FHSC282. Principles of Epidemiology and Biostatistics.

Chapter 5
Cohort design

1
Types of epidemiologic studies

Descriptive Analytical

Individual Population
Observational Experimental

Surveillance Ecological
reports study Analytical Randomized
cross-sectional clinical trials
Case report
Community
Cohort interventions
Case series

Case-control
Descriptive cross-
sectional
2
 “Cohort” definition

Cohort = Ancient Roman term for a military unit that marched

together into battles.

A group of people: The people you recruit must follow the same regulations and roles that you will give them. These
people will be experimented with, asked questions and these group will be followed up for a
§ sharing the same specific
experience
period of time.

§ followed up for a specific period of time to one or more

outcomes

Slide modified from Medipiet M1. Marching towards outcome

3
 General design
Population that won't have the disease will be picked
up, a part of that pop will be exposed and part that
won't be exposed

The incidence
of disease
(outcome) is
compared in
both groups.

Study subjects -initially Exposed and unexposed groups are

free of the disease- followed for the same period of time to
are selected based on identify subjects that develop the disease
their exposure status (outcome) of interest in both groups
You can likely detect the onset of the disease, when it started, based on the outcome.

“Follow-up” or “incidence” studies You can detect the new cases

4
Slide modified from Medipiet M1.
 Types of cohort studies
Depending on the timing of data collection, a cohort study can be:

• Prospective (longitudinal or concurrent): Researchers go forward

in time from the present. Subjects are followed up prospectively
through the study period Researchers recruit a group of individuals without a history of heart disease and
follow them over several years to assess their risk of developing cardiovascular
events based on their dietary habits.

• Retrospective (historical): Researchers go back in time to choose

their cohorts Researchers obtain medical records of a group of individuals who were exposed to asbestos in a workplace
several decades ago. They analyze these records to determine the incidence of lung diseases such as
mesothelioma among the exposed individuals compared to non-exposed individuals.

• Ambi-directional: a combination of retrospective (ascertainment

of past exposure) and prospective (follow up) designs
A study on the effects of a specific drug on long-term health outcomes might combine prospective data collection on current
users of the drug with retrospective data collection on past users and non-users. Researchers would then follow the cohort
both forward and backward in time to assess the incidence of various health outcomes.
5
 PROSPECTIVE DESIGN

§ The investigator identifies the original population at the beginning of the study
§ Can have one or more comparison groups
§ Subjects are followed up prospectively through study period

6
DIAGRAM OF A PROSPECTIVE
COHORT STUDY

Modified from Cahn MA, Auston I, Selden CR, Pomerantz KL. Introduction to HSR, May 23, 1998. National Information Center on Health
Services Research and Health Care Technology (NICHSR), National Library of Medicine. 1998. Available at:
7
http://www.nlm.nih.gov/nichsr/pres/mla98/cahn/sld036.htm. Accessed July 30, 2008.
 RETROSPECTIVE DESIGN

PAST TIME PRESENT

DIRECTION OF INQUIRY

§ Both exposures and outcomes have occurred before the study initiation
§ Can have one or more comparison groups
§ Example: investigating a food poisoning among a cohort of wedding guests
8
 WHY CONDUCT A COHORT STUDY?

Cohort studies are conducted:

§ To describe the incidence of specific outcomes over time in a
defined population
§ To analyze the associations between exposure factors and
outcome(s) of interest

Cohort studies are often conducted after a hypothesized

relationship between an exposure and a disease/outcome
that could have been evaluated in a case-control study
A hypothesized relationship between outcome and exposure is done through observational studies, whether at a
populational level or individual level. To better assess the relationship, we do cohort study, then case study
9
WHAT ARE THE CHARACTERISTICS
OF A COHORT?

1. Being at risk of outcome(s) studied

2. Being free of outcome at start of follow-up

10
 Example

In the study of the association between exposure to chemical solvents

and lung cancer among industrial workers:
710 industrial workers were enrolled in the study, of whom 107 were
exposed to chemical solvents.

• How can we ascertain the exposure status of the enrolled subjects?

