Appendix K Details of the Committee's Models and Assumptions

Michael Stoto and Maria Hewitt

The conclusions and recommendations in this report rely, partly, on statistical calculations of the predictive value of prenatal HIV testing, economic evaluations of prenatal HIV screening programs, and process evaluations of strategies to reduce perinatal transmission. This appendix is intended to provide detailed information about the models and assumptions that the committee used to support its conclusions in Chapters 6 and 7.

Predictive Value Of HIV Testing And Cost-Effectiveness Of HIV Screening And Treatment In Pregnancy

Although the methods for testing pregnant women for HIV are the same as for other individuals, the relatively low prevalence of HIV in pregnant women in most areas (compared to individuals who seek or are referred for testing) affects the predictive value of the test—the lower prevalence rates correspond to higher false positive rates. In addition, the low cost of HIV testing when done routinely in the context of prenatal care (as recommended in this report) affects cost-effectiveness calculations in this setting.

In support of the recommendations in Chapter 7, this appendix estimates the predictive value of prenatal HIV testing, reviews existing economic evaluations of prenatal screening programs, and develops a simple model to evaluate prenatal HIV testing in clinical and economic terms. The primary difference between this and existing models is that the costs of initial ELISA (enzyme-linked immunosorbent assay) tests are limited to the marginal costs of including HIV in the



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 304
--> Appendix K Details of the Committee's Models and Assumptions Michael Stoto and Maria Hewitt The conclusions and recommendations in this report rely, partly, on statistical calculations of the predictive value of prenatal HIV testing, economic evaluations of prenatal HIV screening programs, and process evaluations of strategies to reduce perinatal transmission. This appendix is intended to provide detailed information about the models and assumptions that the committee used to support its conclusions in Chapters 6 and 7. Predictive Value Of HIV Testing And Cost-Effectiveness Of HIV Screening And Treatment In Pregnancy Although the methods for testing pregnant women for HIV are the same as for other individuals, the relatively low prevalence of HIV in pregnant women in most areas (compared to individuals who seek or are referred for testing) affects the predictive value of the test—the lower prevalence rates correspond to higher false positive rates. In addition, the low cost of HIV testing when done routinely in the context of prenatal care (as recommended in this report) affects cost-effectiveness calculations in this setting. In support of the recommendations in Chapter 7, this appendix estimates the predictive value of prenatal HIV testing, reviews existing economic evaluations of prenatal screening programs, and develops a simple model to evaluate prenatal HIV testing in clinical and economic terms. The primary difference between this and existing models is that the costs of initial ELISA (enzyme-linked immunosorbent assay) tests are limited to the marginal costs of including HIV in the

OCR for page 304
--> standard panel of prenatal blood tests with no additional blood samples, and there is no cost for office visits or counseling (since the testing is done in the context of prenatal care). Predictive Value of Prenatal HIV Testing The positive predictive value (PPV) of a test is the probability that an individual with a positive test result is truly infected with HIV. The key assumptions for this calculation are the following. First, a two-stage testing procedure is used, as described by Pins and colleagues (1997). One specimen is subjected to an initial ELISA test, and, if positive, to a second. If repeatedly positive, the same specimen is subjected to a confirmatory Western blot test. Second, the sensitivity of the repeated ELISA test is 100%, and the specificity is 0.999 (Pins et al., 1997). Third, the prevalence of HIV in pregnant women ranges from 1 per 10,000 to 100 per 10,000 (or 1%). This range parallels the range of values found in the 1994 Survey of Childbearing Women. Table K.1 displays the number of true positives (women truly HIV-positive) and the number of positive ELISA tests that would result for every 10,000 pregnant women tested for a given HIV prevalence rate. The table also shows the PPV for a range of prevalence values. As is generally the case, the positive predictive value of the test is lower where the prevalence of HIV is also low. If the prevalence of HIV in the population tested is above 20 per 10,000 (as is the case in about seven states, the District of Columbia, and Puerto Rico), the PPV exceeds 67%. If the prevalence is as low as 2 per 10,000 (as is the case in Utah or Oklahoma), the PPV is only about 17%. This means that there is less than a one TABLE K.1 Cost-Effectiveness of HIV Screening Incorporated into Prenatal Care         Cost of ELISA (dollars) Per 10,000 Women Tested At $5/Test At $3/Test Prevalence TRUE Positives Positive ELISA PPV Total $/True + Total $/True + 0.0001 1 11 0.091 51,100 51,100 31,100 31,100 0.0002 2 12 0.167 51,200 25,600 31,200 15,600 0.0005 5 15 0.333 51,500 10,300 31,500 6,300 0.001 10 20 0.500 51,999 5,200 31,999 3,200 0.002 20 30 0.667 52,998 2,650 32,998 1,650 0.005 50 60 0.834 55,995 1,120 35,995 720 0.01 100 110 0.910 60,990 610 40,990 410 0.02 200 210 0.953 70,980 355 50,980 255 0.05 500 510 0.981 100,950 202 80,950 162

