Transcript January 22,2015 - Risky Business

Some of your questions/thoughts
Why are we more afraid of some things than others?
Who is responsible for making sure that numbers and statistics are reported accurately?
Why are Americans so prone to quoting statistics? Does it really make a difference in our
behavior?
How does the epidemiological definition of risk interact with other definitions of risk we have
encountered?
How do we determine which sources are credible and which are not so credible?
How do we make sense
of health statistics?
Overview of Today
1) Go back to exercise from last week and correct mistakes
2) Story about how this course came together
3) Context
4) Medical Data Interpretation Test
Prompt for Journal Entry 2
***Be sure to have an engaging title
Prompt: Talk with a roommate/friend who is not in the class for about 15-20 minutes. Try
to explain the differences between the study designs to your colleague. Try to use examples
beyond what we discussed in class. How did you use the class readings/texts to help with
your discussion? In your post, describe how the interaction went—what types of examples
did you use? What questions did the friend ask? What questions came to your mind as you
tried to explain the concepts?
In the tag, include the entry Journal entry #2
Comment on 1-2 of your peers articles
(Follow word count and Honor Code guidelines as described for Journal Entry 1)
Activity: How do we relate two
quantities?
Some facts:
1) In the year 2000, there were 275,306,000 people in America
2) 151,268
DIED of external (non-disease) related activities
3) The life expectancy of an individual is 76.9 years
Case control study
Odds 
Probability of event occurring
Probability of event not occurring
Depends on a definition of the event
For example, in sports betting…
Probabilit y of NOT winning
Odds 
Probabilit y of winning
ODDS in our example!!!
One year Odds:
◦ Numerator= # of people in America who were alive
◦ Denominator= # people who died from that disease
Ex. Drowning : 275306000 people alive /3248 died of drowning
79,065 people alive/1 died of drowning
Odds of death by drowning is 1 in 79,065
Lifetime odds: One year odds DIVIDED by the life expectancy i.e. 79065/76.9= 1028
Note: this is not the same as the
proportion of deaths.
Proportion of deaths= Death from drowning/Death from all other
causes;
3248 people died from drowning/ 151,268 DIED in America of
external causes= 0.021*100%= 2.1%
2.1% of people who died of EXTERNAL causes died from accidental
drowning and submersion
Another example
Proportions involve comparing parts to a whole and can only be between 0 to 1.
◦ Proportion: Orange squares/all squares= 3/5=0.6
Odds can range from positive infinity to negative infinity.
Odds: orange squares/brown squares= 3/2 or 1.5/1 ; The odds of picking an orange square is 1.5 times the odds of picking a brown
square
Point: It’s tricky to know whether we are
talking about odds or proportions
ASK QUESTIONS!!!
Know what goes in the numerator and denominator of any fraction!!
This is why numeracy is important!
We will practice with this concept
How do we calculate
these risks?
STUDY DESIGN
Components of an epidemiological study
Population---need to define your population in terms of age, race, SES, etc. ; How many people
will you try to recruit?
Disease—need a case definition
Time frame-How long will you follow people?
Study design- How will you collect information? How will you verify information? What
resources do you have?
We will focus on observational study designs: cohort, case-control, cross-sectional
Take EPID 600 if you are interested in this stuff!
Ways to gather information
1.
Case study
2.
Surveillance—Vital Statistics (Looking at death statistics)
3.
Experimental studies (drug testing, randomized control trials, experiments)
4.
Observational studies (cohort, case-control and cross-sectional)
Cohort Study
Cohort Study
Directionality: Always forward
Data collection: Prospective or Retrospective
Cohort study
Disease
Exposure
Yes (E+)
No (E-)
Yes (D+)
a
b
m1
No (D-)
c
d
m0
n1
n0
N
20
A cohort study: heart attack survivors
after 5 years of follow up
Smoke
Quit
Dead
27
14
41
Alive
48
67
115
Total
75
81
156
What is the 5-year risk of dying among the exposed? 0.36
What is the 5-year risk of dying among the non-exposed? 0.17
What is the ratio of the two risks? 2.1
What is the difference between two risks? 0.19
21
The ratio of two risks
Risk Ratio (RR) a.k.a. Relative Risk
CI1
Risk among exposed
RR 

CI0 Risk among not exposed
Risk among continous smokers
RR 
 ??
Risk among those who quit smoking
Interpretation?
Risk of disease in Exposed is 2.1 times the risk of disease in
non-exposed, over 5 years.
22
Case control study
Probability of event ocurring
P
Odds 

Probability of event not ocurring 1  P
Depends on a definition of the event
For example, in sports betting…
Probabilit y of NOT winning
Odds 
Probabilit y of winning
An easy way to calculate OR
Disease
Exposure
Yes (E+)
No (E-)
Yes (D+)
a
b
m1
No (D-)
c
d
m0
n1
n0
N
ad
OR 
bc
24