	0.18	0.09	2.04 (337)	.042*
Deviance test^a: χ²(2) = 4.63, p = .097 Quasi R² (RI)^b: 7.11% Quasi R² (RS)^c: 11.3%
Model 2 (Passive SI)
Intercept	4.53	0.26	17.19 (72)	<.001
Sleep	0.10	0.11	0.95 (337)	.343
Deviance test^a: χ²(2) = 2.84, p = .240 Quasi R² (RI)^b: − 1.8% Quasi R² (RS)^c: 3.8%
Model 3 (Depression)
Intercept	5.94	0.21	28.26 (72)	<.001
Sleep	0.15	0.13	1.16 (337)	.245
Deviance test^a: χ²(2) = 0.46, p = > .500 Quasi R² (RI)^b: −1.1%Quasi R² (RS)^c: − 0.8%
Model 4 (MIS Sum Score)
Intercept	8.16	0.24	33.54 (72)	<.001
Sleep	0.46	0.27	1.67 (321)	.096
Deviance test^a: χ²(2) = 0.33, p = > .500 Quasi R² (RI)^b: 1.9% Quasi R² (RS)^c: − 1.4%
Model 4a (MIS Item 1)
Intercept	2.16	0.10	22.52 (72)	<.001
Sleep	0.33	0.12	2.67 (321)	.008**
Deviance test^a: χ²(2) = 0.61, p = > .500 Quasi R² (RI)^b: −1.2% Quasi R² (RS)^c: 1.6%
Model 4b (MIS Item 2)
Intercept	1.47	0.06	23.91 (72)	<.001
Sleep	−0.03	0.11	-0.26 (321)	.799
Deviance test^a: χ²(2) = 0.90, p = > .500 Quasi R² (RI)^b: 1.4% Quasi R² (RS)^c: 2.1%
Model 4c (MIS Item 3)
Intercept	2.86	0.10	29.52 (72)	<.001
Sleep	0.15	0.12	1.25 (321)	.212
Deviance test^a: χ²(2) = 0.39, p = > .500 Quasi R² (RI)^b: −1.7% Quasi R² (RS)^c: −1.4%
Model 4d (MIS Item 4)
Intercept	1.66	0.08	22.17 (72)	<.001
Sleep	0.00	0.11	0.03 (321)	.974
Deviance test^a: χ²(2) = 0.65, p = > .500 Quasi R² (RI)^b: 2.7% Quasi R² (RS)^c: 3.1%

Notes. All reported models are random intercept (RI) models. Predictor variable is sleep disturbances. n (level 1 – daily means) = 443; n (level 2 - participants) = 74; Est. = unstandardized regression coefficient; SE = standard error; Model 1 = active suicidal ideation as outcome variable; Model 2 = passive suicidal ideation as outcome variable; Model 3 = depression as outcome variable; Model 4 = impulsivity with MIS sum score as outcome variable; Model 4a = MIS Item 1 as outcome variable; Model 4b = MIS Item 2 as outcome variable; Model 4c = MIS Item 3 as outcome variable; Model 4d = MIS Item 4 as outcome variable;*p < .05, **p < .01; ^aDeviance tests: random intercepts vs. random slopes model, ^bQuasi R indicates the percentage of residual variance in the outcome variable that is explained when adding the models’ level 1 predictors compared to the baseline models (RI models), ^cQuasi R indicates the percentage of residual variance in the outcome variable that is explained when adding the models’ level 1 predictor compared to the baseline models (non significant RS models)

ISSN: 1471-244X