Transcript Educational objectives On completion of your studies you should

Variation in disease by
time, place and person:
A framework for analysis
Raj Bhopal,
Bruce and John Usher Professor of Public Health,
Public Health Sciences Section,
Division of Community Health Sciences,
University of Edinburgh, Edinburgh EH89AG
[email protected]
Educational objectives
On completion of your studies you should
understand:
 That virtually all diseases vary in their incidence
and prevalence over time, across geographical
areas and between population subgroups.
 That apparent disease variations can be artefacts
of errors or changes in data collection systems.
 That variations must be analysed systematically
to check that they are real, and not illusory.
 Real variations are driven by environmental and
social change over the short term, with a genetic
contribution in the long term.
Educational objectives
On completion of your studies you should
understand:

Real variations help in understanding the
causal pathways of disease.

Study of clusters and outbreaks, which
reflect abrupt changes in disease frequency,
may yield both causal knowledge, and
information to control the public health
problem.

Real variations help to develop and target
health policy and health care.

Variations generate observations of
associations, which in turn spark causal
hypotheses.
Educational objectives
On completion of your studies you should
understand:
 Diseases wax and wane in their population
frequency- an axiom of epidemiology
 Diseases patterns have undergone massive
change in incidence within the last 50 to 100
years
 A systematic mode of analysis of disease
variations is vital
 Ensure that observations of variation are real,
and not illusory products of data errors and
artefacts
Exercise

Reflect and comment on the following
graphs before reading on
UK trends in cardiovascular
disease mortality
Standardised mortality ratios for CHD
by sex for selected countries of birth,
1989/92, England/ Wales
Source: Wild, S and McKeigue, P BMJ 1997: 314; 705-10.
Exercise: Benefits of
studying variations


What potential benefits are there from
investigation of changes in disease
frequency?
Is a decline in disease as worthy of
investigation as a rise?
Exercise: Reasons for
variation


Why, in general terms, do diseases vary
over
time, between places and between
subgroups of the populations?
What is the relative importance of genetic
and environmental influences in bringing
about population differences in disease
patterns?
Genetic and environmental
influences





All humans belong to one species
Genetic variation between populations is small
Genetic change arises from a number of
processes including genetic drift and genetic
mutation
Changes in disease frequency in large
populations occurring over short periods of
time are almost wholly environmental
Public health paradox: for populations the
environment is the dominant influence on the
pattern of disease, for individuals genetic
inheritance may be equally or more important
Transitions and disease
variations




A decline in birth rates and death rates leads to
a shift in the age distribution of the population,
with the average age increasing (the
demographic transition)
Industrialisation, wealth creation, ageing of the
population and the other profound changes
alters the pattern of diseases (the
epidemiological transition)
These transitions are reversible
International differences in disease patterns
and disease variations of migration populations
can be conceptualised as a result of
populations being at different stages of the
demographic and epidemiological transition
Variations and associations:
real or artefect?



When changes in disease frequency are
natural, or real
This is an experiment of nature, posing a
challenge to science
Underlying reasons as discussed above
are often exceedingly difficult to pinpoint
Variations and associations:
real or artefect?




First step is to exclude artefact
The second is to develop a hypothesis
stated as an association
The third is to design a test of the
hypothesis
The fourth is to assess the results in
relation to frameworks for causal thinking
Variations and associations: real
or artefect: CHD (see slide)




In the U.K. coronary heart disease mortality
rates rose steadily in the 20th century until the
1970's when they declined
first, demonstrate the association of disease
rates and time periods
attempt to explain the time trend by developing
our understanding of social, environmental and
lifestyle changes over these time periods
test specific hypotheses, e.g. one might be that
the rise and fall reflect the changing levels of
factors that are known to cause CHD, e.g.
exercise patterns
Variations and associations:
real or artefect: CHD



be quantitative, e.g. how much of the decline
in CHD can be explained by the changing
pattern in these factors?
the decline in CHD is too rapid to be a result,
solely, of change in the factors mentioned,
but risk factors and treatments account for
much of the change
think even harder!
Group exercise: Why
variations may be illusory


Consider the possible reasons why a
variation in disease pattern might be an
artefact rather than real.
(You may find 7-10 reasons).
Can you group them into 3/4 categories
of explanation?
Variations as artefact









