Strategy Trials: The Answer?
During the 6th Conference on Retroviruses and Opportunistic Infections (Chicago, January 31 - February 4, 1999), we received much promotional material on drugs now being marketed or about to be approved soon. We do plan to outline some of the major drug-specific news, as there is information worth knowing.But there was also widespread disappointment at the paucity of practical treatment information--despite the promising advances in research, especially immunology. Somehow the piles of faxes, press releases, and other promotional materials--the 12-week, 24-week, or 48-week data, seldom comparable between trials and usually highlighted to promote one drug, or one company's drugs--do not add up to an intelligent approach to treating patients.
Doctors and patients have many treatment options today, but little help with the questions of which treatment strategies are best--which choices now will prove to have been the most beneficial, not only in 12 weeks or 48 weeks, but years later, even after the initial treatment choice may have failed or otherwise been replaced with different treatments.
For several years there has been a growing call for "strategy trials" to answer just these questions--but in practice, few of them are happening. Much of the fault is with the pharmaceutical companies, which are mainly interested in trials likely to promote their products. Government-funded trials help to fill the gap, but often there is difficulty getting assistance from the companies. And public funding can go only so far in doing the drug development which is supposed to be the responsibility of private companies.
Another important difficulty with strategy trials is more inherent, harder to solve with any amount of altruism. Typically these trials will randomly assign patients to a certain starting drug regimen; and later, if and when that regimen fails, will assign them in some protocol-defined way to a second regimen. The goal is to learn which of two or more treatment strategies is best--considering not only the initial drug combination, but also what happens if that combination later needs to be replaced.
Unfortunately it is likely to take much longer to get useful information from such a strategy trial, than from one which tests only a starting regimen, because volunteers must fail one treatment and then have enough experience with the second to test how well it is working for them. Unless there are enough volunteers with substantial experience with the second regimen to allow statistically reliable conclusions about it, the whole scientific point of including it would be lost. And researchers are usually reluctant to release much preliminary information while a trial is ongoing, lest the final results be affected in unpredictable ways. Including all the other delays, such as approvals and recruiting, years are likely to elapse between a strategy trial's design and its results; and with the speed of AIDS treatment development today, the whole picture is likely to have changed greatly by that time, which can easily make the trial results largely irrelevant to the most important issues which doctors and patients will need to address. This has already happened again and again with trials designed years before their results were released; strategy trials are even more vulnerable, both because they usually take longer, and because their protocols are more complex and therefore more vulnerable to a changing medical environment.
One way to approach these problems is to step back from the ideology that medicine should be guided only by scientific trials, and acknowledge that clinical experience and judgment is still as important as clinical trials in making practical treatment decisions, and will probably remain so for the foreseeable future. Then we can more systematically address the issues of how to improve data collection to better support clinical experience. Steps in the right direction include better followup of trials, and also large databases to more systematically collect outcome information from non-randomized medical care (such as the CHORUS project of Glaxo Wellcome, or the community-initiated HIV Treatment Data Project).
A better understanding of clinical trials--including their limitations, and their real-world role in medical decision making--could lead to better trials, and also to more attention on improving the collection and use of non-randomized data to support clinical practice.
