# 3. Choosing the best study design

### Study design

**Learning outcomes:
Observational Studies
**➔ Define cohort studies

➔ Distinguish between prospective and retrospective cohorts

➔ Explain the nested case-control design and strategy

➔ Describe the multiple-cohort design

➔ Define cross-sectional studies

➔ Explain why cross-sectional studies yield weaker evidence for causality than cohort studies

➔ Define case-control studies and their benefits and problems

➔ Describe case-crossover studies.

**Randomized Blinded Trials
➔** Define randomized blinded trials

**Explain how to design RBTs**

➔

➔

**Describe how to choose the intervention and control conditions**

➔

➔

**Describe how to define outcomes and adverse effects**

➔

➔

**Describe how to select participants**

➔

➔

**Describe how to measure baseline and outcome variables**

➔

➔

**Evaluate approaches to randomizing and blinding.**

➔

➔

### Studies of medical tests

**Learning outcomes:**

➔ Understand the definition of studies of medical tests and how these studies differ from therapeutic intervention trials or studies to assess causality

➔ Explain how to select subjects for a study of a medical test

➔ Understand how to measure reproducibility of a test including use of kappa and the coefficient of variation

➔ Define key metrics to use in studies that assess the accuracy of a diagnostic test including sensitivity, specificity, predictive value, ROC curves, and likelihood ratios

➔ Understand how to design studies of clinical prediction rules and the associated limitations and challenges with this design.

### Enhancing causal inference

**Learning outcomes:**

➔ Describe cause-effect relationships and enumerate the four rival explanations

➔ Identify ways to minimize chance

➔ Discuss bias and identify ways to avoid bias

➔ Identify ways to make confounding less likely

➔ Offer several suggestions or strategies for incorporating opportunistic observational designs

➔ Explain how causal inference can be enhanced by positive evidence.

### The methods: matching study designs to research questions

**Learning outcomes:**

➔ Why the methods section is the most important part

➔ How to report study methods accurately and fully

➔ How to report methods to minimise bias and confounding

➔ How to use reporting guidelines for different study types.

### Subjects and variables

**Learning outcomes:**

➔ Define sample and population, and describe how sample and population inform all clinical

research

➔ Identify criteria for a target population

➔ Compare and contrast approaches to sampling

➔ Describe several strategies for recruiting a sample of subjects.

### Sample size and power

**Learning outcomes:**

➔ List the steps for estimating sample size for an analytic study

➔ Explain other considerations in calculating sample size for analytic studies

➔ List the steps for estimating sample size for descriptive studies

➔ Identify strategies to minimize the required sample size

➔ Explain other strategies for estimating sample size when there is insufficient information

### Statistics

**Learning outcomes:**

➔ Define and describe box models

➔ Define and describe standard error

➔ Define and describe p-values

➔ Define null hypothesis

➔ Select the appropriate statistical tests for your study.

### Questionnaires and qualitative research

**Learning outcomes:**

➔ Describe steps an investigator can take to ensure that questionnaires and interviews are as valid and reproducible as possible

➔ Define open-ended questions and closed-ended questions and devise several examples of both types of questions

➔ Identify desirable question elements as well as pitfalls to avoid

➔ Design a one-page instrument that is easy to read, easy to understand, and suitable for data entry.