Study Designs

Dr. Naveed Zafar Janjua Community Health Sciences The Aga Khan University

Epidemiology
Epidemiology is study of distribution and determinants of health events in population. What does this mean?
Distribution Determinants

Environmental Exposure and Health Effects

Environmental Exposure Smoking

Internal Dose PAH

Biological effective dose Activated PAH

Early Effect PAH-DNA Adduct

Late effects Cancers

How these Relations were Explored?

Through posing questions Developing hypothesis Defining objectives Design and conduct of studies Measuring exposure and effect Statistical Analysis Inference

Questions
How many textile mill workers have COPD? What is lead level of children in Karachi City? Does regular use of wood and cow dung increases risk of COPD among rural women? Does chromium causes skin cancer?

Measuring Disease Occurrence

Number of new cases of disease in a specific time period Incidence = Population at riskat that time

Example: 3 cases of lung cancer per 1000 men over age 18 each year in 2001

Measuring Disease Occurrence

Number of cases at a specific time Prevalence = Population at risk

Example: 6% of a textile mill workers had asthma in June 2003 2 men per 1000 had lung cancer in 1999 in Karachi

Measuring Disease Occurrence

Incidence exp Relative Risk (RR) = Incidence unexp

Odds exp Odds Ratio (OR) = Odds unexp

Study Designs: case scenario

You are Medical Advisor of a Petrochemical Industry. There had been concern that workers handling various products may get exposed to the benzene. Benzene is potentially carcinogenic and cause leukemia (blood cancer). Level of benzene vary across work stations. You have been asked by your seniors who were pressurized by trade union and govt. to investigate if there is any effect of benzene exposure on occurrence of leukemia or any other disease?

Cohort Studies
Cross-sectional study or screening to exclude those with disease Exposed to Benzene Reference population free of leukemia Leukemia (-) Sample Leukemia (+)

Leukemia (+)

Not exposed to Benzene

Leukemia (-)

Cohort studies start with an exposure and go forward to diseases.

Drawing by: Nick Thorkelson

Cohort Studies
Leukemia Positive Leukemia Negative

Benzene Exposure

a+b

Non Exposed to Benzene

c+d

a+c

a+d

a+b+c+d

Cohort Studies
Used to test hypothesis about causation of disease Groups of people are defined on particular characteristics before the appearance of disease under study-defining exposed and non exposed Observation of groups over period of time to determine and compare frequency study disease among them

Cohort Studies
Analysis: Relative Risk= (a/a+b)/(c/c+d) Advantages & disadvantages:
Good for rare exposures Can ascertain multiple outcome of single exposure Demonstrate temporal relationship Time consuming and costly Loss to follow up ( participants can leave in between)

Examples of Cohort Studies

Associations between lifetime lead exposure with intelligence among children Lead exposure and motor functioning in 4(1/2)-year-old children: the Yugoslavia prospective study; to investigate associations between lead exposure and early motor development

Study Designs: scenario

The study (cohort) you are planning is
Time consuming Costly and some of your workers may leave till completion of study

What you can do?

Retrospective Cohort Study

If exposure measurements/records and job histories are available can get evidence about exposures If medical records are available you can get information about how many developed Leukemia?

Study Designs
You went back to your friends and discussed? Group work to come with solution?

Work finished!

Study Designs
Information required for retrospective cohort was not available! You went back to your seniors and told them that work is not doable Seniors were insistent on doing some thing and asked you :
Would you be able to tell us causes of leukemia only among workers in a way that cost less?

Study Designs
You went back to your friends and discussed What options do you have? what you can do?

Solution: Case Control Study

Benzene (+) Leukemia pts

Ask about exposures

Benzene (-)

Benzene (+) Healthy Controls benzene (-)

Case-control studies start with a disease and go back to exposures.

Drawing by: Nick Thorkelson

Case Control Studies

Leukemia Patients Benzene Exposed Benzene Non exposed a c a+c Healthy Controls b d a+d a+b c+d a+b+c+d

Case Control Studies

Selection of study subjects is on the basis of disease status Case: Those suffer from disease under study Control: Those are free from the disease under study

Selection of cases and control

Select cases Controls are selected from the same study base as cases were Control if acquire disease, you should able to select it.

Sources of controls
Community Neighborhood Hospitals Friends

Important issues
Recall
Recall bias Imperfect recall

Temporality Selection

Case Control Studies

Analysis Calculation of Odds Ratio- as measure of association OR = ad/bc Advantages & disadvantages: Good for rare disease as cancers Short duration and inexpensive Rely on recall Finding appropriate comparison group difficult Sequence of event is difficult to establish

Examples of Case Control Studies

To asses the risk factors for acute hepatitis B To evaluate the association between lead exposure, as reflected in bone lead levels, and adjudicated delinquency The effect of Pb exposure from automobile exhaust on renal integrity among traffic policemen To test the hypothesis that Essential tremors (ET) is associated with lead exposure.

