Letter Concerning Unethical Placebo-Controlled Studies on HIV-Positive Pregnant Women in Developing Countries
October 23, 1997
Donna Shalala, Secretary
Department of Health and Human Services
200 Independence Avenue SW
Washington, DC 20201
Dear Secretary Shalala:
In the past three weeks we have obtained new information concerning the unethical studies on HIV-positive pregnant women in developing countries being funded by the National Institutes of Health (NIH) and the Centers for Disease Control and Prevention (CDC) which strengthens the case against your ill-advised decision to allow these studies to continue with placebo-controlled arms. We have learned that in one of the placebo-controlled NIH-funded studies, planned for Ethiopia, the researchers have tentatively decided to eliminate one arm of the study in which 300 women were to be given a placebo. In addition, we have obtained detailed, previously unpublished data from the original study which established the efficacy of AZT in reducing maternal-infant HIV transmission (ACTG 076) indicating that in early 1994 it was known that a shorter course of AZT therapy might work as well as longer treatment and that the shorter course was almost certain to be more efficacious than a placebo. On the basis of this new information, we demand that you immediately order the researchers to stop any arm of their studies in which women are denied access to antiretroviral drugs and to provide at least short-term AZT for all women now getting a placebo or other unproven treatments.
Eliminating the Placebo Group in the NIH-funded Ethiopia Study
One of the seven NIH-funded studies which we found to be unethical (a study in the Domnican Republic has been cancelled for other reasons) is a study in Ethiopia involving researchers from that country and the Johns Hopkins University School of Public Health.1 As originally designed, approximately 900 HIV-positive pregnant women were to be randomly assigned to one of three groups (about 300 in each group). In the first group, women were to receive AZT for six to nine weeks during the last part of pregnancy and during labor and delivery, and the newborn infant was to be treated with AZT. Women in the second group were to receive the same AZT treatment during the latter part of pregnancy and labor and delivery, but the infants were not to be given AZT. In the third group, no treatment (a placebo) was to be given to the mothers and the infants. We have learned from four independent sources at Johns Hopkins that in the past several weeks a decision was made by the Johns Hopkins researchers to drop the third (placebo-only) group and to conduct the experiment as a comparison between groups one and two to see if they are equivalent in reducing the rate of infections in the infants.
This restructured (and now ethical) version of the Ethiopian study is similar to the design in another NIH-funded study which is ethical, the study in Thailand involving Dr. Marc Lallemant of Harvard University School of Public Health. In that study, treatments of different prepartum (4 or 12 weeks) or post-partum duration (6 weeks or 3 days for the infants) are being compared in a four-armed trial to determine whether their effectiveness will be approximately equivalent. No women in that study get a placebo and the Chair of the Harvard School of Public Health's ethics review board refused to allow the researchers to be swayed by NIH efforts to change the design to a placebo-controlled study, a design he described as "unethical."2
The logic of the Johns Hopkins researchers in abandoning the use of a placebo for the 300 Ethiopian women assigned to that group and providing some AZT treatment for all women in the study must not be lost on the remaining NIH- or CDC-funded studies. By eliminating the placebo arm of the study, the Johns Hopkins researchers are not only transforming their study into an ethical study, but are also acknowledging that it is possible to conduct a scientifically valid and useful study without the use of a placebo arm. Because the availability of AZT is probably as limited in Ethiopia as in any other country in which these studies are planned or being conducted, the abandonment of the placebo group in that country should be viewed as an ethical standard to which all the remaining studies should adhere. Since these Johns Hopkins researchers have--to their credit--acknowledged the error of their ways, how can any of the other studies involving placebos be allowed to continue as planned?
New Data from ACTG 076 Showing that Shorter Treatments Are Effective
ACTG 076 was stopped in late 1993 because it so clearly showed that AZT significantly reduced HIV transmission to infants born to HIV-positive mothers that it would have been unethical to continue denying treatment to women in the placebo arm of the study. Women were allowed to enroll in ACTG 076 as late as the last weeks of pregnancy; the average duration of maternal treatment was 11 weeks before delivery. In an article published in 1994, the authors stated that the efficacy of AZT in ACTG 076 was observed in all subgroups, including groups differing in the their duration of prenatal treatment,3 but they did not publish the data upon which that conclusion was based.
We have now obtained the actual findings from the ACTG 076 researchers themselves (R. Gelber, personal communication). These previously unpublished data--presented to the NIH Data Safety Monitoring Board for ACTG 076 in February 1994--suggest that among those women getting 12 or fewer weeks of AZT (average duration of treatment of 7 weeks), the shorter courses were significantly better than placebo in this subgroup analysis (7.7% infected infants vs 22.9%), the very question being asked in the current placebo-controlled studies.
