Home 
Contact 
Profile 
Videos 
Presentation at a workshop on Research Methodology in Health Sciences held at the Northern Area Armed Forces Hospital on 28 December 2020 at 7:10 pm by Dr Omar Hasan Kasule Sr MB ChB (MUK), MPH (Harvard), DrPH (Harvard)
DEFINITION OF
EXPERIMENTAL STUDIES:
} Experimental studies involve
deliberate human action or intervention whose outcome is then observed. Their
objective is to establish a definitive causal relation.
} Examples of experimental
studies involving humans: clinical therapeutic trials, community intervention
studies to assess vaccines, community intervention to assess preventive measures,
etc
} Main characteristics of an
experimental design are a random selection, random assignment, manipulation of
independent variables by the experimenter, and control of all other variables
by the experimenter without manipulating the outcome (dependent variable).
} Experimental studies can be natural
experiments or true experiments
NATURAL EXPERIMENTS
}
Natural experiments are not
designed by humans. Humans only observe and benefit from their results
}
Examples of natural
experiments
o
Snow's study of cholera in London: he found cholera to
be related to the source of the water supply
o
In the 1950s air pollution episodes led to
mortality in London and Los Angeles
o
Increased death and leukemia in atomic bomb survivors
in Hiroshima and Nagasaki.
TRUE EXPERIMENTS:
} True experiments are more
involved and more expensive than observational studies or natural experiments
but are still? cheaper and ? easier than laboratory-based research.
} Classical examples of true
experiments before world war 2 were not as complicated as today’s experiments
but were run on similar basic principles
} In 1754 Dr. James Lind published the results of his trial of citrus fruit juice to prevent scurvy among sailors. He
studied 12 sailors who were given a standard diet. They were allocated to 6
treatment groups that received various treatments for a period of 14 days
} 1796 Edward Jenner's cowpox
vaccination to prevent childhood disease. Inserted pus from a cowpox pustule in
the arm of 8 year James Phipps who subsequently became immune to smallpox
(copied from Ottomans by Lady Montagu wife of the British Ambassador in
Istanbul)
} In 1886 Cuban epidemiologist Juan
Carlos Finlay published his experiments on the transmission of yellow fever to
humans by the mosquito. He started by recruiting 20 volunteers, 5 were
experimental and 15 were controls. He used mosquitoes that had bitten yellow
fever to bite his subjects who subsequently developed yellow fever. He
continued the experiment with more volunteers. In 1900 US Army Surgeon Dr
Walter Reed confirmed Finlay’s findings
} 1915 Joseph Goldberger using
dietary intervention discovered the dietary cause of pellagra. In 1937 niacin
deficiency was identified as the cause of pellagra.
} 1905 William Fletcher in
Kuala Lumpur carried out an experiment that proved that beri beri was more
in patients fed white rice than in those fed brown rice. His work was based on
earlier research in 1889 by Christian Eijkman who had found beri beri symptoms
in chickens fed white rice but the symptoms disappeared when they were fed
brown rice.
STRENGTHS OF
EXPERIMENTAL STUDIES:
} The main strength of
experimental studies is a good control for confounding.
} Randomization in
experimental studies is the basis for unequivocal evidence of causality.
} The experimental design
enables the investigator to control extraneous variables.
} The experimental design
enables the investigator to vary the levels of independent variables in order
to make a more thorough and detailed study.
WEAKNESSES
OF EXPERIMENTAL STUDIES:
} The main weakness of
experimental studies is that well-controlled experiments on humans are
difficult to carry out and have ethical problems.
} It is difficult to put
humans under full experimental conditions where they can be observed for 24
hours.
} Ethical controversies and
violations of human rights always arise in such studies.
POPULATIONS:
} The target population is the
big population to which the results of the experiment will be applied.
} Study population: A study population suitable for
the problem being studied must be selected. It may represent the general
population or may be restricted to certain groups. Economic and logistic
considerations affect the selection of the study population.
} Sample are the actual people selected
from the study population for the experiment. A representative sample is not
necessary for the validity of an experimental study.
} Study groups: The researcher determines the
experimental or intervention group (who will get the exposure studied) and the
comparison group. The control group could get the exposure studied unknown to
the investigator. Even after randomization, the experimental and comparison
groups may not be comparable.
RANDOMIZATION: Definition and
objectives
} Randomization is the basic sampling
scheme for experimental studies and clinical trials. It is the main sampling
design used in experimental studies.
} Randomization involves the allocation
of subjects by independent random choice to 2 or more experimental groups.
} The objective of randomization is
to prevent investigator bias in the selection and allocation of subjects to the
treatment and reference groups, ensure comparability of results from the two
groups, and increase confidence in the results of the study.
} Randomization balances the
distribution of confounding as well as prognostic factors and assures the accuracy
of statistical tests
} Randomization is not successful
with small samples and does not always ensure correct conclusions.
RANDOMIZATION: Methods
} Alternate cases
} Sealed serially numbered envelopes
} Random numbers.
RANDOMIZATION: Enhancements
} Randomization has the objective of balancing
the distribution of confounding factors, known and unknown, across groups. This
purpose is not always achieved to 100% perfection.
} To obtain more balancing of
confounding randomized block design is used in which the randomization process
is carried out separately for each 'block'. The block could be a gender or age
group.
} Stratified randomization is akin to the block design of experimental studies.
RANDOMIZATION:
Parallel and crossover
} Once the 2 groups are identified,
the study may be carried out in parallel or it may use the cross-over design.
} In the parallel design the
experimental and comparison groups are followed until the end.
} In the crossover there is an
exchange. The experimental group becomes the comparison group during the trial
and vice versa. Thus subjects in both groups will take turns receiving the
material being tested.
RANDOMIZATION:
alternatives
} Where randomization and selection
of a good control group are not possible, quasi-experimental studies may be
carried out.
} Non-randomized trials can be
carried out where randomization proves difficult but they are far less valid
than randomized ones. They obtain comparison groups by using either
historical controls or by carrying out pre and post-test assessments.
ETHICAL
ISSUES IN CLINICAL TRIALS:
} Individual rights/welfare vs
community benefit: why should a patient take a risk to discover a new treatment
that will benefit future patients:
} Randomization: control/placebo
group does not benefit or the experimental group is exposed to danger.
} Informed consent and coercion in
the vulnerable
} Conflict of interest
} Confidentiality
} Publication malpractices: no
publication, publication bias
FURTHER READING:
} The Oxford Textbook of Clinical Research Ethics