Study Designs: another scenario

Social Security Institute was asked by govt. that there are reports of poor safety measures in leather industry and handling of chromium is very poor? Occupational health expert was called and asked go and plan an investigation and tell us what is the magnitude of problem say prevalence of chrome ulcer among workers.

Solution: Cross-sectional Study

Reference population Chrome ulcer (+) Gloves (-) Sample Gloves (+) Chrome ulcer (-) Gloves (-)

Gloves (+)

Cross-sectional Studies
Disease Non disease b d a+d a+b c+d a+b+c+d

Exposed Non Exposed

a c a+c

Cross-sectional Studies
Objectives:
To examine the health problem and disease frequency To examine relationship between exposure and disease frequency or health problem

Unit of Analysis: individual Exposure and disease status is assessed at the same time

Advantages
Short duration study Start with reference population so generalization possible Provide prevalence estimates

Examples of Cross-sectional Studies

Blood lead levels and risk factors for lead toxicity in children from schools and an urban slum in Delhi Prevalence of dermatitis among leather tannery workers

Design in Epidemiology
Design choice depends on the objectives of study. When you want to know How much is disease: distribution of disease descriptive study and when you want to answer why Analytical studies

Analytical Designs
Observational
Case-control Cohort

Interventional
Randomized Controlled Trials Clinical and Community Trials Pre post comparison

Randomized Controlled Trial

These studies are used test the interventions Interventions could be
Drugs or clinical procedures Health education programme A new measure for hazard prevention or decreasing exposure at workplace

Randomized Controlled Trial

Radical mastectomy Recurrence of cancer

Patient with Breast Cancer

Random allocation Radical mastectomy with more limited resection

No Recurrence of cancer Recurrence of cancer

No Recurrence of cancer

Randomized Controlled Trial

Intervention Recurrence of breast cancer Total

+ Radical mastectom y (RM) Mastectomy with more limited resection (MLR) Total

a+b

c+d

a+c

b+d

a+b+c+d

Randomized Controlled Trail

Directly observed treatment Patients of MTB positives

Random allocation

MTB(+)

MTB (-)

MTB(+) Conventional treatment MTB (-)

Examples
Placebo-controlled, randomized trial of up to three courses of succimer in children with blood lead levels of 20-44 microg/dL (1.0-2.1 microM). Trail to assess the safety and efficacy of meso-2,3-dimercaptosuccinic acid in the treatment of children with lead toxicity

Cohort & Intervention study

Cohort Study (Prospective) To test hypothesis regarding the causation of disease Intervention Study To evaluate the efficacy of preventive or therapeutic agent or procedure

Objective

The role of investigators

To conduct To allocate the observation of exposure to the exposure causing study subjects, the disease and observe the outcome caused by exposure

Descriptive Epidemiology
Use to describe health states or events Types
Case Report & Case series Correlative studies Time series analysis Ecological Studies Surveillance Cross-sectional studies

Case Report & Case Series

Case report: A careful and detailed report of some new finding about one case is case report. e.g., chronic CO poisoning in a child Case report: Case report describe characteristics of number of cases e.g., distribution of characteristics of cancer patients in a hospital

Correlation studies
Objective: To correlate general characteristics of population with their disease frequency at same period of time; within same population at different time periods Unit of analysis: group

Ecological correlations
To correlate characteristics of population with disease frequency among several group of people during same period of time Example: Average TSP in ambient of big cities and respiratory tract infections

Surveillance
Continuous, systematic process of collection, analysis, interpretation and dissemination of information for monitoring health problems

Study Designs
Descriptive studies Analytical studies

Case report

Case series

Surveillance

Correlational

Cross-sectional

Observational

Interventional

Case-control

Cohort

Clinical Trail

Community trail

Traditional Study Designs

Design
Prospective follow-up Retrospective follow-up Nested case-control Registry-based case-control Cross-sectional Ecologic; PMR Descriptive & surveillance

Environmental Epidemiology
Common Rare Rare Common Common Becoming rare; rare Common for exposure & disease

Occupational Epidemiology
Rare Common Common Common but of limited value Common Rare; becoming rare Common for exposure & disease Common

Meta-analysis or collaborative Common reanalysis

Specific Study Designs in Relation to Research Objectives & Design Determinants

Objectives/Design Determinants
Etiology Design* PF RF NCC RBCC CS Ecologic Surv. Meta + + + + + + ++ + + + ++ +++ Public Health Intervention + Hypothesis + + ++ Feasibility Credibility +++ + ++ ++ -

*PFS, prospective follow-up; RFS, retrospective follow-up; NCCS, nested case-control; RBCC, registry-based case-control; CS, cross-sectional.