Table 4. One-way and two-way sensitivity analyses of point-of-care CD4 test characteristics on misclassification of patients, clinical outcomes, costs, and cost-effectiveness compared to laboratory-based CD4 tests.

Analysis	CV (Percent)	Bias (Percent)	Sensitivity (Percent)a	Specificity (Percent)a	False Negatives (Percent)b	False Positives (Percent)c	Cost (US Dollars)d	Life Expectancy (Years)d	ICER (US Dollars/YLS)
One-way sensitivity analysis on CV	0		100	100	0.0	0.0	2,850	10.4	470
	5		98	98	2.1	1.6	2,830	10.4	470
	15		95	94	5.4	5.5	2,810	10.3	490
	25		92	89	8.0	10.8	2,800	10.3	500
	32.6		90	85	9.6	15.1	2,800	10.3	500
	45		88	79	11.9	21.2	2,790	10.2	520
	55		87	75	13.4	25.2	2,780	10.2	530
	70		85	70	15.3	29.6	2,770	10.2	550
	100		82	65	18.3	35.2	2,760	10.1	570
One-way sensitivity analysis on bias		−20	96	70	4.0	30.3	2,900	10.5	500
		−15	95	74	5.1	25.9	2,880	10.5	500
		−10	94	78	6.3	21.9	2,850	10.4	500
		−5	92	82	7.6	18.2	2,830	10.3	500
		0	90	85	9.6	15.1	2,800	10.3	500
		5	89	88	10.9	12.0	2,770	10.2	510
		10	87	90	12.8	9.6	2,750	10.2	510
		15	85	93	14.7	7.5	2,720	10.1	520
		20	83	94	16.7	5.8	2,700	10.1	540
Two-way sensitivity analysis	0.15	−20	99	78	0.6	21.7	2,900	10.5	480
	0.15	0	95	95	5.0	5.5	2,800	10.3	480
	0.15	20	84	99	15.7	0.5	2,690	10.0	520
	0.25	−20	98	73	2.5	26.9	2,910	10.5	490
	0.25	0	93	89	7.5	10.6	2,800	10.3	490
	0.25	20	84	97	16.1	3.0	2,690	10.1	530
	0.35	−20	95	69	4.6	31.3	2,900	10.5	500
	0.35	0	90	84	9.6	16.1	2,800	10.3	510
	0.35	20	83	93	16.9	6.8	2,700	10.1	540

Base case in bold.

^aDerived using test performance of percent SD  = 32.6% [36], mean CD4 count (SD) of cohort  = 300/µl (230/µl), and ART-eligibility threshold at CD4 count ≤250/µl,

^bFalse positives: patients who are ART ineligible (true CD4 count>250/µl) but are misclassified as ART eligible.

^cFalse negatives: patients who are ART eligible (true CD4 count ≤250/µl) but are misclassified as ART ineligible.

^dDiscounted at 3%/y.