Experimenting To Improve Clinical Practice

Clinical practice often involves sequential decisions. For example, altering the intensity and type of treatment over time is crucial for many reasons, such as to obtain improvement if the individual is not responding or to reduce costs and burden when intensive treatment is no longer necessary. Clinicians adapt the treatment based on an individual’s background (e.g. severity of disorder, preferences) and then dynamically utilize the individual outcomes (response to treatment, adherence, side effects) to readapt treatment.  Indeed, Weisz et al.  in a 2004 discussion of dissemination and evidence-based practice in clinical psychology, maintained that evidence-based practice should ideally consist of much more than simply obtaining an initial diagnosis and choosing a matching treatment:

Evidence-based practice… is not a specific treatment or a set of treatments, but rather an orientation or a value system that relies on evidence to guide the entire treatment process. Thus, a critical element of evidence-based care is periodic assessment to gauge whether the treatment selected initially is in fact proving helpful. If it is not, adjustments in procedures will be necessary, perhaps several times over the course of the treatment. (pp. 302--303)

Important sequential questions that arise in this setting include:  “Which treatment should we try first and which should come second,” “How long should we try the first treatment before giving up and trying another treatment,” “How long should we use a treatment that is working prior to moving the individual to a lower burden maintenance treatment” and “What information should we use to make these decisions.”   Unfortunately classical two or three group experiments provide little evidence to address these types of tactical questions. However a recent innovation in the statistical field has led to new experimental designs that can be used to directly inform sequential clinical decision making. In a Sequential Multiple Assignment Randomized Trial (SMART) participants move through treatment stages; each stage coincides with one of the sequence of decisions and at each stage participants are (re-)randomized among treatments.  We introduce the SMART and highlight its advantages in informing adaptive sequential decision making as compared to alternative experimental designs. We illustrate how different questions, questions more relevant to adaptive, sequential decision making can be addressed by the SMART.  Although this experimental design is in its infancy, SMART studies have been conducted or underway in a variety of clinical fields, including cancer, schizophrenia, depression, alcohol dependence, substance abuse and autism. Selected examples of SMART studies will be discussed. We illustrate the types of results that can be obtained via the data analysis of a SMART study involving children with attention deficit and hyperactivity disorder.