Table 2.

Characteristics of included studies (cost-effectiveness). AIDS—acquired immunodeficiency syndrome.

Study	Quality/Drummond Score	Design	Population	Intervention	Cost Effectiveness	Resource Requirements
What are the Cost-Effectiveness and Resource Requirements of HIV Testing?
Farnham et al., 1996 [30]	Allowance was made for uncertainty, sensitivity analysis performed around a variety of model inputs. One-way sensitivity analysis in a decision analytic framework. Sensitivity analysis compares basic value with the breakeven value that makes the two strategies equally cost-effective. No range of values tested and no a priori justification for values tested in sensitivity analysis. There was no assumed range, as noted above, but results seem to be sensitive to plausible changes in some model inputs, especially waiting and counselling times.	Decision analytic model, societal perspective. Costs measured in 1992 U.S. dollars.	United States of America	ELISA test, and counselling and testing (C/T) vs. rapid C/T vs. no intervention	ELISA C/T: Average not incremental cost-effectiveness ratios: $1165 per correctly identified case vs. no intervention; rapid C/T $940 per correctly identified case Rapid vs. ELISA: $596 per correctly identified case	ELISA C/T: positive individual $103 per person, negative individual $33. Rapid C/T: positive individual $135 per person, negative individual $33 per person. Low to moderate costs of both strategies.
Kassler et al., 1997 [31]	No allowance was made of uncertainty. No sensitivity analyses.	Cost comparison, societal perspective. Comparison of testing strategies in an HIV clinic. Costs measured in 1993 U.S. dollars.	Individuals attending an anonymous testing clinic and a sexually transmitted disease (STD) clinic in Dallas, Texas	Standard C/T vs. rapid C/T	No incremental cost-effectiveness ratio calculated, not a full economic evaluation.	Cost per person receiving results and counselling: standard $151, rapid $131. Low to moderate cost savings of rapid C/T over standard C/T.
Wilkinson et al., 1997 [32]	No allowance was made for uncertainty. No sensitivity analyses.	Cost comparison, prospective comparison of testing strategies in a South African hospital. Costs measured in 1996 South African rand.	Resource-poor setting: adult inpatients of a rural South African district hospital	ELISA C/T vs. single rapid C/T vs. double rapid C/T. The double rapid strategy consists of two different rapid tests: a Capillus test and an Abbott test.	N/A	Cost per person counselled post-test: single rapid R 14–31.2, double rapid R 45.2, ELISA R 83.8. Cost savings of single rapid test.
Kallenborn et al., 2001 [33]	Some allowance made for uncertainty. Sensitivity analysis limited and discursive. Not statistically rigorous. No range of values for sensitivity analysis provided. Results overall not sensitive to whether basic or expanded regimen used, but as noted sensitivity analysis was incomplete.	Cost comparison study, retrospective chart review of Health Care Workers in an emergency department. Costs measured in 1999 U.S. dollars.	Healthcare workers	Rapid testing vs. ELISA testing	N/A. This is just a cost comparison	Total costs for 17 patients: ELISA $5966, Rapid test $466. Cost savings of switching from ELISA testing to rapid testing in health care workers.
Ekwueme et al., 2003 [34]	Allowance made for uncertainty. One-way sensitivity analysis performed in a cost analysis model. Range of sensitivity analysis is +/− 50% of the base value, or as wide as possible in the absence of hard data. Rank order of two-step rapid relative to standard C/T sensitive to the return rate for standard C/T, but one-step rapid consistently least expensive.	Cost analysis study using a decision analysis model, costs estimated from both societal and provider perspective, in 2000 U.S. dollars	United States of America	Standard ELISA C/T vs. both one-step (multiple rapid assays) and two-step rapid C/T (i.e., with a confirmatory Western blot test)	N/A	From both a provider and societal perspective, costs vary based on sero-status. However, one-step rapid testing is consistently the lowest cost option, and two-step rapid testing tends to be the highest cost. There appear to be cost savings of using a one-step rapid C/T protocol vs. standard ELISA testing or two-step rapid C/T.
