Slides supporting chapter 6 of the book:

1
Slides supporting chapter 6 of the book Bhopal
R S. Concepts of Epidemiology. Oxford, Oxford
University Press, 2002, pp317 http//www.oup.co.uk
/isbn/0-19-263155-1
2
Natural history, spectrum, iceberg, population
patterns and screening interrelated concepts in
the epidemiology of disease Raj Bhopal, Bruce
and John Usher Professor of Public Health,
Public Health Sciences Section, Division of
Community Health Sciences,University of
Edinburgh, Edinburgh EH89AGRaj.Bhopal_at_ed.ac.uk
3
Educational objectives

• On completion of your studies you should
  understand
• That the natural history of disease is the
  unchecked progression of disease in an
  individual.
• Natural history ranks alongside causal
  understanding in importance for the prevention
  and control of disease.
• The technical and ethical challenges posed in
  elucidating the natural history of disease are
  great.
• That the changing pattern of disease in
  populations over time and the spectrum of the
  presentation of disease are related yet separate
  concepts.

4
Educational objectives

• That the iceberg of disease is a metaphor
  emphasising that for virtually every health
  problem the number of cases of disease
  ascertained (those visible) is outweighed by
  those not discovered (those invisible).
• How the iceberg of disease phenomenon thwarts
  assessment of the true burden of disease, the
  need for services and the selection of
  representative cases for epidemiological study.
• Screening is the application of tests to diagnose
  disease (or its precursors) in an earlier phase
  of the natural history of disease (often in well
  people) than is achieved in routine medical
  practice.
• The key to successful screening is a simple test
  which can be applied to large populations with
  minimum harm and has a high degree of accuracy.
• The potential of screening is vast but there are
  important limitations.

5
Natural history of disease class exercise

• This is the uninterrupted progression in an
  individual of the disease from the moment of
  exposure to the causal agents.
• Reflect on the four major possible outcomes in an
  individual of exposure to a causal agent.

6
Natural history responses

• First, the exposure may have no discernible
  effect.
• Second, there may be demonstrable damaging effect
  of the exposure which may be repaired.
• Third, the effect may be an illness that is
  rapidly contained by the body's defence
  mechanism.
• Finally, the illness may progress until it leads
  to continuing long term problems, irreversible
  damage or death.
• The outcome will depend on the interactions of
  host, agent and environmental factors.

7
Natural history graphic representation

• Figure 6.1 provides an idealised view of the
  concept.
• The same concept can be applied to individual
  diseases.
• Tuberculosis provides an excellent example, which
  is illustrated in figure 6.2.
• Figure 6.3 shows a typical path for the natural
  history of CHD.

8
Figure 6.1
9
Figure 6.5
10
Figure 6.3
Recurrence and death
Disease and first manifestation can be diagnosed
here, eg. MI
Causes begin to exert their influence here
11
Obstacles to studying the natural history of
disease

• Information on natural history is very hard to
  obtain.
• What difficulties can you see in studying the
  true natural history of disease?
• Would you be willing to participate in a natural
  history study?
• What might be the effect on you of being in such
  a study?

12
Natural history studies consequences

• First, the mere act of diagnosis and follow-up by
  a physician may initiate changes in the disease
  process.
• Second, the scientific objective of observing the
  natural history of disease clashes with the
  ethical medical imperative to act to alleviate,
  contain or treat the disease.
• Studies of the natural history of disease are
  potentially ethically explosive e.g the US Public
  Health Services Tuskegee syphilis study, where
  600 "negro" men with syphilis in the state of
  Alabama in the USA were followed up for a period
  of about 40 years.
• Follow up, or cohort, studies are needed to
  define the natural history of disease and such
  long-term observations may prove costly or
  impossible.
• Natural history is, therefore, usually pieced
  together from a mixture of observations.

13
Natural history and incubation period

• Time between exposure to the agent and the
  development of disease is called the incubation
  period.
• Diseases that have long incubation periods
  generally have a long clinical course and, if so,
  by convention they are called chronic diseases.
• Some chronic diseases, paradoxically, lead to
  sudden and unexpected death e.g. a stroke or
  heart attack.
• The label chronic disease is based on the natural
  history as defined in many individuals.

14
Natural history and incubation period

• Diseases with a short incubation period usually
  have a short course, and by convention are known
  as acute diseases.
• These include most infections and many toxic
  disorders.
• The effects of acute disease may also be severe
  and prolonged, eg post-viral syndromes.
• The incubation period, together with minimal
  clinical information of the nature of the illness
  (e.g. a rash and fever), may be sufficient to
  identify the disease.

