Epidemiology in Action

1
Epidemiology in Action

• Unit III Measurement and Hypothesis Testing
• Measuring Disease Frequency
• Proportions and rates
• Comparing Disease Rates
• Measures of association used to
• test hypotheses

Activity Show What You Know
2
Epidemiology in Action
2
Measuring Disease Frequency
Disease Yes Disease No TOTAL
Exposed Yes A B AB
Exposed No C D CD
TOTAL AC BD ACD
In any study, one of the first steps is to
gather data. The chart above is called a Chi
Square, and it is the basic statistic used to
test for the significance of any differences
noted in distribution of disease or risk. Rates
needed for epidemiological analysis can be
calculated from this basic table.
3
Epidemiology in Action
3
Disease Yes Disease No TOTAL
Exposed Yes A B AB
Exposed No C D CD
TOTAL AC BD ACD
Chi Square
A the who are exposed and have the
disease Bthe number who are exposed and do not
have the disease Cthe number who are not exposed
and have the disease Dthe number of individuals
who are not exposed and dont have the
disease ABCD the total number in the
population being studied ABthe total number
exposed CDthe total number not exposed ACthe
total number with disease BDthe total number
without disease
4
Epidemiology in Action
4
Measuring Disease Frequency In epidemiology,
occurrence of a condition must be related to the
population at risk. Common measures of disease
frequency are Prevalence rate and Incidence
rate. Prevalence rate the proportion of a
population that are cases at a point in
time. For example, what PROPORTION Of the the
U.S. population Has HIV/AIDS right now? For
current HIV/AIDS statistics, check the link
below Link CDC - Division of HIV/AIDS
Prevention - Basic Statistics
Source LC,/1985,David Lance Goines.
5
Epidemiology in Action
5
Prevalence Rate Prevalence Rate the proportion
of people in a population who have a
condition/attribute at a specified
time. Prevalencenumber of existing cases divided
by total population For example, a hypothetical
survey of 2477 persons showed that 310 were heavy
smokers. The prevalence of the condition, heavy
smoking, was the number of people who smoke
heavily at a specified time divided by the total
population 310 x 100 12.5
2477

6
Epidemiology in Action
6
Incidence Rate rate at which NEW cases occur in
a population during a specified time period.
For example, what is the Incidence Rate for
new cases of AIDS in the U.S. during
the last year? For current HIV/AIDS statistics,
check the link below Link CDC - Division of
HIV/AIDS Prevention - Basic Statistics
Source LC,/1985,David Lance Goines.
7
Epidemiology in Action
7
Incidence Rate rate at which NEW cases occur in
a population during a specified time period. For
example, in a hypothetical study of 2390 college
women, 482 drank two or more glasses of alcohol a
day. Twenty-seven of the 482 two-a-day women
drinkers over a specified time period were in car
accidents related to alcohol use. The incidence
rate of alcohol-related car accidents
was Alcohol-related car accidents divided by
two-a-day drinkers. 27 482 X 100
5.6
Wartenberg, Daniel. (2000). Analytical Methods
Used by Epidemiologists.. FACSNET Reporting
Tools online, September 27, 2002
8
Epidemiology in Action
8

• Measures Used to Show Statistical Correlation
• Is this disease increasing in incidence?
• Is it more frequent in my area?
• Does its incidence correlate with
• some suspected cause?
• Have things changed since we
• introduced control measures?
• These are all questions we answer by setting two
  sets of rates side by side and comparing them.
  Three common measures we use to compare rates are
  
• Relative Risk, Odds Ratio, and Attributable Risk

SourceLC/Jerome Henry Rothstein, 1936-38
For information on cancer and the link to
nutrition go to Epi Research Nutritional
Epidemiology
9
Epidemiology in Action
9

• Relative Risk
• Calculated to identify differences in disease
  rates between exposed and unexposed groups. It
  is the most common measure of association used by
  epidemiologists, because it can be estimated by a
  wider range of study designs.
• Relative Risk Rate of condition among exposed
  divided by rate of condition among unexposed
• For example, if lung cancer mortality rate
  among smokers is 131 per 100,000, and the lung
  cancer rate among non-smokers is 11 per 100,000,
  then
• Relative Risk 131/1111.9
• Relative Risk 1 means no difference
• Relative Risk gt 1 shows existence of an
• association between exposure and disease,
• and there is an association between
• smoking and lung cancer.

