Transcript Epidemiologic Tables and Transforming Variables

Day 4
Epidemiologic Tables
and Transforming Variables
Incidence Rates
• Is there an association between breast cancer
and exposure to multiple X-ray
fluoroscopies (Howe, 1982)?
Exposed
Breast cancer cases 41
Person years
28010
Unexposed
15
19017
Incidence Rates
iri 41 15 28010 19017
Cohort Studies
• Is there an association between mortality
and low birthweight among infants (defined
by birthweight < 2500g) in New York City
in 1974?
Died
Lived
Low Birthwieght
618
4597
Normal Weight
422
67093
Cohort Studies
csi 618 422 4597 67093
Case Control Studies
• Is there an association between average
daily alcohol consumption and esophageal
cancer? (Tuyns, 1977).
Cases
Controls
80+ g
96
109
0-79 g
104
666
Case Control Studies
cci 96 104 109 666
Using cc or cs on “real data”
• What is the risk of death for individuals
who have septicemia as a complication of
abdominal aortic surgery compared to those
who do not?
cc death septicem
cs death septicem
Matched Case Control Studies
• Are individuals who drink 6 or more cups
of coffee per day at increased risk of
myocardial infarction?
Cases
6+ cups
<6 cups
controls
6+ cups
< 6 cups
8
8
3
8
Matched Case Control Studies
mcci 8 8 3 8
Transforming Variables
What “transformation” should we use to
normalize
los?
totchg?
icuday?
age?
Transforming Variables
gladder
sktest
[ladder]
Conditional Statements
• if
– comes before the comma
– e.g.
• if age<50
• if age>=50
• if age==50
– ci los if age<50
– ci los if age<50, by(sex)
Conditional Statements
• and (&), or (|)
– ci los if age>50 & age<70
– ci los if age<50 & sex==1
– ci los if septic==1 | reintub==1