MI580 Lecture 4 - Observational Studies

Observational Studies

Patients are ‘observed’ in normal clinical practice (‘real world setting’), self-assigned exposures.

Subject to a number of biases.

Real World Data

Data relating to patient health status and/or the delivery of health care routinely collected from a variety of sources

Real World Evidence

The clinical evidence regarding the usage and potential benefits or risks of a medical product derived from analysis of RWD

Ecologic Studies

Observational studies of population group characteristics.

Cannot demonstrate causality. Mostly focused on hypothesis generating, not testing.

Ecological Fallacy

Assigning the characteristics of a group to individuals within the group who may not have that characteristic

Case-Series/Case Study

Prevalence survey, typically involves small numbers of subjects. No comparison or control group.

Often used to characterize rare diseases.

Cross-Sectional Studies

An observational study in which exposure and disease (or disease outcome) are determined at the same point in time in a given population.

Can show possible exposure/outcomes associations, but cannot determine causality due to lack of temporal data.

AKA Prevalence study

The Four Possible Categories in a Cross-Sectional Study

1. Exposed, have disease

2. Exposed, no disease

3. Not exposed, have disease

4. Not exposed, no disease

Prevalence of Exposure in Cross-Sectional Study Calculation

Prevalence of Disease in Cross-Sectional Study

Cohort studies

Exposure-based subject selection, longitudinal studies. Groups followed for outcomes.

Can be retrospective or prospective.

Steps to Design a Cohort Study (3)

1. Define population of interest

2. Identify exposure status of individuals

3. Follow over time for outcomes

Advantages of Prospective Cohort Study (3)

1. Can ensure good measurements of exposure

2. Can ensure relatively accurate measure of disease onset dates

3. Can add new measures over time as new information on risks and exposures becomes available

Limitations of Prospective Cohort Studies (3)

1. Time-consuming

2. Expensive

3. Biases from nonresponse and loss to follow-up

Retrospective Cohort Study

Investigator uses existing historical data collected previously to identify the population and the exposure status (exposed vs non-exposed groups).

Determine current patient status with no follow up.

Retrospective Cohort Study Advantages (2)

1. Typically quick to establish cohort within an existing database (e.g. EMR, registers, or insurance claims database)

2. Relatively inexpensive

Retrospective Cohort Study Limitations (2)

1. Must rely on “pre-recorded” data or memories

2. Data may not contain accurate diagnoses; chart abstraction or other means of validation may be necessary for outcome (and possibly exposure)

When to Choose Retrospective Cohort Study

Exposure and outcome have already occurred, medical history for both exposure and outcome is considered accurate

When to Choose Prospective Cohort Study

Neither exposure nor outcome have occurred or only exposure has occurred, medical history information is inadequate to capture outcome

Advantages of Cohort Studies (4)

1. Ability to explore many outcomes

2. Permit calculation of direct risk

3. Can study uncommon exposures

4. Can study temporal relationship between exposure and health outcome

Disadvantages of Cohort Studies (4)

1. Expensive

2. Time-consuming (prospective)

3. Not ideal for rare diseases

4. Lost to follow-up

When a Cohort Study is Warranted (4)

1. When the (alleged) exposure is known

2. When exposure is rare and incidence of disease among exposed is high

3. When the time between exposure and disease is relatively short

4. When adequate funding is available

Case Control studies

Cases and controls are selected based on outcome, then sampled for previous exposures.

If the exposure is in fact related to the disease, then it is anticipated that the prevalence of history of exposure in the cases will be greater than the prevalence of history of exposure in the controls

How to Identify Cases for Case-Control Studies (5)

1. Self-report

2. Hospital records

3. Patient charts

4. Laboratory records

5. Surveillance data

How to Select Case-Control Controls

Conceptually, controls should come from the same population at risk of disease from which cases develop.

But practically, controls are often selected to be similar to cases on key factors but without the disease.

Matching in Case-Control Studies

Matching controls to cases on potential confounders or key characteristics (e.g., age, sex, ethnicity)

Recall Bias in Case-Control Study

Information on some past exposures depends on memory of events from both cases and controls, which can be flawed.

Recall bias occurs when the recall is better among cases than controls because of the presence of disease

Case-Control Study Advantages (3)

1. Can measure multiple exposures for the disease

2. More efficient than a cohort study for rare outcomes

3. Costs relatively less and typically can be conducted in a shorter time

Case-Control Study Disadvantages (4)

1. Must have a fairly common exposure

2. Can’t always confirm timing of exposure

3. Higher chance of certain biases

4. Provides indirect estimate of risk