Overview of study designs

A study design is a scientific method that a

researcher follows to assess the association
between an exposure and an outcome
It is the method based on which subjects
included in the study are selected, observed,
followed and studied.
In clinical research, there are two broad
categories of study designs
observational and experimental
 Clinical studies
Descriptive studies
Descriptive studies report unusual or new events
such as the occurrence of sudden infant death
syndrome (SIDS) in several siblings within a
single family, prevalence of albinism in a single
family etc
The researcher simply records the observations
and co-relates the events observed with possible
reason.
 Descriptive studies

It is where the main objective of

the research is to describe the main
features of the population
These are neither randomized nor
pre-designed researches
They may be presented as case
reports/ case series
 Case report

It is a report that documents unusual medical

occurrences that can represent the first clue in
the identification of new disease or adverse
effect of exposures (based on one patient).

It includes a summary of the disease, such as the

presentation, signs, symptoms, diagnostic
studies, treatment course and outcome
 Case reports
case reports whereby certain individual patients
with distinguished clinical characteristics are
included in the study.
All the baseline characteristics are recorded and
the individual patient is treated as unique case
with control over all the variables.
The patient is observed and evaluated for the
possible outcome.
 case reports
The results are compared with baseline values or
are expressed as success or failure of the
treatment given.
If the treatment succeeded, a hypothesis is
generated for an expanded and more rigorous
study to find the relationship between the
treatment and the outcome observed
 Case series

It is a collection of different case

reports, thus based on more
than one patient
 Case series
In case series, observations are documented at
regular intervals from patients exposed to a
particular drug or a group of drugs.
They may also cover prior histories of patients
with the same outcome, to find a possible cause-
effect relationship if exists.
These are useful in predicting the incidence of an
adverse event of newly-marketed drug when
reports on such events are limited.
 Exploratory studies
Observation studies
In an observational study, the subject to be
observed chooses whether or not to take the drug
or to have the surgery being studied

Errors that are likely to occur include the

differences in profile of the subjects since
variables such as age, family history of disease,
cause and severity of disease etc. may not be
defined
 Observation studies

For example, two patients have left ventricular (LV)

dysfunction, in one it is because of ischemic heart
disease (IHD) and in another it is because of severe
mitral valve stenosis.

Thus, the therapy of both the diseases differs due to

different oetioes and hence both the patients can
not be compared in one study
 Observation studies

Another example is of two patients suffering from

headache, one because of migraine and the other
because of common cold.

These two patients can not be compared for the

analgesic activity of one drug since the cause and the
severity of headache and hence the analgesic activity of
the drug would vary greatly.

Observational studies can never be blinded. Hence,

biases from patients, observer and experimenter may
result into systematic and random errors.
 Aggregate observation studies

Pandemic and epidemic studies on

communicable diseases and their
treatments are generally carried out as
aggregate observation studies
e.g. occurrence and effective treatment of
malaria and its relapse in particular
geographical area
 Individual observation studies

In individual observational study, the patients/

subjects are individually observed and they are
assembled in groups on the basis of outcome or
exposure or both.
Depending upon the basis of the grouping, the
individual observational study is sub-classified as
1) Case-control
2) Cohort
3) Cross sectional
 Case-control study
Case-control study involves assembling of
subjects in groups on the basis of the outcome
found in those subjects.
It compares the subjects with outcome in
question (the group behaves as a case group)
with the subjects without the outcome (the
group acts as a control) e.g. occurrence or non
occurrence of myocardial infarction (MI) in
patients with hypertension (HT).
 Definition
case-control study is an epidemiological study
where participants are selected based on their
disease status.

Two groups of subjects are included in this type of

study, cases (having the disease under
consideration) and controls (not having the
disease).
 Definition
Accordingly, both cases‟ and controls‟ history of
the exposure and other risk factors is checked.

Exposure history is compared between the cases

and controls to assess whether an association
exists between the exposure and the outcome.
It generally follows the retrospective design and
evaluates how the exposure is related to the
well- defined outcome using control group.

However, grouping on the basis of outcome

incorporates subjects with variety of
distinguished characteristics.
It is quick and inexpensive
Further, patients with rare outcome can be assembled in a
group to study oetioes, pathophysioes and prognosis of a
disease.

Results are generally expressed in terms of odds ratio (OR)

and risk ratio/ relative risk (RR).

Although multiple exposure variables can be correlated

with outcome, it does not allow the correlation of
temporal sequence of cause and effect with the final
outcome
Case control studies are also called retrospective
studies as they assess exposure in the past, and
since the direction of the study is backwards.