• How can we ascertain the outcome status of the enrolled subjects?
...... by doing a cohort study

11
 EXPOSURE ASCERTAINMENT
YOU NEED TO VERIFY IF THE EXPOSURE STATUS DOCUMENTED IS VALID OR NOT TO
PREVENT MISCLASSIFICATION.

}It is important to obtain complete and unbiased information on exposure

and outcome of every subject.
}Exposure should be well defined using a clear and standardized definition.
}Depending on the type of exposure, ascertainment of exposure varies.

Example of exposure Exposure ascertainment

Smoking status, reporting having Is exposure reported verbally?
consumed raw meat at dinner did u state it and proved it correctly?

High cholesterol level, anemia, high Is it documented in medical

to prove if REALLY the recruit has high
blood pressure,… tests? cholesterol
hypertension
levels by testing for LDL lvls and

Working hours, position, exposure Is it documented in employment

to chemical or physical factors,… records?
12
 OUTCOME ASCERTAINMENT
• Methods used to assess the occurrence of outcome should be
equally applied to all study subjects (whether exposed or not
exposed) to ensure the validity of the study.
• Methods used to assess the outcome(s) differ according to the
outcome type.
Outcome type Methods used
Birth/Death Vital registries (birth registries, death
certificates,…) to prove if the recruit was born on that day
Medical Hospital discharge summaries,
condition medical files, registries, surveillance,…
Signs and Reported verbally ifseveral
you have doubt, you can also do
tests to ascertain the signs and
symptoms symptoms
13
2x2 contingency table

(a)

(b)

(c)

(d)

14
 Back to the example
From the cohort study of industrial workers, over time: we find
that 3 persons who worked with solvents developed lung cancer
and 104 exposed workers did not. Two cases of lung cancer
occurred among non-exposed workers. The remaining 601 non-
exposed workers did not develop lung cancer.
1) Construct the contingency table

Lung

How can we study whether there is an association

between exposure and outcome? 15
 RELATIVE RISK (RR)

The relative risk RR compares the likelihood of developing the disease

(incidence) in the exposed group compared to that among the non-
exposed group.
Incidence of disease (a or c) among exposed
Incidence among
a/(a+b)
exposed
=Incidence =
c/(c+d)
among not-exposed
Incidence of diseased (a or c) among
non-exposed

16
 Relative risk interpretation

Incidence of disease among exposed a/(a+b)

Relative Risk RR = =
Incidence of disease among not-exposed c/(c+d)
RR=1 No association between exposure and disease
Identical incidence rates among exposed and not-exposed groups
RR>1 Exposure is a risk factor for the disease, increased risk among exposed
The risk of developing the disease is higher among exposed
RR<1 Exposure is a protective factor for the disease, decreased risk among
exposed. The risk of developing the disease is higher among not exposed

Careful!!!
1- the 95% CI for the RR should not contain 1, or be too wide
2- the distribution of the cell count is significantly different between exposed
and not-exposed (p-value <0.05)

17
 Relative risk (RR): example

Relative risk RR
Incidence among exposed= 3/107
Incidence among
3/107
----------------------------------------------- ---> 8.48 >>1
exposedamong non-exposed= 2/603
Incidence

=Incidence = = 8.48
RR>1 Exposure to the solvents is a RISK FACTOR, which
among
occurs not- more frequently
8 times 2/603
with people exposed to
exposed
solvents than those not exposed to the solvents.

The risk of developing lung cancer is eight times greater

among workers who were exposed to solvents than among
those who were not exposed to solvents

18
 ATTRIBUTABLE RISK (AR)
marks the extent of a certain risk factor on the cause of the disease, and it can be prevented if
the exposure was eliminated.