15
Natural history applications

• Natural history is vital for disease prevention
  policies.
• It underlies secondary prevention based on
  screening
• It provides a rationale for all health care.
• Purpose of health care, including medicine, is to
  influence the natural history of disease by
  reducing and delaying ill-health.
• When achieved through deliberate actions by
  societies the collective endeavour is public
  health.

16
Figure 6.4
17
The population pattern of disease

• Natural history of disease should not be (but is)
  confused with the changing pattern of disease in
  populations.
• The distribution of a disease across
  socio-economic groups may change as it has for
  coronary heart disease.
• I call this the Population pattern of disease
• Main measures of PPOD are the disease incidence
  and prevalence.

18
Interrelationship between natural history and
population pattern of disease exercise

• Assuming there are no changes to exposure to the
  causal agent, what effect would changing the
  natural history have on the population pattern?
  Consider, for example, the effect of
• Reduced and enhanced susceptibility
• A shorter or longer course of disease
• A longer and shorter incubation period
• A more severe or less severe disease

19
Changing natural history and population pattern
of disease

• Reducing the population's susceptibility would
  diminish the number of cases of overt, diagnosed
  disease.
• If the changes in susceptibility were uneven
  across a population, there will be other changes
  in the PPOD too, e.g. the reversal of
  inequalities in CHD.
• A shorter course is also likely to have a better
  outcome, with less long-term morbidity so a lower
  prevalence, or lower mortality.
• If the incubation period lengthens in a chronic
  disease from 20 to 30 years, then the disease
  burden will decline, at least in the short-term.
• The idea that an exposure can lead to variants
  (and varying severity) of the same disease is the
  spectrum of disease.

20
Spectrum of disease

• Disease may present with varying signs, symptoms
  and severity.
• Tuberculosis is another particularly good example
  and as illustrated in table 6.1.
• The spectrum of disease is, primarily, a
  population concept (while natural history is
  primarily a concept relating to individuals).
• Diseases may be mild or even silent -one of
  the many explanations for undiagnosed disease in
  the community.
• This phenomenon is described by the metaphor of
  the iceberg of disease.

21
The unmeasured burden of disease the metaphors
of the iceberg and the pyramid

• For most health problems there are large numbers
  of undiscovered or misdiagnosed cases of disease.
  
• Serious and killing disorders such as diabetes,
  atrial fibrillation and hypertension are other
  good examples of this iceberg phenomenon.
• Cases that have been correctly diagnosed can be
  likened to the tip of the iceberg, visible and
  easily measured.

22
Iceberg/pyramid of disease

• In most diseases, as with the iceberg, the larger
  presence lurks unseen, unmeasured and easily
  forgotten.
• Figure 6.6 illustrates this idea and develops the
  iceberg concept in the form of a pyramid of
  disease by using its clear structure and shape.
• Blocks 1 and 2 correspond to the iceberg above
  the sea-level and 3 to 5 below sea level.
• Epidemiology that forgets the iceberg phenomenon
  of disease Is weak and potentially misleading.

23
Figure 6.6 The pyramid and iceberg of disease
24
Iceberg/pyramid of disease

• Unidentified cases may be different to identified
  ones, both in terms of the natural history or
  spectrum of disease.
• Where symptoms and disease progression and
  outcome are related, the undiagnosed cases are
  likely to be less severe.
• When symptoms and signs are not evident in the
  early stages of disease, as in high blood
  pressure or chronic glaucoma, undiagnosed cases
  may be just as severe as diagnosed ones.
• Epidemiological studies based on selected cases
  from the tip of the iceberg may give an erroneous
  view.

25
Iceberg/pyramid severity of disease

• Prostate cancer based on cases diagnosed in
  hospital would lead to the view that the disease
  is usually, if not always, progressive.
• Unselected cases show that prostatic cancer can
  in some cases be a static, or slowly progressive,
  phenomenon.
• Patients who are at the tip of the iceberg are
  more likely to have multiple health problems than
  others.
• People with cardio-respiratory problems and
  diabetes are more likely to be admitted to
  hospital, than people with only one of these two
  problems. This is the basis of the bias known as
  Berkson's bias.

26
Screening

• Screening is the use of tests to help diagnose
  diseases (or their precursor conditions) in an
  earlier phase of their natural history or at the
  less severe end of the spectrum than is achieved
  in routine clinical practice.
• Screening attempts to uncover the iceberg of
  disease.
• On the pyramid model in 6.7 screening is applied
  to block 3, and less commonly, to block 4.
• Aim is to reverse, halt or slow the progression
  of disease.
• Screening is also done to protect society.
• Screening may be done to select out unhealthy
  people e.g. for a job.
• Screening is sometimes done to help allocate
  health care resources.
• Screening may be done simply for research, for
  example, to identify disease at an early stage to
  help understand the natural history.

