Table 3.

Binary logistic regression modelling

	Omnibus test df = 1		Cox & Snell	Nagelkerke	Correctly classified	B	SE	Wald df = 1		Odds ratio/Exp (B)	95% CI of odds ratio
	X 2	p	R 2	R 2	%			X 2	p		Lower	Upper
Outcome: Positive moves
MODEL 1	8.7	0.003	0.139	0.187	65.5
DUNDRUM-3 clinician rated						−0.124	0.045	7.55	0.006	0.884	0.809	0.965
Constant						1.725	0.753	5.25	0.022	5.614
MODEL 2, add HCR-20 (C + R)	15.7	0.001	0.237	0.317	75.9
HCR-20 (C + R)						−0.264	0.078	11.33	0.001	0.768	0.659	0.896
Constant						1.423	0.543	6.87	0.009	4.149
Outcome: Negative moves
MODEL 3	9.06	0.003	0.145	0.262	86.2
DUNDRUM-4 clinician-patient difference						0.319	0.125	6.54	0.011	1.375	1.077	1.756
Constant						−4.730	1.347	12.33	0.001	0.009
MODEL 4, add HCR-20 (C + R)	9.06	0.003	0.145	0.262	86.2
DUNDRUM-4 clinician-patient difference						0.319	0.125	6.54	0.011	1.375	1.077	1.756
Constant						−4.730	1.347	12.33	0.001	0.009
Outcome: Conditional discharge
MODEL 5	26.2	0.001	0.336	0.604	95.3
DUNDRUM-3 clinician rated						−0.553	0.214	6.69	0.010	0.575	0.378	0.875
Constant						3.965	1.864	4.53	0.033	52.737
MODEL 6 add HCR-20 (C + R)	26.2	0.001	0.336	0.604	95.3
DUNDRUM-3 clinician rated						−0.553	0.214	6.69	0.010	0.575	0.378	0.875
Constant						3.965	1.864	4.53	0.033	52.737

All models include clinician rated DUNDRUM-3 and DUNDRUM-4, patient self-rated DUNDRUM-3 and DUNDRUM-4 and the differences between pairs. Models 2, 4 and 6 also include the HCR-20 (C + R) dynamic score. All models: forward stepwise likelihood ratio.