The Attributable Risk (AR) in a cohort study refers to the disease incidence
that can be attributed to a specific exposure, and that can be prevented if
the exposure in question is eliminated (assuming causal association).
It is calculated as the difference between the incidence among exposed and
that among non-exposed group:
AR = Iexp - Inon exp
Back to the example:
• Incidence rate of lung cancer among exposed: 3/107=0.028
• Incidence rate of lung cancer among non-exposed : 2/603=0.003
• Attributable risk: 0.028-0.003=0.025 or 25 per 1,000
The incidence rate of lung cancer associated with exposure to chemical solvents.
For every 1000 population, 25 new cases of lung cancer can be prevented if
exposure to chemical solvents is eliminated. ifnew
u prevent the exposure to solvents, the 25
cases per 1000 won't happen
19
 Consider the previous example to answer the
following questions. Indicate whether it is an
advantage or a limitation of the cohort design

• Does the collected information on outcome refer to

prevalence or incidence? Incidence
• Is it useful to conduct a cohort design to study a rare health
outcome? No CUZ the ppl who have the exposure and the outcome would be soo small, even
negligible, do we can't use cohort

• Is it useful conduct a cohort design to study a rare exposure?

Yes 2 or 3 exposure cases are enough

• How many outcomes can we study? Multiple outcomes

of a single exposure
!!!!!!! In a cohort study that measured all three exposures (smoking, low vitamin D intake, and severe cold weather) at baseline,
the number of different outcomes that could be examined is essentially infinite. This is because each exposure has the
potential to impact various health outcomes, and there is continuous variation within each outcome. Given the continuous
variation in exposure levels, individual susceptibilities, and complex interactions between exposures and outcomes, the
number of potential outcomes that could be examined in a cohort study measuring these three exposures is virtually infinite. It
highlights the importance of carefully selecting and defining the outcomes of interest in order to capture the full range of health
effects associated with the exposures under investigation.!!!!!
20
Consider the previous example to answer the
following questions. Indicate whether it is an
advantage or a limitations of the cohort design

• What benefit(s) do we have when conducting a study with

exposure preceding the outcome? Temporal relationship 21
can be established and
determine
the onset of
the disease
• What might be inconvenient in following people up for a certain time?
Attrition or loss to follow-up

• If this study had a retrospective design, do you think employment

records would have complete information about occupational
exposure? No
 COHORT STUDIES (CONT.)

ADVANTAGES DISADVANTAGES
§ Permit direct § Expensive and time
observation of risk. consuming.
§ Exposure factor is well § Complicated and difficult
defined. to carry out.
§ Can study exposures that § Subjects may be lost to
are uncommon in the follow-up during the
population. course of the study.
§ The temporal § Exposures can be
relationship between misclassified.
factor and outcome is
known.
22
 ADVANTAGES

§ As incident cases are identified as they occur, cohort studies

provide the most direct measurement of the risk of
developing disease
§ Multiple outcomes of a single exposure can be studied
§ Selection bias is minimized since exposure is well-defined
§ Since exposure precedes the outcome, temporal relationship
(between exposure and disease) can be established, providing
useful information to establish causation CAUSALITY CAN BE ESTABLISHED
§ Cohort studies are suitable for studying the late or chronic
effects of rare exposure (occupational, environmental,
industrial accident,…)
23
 LIMITATIONS

§ Expensive and time consuming, compared to other

analytical designs
§ Loss to follow up: Higher attrition rate with longer study
duration
§ Not convenient for rare disease
§ Reporting bias, when the exposure and/or outcome is
subjective difficulty to accurately recall the exposure or any misclassification of exposure or outcome

§ In retrospective cohort designs: past records might not

contain complete information about exposure
Sometimes exposure status is not clear when it is necessary to go back in time and use
whatever data is available, especially because the data being used was not designed to 24
answer a health question..
EXAMPLE OF COHORT STUDY:
FRAMINGHAM HEART STUDY
¡ 1948-Present
¡ A prospective cohort study studying the risk factors for heart disease
in an initial sample of 5100 residents of Framingham, Massachusetts
http://www.framinghamheartstudy.org/
¡ “Over the years, careful monitoring of the Framingham Study
population has led to the identification of the major CVD risk factors
- high blood pressure, high blood cholesterol, smoking, obesity,
diabetes, and physical inactivity - as well as a great deal of valuable
information on the effects of related factors such as blood
triglyceride and HDL cholesterol levels, age, gender, and psychosocial
issues.
¡ Although the Framingham cohort is primarily Caucasian, the
importance of the major CVD risk factors identified in this group have
been shown in other studies to apply almost universally among racial
and ethnic groups, even though the patterns of distribution may vary
from group to group. ”

25
BHOPAL STUDY

26