Drs. Burke and Brown suggest that test costs may be lower than the estimates we used and that geographically targeted screening may be more cost-effective than screening all surgeons. We agree that some purchasers will be able to negotiate for relatively low-cost HIV tests. The military HIV screening program is particularly efficient [1]. Our estimate of $57, however, represents the total cost per patient screened and includes the cost of the enzyme immunoassay and any necessary confirmatory tests, counseling before and after the test, and administrative overhead. The cost of the tests alone is a minor component of the overall expenditures per person screened. Our costs reflected those paid by a large purchaser of health care. If, however, the total cost per person screened could be reduced from $57 to $10, the cost-effectiveness of one-time screening program improves from $458000 to $402000 per year of life saved.

As noted by Drs. Burke and Brown, if a screening program were targeted geographically, the prevalence among the surgeons might be higher than that in a national screening program. Our base-case analysis assumed that the prevalence of HIV infection among surgeons was approximately 0.1%. If the prevalence of HIV infection among screened surgeons in a targeted program were 1% (10 times higher than our base-case estimate and substantially higher than any published estimates of HIV infection among physicians) and if the total program costs were reduced to $10 per person screened, one-time screening costs $382000 per year of life saved. Thus, the cost of tests and counseling and the prevalence of HIV infection among surgeons only modestly influence the cost-effectiveness. Their effect is modest because the induced costs (the cost due to the decreased services provided by surgeons identified as having HIV infection) of a program to screen surgeons are high [2]. In contrast, the prevalence of HIV infection in the screened population and the test and counseling costs have a more marked effect on the cost-effectiveness of programs to screen patients because these programs do not incur similar induced costs. In summary, even an efficient, geographically targeted screening program for surgeons would cost substantially more per year of life saved than most commonly accepted medical interventions.