Course Title: Scientific Research

Week 5: [Research designs Part 1]

Instructor: [Mahmoud Salah]

Date: [18/10/2025]

License: CC BY-NC-SA 4.0

Once an investigator has formulated the research question and formulated a particular, testable
hypothesis, the next is choosing a sufficient research design.

Research designs come in two broad forms:

➢ Observational Studies (Non-Experimental/Non-Interventional): The researcher is a passive

observer and measures outcomes as they naturally happen, without controlling exposure or
intervention.
Those studies have mainly two types:

• Descriptive
• Analytical

➢ Experimental/Intervention Studies: The researcher actively does something, i.e., an

intervention or an action (e.g., a new rehabilitation technique) and assesses the impact.

Decision Making for Study Selection

Selection of study design is ultimately a matter of how well-established existing knowledge of

the problem and the nature of the research question is asked.
• If little is known about the nature of, or the causes of, a problem (e.g., "How large is the
problem?" or "Who is affected?"), descriptive Studies are selected.

• If some factors are suspected to be among the cause of the problem (e.g., ascertaining whether
factors do relate to the problem), Analytical Studies (Cohort or case-control studies) are
employed.

• If enough information is known to devise an intervention and try it out (e.g., "What is the
effect of a particular intervention?"), Experimental Studies (randomized/non-randomized
controlled trials) are utilized.
Example: If the problem exists, e.g., a high rate of patient non-compliance with exercises at
home, but no reason why, then a Descriptive study is required in order to identify who is
affected and what they perceive about the problem. But if the question is framed differently,
i.e., testing whether or not a new intervention into pain management is effective, then an
Experimental study is required.
 Observational Studies

Descriptive studies

The researcher has no control over the interest exposure since he/she remains a passive
observer. The character of a given situation, event, or case is established. These studies
establish a hypothesis and describe the event in relation to time, place, and person.

• Time: Year, season, duration of illness, etc.

• Place: Nation, region, city, country, institutions, etc.
• Person: Age, gender, profession, personal habits (e.g., smoking, brushing teeth), height,
weight, etc.
Specific Descriptive Designs

1. Case Studies

Give a concise description of the characteristics of a single or very few 'cases' (i.e., a patient, a
health unit, or a village).
In clinical medicine, this would typically be a lead-in to the description of a clinical picture of
a previously unknown disease.

Example: Describing the step-by-step process of recovery, unusual complications, and

intervention steps taken for an extremely uncommon neuromuscular disorder patient.

2. Case Report Series

A concise and objective report of a clinical sign or observation in a series of clinical subjects.
Generalization of the results is limited on the basis of potential biased selection, lack of control
group, or subject unrepresentativeness.
3. Correlation Studies:
Allows hypothesis creation for conducting analytic research. They are quick and inexpensive
as data are normally readily available.
Example: Charting geographic levels of physical activity (population data) against geographic
rates of Type 2 Diabetes (population outcome data) to make an inference toward a hypothesis
for an analytic study.

4. Cross-Sectional Studies (Prevalence/Instantaneous Study):

A cross-sectional study where the investigator identifies a particular population at a single point
in time and assesses a collection of factors and the outcome of interest at the same point in
time.

Since exposure and outcome are determined at the same point in time, no directionality as such
exists.
The study can comprise:

• Physical characteristics (e.g., disability, blood pressure).

• Socio-economic characteristics (e.g., age, education, income).
• Behavior or habit (KAP studies: Knowledge, Attitudes, Practices).
• Health and disease attributes (e.g., exposure to environmental factors).

Example: An analysis on a large scale evaluating the prevalence of chronic knee pain among
manual workers in a particular industry and simultaneously obtaining data on their relevant job
activities and use of supportive bracing (exposure and outcome determined simultaneously).

Analytical Studies:

1. Case-Control Studies (Case-Reference/Retrospective Studies):

A Quite simple and inexpensive, often used to examine etiologies of rare diseases.

• Design: investigator looks back from the outcome (disease) to likely exposure (cause).
• Study contrasts cases (with disease/outcome) with a suitable control group (without
disease).
• Primary Measurement: Odds Ratio (OR).
The odds ratio (OR) is a measure to report the extent to which an event is more or less likely
to occur in one group than another. In short, it provides the odds that an event will occur in one
group relative to the odds that the event will occur in some other group. "Odds" is equal to the
probability that something happens divided by the probability that it does not happen. For
example (showed in the table below), in a test of whether regular exercisers among
physiotherapists are at reduced risk of experiencing low back pain, we could compare the risk
of back pain between exercisers and non-exercisers. The odds ratio is 1 if there is no group
difference. An odds ratio of more than 1 reflects the event to be more probable in the first group,
while an odds ratio of less than 1 reflects it to be less probable. For instance, if the odds ratio
was 0.26, then it would reflect that physiotherapists who exercise regularly have about 74%
lower likelihoods of low back pain compared to those who don't. To summarize, the odds ratio
enables us to understand the degree and direction of association between an exposure (e.g.,
exercising) and an outcome (e.g., back pain), but never proves one causes the other.