OCR for page 304
--> in five chance that a women with a repeatedly positive ELISA test is truly infected with HIV. Note that these rates apply to repeated ELISA testing only. When the original blood samples are subjected to Western blot confirmatory testing, most of the false positive results would test negative. Some fraction of Western blot results are indeterminate (depending on the testing procedure used and the laboratory), but some of these indicate an early-stage infection (Pins et al., 1997). Economic Evaluations of HIV Testing and Treatment in Pregnancy There have been numerous economic evaluations of HIV testing and treatment in pregnancy, each with different assumptions and different specific questions. Taken as a group, however, they generally establish the cost-effectiveness of prenatal HIV screening and treatment programs. Mauskopf and colleagues (1996) have estimated the economic impact of treating pregnant women who are HIV-positive with ZDV (zidovudine), and have found that such treatment is cost saving over a wide range of assumptions. They further find that voluntary prenatal HIV screening programs are cost saving if the prevalence exceeds 4.6 per 1,000 (under certain assumptions). Under the assumption that the prevalence rate is 1.7 per 1,000 (the national average), the cost per case avoided of a voluntary screening program with comprehensive counseling and 100% acceptance is $155,000. The same program with limited pre-test counseling is actually cost saving. In their base case analysis, assuming a prevalence rate of 1.7 per 1,000 in pregnant women, Gorsky and colleagues (1996) find that implementation of the Public Health Service (PHS) counseling and testing guidelines nationally would prevent 656 pediatric HIV infections annually and would result in a medical care cost saving of $105.6 million. Varying the maternal seroprevalence rate, they find that screening is cost saving as long as the prevalence rate is above 1.1 per 1,000. Myers and colleagues (1998) have determined the cost-effectiveness of mandatory versus voluntary prenatal HIV screening. They conclude that mandatory screening will prevent more cases of pediatric AIDS, but at a somewhat higher cost than voluntary screening. Under their base assumptions, including a maternal seroprevalence rate of 1.7 per 1,000, the cost per case averted was $255,000 for mandatory screening and $367,000 for voluntary screening. The incremental cost-effectiveness of mandatory compared with voluntary screening was $29,500. Cost-Effectiveness of Universal, Routine HIV Testing in Prenatal Care Two additional assumptions are needed for this calculation. First, the marginal cost of including an HIV test in the standard prenatal panel is $3 to $5. Costs of testing vary markedly according to the circumstances in which the

