The Five Kinds of Control a Drug Trial Can Use

Answer first: the purpose of a control is to rule out three other explanations

The regulation that defines an adequate and well-controlled study, 21 CFR 314.126, opens with the reason any of it exists. The purpose of conducting clinical investigations of a drug, it says, is to distinguish the effect of a drug from other influences, such as spontaneous change in the course of the disease, placebo effect, or biased observation.

Those three named influences are the whole argument for a control group. A disease can improve on its own. People can improve because they believe they are being treated. Observers can see what they expect to see. A result from a group that received a drug, on its own, cannot separate the drug from any of them.

The regulation lists characteristics that the scientific community recognizes as the essentials of an adequate and well-controlled clinical investigation, and states that reports of adequate and well-controlled investigations provide the primary basis for determining whether there is substantial evidence to support claims of effectiveness for new drugs. It also says the study report should provide sufficient detail of design, conduct and analysis to allow critical evaluation — which is a standard about transparency, not just about design.

The five types, in the order the regulation puts them

Section 314.126(b)(2) says that generally, the following types of control are recognized, and then names five.

Placebo concurrent control. The test drug is compared with an inactive preparation designed to resemble the test drug as far as possible. The regulation notes that such a study may include additional treatment groups, and usually includes randomization and blinding of patients or investigators, or both.

Dose-comparison concurrent control. At least two doses of the drug are compared. This too usually includes randomization and blinding.

No treatment concurrent control. Where objective measurements of effectiveness are available and placebo effect is negligible, the test drug is compared with no treatment. These trials usually include randomization. The two conditions in that sentence are the point: it is not a general-purpose design, it is one for situations where there is nothing for a placebo effect to move.

Active treatment concurrent control. The test drug is compared with known effective therapy — for example, the regulation says, where the condition treated is such that administration of placebo or no treatment would be contrary to the interest of the patient.

Historical control. Results with the test drug are compared with experience historically derived from the adequately documented natural history of the disease or condition, or from the results of active treatment, in comparable patients or populations.

Two of the five carry warnings written into the regulation itself

The active-control entry contains a caution that is easy to miss and important to anyone reading a head-to-head claim. If the intent of the trial is to show similarity of the test and control drugs, the regulation says, the report should assess the ability of the study to have detected a difference between treatments. It then states the reason in one blunt sentence: similarity of test drug and active control can mean either that both drugs were effective or that neither was effective.

That is the failure mode of a comparison without a placebo arm. A trial that finds no difference between two treatments has not, by that fact alone, shown that both worked. It may have shown that the trial could not tell anything apart. The regulation adds that the analysis should explain why the drugs should be considered effective in the study, for example by reference to results in previous placebo-controlled studies of the active control drug.

The historical-control entry carries its own limit. Because historical control populations usually cannot be as well assessed with respect to pertinent variables as concurrent control populations, the regulation says, historical control designs are usually reserved for special circumstances. It gives examples: diseases with high and predictable mortality, such as certain malignancies, and studies in which the effect of the drug is self-evident, such as general anesthetics.

The words "usually reserved for special circumstances" are doing real work. A comparison against past experience is recognized, and it is the weakest of the five for a reason the regulation states plainly — the comparison group cannot be characterized as carefully as one enrolled alongside.

The control is one of seven characteristics, not the whole test

Choosing a control is one item on a longer list. Section 314.126(b) sets out seven characteristics of an adequate and well-controlled study, and the other six are about everything around the comparison.

There must be a clear statement of the objectives and a summary of the proposed or actual methods of analysis, in the protocol and in the report. The method of selecting subjects must give adequate assurance that they have the condition being studied. The method of assigning patients to treatment and control groups must minimize bias and be intended to assure comparability with respect to pertinent variables such as age, sex, severity of disease, duration of disease, and use of other drugs — and the regulation says that ordinarily, in a concurrently controlled study, assignment is by randomization, with or without stratification.

Adequate measures must be taken to minimize bias on the part of subjects, observers and analysts, and the protocol and report should describe the procedures used, such as blinding. The methods of assessing subjects' response must be well-defined and reliable. And there must be an analysis adequate to assess the effects of the drug, which should assess among other things the comparability of test and control groups and the effects of any interim data analyses performed.

One sentence at the end of the section constrains the drug itself rather than the design: for an investigation to be considered adequate for approval, the test drug must be standardized as to identity, strength, quality, purity and dosage form, to give significance to the results.

What this means when a page cites a trial

The practical use of this list is that it turns a vague question — was this a good study — into a specific one. What was the comparison group, and was it enrolled at the same time as the treated group.

For the drug class this site covers, FDA's draft guidance on developing weight-reduction drugs states a design expectation directly. In general, it says, phase 3 clinical trials examining the efficacy and safety of weight-reduction drugs should be randomized, double-blind, and placebo-controlled, with the investigational drug used as an add-on to standardized recommendations for diet and physical activity in all randomized subjects. The last clause is worth reading twice: in that design both arms receive the lifestyle program, so the comparison is a drug against a placebo on top of the same background, not a drug against nothing.

Three reading habits follow. First, a percentage from a single arm with nothing to compare it to is not a controlled result, whatever the size of the trial. Second, a claim that one drug matched another needs the trial's own assessment of whether it could have detected a difference, because the regulation says similarity can mean neither drug worked. Third, a comparison against historical experience is a recognized design with a stated limitation, so a page relying on one should say that is what it is.