Seeking the perfect protocol

The study design is the framework for all research

ILLUSTRATION: SEAN KENNEDY

In truth, the experimental design is central to all trials, and it contributes to their successes and failures. In fact, no matter how safe and effective a drug may actually be, a poorly conceived research program ultimately will fail to prove its virtues; for example, the FDA will not approve a medication if the research meant to demonstrate its worth included insufficient sampling sizes (a statistical problem), was analyzed using suspect mathematical models, or failed to support the declared claims of therapeutic value.

As a result, a new pharmaceutical product (also known as a “new chemical entity”, NCE) is never investigated haphazardly. Further, because of research costs and various inherent time elements associated with the development process (see “We are the world?”, June 2001, p 36), it is critical that clinical experiments be designed carefully and accurately to address key questions of which patient populations should be studied, what the hoped-for results will be, and how the study design should be constructed to provide a credible foundation from which to interpret the data and reach meaningful conclusions.

Before delving too deeply into the design of the clinical trial itself, though, it is important to describe the scope and content of its documentation, known as the “protocol”. This document delineates the purpose of a trial, the nature and composition of an NCE, the patient population, the number of clinic visits the subjects will be expected to make, the assessments that will be included at each visit, the manner in which those assessments will be conducted, the investigator’s responsibilities (beyond those in the U.S. Code of Federal Regulations), the oversight delegated to contract research organizations, and the statistical review that will analyze the collected data.

The framework

Phases of clinical trials in humans (specifics for each phase may vary)
Phase	No. patients	Duration	Location	Purpose
I	20–80 (healthy)	36 hours, over 2–6 weekends	Single site	Bioavailability, pharmacokinetics, safety, dosing regimens/ranges
II	50–200 (diseased)	36 hours, over 1–4 weekends	Single site	Safety and preliminary efficacy
III	100–2000 (diseased)	6 months– 3 years	30–40 sites	Immediate and long-term safety and efficacy
IVa	100–2000 (diseased)	6 months– 3 years	30–40 sites	Efficacy for alternate indications
aPhase IV studies are known as “postmarketing” and are conducted only after an NCE is approved by the FDA and the pharmaceutical company wishes to explore a further use of their medication in treating different indications (i.e., medical conditions).

All clinical trial designs begin by addressing the issues of study intent, size, and population. Clinical trials consist of four phases, and each one requires a different set and number of patients (see box, “Phases of clinical trials in humans”). The participant population is defined on the basis of phase and therapeutic need; the protocol outlines the specific inclusion and exclusion criteria. The purpose of the trial is also outlined in the protocol: Generally, only one study each is conducted in Phases I and II, and their purposes are universally defined. However, later phase trials may be multiple as different effectiveness measurements are obtained. For instance, one Phase III study might test the NCE versus a placebo, another might pit the NCE against a competitor’s product, and still another might consider the combined effect of the NCE and an additional, marketed medication.

Once the intent of the study is established and the patient population identified, the specifics of the trial must be determined. A primary consideration involves how many “arms” the study will have and what those arms will be. For example, in a two-armed study, the total patient population would be divided in half, with each half receiving a different course of treatment, such as the NCE or a competitor’s drug. Once the number and type of arms are defined, a choice must be made on whether the research will be blinded, and if so, to what extent. The term “blinding” refers to how much information the patients, researchers, and study monitors will possess about a particular patient’s treatment course. In general, to remove all biases, clinical trials are either double- or triple-blinded; commonly, only the statistician who developed the methodology for the trial can determine to which arm a particular patient has been assigned. Although this may sound unsafe, it’s not. The investigating physicians are provided with codes that can identify a particular patient’s medication in the case of an emergency. This procedure is known as “breaking the blind”.

The experiment
The “meat and potatoes” of the study design, however, involves the actual patient visits and the events that occur during them. Obviously, it is not possible to describe the vast array of schedules that are used within different studies, but a general review of their high points is instructive.

During Phases I and II, there are commonly only a few short visits, which are carefully controlled to reduce the number of extraneous variables. Specifically, patients are required to eat identical foods, consume the full portions of those meals, record the quantity of all additional ingestions such as water, and sleep or at least rest quietly for a pre-set period. In addition, patients in these trials start and stop their meals, are dosed, and have their blood drawn at carefully timed and precise intervals so that metabolic information regarding the medications can be accurately determined.

These trials are known commonly as “bleed and feeds”, because the primary assessment method involves blood assays. As such, patients begin a typical visit by arriving at night for dinner and then beginning the next day with breakfast, followed by a dosing of the NCE, and then a series of 10–15 blood draws before lunch. Often, a total of five more draws are then performed throughout the afternoon and evening, concluding with a final draw before dismissal the next morning.

Phase III and IV trials, by contrast, are markedly uncontrolled. Their purpose is, in part, to test the NCEs under “real-life” conditions; patients make visits just as they would go to a doctor’s office, and the visits typically last just as long. At each visit, patients are asked about their physical well-being and any adverse events or concomitant medications that are new since the last visit. (Visits are often spaced a month apart and may continue for many years.) In addition, the patient’s dosing regimen is reviewed and altered if necessary, laboratory tests are completed, and assessment exams are conducted as appropriate (see “In the mind’s eye”, June 2001, p 21).

The protocol details every visit’s events (these assessments often vary from visit to visit, particularly in later phase trials), the allowable interval between patient visits, the requirements for record keeping, and instructions for completing unique facets of the trial. For example, the protocol for a study that includes patient diaries might feature directions for their use.

The nitty-gritty
A key component of every protocol is the scientific/mathematical justification for the research design and the corresponding explanation of the methodologies to be used in the data analysis. (The rigid design of the clinical database—the repository of all collected research data—will be detailed in next month’s Clinical Trials Track.)In addition, it is here that the “randomization scheme” is justified; this is the instrument used to assign patients, randomly and blindly, into various arms of the trial. Finally, it is within this statistical review that the expected patient enrollment is shown to be sufficient for providing a statistically meaningful data set.

Ultimately, the goal of a clinical trial is to prove an NCE’s safety and efficacy. Thus, the goal of the experimental design is to structure a study that can do just that. In addition, it is important to realize that issues of patient safety and health are paramount to the design. Scientists and researchers must be very careful to create clinical trials that do not jeopardize or needlessly complicate a patient’s health or quality of life. The reasons are obviously ethical and legal; for example, U.S. law forbids the incorporation of a placebo arm in a study that will include patients suffering from life-threatening diseases for which an alternative therapy is available. Another reason, however, goes to the statistical heart of a trial: Patient enrollment may suffer if the chosen clinical assessments or dosing regimens are extreme or complex.

Finally, it is worth noting that protocols are routinely amended even after a trial begins (with investigational review board and FDA approval). Amendments to correct errors in design are obviously a much cheaper approach than discarding an approved study and starting fresh. Of course, spending the time and consideration to develop an effective protocol at the beginning of a study can ensure better, faster, and more accurate and complete results and conclusions, which are critical in achieving FDA approval.