Correspondence

More on Routine HIV Screening

N Engl J Med 1993; 328:1715-1717June 10, 1993

Article

To the Editor:

The article by Janssen et al. (Aug. 13 issue)1 and the letter by Lurie et al. (Dec. 17 issue)2 raise important issues with regard to the effectiveness of widespread screening for the human immunodeficiency virus (HIV) in acute care hospitals in the United States. These issues are also relevant in less industrialized nations. More than 10 million of the approximately 12 million cases of HIV infection in the world occur overseas; almost 8 million of these are in sub-Saharan Africa alone3. Different conclusions may be drawn when one considers the application of testing policies abroad.

Screening patient populations for HIV demands effective tests. Janssen et al. cite the sensitivity and specificity of the combined testing sequence involving the enzyme-linked immunosorbent assay (ELISA) and the Western blot assay as 0.99 and 0.999, respectively. Lurie et al. suggest that the specificity might be one order of magnitude lower if testing were expanded to include less experienced laboratories. These numbers do not reflect the sensitivity and specificity of tests performed in less industrialized areas of the world. In Africa, the sensitivity of ELISAs ranges from 0.92 to 0.998, and the specificity from 0.514 to 1.0, depending on the test used4-6. The prevalence of HIV type 2, as well as other host factors, may further decrease the specificity7. In addition, the number of indeterminate Western blot results is high, ranging from 2.3 percent to 20 percent4,8. Interpreting these results is difficult, considering that only 6 percent of samples with indeterminate results in one study eventually had positive Western blot results4. These data on test efficacy are reflected by a recent study conducted by the World Health Organization that showed considerable variation among laboratories. Standardized samples with a prevalence of 26 percent that were sent to 28 national reference laboratories in Africa, Europe, the Americas, the eastern Mediterranean region, and Southeast Asia showed an average false positive rate of 13 percent after ELISA and confirmatory tests9. Consequently, screening of patients will produce many more false negative, false positive, and indeterminate results in Africa than in the industrialized world.

In many areas, economic considerations overshadow issues of test performance. Despite high HIV seroprevalence, many countries are unable to screen their blood supply. In Kinshasa, Zaire, a city in which the seroprevalence of HIV among adult blood donors is 5 percent by conservative measurement, one fourth of transfusions remained unscreened in 199010. The conditions affecting health care delivery in the areas of Africa most affected by HIV make hospital-based screening programs unrealistic. In addition to screening, treatment, counseling, and social services are often unavailable. Current discussions of HIV testing should be expanded to include many of the complex issues related to the epidemic throughout the world.

Jonathan Mermin, M.D.
University of California, San Francisco, San Francisco, CA 94105

Reuben Granich
Stanford University School of Medicine, Stanford, CA 94305

10 References

1. 1

   Janssen RS, St Louis ME, Satten GA, et al. HIV infection among patients in U.S. acute care hospitals -- strategies for the counseling and testing of hospital patients. N Engl J Med 1992;327:445-452
   Full Text | Web of Science | Medline

2. 2

   Lurie P, Lowe RA, Avins AL, et al. Undiagnosed HIV infection in acute care hospitals. N Engl J Med 1992;327:1815-1816
   Full Text | Web of Science | Medline

3. 3

   Mann J, Tarantola DJM, Netter TW, eds. AIDS in the world. Cambridge, Mass.: Harvard University Press, 1992:105.
   

4. 4

   Behets F, Disasi A, Ryder RW, et al. Comparison of five commercial enzyme-linked immunosorbent assays and Western immunoblotting for human immunodeficiency virus antibody detection in serum samples from Central Africa. J Clin Microbiol 1991;29:2280-2284
   Web of Science | Medline

5. 5

   Nkengasong J, Van Kerckhoven I, Carpels G, Vercauteren G, Piot P, van der Groen G. HIV screening and confirmation: a simplified and less expensive testing algorithm. Ann Soc Belg Med Trop 1992;72:129-139
   Medline

6. 6

   Danaby RS, Constantine NT, Daise RL, Watts DM. Assessment of serological assays for the diagnosis of HIV-1 infections. J Egypt Public Health Assoc 1991;66:477-490
   Medline

7. 7

   Holzer TJ, Allen RG, Heynen CA, et al. Discrimination of HIV-2 infection from HIV-1 infection by Western blot and radioimmunoprecipitation analysis. AIDS Res Hum Retroviruses 1990;6:515-524
   CrossRef | Web of Science | Medline

