Is HIV testing a useful tool for prevention? This was the question asked
by Dr. David Wilson in his recent posting. Wilson goes on to ask,
"Do any readers of this forum know of any peer reviewed publication which
proves that HIV testing really leads to change of behavior in significant
numbers of cases?"
Rather than focus on behavior change per se, I suggest a better outcome
measure to answer Dr. Wilson's question is reduced HIV transmission. Thus
the hypothesis would be that testing reduces HIV transmission via a change
in sexual behavior or blood practices. Once treatment comes into the picture,
the potential benefits of testing to induce preventive behavior becomes
clouded. Hence the peer reviewed publication that Dr. Wilson is asking
for should either be studies done in wealthier countries before 1995-96
when more effective therapy started to appear, or coming from the developing
world where treatment is not available.
Most researchers agree that the best proof of a hypothesis comes from
randomized clinical trials. In such a trial, the different participating
groups should have the same inherent risk of HIV independent of whether
or not they received testing. While the information would be useful, doing
such a trial would be difficult due to ethical considerations. For example,
assume that HIV infected person were identified in a blood testing program
in developing country XXX where expensive treatments are not available.
Typically in this mythical (but realistic to Southeast Asia) country, blood
donors are not told that their blood is infected. To address Dr. Wilson's
question, researchers might allocate the detected HIV positive persons
to two groups, the "intervention" group that is told of their infection
and the "control" group that is not told. The researchers would then follow
both groups over time. Because the outcome of interest is further HIV transmission
to others, the study would be difficult to do unless the sexual partners
were also included, which adds to the complexity and expense of the study.
Thus our hypothetical research study might be made more efficient by limiting
participants to those who are married. The researchers would then study
the spouses of both the HIV infected "intervention" and "control" groups
to assess the transmission rate. By now it should be clear to all that
such a study would be highly unethical because the researchers would not
have informed the "control" group that they were infected and might pass
the virus on their loved ones. Thus I doubt that this kind of research
study could be done.
Instead of a randomized clinical trial, we more likely need to turn
our attention to cohort studies of discordant couples. Here all persons
are told that one member of the couple is HIV infected. The researchers
can then study the subsequent behavior of the couples and the rate of HIV
transmission to see if knowledge of infection which comes from testing
had a beneficial effect. Unfortunately with this type of research there
is no comparison group under observation. Fortunately, however, comparisons
can be made with what occurs in the general society.
The best study of
this kind that might answer Dr. Wilson question was published in 1993 by
Dr. Alberto Sarracco and colleagues [J Acquired Immune Deficiency Syndrome
6(5)497-502, 1993] and describes the HIV seroconversion experience of 343
steady female partners of HIV infected men in Italy. The HIV transmission
rate among those who did not always use condoms was 9.7 per 100 person-years
(roughly interpreted as 9.7 percent sero-conversion in one year), and 1.1
per 100 person-years among those who reported always using condoms. A third
group among the 343 couples suspended sexual relations with their infected
partners. They had no new infections. But did testing lead to preventive
behavior change? That is the question that Dr. Wilson is trying to address.
When told that the male member of the couple was infected, 61 percent
of the couples either abstained from sex (38 or 11%) or always used condoms
(171 or 50%). These two groups are the ones that benefited most from knowing
that the partner was infected. Thus it appears that at maximum 61 percent
changed their behavior based on the test results and 39 percent did not.
Yet without a control group (which would have been unethical), we cannot
say for sure what percentage of stable couples would normally have abstained
from sex or always used condoms.
My impression is that few married couples
practice abstinence or regularly use condoms (perhaps less than 5 percent).
Thus I suspect that the vast majority of the behavior change in the Sarraco
study can be attributed to the beneficial impact of testing. Others will
have to decide for themselves.