RESEARCH METHODOLOGY

Study designs
Shraddha Parab, Supriya Bhalerao1
Department of Clinical Pharmacology, Seth GS Medical College and KEM Hospital, Parel, Mumbai - 400 012, 1Department of Clinical
Pharmacology, TNMC and BYL Nair Hospital, Mumbai Central, Mumbai - 400 008, India

In the last issue, we discussed “sample size”, one of the crucial a linkage that a specific prakriti is a causative factor for
aspects when planning a clinical study. This article discusses hypertension. Types of descriptive studies are prevalence
another statistically important issue, Study designs. Study surveys, case series, surveillance data and analysis of routinely
design is a process wherein the trial methodology and statistical collected data, etc.
analysis are organized to ensure that the null hypothesis is
either accepted or rejected and the conclusions arrived at Case series and case reports
reflect the truth. A case report is a descriptive study of a single individual,
whereas case series is a study of a small group. In these studies,
The design of any study is more important than analyzing its the possibility of an association between an observed effect
results, as a poorly designed study can never be recovered, and a specific environmental exposure is studied based on
whereas a poorly analyzed study can be reanalyzed to reach detailed clinical evaluations and histories of the individual(s).
a meaningful conclusion.[1] Rather, the design of the study
They are most likely to be useful when the disease is
decides how the data generated can be best analyzed. The
uncommon and caused exclusively by a single kind of exposure
scientific integrity of the study and the credibility of the data
(e.g. vinyl chloride and angiosarcoma or diethylstilbestrol
from the study thus substantially depend on the study design.
(DES) and clear-cell carcinoma of the vagina).[2] Case reports
(or case series) may be first to provide clues in identifying a
The various aspects of clinical research can be broadly divided
new disease or adverse health effect from an exposure.
into two types, viz., observational and experimental. The
basic difference between these two types is that the earlier
does not involve any intervention (drug treatment/therapeutic ANALYTICAL STUDY
procedures/diagnostic tools), whereas in an experimental
study, the investigator administers an intervention to patients These studies are generally (although not always) used to
and the effect of this intervention on the course of events test one or more specific hypotheses, typically whether an
is documented. Let us see the different designs which are exposure is a risk factor for a disease or an intervention
commonly used to conduct these two types of researches. is effective in preventing or curing disease (or any other
occurrence or condition of interest). Of course, data obtained
in an analytic study can also be explored in a descriptive mode,
DESCRIPTIVE STUDY and data obtained in a descriptive study can be analyzed to
test hypotheses, making it analytical. In short, these studies are
This is the first foray into research. These studies describe designed to examine etiology and causal associations. Types
the frequency, natural history and determinants of a factor/ of analytical studies are cross sectional, case-control, cohort
disease. It is a study to identify patterns or trends in a situation, (retrospective and prospective) and ecological.
but not the cause and effect (causal) linkages among its
different elements, e.g. a study to assess the predominant Cross sectional
prakriti in hypertensive patients only helps in determining Cross-sectional study is also known as a prevalence study. It
the predominant prakriti in these patients, it does not establish measures the cause and effect at the same time, but does not
tell us the relationship, i.e. which one is the cause and which
Address for correspondence: one is the effect. This is the commonest study design used in
Dr. Supriya Bhalerao, Department of Clinical Pharmacology, TNMC general practice and research, in general. These studies are
and BYL Nair Hospital, Mumbai Central, Mumbai - 400 008, India.
E-mail: [email protected] relatively easy to do, inexpensive and can be carried out in a
short time frame.
Submission Date: 23-05-10 Accepted Date: 29-05-10
Case-control
DOI: 10.4103/0974-7788.64406
In studies using this particular design, patients who already

