Antenatal HIV testing is undertaken primarily to offer interventions that will reduce the risk for HIV transmission from mother to child. However, diagnosis of HIV during pregnancy also has implications for the mother. In a recent analysis of over 8000 patients with AIDS from 11 European countries including the United Kingdom, 21% of HIV-infected persons remained undiagnosed until they develop AIDS . Any economic analysis of antenatal screening strategies must, therefore, take into account both the potential health gains and the costs occasioned by earlier diagnosis of the woman‚s HIV infection . As part of a study of the cost- effectiveness of antenatal HIV testing in the United Kingdom, we have estimated the health service costs and benefits of earlier HIV diagnosis in women through antenatal testing. This is achieved by applying cost and efficacy estimates of antiretroviral therapy (ART) from clinical trials to a model of HIV disease progression.
Disease progression model
Disease progression is conceptualized as a series of states characterized by the level of the CD4 cell count (Fig. 1). Each woman enters the flow diagram at a point determined by her CD4 cell count at the time of her HIV diagnosis and progresses at a rate determined by the rate of decline of her CD4 cell count. The states are defined in terms of CD4 cell count ranges (>350, 200-350, 60-200 and <60¥106cells/l) and death. Both the woman‚s initial CD4 cell count, at the time when she would participate in antenatal testing, and her rate of decline are selected at random from statistical distributions. At every stage of HIV disease, women are also assumed to be subject to mortality from other causes. The time spent in each CD4 cell count state, the time to AIDS, the time to death and the health service costs incurred in each state were calculated.
The model was applied to two matched simulated cohorts of women, with identical initial CD4 cell counts and decline rates. In women not diagnosed until later (the untested cohort), each woman experiences disease progression according to her rate of decline in CD4 cell count until it has fallen by a fixed amount D, at which point it is assumed her HIV infection is diagnosed. It is not until then that she occasions any health care costs. In the antenatally diagnosed women (tested cohort), each woman occasions health care costs immediately. Different strategies are explored as to when ART might be introduced; the effect of ART is conceptualized as decreasing the rate of decline in CD4 cell count by a fixed amount from the moment when treatment begins until death. The tested woman is treated earlier and will, therefore, incur higher health service costs, but the assumption is that she will live longer and have more AIDS-free years of life.
Disease progression parameters and their ranges
The median CD4 cell count in tested women was assumed to be 360¥106cells/l. This was based on the CD4 cell count of 86 women diagnosed antenatally (at an average of 25 weeks of gestation) between 1990 and 1997 in the National Study of HIV in Pregnancy [3,4]. The square root of CD4 count is assumed to be normally distributed with a 95th percentile range of 43 to 984 cells. This distribution was truncated at 60¥106cells/l, which is the average CD4 cell count at AIDS onset in women in the United Kingdom [Communicable Disease Surveillance Centre (CDCC), personal communication], on the basis that a woman with AIDS is already diagnosed and, therefore, not eligible for antenatal HIV screening. The median baseline rate of decline was taken to be 47¥106cells/l per year in the absence of ART, based on data up until May 1998 from 795 asymptomatic adults enrolled into the placebo arm of the Concorde trial  (A. Babiker, personal communication). The 5th and 95th centile rates of decline in the Concorde data were 199 and -29¥106cells/l per year, respectively, but it was assumed that measurement error and other short-term variation contributed substantially to the observed variation and the rate of decline in the model was represented by two half-normal distributions with median 47¥106cells/l per year, and 95 percentiles at 137 and -5¥106cells/l per year. Women with no decline in CD4 cell counts are assumed not to progress at all.
