Observational Epidemiology: Survey, Cohort and Case-Control Studies

Language of Instruction

English

Pre-requirements

Students admitted to a Master`s degree Programme may join this course (e.g. TropEd Europe network). Good working knowledge of English (TOEFL score of at least 550 points paper-based or 213 points computer-based, or an equivalent approved test).

Learning Outcomes

At the end of the course the students should be able to

• distinguish the principles of surveys, case-control and cohort studies - and how the three designs differ from each other and from the design of randomized controlled trials
• calculate sample sizes for surveys, cohort studies, and matched and unmatched case control studies, based on simple random sampling and two-stage cluster sampling with stratification
• compare alternative sampling methods (stratified, systematic, cluster, non-random)analyse data sets from surveys, cohort, and case-control studies
• calculate precision and account for design effect in cluster sample surveys
• distinguish the different types of cohort studies, i.e. prospective, retrospective and double cohorts
• distinguish the different types of case-control studies
• suggest relevant designs (plan) for case control and cohort studies and surveys
• compare principles and consequences of density based sampling of controls in case control studies and the importance of using incident rather than prevalent cases
• evaluate the direction and magnitude of selection- and information biases in case-control studies, cohort studies and surveys and discuss how to minimize the above mentioned biases during design and conduct of studies
• distinguish in stratified analysis potential confounding and interaction and ways to differentiate between the two, i.e. adjust for confounding factors using Mantel-Haenszel adjusted relative risk estimates and how best to present and interpret a stratified presentation of effect measures when interaction is present. This knowledge should be based on an understanding and ability to identify effect measure modification
• critically appraise the design, analysis and interpretation of studies conducted by other investigators
• communicate effectively with those involved in conducting public health research

Contact Information

Centre for International Health

Tel.: +47 55588560; e-mail: studie.cih@uib.no

Course Coordinator

Course coordinator:

Professor Knut M. Fylkesnes

Course offered (semester)

Spring

Language of Instruction

English

Course Unit Level

Master and PhD

Aim and Content

This course addresses critical methodological aspects of clinical and epidemiological studies relevant for interventions against poverty related diseases, including HIV infection, tuberculosis and malaria.

The lectures in the course cover the following:

- Epidemiology: - an overview.

-What is a survey?

- Random v non-random sampling, Simple random sampling,

- Sampling variability, estimates and confidence intervals,

- Alternative sampling (Systematic, Stratified, Cluster, Two-stage),

- Consequences for sample size calculation and data analysis exemplified with a simple data set.

- Stratification and clustering in survey.

- HIV related surveys.

- Demographic and health surveys: use of sample weights.

- Exercise: Estimation of sample size and confidence intervals (using formula, monograms, tables and computer)

- Overview of the cohort design

- Types of cohort studies

- Demographic surveillance sites

- Sample size calculations for surveys, cohort studies and case control studies

- Measures of disease occurrence, incidence, person-time

- Calculation of person years and incidence rates

- Bias in cohort, case control and survey studies

- Confounding and effect modification

- Interaction and confounding

- Poisson regression

- Points to remember in the planning and evaluation of cohort studies

- General principles of case-control studies

- Comparing case control studies with cohort studies

- Types of case-control studies, including nested case-control studies

- Sample size adequate power and matched case-control studies

- Sample size precision(didn¿t we include this Rajiv)

- Measurement of disease and exposure

- In a cohort and in the corresponding case-control dataset

- Odds vs. prevalence of disease

- Odds vs. prevalence of exposure

- Rare disease assumption is not always required

- Case selection to reduce bias

- Control selection to reduce bias

- Measures of association odds ratio as a measure in itself, as an approximation of relative risk and as a measure of incidence rate ratio

- Analyses of matched case-control studies

- Analyses of nested case-control studies

- Interaction

- Evaluation of bias and confounding

- Controlling confounding

- Mantel-Haenszel pooled OR estimate

- Brief introduction to logistic regression

The group work also covers the development of proposal and protocol.