128 International Journal of Ayurveda Research | April-June 2010 | Vol 1 | Issue 2

 Parab and Bhalerao: Study designs

have a certain condition (cases) are compared (e.g. diabetic Cohort (Longitudinal studies)
patients with hospitalization) with people who do not have A cohort study begins with a group of subjects with some
that condition (controls) (e.g. diabetic patients without causative factor (e.g. daily intake of Virrudha ahar) but free of
hospitalization). The researcher goes through the past records the condition of interest (e.g. skin diseases). All the subjects are
of these subjects (both cases and controls) to find out whether followed up and observed for the occurrence of the condition
the development of the condition only in one group of patients of interest.
is due to presence of some causative factor (exposure). Thus,
in a typical case-control study, the data collection is mainly In contrast to the case-control study, a cohort study is usually
retrospective (backward in time) [Figure 1]. prospective (forward in time). It provides the best information
about the cause of disease plus the most direct measurement of the
These studies are less reliable than either randomized risk of developing a particular outcome due to exposure [Figure 2].
controlled trials or cohort studies. A major drawback to
case-control studies is that one cannot measure the risk of These studies, however, require a large number of subjects and
developing a particular outcome because of an exposure. a long period of follow up to assess whether the event of interest
Additionally, in these studies one has to mainly rely on the has occurred, due to which these studies are very expensive to
memory of patients to identify what in the past might have conduct. The main drawback of these studies due to long follow
caused their current disease, which is most often of long up is that there are high chances of subjects getting lost to follow
latency. This might induce a bias while analyzing the results, up.[7] If in the two groups, the degree of such losses is substantially
which is known as “recall bias”. Because human memory is different, it can lead to bias and false positive results.
frequently imprecise, recall bias is commonly believed to be
“pervasive in case-control studies”.[3] Correlational studies
These studies (sometimes called ecologic studies) explore the
The presence of disease affects both the patient's perception statistical connection between disease in different population
of the causes and his search for possible exposure to a groups and estimated exposures in groups rather than individuals.
hypothesized risk factor. Therefore, the recall of remote For example, they may correlate death rates by country with
exposures in case-control studies is commonly presumed estimates of exposure, such as factory emissions in a given
to be differential among study subjects depending on their geographic area, proximity to waste sites, or air or water pollution
disease status.[4] Data, even about irrelevant exposures, are levels. The geographical information system (GIS) is a very
often remembered better by cases or/and underreported by useful new tool that improves the ability of ecologic studies to
controls.[5] This trend in exposure recall tends to inflate the be able to determine a link between health data and a source of
risk estimate in case-control studies. Also, recalling the exact environmental exposure.
timing of exposure, which is often important in determining
temporality of an association and in estimating induction CONTROLLED STUDIES
period of a disease, can be differential among exposed cases
and exposed controls. These studies have control groups (i.e. comparator that can
either be a standard drug or placebo). Controlled trials can
Despite the fact that recall bias is a major limitation of be clinical trials (unit of randomization is an individual) or
case-control studies, a number of methodological strategies community trials (unit of randomization is a community or
documented in the literature can minimize the recall bias.[6] cluster).

The advantages of case-control studies are that they can be Nonrandomized controlled
done quickly and are very efficient for conditions/diseases This is an experimental study in which people are allocated to
with rare outcomes.[7] different interventions using methods that are not random. In

Disease
Exposed
Adverse Event Exposed
Not Exposed No disease
Popula
on Popula
on

Exposed Disease
No Event
Not Exposed

Not Exposed Prospec
ve No disease

Retrospec
ve

Figure 1: Case-Control design Figure 2: Cohort (Longitudinal studies) design

International Journal of Ayurveda Research | April-June 2010 | Vol 1 | Issue 2 129

 Parab and Bhalerao: Study designs

these studies, allocation to different groups is done arbitrarily. Assumptions

This kind of study design may sometimes overestimate the The effects of intervention during the first period do not carry
advantages of one treatment over other. over into the second period.

Randomized controlled Internal factors (e.g. disease severity) and external factors
Randomized controlled trials (RCTs) are considered the (e.g. season), which can affect the efficacy of the drug/s, are
“gold standard” in medical research since they offer the constant over time.
best answers about the effectiveness of different therapies
or interventions. Ideally, the patient’s disease condition should return to its
baseline state after discontinuation of the first treatment.
The important aspect of this study design is that the patients are
randomly assigned to the study all groups that help in avoiding In case of Ayurvedic studies, this design can prove useful
bias in patient allocation-to-treatment that a physician might as each patient serves as his own control and this way the
be subject to [Figure 3]. It also increases the probability that individualistic approach of Ayurveda gets conserved even in
the differences between the groups can be attributed only to clinical studies. However, at the same time it is difficult to
the treatment(s) under study. implement as the “wash out period” (duration between two
treatments to wash out the effect of the first so that it does not
There are certain types of questions where randomized get carry over) cannot be defined in view of unavailability of
controlled studies cannot be done for ethical reasons, for pharmacokinetic data of Ayurvedic treatments.
instance, if patients are asked to undertake harmful experiences
(like smoking) or denied any treatment beyond a placebo when Factorial
there are known effective treatments. Studies involving two or more factors while randomizing
are called factorial designs [Figure 6]. A factor is simply a
There are different types of randomized studies as follows.[8] categorical variable (e.g. age and prakriti) with two or more
values, referred to as levels.
Parallel
In parallel studies, treatment and controls are allocated to Factorial design permits researchers to investigate the joint
different individuals. This is unlike a crossover study where effect of two or more factors on a dependent variable (e.g.
at first one group receives treatment A, followed by treatment weight). The factorial design also facilitates the study of
B later, while the other group receives treatment B followed interactions, illuminating the effects of different conditions
by treatment A [Figure 4]. of the experiment on identifiable subgroups of subjects
participating in the experiment.
In case of Ayurvedic studies, an extension of this design
known as “add-on” design is useful, where one group receives Cluster
standard treatment, while the other group receives standard It is a type of randomized controlled trial wherein groups
treatment along with Ayurvedic treatment. Using these studies, of participants (as opposed to individual participants) are
comparison of relative or absolute efficacy can be obtained in randomized. Cluster randomized controlled trials are also
a short period. However, these studies generally require large known as cluster randomized trials, group randomized trials,
number of patients for the analysis. and place randomized trials.

