Routine blood tests are associated with short term mortality and can improve emergency department triage: a cohort study of >12,000 patients

Table 3 Distribution of patients by formalized triage and blood test prediction model

	Validation cohort (n = 5738)
	DEPT	Blood test prediction^a No calibration	Blood test prediction^a Recalibrated
Green	1876 (2.1% [1.4;2.7])	235 (0% [0;0])	2030 (0.4% [0.1;0.7])
Yellow	2272 (3.9% [3.1;4.7])	4205 (1.3% [0.9;1.6])	3174 (3.3% [2.7;3.9])
Orange	1557 (6.0% [4.9;7.2])	821 (8.0% [6.2;9.9])	365 (15.1% [11.4;18.7%])
Red	33 (36.4%[20.0;52.8%])	477 (23.9% [20.1;27.7])	169 (39.6% [32.3;47.0])

Distribution of patients, risk stratified by formalized triage algorithm (DEPT) or the blood test prediction model. 30-day mortality with 95% confidence intervals for individual strata in parenthesis. Initially blood test prediction model overestimates risk of mortality, seen by large number of high risk patients. After recalibration. ^aPredicted risk of 30-day mortality: green <1%, yellow 1–10%, orange 10–25%, red >25%