Table 2.

Staff primary outcomes^a

	Coefficient (95% CI)	P
Views of the Therapeutic Environment (VOTE)b
System
Routine care system	reference
Triage system	1.68 (−2.05 to 5.41)	0.379
Interaction: system (routine) × phase
Baseline (routine)	reference
Phase 1	6.58 (3.83 to 9.33)	<0.001*
Phase 2	0.25 (−3.11 to 3.16)	0.987
Phase 3	4.80 (1.30 to 8.30)	0.007*
Interaction: system (triage) × phase
Baseline (triage)	reference
Phase 1	0.94 (−1.81 to 3.70)	0.503
Phase 2	−0.13 (−2.90 to 2.63)	0.924
Phase 3	4.78 (1.93 to 7.63)	0.001
Maslach Burnout Inventory Human Services Surveyb
System
Routine care system	reference
Triage system	0.86 (−3.75 to 5.47)	0.715
Interaction: system (routine) × phase
Baseline (routine)	reference
Phase 1	5.02 (1.44 to 8.60)	0.006*
Phase 2	2.47 (−1.47 to 6.40)	0.220
Phase 3	8.61 (4.21 to 13.01)	<0.001*
Interaction: system (triage) × phase
Baseline (triage)
Phase 1	−0.87 (−4.60 to 2.86)	0.648
Phase 2	1.06 (−2.60 to 4.72)	0.569
Phase 3	1.76 (−1.95 to 5.46)	0.352
Index of Work Satisfactionb
System
Routine care system	reference
Triage system	8.06 (0.95 to 15.16)	0.026*
Phase
Baseline	reference
Month 6	8.13 (3.23 to 13.05)	0.001*
Month 12	5.06 (−0.16 to 10.27)	0.057
Month 18	11.90 (6.36 to 17.43)	<0.001*

^a.Intraclass correlation coefficients were 0.71, 0.64 and 0.72 for the three outcomes respectively.

^b.Mixed-model linear regression including a random effect for individuals and covarying for phase (categorical, coefficients use the baseline as a comparison). A system × phase interaction was tested in each model and results are presented where this interaction was shown to be significant from a likelihood ratio test.

^*P<0.05 after performing a sensitivity analysis of the same model with an additional level of clustering at the ward level. Significance of all other results did not change.