Table 3.

CSC Algorithm performance in small group comparisons^a

Algorithm	(A) Decisions not requiring lottery tie-breaker		(B) Correct selections among decisions not requiring lottery		(C) Overall performance for correct selections
	%	95% CI	%	95% CI	%	95% CI
Groups of two
New Yorkb	52	48–55	72	66–77	61	57–65
Colorado	77	74–82	72	68–76	67	63–71
Raw SOFA	89	87–92	65	62–70	64	60–68
Algorithms + age as tie-breaker
New York + age	90	87–93	70	66–75	68	65–72
Colorado + ageb	93	91–96	69	65–73	68	65–72
Raw SOFA + age	98	97–99	66	62–70	66	62–69
Groups of five
New Yorkb	6	5–7	64	51–75	61	58–63
Colorado	58	56–61	74	70–77	70	67–72
Raw SOFA	78	76–81	66	63–69	64	62–67
Algorithms + age as tie-breaker
New York + age	68	65–71	73	69–76	71	69–74
Colorado + ageb	83	80–85	72	69–75	71	68–74
Raw SOFA + age	95	93–96	67	64–70	66	63–69

^aTriage decisions by CSC algorithms in a simulation of 1,000 random groups of two or five patients. Column A, i.e., two or more patients not tied for the “best” (lowest) priority score. Column B, i.e., survival. Column C, i.e., selecting a surviving patient across all decisions (i.e., all decisions regardless whether selected by priority score or requiring a tie-breaking lottery. Unpaired t tests were conducted to compare all algorithms (with and without age as a tie-breaker) to each other. Nearly all comparisons were significant at p < 0.01. The only non-significant comparisons were New York versus Colorado for groups of two in column B, New York + age versus Colorado + age for groups of two and groups of five in Column B, and New York + age versus Colorado + age for groups of two and groups of five in Column C.

^bIndicates the algorithm that is closest state guidelines. New York’s algorithm as written in the state guidelines does not use a tie-breaker, whereas Colorado’s algorithm does.