OCR for page 304
--> testing is done. ELISA tests done by private laboratories range from $15 to $65, but the cost to state laboratories is about $5. It costs the U.S. Army only $2.50 per serum specimen in its routine screening of all recruits because of the number tested and the established infrastructure for transporting specimens to the laboratory (all of these figures are from Mauskopf). In New York, the marginal cost of testing infant heel-stick samples for HIV is only about one dollar (Birkhead, 1998). Second, the follow-up cost for a repeatedly positive ELISA test (including the cost of the Western blot test and counseling those who are positive, but not treatment costs) is $100. Table K.1 also shows the marginal cost of prenatal testing (per 10,000 women in prenatal care) and the cost per true positive case found. The results show that in high-prevalence areas the cost per case found is extremely low—hundreds of dollars. Even in low-prevalence areas the cost exceeds $50,000 per case found only if the marginal cost per ELISA test is $5 and the prevalence is 1 per 10,000. In a more reasonable low-prevalence scenario ($3 dollars per test and a prevalence of 2 per 10,000), the cost per case found is only $15,600. While these numbers are not precise, they clearly indicate that universal routine HIV testing integrated into prenatal care can be very cost-effective, even in low-prevalence areas. Strategies To Reduce Perinatal HIV Transmission Inadequate prenatal care among women at high risk for HIV, health care providers' lack of adherence to PHS guidelines, and women's rejection of HIV testing and ZDV use all limit the ability to further reduce perinatal transmission. This section provides estimates of each potential barrier to HIV transmission reduction and presents a simplified model with which to assess the implications of different intervention strategies. If a hypothetical population of 7,000 HIV-infected pregnant women all obtained early prenatal care; if their providers were in complete compliance with PHS recommendations regarding counseling, testing, and ZDV treatment; and if women all accepted HIV tests and ZDV treatment, and all pregnancies resulted in a live birth—the committee estimates that 350 HIV-infected babies would be born (i.e., the risk of transmission under optimal care is 5%). If, however, the onset of prenatal care, provider behavior, or other factors affecting perinatal HIV transmission are not optimal, the number of HIV-infected babies increases. Table K.2 shows the effects of varying some of the factors affecting perinatal HIV transmission. Column 2 shows the committee's estimates of the current environment: an estimated 85% of HIV-positive women seek prenatal care, 75% of women are counseled regarding HIV testing, 80% of women accept the test, 90% of HIV-positive women are offered ZDV, and 90% of women accept and comply with ZDV treatment when it is offered. Given this scenario, 1,172 babies would be born to the hypothetical cohort of 7,000 HIV-infected women, a 235% in-

OCR for page 304
--> TABLE K.2 Alternative Prenatal Transmission Scenarios for 7,000 HIV-infected Pregnant Women—Change Current Environment Factors Affecting HIV Transmission Rates Currently Achievable (1) Estimate of Current Environment (2) Increase Prenatal Care Attendance (3) Increase Providers' Offering of Test (4) Increase Women's Test Acceptance (5) Increase Providers' Offering of ZDV (6) Increase Women's Acceptance of/Compliance with ZDV (7) Increase Provider's Offering of Test and ZDV (8) Increase Women's Acceptance of Test/ZDV (9) Women with prenatal care (%) 100.00 85.00 100.00 85.00 85.00 85.00 85.00 85.00 85.00 Women counseled (%) 100.00 75.00 75.00 100.00 75.00 75.00 75.00 100.00 75.00 Women accepting HIV test (%) 100.00 80.00 80.00 80.00 100.00 80.00 80.00 80.00 100.00 Women offered ZDV treatment (%) 100.00 90.00 90.00 90.00 90.00 100.00 90.00 100.00 90.00 Women accepting/ complying with ZDV treatment (%) 100.00 90.00 90.00 90.00 90.00 90.00 100.00 90.00 100.00 Babies exposed to low transmission rate (.05) (%) 100.00 41.31 48.60 55.08 51.64 45.90 45.90 61.20 57.38 Babies exposed to high transmission rate (.25) (%) 0.00 58.69 51.40 44.92 48.36 54.10 54.10 38.80 42.63 Expected number HIV-infected babies 350.00 1,071.66 1,069.60 978.88 1,027.08 1,107.40 1,107.40 893.20 946.75 Reduction in number HIV-infected babies from current scenario (%) 70.13 NA 8.71 16.45 12.34 5.48 5.48 23.77 19.20 Increase in number of HIV-infected babies from achievable scenario (%) NA 234.76 205.60 179.68 193.45 216.40 216.40 155.20 170.50 NOTE: Model assumes no fetal loss and two perinatal HIV transmission rates (.25 and .05); NA = not applicable. crease over the currently achievable state (i.e., from 350 to 1,172 HIV-infected babies).1 If we hold all but one condition constant, and change one parameter at a time, the impact of changes in the current environment can be assessed. 1   The model assumes only two HIV transmission rates, .25 if women are not treated and .05 if they are treated. These transmission rates actually vary according to the HIV-infected woman's clinical state, and the onset and completeness of treatment. The model also assumes that testing rates for HIV-positive women are similar to those observed in the general population of pregnant women.

