Adherence to inclusion and exclusion criteria is essential to the successful execution of a clinical trial. Deviations from these criteria must be avoided because they can jeopardize scientific integrity, regulatory acceptability, or the safety of subjects in the study. This presentation provides suggestions and advice on formulating inclusion and exclusion criteria to enhance the quality of subject selection, minimize protocol violations, and avoid protocol amendments.

1. Improving Inclusion/Exclusion
Criteria in Clinical Trials

2. JackModell,MD,Senior
MedicalOfficer
• Dr. Modell is a board-certified psychiatrist with 30 years of
experience in clinical research, teaching, and patient care
including 10 years of experience in clinical drug development
(phases 2 through 4), medical affairs, successful NDA filings,
medical governance, drug safety, compliance, and
management in the pharmaceutical industry.
• He has authored over 50 peer-reviewed publications in
addiction medicine, anesthesiology, psychiatry, neurology,
and nuclear medicine. Dr. Modell is a key opinion leader in
the neurosciences, has served on numerous advisory and
editorial boards, and is nationally known for leading the first
successful development of preventative pharmacotherapy
for the depressive episodes of seasonal affective disorder.

3. Purpose
• To provide recommendations to clinical teams
on the approach to formulating inclusion and
exclusion criteria in clinical trials:
– to enhance the quality of subject selection
– to simplify and expedite the enrolment process
– to minimize protocol violations and amendments
to protocols

4. General Principles
Inclusion/exclusion criteria define the
study population to which results of the
clinical trial can reasonably be
generalized.
• Population may or may not be a good
representation of the entire population
with the disease or condition of concern.
Deviations from inclusion/exclusion
criteria (“waivers”) are not allowed
because they potentially jeopardize the
scientific integrity of the study, regulatory
acceptability, and subject safety.
• Therefore, adherence to the criteria as
specified in the protocol is essential.

5. General Principles (2)
• Inclusion criteria contain only those items that
define the target population desired for study.
– Within limits of known safety coverage, should
generally allow inclusion of the broadest
representation of the population for intended use,
but usually not beyond.
– For each and every inclusion criterion written, the
study team should have justification for why that
criterion, exactly as written, is necessary to
define the desired study population.

6. General Principles (3)
• Exclusion criteria shape the desired,
“included” population by defining which
subsets need to be excluded for other
reasons, such as safety.
– For every exclusion criterion written, the study team
should have justification for why that criterion,
exactly as written, is necessary to prevent subjects,
in some cases who may otherwise be qualified,
from participating in the study.

7. General Principles (4)
• Inclusion and exclusion criteria need to be
written as clearly and unambiguously as
possible.
– There should also be considerable forethought
into the
• necessity,
• desirability, &
• potential unintended consequences
of each criterion and any associated parameters.

8. General Principles (5)
• Care should be taken to write the fewest
number of inclusion and exclusion criteria
sufficient to describe the population to be
studied.
– Duplicative, overlapping, and potentially
contradictory criteria should be avoided.
Challenge yourself to determine whether all of the
inclusion criteria are truly required to define the
target population, and all of the exclusion criteria are
truly required for safety or other limitations to the
target population.

9. General Principles (6)
Inclusion/exclusion criteria should
not be copied indiscriminately from
one protocol to the next.
• Criteria should be selected anew for each
protocol, though previously written criteria
can be used for reference.
The same applies to labs & other
clinical exams.
• “I have no idea how to interpret it or why
it’s there – Oh, wait, it was in the last
protocol we did!”

10. General Principles (7)
• Think about inclusion/exclusion criteria in
terms of where you want to draw the line
between sensitivity and the “false positive
exclusion” rate;
– there is usually a reciprocal relationship between
the two.
False
positive
rate
Detection sensitivity

11. Example: No hypertensive
subjects please!
• Approach #1: Exclude all subjects with a
systolic blood pressure > 120 mmHg
– This will likely exclude nearly all subjects with at
least moderate hypertension.
• The criterion provides extremely high sensitivity for
detecting hypertension.
– It will also inappropriately exclude a very high
percentage of subjects who do not have
hypertension.

12. No hypertensive subjects please
(but don’t falsely exclude any subjects
either)
• Approach #2: Exclude all subjects with a
systolic pressure of > 180 mmHg
– A small proportion of all patients with
hypertension would be excluded.
• The criterion provides extremely low sensitivity for
detecting hypertension.
– But no non-hypertensive subjects would be
excluded

13. No hypertensive subjects
please! (3)
• No matter where a criterion is set, there will
be a trade-off between subjects who are
unnecessarily excluded and those who
probably should be excluded but do not meet
the specified criterion.
– It is up to the study team to consider this for each
criterion selected and to decide how best to
optimize this inherent trade off.
 there is no absolute or perfect solution to this
problem.

14. Investigator Education
Study teams should be diligent in reviewing I/E criteria with
investigators to help ensure that:
•the protocol-specified details are completely understood
•investigators understand that inclusion and exclusion criteria must not
be violated
To aid in understanding the I/E
criteria, explanatory footnotes in
the protocol should be
considered for any criteria for
which the rationale may not be
evident

15. Phase of Development
• Early in development when limited safety
information is known, more conservative
inclusion/exclusion criteria may be needed to ensure
patient safety.
• Once the safety profile begins to be established, less
restrictive criteria can usually be applied.
– Allows broader inclusion of patients into late-stage trials,
which provides more “real-world use” safety profile.
– Helps define appropriate labeling for this target
population.