Doyle et al., 2005 [35]	Allowance made for uncertainty. One-way sensitivity analysis on “sensitivity, specificity, and positive predictive values of each screening test and confirmatory Western blot test, the costs of each test, and the costs of treatments” in a decision analytic model. Range of values tested in sensitivity analysis appears to be based on published estimates but this is not explicitly stated. The Oraquick rapid test is the dominant strategy over a wide range of assumptions. Results not sensitive to plausible changes.	Decision analysis techniques: decision tree	Low risk Mexican American population, incidence 0.05%	(1) testing with enzyme linked immunosorbent assay that was confirmed by Western blot (2) testing with Oraquick rapid testing that was confirmed by Western blot	Oraquick as the primary screening test for the unknown HIV status of women who were in labor was the most cost-effective at $217,718 per HIV case that was prevented. Assuming a 70-year lifespan, this equals $3111 per life-year gained.	Oraquick cost $98 spent for each child who was HIV negative, ELISA screening cost $491. High cost of ELISA screening in a low-prevalence Mexican American population were from unnecessary treatment of women and infants with false-positive test results. Oraquick has a relatively modest costs.
Paltiel et al., 2005 [36]	Allowance made for uncertainty One-way sensitivity analysis performed. Range of estimates seems to be derived from published estimates/a plausible a priori estimate. Not explicitly stated. Some sensitivity to assumptions regarding background therapy, adherence to ARV therapy, and rates of linkage to care. This does not significantly change the results, may simply change whether screening every three or five years is preferable. Results regarding rapid vs. conventional testing are unclear and sensitive to plausible changes in background testing rates, acceptance and linkage to care, and rate of secondary transmission.	A stochastic model (individual model) of HIV screening and treatment: The cost-effectiveness of preventing AIDS, and a complications model (CEPAC model)	United States of America	(i) routine, voluntary HIV, CTR (counselling, testing and referral); (ii) current practice: background testing OR presentation with opportunistic infections in three target populations:	Compared to current practice, current practice plus one time ELISA costs $36,000 per Quality-Adjusted Life Year (QALY) gained; current practice plus ELISA every 5 years costs $50,000 per QALY gained; current practice plus ELISA every 3 years costs $63,000; current practics plus ELISA every year costs $100,000 per QALY gained	For HIV infected persons only: current practice costs: $78,100 lifetime cost per person; current practice plus one time ELISA costs $80,700; current practice plus ELISA every five years costs $89,000; current practice plus ELISA every 3 years costs $92,500; current practices plus ELISA every year costs $98,600 For general population: current practice costs $32,700 lifetime cost per person, current practice plus one-time ELISA costs $33,800; current practice plus ELISA every five years costs $37,300 current practice plus ELISA every three years costs $38,900; current practices plus ELISA every year costs $41,700. More frequent screening produced large costs, due to screening test cost plus the cost of managing false positives.
Vickerman et al., 2006 [37]	Allowance made for uncertainty. Univariate sensitivity analysis for a number of dimensions. Model input ranges derived from published estimates. The cost-effectiveness of the POC rapid test is sensitive to test cost.	Dynamic compartmental model	Female sex workers	A range of sensitivities of point of care (POC) tests.	If the POC test cost $2 per test (2004 $US), and was 70% sensitive, then POC test would cost $152 per additional HIV infection averted, which is cost-effective. If the cost of the POC test was $1 and the sensitivity was 80%, the cost per HIV infection averted would have been $58, which is cost-effective. When the POC test has a low sensitivity, of 50%, POC is not cost-effective.	Possible cost savings from using POC tests include the reduction in the number of STI clinic attenders receiving treatment. Assuming each test takes an extra 0.3 h to undertake, POC testing costs for 4 years is $13,399 if the test cost $1 and $34 621 if the test cost $3 (in 2004 U.S.$). Moderate costs.