Crossover Advantages of cluster randomized controlled trials over

In these types of studies each patient serves as his own control. individually randomized controlled trials include the ability
Each patient gets both drugs; the order in which the patient to study interventions that cannot be directed toward selected
gets each drug is randomized [Figure 5]. Generally, it requires individuals (e.g. a radio show about lifestyle changes) and
a smaller sample size.[9] the ability to control for "contamination" across individuals

Random assignment
Treatment
Treatment Follow up
group Pa
ents randomize
Pa
ents Compare results

Random assignment Control group Follow up

Control

Figure 3: Randomised clinical trial Figure 4: Parallel design

130 International Journal of Ayurveda Research | April-June 2010 | Vol 1 | Issue 2

 Parab and Bhalerao: Study designs

Treatment B
Treatment A Treatment B No Treatment

Pa
ents Pa
ents
Treatment A Treatment A + B
Treatment B Treatment A
Figure 6: Factorial design

Figure 5: Cross over design

(e.g. one individual's change in behavior may influence REFERENCES

another individual to do so too). Disadvantages compared
with individually randomized controlled trials include greater 1. Campbell MJ, Machin D. Medical Statistics: A common-
sense approach. 3rd ed. Chichester: Wiley; 1999.
complexity in design and analysis and a requirement for more 2. Lilienfeld AM, Lilienfeld DE. Foundations of Epidemiology.
participants to obtain the same statistical power. London: Oxford University Press; 1980.
3. Grimes D, Schulz K. Bias and causal association in
observational research. Lancet 2002;359:248-52.
Quasi-randomized 4. Gabbe B, Finch C, Bennell K, Wajswelner H. How valid is a
In these studies, participant allocation is done using schemes self-reported 12 month sports injury history? Br J Sports Med
such as date of birth (odd or even), number of the hospital 2003;37:545-7.
5. Choi B, Noseworthy A. Classification, direction, and
record, date at which they are invited to participate in the study prevention of bias in epidemiologic research. J Occup Med
(odd or even), or alternatively into different study groups. 1992;34:265-71.
6. Kaerlev L, Lynge L, Sabroe S, Oslen J. Colon cancer
controls versus population controls in case-control studies of
A quasi-randomized trial uses quasi-random method of allocating occupational risk factors. BMC Cancer 2004;4:15.
participants to different interventions. There is a greater risk of 7. Zailinawati AH, Nik-Sherina H, Teng CL. What study design
selection bias in quasi-random trials where allocation is not should I choose? Malays Fam Physician 2006;1:29-31.
8. Hopewell S, Dutton S, Yu LM, Chan AW, Altman DG. The
adequately concealed compared with randomized controlled quality of reports of randomised trials in 2000 and 2006:
trials with adequate allocation concealment. Comparative study of articles indexed in PubMed. BMJ
2010;340:c723.
9. Byron J, Kenward MG. Design and Analysis of Cross-Over
ACKNOWLEDGMENT Trials. 2nd ed. London: Chapman and Hall; 2003.

The authors sincerely acknowledge Dr. Jaideep Gogtay, Medical

Director, Cipla Ltd., for his technical help. Source of Support: Nil, Conflict of Interest: None declared

International Journal of Ayurveda Research | April-June 2010 | Vol 1 | Issue 2 131

Copyright of International Journal of Ayurveda Research is the property of Medknow Publications & Media
Pvt. Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email articles for
individual use.