Chance
Errors of observation
Changes in the size and structure of the
population
The likelihood of people seeking health care and
hence being diagnosed
The likelihood of the correct diagnosis being
reached
Changes in the clinical approach to diagnosis
Changes in data collection methods
Changes in the way diseases are diagnostically
coded
Changes in the way data are analysed and
presented
Variations as artefacts:
diagnostic activity




Diagnostic activity measured by the number of
tests can be related to the number of cases
diagnosed
Test a hypothesis that a high number of cases
in a locality or time period reflects excessive
diagnostic activity
Predict that a large number of tests would be
done for each case diagnosed
By contrast, if a high incidence of disease, and
no excessive testing, then the test to case ratio
would be low
Figure 3.1
(test/case ratio)/
Tests done per case detected
high
low
Scenario 1:
Testing common
disease uncommon
Scenario 2:
Testing common
disease common
Scenario 3:
Testing uncommon
disease uncommon
Scenario 4:
Testing uncommon
disease common
Variations over time, pretending
to be variations in space!


One real, yet potentially misleading cause
of geographical variation, is short term
fluctuation in disease incidence
This can seem like geographical variation,
when in the long-term there is none
Figure 3.2
Place A
No. cases
No. cases
Place B
1
2
year
3
1
2
year
3
Exercise: Explanations for real
changes in disease frequency


What explanations can you think of for a
real change in disease frequency?
Can you group these into three or four
categories of explanation?
Summary of real explanations of
disease variations: the causal triad



Host e.g. genetics, behaviour
Agent e.g. virulence, introduction of a
new agent
Environment e.g. housing, weather
Applying the real-artefact
framework: Legionnaires’ disease






You are the epidemiologist responsible
for surveillance of infectious diseases in a
city of about 1 million people
You are examining the statistics on the
numbers of cases of Legionnaires'
disease
This pneumonia, acquired by inhaling
contaminated aerosol, is rare
About 8 cases per million in your country
Examine the surveillance data in table 3.3
Now, make a judgement on whether the
findings represent an outbreak of
Legionnaires' disease
Applying the real-artefact
framework: Legionnaires’ disease






At 72 cases per million population the
incidence rate in this city is exceedingly high
Chance (random fluctuation) seems to be a
remote possibility
Is there a problem in the techniques used to
handle laboratory specimens leading to false
positive results?
Could information on other diseases, say
pneumococcal pneumonia or influenza, have
been miscoded as Legionnaires' disease?
Has there been a batch of reports in June?
Has the number of people at risk altered?
Applying the real-artefact
framework: Legionnaires’ disease





is it possible that the cases could be returning
from a package tour to a particular destination?
Has the likelihood of diagnosis increased,
either because of greater vigilance by doctors
or of people using the health care system?
Have there been changes in diagnostic fashion
or disease definition?
Have there been changes in the completeness
of the data collection methods?
Has there been a deliberate change in the way
diseases are coded, analysed or data
presented?
Applying the real-artefact
framework: Legionnaires’ disease



Once error is excluded, the date of onset
of illness in the cases has been checked,
and the symptoms and signs found to be
of a pneumonic illness..
… the likelihood is that the rise in case
numbers is real and there is an outbreak
The challenge now is to develop a
testable explanation, a hypothesis, to
unveil the underlying reason for the rise
in the disease
Applying the real-artefact
framework: LD- hypotheses






Is their increased susceptibility to disease?
Is there increased virulence of microorganisms?
Is there an increase in the level of exposure
to the micro-organisms in aerosol? If so,
whyHas the weather changed?
Have winds and humidity changed?
Have protective mechanisms (such as the
drift elimination mechanisms) broken down?
Applying the real-artefact
framework: LD- hypotheses