16. Target Population & Individual
Subject Characteristics
• Generally healthy subjects can have physiologic or
laboratory values outside of the defined “normal
range,” yet still be very healthy;
– for example an athlete or runner with a resting heart rate
of 45.
• Yet in a cardiac patient, bradycardia often signals
significant underlying pathology or drug effect.
– The study team (and protocol) need to be able to
differentiate between normal physiology that may produce
a testing abnormality and a truly pathologic condition.

17. Disease Area
• Exclusionary values/ranges need to be adapted to
the population under study
– In some diseases, lab values that would be considered
abnormal in healthy populations may be acceptable, such
as low hemoglobin in patients with chronic disease; or
rapid heart rate & hypertension in anxiety.
– If there is intent to exclude a specific medical problem, for
example, serious cardiac arrhythmia, the exclusion
criterion needs to refer specifically to the excluded rhythm
– and not, in the case of this example, to something more
general like “cardiac arrhythmia,” which could include
several benign rhythm disturbances.

18. Demographics and “Standard
I/E Criteria”
• Conventional demographics or labs such as age, BMI,
renal function, etc. should not be routinely included
in I/E criteria without any consideration.
– Do you actually need BMI, age, or the specific laboratory
criteria? If so, for what reasons?
– Could the criteria instead be written with regard to co-
morbid conditions; e.g., a healthy 80 year-old versus a 50
year-old in poor condition?
– Can a more general criterion be used e.g. “no major illness
or debility that in the investigator’s opinion is likely to
jeopardize subject safety or interfere with the subject’s
ability to comply with study requirements”
– but…

19. Flexible vs. Fixed
Inclusion/Exclusion Criteria
– The use of more flexible criteria requires strong
assurance that these decisions can and will be
properly made, and that variability or errors in
investigator judgment are not likely to skew the
intended population or jeopardize patient safety.

20. “…that these decisions can and will
be properly made…”
“I’m a psychiatrist: how was I to know the subject
had pneumonia?”
“I knew his liver enzymes were elevated but I’ve
seen much worse and he really wanted to be in
the study.”
“The rating forms didn’t show it, but in my clinical
opinion he had it, so I put him in the study.”
“It said, ‘no current major medical illnesses,’ and
her colon resection [for cancer] was seven months
ago.”
“Beta thalassemia is a major illness?”
Actual quotes from investigators

21. Fixed Exclusion Criteria
• It is also permissible, and often desirable, to use
fixed (“rigid”) exclusion criteria, e.g.,
“serum Na < 130 or > 150”; “systolic BP < 60 or >
160”
• If there is clinical justification for why it is desirable
to exclude these subjects and the values chosen are
intended to be unequivocal.
– The advantage: It sets absolute standards for subject
participation that are not open for opinion or debate.
– The disadvantage: If the criteria are too stringent, they
may unnecessarily exclude some subjects who are not at a
particular risk that the criteria are intended to prevent.

22. Studies of Long Duration
• Many studies are now requiring data
collection over many years.
• Since guidelines and prescribing information
change over time, this should be taken into
account by using wording such as,
“contraindications as per the prescribing
information,” to allow label and guideline
changes to be incorporated without
modifications to the protocol.

23. Regulatory & Ethical
Considerations
• In later stage development, I/E criteria may have
regulatory implications.
– The population under study may affect product labeling;
this needs to be considered carefully.
– Regardless of potential labeling considerations, I/E criteria
are always written to optimize the benefit-risk for patients.
• Prospective subjects should never be “groomed” to
meet I/E criteria for the primary purpose of study
participation prior to obtaining informed consent.

24. Operational Issues
• Subjects are sometimes inappropriately
randomized before investigators receive all
information relevant to I/E criteria, often due
to delayed lab results.
• Check-steps should be considered in the
programming of the randomization such that
randomization cannot occur until all I/E
criteria have been entered and noted to be
consistent with protocol specifications.

25. Discussion
• Using no more than four inclusion and six
exclusion criteria*, define a study population
for late phase II testing of:
– a drug for depression (n = 150 per arm),
– which in a healthy volunteer study was well
tolerated,
– and although prolonged QTc interval was observed
in 4 of 30 on drug compared to 2 of 30 on placebo
(p ≈ .7), no dysrhythmias were observed
*A table of lab values – if you’d like to use one – can count as one criterion;
just have a few examples of any values you think might be particularly important.

26. Summary
Inclusion/Exclusion Criteria:
 define the study population to which results of the
clinical trial can be generalized – waivers not appropriate
 need to be written clearly and unambiguously, with
consideration of “unintended consequences”
 must be highly study specific; avoid “copying”
 establish the line between sensitivity and “false positive
exclusion” rate
 that are “flexible” often lead to “flexible populations”
and “flexible results”
 may have regulatory implications
 must always written to optimize the benefit-risk
for patients