Study Designs

Hani Tamim, MPH, PhD

Assistant Professor Epidemiology and Biostatistics Research Center / College of Medicine King Saud bin Abdulaziz University for Health Sciences Riyadh Saudi Arabia

Scientific methods

Scientific research: answer questions and acquire new knowledge Scientific methods: includes the following:

Empirical approach Observations

Questions
Hypotheses Experiments Analyses Conclusions Replication

Research

Two types:

Qualitative research

Involve interviews and observations No formal measurement

Quantitative research

Systematic measurement and quantification of results Use of statistics

Study Design
I- Observational studies

Descriptive studies - Case report / case series - Cross-sectional studies

Analytic studies - Case control studies - Cohort Studies

II- Experimental studies

Clinical trials

Study Design
I- Observational studies

Descriptive studies - Case report / case series - Cross-sectional studies

Analytic studies - Case control studies - Cohort Studies

II- Experimental studies

Clinical trials

Descriptive studies

Case report:

Documents unusual medical occurrences that can represent the first

clues in the identification of new diseases or adverse effects of exposures (based on one patient)

Descriptive studies

Case series:

Are collections of individual case reports (based on more than one

patient

Study Design
I- Observational studies

Descriptive studies - Case report / case series - Cross-sectional studies

Analytic studies - Case control studies - Cohort Studies

II- Experimental studies

Clinical trials

Study Design
I- Observational studies

Descriptive studies - Case report / case series - Cross-sectional studies

Analytic studies - Case control studies - Cohort Studies

II- Experimental studies

Clinical trials

Cross-sectional study
Also called surveys

Exposure and disease are assessed simultaneously among

individuals in a defined population, thus at one point in time

No sampling of individuals based on a exposure or an outcome

Cross-sectional study
Exposure + , disease + Defined population

Exposure + , disease Sampl e

Exposure - , disease +

Exposure - , disease Time of study

Time

Two by two table

Disease Exposure Yes No Total

Yes
a c a+c

No
b d b+d
=a/a+b

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

Prevalence of outcome in exposed

Prevalence of outcome in non-exposed = c / c + d

Prevalence Rate Ratio (PRR) =

=a/a+b c/c+d

Cross-sectional study: example

Estimate the rate of prescribing evidence-based Myocardial Infarction medications between males and females Gender Evidence based Prescription Yes No

Total

Male
Female Total

2228
737 2965

1281
1280 2561

3509
2017 5526

Prevalence of outcome in males Prevalence of outcome in females Prevalence Rate Ratio (PRR) =

= 2228 / 3509 = 63.5% = 737 / 2017 = 36.5% = 63.5 / 36.5 = 1.7

BMC Family Practice, 2004; 5; 6

Cross-sectional study

Advantages:

Cheap Fast Simple

Disadvantages:

Representatives Failing to take time Vulnerability towards biases

Study Design
I- Observational studies

Descriptive studies - Case report / case series - Cross-sectional studies

Analytic studies - Case control studies - Cohort Studies

II- Experimental studies

Clinical trials

Study Design
I- Observational studies

Descriptive studies - Case report / case series - Cross-sectional studies

Analytic studies - Case control studies - Cohort Studies

II- Experimental studies

Clinical trials

Case-control study

Exposure + Case Exposure -

Defined population

Sampl e
Exposure +

Control
Exposure Direction of study

Time

Two by two table

Disease Exposure Yes No Total

Yes
a c a+c

No
b d b+d =a /b

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

Odds of outcome in exposed

Odds of outcome in non- exposed

Outcome odds ratio = (a/b) / (c/d)

=c /d
= ad / bc

Case-control study: example

Assess the risk of ischemic stroke in association with COX-2 NSAIDs (Rofecoxib) NSAIDs Ischemic stroke Yes No

Total

Yes
No Total

121
2973 3094

334
11525 11859 = 121 / 334 = 1.4

455
14498 14953 = 0.36 = 0.26

Odds of outcome in exposed Outcome odds ratio = (a/b) / (c/d)

