This document discusses randomized trials, including their uses in evaluating new treatments and prevention methods. It defines key concepts like randomization, stratified randomization, and masking. Randomized trials randomly assign subjects to treatment groups to reduce bias and make groups comparable. Issues like noncompliance, sample size calculations, and ethical concerns regarding informed consent are also addressed.

1. Randomized Trials
Linda Ahdieh Grant
Epi 504D

2. Reading
ò Chapter 7: Assessing the efficacy of preventive
and therapeutic measures: Randomized trials.
ò Chapter 8: Randomized trials: Some further
issues.

3. Learning objectives
ò Understand the uses of randomized trials
ò Define the concept of randomization and its role on
randomized trials.
ò Know the main variables collected from subjects in a
randomized trial.
ò Distinguish crossover types in randomized trials
ò Recognize internal and external validity of a randomized
trial
ò Understand the principle of sample size and type I/II
error

4. Selected concepts
ò Randomized trial
ò Community-based trial
ò Randomization
ò Stratified randomization
ò Masking
ò Crossover
ò Noncompliance
ò Validity of a study
ò Sample size
ò Informed consent

5. Randomized trials
ò Terminology:
ò In clinical settings →
randomized clinical trials
ò Outside clinical settings →
randomized community-
based trials.
ò Uses
ò Evaluation of new approaches to treatment and
prevention of disease such as:
ò New drugs and other treatments for disease
ò New screening programs
ò New ways of organizing and delivering health
services

6. Randomized trials
An epidemiologic design where subjects are randomly allocated
into treatment groups.

7. Random allocation of
subjects
ò Advantages
ò Increases the likelihood that treatment groups are comparable
with regards to measured and unmeasured characteristics that
may affect prognosis
ò Eliminates subjective biases that may be introduced into the
process of selecting patients for one treatment group versus
the other
ò Concerns
ò Should always aim to compare new treatment with currently
recommended treatment
ò It is not always ethical to randomize subjects (this is the main
reason why most epidemiologic studies are observational in
nature)

8. Observational (I) versus
Experiemental (II) Studies
ž The presence of
arrhythmia affects
prognosis. Case-fatality is
higher in those with
arrhythmia
ž With randomization the
distribution of this
condition is similar in
comparison groups
ž Conclusions about the
effectiveness of the
intervention are not
explained by differences in
the prevalence of
arrhythmia

9. Stratified randomization
ò Randomization does not
guarantee the likelihood
that treatment groups are
fully comparable
ò One solution is to stratify by
each variable considered
important and THEN
randomize participants to
treatment groups
ò This increases the likelihood
of successful randomization

10. Data collection on
subjects
ò Need to know to which treatment group was
the subject assigned to
ò Outcomes must be measured in the same way
in all treatment and non-treatment groups
ò Check that randomization made treatment and
non-treatment groups similar regarding risk
factors for an outcome
ò à remember arrhythmia example?

11. Masking (Blinding)
ò Masking
ò Also known as blinding
ò Two approaches (can use one or both in the
same study)
ò Keep study subjects from knowing which therapy
they received →
use of a placebo
ò Keep the observer or data collectors unaware as
to which group a patient has been assigned

12. Crossover
ò Planned crossover
ò A method of comparing two or more treatments in
which the subjects, after being observed for some
time on one therapy, are switched to another therapy
and observed for a certain period of time.
ò Issues:
ò Carryover
ò Order in which therapies are given may elicit
psychological responses
ò Not possible if new therapy is surgical or cures the
disease

13. Crossover
ò Unplanned crossover
ò Poses a serious problem in analyzing the data
ò Analysis options
ò Analyze data according to intention to treat
ò Analyze data after breaking the randomization
ò If the number of crossover is large the meaning
of the results is questionable

14. Problems with randomized
trials
ò Noncompliance
ò Failure of patient to comply with the assigned
treatment
ò Reduces any observed differences among treatment
groups
ò People who do not comply or may not participate in
studies differ from those who do comply and who do
participate
ò Drop-ins
ò Patients in one group may inadvertently take the
agent assigned to the other group

15. Measures of disease
occurrence
ò Measures typically used in randomized trials
ò Incidence risk/rate
ò Mortality rates
15

16. Validity of a randomized trial
ò Internal validity – whether
the study was well done
and the findings
represents the true value
in the source population
ò External validity – ability
to generalize the findings
beyond the study
population itself

17. Sample Size
ò In order to make sure that a study gives
meaningful results, a sample size calculation
tells you how large of a sample you need with
the minimum amount of error
ò See Figures 8-4 and 8-5 for description of
ò Type I error
ò Type II error

18. Sample Size - Example
ò E.g., if doing a trial of on effect of statins on heart
disease, one bases sample size calculation on the
effect of statins, and not the effect of statins among
smokers
ò Type 1 error: study shows statins do prevent
heart disease, but in reality there is no
association
ò Type II error: study shows statins do not
prevent heart disease, but in reality there is an
association

19. Ethical conduct of research
ò In the U.S., research conducted, supported or otherwise
regulated by Federal institutions, must follow the policy
established for the protection of human subjects
ò It requires the establishment of an institutional review
board (IRB) in accord with and for the purposes
expressed in the policy
ò IRB approval means the determination of the IRB that
the research has been reviewed and may be conducted
at an institution within the constraints set forth by the
IRB and by other institutional and federal requirements
ò Without IRB approval it is not possible to obtain funding
to conduct a study
http://www.hhs.gov/ohrp/humansubjects/guidance/45cfr46.html

20. Ethical conduct of research
ò Informed consent
ò The U.S. code of Federal Regulations states that
no investigator may involve a human being as a
subject in research unless the investigator has
obtained the legally effective informed consent of
the subject or the subject's legally authorized
representative.