As can be seen in the figure below, short-term AZT use was associated with a 66.4% reduction in HIV transmission compared to placebo. In addition, when those women who had received AZT or a placebo for equal to or less than 12 weeks (average of 7 weeks) were compared with those women getting AZT or a placebo for more than 12 weeks (average 17 weeks), the reduction in HIV infection to their infants was almost identical. Compared to a placebo, there were 66.4% fewer HIV-infected babies in women getting 12 or fewer weeks of AZT treatment and 65.1% fewer HIV-infected babies in women who got more than 12 weeks of treatment. Thus, these data suggest that 1/ the shorter regimen is almost certainly more effective than a placebo, making subsequent placebo-controlled studies unethical, and 2/ the shorter and longer AZT treatments may have equal efficacy, a question that should be explored in subsequent equivalency studies in which all women get some active treatment. While this is a subgroup analysis of the ACTG 076 data, the randomized design and the strength of the association make it likely that the effects observed are real.
The ACTG 076 data presented above are supplemented by additional data that had already been published by the time ACTG 076 was completed. Among non-breast- feeding women, 65% of HIV transmission to infants is estimated to occur during delivery and 95% of the remaining 35% occurs during the last two months of pregnancy.4 Regimens like that in the Ethiopia study, which offers AZT for six to nine weeks before delivery and during delivery, cover the periods when HIV transmission is most likely to occur and are thus very likely to be effective, perhaps even as effective as the ACTG 076 regimen itself. To proceed to placebo-controlled trials after these data were in hand is clearly unethical because the question posed by such a study, "Is a short course of AZT better than nothing?" has been tentatively answered in the positive. To deny women access to a drug likely to save their infants' lives is in clear violation of all relevant ethical codes. Any researcher designing placebo-controlled trials without integrating such information concerning ACTG 076 into their study design has not carefully conducted the first, essential step in research, namely to re-search what has already been done.
More Recent Studies Further Undermine the Current Placebo-Controlled Studies
Since the publication of the ACTG 076 results, data suggesting the equivalency of shorter and longer regimens of AZT, or that the shorter regimen is better than nothing, have continued to accumulate. In the Bahamas, the transmission rates of women receiving less than two months and more than two months of prepartum AZT were similar (12.5% vs. 11.5%).5 In a multicenter U.S. study, duration of AZT administration prior to delivery was also not associated with the HIV transmission rate.6
Data presented at the recent Conference on Global Strategies for the Prevention of HIV Transmission from Mothers to Infants in Washington, DC (September 3-6, 1997) also make a mockery of several other persistent claims by the trial defenders. For example, a study conducted by the CDC itself in Thailand demonstrated similar blood AZT levels in women receiving oral or intravenous AZTduring delivery,7making it clear that the argument that the intravenous administration of AZT in ACTG 076 somehow makes the results of that trial inapplicable to and too expensive for the developing world is without merit.
Furthermore, there is now ample evidence that HIV transmission rates among women provided the ACTG 076 regimen in developing or underdeveloped countries rivals or is identical to that observed in the treatment arm of ACTG 076. The overall mother-to-infant transmission rate among women administered the ACTG 076 regimen was 7.9% in Thailand (vs. 7.6% in ACTG 076)8, 7.7% in Poland9, and 11.9% in the Bahamas.10 There are therefore no data to support the notion that the ACTG 076 regimen will somehow become ineffective when administered in developing countries.
Finally, the oft-repeated claim that the presence of breastfeeding in some developing countries makes the results of ACTG 076 inapplicable in those settings ignores estimates of the probability of HIV transmission from breastfeeding published in 1992, prior to the completion of ACTG 076. Since breastfeeding is estimated to represent a 14% risk of transmission from mother to infant,11 the only way breastfeeding could render the ACTG 076 regimen ineffective (it reduced HIV transmission in non-breastfeeders in ACTG 076 by 15%) would be if every woman whose transmission of HIV to her infant was prevented by the ACTG 076 regimen now passed on HIV by breastfeeding. This seems unlikely in the extreme.
The CDC's own protocol reveals that they believed the short course of AZT was likely to be as effective as ACTG 076 itself. A January 15, 1996 version of the protocol for the CDC-funded, placebo-controlled Thailand study (placebo vs 4 weeks of AZT) states that "This ZDV [AZT] study is proposed in the belief that short-course oral therapy may be as effective or nearly as effective as the full ACTG 076 regimen." This belief should lead to an equivalency study, not to a placebo-controlled study. Although there were already data strongly suggesting that the short course of AZT was more effective than a placebo and perhaps as effective as a longer course (see above), the absence of this information in the informed consent sheets we have seen (the CDC Thailand and CDC Cote d'Ivoire studies) is unethical.