Odds of outcome in non-exposed = 2973 / 11525

Stroke, 2006; 37; 1725-1730

Case-control study: challenges

Selecting cases

Eligibility

Selecting controls

Representativeness

Exposure assessment

Accurate

Case-control study

Advantages:

Cheap Require short time when outcomes are delayed Study multiple exposures

Appropriate for rare diseases

Disadvantages:

Biases

Selection (control selection) Information (recall) Confounding (incomplete information on confounders)

Study Design
I- Observational studies

Descriptive studies - Case report / case series - Cross-sectional studies

Analytic studies - Case control studies - Cohort Studies

II- Experimental studies

Clinical trials

Study Design
I- Observational studies

Descriptive studies - Case report / case series - Cross-sectional studies

Analytic studies - Case control studies - Cohort Studies

II- Experimental studies

Clinical trials

Cohort study

Defined population

Disease + Exposure + Disease -

Sampl e
Disease +

Exposure Disease Direction of study

Time

Two by two table

Disease Exposure Yes No Total

Yes
a c a+c

No
b d b+d = a / a+b

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

Risk of outcome in exposed

Risk of outcome in non-exposed

Relative Risk

= c / c+d
= (a / a+b) / (c / c+d)

Cohort study: example

Assess the efficacy of warfarin compared to aspirin in patients after aortic valve replacement. Drug Death at follow up Yes No

Total

Warfarin
Aspirin Total

5
3 8

103
138 241

108
141 249

Risk of outcome in exposed Risk of outcome in non-exposed Relative risk

= 5 / 108 = 4.6% = 3 / 141 = 2.1% = 2.1 (not significant)

Circulation (Journal of the American heart Association), 2004; 110; 496-500

Cohort study

Advantages:

Study multiple outcomes Appropriate for rare exposures

Disadvantages:

Expensive Long follow-up Loss to follow-up

Study Design
I- Observational studies

Descriptive studies - Case report / case series - Cross-sectional studies

Analytic studies - Case control studies - Cohort Studies

II- Experimental studies

Clinical trials

Study Design
I- Observational studies

Descriptive studies - Case report / case series - Cross-sectional studies

Analytic studies - Case control studies - Cohort Studies

II- Experimental studies

Clinical trials

Clinical trial

Defined population

Disease + Exposure + Disease -

Sampl e
Disease +

Exposure -

Intervention
Direction of study

Disease Time

Clinical trial: example

Assess the association between macrolide clarithromycin and mortality in patients with stable coronary heart disease. Mortality
Clarithromycin

Yes

No

Total

Yes
No (placebo) Total

344
307 651

1828
1893 3721

2172
2200 4372

Risk of outcome in exposed Risk of outcome in non-exposed Relative risk

= 344 / 2172 = 16% = 307 / 2200 = 14% = 1.14

British Medical Journal, 2006; 332; 22-27

Clinical trial: Different issues

Ethical consideration:

Enough evidence of efficacy Should not be superior

Randomization

To assure equal distribution of confounders Each subject will have same probability of taking the intervention Done by hand or computer Randomized and non-randomized clinical trials Randomized clinical trial are superior

Clinical trial: Different issues

Allocation concealment

Assessing eligibility criteria and knowledge of allocation group

Control group

Placebo or other standard exposure

Blinding

Open labeled, single, double, triple

Clinical trial: Different issues

Efficacy:

Potential effect of treatment: whether treatment can have an effect on outcome.

Compare subjects according to the treatment actually received.

Exclude subjects who complied poorly, switched over, or withdrew

Effectiveness:

Actual effect of treatment in the real world of people who comply

poorly, or change treatment

Clinical trial

Advantages:

Reduced of biases (selection, information, and confounding)

Disadvantages:

May be impractical or unethical Expensive

Generalizability

Study Design
I- Observational studies

Systematic review

Descriptive studies - Case report / case series - Cross-sectional studies

Analytic studies - Case control studies - Cohort Studies

II- Experimental studies

Clinical trials

Systematic review

Assess the association between Homocysteine and Risk of Stroke

JAMA, 2002; 288; 2015-2022

THE END