Transcript Do Cancer Cluster Really Exist?

Do Cancer Cluster Really
Exist?
Definitions Are Important
► Robert
N. Hoover, M.D National Cancer
Institute defines a cancer cluster as
 “the occurrence of a greater than expected
number of cases of a particular disease within a
group of people, a geographic area, or a period
of time."
Do Cancer Cluster Exist?
►
If we mean any group of cancers that have
a common cause - Yes

Clusters have been found in people who share the same
ancestry, in people who share the same occupation, and in
people who share some other common trait such as smoking.
►
If we mean an unusually high number of
cancer cases found in a defined geographic
area, assumed to be caused by the
presence of some carcinogen in the
ambient environment. – Probably not
►
Even if such a cluster exists,

it is virtually impossible to properly validate it or to distinguish it
from clusters that are due to purely to chance
What Do The Experts Say
Paraphrased quotes from witnesses
testifying in 2001 before the Senate
Cancer Coalition.
►





Robert N. Hoover, M.D., Director, Epidemiology and
Biostatistics National Cancer Institute
David W. Fleming, M.D., Centers for Disease Control
and Prevention
David A. Savitz, Professor and Chair in Epidemiology,
University of North Carolina;
Sue-Min Lai, Ph.D., Associate Professor for
Preventive Medicine, University Medical Center,
University of Kansas;
William Wright, Ph.D., Chief, Cancer Surveillance
Section, California Department of Health Services.
► Robert
N. Hoover
 Investigating small disease clusters usually does not
yield definitive results because cancer resulting from an
environmental exposure may take many years to appear
and populations in a particular geographic setting are
often in flux.
► David
Fleming
 85-90 per cent of cancer cluster investigations do not
find evidence of increased cancer risk.
 Purely random events also cluster by chance, and
statistical tests cannot separate observed clusters
caused by chance from those due to an unrecognized
common cause.
► David
A. Savitz
 Studying cancer clusters over short periods of
time and by locations have not been successful
because 1) most environmental exposures
happen very slowly, often over decades, and
thus, an elevated cancer rate for 6 months or 1
year is not likely to point to some specific,
identifiable cause, 2) causes of cancer operate
over many years, so that clustering at the time
of diagnosis does not correspond to clustering
at the time when they were actually caused, 3)
exposures that tend to cluster in time and space
may well not be the dominant causes of cancer
► Sue-Min
Lai
 Most cancer cluster investigations conducted to date
have not reached the feasibility stage. The following
points make disease cluster investigations problematic.
Geographic boundaries for a cluster study are influenced
by the observed result and may be in error. The effect
of population emigration and immigration will dilute
disease clusters with a specific localized cause and
diminish their detection by geographical studies.
► William
E. Wright
 Studies of cancer clusters to identify an environmental
cause of cancer in a selected neighborhood are almost
certainly doomed to failure
Fear caused by the perception of
cancer clusters is real
To quote David Fleming
►

There are few things more frightening to a
community than the notion that an unseen
killer is lurking in its midst. Such is the fear
inspired by the belief that cancer is
disproportionately striking down the people
in a town.
Southside High School
One of many cancer cluster scares
around the country
►
"From the high rates of cancer at a high
school in Elmira to the alarmingly high
breast cancer rates on Long Island,
communities in New York and around the
country people want to know what's
making them sick - it's time for us to step
up our efforts to find the answers," Clinton
said last week.
Southside Makes National News
Poisoned Schools?
Schools Near Toxic Waste Sites are Making Children Sick
By Michele Norris
An Elmira mother and her son will appear at 9 a.m. today on MSNBC's "The
Hour Show" to discuss the state's cancer investigation at Southside High
School.
The Patros family has also been contacted by ABC's "Good Morning America"
to discuss their views about the investigation, said husband and father, Andy
Patros, who remained in Elmira.
►
►
►
►
►
►
I am beyond furious after attending a meeting at Southside High
School on May 2. I was so insulted by the "answers" given by the
panel of "experts" that ran the meeting that evening. We were told
everything was safe, but there were no answers about why children
are getting sick.
Why are they developing cancer and other illnesses that normally could
not be found in such concentrations in an area as small as Elmira's
Southside?
The community itself needs to become so much louder. We cannot
allow any cover-ups. What was in a report that was supposedly
detailed in 1997? In the meantime, children are developing thyroid
problems, diseases of the bone marrow, cancer, skin diseases and
stomach and colon problems. And those are only the problems I heard
that evening.
How many children are being treated for what may appear to be a
fairly normal problem, but may have been brought on by whatever is in
the soil, water and air?
And what about the surrounding neighborhoods? How many people
have become sick and have died needlessly?
There has to be an answer. Until we as a family can figure out a
suitable alternative, we are taking our daughter out of school near the
end of her junior year.
SOUTHSIDE HIGH SCHOOL
Draft for Public Comment
U.S. Department of Health & Human Services Public
Health Service
Agency for Toxic Substances and Disease Registry
►
“The NYS DOH looked at the overall time period
since the opening of Southside High School, and
did not find an unusual pattern of cancer among
current and former students, or among children
living in the area served by the school. The
numbers and types of cancers diagnosed were
similar to what would be expected for this age
group.”
Cancer Maps, A New Tool To Scare
The Daylights Out Of People
Why Is This Seemingly Simple Task
So Hard
►
Problems Defining the Relevant Cancer Cases
Do you include?
►
All cases currently living in a geographic area


