The DUNDRUM-1 structured professional judgment for triage to appropriate levels of therapeutic security: retrospective-cohort validation study

Table 4 Item to outcome correlations and exact agreement with outcome

	Observed proportion in Agreement (n = 316)^a	95% CI of observed proportion in agreement	Spearman rank correlation coefficient^b
TS1: serious violence	0.75	0.70 - 0.80	0.803
TS2: serious self harm	0.61	0.56 - 0.67	0.259
TS3: immediacy of violence risk	0.75	0.70 - 0.80	0.879
TS4: immediacy of self harm risk	0.67	0.62 - 0.73	0.236
TS5: specialist forensic need	0.78	0.74 - 0.83	0.908
TS6: absconding risk	0.80	0.76 - 0.85	0.879
TS7: preventing access	0.78	0.74 - 0.83	0.831
TS8: victim sensitivities	0.80	0.76 - 0.85	0.806
TS9: complex risks	0.72	0.67 - 0.77	0.828
TS10: institutional behaviour	0.71	0.66 - 0.76	0.758
TS11: legal procedure	0.92	0.90 - 0.95	0.921

Exact agreement i.e. those allocated to the level of therapeutic security corresponding to the rating - 0 not followed up or followed up as an out-patient, 1 admitted to open ward, 2 admitted to PICU, 3 or 4 admitted to forensic (medium or high secure) service. Spearman rank correlation coefficient for five point rating (0 to 4) against outcome in four levels. N = 316. ^aBinomial test for observed proportions differing from test proportion 0.5, based on Z approximation all p < 0.001 (2-tailed), ^b all p < 0.001

ISSN: 1471-244X