8. 8

   Delaporte E, Peeters M, Simon F, et al. Interpretation of antibodies reacting solely with human retroviral core proteins in western equatorial Africa. AIDS 1989;3:179-182
   CrossRef | Web of Science | Medline

9. 9

   Snell JJ, Supran EM, Esparza J, Tamashiro H. World Health Organization quality assessment programme on HIV testing. AIDS 1990;4:803-806
   CrossRef | Web of Science | Medline

10. 10

    N'tita I, Mulanga K, Dulat C, et al. Risk of transfusion-associated HIV transmission in Kinshasa, Zaire. AIDS 1991;5:437-439
    CrossRef | Web of Science | Medline

To the Editor:

Janssen et al. conclude that routine voluntary testing for HIV infection in patients 15 to 54 years of age is warranted in hospitals at which the number of AIDS cases per 1000 discharges is greater than 11. We undertook a study to validate the association between HIV seroprevalence and hospital-discharge data and to estimate the cost of implementing such a policy at the Johns Hopkins Hospital.

Discharge data for the department of medicine in 1991 revealed 13.6 cases of symptomatic HIV infection or AIDS per 1000 discharges, and 87.9 cases per 1000 discharges for the group 15 to 54 years of age. When one applies the formula of Janssen et al., the expected overall HIV prevalence would be 13.6 (per 1000) × 10.4 = 14.1 percent (95 percent confidence interval, 11.96 to 16.32). To compare this estimate with the actual prevalence of HIV, serologic and demographic data were collected for one month for patients on medical wards. Of 719 consecutive admissions, 14.4 percent of all patients (103 of 719) and 26.8 percent of patients 15 to 54 years old (98 of 366) tested HIV-positive. The overall prevalence (14.4 percent) was nearly identical to the calculated rate (14.1 percent), supporting the use of discharge data as a surrogate for HIV prevalence.

From the data of Janssen et al., we next developed a mathematical model of the cost and yield of HIV testing, based on a cost of $45 per patient tested (Figure 1Figure 1Cost and Yield of Routine Voluntary HIV Screening among Hospital Patients 15 to 54 Years of Age, According to the AIDS-Diagnosis Rate per 1000 Discharges.). After calculating the hospital AIDS-diagnosis rate per 1000 discharges, physicians and hospital administrators can estimate their yield and cost per HIV infection identified. In our seroprevalence study, 366 of 719 patients admitted (50.9 percent) were in the 15-to-54-year-old age group; 80 were known to be HIV-positive, 15 were known to be HIV-negative, and 271 had unknown HIV status. The yield of detection among the patients of unknown HIV status was 6.6 percent (18 of 271), at a cost of $12,195. This cost would identify 18 otherwise unknown HIV infections per month at a cost of $677.50 per newly identified case. The benefits of this policy include the initiation of therapy and HIV counseling to prevent transmission. This cost is relatively low, as compared with those of other preventive screening tests, such as mammography, Pap smears, and stool testing for occult blood2-4.

The yield of detection (6.6 percent) and the HIV prevalence among patients 15 to 54 years of age (26.8 percent) in our hospital are at variance with our model, because the formula of Janssen et al. does not adjust for patients with known HIV status. If we had ignored the results of previous HIV testing, our costs would have been lower and consistent with the model. When HIV prevalence is high, this adjustment lowers the yield and increases the cost of each new case identified. In hospitals with a substantially lower HIV prevalence, the cost of routine voluntary testing will be exponentially higher and will probably remain the subject of debate.

Linda M. Mundy, M.D.
Ramana Gopalan, M.D., M.P.H.
Thomas C. Quinn, M.D., M.Sc.
Johns Hopkins Hospital, Baltimore, MD 21205

4 References

1. 1

   Janssen RS, St Louis ME, Satten GA, et al. HIV infection among patients in U.S. acute care hospitals -- strategies for the counseling and testing of hospital patients. N Engl J Med 1992;327:445-452
   Full Text | Web of Science | Medline

2. 2

   Brandeau ML, Owens DK, Sox CH, Wachter RM. Screening women of childbearing age for human immunodeficiency virus: a cost-benefit analysis. Arch Intern Med 1992;152:2229-2237
   CrossRef | Web of Science | Medline

3. 3

   Udvarhelyi IS, Colditz GA, Rai A, Epstein AM. Cost-effectiveness and cost-benefit analyses in the medical literature: are the methods being used correctly? Ann Intern Med 1992;116:238-244
   Web of Science | Medline

4. 4

   Guide to clinical preventive services: report of the U.S. Preventive Services Task Force. Baltimore: Williams & Wilkins, 1989.