The phrase clinical significance is used in various ways in drug, biotech, and device development. The goals of this chapter are to (a) define the term clinical significance and to differentiate it from statistical significance, (b) consider how to assess and use the concept of clinical significance when interpreting the results from a clinical trial and how to assess its relevance for medical practice, and (c) discuss under what circumstances a patient’s laboratory abnormalities should be assessed as clinically significant (i.e., whether this determination should be made before or after the determination is made that a patient’s laboratory abnormalities are in fact an adverse event). A more extensive discussion on the entire topic of clinical significance and its relationship to statistical significance and the relevance for both of these to medical practice is found in several other chapters within the Clinical Activities and Issues section of this book and in the book Guide to Clinical Trials (Spilker 1991).

The statistical significance of data is calculated as part of the data analyses done after the data are entered into computers and the tables and listings are created that are specified in the Statistical Analysis Plan. The magnitude of effect observed for the primary endpoint in the treatment group(s) compared with the trial’s control group that will be defined as clinically significant is generally prespecified in the protocol. The effect size is
used to help the statistician determine the number of subjects to enroll. However, in some cases, the clinical significance of the data is determined as part of the clinical interpretation of data, which is made after the statistical analyses have been completed (at least in draft form). One important principle is that statistical data analyses and clinical interpretation of the data are two separate processes.

There are multiple definitions of clinical significance relating to the effects of a specific drug or device. The most appropriate definition depends on several factors, including the perspectives from which the data are judged (e.g., company staff, regulators, physicians, or patients). The definition may also depend on whether one is dealing with a specific patient or the entire patient population in a clinical trial (or those under treatment in the general population for an approved product). Clinical significance can be applied to efficacy, as assessed with a specific parameter, overall drug benefits, one or more adverse events, or overall risk.

Clinical significance is most often used to refer to the primary parameter or endpoint used to assess efficacy, and is defined as the amount of change of that endpoint versus placebo, an active control, or other control that would encourage physicians to use (or try) the drug or device in their patients. Another definition of clinical significance that is sometimes mentioned is the level of effect that a regulatory agency requires to approve a new therapy; however, in case of a conflict between the regulatory view and practicing physician view, the definition should be based on the amount of effect required for physicians to try (or use) the product in their patients.

The phrases clinically important and clinically meaningful are defined the same as clinically significant, but they are not used in this text because the phrase clinically significant emphasizes the relationship with the phrase statistically significant. The phrase clinical benefit is vague because it can refer to almost any magnitude of effect that is considered positive.

Clinically significant efficacy for a single patient in a trial or for all patients in a trial can be considered as the magnitude or extent of change in a specified endpoint that would define those patients as “responders.” The level of change that is used to define responders is ideally one that would be agreed to by practicing physicians. Trial sponsors must be careful not set the standards for defining a clinically significant response unreasonably high or low.

Clinical significance from the perspective of the patient himself or herself depends in large measure on subjective factors that relate to the amount of improvement in physical and/or psychological status. The patient’s definition will also depend on whether any adverse events were experienced and how the patient assesses the overall impact of the drug on their well-being or quality of life. The patient’s perspective comes close to being one in which the patient will consider the clinical significance of the drug to be positive if the benefits outweigh the risks. There is apt to be much more variability in the concept and magnitude of clinical significance among individual patients than from any other perspective. For drug development, this is not the definition that will prove useful. However, the critically important question of who decides whether the benefits outweigh the risks is important to discuss and is mentioned in Chapter 77.

From the perspective of a practicing physician, the definition of clinical significance in terms of efficacy is the same as that used by a pharmaceutical company (i.e., the amount of change versus placebo or an active control that would encourage physicians to use, or at least to try, the new treatment in their patients).

Clinical significance in terms of safety for a patient in a clinical trial is discussed later in this chapter, specifically with respect to abnormal laboratory data.

Most protocols do not include the phrase clinical significance (or clinical importance), particularly in terms of identifying the magnitude of effect that will be considered as clinically significant. Nonetheless, the data are present in most protocols. This information is in the objectives in some cases (e.g., the trial is testing the hypothesis that the drug at a dose of X will decrease the primary clinical endpoint by 45%). More often, this information is found in the statistical section in the discussion on determining the number of subjects to enroll. The statistician will have discussed the magnitude of effect of the primary endpoint that the clinician believes must be observed for the trial to be successful. While there is no guarantee that the clinician has chosen what he or she believes to be the magnitude of effect that is clinically significant, this is what is most often done. Therefore, the statistical section will state that, in order to find an effect of 45% increase (or decrease) in the primary clinical endpoint with a power of X%, there will need to be Y subjects in the treated and control groups. Other ways of expressing the magnitude of effect may be presented, but this is the level believed to be clinically significant.

When judged in comparison to placebo, an active comparator, no treatment, or historical (or anecdotal) data, how do we know if the aggregate clinical trial results are “clinically significant?”

Usually, a positive result from a clinical trial would be based on demonstrating a statistically significant difference between the drug and placebo. However, whether the results of the study are also clinically significant depends in large measure on the magnitude of effect on the primary endpoint and whether that result is considered to represent an important change in the clinical condition. For example, if a drug is being tested to reduce a highly elevated heart rate in patients and it caused a three beat per minute decrease, then it might be a highly statistically significant result if a large number of measurements were taken; however, it would clearly not be clinically significant because of the small magnitude of the effect. Of course, even the absolute change in a parameter is not as clinically or statistically important as the difference between drug and placebo (or other comparator in a superiority trial). Thus, if a relatively large favorable effect was found in the placebo group, then the difference
between the results for placebo and drug must still be sufficiently large to achieve statistical significance in favor of the drug. Such results may or may not be clinically significant, depending on the magnitude of the absolute and relative changes, as well as on more qualitative aspects of the effects of the drug and the availability and performance of alternative therapies.

If an active comparator is used in the trial (without a placebo arm), it is important to know how great a difference between the active and test therapies is sufficient to say that the difference between the two is clinically significant. This assumes that one is conducting a superiority trial (i.e., a trial designed to show that the test therapy is better than the comparator). This requires not only a statistically significant difference between treatments, but also a clinically significant difference. If a trial is meant to demonstrate either equivalence of the two treatments or noninferiority (of one treatment in comparison to the other), then it is not necessary to determine the clinically significant magnitude of the difference in effect between the two drugs. In this situation, one is primarily concerned with the overall magnitude that must be achieved by the test drug in order to say that the effect it elicited is clinically significant. It is implicit that the effects shown by the comparator, a known and approved product, are acknowledged as clinically significant, although many studies have shown that the active control is not always sufficiently active in a trial or that the active control does not demonstrate a statistically significant improvement over that caused by placebo.