Group Have low back pain No low back pain Total

Exercise regularly 10 90 100

Don’t exercise 30 70 100

Exercisers: Odds = 10/90 = 0.11

Non-exercisers: Odds = 30/70 = 0.43

Then, the odds ratio (OR) = 0.11 ÷ 0.43 = 0.26

OR = 1 → no effect
OR > 1 → exposure increases risk
OR < 1 → exposure decreases risk

2. Cohort Studies:

A cohort of persons moving forward or backward in time, often sharing one thing in common
(the cohort).

• Design: Observational study attempting to link a suspected (exposure) to a subsequent

risk of developing disease (effect).
• Directionality:
➢ In prospective studies, participants are followed forward in time to see who
develops the disease.
➢ In retrospective studies, existing records are used to look back at exposures and
outcomes that have already happened.
• Primary Measurement: Incidence Rate and Relative Risk (RR). (Explained in the
examples below)

Prospective Cohort Study Example:

Two groups of athletes are followed for one year by a physiotherapist in a prospective cohort
study to determine if regular strength training reduces the risk of incurring shoulder injury.
None of the athletes is injured at the start. The physiotherapist groups them into an exposed
group and an unexposed group: those who receive strength training and those who do not. After
one year, 10 of 100 in the strength-training group and 30 of 100 in the non-training group
develop injuries. The rate of incidence (new cases ÷ total at risk) is thus 10/100 = 0.10 (10%)
for the exposed group and 30/100 = 0.30 (30%) for the unexposed group. The relative risk (RR)
is calculated by dividing these incidence rates: 0.10 ÷ 0.30 = 0.33. This shows that athletes who
perform strength training have about one-third the risk of shoulder injury of non-strength-
training athletes — and this would suggest that strength training may be protective, but cannot
be said to cause anything.

Retrospective Cohort Study Example:

In a retrospective cohort study, the investigator goes back to hospital records for the past five
years to examine if patients who attended a supervised physiotherapy program after surgery on
their knee were less likely to develop chronic knee pain than those who did not attend a
physiotherapy program. The patients are classified based on their past exposure (whether they
attended physiotherapy). Based on the reports, 20 out of the 80 patients who had come for
physiotherapy had developed chronic pain and 40 out of the 80 who had not come had
developed chronic pain. The proportion of development of chronic pain (incidence rate) for the
physiotherapy group is 20/80 = 0.25 (25%) and for the non-physiotherapy group is 40/80 =
0.50 (50%). The relative risk (RR) is therefore 0.25 ÷ 0.50 = 0.5, indicating that patients
undergoing physiotherapy had half the probability of developing chronic knee pain compared
to those who did not.

Case-Control Study Example:

In a case-control study, the investigation begins with the outcome, not the exposure. Using the
same hospital setting, the researcher finds 50 patients who are suffering from chronic knee pain
(cases) and 50 patients free from chronic knee pain (controls), all undergoing knee surgery
within the past five years. The researcher then looks back at how many patients in each group
attended the physiotherapy program after surgery. 15 out of 50 of the cases attended
physiotherapy, while 35 out of 50 of the controls attended. To measure the association, the
researcher calculates the odds of exposure:

Odds in cases = 15/35 = 0.43

Odds in controls = 35/15 = 2.33

The odds ratio (OR) = 0.43 ÷ 2.33 = 0.18.

This means that patients who attended physiotherapy had about 82% fewer opportunities for
chronic knee pain compared to those who did not attend.

Types of Variables:

Independent Variable The variable that causes or influences the dependent variable; it is not
dependent on the outcome to be measured (e.g., specific exercise protocol).

Dependent Variable The outcome or consequence being measured, depending on the

independent variable (e.g., improvement in gait speed).
Confounding Variable A variable which is associated both with the problem and with a
suspected cause of the problem. It can strengthen or weaken the perceived association between
independent variable and dependent variable and needs to be controlled for in planning or
analysis.
References:

1. Goyal, R. C. (2010). Research Methodology for Health Professionals. Jaypee

Brothers Publishers.

License

This document is licensed under a Creative Commons Attribution. 4.0 International

License.