►
Everybody who has ever lived in a geographic area

►
What about newcomers
What about people who have moved away
How far back into the past is relevant
Example, a person is living in an area with a high rate of
cancer and they moved there 5 years ago from an area with
a high rate of cancer. To which population/area do you
assign their cancer?
Problems Defining The Relevant
Geography
► Using
predefined political or administrative
geographic areas such as zipcodes makes
no sense as the geographic area used to
determine the affected population has no
relationship to the suspected exposure
Examine the map below you see adjacent
zipcodes with high and low rates
How Do You Handle This Situation
Zipcode
17643
Zipcode
17644
Is This
►
2 cases in zipcode 17644 and 3 cases in zipcode
17643 or 5 cases clustered in their own area
►
If 5 cases clustered in their own area, how do you
draw the area to determine the population to be
associated with the cancer cases to calculate the
rate.

May get substantially different results using the dashed as
opposed to the dotted geographic area and these are only
2 out of an infinite set of areas you could draw.

If after the fact you draw areas around clusters designed
to pack them in, you naturally get exaggerated rates.
From the National Cancer Institute
► Before
a cluster can be considered "true,"
epidemiologists must show that the number of
cancer cases which have occurred is significantly
greater than the number of cases that would be
expected, given the age, gender, and racial
distribution of the group of people at risk of
developing the disease. However, it is often very
difficult, if not impossible, to accurately define the
group of people who should be considered "at
risk."
From the National Cancer Institute
► One
of the greatest pitfalls of defining clusters is
the tendency to extend the geographic borders of
the cluster to include additional cases of the
suspected disease as they are discovered. The
tendency to define the borders of a cluster on the
basis of where one knows the cases are located,
rather than to first define the population and then
determine if the number of cancers is excessive,
creates many "clusters" that are not genuine.
From Me
► There
is no scientific way to select the
cancer cases to be included in the cluster or
to define the population or the geographic
area to be used to define the cluster.
► Inevitably
the cluster is measured using
arbitrary methods.
Lets Look More Closely At
Random Distributions Over Space
If you throw confetti in the air and it lands on
the floor does it look like this
Most people understand that the confetti will fall
in clumps with bare spaces between the clumps.
Why should randomly distributed cancer cases
look any different?
To illustrate, try this Excel exercise to generate and
examine random distribution over space. Use
Tools/Data Analysis/Random Number Generator to
generate a Uniform distribution of 2 sets of 400
random numbers between 0 and 100
Make an xy scatter diagram using one variable as the
x axis and the other as the y axes. You now have
400 randomly distributed dots over a 100 by 100
graphical space. Below is one result
100
90
80
70
60
50
40
30
20
10
0
0
10
20
30
40
50
60
70
80
90
100
100
90
80
70
60
50
0
40
30
20
10
0
0
10
20
30
40
50
60
70
80
90
100
Like confetti thrown on the floor clustering and
linear structures are the expected result of a
random distribution over space
►
Easy to draw clusters large and small
►
Easy to find linear structures in the data. The
diagram shows only a few of the many clusters
that could be identified.
►
Not surprising that there will be instances where
clusters center around a factory site or some
other assumed source of pollution, or where
linear clusters appear to follow a river, highway,
pipeline, or transmission line.
We can add grid lines to the diagram to create
large and small geographic areas. How will
deviation from the expected value be affected by
the size of the geographic area?
100
75
50
0
25
0
0
25
50
75
100
As geographic areas get smaller, deviation
from the expected get larger and more
common
100
90
80
70
60
50
0
40
30
20
10
0
0
10
20
30
40
50
60
70
80
90
100
Frequency Chart
25 Cells
6
5
Frequency
4
3
2
1
0
38%
44%
50%
56%
63%
69%
75%
81%
88%
94% 100% 106% 113% 119% 125% 131% 138% 144% 150%
Percent Above/Below Expected
Frequency Chart
100 Cells
25
Frequency
20
15
10
5
0
0%
25%
50%
75%
100%
125%
150%
Percent Above/Below Expected
175%
200%
225%
►
With 25 cells only 1 of 25 cells, 4% of the
cells, is 150% more than the expected
value of 16.
►
With 100 cells 12 of 100 or 12% of the
cells, are 150% more than the expected
value of 4.
►
With 25 cells the largest difference is
150% of the expected value. With 100
cells the largest difference is 225% of the
expected value.
As the geographic areas get
smaller the number of areas that
deviate significantly from the
expected increases and the
magnitude of the deviation
increases.
How To Identify A Real Cancer
Cluster
► If
A causes B then wherever you find A you
will tend to find B.
► If
A causes B then more A will tend to cause
more B
► Not
logical that minute exposure to a
chemical in one location causes cancer but
massive exposure to the same chemical in
another location doesn’t.
► Not
logical that short term exposure to a
chemical causes cancer but long term
exposure doesn’t.
► In
short, you would see a pattern that
would be consistent and would persist over
time and space.
► This
is not what is seen in virtually every
neighborhood cancer cluster scare.