10
Epidemiology in Action
10

• Odds Ratio
• Calculated to identify likelihood of exposure
  to risk when comparing two groups, one with and
  one without disease.
• Odds Ratio Exposure odds in disease group
  divided by exposure odds in non-disease group
• If the prevalence of smoking among lung cancer
  patients is 95 per 100, and the prevalence of
  smoking among people without lung cancer is 25
  per 100, then
• Odds Ratio 95/253.8
• Ratio 1, no association
• Ratio gt 1, association between
• exposure and disease
• ..And there is an association
• between lung cancer and smoking.
• Read about some interesting studies on youth
  and smoking at
• Epidemiology of Youth Drug Abuse - Research
  Findings 5/01

11
Epidemiology in Action
11

Based on the Odds Ratio formula, what is the Odds
Ratio for each disease status in the famous
smoking study?
Smoking and Carcinoma of the Lung
Disease Status of smokers of nonsmokers P-value
Males Lung cancer 647 (99.7) 2 (0.3)
Males Controls 622 (95.8) 27 (4.2) 0.00000064
Females Lung cancer 41 (46.7) 19 31.7)
Females Controls 28 (46.7) 32 53.3 0.016
Doll R. Bradford, Hill A. Smoking and carcinoma
of the lung preliminary report. British
Medical Journal 1950, 2 739-748. Epidemiology
of Youth Drug Abuse - Research Findings 5/01
12
Epidemiology in Action
12
Attributable Risk Attributable Risk is
the proportion of exposed cases that would not
have gotten the disease if they had not been
exposed. For example How many MORE people will
get cancer in a town that has arsenic contaminati
on due to industrial pollution? To find out more
about the link between arsenic poisoning and
cancer link to UC Berkeley Arsenic Research
Program

Source LC/1951
The Attributable Risk would be useful to know in
this case, because we may then be able to
attribute this excess of disease to the risk
factor were studying.
13
13
Epidemiology in Action
Attributable Risk If we subtract the rate of
disease in a population that does not have a risk
factor from the rate of disease in a population
that DOES have a risk factor, we get the
Attributable Risk. Rate of Disease/
exposed minus Rate of Disease/ not exposed
Attributable Risk. For example Lets say that
in a town exposed to arsenic from pollution, we
find ten cases of one type of cancer in 2000
people over a 10-year period of observation. The
rate, then, is 4 cases per 10,000 people per
year. Comparing the rate in this town to that
estimated for the entire United States, we find
that this town had an excess of four cases per
10,000 people per year (5 cases in the town minus
one case expected based on national data, per
10,000 persons per year. We may then attribute
that excess to arsenic from pollution. Wartenberg,
Daniel. (2000). Analytical Methods Used by
Epidemiologists.. FACSNET Reporting Tools
online, September 27, 20

Source LC/1951
14
Epidemiology in Action
14

• Show What You Know Assignment
• STEP I Using the statistical databases we have
  become familiar with so far, find REAL examples
  of the following things. Then, explain what that
  statistic MEANS in terms of the data you are
  looking at.
• Note you can find them already calculated, or
  you can construct them by working
  backward/forward with a set of data. For all
  examples, document correctly where the data comes
  from.
• A chi square. You may find one already in place,
  or you can construct one working backward from
  data you find.
• A prevalence rate calculate, if necessary.
• An incidence rate calculate, if necessary.
• An example of relative risk calculate, if
  necessary.
• An example of odds ratio calculate, if
  necessary.
• An example of attributable risk calculate, if
  necessary.

15
Epidemiology in Action
15
Show What You Know Assignment STEP II Create
two hypothetical studies, in which you use all of
these statistical measures (you dont have to use
all six in each study- use each at least once in
one or the other of the studies). For your
hypothetical study, identify the following
Target population, type of study.

Useful sources for data that measures frequency
Office of Population Research, Princeton
University PopNet Population Statistics