Table 4.

Comparison of the S₁- and S₂-based tripartite rules. The optimal cut-off points based on S₁ are transformed back to the original scale of CD4 count. The numbers in parentheses are standard errors.

	λ	φ	cut-off points		FNR	FPR	R3	TMR
			lower	upper
S1	.25	.00	17 (17)	17 (17)	.98 (.02)	.00 (.00)	.06 (.01)	.24 (.02)
		.15	17 (9)	201 (18)	.72 (.04)	.00 (.00)	.04 (.00)	.18 (.01)
		.30	17 (13)	284 (13)	.46 (.04)	.00 (.00)	.03 (.00)	.12 (.01)
	.50	.00	120 (64)	120 (64)	.93 (.12)	.03 (.04)	.12 (.01)	.26 (.02)
		.15	90 (16)	216 (39)	.63 (.06)	.02 (.01)	.08 (.01)	.17 (.01)
		.30	17 (15)	284 (32)	.45 (.04)	.01 (.01)	.05 (.01)	.12 (.01)
	.75	.00	302 (45)	302 (45)	.43 (.06)	.26 (.06)	.13 (.01)	.30 (.04)
		.15	216 (51)	317 (50)	.40 (.07)	.13 (.06)	.10 (.01)	.20 (.04)
		.30	226 (67)	417 (81)	.30 (.07)	.14 (.06)	.08 (.01)	.18 (.04)
S2	.25	.00	.64 (.04)	.64 (.04)	.91 (.04)	.01 (.00)	.06 (.00)	.23 (.01)
		.15	.39 (.01)	.75 (.04)	.62 (.04)	.00 (.00)	.04 (.00)	.16 (.01)
		.30	.29 (.01)	.71 (.04)	.42 (.04)	.00 (.00)	.03 (.00)	.11 (.01)
	.50	.00	.53 (.07)	.53 (.07)	.79 (.08)	.04 (.02)	.11 (.01)	.23 (.01)
		.15	.37 (.01)	.66 (.06)	.60 (.05)	.01 (.01)	.07 (.01)	.16 (.01)
		.30	.28 (.01)	.67 (.06)	.43 (.04)	.01 (.01)	.05 (.01)	.11 (.01)
	.75	.00	.26 (.04)	.26 (.04)	.35 (.08)	.30 (.08)	.12 (.01)	.31 (.04)
		.15	.26 (.04)	.34 (.05)	.32 (.08)	.21 (.07)	.09 (.01)	.24 (.03)
		.30	.19 (.03)	.34 (.08)	.25 (.06)	.13 (.05)	.07 (.01)	.16 (.03)