In our series on research methods, we've touched on randomized controlled trials (RCTs) and one type of RCT, the pragmatic clinical trial. But there are several other types of RCTs that can be useful to researchers. These study designs might be more appropriate for studying chronic diseases, determining treatment when there is therapeutic uncertainty, evaluating health system changes, or addressing the need for specific clinical endpoints. Let’s take a closer look.
A crossover design allows patients to act as their own controls, enabling comparisons between and within groups. Patients receive a sequence of treatments over successive periods of time, crossing over to an alternative therapy as part of the sequence. At the start of the study, every patient is assigned to a sequence (eg, AB vs. BA), with successive treatments typically separated by a washout period (a specified period of nonuse prior to initiation of therapy). Because all participants receive the same number of treatments, usually in different sequences, it is possible to determine how patient characteristics influence response to treatment. There are several types of crossover designs with different numbers of periods and sequences, and all these designs have their own sets of considerations. We won’t get into the different designs and considerations in this post, but generally, crossover designs are most appropriate to study treatments for stable and chronic diseases.
N of 1 Randomized Controlled Trials.
N of 1 trials are individual randomized, double-blinded crossover trials that compare two or more treatment options for a single patient. In such a study, the patient undergoes multiple sequential treatment periods during which an active intervention is paired with a matched placebo or an alternative therapy. For each period, the order of administration of the active therapy or comparator is assigned randomly, such as by a coin toss, and ideally both the patient and clinician are blind to the assignment. Appropriate outcomes (those that are of interest to and readily reported by the patient) are often measured through the use of a diary or questionnaire. N of 1 trials tend to be used by clinicians when they are faced with therapeutic uncertainty for a particular patient. This approach offers an alternative for making individualized treatment decisions based on objective data, patient values, and patient-centered outcomes.
Cluster Randomized Controlled Trials.
Cluster RCTs are studies in which patients are grouped (clustered) on the basis of geography of practice (eg, caregivers, hospitals, communities), and then randomized as a group to either the intervention or control arm. Community-based cluster RCTs are generally characterized by small numbers of clusters (eg, hospitals) with a large number of patients in each cluster. They are viewed as a pragmatic methodology to measure the effectiveness of an intervention on a large scale. Cluster RCTs have commonly been used to evaluate the delivery of healthcare services, the effects of educational interventions, or the effects of organizational changes. More recently, researchers have suggested that cluster RCTs can be used to determine the comparative effectiveness of two or more therapies where there exists a natural variation in practice patterns.
Delayed Start Designs.
Delayed-start designs, or randomized-start designs, allow studies to determine whether an intervention acts by reducing symptoms or by modifying disease, an important distinction for slowly progressing diseases such as Parkinson’s disease. Delayed-start study designs are useful when there is a need for a clinical end-point that reliably measures disease progression and is not confounded, or confused, by the study intervention’s effects on symptoms.