In practice, teasing out the different
explanations is a complex task
In studies of the geographical epidemiology
of Legionnaires' disease in Scotland, 19781986, I prepared a case-list of all 372
potential cases diagnosed over the period
The chart showing the plan of the studies is
in figure 3.2
Such an analysis and overview is necessary
in all investigations of disease variations
Figure 3.3
Prepare a case list
Does the incidence vary?
Incidence by
health board,
and city of
residence
Incidence Dot maps
Map by Incidence
by postof place of
place of over time
code sector residence
work
of residence
Yes, incidence varies
Why?
Artefact?
Error in
case-list
and data
Differential use
of diagnostic
facilities
Cross check
case-lists,
Count
compare
consultants’ serology
tests
and GP’s
opinions on
diagnosis, and
survey of
patients
Real?
HostAgent
susceptibility
virulence
differs by place differs
Seek
variation
for other
Examine
approach to respiratory
diagnosis of
consultants disease
and
laboratories
Examine Not
data on studied
socioeconomic
status by
place
Environment
differs
Study
water
supply
Cooling tower
maintenance
and location
study
Figure 3.7
Figure 3.8
Disease clustering and
clusters in epidemiology




A cluster is a collection of things of the same
kind
A disease cluster is an aggregation of relatively
rare events or diseases in time or place, or both
A cluster is a mini-epidemic or outbreak of a
rare event
The concept of cluster is not used for common
diseases because clustering is inevitable due to
chance alone, or,
for infectious diseases that spread from
person-to-person for clustering is the norm
Disease clustering and
clusters in epidemiology





A cluster presents a public health problem, and
a difficult epidemiological puzzle
Clustering is merely a specialised variant of
disease variation so the analysis of clustering
follows the principles discussed
Alistair Gregg observed in 1941 that the number
of cases of congenital cataract, an exceptionally
rare problem, far exceeded the normal
He saw 13 cases of his own, and 7 of his
colleagues
You know the story!
Do the 5 grapes
comprise a cluster?



Reflect on whether the 5 grapes in figure
3.4 comprise a cluster.
What characteristics of the grape makes
you think they may be?
Imagine 5 case of acute leukaemia are
reported from a single street in a small
town i.e. these are the grapes
Figure 3.4
Is this a cluster?
Perhaps.
The challenge is
statistical and causal
Assessing whether the cluster of
grapes and of leukaemia is an artefact
or whether there is a common cause



Reflecting on both the cluster of grapes
and 5 cases of childhood leukaemia
What evidence would you seek to help
you exclude artefact and to ascertain a
common cause?
Start with the grapes-what would
convince you that they are part of a single
cluster
Figure 3.5
Is this a cluster?
Yes, but, significance unclear
i.e. how or why the grapes are
together.
The challenge is causal.
Assessing whether the cluster of
grapes and of leukaemia is an artefact
or whether there is a common cause




Evidence that the grapes are bound together
by a common stalk would be compelling
Close occurrence of leukaemia cases could be
an artefact
If our investigation of leukaemia cases had
shown these cases were all bound by common
factors such as type of leukaemia, age group,
residence, time of disease onset and
exposures to causal factors we could be
convinced the cluster is real
The next step is to explain mechanisms
Figure 3.6
Is this a cluster?
Yes. Why?
We know that
grapes are held
together by stalks
and by a vine.
Value of studying variations



Variations in disease patterns are of
practical value in helping guide the
clinician in both diagnosis and
management of disease
Outbreaks and clusters alert clinicians to
otherwise rare diseases
Long term trends are important to clinical
practice, for example, the changing nature
and decline of tuberculosis
Value of studying variations




Variations over decades (known as secular
trends) are of special importance in setting
priorities and for evaluating whether health
objectives have been achieved
Variation in disease by place and by socioeconomic status are a guide to the level of
inequity in health status
Disease variations help to match resources to
need
Health promotors can tailor both the timing
and the content of interventions
Epidemiological theory
underpinning this subject





Disease variation arises because of either (a)
changes in the host, the agent of disease or
the environment or (b) changes in interaction
between the host, agent and environment
Changes occur at a different pace in different
places and sub-populations
Disease variations are, therefore, inevitable
In epidemiology we are seeking to uncover the
natural forces that caused them
First, the epidemiologist must ensure that
variations are not merely artefacts
Summary





Diseases wax and wane in their population
frequency
The causes of such variations are often
difficult to detect and may remain a mystery
Three principal reasons for investigating
variations:
1. Bring under control an apparent abrupt rise
in disease incidence
2. Gain insight into the causes of disease
3. Make predictions about the future, both in
terms of health policy and health care, and the
frequency of disease
Analysis of variation in disease begins by
differentiating artefactual change from real
change
For real change the epidemiological challenge
is to pinpoint the causal factors