The New England Journal of Medicine -- September 18, 1997 -- Volume 337,

Number 12

SOUNDING BOARD

Unethical Trials of Interventions to Reduce Perinatal Transmission of the

Human Immunodeficiency Virus in Developing Countries

Inadequate Analysis of Data from ACTG 076 and Other Sources

The NIH, CDC, WHO, and the researchers conducting the studies we

consider unethical argue that differences in the duration and

route of administration of antiretroviral agents in the shorter

regimens, as compared with the ACTG 076 regimen, justify the use

of a placebo group. (15,16,17,18) Given that ACTG 076 was a

well-conducted, randomized, controlled trial, it is disturbing

that the rich data available from the study were not adequately

used by the group assembled by WHO in June 1994, which

recommended placebo-controlled trials after ACTG 076, or by the

investigators of the 15 studies we consider unethical.

In fact, the ACTG 076 investigators conducted a subgroup analysis

to identify an appropriate period for prepartum administration of

zidovudine. The approximate median duration of prepartum

treatment was 12 weeks. In a comparison of treatment for 12 weeks

or less (average, 7) with treatment for more than 12 weeks

(average, 17), there was no univariate association between the

duration of treatment and its effect in reducing perinatal HIV

transmission (P = 0.99) (Gelber R: personal communication). This

analysis is somewhat limited by the number of infected infants

and its post hoc nature. However, when combined with information

such as the fact that in non-breast-feeding populations an

estimated 65 percent of cases of perinatal HIV infection are

transmitted during delivery and 95 percent of the remaining cases

are transmitted within two months of delivery, (19) the analysis

suggests that the shorter regimens may be equally effective. This

finding should have been explored in later studies by randomly

assigning women to longer or shorter treatment regimens.

What about the argument that the use of the oral route for

intrapartum administration of zidovudine in the present trials

(as opposed to the intravenous route in ACTG 076) justifies the

use of a placebo? In its protocols for its two studies in

Thailand and Cote d'Ivoire, the CDC acknowledged that previous

"pharmacokinetic modelling data suggest that [zidovudine] serum

levels obtained with this [oral] dose will be similar to levels

obtained with an intravenous infusion." (20)

Thus, on the basis of the ACTG 076 data, knowledge about the

timing of perinatal transmission, and pharmacokinetic data, the

researchers should have had every reason to believe that

well-designed shorter regimens would be more effective than

placebo. These findings seriously disturb the equipoise

(uncertainty over the likely study result) necessary to justify a

placebo-controlled trial on ethical grounds. (21)