AIDS was assumed to occur at a CD4 cell count of 60¥106cells/l and death at a notional CD4 cell count of -22.3¥106cells/l. Therefore, time to a CD4 cell count of 60¥106cells/l is taken as an estimate of ‚time to AIDS‚, and time to -22.3 as an estimate of ‚time to death‚. The latter value was selected as in the model it would lead to death a median 21 months after AIDS onset. This is compatible with UK data on survival after AIDS onset in over 7000 patients before the era of highly active antiretroviral therapy (HAART) [6,7]. Mortality from other causes was modelled as the age-specific all-cause mortality rate , and for simplicity it was assumed that all women were aged 28.5 years at the time of antenatal attendance, as this was the average age of infected women giving birth in the United Kingdom.
A base-case hazard ratio of 0.60 was assumed for the effect of ART on disease progression. This was applied to the decline of the CD4 cell count (for example lowering a rate of decline in CD4 cell counts of 47 to 28¥106cells/l per year) with a range of 0.40 to 0.80. This was intended to represent the effects of triple ART, costing the equivalent of two nucleoside analogue reverse transcriptase inhibitor (NRTI) drugs and one protease inhibitor (PI) drug used from the time of treatment onset until death. The hazard ratio is higher than observed in trials of triple ART in the short term [9-12] but lower than observed in trials of two NRTI versus monotherapy . Two ART regimen were considered: women could either be treated at diagnosis or when their CD4 cell count reached 350¥106cells/l as recommended in recently published UK guidelines .
The model was run for 999 women tested antenatally, with initial CD4 cell counts and initial rates of decline drawn from the distributions described above, and for 999 ‚matched‚ untested women with the same initial CD4 cell counts and decline rates. The median CD4 cell count for all women in England and Wales at the time of HIV diagnosis (only a small minority of whom would have been diagnosed at antenatal screening) is 280¥106cells/l (CDSC, personal communication). It was assumed, therefore, that the untested woman is diagnosed at a time when her CD4 cell count is 80¥106cells/l lower (i.e., 360-280¥106cells/l) than the tested woman, or at AIDS onset (60¥106cells/l), whichever is first. The untested woman, therefore, enters the flow chart at a point that is 80 cells further on than her tested ‚twin‚, 80/r years later, where r is the initial rate of decline in CD4 cell counts of each pair.
As the presumed time between antenatal HIV diagnosis and the diagnosis of the untested women is a critical factor, sensitivity analyses were run in which the difference in CD4 cell count between the two was 40 or 200¥106cells/l, rather than 80¥106cells/l. Given the CD4 cell count rate of decline of 47¥106cells/l described above, a difference of 80¥106cells/l represents a 20.4-month delay in diagnosis for the untested woman, while differences of 40 and 200¥106cells/l would represent delays of 10 months and 51 months, respectively.
Health service costs per year spent in each CD4 cell count ‚state‚ were taken from Chancellor et al . This study included the in-patient care, laboratory tests, day-care and community health care costs of 389 HIV-infected adults treated in a London centre in 1995. Excluding costs of ART, the annual costs (updated to 1996-1997 prices) were £2793, £2793 and £3093 for patients with CD4 cell counts >350, 200-350 and <200¥106cells/l but prior to AIDS onset, respectively, and £9127 for patients with AIDS. Costs of ART were based on the average cost of two NRTI and one PI drug, at British National Pharmacopoeia 1996 prices , which produces an annual antiretroviral drug cost of £9596 per patient on ART. Both costs and life-years were discounted in accordance with standard UK practice : costs at 6%  and life-years at 2% per year .
The differences in health service costs and life expectancy between tested and untested women were expressed as an incremental ratio (IR) where IR is given by (costs for tested minus costs for untested women)/(life-years for tested minus life-years for untested women).
Formal specification of the disease progression model
Initial CD4 cell count (¥106cells/l) x at age 29 is given by ÷x ≊ N(÷360,6.3252) if x >÷60.
CD4 cell count of the unscreened women diagnosed at age (29 +(D/r)) is max(x-D,60); D=80with an sensitivity analysis range of 40-200.
Underlying annual rate of decline in CD4cell count (¥106cells/l per year): r≊N(47,272) if r <47 and r≊N(47,462) if r £47.
