Self Study Cohort Study Design
Stanley Hatch, OD, MPH,  Salus University

STUDY DESIGN

Study Design Graphic 

The above chart shows the types of studies in order of their strength.  Case reports and series are interesting anecdotal reports that may or may not occur in populations.  There are two main categories of epidemiology study types:  1)  DESCRIPTIVE STUDIES as covered in lecture 2 and 2)  EXPLANATORY STUDIES.  Explanatory  studies can be further categorized as EXPERIMENTAL and OBSERVATIONAL.  This self study work unit will address one type of observational study-  the COHORT study.

 

PROSPECTIVE STUDIES are designed in the present and implemented in the future.  Though most any study design (case series, case control, cohort, clinical trial) can be prospective, cohort studies and clinical trials are almost always prospective.

·       Prospective-  potential, future, forthcoming,

RETROPECTIVE STUDIES are designed in the present and assess data or outcomes from the past.  Case reports, case series, and case-control studies are almost always retrospective.

·       Retrospective- previous, past, demonstrated

 

COHORT STUDIES

What is a cohort?

              A group of subjects, usually people, that is observed over time.

              Cohort is derived from the Roman army term of legion.  A cohort would march into battle together but not all would survive.

Cohort

A cohort study assembles a group of subjects, makes baseline measures, then observes the group over time for outcomes.  Survival curves are a good method to see conceptually what a cohort study does.

Figure 1.  Overall survival rate in whites from time of diagnosis of melanoma all ages.  (Source:  SEER)

Figure 1

THE KEY TO COHORT STUDY DESIGN

The cohort is assembled and then followed over time to see what happens.    Subjects need not all be enrolled at the same time (left graphic), or they can be staggered (right table).

1

 

 

 

 

 

2

 

 

 

 

 

 

3

 

 

 

 

4

 

 

 

 

 

5

 

time graphic 

                                                                                               T1         T2         T3         T4         T5       T6       T7                                                                                 

Subject group                 Subject group                              Subjects enter at different times .  Each subject

Time 1                             Time 2                                          followed for same amount of time.      

Baseline measures         Outcome measures                    Baseline measure on entrance, outcome on exit

Feedback     What is the time referent for cohort studies?   Answer

EXPOSURES OF INTEREST HAVE TO BE MADE AT THE BEGINNING

Cohort studies can be designed without specific outcomes in mind.  In the Nurses Health Study, the original outcome of interest was cancer.  The exposure of interest was oral contraceptives.  It was imperative that each subject be categorized accurately as to whether they took contraceptives and how long during the study.  In order to find out if oral contraceptives were associated with cancer, there had to be a sufficient number of subjects who used and did not use them.

Environmental health studies require accurate assessment of exposure prior to enrollment.  To see if lead causes learning disability, blood lead levels, amount of lead paint, traffic density, etc. are measured early in life or even prenatally, and then regularly over the length of the study.  Outcomes, presence or magnitude of learning disability are then determined later- probably school age.

OUTCOMES AND STUDY GOALS CAN BE CHANGED

In a cohort study, you usually can’t go back and collect new exposure data, but outcomes can be added.  This is a big advantage of prospective cohort studies.  If different outcomes appear to be prevalent, you can assign case and control status  and see if the exposure(s) are associated.

EXPOSURES OF INTEREST HAVE TO BE MADE AT THE BEGINNING

Cohort studies can be designed without specific outcomes in mind.  In the Nurses Health Study, the original outcome of interest was cancer.  The exposure of interest was oral contraceptives.  It was imperative that each subject be categorized accurately as to whether they took contraceptives and how long during the study.  In order to find out if oral contraceptives were associated with cancer, there had to be a sufficient number of subjects who used and did not use them.

Environmental health studies require accurate assessment of exposure prior to enrollment.  To see if lead causes learning disability, blood lead levels, amount of lead paint, traffic density, etc. are measured early in life or even prenatally, and then regularly over the length of the study.  Outcomes, presence or magnitude of learning disability are then determined later- probably school age.

Example:

Clinical question and target population:  Does mercury accumulation in fish lead to cancer in workers?

 Mercury Study Design

MULTIPLE EXPOSURES CAN BE MEASURED

Framingham Heart Study 1948-present

 

Exposures

 

Outcomes

Findings  Risk for any outcome

Age

 

Atrial fibrilllation

Increases with age

Gender

 

Congestive heart disease

Increases with males

Blood pressure

 

Coronary artery disease

Increases with higher

Hematocrit

 

Stroke

No association

Blood cholesterol levels

 

 

 

Newer added outcomes

Decreases with increased HDL

Increases with increased LDL

Increases with increased total chol

Weight BMI

 

Hypertension

Increases with obesity

Diabetes

 

Dementia

Increases with diabetes

Smoking

 

 

Increases

Thyroid measures

 

 

No association

Gout

 

 

No association

Alcohol use

 

 

No association

Physical activity

 

 

Decreases with increased activity

 

Source:  www.framinghamheartstudy.org

Complete the table below regarding cohort studies then click on the link to compare your answers.

Study characteristics

Advantage

Disadvantage

Cost

 

Very high

Time

 

 

Common condition

 

 

Rare condition

 

 

Easily defined and measurable exposure

 

 

Easily defined and measurable case condition

 

 

Subject compliance

 

 

Incidence rate

Directly measured

 

Incidence rate ratio

 

 

Sample size

 

 

 

Advantages and Disadvantages of Cohort Studies 

Read Neil Pearce's (Chapter 9) Epidemiology ebook

Read Jones-Jordan and Hoppe's chapter pg 19-26