Why are clinical trials so important? Short answer: They are an essential means for growing our clinical professions. Clinical trials are gateways through which new diagnostic tests and new treatments pass in becoming the professional tools of clinicians. Not surprisingly then, clinical trials place prominently in the process of evidence-based practice (for a review of evidence-based practice, see Dollaghan's article in the April 13, 2004 issue of The ASHA Leader as well as ASHA's "Evidence-Based Practice in Communication Disorders: Position Statement," 2005).
Clinical trials have been around for some time. Clinical researchers in many fields, third-party payers, regulators, and policy makers all recognize a model of clinical research for assessing the potential value of new clinical protocols (e.g., a new test or treatment). Within that model, a clinical trial is the standard means for testing the efficacy of new protocols as well as the efficacy of existing protocols when used under new circumstances or with new measures of outcome. If the test is successful, the protocol may become a legitimate part of clinical practice, particularly if other clinical trials confirm the positive result.
For that reason, many schemes for coding levels of evidence in the critical-appraisal portion of evidence-based practice have as the highest grade "a high-quality clinical trial." The only better form of evidence is a positive outcome in a meta-analysis or systematic review of multiple high-quality trials.
Two examples demonstrate this point. The first is a levels-of-evidence scheme for coding the quality of evidence found in a body of treatment studies. The second is a scheme for coding the quality of evidence in a body of diagnostic studies. In each case, clinical trials constitute the preeminent form of evidence for supporting the use of a certain protocol in clinical practice. From that perspective, high-quality clinical trials are the most important contributions that researchers in audiology and speech-language pathology can make in providing practitioners with best and current evidence.
Although clinical trials are certainly not the only form of research for supporting clinical practice, in the long run, evidence-based practice leads to equivocal conclusions without them. So why don't we have more clinical trials? A high-quality trial is a tremendously complex and expensive proposition requiring extraordinary scientific rigor.
Complexity, expense, and rigor notwithstanding, individuals with communication disorders and the future of our clinical professions will be well served by increasing our productivity in this important undertaking.
What Is a Clinical Trial?
A clinical trial is prospective research aimed at answering a specific question emerging out of clinical practice. The clinical question particularizes one or more populations and a new clinical protocol. For instance, a clinical trial might assess a new protocol for preventing a communication disorder. Alternatively, a clinical trial might assess a new screening protocol, a new technology, a new diagnostic protocol, a new treatment protocol, or a new dimension of outcome (e.g., quality of life).
No matter the type of clinical protocol being tested, all clinical trials share a common trait: precise, thorough, and systematic evaluation of the protocol. If done well, the results observed within a trial generalize to members of the target population through statistical inference. That is, a good clinical trial produces scientifically valid evidence for answering the clinical question.
Demonstrating that a new clinical protocol alone is responsible for observed results is absolutely essential. Accomplishing that goal requires extraordinary rigor in experimental design. For instance, the experimental protocol must be tested against some control or reference population. This is why clinical trials are often termed controlled clinical trials.
Sometimes, persons without a communication disorder constitute the control population (e.g., typically developing children). Other times, the control population is the target clinical population but without exposure to the new clinical protocol. This is a no-treatment control (e.g., children with conductive hearing loss and middle-ear effusion who do not receive tympanostomy tubes).
Individuals participating in the no-treatment control arm of a trial may be wait-listed and offered the treatment at the conclusion of a trial. Sometimes, the control population is the target clinical population with exposure to a standard, or best-practice, protocol (e.g., standard-treatment control). For example, a trial might compare an experimental method of aural rehabilitation designed for new hearing aid users versus another method that is widely used in clinical practice.
Simply put, demonstrating that a new clinical protocol alone is responsible for observed results in a trial means ruling out all counter explanations. The most objective means for ruling out bias in the composition of the experimental and control groups is randomly assigning participants to group membership. A trial with random assignment is termed a randomized-controlled clinical trial, or more simply, a randomized-controlled trial (RCT).
However, random assignment is sometimes undesirable and sometimes impossible. For instance, when the control population consists of individuals without a communication disorder and the experimental population consists of individuals with a certain communication disorder, random assignment is impossible and the trial is termed a nonrandomized-controlled trial.