Antiretroviral therapy hazard ratio (t)=0.60, sensitivity analysis range 0.40 to 0.8.
Non-HIV-related mortality rate, m, is estimated from the England and Wales age-specific female all cause mortality rate .
In the base-case scenario, it is assumed that untested women are diagnosed at a CD4 cell count 80¥106cells/l lower than tested women, that the ART efficacy hazard ratio is 0.60 and that therapy is initiated at a CD4 cell count of 350¥106cells/l. The median time from diagnosis to AIDS calculated from the model is 10.0 years in tested women and 9.3 years in untested (Table 1), compared with an estimated median of 6.8 years without any ART. The estimated median survival times from antenatal attendance to death are shown in Fig. 2: 12.7 years in tested and 12 years in untested women compared with an estimated 8.3 years without ART (Table 1). It is evident that tested and untested cohorts differ only marginally in time to AIDS and time to death. It is useful to compare survival for both tested and untested women receiving ART with the two extreme situations represented by untreated infected women, who live for a median additional 8.3 years after antenatal attendance, and by uninfected women, who are expected to live a median 53.3 additional years (Table 1). Discounted average time to death (life expectancy) is also shown in Table 1.
Additional costs per life-years saved by screening
The discounted costs incurred under the above conditions are £102811 ($146873) for tested and £79494 ($118648) for untested women, a difference of £23317 ($34801). This can be compared with the discounted life expectancy after antenatal attendance: 13.96 years in tested and 13.5 years in untested women, a difference of 0.46 years. The ratio of these differences, £51258 ($76504), represents the additional costs occasioned by antenatal diagnosis per additional life-year gained.
The difference in life expectancy is widened if treatment is initiated at diagnosis rather than at a CD4 cell count of 350¥106cells/l. Holding treatment efficacy constant, this would extend additional life expectancy to 14.75 years in tested and 14.02 years in untested women, a difference of 0.73 years (Table 1). However, costs incurred by tested women would also be increased by a longer period of treatment to £132231 ($197360). Costs in untested women would increase only marginally because the earlier treatment is not available to them until they are diagnosed. As a result, the additional cost per life-year gained remains virtually unchanged at £52622 ($78540).
The difference between tested and untested women in AIDS-free survival and in life expectancy is also widened by more effective ART. If the hazard ratio is 0.4, this difference is 0.835 years with treatment started at a CD4 cell count of 350¥106cells/l, and 1.26 years with immediate treatment at the time of diagnosis. Similarly, if a hazard ratio of 0.8 is assumed, it is narrowed to 0.19 years with treatment started at 350¥106cells/l, and 0.22 years with immediate treatment. As treatment efficacy has only minor effects on costs, increasing treatment efficacy will result in greater cost-effectiveness of earlier identification through antenatal screening from the perspective of the mother‚s costs and life-years gained.
Sensitivity to difference in time to diagnosis between tested and untested
All the above results are based on a difference of 80¥106cells/l between tested and untested women at the time of their diagnosis, representing a 20.4-month delay. The longer the delay, the greater the total cost difference between tested and untested women, and the greater the potential benefits of testing. The base-case costs of £51258 ($76504) per additional life-year increase to £66340 ($99015) and decrease to £39208 ($58519) by changing the difference in CD4 cell count between tested and untested women from 80 to 40 and 200¥106cells/l, respectively (Table 2).
The highest incremental ratio, £143757 ($214563) per life-year gained, occurs with less-effective ART (hazard ratio 0.80), the least difference between tested and untested women at time of diagnosis (40¥106cells/l) and ART initiated at 350¥106cells/l. The lowest incremental ratio, £24628 ($36758) per life-year saved, is generated by immediate ART, a hazard ratio for efficacy of 0.40, and a CD4 cell count difference of 200¥106cells/l between tested and untested women (Table 2). The estimated survival time from antenatal attendance to death for tested and untested groups under these two extreme scenarios are presented in Fig. 2 and can be compared with the base case.
