1) Non-interventional / Observational Study
Definition:
In this study, the researcher only observes and does not give
any treatment or change anything.
Examples (2):
Study of smoking habits and lung disease in people.
Survey of diet patterns in diabetic patients.
2) Interventional / Experimental Study
Definition:
In this study, the researcher gives a treatment or intervention
and observes the result.
Examples (2):
Giving a new drug to TB patients to check its effect.
Vaccination trial to see prevention of disease.
Non-Interventional (Observational) Study – Types
A) Descriptive Study
Definition:
Describes who, where, when about disease (no comparison).
Example:
Study showing number of dengue cases in Lahore in 2024.
B) Analytical Study
Definition:
Compares groups to find cause and effect.
Example:
Comparing heart disease in smokers vs non-smokers.
C) Cross-Sectional Study
Definition:
Data collected at one time only.
Example:
Survey of blood pressure in students in one college.
D) Case-Control Study
Definition:
Compares patients (case) with disease and healthy (control)
group to find cause.
Example:
Comparing past smoking history in lung cancer patients and
healthy people.
1) Cohort Study
Definition:
A study where a group of people (cohort) is followed over time
to see who develops disease.
Key point:
Exposure → Follow up → Outcome
Example (2):
Following smokers for 10 years to see who develops lung
cancer.
Following obese people to see who develops diabetes.
2) Randomized Clinical Trial (RCT)
Definition:
An experimental study where participants are randomly divided
into groups to test treatment.
Groups:
Treatment group
Control group (placebo/old treatment)
Case-Control Study (Easy Explanation)
Basic Steps
1) Selection of cases & controls
Cases: People who already have the disease
Controls: Healthy people (no disease)
2) Matching
Cases and controls are matched by:
👉 Age
👉 Gender
👉 Socioeconomic status
To avoid bias.
3) Measurement of exposure
Ask about past history
(smoking, diet, job, habits etc.)
4) Analysis & interpretation
Compare exposure in cases vs controls
Find risk factors
Selection of Cases
Case definition includes:
i) Diagnostic criteria
Disease must be clearly defined
Same rules for all patients
ii) Eligibility criteria
Prefer new cases
Old/advanced cases avoided
Source of Cases
Hospitals
From one hospital
Or from many hospitals
General population
All cases from one area
Found by survey / registry
Selection of Controls
Controls must be disease-free
Problem:
If disease is hidden (subclinical)
Diagnosis becomes difficult
Sources of Controls
Hospital patients
Relatives
Neighbours
General population
How many controls needed?
Common ratio
✔1:1
(1 case : 1 control)
If disease is rare
✔1:2
✔1:3
✔1:4
(More controls increase accuracy)
What is Matching?
Definition:
Matching means selecting controls similar to cases in important
factors like:
Age
Gender
Socioeconomic status
Area
Education
👉 Purpose:
To reduce bias and make fair comparison.
Kinds of Matching Procedures
1) Individual Matching
Meaning:
Each case is matched with one specific control.
Example:
Case: 40-year-old female
Control: 40-year-old female
2) Group Matching (Frequency matching)
Meaning:
Overall group is matched, not one-to-one.
Example:
60% males in cases
60% males in controls
3) Pair Matching
Meaning:
One case matched with one control
(Another name for individual matching)
Measurement of Exposure
• Clear definition and criteria about exposure
• Obtained by:
Interviews
Questionnaires
Past records (hospital, job records)
• Bias occurs if exposure is not measured properly
4) Analysis
To find:
A) Proportion of exposed cases and controls
→ Compare using chi-square test
B) Estimation of disease risk
→ Using odds ratio
Exposure Rates
• Case control study gives direct estimation
• Shows frequency of exposure in:
Diseased group
Non-diseased group
Method to Calculate Exposure
Example:
Smoking & lung cancer
Using 2×2 table
Limitations of Case Control Study
• Does NOT give incidence rate
• Relative risk cannot be calculated
• Relative risk is calculated from cohort study
Odds Ratio
• Derived from case control study
• Measures strength of association
• Mostly used for chronic diseases
How to Calculate Odds Ratio
Odds ratio = ad / bc
Example:
= 33 × 27 / 55 × 2
= 8.1 times higher
Bias in Case Control Study
Definition:
Bias = Systematic error in association between exposure &
disease
Disadvantage:
Risk estimate may increase or decrease
Varieties of Bias
Confounding bias
→ Removed by matching
Recall bias
→ Cases remember past better
Selection bias
→ Cases & controls not true population
Berksonian bias
→ Due to hospital admission differences
Interviewer Bias
Definition:
When the interviewer influences answers of the participant.
How it happens:
Asking leading questions
Different tone for cases & controls
Showing personal opinion
Example:
Doctor asks smokers more detailed questions than non-
smokers.
Measurement Bias
Definition:
When wrong method or tool is used to measure exposure or
disease.
How it happens:
Faulty instruments
Improper records
Wrong technique
Example:
Using a broken BP machine gives wrong readings.
Prospective Study
Definition:
Study that moves forward in time
(From present → future)
How:
Start with exposure → follow people → see disease
Example:
Following smokers for 10 years to see lung cancer.
Retrospective Study
Definition:
Study that looks back in time
(From present → past)
How:
Start with disease → check past exposure
Example:
Lung cancer patients → check past smoking history.
Experimental (Interventional) Study
Definition:
A study in which the researcher gives treatment/intervention to
participants and observes the result.