OCR for page 304
--> Increase Women's Test Acceptance (5) Increase Providers' Offering of ZDV (6) Increase Women's Acceptance of/Compliance with ZDV (7) Increase Providers' Offering of Test and ZDV (8) Increase Women's Acceptance of Test/ZDV (9) 85.00 85.00 85.00 85.00 85.00 75.00 75.00 75.00 100.00 75.00 100.00 80.00 80.00 80.00 100.00 90.00 100.00 90.00 100.00 90.00 90.00 90.00 100.00 90.00 100.00 51.64 45.90 45.90 61.20 57.38 48.36 54.10 54.10 38.80 42.63 1,027.08 1,107.40 1,107.40 893.20 946.75 12.34 5.48 5.48 23.77 19.20 193.45 216.40 216.40 155.20 170.50 Increasing the receipt of prenatal care from 85% to 100% reduces the number of HIV-infected babies by 9% (i.e., from 1,172 to 1,070) (column 3). Increasing the rate at which providers offer HIV tests from 75% to 100% reduces the number of HIV-infected babies by 16% (i.e., from 1,172 to 979) (column 4). Increasing women's acceptance of HIV tests from 80% to 100% reduces the number of HIV-infected babies by 12% (i.e., from 1,172 to 1,027) (column 5). Increasing the providers offering ZDV treatment from 90% to 100% reduces the number of HIV-infected babies by 5% (from 1,172 to 1,107) (column 6).

OCR for page 304
--> TABLE K.3 Alternative Prenatal Transmission Scenarios for 7,000 HIV-Positive Pregnant Women—Change Current Environment Factors Affecting HIV Transmission Rates Currently Achievable (1) Estimate of Current Environment (2) Increase All but Prenatal Care Attendance (3) Close Gap Between Current and Achievable by 10% (4) Close Gap Between Current and Achievable by 20% (5) Close Gap Between Current and Achievable by 30% (6) Close Gap Between Current and Achievable by 40% (7) Close Gap Between Current and Achievable by 50% (8) Close Gap Between Current and Achievable by by 78% (9) Women with prenatal care (%) 100.00 85.00 85.00 86.50 88.00 89.50 91.00 92.50 96.70 Women counseled (%) 100.00 75.00 100.00 77.50 80.00 82.50 85.00 87.50 94.50 Women accepting HIV test (%) 100.00 80.00 100.00 82.00 84.00 86.00 88.00 90.00 95.60 Women offered ZDV treatment (%) 100.00 90.00 100.00 91.00 92.00 93.00 94.00 95.00 97.80 Women accepting/complying with ZDV treatment (%) 100.00 90.00 100.00 91.00 92.00 93.00 94.00 95.00 97.80 Babies exposed to low transmission rate (.05) (%) 100.00 41.31 85.00 45.52 50.05 54.92 60.14 65.74 83.56 Babies exposed to high transmission rate (.25) (%) 0.00 58.69 15.00 54.48 49.95 45.08 39.86 34.26 16.44 Expected number HIV-infected babies 350.00 1,171.66 560.00 1,112.70 1,049.26 981.10 907.97 829.62 580.17 Reduction in number HIV-infected babies from current scenario (%) 70.13 NA 52.20 5.03 10.45 16.26 22.51 29.19 50.48 Increase in number of HIV-infected babies from achievable scenario (%) NA 234.76 60.00 217.91 199.79 180.31 159.42 137.03 65.76 NOTE: Model assumes no fetal loss and two perinatal HIV transmission rates (.25 and .05); NA = not applicable. Increasing women's acceptance of ZDV treatment from 90% to 100% reduces the number of HIV-infected babies by 5% (i.e., from 1,172 to 1,107) (column 7). Given the current environment, the most effective single intervention to reduce perinatal transmission is to increase the number of providers offering HIV tests (reduces perinatal HIV transmission by 16%). If providers were in complete compliance with the PHS guidelines (i.e., they offered HIV tests and ZDV treatment to all women), there would be a 24% decrease in the number of HIV-infected babies (from 1,172 to 893) (column 8). Alternatively, if the current environment remained the same, but all HIV-infected women accepted HIV testing

