Table 4.

Associations of the Intervention With Depression and Functional Status of Evaluators^a

Outcome/Predictors	Patients			Family
	Trainees Evaluated/Evaluations, No.	b (95% CI)b	P Valuec	Trainees Evaluated/Evaluations, No.	b (95% CI)b	P Valuec
Depression scored
Interventione	193/1121	2.20 (0.62 to 3.78)	.006	108/463	1.07 (−1.09 to 3.23)	.33
Groupf		−0.77 (−1.93 to 0.40)	.20		0.20 (−1.56 to 1.96)	.82
Study periodg		−0.85 (−1.84 to 0.14)	.09		−0.28 (−1.49 to 0.92)	.64
Study siteh		0.11 (−0.64 to 0.86)	.78		−0.58 (−1.89 to 0.74)	.39
Stratumi			.003
R1		0.000			0.000
R2		−0.42 (−1.58 to 0.74)	.48		0.55 (−0.86 to 1.96)	.45
R3/fellow		−1.63 (−2.54 to −0.72)	<.001		−0.48 (−2.32 to 1.36)	.61
Physical statusj	178/938
Interventione		1.26 (−1.22 to 3.75)	.32
Groupf		0.10 (−1.56 to 1.75)	.91
Study periodg		−0.46 (−2.04 to 1.13)	.57
Study siteh		0.53 (−0.71 to 1.77)	.41
Stratumi			<.001
R1		0.000
R2		1.83 (0.34 to 3.33)	.02
R3/fellow		2.87 (1.11 to 4.64)	.001
Mental statusk	178/938
Interventione		0.34 (−2.52 to 3.20)	.82
Groupf		−0.76 (−2.83 to 1.31)	.47
Study periodg		0.59 (−1.08 to 2.25)	.49
Study siteh		−0.24 (−1.59 to 1.12)	.73
Stratumi			.003
R1		0.000
R2		0.78 (−1.11 to 2.68)	.42
R3/fellow		3.01 (1.17 to 4.86)	.001

Abbreviations: R1, first-year resident; R2, second-year resident; R3, third-year resident.

^aAll outcomes were modeled within evaluator types (patient or family) using robust linear regression models, evaluators clustered under trainees, and estimates based on restricted maximum likelihood. The model for each outcome and evaluator group included only the predictors shown on the rows, included only 1 evaluation per evaluator, and included only those trainees who had at least 1 valid outcome score for both the preintervention and postintervention periods. Results with mean scores are shown in eTables 5–7 in Supplement.

^bb = expected change in the outcome score with a 1-point increase in the row predictor.

^cAll P values and 95% CIs were based on 2-tailed tests. P values for overall differences between strata were based on likelihood ratio tests; other P values were based on Wald test.

^dStandard composite measure for the PHQ-8 Depression Questionnaire.

^eInteraction term computed as randomization group × time point (0 = control group or preintervention evaluation for intervention group; 1 = postintervention evaluation for intervention group).

^fMain effect of randomization group (0 = control, 1 = intervention).

^gMain effect of time: effect of the pre/post indicator (ie, whether the evaluator’s last encounter with the trainee occurred before [0] or after [1] the workshop series to which the trainee was assigned).

^h0 = University of Washington; 1 = Medical University of South Carolina.

ⁱStratified randomization was done within the 2 study sites, using 5 randomization strata: (R1; R2; R3/fellow; students in nurse practitioner or nursing education programs; and community nurses, nurse practitioners or registered nurses). Neither of the nursing strata had cases in these analyses and are omitted from the Table. The R1 stratum was the reference group in all analyses.

^jNorm-based standardized physical component score from the 12-item Short-Form Health Survey questionnaire. Intercept (estimated mean for cases scoring 0 on all predictors) = 28.777. Assessed in patients only.

^kNorm-based standardized mental component score from the 12-item Short-Form Health Survey questionnaire. Intercept (estimated mean for cases scoring 0 on all predictors) = 42.930. Assessed in patients only.