Both exposure and outcome would have already

occurred in a case-control study
For example, a study was carried out to identify
the prominent maternal and neonatal risk factors
associated with early-onset group B streptococcus
(EOGBS) disease in neonates.
The authors carried out a case-control study
where cases were infants < 7 days of age with
invasive group B streptococcus (GBS) disease and
controls were healthy infants born in the same
hospital during the same period having the same
birth weight and gestational age category
 Identifying cases
Case ascertainment can either be retrospective
(prevalent cases) or concurrent (incident cases).
Incident cases are those derived from ongoing-
ascertainment of cases over time, whereas
prevalent cases are derived from a cross-
sectional survey.
One very important issue in selection of the
cases is to ensure the representativeness of the
selected sample.
Ideally, cases should represent a random sample
of all cases of interest in the source population
(e.g. from vital data, registry data).
More commonly cases are a selection of
available cases from a medical care facility (e.g.
from hospitals, clinics).
To insure comparability between the cases and
controls, different types of controls have been used
in different case-control studies, and these include:
hospital controls (those admitted to the same
hospital for reasons not related to the disease
under study), neighbours (door to door, phone,
etc.), friends or associates of cases, or population-
based controls (a members of a population of a
defined area).
The selection of one type of controls over the other
depends on factors such as feasibility,
representativeness, convenience, logistics, cost, etc.
The number of controls selected for each case
varies between different studies.
It is well established that as the ratio of controls
to cases increase, the efficiency of the study will
increase, which will reach a plateau at 4 controls
per case.
Thus, most case-control studies include between
2 to 4 controls per each case
 Exposure assessment
Once the cases and controls are identified,
information on past exposure history, as well as
information on other risk factors are collected
through different means. Most frequently used
method of exposure collection is through
questionnaires, where the participants are
interviewed for detailed recollection of their past
exposures. A more standardized way of data
collection is using already existing databases,
such as drug registry
 Identifying controls
If cases are a random sample of cases in the
population, then controls should be a random
sample of all non-cases in the population
sampled at the same time. One general rule is
that controls should be at risk of the disease. The
controls should resemble the cases in all aspects
except for the presence of disease (and any as yet
undiscovered risk factors for disease).
 Matching

To ensure comparability between cases and

controls in terms of strong confounding variables,
control selection is usually done through
matching.
This indicates the process of selecting controls so
that they are similar to the cases in regard to
certain characteristics. The three most commonly
used matching criteria are usually age, gender,
and time
In the paper summarized earlier, cases and
controls were matched on time, weight and
gestational age, where controls had to fall in the
same category as the cases.
One of the disadvantages of matching is that
matching on many variables may make it difficult
or impossible to find an appropriate control.
Moreover, once a variable has been used to
match on, it cannot be used to explore possible
association of disease with it
 Analysis
B streptococcal
infection
Cases Control Total
s

Fetal Yes 22 7 29
tachycardia

No 77 193 270

Total 99 200 299

 Association between fetal tachycardia and B streptococcal
infection

The calculation of the prevalence rate ratio or the

risk ratio will yield wrong estimates, since any of
those measures will be dependent on the number
of controls per case recruited into the study, which
is decided upon by the investigator. Thus, an
alternative measure is calculated which is specific
to case-control studies, and is called the Odds
Ratio (OR).
The OR is calculated by dividing the odds of exposure
among the cases over the odds of exposure among the
controls. To illustrate the concept of the OR, the following
is a hypothetical table

Cases Controls Total

Exposure Yes a b a+b

No c d C+ d

Total a +c b+d a + b +C+ d

The definition of odds is a ratio between two numbers
The odds of exposure among cases = a / c
The odds of exposure among controls = b / d
Thus, the OR will be calculated by dividing the two odds
which will be equal to (a * d) / (b * c)

An OR can approximate the relative risk in instances where

the disease prevalence is low (< 10%). The OR is interpreted
in the same manner as the relative risk with an OR = 1.0
indicating no association, an OR > 1.0 indicating a positive
association, and an OR < 1.0 indicating a negative, or
protective association. In the above example, the OR will
be 7.88, which indicates a positive association between
fetal tachycardia and B streptococcal infection.
 The few strengths of case-control studies are:

a. Being fast to be carried out, as no follow-up is needed, and

at the time of conducting the study, both exposure and
outcome have already took place.
b. It involves low financial cost, again due to the fact that all
the data is collected in a relatively short period of time, and
the absence of follow-up.
c. Appropriate for studying rare diseases, since patients are
selected based on the outcome status, which gives the
opportunity of recruiting an adequate sample of the cases
from a specialized health care facility.
d. Appropriate for assessing the association between
different risk factors and one outcome.
 As for the limitations of case-control studies, they include the following :

a. Prone to selection bias, if the cases or controls are not

representative of the underlying population.
b. Prone to information bias, since information collected
has happened long in the past, and it might not be
accurate due to recall (people with a condition will be
more motivated to recall details about past exposures).
c. Affected by confounding, as it might not be feasible to
collect information on a wide range of confounding
variables.
d. Not appropriate for studying rare exposures, since not
enough subjects (whether cases or controls) will be found
to be exposed.