STAR*D Starts to Resemble Practice

When clinical practice appears to diverge from evidence-based medicine, is the clinician departing from science, or are the data not applying to practice? The challenge of developing clinical research data to inform treatment strategies for the inconstant course of psychiatric illness was recently considered by Susan Murphy, PhD, of the University of Michigan's Institute for Social Research, with colleagues from the MCATS (Methodology for Constructing Adaptive Treatment Strategies) network, and John Rush, MD, of the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) investigators group.

When clinical practice appears to diverge from evidence-based medicine, is the clinician departing from science, or are the data not applying to practice? The challenge of developing clinical research data to inform treatment strategies for the inconstant course of psychiatric illness was recently considered by Susan Murphy, PhD, of the University of Michigan's Institute for Social Research, with colleagues from the MCATS (Methodology for Constructing Adaptive Treatment Strategies) network, and John Rush, MD, of the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) investigators group.1

The NIMH-funded STAR*D trial is the largest and longest study ever conducted to evaluate depression treatment, and it was designed to better reflect conditions of practice than have traditional clinical trials. While researchers traditionally evaluate specific treatments in selected populations over short periods, clinicians treat a varied population of patients with different treatment responses, comorbid illnesses, and concurrent medications and symptoms that wax and wane over time. Treatment strategies must be adaptive, then, for the dynamics of illness and such factors as the emergence of comorbid conditions, adverse drug reactions and interactions, and how the patient adheres to treatment regimens.

The STAR*D study design of sequenced treatments "reflects what is done in clinical practice," according to the NIMH, in a statement issued in November, "because it allowed study participants to choose the treatments most acceptable to them and limited the randomization of each participant only to his or her range of acceptable treatment strategies."

The STAR*D study investigators described this as "equipoise-stratified randomization," with subjects and clinicians determining whether certain sets of options would be unacceptable.2

"One of the imperatives guiding the development of STAR*D," the investigators related, "was to preserve the integral role participants typically play in negotiating treatment decisions in clinical settings. As medical practice has evolved and public access to information regarding treatment of mental illness has grown, clinicians have increasingly recognized the value of patient participation when decisions need to be made."3

This aspect of the STAR*D design has its detractors, however. In an editorial in the November 2006 issue of The American Journal of Psychiatry, Craig Nelson, MD, indicated, "The greatest disappointment of the study-depending on your perspective-was that patients were not randomly assigned to all treatments at level 2, and as a result comparisons between treatment strategies were limited."4

This contention notwithstanding, there is widespread support for this effort to conduct clinical research with patients and conditions more reflective of clinical practice and to assess treatment strategies for the varied and changing circumstances encountered in practice.

Data for sequential decisions

Murphy and colleagues suggest that sound sequential decisions follow from observing important clinical outcomes that mark "critical decision points." From this, they indicate, decisions are made to optimally treat the disorder, maximize function, and minimize the burden of illness. In the typical dilemma of a patient's symptoms improving but not fully remitting, the clinician confronts the choice of switching to a different treatment, with risk of losing the initial benefit or increasing adverse effects, or maintaining the marginally successful treatment in hope that more improvement will occur with time.

"Adaptive treatment strategies, treatment algorithms, and expert systems provide a framework for operationalizing these key clinical decisions," Murphy and colleagues explain. "By operationalizing these decisions, they can be studied and improved upon, with the aim of reducing inappropriate variance in treatment delivery while retaining appropriate flexibility to tailor these decisions to individual patients."1

Murphy and colleagues suggest that studies based on the Sequential Multiple Assignment Randomized Trial (SMART) design,5,6 which uses multiple randomizations at each critical decision point, are more likely to inform sequential decisions than traditional randomized clinical trials. They note that these studies are not confirmatory, however, and may not involve a control condition. Murphy and colleagues also indicate that several SMART design trials may be necessary to fully inform an adaptive treatment strategy and that the validity of that strategy may need further testing against alternatives in a confirmatory randomized clinical trial.

The STAR*D study does not appear to be an exception to these characteristics. Despite enrolling 4041 patients over a 7-year period, there was a lack of differences between treatments at levels 2, 3, and 4, which, Nelson remarks, "leaves us without a roadmap to guide treatment selection and leaves us wanting more."

The characteristics of psychiatric illness, however, "motivate the development of adaptive treatment strategies," Murphy and colleagues declare. "Addressing tactical questions concerning the length of time to wait for treatment response and the choice of subsequent treatment are crucial in this endeavor."

References:

References1. Murphy SA, Oslin DW, Rush JA, et al. Methodological challenges in constructing effective treatment sequences for chronic psychiatric disorders. Neuropsychopharmacology. 2007;32:257-262.
2. Alpert JE, Biggs MM, Davis L, et al. Enrolling research subjects from clinical practice: ethical and procedural issues in the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) trial. Psychiatry Res. 2006;141: 193-200.
3. Fava M, Rush AJ, Trivedi MH, et al. Background and rationale for the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study. Psychiatr Clin North Am. 2003;26:457-494.
4. Nelson JC. The STAR*D study: a four-course meal that leaves us wanting more. Am J Psychiatry. 2006;163: 1864-1866.
5. Dawson R, Lavori PW. Placebo-free designs for evaluating new mental health treatments: the use of adaptive treatment strategies. Stat Med. 2004;23:3249-3262.
6. Murphy SA. An experimental design for the development of adaptive treatment strategies. Stat Med. 2005; 24:1455-1481.