Chapter  7: Human Experimentation

Section 4. Reading

Title: Continuing Unethical Conduct of Underpowered Clinical Trials

Authors: Scott D. Halpern, Jason T. Karlawish, and Jesse A. Berlin

Publication Information: JAMA, Vol 288, no.3 (2002), pp.358-362.  


A long-standing understanding among the investigators is that the
statistically underpowered clinical trials are “scientifically useless”.
These Random Clinical Trials (RCTs) fail to test the underlying
hypothesis adequately. Hence, they expose individuals to risks and
burden of human testing through such unethical trials. Nevertheless,
RCTs are widespread. Some recent publications support such underpowered
RCTs. Authors of these publications argue that by combining the results
of similar small studies, the investigators can estimate the efficacy of
an intervention.  Secondly, even though the small studies may not test
the hypothesis, they still provide valuable estimates of treatment
effects using confidence intervals. Halpern, Karlawish, and Berlin
assess these arguments in the light of ethical and moral issues and
conclude that underpowered trials can be ethical only in the following

(1)   If the trials are for a rare disease and investigators document
explicit plans for combining the results with similar trails in a
prospective meta-analysis

(2)   Early-phase trials in the development of drugs, devices, or other
interventions can be underpowered  

The trio-authors also stress that the researchers must inform the
prospective participants that their participation may only indirectly
contribute to future health care benefits.  

In the world of null hypothesis testing, there are two types of errors –
type I and type II. Type I error occurs when the experiment data
incorrectly rejects the null hypothesis and type II happens when the
experiment data fails to reject a false null hypothesis. Investigators
would like to minimize both of such errors in their trials. They would
normally consider an experiment statistically powered if there is at
least 80% chance of detecting a clinically significant effect when one
exists. The number of individuals enrolled in a trial and variability of
their outcomes are the factors used by investigators to calculate the
power of a study.  RCTs lacking enough participants are hence
statistically underpowered.  

The authors of other publications supporting underpowered small studies
argue that prohibiting such studies means diminishing of knowledge.
Halpern, Karlawish and Berlin, however, find practical and ethical
problems with this argument. It is not practical to become too
optimistic about the usefulness of small studies. It is also not
practical to synthesize the results from various such studies all the
time. Furthermore, the underpowered trials are more likely to produce
negative results and consequently the investigators may not publish the
results. This practice makes the meta-analysis incomplete and useless.
Then there are ethical problems when for different reasons, the
investigators do not inform the participants about the limited value of
the underpowered studies. Normally people participate in such studies to
help the humanity. However, if they are not fully aware of the value of
the studies they are about to participate in, their ability to decide
autonomously diminishes.  Due to these practical and ethical issues, the
trio-authors do not approve underpowered studies across the board.  

The two exceptions when investigators can consider underpowered studies
are in the case of rare diseases and early-phase studies of experimental
intervention with certain restrictions. It is understandable that it is
difficult for investigators to have a large number of participants in
cases of rare diseases.  A mere argument that having some evidence is
better than none, is not enough for approving underpowered studies for
rare diseases. The investigators must explicitly plan to make the
results of a small trial available for inclusion in a prospective
meta-analysis. Only then, it provides sufficient assurance to the
participants that a study’s results will eventually contribute to
valuable or important knowledge.  The prospective meta-analysis has
innovation and organization that will enable the synthesizing of various
results. Similarly, plans for large, comparative trials of experimental
interventions can justify the conduct of small studies in earlier phases
of drug or device development. Each phase of trial must aim adequately
for the next phase and the investigators must inform the participants
that their participation will guide only the future studies.

Summary: Hita Gurung (QCC,2003)

Return to the Readings section>>> Readings.

© Copyright Philip A. Pecorino 2002. All Rights reserved.

Web Surfer's Caveat: These are class notes, intended to comment on readings and amplify class discussion. They should be read as such. They are not intended for publication or general distribution.

Return to:                 Table of Contents for the Online Textbook