You are here
July 2, 2019
Understanding Study Size
By Aaron Cypess, M.D., Ph.D., M.M.Sc.
Acting Section Chief, Translational Physiology Section, Diabetes, Endocrinology, and Obesity Branch, National Institute of Diabetes and Digestive and Kidney Diseases
It is a common misconception that the larger a clinical trial, the better the study is and the more important the results. That perspective is partly understandable, in that many of the studies published in the highest-impact journals involve thousands of patients, and the definitive clinical trials required by the FDA for drug approval involve multiple centers throughout the world. However, what is not appreciated about clinical research is that every trial, from proof-of-concept to Phase 3 drug study, is in fact the smallest it can be.
When you’re evaluating a study, the first step is to consider what type of study it is. Clinical studies can be classified into two distinct categories: observational studies and clinical trials. The former is more for information gathering and the generation of hypotheses. The size is not usually prespecified. Rather, it is determined more by the number of patients or medical records available. Generally speaking, the larger the number of people studied, the better the chance to detect small but meaningful relationships. There are many variables that are not controlled, so the conclusions must be regarded as preliminary—exciting, perhaps, but only correlations, not proofs of causation.
Clinical trials are entirely different. They are designed to confirm and expand a hypothesis, such as the effectiveness of a drug or how a specific process in the body works. Clinical trials are prospective—that is, the assignments of people to groups or interventions is predetermined. Prospective trials are considered the optimal scientific way to prove something about the human body or behavior.
If bigger is better, then why should a clinical trial be small? Two obvious factors are money and time. However, the most important reason to make a clinical trial as small as possible is ethical: risks to patients must be minimized. The best way to reduce risks is to study as few patients as possible. To achieve this goal, prospective clinical trials undergo rigorous vetting in the design stage. The critical decision by the principal investigator is what will be the prespecified endpoints. There may be just one—the primary endpoint—or more. So the most important question to ask when reviewing a clinical trial is determining what those endpoints were, then determining what the investigators found about them.
If a well-designed clinical trial is small, then it means that the anticipated effect had to be large and often conceptually groundbreaking. An excellent example is the history of statins. We all know of statins as the principal way to lower LDL cholesterol and reduce the risks of cardiovascular disease and stroke. However, most people are unaware that the first proof-of-concept study showing the ability of statins to achieve their effects was demonstrated in seven patients with familial hypercholesterolemia.
Another example is in my field of human brown adipose tissue (BAT) research. When I started studying BAT in 2004, I was told that there was no BAT in adult humans. That was too bad, I thought, because in animal models, BAT activation had the ability to consume fat and glucose at high rates and improve obesity and insulin sensitivity. It turns out the conventional wisdom was wrong, and today it is universally accepted that nearly all humans have some BAT. How did this revolution happen? The critical proof-of-concept studies to show that BAT was functional in adult humans and could respond to drug treatment were accomplished in studies with as few as five to twelve healthy volunteers. (See Virtanen et al NEJM 2009, Cypess et al. Cell Metab 2015, and Broeders et al. Cell Metab 2015.)
An often-overlooked way of reducing patient risk is through increasing the precision of the clinical trial data. In this area, the NIH Clinical Center has a unique gem: The Metabolic Clinical Research Unit. This unit conducts rigorous studies that require strict control over diet, physical activity, and environmental temperature that cannot be replicated elsewhere.
A product of that attention to detail enabled Kevin Hall’s group in the NIDDK to study inpatient adults who were exposed to ultraprocessed versus unprocessed diets for 14 days each, in random order. They made the discovery that an ultraprocessed diet caused increased food intake and weight gain despite being matched in calories, macronutrients, and other factors. This important study involved only 20 healthy volunteers. Small clinical trial, global impact.
What are the takeaway lessons about clinical trial size?
- Don’t first consider the size but the design—was it an observational study or a clinical trial? Clinical trials of any size can prove a point better than an observational study.
- When looking at a clinical trial, don’t fixate on the size, given that each one is as small as it can be. Rather, focus on the prespecified endpoints, particularly the primary one. Ask what the investigators expected to see and whether the results confirm their hypotheses.
- Consider that larger clinical trials are often meant to confirm something important to change our practice. The small trials are the ones making conceptual breakthroughs that change our thinking.