Table 2.

Predicting Next Day’s PTSS by Previous Night’s Sleep Characteristics

Sleep characteristics	$\widehat{\beta }$	CI	p
Univariate sleep characteristicsa
Sleep duration
Person mean	−7.10	[−14.59, 0.390]	.063
Last nightb	−1.45	[−2.18, −0.73]	<.001
Number of awakenings
Person mean	1.03	[−4.84, 6.88]	.727
Last night	0.70	[−0.07, 1.47]	.073
Sleep quality
Person mean	−18.98	[−42.34, 4.38]	.109
Last night	−2.82	[−4.42, −1.23]	<.001
Trouble falling asleep
Person mean	41.52	[12.35, 70.69]	.006
Last night	10.54	[5.99, 15.09]	<.001
Difficulty staying asleepc
Person mean	73.99	[31.63, 116.35]	.001
Last night	18.73	[13.68, 23.78]	<.001
Multivariate sleep characteristicsd
Sleep duration
Person mean	−2.28	[−9.96, 5.41]	.555
Last night	−0.93	[−1.73, −0.12]	.024
Sleep quality
Person mean	−6.91	[−29.72, 15.91]	.546
Last night	0.41	[−1.45, 2.27]	.667
Trouble falling asleep
Person mean	21.74	[−10.51, 54.00]	.182
Last night	5.65	[0.92, 10.39]	.019
Difficulty staying asleep
Person mean	53.95	[6.98, 100.93]	.025
Last night	16.61	[11.16, 22.06]	<.001

Note. ^aSingle variable analysis adjusted for demographic covariates. ^bThe partitioned last night variable was created as the difference between the person mean and the last night. ^cRandom slope was tested for each sleep characteristic and there was a significant random slope of difficulty staying asleep, and the corresponding fixed effects (person mean $\widehat{\beta }$ = 71. 34, p = .001, and last night $\widehat{\beta }$ = 17.69, p < .001) were similar to the model without the random slope. ^dPerson mean and last night variables of sleep quality were not statistically significant and were removed from the final model. The model included sleep duration, trouble falling asleep, difficulty staying asleep, and demographic covariates