Epidemiology Exam

1
Epidemiology Exam 3 Review

• By
• Dr. Shiaw-Fen Ferng

2
The Purpose of All Study Designs

• Exposure/intervention
• Outcome

To investigate the relationship between exposure
and outcome
3
1. Rate Calculation

• Incidence ratePrevalence rate
• PI x D

4
Cohort Study Design

• Source Population (healthy)

5
Cohort Study Design
6
How Do You Put All Information Into the 2x2 Table?
7
Basic Analysis of Cohort Study
Disease
Non-disease
Total
Exposure
a
b
ab
Non-exposure
c
d
cd
ac
bd
N
Always consider bad factor as the Exposure.
IR(E) a/(ab)IR (non-E) c/(cd)
8
When the EXPOSURE is not the Risk Factor

• If the EXPOSURE has nothing to do with the
  OUTCOME (disease)
• Then you should expect to see the IR (E) is
  similar to IR (non-E)
• Because the chance of getting sick are similar in
  both group (random effect)

9
RR in Three Possible Results
a
IR (E)
ab

RR (rate ratio)
c
)
IR (
E
cd
RR gt1 ? IR (E) gt IR (E) ? E risk factor RR 1 ?
IR (E) IR (E) ? E no effect RR lt 1 ? IR (E) lt
IR (E) ? E protective
factor
10
RR Example

• IR (E) (not participating Health Education
  program) 8 per 1,000 per year
• IR (E) (participating Health Education program)
  2 per 1,000 per year
• RR 8/2 4
• Students who did not participate health education
  program has 4 time risk of getting STD than those
  who participated

11
Another Application of Incidence Rates (1)

• How many STD can be prevented, if all students
  attend the STD health education program?
• Since everyone has the chance to have STD
• the baseline rate of STD is the IR of
  non-exposure group
• The contribution of the health education to
  prevent the STD
• The contribution is the difference between two
  rates

12
Another Application of Incidence Rates Rate
Difference (RD)/ attributable risk (2)

• The contribution is the difference between two
  rates 8 per 1,000 per year - 2 per 1,000 per
  year 6 per 1,000 per year
• For every year, among 1000 students, 6 STD cases
  can be prevented, if the they attended the health
  education program
• For a school district with 5,000 students, the
  total STD can be prevented will be 6 per 1,000
  per year x 5,000 30 per year

13
Another Application of Incidence Rates (3)

• How many STD can be prevented in a school
  district with 5,000 students(RD 6 per 1,000
  per year)
• 6 per 1,000 per year x 5,000 Rate X total
  population 30 per year

14
Estimate The Total Existing Cases In A Community
After Conducting a Cohort Study

• Before start a cohort study, need to conduct a
  screening test to confirm all participants are
  free of disease of your study interest
• Those who have the disease will be excluded from
  participating the study.
• These subjects will be counted as EXISTING
  CASES
• Knowing the total population (the total
  number of your samples)
• You can calculate PREVALENCE RATE
• TOTAL EXISTING CASES IN A COMMUNITY
  PREVALENCE RATE X TOTAL POPULATION

15
Set Up Null Hypothesis for a Cohort Study
16
Null hypothesis H0 RR1

• H0 There is no association between exposure and
  disease
• So when the exposure has nothing to do with the
  disease ? then whether a person has been exposed
  to the risk factor or not ? their chance to get
  disease will be the same
• It will translated into IR (E) IR (non-E)
• Remember always set up Exposure as Bad Factor
• Not to participate the health education
• Exposed to Environmental pollutants
• Exposed to Risk Factor at work

17
Null hypothesis H0 RR1

• Since IR (E) IR (E) so

18
Statistic Test for H0 RR1
19
H0 RR1

• When 95 confidence interval of RR does not cover
  1
• statistically significant
• reject H0
• I.e. Accept Ha RR ? 1
• exposure associate with disease

20
H0 RR1

• When 95 confidence interval of RR does cover 1
• not statistically significant
• accept H0
• ie. H0 RR1
• exposure does not associate with disease

21
H0 Test Based on p Value

• From Epi Info Yates Corrected p value (more
  conservative)
• P value gt 0.05 ?Accept the H0
• P value lt 0.05 ?Reject the H0
• Using p value for statistic test will have the
  same result as using 95 CI

22
Epi Info Results
A RR 1.34, 95 CI (0.73 to 2.38)
A
B
B Yates corrected p value 0.388 gt 0.05? accept
H0
23
When to Use Cohort Study? Advantage/Disadvantage
of Using Cohort Studies

• Factors need to be considered
• Incidence Rate
• Incubation Time
• How much do you have?
• How much time do you have?
• What do you want to know?

24
Case-Control Study
25
Case-Control Study Design
26
Case-Control Study 2 X 2 Table
Approximating the Rate Ratio
Disease
no Disease
Total
Exposure
a
b
M1
Non-Exposure
c
d
M2
N1
N2
T
N1 (subjects with disease, collected at the
beginning, then you ask their exposure
experience) N2 (subjects without disease,
collected at the beginning)
You do not know how many people are at risk to
begin with, no Incidence rates can be calculated.
27
Disease Odds Ratio (OR)
Interpretation of OR is the same as RR
28
Hull Hypothesis Setting and Test

• Same as Cohort Study except need to read OR data
  from EPI Info
• H0
• Decision making based on
• 95 confidence interval
• Yates corrected p value

29
Case-Control Study Calculation Results From Epi
Info
A the OR 2.55 and 95 CI (1.27 to 5.12)

• B

A
B
B Yates corrected p value0.0067 lt 0.05
30
When to Use Case-Control Study?
Advantage/Disadvantage of Using Case-Control
Studies

• Factors need to be considered
• Incidence Rate
• Fatality Rate
• Incubation Time
• How much do you have?
• How much time do you have?
• What do you want to know?

31
 Cross-Sectional Study

• All data analysis, statistic test are same as
  CASE-CONTROL studies
• Except cross-sectional study is a survey
• No need to specify your subject at beginning of
  your project
• Do not need to know they have disease or not
• Do not know they have been exposed or not
• Ask both exposure and disease status when you do
  your survey
• Use cross-sectional study to explore the status
  of a disease then decide for the next step of a
  research

32
Experimental Epidemiology

• advantage of randomization
• Control the confounding factors
• how to prevent the biased assessment of clinical
  trial
• Double blind study design

33
Selection of Different Study Designs (1)

• a special disease which occurs among a community,
  only male patient can survive.  If you want to
  investigate the relationship between exposure and
  outcome (disease occurrence), what kind of study
  design should you use? why?

34
Selection of Different Study Designs (2)

• the incidence rate of a special disease is 1.2
  per 1,000,000 per year (considered as a rare
  disease)
• If you want to investigate the relationship
  between exposure and outcome (disease
  occurrence), what kind of study design should you
  use? why?

35
The End

• Remember to do the Course Evaluation!!!
• Get the link from Calendar.