OCR for page 304
--> Closing Gap Between Current and Achievable by 20% (5) Closing Gap Between Current and Achievable by 30% (6) Closing Gap Between Current and Achievable by 40% (7) Closing Gap Between Current and Achievable by 50% (8) Closing Gap Between Current and Achievable by 78% (9) 88.00 89.50 91.00 92.50 96.70 80.00 82.50 85.00 87.50 94.50 84.00 86.00 88.00 90.00 95.60 92.00 93.00 94.00 95.00 97.80 92.00 93.00 94.00 95.00 97.80 50.05 54.92 60.14 65.74 83.56 49.95 45.08 39.86 34.26 16.44 1,049.26 981.10 907.97 829.62 580.17 10.45 16.26 22.51 29.19 50.48 199.79 180.31 159.42 137.03 65.76 when offered, and accepted and complied with ZDV treatment, there would be a 19% reduction in the number of HIV-infected babies (i.e., from 1,172 to 947) (column 9). If both providers and HIV-infected women had optimal rates (i.e., if all but prenatal care is set to 100%), there would be a 52% decline in the number of HIV-infected babies (i.e., from 1,172 to 560) (Table K.3, column 3). This simplified model illustrates the need for multifaceted approaches to significantly reduce perinatal HIV transmission. But even with a multifaceted approach, significant further reductions in the number of HIV-infected babies will be difficult to achieve. Table K.3 shows the effects of closing the gap between current and optimal rates by 10% to 50% (columns 4 through 8). Even if the gap was reduced by 50% (e.g., prenatal care increases from 85% to 92.5%),

OCR for page 304
--> there would be only a 29% decline in the number of HIV-infected babies (i.e., from 1,172 to 830). Here it is assumed that 92.5% of HIV-infected pregnant women obtained early prenatal care, 87.5% of women were offered HIV testing, 90% of women accepted testing, 95% of HIV-positive women were offered ZDV, and 95% of women accepted and complied with ZDV therapy. To achieve a further 50% decline in the number of HIV-infected babies (i.e., from 1,172 to 580 infected babies) and be within reach of the currently achievable state (i.e., 350 infected babies), the gap between observed and achievable rates would have to close by 78%, and rates for factors related to transmission would have to be very high (e.g., 96.7% of women with prenatal care) (column 9). References Birkhead GS. New York State AIDS Institute. Personal communication. 1998. Gorsky RD, Farnham PG, Straus WL, Caldwell B, Holtgrave DR, Simonds RJ, Rogers MF, Guinan ME. Preventing perinatal transmission of HIV: Costs and effectiveness of a recommended intervention. Public Health Rep 111(4):335–341, 1996. Mauskopf JA, Paul JE, Wichman DS, White AD, Tilson HH. Economic impact of treatment of HIV-positive pregnant women and their newborns with zidovudine. Implications for HIV screening. JAMA 276(2):132–138, 1996. Myers ER, Thompson JW, Simpson K. Cost–effectiveness of mandatory compared with voluntary screening for human immunodeficiency virus in pregnancy. Obstet Gynecol 91(2):164–181, 1998. Pins MR, Teruya J, Stowell CP. Human immunodeficiency virus testing and case detection: pragmatic and technical issues. In: Cotton D and Watts DH, eds. The Medical Management of AIDS in Women. Wiley-Liss: New York City, 1996.