27
Figure 6.7
28
Screening ethics and limitations

• The ethical viewpoint, that the natural history
  of disease must be influenced favourably, sets
  limits on the scope of screening.
• Screening could be done for every disease for
  which there is a diagnostic test or diagnostic
  signs and symptoms.
• Criteria, usually variants of those of Wilson and
  Jungner.
• These can be crystallised as six questions

29
Criteria for screening

• Is there an effective intervention?
• Does intervention earlier than usual improve
  outcome?
• Is there an effective screening test that
  recognises disease earlier than usual?
• Is the test available and acceptable to the
  target population?
• Is the disease one that commands priority?
• Do the benefits exceed the costs?
• If the answer to these six questions is yes then
  the case for screening is sound

30
Screening evaluating the case

• Screening programmes need more careful evaluation
  than clinical care and we would make the case if
  Wilson and Jungner's criteria are met as for
  hypertension-
• The benefits of screening for hypertension far
  exceed the costs.
• The screening test is measurement of the blood
  pressure, usually using a sphygmomanometer.
• The diagnostic test is, effectively, repetition
  of the same test on several occasions combined
  with a clinical history, examination and other
  tests to check for other diseases, particularly
  those that cause specific forms of hypertension.
• Additional tests of high blood pressure are
  possible but used infrequently, including 24-hour
  readings using equipment that permits measurement
  while the person is ambulatory.

31
Screening hypertension

• The ideal test would pick up all cases of
  hypertension in the population tested. This
  attribute of the test is known as high
  sensitivity (or true positive rate).
• The ideal test would also correctly identify all
  people who do not have the disease, that is, the
  test is specific to those who have the disease
  i.e. high specificity (or true negative rate).
• When cases go for more detailed clinical
  examination, the screening test result is
  confirmed, so
• A positive test predicts with accuracy the
  presence of hypertension, and similarly a
  negative test predicts its absence.

32
Screening tests performance

• These four measures, sensitivity, specificity and
  predictive power of a positive and negative test,
  are the main way to assess the performance of a
  screening test.
• Measures of performance can be calculated from
  the 2 x 2 table as shown in table 6.3.
• As the definitive test is never 100 accurate.
• Screening test is being evaluated against another
  imperfect, albeit better, test.

33
The 2x2 table - validating the screening test
34
Exercise Calculating sensitivity and
specificity, and predictive power

• 500 patients known to have a particular disease
  were screened with a new test.
• 500 controls without this disease were also
  screened.
• Of the 500 patients 473 had a positive test.
• Of the healthy group without the disease 7 had a
  positive test.
• Create a 2 x 2 table based on table 6.3 and
  reflect on the interpretation of the data.
• Calculate sensitivity and specificity of the
  test.
• Is this a good performance?
• What are the implications for those wrongly
  classified by the test?

35
Table 6.4 Calculation of sensitivity and
specificity based on data in box 6.4
36
Sensitivity and specificity

• The sensitivity (94.6) and specificity (98.6)
  of the test are very high.
• The test will correctly identify most people who
  have the disease and correctly identify most
  people who are disease free.
• About one person in twenty who does have the
  disease will be misclassified as disease free.
• Far fewer people without disease will be
  misclassified as having the disease.
• Individuals and their doctors who want to know
  the implications of their individual results and
  this is given by predictive power.

37
Predictive Powers

• If a person is positive on the screening test and
  asks what is his chance of having the disease
  once all the tests are done, what can we advise?
  
• Similarly, what do we advise if the test is
  negative on the screening test?
• From table 6.4 calculate predictive powers.

38
predictive powers and prevalence

• Predictive power of a positive test is a/ab
  473/480 98.5
• and of a negative test is d/cd 493/520
  94.8.
• Only one or two percent of those testing positive
  will have this result overturned by the
  definitive test.
• More of those with a negative test, however, will
  have this result overturned.
• The prevalence of the disease has a profound
  effect on the predictive powers.

39
Predictive powers

• Imagine that the prevalence of a disease is
  actually zero.
• Then all screening test positive cases must, of
  necessity, be false positives.
• If the prevalence of a condition is 100 then,
  logically, all screen positive cases will have
  the condition (and screen negatives will all be
  false), so the predictive power of a positive
  test is 100.
• Most diseases are uncommon, so the predictive
  power of a positive screening test tends to be
  low.

40
Figure 6.8
41
sensitivity and specificity cut-off

• The sensitivity and specificity are, however,
  profoundly affected by the "cut-off" value of the
  measure at which a test is defined as positive.
• This is a very difficult decision. How do we make
  it?
• For blood pressure, for example, we could take
  any cut-off value that is associated with a
  higher risk of disease.
• This could mean a cut-off value less than 120/80
  mmHg.
• About half of the population would therefore be
  defined as hypertensive.
• For most people so defined the true additional
  risk of disease would be very low.
• At a cut-off of 180/120 few people would be
  defined as hypertensive and for those that were
  the target organ damage and incidence of disease
  would be high.
• We would miss people who are at risk with, say, a
  blood pressure of 150/95
• There is a price to be paid for each choice of
  cut-off point.

