|
|
The Nature of the Research Question.
In their article in The New England
Journal of Medicine, Lurie and Wolfe argued that researchers were asking
the wrong question in their conduct of the clinical trials--"Is the shorter
regimen [of AZT] better than nothing?"--a question which only a placebo arm
could answer.
Taking the "more optimistic view"
that researchers would be capable of designing an effective short course of
treatment, Lurie and Wolfe maintained that the better research question was,
roughly, "Can the duration of treatment with AZT be reduced without compromising
the demonstrated efficacy of the 076 regimen?"
This research question, they wrote,
could be answered by equivalency studies, in which the proven therapy was the
comparison arm against which less expensive or toxic treatments would be measured.
Such studies, they added, would "provide even more useful results than placebo-controlled
trials, without the deaths of hundreds of newborns that are inevitable if placebo
groups are used."
Lurie and Wolfe based their optimism
about the likely success of trials of shorter courses of AZT on clinical data
about the timing of HIV transmission and on findings from the ACTG 076 study
itself, in particular, a "subgroup
analysis" from that study, which suggested that the 076 regimen was effective
for shorter durations than twelve weeks. The existence of these data, they further
contended, disturbed the "equipoise," or uncertainty about the outcome of a
trial, that was necessary to justify the use of a placebo arm.
Proponents of placebo-controlled
trials, on the other hand, argued that the question these studies asked was
more nuanced than depicted by Lurie and Wolfe. The trials in Thailand and sub-Saharan
Africa sought to determine not simply whether a shorter regimen of AZT was better
than nothing, but rather, in the words of one analysis, whether it was "safe
in these populations, and, if so, whether the demonstrated efficacy is large
enough, as compared to the placebo group, to make it affordable to the governments
in question."
To such questions, they maintained, the answers were by no means assured
and researchers, therefore, remained in a state of equipoise.
Researchers, many argued, could
not confidently apply the results of 076, gathered from trials in two highly
industrialized nations, to sub-Saharan Africa, where starkly different conditions--economic,
environmental, cultural, and biological--prevailed.
"These are two different populations,"
said Dr. H.M. Coovadia, head of the Department of Pediatrics at the University
of Natal in Durban, South Africa, during an
appearance on National Public Radio's Talk of the Nation.
For example, he and others pointed
out, women in Africa breastfed their infants, which the subjects in the 076
trials did not. Although breastfeeding increased the risk of HIV transmission,
health officials in Africa generally continued to recommend the practice on
a number of grounds: the cost of formula; the lack of clean water; the nutritional
benefits of breast milk; and, in a culture where breastfeeding was the norm,
the risk of being stigmatized by others as an HIV-infected mother.
In addition, women in African countries
frequently suffered from anemia--a common side effect of AZT therapy--as well
as from malaria, genital infections, and vitamin deficiencies, all of which
could potentially produce different effects from those observed in the 076 trials
in the US and France. Consequently, Coovadia said, he and others were not convinced
that 076 was "a sort of universal standard by which you could measure effectiveness
to reduce mother-to-infant transmission in all parts of the world."
Similarly, those who favored placebo-controlled
trials disputed the contention by Lurie and Wolfe that the 076 subgroup analysis
was a useful predictor of the likely effectiveness of shorter courses of AZT.
The data from the sub-analysis, they maintained, did not "determine the efficacy
of short course treatment" or obviate the need for additional trials.
At most, they indicated the direction further research might take. 
|