Whether group membership results from random assignment or some human attribute, it is important to observe a sufficient number of participants to establish scientific validity. In fact, starting out with more than a sufficient number is necessary since participant attrition is a fact of life for researchers. Often, the necessary sample sizes just can't be obtained through a single research facility. When multiple research facilities collaborate in carrying out a clinical trial, the trial is termed a multi-center trial.
A host of other factors that might influence the outcome of a trial require accounting or control. Blinding is one such control. In a single-blind trial , participants are unaware of their status regarding placement in the control arm or the experimental arm. When the individuals administering the clinical protocol(s) are also unaware of group membership, the term double-blind trial applies. A triple-blind trial requires that group identity is also concealed from the data analyst.
External funding is required to meet the exceptional expense of conducting a high-quality clinical trial. Funding agencies consider the certain costs of recruitment, personnel, facility, equipment, and material and set these against the uncertain potential for improving the care of individuals with a communication disorder. Because a trial combines high stakes and uncertain outcome, researchers test protocols that are important and reasonably well understood (e.g., why it works, how it is best implemented, how it is best measured, for whom it applies or does not apply, the potential for risk, the potential for beneficial outcome).
Determining When a Clinical Protocol Is Ready for Trial
Clinical research on humans occurs in five stages or phases of testing. Clinical trials constitute Phase III testing. In Phases I and II, testing is dedicated to all of the experimental inquiry that necessarily precedes a decision to take a clinical protocol to trial. Scientific inquiry regarding a new clinical protocol may even begin prior to Phase I in animal or pre-clinical human studies in the form or translational research designed to determine the physiologic bases of safety and post-treatment change.
From the earliest beginnings of experimentation and through the completion of Phase II, researchers estimate the magnitude of effect size, define the clinical population, assess potential risks, potential benefits, and safety (i.e., the potential benefits outweigh the potential risks), finalize inclusion and exclusion criteria, determine the primary outcome and how to measure it, finalize the clinical protocol, establish reliability, develop administration manuals, and determine optimal dosage.
When a new protocol is ready, efficacy is tested in a Phase III clinical trial. Efficacy is defined throughout the clinical-research and public health communities as the potential or probability of benefit expected when a new clinical protocol is applied under the best of all possible conditions. Efficacy indexes maximum benefit using highly skilled practitioners in the best possible circumstances with best possible participants.
A successful Phase III clinical trial establishes the expectation that a new protocol will bring about the desired benefit when applied to members of the target population. Prior to a successful trial, there is no reasonable expectation of benefit, and Principle 1, Rule G of ASHA's Code of Ethics applies (2003). Therefore, in a Phase III trial, the new protocol is offered to volunteers rather than to clients on a for-pay basis.
When the purpose of a clinical trial is testing the efficacy of a new treatment, researchers likely use one of three research designs. A cross-over trial is appropriate when the effect of a new treatment on human physiology vanishes shortly after it is withdrawn. Because the effect of such a treatment washes out, a subject may be exposed to both the control and the experimental conditions. Examples include noise reduction circuits in hearing aids, adaptive directional microphones, and pharmaceuticals. In contrast, behavioral treatments (e.g., Lindamood-Bell Phoneme Sequencing Program, Vestibular Rehabilitation Therapy, Smith Accent Voice Therapy, Cueing-Verb Treatment) are successful only when therapeutic change persists.
This carry-over effect invalidates the cross-over design and the correct research design becomes a parallel-groups trial. Occasionally, a trial consists of intensive and extensive study of a few special participants. These studies are termed n-of-1 trials. Regardless of the type of research design, the most important characteristic of a trial is always quality: Does the trial produce valid evidence for answering the clinical question?
Although the fine points in assessing the quality of a report describing a clinical trial are not entirely self-evident, the skill set acquired in graduate school for assessing the validity of any research report applies to reports of clinical trials.
Furthermore, some excellent resources are available for guiding consumers in assessing reports of clinical trials. The organization Standards for Reporting of Diagnostic Accuracy (STARD; http://www.stard-statement.org/) provides readers of diagnostics studies with a systematic means for assessing quality. A similar organization called Consolidated Standards of Reporting Trials (CONSORT: http://www.consort-statement.org/consort-statement/) provides a similar means for assessing parallel-group studies (much of this work also applies to cross-over designs). Transparent Reporting of Evaluations with Nonrandomized Designs (TREND; www.trend-statement.org/asp/trend.asp) provides a systematic means for assessing the quality of non-randomized trials.