Additional costs per life-year saved by early treatment
It is also of interest to assess the costs and effects of earlier versus later treatment, per se, generated by the model (Table 3). Compared with treatment at a CD4 cell count of 350¥106cells/l, treatment at diagnosis (depending on the lifetime efficacy of ART) in the tested and untested cohorts implies an additional £24945 to £75167 and £19217 to £56745 per life-year gained, respectively,
In this study, we have developed a model of HIV disease progression and applied it to data from the United Kingdom so that the costs and benefits of earlier HIV diagnosis for the woman as a result of antenatal HIV testing can be quantified and included in an overall cost-effectiveness analysis of antenatal HIV testing. Most previous studies evaluating the economics of antenatal HIV testing from developing or developed world perspectives have not included future costs and benefits for the woman in their models of the cost-effectiveness of testing [20-26], although it should be recognized that in developing countries women may not have access to ART after they have delivered their babies. An attempt to compare the costs and benefits for tested and untested HIV-infected women has not previously been undertaken. Ecker included a fixed cost associated with earlier detection of HIV in the mother, but no additional benefit . Mauskopf et al. included maternal costs only in the sensitivity analysis and concluded that cost-effectiveness of antenatal HIV screening was highly sensitive to these additional costs but again included no benefits in terms of life-years gained .
The importance of including the additional downstream costs and benefits for the woman resulting from an earlier HIV diagnosis in any economic analysis of antenatal HIV-screening strategies has increased recently with the advent of more efficacious but costly combination therapies. The short-term cost-effectiveness of HAART has been evaluated in several studies in the United States [27-29]. Because of short-term savings on hospital inpatient care costs, relatively low incremental cost-effectiveness ratios were demonstrated relative to one- or two-drug ART (US$30000 or less per life-year gained). However, the longer-term costs and benefits of HAART and of commencing HAART earlier in the course of HIV disease remain unknown and have not been addressed.
The validity of our estimates of the cost per life-year gained of diagnosing women antenatally depends on how well the disease progression model and estimates of the efficacy of ART capture the costs and effects of long-term ART. Disease progression in untreated women in our model was based on the rate and ranges of decline in CD4 cell counts observed during long-term follow-up of individuals randomized to the placebo group in the Concorde trial , similar to natural history studies . While this is imperfect and does not include a parameter for HIV RNA viral load, which predicts disease progression independently of the CD4 cell count [30,31], our model-derived estimates of the median time to AIDS and to death from a CD4 count of 360¥106cells/l are comparable with data from cohort studies where little or no ART was used [30,32,33]. In a retrospective cohort of 1056 Africans and 992 non-Africans with HIV infection in London between 1982 and 1995, with CD4 cell counts of 238 and 371¥106cells/l, respectively, at baseline, median time to death was similar in both groups (82 and 78 months, respectively) and similar to our model . Data from the UK Womens Cohort on 84 women followed from CD4 cell counts between 300 and 400¥106cells/l, of whom only 25% received any ART (mainly zidovudine monotherapy), showed a 75% survival at 29.8 months (A. Copas, personal communication), a result also not dissimilar to survival in untreated women in our model.
In our base-case model, time to AIDS is extended by 3.2 years and time to death by 4.4 years in tested women receiving triple ART after reaching a CD4 cell count of 350¥106cells/l. For simplicity, we have assumed that the effect of ART is continuous from diagnosis until death and have chosen a relatively modest effect (reduction in the rate of decline in CD4 cell count by 40%) in the base-case scenario. Reductions in progression to AIDS of at least 60-80% (hazard ratios 0.4-0.2) have been reported from studies of triple ART [9-12] but these have been based on relatively short-term follow-up of 1-2 years, and there are no data of the overall long-term efficacy of HAART. Increasing the efficacy by modelling a 60% reduction in the rate of decline in CD4 cell count has the effect of extending AIDS-free survival by 7.3 years and overall survival by 9.6 years compared with untreated women. It is unlikely that the assumption of life-long constant effects of ART, as opposed to short-term temporary effects, would have a major effect on the cost and benefit differences between screening strategies as long as predicted life-expectancy lay between the limits of our maximum and minimum ART efficacy scenarios. Application of maximum efficacy (reduction in the rate of decline in CD4 cell counts by 60% and maximum delay between diagnosis in tested and untested women) generates the lowest incremental ratio with additional costs of early diagnosis of around £25000 ($37313) per life-year gained.