The computer laboratory exercises include generating random numbers, calculating trial size, importing files, data exploration, baseline comparisons, main effects, adjustment for confounding, adjustment for confounding, sub-group analysis and interaction

By the end of this optional module students should be able to:

At the end of the course, the students shall:

1. be able to define and discuss the principles of case-control studies, cohort studies and survey research - and know how the three designs differ from each other and from the design of randomized controlled trials

2. be able to distinguish between the different types of cohort studies, i.e. prospective, retrospective and double cohorts

3. be able to suggest relevant designs (plan) for case control and cohort studies and surveys

4. be able to discuss the principles and consequences of density based sampling of controls in case control studies and the importance of using incident rather than prevalent cases

5. be able to identify and evaluate the direction and magnitude of selection- and information biases in case-control studies, cohort studies and surveys and discuss how to minimize the above mentioned biases during design and conduct of studies

6. be able to calculate sample sizes for cohort studies, matched and unmatched case control studies, and surveys based on simple random sampling and two-stage cluster sampling with stratification

7: be able to analyze data sets from case-control, cohort studies and surveys

8. using STATA - accounting for precision and design effect in cluster sample surveys

9. define alternative sampling methods (stratified, systematic, cluster, two-stage, non-random)

10. using STATA - do cluster sampling with probability proportional to cluster size based on a given data set

11. with an emphasis on stratified analysis, know how to identify potential confounding and interaction and ways to differentiate between the two. Know how to adjust for confounding factors using Mantel-Haenszel adjusted relative risk estimates and how best to present and interpret a stratified presentation of effect measures when interaction is present. This knowledge should be based on an understanding and ability to identify effect measure modification

12. critically appraise the design, analysis and interpretation of studies conducted by other investigaators

13. communicate effectively with those involved in conducting public health research

Learning Outcomes

At the end of the course the students should be able to

• distinguish the principles of surveys, case-control and cohort studies - and how the three designs differ from each other and from the design of randomized controlled trials
• calculate sample sizes for surveys, cohort studies, and matched and unmatched case control studies, based on simple random sampling and two-stage cluster sampling with stratification
• compare alternative sampling methods (stratified, systematic, cluster, non-random)analyse data sets from surveys, cohort, and case-control studies
• calculate precision and account for design effect in cluster sample surveys
• distinguish the different types of cohort studies, i.e. prospective, retrospective and double cohorts
• distinguish the different types of case-control studies
• suggest relevant designs (plan) for case control and cohort studies and surveys
• compare principles and consequences of density based sampling of controls in case control studies and the importance of using incident rather than prevalent cases
• evaluate the direction and magnitude of selection- and information biases in case-control studies, cohort studies and surveys and discuss how to minimize the above mentioned biases during design and conduct of studies
• distinguish in stratified analysis potential confounding and interaction and ways to differentiate between the two, i.e. adjust for confounding factors using Mantel-Haenszel adjusted relative risk estimates and how best to present and interpret a stratified presentation of effect measures when interaction is present. This knowledge should be based on an understanding and ability to identify effect measure modification
• critically appraise the design, analysis and interpretation of studies conducted by other investigators
• communicate effectively with those involved in conducting public health research

Pre-requirements

Students admitted to a Master`s degree Programme may join this course (e.g. TropEd Europe network). Good working knowledge of English (TOEFL score of at least 550 points paper-based or 213 points computer-based, or an equivalent approved test).

Teaching Methods

The pre-reading provides necessary background information to follow the course. Each day has a mixture of lectures and practical sessions, with group work or individual work on specific assignments and the use of the computer laboratory for data analysis under supervision. The lectures are interactive, and course participants are encouraged to ask questions and discuss during all sessions. The reference literature will be made available on the first day of the course through internet ("My Space"). Each week new scientific papers will be handed out for reading, group work group and presentations/discussions in plenary together with the course facilitators/lecturers.

About 40% of the course is lectures, 40% individual assignments or group assignments with supervision and work/discussions and 20% individual reading and lab exercises.

Assessment methods

4-hour written exam consisting of short questions and problem-solving questions and calculation.

Grading Scale

Pass/fail

Course Unit Evaluation

Students evaluate the teaching according to the quality assessment requirements of the University of Bergen. The evaluation method is through an online electronic form.

Contact Information

Centre for International Health

Tel.: +47 55588560; e-mail: studie.cih@uib.no