Also called:
✔ Randomized Controlled Trial (RCT)
Types of Experimental Studies
Randomized Controlled Trial (RCT)
– Participants are randomly divided into groups
Non-randomized trial
– Groups made without random method
Field trial
– Done in community (e.g. vaccine trial)
Clinical trial
– Done in hospital patients
Three Main Characteristics
1) Randomization
Meaning:
Participants are randomly divided into groups
Purpose:
– To avoid bias
– To make groups equal
Example:
Using lottery method to make groups
2) Manipulation
Meaning:
Researcher gives treatment/intervention
Example:
– New drug
– Exercise program
– Vaccine
3) Control
Meaning:
One group does not receive new treatment
(Placebo or old treatment)
Schematic Diagram of Experimental Study
Population
↓
Randomization
↓
Two groups formed
Group 1:
→ Receives treatment
→ Outcome observed
Group 2 (Control):
→ No treatment / placebo
→ Outcome observed
↓
Compare results
Quasi-Experimental Study
Definition:
A study where treatment is given but no randomization is done.
(Groups are already formed)
👉 Researcher cannot randomly divide people.
Examples of Quasi-Experimental Study
1) Person by Treatment
Meaning:
Different people receive different treatments naturally.
Example:
One group does exercise,
Another group takes medicine
→ Compare results.
2) Natural Experiments
Meaning:
Event happens naturally (not by researcher) and effect is
studied.
Example:
– Effect of flood on disease spread
– Effect of lockdown on mental health
3) Ecological Study
Meaning:
Study done on groups or communities, not individuals.
Example:
– Compare cancer rate in two cities
– Compare obesity in rural vs urban areas
Hybrid Study Design
Definition:
Hybrid study design means combination of two study designs in
one study.
(Observational + experimental together)
Example:
– Study starts as cohort study
– Then intervention is given like trial
→ This becomes hybrid design.
Another example:
Observing diabetic patients
Then giving diet program and checking results.
Clinical Trial
Definition:
A study done on patients to test
✔ New drug
✔ New treatment
✔ New therapy
Use of clinical trial:
• Check safety
• Check effectiveness
• Compare with old treatment
• Find side effects
Example:
Testing new BP medicine in hospital patients.
Types of Blinding
1) Single Blind
Patient does NOT know
Which treatment he is getting.
2) Double Blind
Both patient and doctor
Do NOT know who gets treatment.
3) Triple Blind
Patient, doctor and data analyzer
Do NOT know treatment groups.
Randomized Controlled Trials (RCT)
Definition:
A study where participants are randomly assigned to groups to
receive treatment or control.
Example:
Testing a new vaccine: one group gets vaccine, one group gets
placebo (randomly assigned).
Key point:
Randomization reduces bias and makes groups comparable.
3) Non-Randomized Controlled Trials
Definition:
Participants are assigned to groups without randomization.
Example:
Giving a new diet to patients in one hospital, old diet to another
hospital.
Key point:
More chance of bias because groups may differ.
3) Sequential Control Trials
a) Self-Control Trials
Definition:
The same person acts as control and treatment.
Example:
First month without drug → second month with drug → compare
results.
b) Cross-Over Trial
Definition:
Participants switch treatments after a period.
Example:
Group 1: Drug A → Drug B
Group 2: Drug B → Drug A
c) Trials with External Controls
Definition:
Compare treatment group with another group outside the trial.
Example:
Patients getting new drug vs previous hospital records of old
treatment.
4) Experimental Event Rate (EER)
Definition:
Percentage of patients who develop outcome in the treatment
group.
Formula:
EER = (Number of patients with event in treatment group ÷
Total patients in treatment group) × 100
Example:
100 patients take new drug
10 develop disease
EER = (10 ÷ 100) × 100 = 10%
Control Event Rate (CER)
Definition:
Percentage of patients who develop the event in the control
group.
Formula:
CER = (Number of patients with event in control ÷ Total
patients in control) × 100
Example:
100 patients in control group
20 develop disease
CER = 20 ÷ 100 × 100 = 20%
2) Relative Risk (RR)
Definition:
Shows risk in treatment group compared to control group.
Formula:
RR = EER ÷ CER
Example:
EER = 10% (treatment group)
CER = 20% (control group)
RR = 10 ÷ 20 = 0.5 → treatment reduces risk by 50%
4) Absolute Risk Reduction (ARR)
Definition:
Difference in risk between control and treatment.
Formula:
ARR = CER – EER
Example:
CER = 20%, EER = 10%
ARR = 20 – 10 = 10%
Meaning:
Treatment reduces 10% absolute risk.
5) Relative Risk Reduction (RRR)
Definition:
Percentage risk reduction relative to control.
Formula:
RRR = ARR ÷ CER × 100
Example:
ARR = 10%, CER = 20%
RRR = 10 ÷ 20 × 100 = 50%
Meaning:
Treatment reduces risk by 50% compared to control
6) Number Needed to Treat (NNT)
Definition:
Number of patients you need to treat to prevent 1 event.
Formula:
NNT = 1 ÷ ARR (as a proportion, not %)
Example:
ARR = 10% = 0.1
NNT = 1 ÷ 0.1 = 10
Meaning:
Treat 10 patients to prevent 1 disease/event
Meta-Analysis
Definition:
Meta-analysis is a study that combines results of many previous
studies on the same topic to get a stronger, overall conclusion.
It does not collect new patients
Uses data from already published studies