In reality, it is likely that the greatest effect of ART might be at the initiation of therapy, and that efficacy will also be different depending on the stage of disease at which ART is initiated. If both these occurred in favour of earlier therapy, they would tend to increase the cost-effectiveness of an antenatal HIV diagnosis from the woman‚s perspective but would be unlikely to have a major effect on overall cost-effectiveness.
The model could be made more realistic by inclusion of maternal age distribution, age-related disease progression rates and random distributions of CD4 cell count at diagnosis and AIDS; however, preliminary explorations suggest that incorporation of these parameters would have little effect on the expected differences between tested and untested cohorts, particularly in comparison with the uncertainty in long-term ART efficacy. In addition, temporary changes in CD4 cell counts have been reported to occur during pregnancy [35-37]. However, calculations suggest that this would have only a minor effect on the additional cost per life-year gained, whether treatment is assumed to start at 350¥106cells/l or at diagnosis.
Data on stage of HIV disease and CD4 cell count at HIV diagnosis among women show that in the past a substantial proportion (approximately 25%) of women did not present until the onset of symptoms or AIDS [1,38]. Although there are no data comparing early with late commencement of HAART, there is evidence from trials of dual NRTI therapy that delaying ART until AIDS onset may be associated with a poorer response to treatment and would not be currently recommended.
It is of interest to compare the costs and benefits for the mother of an antenatal HIV diagnosis with those relating to the child. Assuming that vertical HIV transmission can be reduced from 25 to 5% with screening and interventions and that the discounted life-expectancy for an uninfected child is 40 years, for infected children with HIV status known from birth is 11.6 years and for infected children with HIV status not known from birth is 10.7 years , then knowledge of the mother‚s infection status can generate an additional 5.9 paediatric life-years [(0.2¥40)+ (0.05¥11.6)- (0.25¥0.7)]. Assuming that the infected child accrues life-time costs of £73855 ($110231) if HIV status is known from birth and £59004 ($88506) if not , this comes with a saving of £11058 ($16504) [(0.25¥£59004)- (0.05¥£73855)], mainly resulting from the large number of life-years gained from the group of uninfected children [39,40]. This contrasts with the additional cost of £23318 ($34803) to give an additional 0.46 life-years for the mother. It is interesting that inclusion of maternal costs and life-years gained in economic analyses of antenatal screening will have a major effect on downstream costs of screening while contributing only a little to the life-years gained. Putting the paediatric and maternal components together, we obtain an additional downstream cost of £12260 ($18390) for a gain of 6.36 life-years. Setting aside other benefits of maternal diagnosis, such as the potential to reduce the risk of transmission to sexual partners, if we assume decision-makers would pay £10000 ($15000) per additional life-year, then they should be willing to pay [(6.36¥£10000)- £12260], or just over £50000 ($75000), to diagnose a single maternal infection. From this it can be deduced, for instance, that antenatal HIV testing might be considered cost-effective in a population with HIV prevalence 1:10000, as long as the cost of testing and pre-test information is less than £5.
In conclusion, we have developed a model of maternal HIV disease progression and applied it to UK data in order to estimate the cost-effectiveness of antenatal HIV diagnosis from the woman‚s perspective. The estimated costs per additional life-year gained for the mother of antenatal HIV testing are relatively high and there are many uncertainties in the model parameters. However, it is important that in the era of HAART, these are included in any economic analysis of antenatal HIV screening.