A recent series of interviews with women participating in the CDC-funded placebo-controlled study in Cote d'Ivoire revealed that many of them, even after giving informed consent, were not aware of important aspects of the studies.12 Some agreed to participate in the studies only to gain access to medical care. Others believed that various other pills they had received would treat HIV infection. Thus, aside from deficiencies in the content of the informed consent documents, there is a serious logistical question as to how much of this information is comprehended by the women in the studies. In our original published material we did not allege infringements of informed consent since, even if informed consent were optimal, the studies would still be unethical because of their design. At the time there was no evidence that informed consent was being abrogated. Now the evidence is in.
We agree with a virologist from Zimbabwe, Dr. Oyewale Tomori, who wrote to us stating that "While the current AZT trials in Africa may not be strictly another Tuskegee affair, there is only an imaginary divide between the two. In an environment where the majority can neither read nor write and is wallowing in poverty and sickness, hunger and homelessness, and where the educated, the powerful, the rich or the expatriate is a semi-god, how can you talk of informed consent?"
Comparisons to Tuskegee
There has been much controversy over comparisons between the present studies and the Tuskegee study of untreated syphilis. While no analogy is perfect (that is why they're called analogies, after all), there are numerous critically important similarities between the present studies and the Tuskegee study:
1. They are both conducted or funded by the Public Health Service;
2. Both involve the denial of therapeutic regimens known to be effective;
3. Both involve studies of peoples of color;
4. Both involve violations of informed consent;
5. Both have been justified by arguing that their methods are the only way to get the study results;
6. In both studies it was argued that differences between the study population and the population previously studied (in the case of Tuskegee, studies in Norway describing the natural history of syphilis) justified the studies; and
7. Both have been justified by arguing that in the absence of the studies the subjects would not have been treated anyway.
The Tuskegee analogy may make some people uncomfortable, but that may be because they sense its accuracy.
Since we last corresponded with you, we have obtained important new information which makes the case even stronger for requiring that no more HIV-positive pregnant women in Africa or Thailand be denied access to proven therapies in these U.S.-funded studies. It is time to admit that serious mistakes have been made and to put an immediate stop to those parts of the studies which bear frightening similarities to Tuskegee and which continue to lead to unnecessary deaths among poor infants in the developing world.
Sidney M. Wolfe, M.D.
Public Citizen's Health Research Group
Peter Lurie, M.D., M.P.H.
Public Citizen's Health Research Group
1. Lurie P, Wolfe SM. Unethical trials of interventions to reduce perinatal transmission of the human immunodeficiency virus in developing countries. N Eng J Med 1997;337: 853-856.
2. Brennan TA. Letter to Gilbert Meier, NIH Division of Research Ethics, December 28, 1994.
3. Connor EM, Sperling RS, Gelber R, et al. Reduction of maternal-infant transmission of human immunodeficiency virus type I with ziduvudine treatment. N Eng J Med 1994;331:1173-80.
4. Rouzioux C, Costagliola D, Burgard M, et al. Timing of mother-to-child HIV-1 transmission depends on maternal status. AIDS 1993;7 suppl 2:S49-S52.
5. Gomez MP, Bain RM, Dean S, McNeil P, Read SE. Zidovudine reduces vertical transmission of HIV in the Bahamas. Presented at Conference on Global Strategies for the Prevention of HIV Transmission from Mothers to Infants, Washington, DC, September 3-6, 1997.
6. Frenkel LM, Cowles MK, Shapiro DE, et al. Analysis of the maternal components of the AIDS Clinical Trial Group 076 zidovudine regimen in the prevention of mother-to-infant transmission of human immunodeficiency virus type 1. J Inf Dis 1997;175:971-974.
7. Chearskul S, Chuachoowong R, Chopitayasunondh T. Presented at Conference on Global Strategies for the Prevention of HIV Transmission from Mothers to Infants, Washington, DC, September 3-6, 1997.
8. Phanuphak P. Letter to Peter Lurie, Public Citizen's Health Research Group. Program on AIDS, Thai Red Cross Society, October 6, 1997.
9. Lipniacki A, Burkacka E, Putz-Szczepanska M, et al. Prevalence of vertically-transmitted HIV in Poland. Presented at Conference on Global Strategies for the Prevention of HIV Transmission from Mothers to Infants, Washington, DC, September 3-6, 1997.
10. Gomez MP, Bain RM, Dean S, McNeil P, Read SE. Zidovudine reduces vertical transmission of HIV in the Bahamas. Presented at Conference on Global Strategies for the Prevention of HIV Transmission from Mothers to Infants, Washington, DC, September 3-6, 1997.
11. Dunn DT, Newell ML, Ades AE, Peckham CS. Risk of human immunodeficiency virus type 1 transmission through breastfeeding. Lancet 1992; 340:585-588.
12. French HW. AIDS research in Africa: juggling risks and hopes. New York Times, October 9, 1997, pp.A1, A8.