Epidemiological Study Designs

1
Epidemiological Study Designs

• Chapters 9, 10, 7, 8, 13

2
Epidemiological Study Designs

• C
• Ecological Studies
• C
• Case Control
• C
• Randomized Clinical Trials (RCTs)

3
Observational Study Designs

• Descriptive Studies
• Analytic Studies

4
Experimental Study Designs

• Randomized clinical trials
• Community-based interventions/trials

5
Case Series

• Describes the experience of a single patient or
  group of patients with a similar diagnosis.
• Describes the problem based on person, place, and
  time factors.

6
Ecologic Studies

• Used to determine an association between two
  variables by comparing groups.
• Exposure and outcome measures based on group
  data.
• Groups are formed based on a variable that is not
  being studied. (i.e. geographic location)

7
Cross Sectional Studies

• Researcher assesses exposure and disease at the
  same time.
• Exposure and outcome are based on individuals.

8
Cross Sectional Study
Exposure Status Disease Status
Delinquent Fought
Delinquent, Did not Fight
Sample of High School Students
Prior delinquency (exposure) fighting
(disease/outcome)
Not Delinquent, Fought
Not Delinquent, Did not Fight
9
Case Control Study

• Participants are selected on the basis of disease
  status.
• Are compared based on past exposures.

10
Case Control Study
Time
1970
Present
1970
Exposure
Disease
No Exposure -
Exposure
No Disease -
No Exposure -
11
Case Control Study

• Exposures are obtained retrospectively.
• Cases
• Cases

12
Selection of Cases

• I
• typically preferable to use
• more difficult to find cases
• P
• exposures may be related to duration of illness
  instead of development of the illness.
• the exposure status of prevalent cases will
  differ from that of all cases.

13
Case Selection

• Hospital Cases
• Disease registries
• Occupation based

14
Example of Case Selection

• You want to determine the antecedents to teen
  fatherhood. Who are your cases?

15
Example of Case Selection

• You want to determine incidence cases in a study
  assessing the effectiveness of bike helmets in
  preventing head injuries. Who are your cases?

16
Control Selection

• Do not have to be representative of the general
  population.
• Ideally, chosen at random from the referent
  population.
• Should have or have had the same opportunity to
  become cases.

17
Sources of Controls

• Population controls - random digit dialing
• H
• N
• F
• M

18
Matching

• Characteristics used for matching

19

• Matching on too many variables makes control
  selection difficult.
• Cannot study the characteristics upon which you
  match.
• Unplanned matching-neighborhood controls are
  inadvertently matched with cases on SES.
• Overmatching-occurs when variables related to the
  disease are inadvertently matched upon.

20
Selection of Controls

• You want to determine the antecedents to teen
  fatherhood.
• Who are your controls?
• What variables do you match?

21
Cohort Studies

• Looks to determine the effects of a particular
  exposure.
• All participants are free of disease at the
  outset.

22
Types of Cohort Studies

• C
• prospective cohort
• longitudinal
• N
• retrospective cohort
• historical cohort

23
Concurrent Cohort Study
Time
Present
2025
physically active
Cancer
No Cancer
Defined Population
Cancer
not physically active
No Cancer
24
Nonconcurrent Cohort Study
Time
1985
2005
physically active
Cancer
No Cancer
Defined Population
Cancer
not physically active
No Cancer
25
Selection of Exposed Group

• Select a random sample of the population, survey
  to determine exposure, then stratify into
  exposed/unexposed groups.

26
Selection of Unexposed Group

• Groups should be as similar as possible, with the
  exception of the exposure in question.
• Internal comparison -
• External comparison -

27
Exposure Assessment

• Exposure must be clearly defined and measurable.
• Some exposures may be acute and some may be
  long-term.

28
Outcome Assessment

• Participants cannot have the disease in question
  at the start of the study.
• Measurement of outcome status

29
Cohort Example

• The association of drinking water source and
  chlorination by-products with cancer incidence
  among postmenopausal women in Iowa.
• How do you assess exposure?
• How do you assess outcome?

30
Exposure Assessment - Example
31
Outcome Assessment Example
32
Advantages of Cohort Studies

• Provide a direct estimate of risk (relative risk)
• Establish temporal trends
• Strongest observational design for establishing
  cause and effect relationships.

33
Disadvantages of Cohort Studies

• Requires large sample size
• Not good for studying rare diseases
• Changes over time of diagnostic tests may bias
  results.

34
Comparison of Cohort Studies
35
Nested Case Control

• Combination of a cohort and case control study.
• Utilizing a cohort study

36
Experimental Studies

• Community intervention trials
• Clinical trials

37
Experimental Studies

• Characteristics
• p
• c
• intervention is provided drug, treatment,
  education
• researcher assigns participants to treatment
  groups
• b

38
Randomized Clinical Trial
restenosis
stent rapamycin group
no restenosis
Sample of Patients with CAD
Randomized into new tx and old tx groups
restenosis
stent w/o rapamycin group
no restenosis
39
Randomization

• Each participant has equal probability of
  assignment to intervention or control group.
• Increases the chances that the two groups will be
  comparable in terms of known and unknown
  variables that affect prognosis.

40
Advantages of RCTs

• M
• Balances prognostic indicators across study
  groups
• D
• Assumptions of statistical tests are typically
  met.
• R

41
Disadvantages of RCTs

• E
• Subject exclusions may limit generalizability
• T
• Ethical concerns--who gets treatment, who does
  not
• Noncompliance by study participants
• C

42
RCT Example

• Study Question
• Assess the effectiveness of a violence prevention
  curriculum among children in elementary schools
  in Seattle, Washington
• How do you select your subjects? Control group?
• How do you randomize?
• What are your end points (outcomes)?

43
Subject Selection
44
How to Randomize
45
End Points?