We thank the following for providing data: Abdel Babiker (MRC HIV Clinical Trials Unit) for providing data on CD4 cell counts from the Concorde Trial, Dr Angus Nicoll and R. Gilbert at CDSC, and Mr A. Copas (Department of Sexually Transmitted Diseases, University College London Medical School, London). We would like to thank Mr A. Copas and Dr D. Mercey (Department of Sexually Transmitted Diseases, University College London Medical School, London) and Dr K Porter (MRC HIV Clinical Trials Unit) for their comments on the paper. We also thank C. Davies for help with preparation of the manuscript.
1. Hamers FF, Delmas MC, Alix J. Unawareness of HIV seropositivity before AIDS diagnosis in Europe
. XII International Conference on AIDS.
Geneva, June1998 [abstract 43105].
2. Briggs A, Gibb D, Sculpher M, MacDonagh SE, Ades AE. Antenatal testing for HIV infection. Developing a model for economic evaluation.
HERGDiscussion Paper 14: June, 1996 (ISBN 18721 66 679).
3. Ades AE, Davison CF, Holland FJ, et al. Vertically transmitted HIV infection in the British Isles
4. Nicoll A, McGarrigle C, Brady T, et al. Epidemiology and detection of HIV-1 among pregnant women in the United Kingdom: results from national surveillance 1988-1996
5. Concorde Coordinating Committee. Concorde: MRC/ANRS randomised double-blind controlled trial of immediate and deferred zidovudine in symptom-free HIV infection
6. Mocroft A, Youle M, Morcinek J, et al. Survival after diagnosis of AIDS: a prospective observational study of 2625 patients. Royal Free/Chelsea and Westminster Hospital‚s Collaborative Group.BMJ
7. Rogers PA, Whitmore Overton SE, et al.Survival of adults with AIDS in the United Kingdom.Commun Dis Rep CDR Rev
8. Office of Population Census and Surveys. 1992 Mortality Statistics - General, England and Wales. Series DH1 No 27.
9. Palella FJ, Jr, Delaney KM, Moorman AC, et al. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. HIV Outpatient Study Investigators.N Engl J Med
10. Mouton Y, Alfandari S, Valette M, et al. Impact of protease inhibitors on AIDS-defining events and hospitalizations in 10 French AIDS reference centres. Federation National des Centres de Lutte contre le SIDA.AIDS
11. Cameron DW, Heath Chiozzi M, Danner S, et al. Randomised placebo-controlled trial of ritonavir in advanced HIV-1 disease. The Advanced HIV Disease Ritonavir Study Group
12. Egger M, Hirschel B, Francioli P, et al. Impact of new antiretroviral combination therapies in HIV infected patients in Switzerland: prospective multicentre study. Swiss HIV Cohort Study.BMJ
13. Delta: a randomised double-blind controlled trial comparing combinations of zidovudine plus didanosine or zalcitabine with zidovudine alone in HIV-infected individuals. Delta Coordinating Committee.Lancet
14. Gazzard B, Moyle G. 1998 revision to the British HIV Association guidelines for antiretroviral treatment of HIV seropositive individuals
15. Chancellor JV, Hill AM, Sabin CA, Simpson KN, Youle M. Modelling the cost effectiveness of lamivudine/zidovudine combination therapy in HIV infection.PharmacoEconomics
16. British Medical Association, Royal Pharmaceutical Society. British National Formulary,
No. 31. London: BMA; 1996.
17. Drummond MF, O‚Brien BJ, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes.
New York: Oxford University Press; 1997.
18. HM Treasury. Economic Appraisal in Central Government: a Technical Guide for Government Departments.
London; HMSO; 1991.
19. Department of Health. Policy Appraisal and Health.
London: Department of Health; 1995.
20. 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
21. Remis RS, Vandal AC. Cost-effectiveness of universal and selective HIV screening of pregnant women in Quebec.Fifth Annual Conference on HIV/AIDS Research in Canada.
