SMART tool moves beyond 'one treatment fits all'
for chronic conditions
Not only does one treatment not fit all, but over the long haul, one treatment probably won't fit anyone suffering from depression, substance abuse problems, HIV infection and many other chronic conditions, according to a U-M researcher.
With funding from the National Institutes of Health and the National Institute of Drug Abuse, statistician Susan A. Murphy, a research professor at the Institute for Social Research (ISR), has developed a new tool to help doctors make treatment choices and other complex decisions that go into individually tailored, adaptive, treatment strategies that change along with the individual's response to the treatment.
She calls the tool SMART: sequential multiple assignment randomized trials.
Tailoring treatment strategies to individuals and using different approaches at different stages of their treatment is the future of medicine. But until recently, clinicians lacked a means to make the complex series of repeated adjustments to treatment regimens for each patient struggling to lose weight, stop drinking, quit using illegal drugs or to keep functioning in spite of depression or AIDS.
How long should clinicians give the treatment a chance to produce a response for patients? That is, how long should patients continue on a treatment when so far there has been little or no sign of response or remission? And if a different type of treatment is indicated, what should it be? Are there patient outcomes that signal which treatment should come next? When is it best to offer additional therapies to patients who are having trouble adhering to treatment?
"Currently, scientists use a combination of clinical experience; trial and error; behavioral; psychosocial and biological theories; results from observational studies; and randomized experiments conducted for other purposes to formulate the decision rules for adaptive treatment strategies," Murphy says in describing how such questions now are answered. She hopes her SMART method will maximize the effectiveness of treatment by avoiding both the negative effects of overtreatment and by providing timely increases in treatment to those who can benefit.
"SMART is designed to enhance, rather than replace, the clinical judgment of practitioners," she says.
A treatment that appears best initially may not be best in the long run, Murphy says. So adapting treatment regimens not only makes sense for individual patients, but it also allows researchers to collect data on intermediate outcomes that can be analyzed to guide treatment of others.
In a recent issue of the journal Neuropsychopharmacology, Murphy and colleagues David Oslin, John Rush and Ji Zhu, members of a national network of computer scientists, statisticians, chemical engineers, psychiatrists, psychologists and others interested in SMART and similar techniques, review the methodological challenges in constructing effective treatment sequences for chronic psychiatric disorders. As part of the network, Murphy has established a Web site, neuromancer.eecs.umich.edu/dtr/twiki/bin/view, designed for physicians interested in learning about adaptive treatment strategies.
"Most of the experimental designs and data-analysis methods best suited for improving sequential clinical decision-making are found in nonmedical fields," Murphy says. "With input from all of these disciplines, we have the potential to jump-start the development of experimental data collection and data analysis methods like SMART that will inform and evaluate adaptive treatment strategies."
Murphy and other scientists from a variety of fields will discuss the development of adaptive treatment strategies in substance abuse research in a special issue of the journal Drug and Alcohol Dependence, to be published in late 2007. On Feb. 21 Murphy will lecture on this topic in Washington, D.C., at the annual meeting of the International Society for CNS Clinical Trials and Methodology.
"Our eventual goal," Murphy says, "is to combine new data on patient response and relapse with what we already know from engineering, computer science, infectious disease research, psychiatry and statistics in order to improve clinical judgment and thus help clinicians help their patients combat chronic conditions and prevent relapses among people who are working hard to adopt healthier lifestyles."