Designing a Randomized Blinded Trial

European University
Giorgi Lomidze, MD, PhD
2024
Randomized Blinded Trial – key
points
• In clinical trials, the investigator applies an intervention
and observes the effect on outcomes;

• The major advantage of a trial over an observational

study is the ability to demonstrate causality;

• In particular, randomly assigning the intervention can

eliminate the influence of confounding variables, and
blinding its administration can eliminate the possibility
that the observed effects of the intervention are due to
differential use of other treatments in the treatment and
control groups or to biased ascertainment or
adjudication of the outcome;
Randomized Blinded Trial - design
Choice of Intervention
• The best balance between effectiveness and safety depends
on the condition being studied;

• Trials to test single interventions are generally much easier to

plan and implement than those testing combinations of
treatments;

• However, many medical conditions, such as HIV infection or

congestive heart failure, are treated with combinations of
drugs or therapies.

• Simple interventions are generally better than complicated

ones (patients are more likely to take a pill once a day than
two or three times).
Choice of Control

• The best control group receives no active

treatment in a way that can be blinded, which for
medications generally requires a placebo that is
indistinguishable from active treatment;

• The cleanest comparison between the

intervention and control groups occurs when
there are no cointerventions—medications,
therapies or behaviors (other than the study
intervention) that reduce the risk of developing
the outcome of interest.
Choice of Control

• Often it is not possible to withhold treatments

other than the study intervention;

➢ For example, in a trial of a new drug to reduce

the risk of myocardial infarction in persons with
known coronary heart disease (CHD), the
investigators cannot ethically prohibit or
discourage participants from taking medical
treatments that are indicated for persons with
known CHD, including aspirin, statins and beta-
blockers;
Choice of Control

• One solution is to give standard care drugs to

all participants in the trial; although this
approach reduces the event rate and therefore
increases the required sample size, it minimizes
the potential for differences in cointerventions
between the groups and tests whether the new
intervention improves outcome when given in
addition to standard care.
Choosing Outcome Measurements

• Outcomes measured as continuous variables,

such as quality of life, can generally be
studied with fewer subjects and shorter
follow-up times than rates of a dichotomous
clinical outcome, such as recurrence of
treated breast cancer.
Choosing Outcome Measurements
• Intermediate markers, such as bone density, are
measurements that are related to the clinical outcome;

• Trials that use intermediate outcomes can further our

understanding of pathophysiology and provide information
to design the best dose or frequency of treatment for use
in trials with clinical outcomes;

• The clinical relevance of trials with intermediate outcomes

depends in large part on how accurately changes in these
markers, especially changes that occur due to treatment,
represent changes in the risk or natural history of clinical
outcomes;
Choosing Outcome Measurements

• Intermediate markers can be considered surrogate

markers for the clinical outcome to the extent that
treatment induced changes in the marker consistently
predict how treatment changes the clinical outcome;

• A good surrogate measures changes in an

intermediate factor in the main pathway that
determines the clinical outcome.

• HIV viral load is a good surrogate marker because

treatments that reduce the viral load consistently
reduce morbidity and mortality in patients with HIV
infection;
Number of Outcome Variables

• It is often desirable to have several outcome

variables that measure different aspects of the
phenomena of interest.

• However, a single primary endpoint was

designated for the purpose of planning the
sample size and duration of the study and to
avoid the problems of interpreting tests of
multiple hypotheses;
Adverse Effects
• The investigator should include outcome measures that
will detect the occurrence of adverse effects that may
result from the intervention;

• Revealing whether the beneficial effects of an

intervention outweigh the adverse ones is a major goal of
most clinical trials, even those that test apparently
innocuous treatments like a health education program;

• Adverse effects may range from relatively minor

symptoms such as a mild or transient rash, to serious and
fatal complications;
Selecting The Participants

Define Entry Criteria

• Inclusion criteria should produce a sufficient
number of enrollees who have a high enough
rate of the primary outcome to achieve adequate
power to find an important effect on the
outcome;

• On the other hand, criteria should also maximize

the generalizability of findings from the trial and
ease of recruitment.
Selecting The Participants

• Exclusion criteria should be parsimonious

because unnecessary exclusions may diminish
the generalizability of the results, make it more
difficult to recruit the necessary number of
participants, and increase the complexity and
cost of recruitment.
Design an Adequate Sample Size and
Plan the Recruitment Accordingly
• Estimating the sample size is one of the most important early
parts of planning a trial;

• Outcome rates in clinical trials are commonly lower than

estimated, primarily due to screening and volunteer bias;

• Recruitment for a trial is usually more difficult than

recruitment for an observational study;

• For these reasons, the investigator should plan an adequate

sample from a large accessible population, and enough time
and money to get the desired sample size when (as usually
happens) the barriers to doing so turn out to be greater than
expected.
Randomizing and Blinding

• The random allocation of participants to one

or another of the study groups establishes the
basis for testing the statistical significance of
differences;

• Randomization, which eliminates bias due to

baseline confounding variables, should be
tamperproof; matched pair randomization is
an excellent design when feasible.
Randomizing and Blinding
• Whenever possible, the investigator should
design the interventions in such a fashion that
the study participants, staff who have contact
with them, persons making laboratory
measurements, and those adjudicating outcomes
have no knowledge of the study group
assignment;

• When it is not possible to blind all of these

individuals, it is highly desirable to blind as many
as possible (always, for example, blinding
laboratory personnel);
Randomizing and Blinding

• In a randomized trial, blinding is as important

as randomization: it prevents bias due to use
of co-interventions and biased ascertainment
of outcomes.