Winnipeg, June 1995 [abstract 338].
22. Ecker JL. The cost-effectiveness of human immunodeficiency virus screening in pregnancy.Am J Obstet Gynecol
23. Le Gales C, Moatti JP. Cost-effectiveness of HIV screening of pregnant women in hospitals of the Paris area. The Paris
-Tours study group of antenatal transmission of HIV, Group ‚9 maternites‚.Eur J Obstet Gynecol Reprod Biol
24. Wilkinson D, Floyd K, Gilks CF. Antiretroviral drugs as a public health intervention for pregnant HIV-infected women in rural South Africa: an issue of cost-effectiveness and capacity.AIDS
25. Marseille E, Kahn JG, Saba J. Cost-effectiveness of antiretroviral drug therapy to reduce mother-to-child HIV transmission in sub-Saharan Africa.AIDS
26. Newell ML, Dabis F, Tolley K, Whynes D. Cost-effectiveness and cost-benefit in the prevention of mother-to-child transmission of HIV in developing countries
27. Rahman A, Deyton LR, Goetz MB, Rimland D, Simberkoff MS. Inversion of inpatient/outpatient HIV service utilization: impact of improved therapies, clinician education and case management in the US Department of Veteran Affairs
. XII International Conference on AIDS.
Geneva, June1998 [abstract 42429].
28. Anis A. The cost effectiveness of successive antiretroviral therapeutic drug regimes for the treatment of HIV/AIDS
. XI International Conference on AIDS.
Vancouver, July 1996 [abstract 24377].
29. Bartlett JG, Chaisson R, Kerruly J, Moore R. HAART: one of the most cost-effective therapies of the past decade.XI International Conference on AIDS.
Vancouver, July 1996 [abstract 24380].
30. Mellors JW, Munoz A, Giorgi JV, et al. Plasma viral load and CD4 lymphocytes as prognostic markers of HIV-1 infection. Ann Intern Med
31. Feinberg MB, Carpenter C, Fauci AS, et al. Guidelines for the use of antiretroviral agents in HIV-infected adults and adolescents.Ann Intern Med
32. Oddone EZ, Cowper P, Hamilton JD, et al. Cost effectiveness analysis of early zidovudine treatment of HIV infected patients.BMJ
33. Joint Concorde and Opal Coordinating Committee. Long-term follow-up of randomized trials of immediate versus deferred zidovudine in symptom-free HIV infection
34. Del Amo J, Petruckevitch A, Phillips A, et al. Disease progression and survival in HIV-1 infected Africans in London. AIDS
35. European Collaborative Study. Immunological markers in HIV-infected pregnant women. The European Collaborative Study and the Swiss HIV Pregnancy Cohort
36. Tuomala RE, Kalish LA, Zorilla C, et al. Changes in total, CD4+, and CD8+ lymphocytes during pregnancy and 1 year postpartum in human immunodeficiency virus-infected women. The Women and Infants Transmission Study.Obstet Gynecol
37. Brettle RP, Raab GM, Ross A, et al. HIV infection in women: immunological markers and the influence of pregnancy. AIDS
38. Porter K, Wall PG, Evans BG. Factors associated with lack of awareness of HIV infection before diagnosis of AIDS.BMJ
39. Sculpher MJ, Gibb D, Ades AE, Ratcliffe J, Duong T. Modelling the costs of paediatric HIV infection and AIDS: comparison of infected children born to screened and unscreened mothers.AIDS
40. Ratcliffe J, Ades AE, Gibb D, Sculpher MJ, Briggs AH. Prevention of mother-to-child transmission of HIV-1 infection: alternative strategies and their cost-effectiveness.AIDS
Keywords:© 1999 Lippincott Williams & Wilkins, Inc.
antenatal testing; maternal HIV; treatment costs; treatment benefits; modelling; health service