42
Sensitivity and specificity setting the cut-off
value

• The underlying reason for the reciprocal nature
  of the sensitivity and specificity is that, for
  most diseases, cases and non-cases belong to one,
  not separate distributions of values.
• In figure 6.9(a) there are three distributions
  which could be described as low, medium and high
  blood pressure with varying levels of risk of
  hypertensive end-organ disease.
• Figure 6.9(b) shows a more realistic, so-called
  bi-modal (two peak), distribution.
• This type of distribution is not common but it
  illustrates the idea behind screening.
• Figure 6.9c, however, is the picture portraying
  the distribution of the risk factors for many
  common disorders.
• No natural separation between people at risk of
  disease and not at risk.

43
Figure 6.9
44
Sensitivity and specificity setting the cut-off
value

• Cut-off point is set solely on a judgement
  balancing the importance of avoiding false
  positives (achieving high specificity) versus
  avoiding missing true positives (achieving high
  sensitivity).
• Screening will make blocks 1 and 2 in the pyramid
  of disease (figure 6.7) grow and block 3 shrink.
  
• Danger is that through false positive tests
  people in blocks 4 and 5 are wrongly placed in
  blocks 1 and 2, and through false negative tests
  people in blocks 1 and 2 are placed in blocks 4
  and 5.

45
Setting cut-off points

• Three actions are essential to help define the
  cut-off point.
• Understanding of the natural history of the
  disease.
• Weighing up the adverse consequences of
  treatment.
• Judgments on the required sensitivity,
  specificity, and predictive powers of the
  screening test in the population to be screened.

46
Applications of the concepts of natural history,
spectrum and screening

• Health policy that has the objective to shift the
  natural history of disease to the right and alter
  the spectrum so disease is less severe.
• Public health and medical action can be seen as
  the force spearheading the attack against
  ill-health and disease.
• Knowledge of the natural history of disease can
  radically alter the organisation of health care
  so care is proactive.
• Knowing the role of early life events in the
  genesis of heart disease and diabetes alters
  fundamentally our approach to these problems.
• The need to influence the policies which foster
  good education and health of mothers and their
  infants is crystal clear.
• The scientific rationale for health care agencies
  to seek partnership with other agencies such as
  education, housing and social services is overt
• Cross-disciplinary working within health care
  (primary health care, paediatrics, obstetrics,
  nutrition and adult medicine) is seen as
  essential.

47
Applications of the concepts of natural history,
spectrum and screening

• Researchers studying people with disease now may
  need to obtain information about the life
  circumstances of the patient in childhood and
  even in-utero (the fetal origins hypothesis).
• Epidemiological methods are needed that help
  people to recall information on causal factors.
• New methods such as the life-grid approach where
  questioning is linked to memorable life events.
• Prospective epidemiological studies require
  timescales measured in the same order of time as
  the natural history of the disease.
• The timing of prevention interventions .
• The iceberg of disease phenomenon requires that
  health policy should be based on a realistic
  estimate of the size of the unidentified
  population of cases and those at risk.

48
Epidemiological theory symbiosis with clinical
medicine and social sciences

• Theory that many diseases are initiated by events
  acting years, or decades, before any clinical
  manifestation.
• Diseases may manifest themselves in many ways,
  including asymptomatic yet damaging forms.
• To understand why some people with symptoms and
  signs of disease seek care, and hence are
  diagnosed, while others do not, epidemiology
  crosses to the social sciences, linking into
  theories of illness seeking behaviour.

49
Summary

• Natural history of disease is the uninterrupted
  progression of disease from its initiation by
  exposure to the causal agents to either
  spontaneous resolution, containment by the bodys
  repair mechanisms, or to a clinically detectable
  problem.
• The primary purpose of public health and medicine
  is to influence favourably the natural history of
  disease.
• Natural history of disease is related to (and
  influences) the changing pattern of disease in
  populations or the different levels of severity
  with which a disease may present (spectrum of
  disease).
• For most health problems the number of cases
  identified is exceeded by those not discovered.

50
Summary

• The iceberg phenomenon thwarts epidemiological
  efforts to assess the true burden of disease.
• It is impossible to identify truly unselected and
  representative cases for epidemiological studies.
  
• Screening is the application of tests to diagnose
  disease (or its precursors) in an earlier phase
  of the natural history of disease (often in well
  people) or in a less severe part of the disease
  spectrum than is achieved in routine medical
  practice.
• These concepts are highly interrelated.