Clinical epidemiology

Introduction:
• Clinical epidemiology is the application of epidemiological principles
and methods to the practice of clinical medicine.
• It usually involves a study conducted in a clinical setting, most often
by clinicians, with patients as the subjects of study.
• The aim of clinical epidemiology is to aid decision-making about
identified cases of disease.
• The central concerns of clinical epidemiology are:
✓ Definitions of normality and abnormality.
✓ Accuracy of diagnostic tests.
✓ Natural history and prognosis of disease.
✓ Effectiveness of treatment.
✓ Prevention in clinical practice.
Definitions of normality and abnormality:
• The first priority in any clinical consultation is to determine whether
the patient’s symptoms, signs or diagnostic test results are normal
or abnormal. This is necessary before any further investigations or
treatment.
• Measurements of health-related variables can be expressed as
frequency distributions in the population of patients.
• There are three ways of distinguishing results in such a distribution:
✓ Normal as common.
✓ Abnormal as associated with disease.
✓ Abnormal as treatable.
Normal as common:
• This definition classifies values that occur frequently as normal and
those that occur infrequently as abnormal.
• We assume that an arbitrary cut-off point on the frequency
distribution (often two standard deviations above or below the mean)
is the limit of normality and consider all values beyond this point
abnormal. This is called an operational definition of abnormality.
• An alternative approach, which does not assume a statistically normal
distribution, is to use percentiles: we can consider that the 95th
percentile point is the dividing line between normal and abnormally
high values, thus classifying 5% of the population as abnormal
Abnormality associated with disease:
• The distinction between normal and abnormal can be based on the
distribution of the measurements for both healthy and diseased people,
and we can attempt to define a cut-off point that clearly separates the
two groups.
• A comparison of two frequency distributions often shows considerable
overlap - as illustrated by serum cholesterol distributions for people
with and without coronary heart disease. Choosing a cut-off point that
nearly separates cases from non-cases is clearly impossible.
• There are always some healthy people on the abnormal side of the cut-
off point, and some true cases on the normal side.
• These two types of classification error can be expressed quantitatively
in terms of the sensitivity and specificity of a test, as discussed
before:
➢ Sensitivity is the proportion of truly diseased people who are
categorized as abnormal by the test.
➢ Specificity is the proportion of truly normal people categorized as
normal by the test. A balance always has to be struck between
sensitivity and specificity; increasing one reduces the other.
Abnormal as treatable:
• These difficulties in distinguishing accurately between normal and
abnormal have led to the use of criteria determined by evidence from
randomized controlled trials, which can be designed to detect the
point at which treatment does more good than harm. Unfortunately,
many treatment decisions have to be made in the absence of such
evidence.
Diagnostic tests:
• The first objective in a clinical situation is to diagnose any treatable
disease. The purpose of diagnostic testing is to help confirm possible
diagnoses suggested by the patient’s signs and symptoms. While
diagnostic tests usually involve laboratory investigations (genetic,
microbiological, biochemical or physiological).
Value of a test:
• A disease may be either present or absent and a test result either
positive or negative. There are thus four possible combinations of
disease status and test result:
Relationship between a diagnostic test result and the occurrence of disease

Disease

Present Absent

Test Positive True positive False positive

Negative False negative True negative

• We can only use these categories when there is an absolutely accurate
method of determining the presence or absence of disease.
• Rarely is such a method available, particularly where chronic non-
communicable diseases are concerned. For this reason, and because
wholly accurate tests are likely to be expensive and invasive, simpler
and cheaper tests are used in routine clinical practice.
• Even then, we still need to know and account for the validity, accuracy
and precision of these tests when interpreting the results.
• To determine the practical utility of a given test, we need to know
more about how it performs. A test’s positive and negative predictive
value are particularly important.
• The positive predictive value is the probability of disease in a patient
with an abnormal test result, while the negative predictive value is the
probability of a patient not having a disease when the test result is
negative.
• Predictive value depends on the sensitivity and specificity of the test
and, most importantly, on the prevalence of the disease in the
population being tested. Even with a high sensitivity and high
specificity, if the prevalence is low the positive predictive value of a
test may be very low.
• The predictive values of a test in clinical practice depend critically on
the prevalence of the abnormality in the patients being tested.
Effectiveness of treatment:
• In the table below, the results of a new diagnostic test for cancer are
compared with the complete diagnostic package in current use. What
are the sensitivity and specificity of the new test? Would you
recommend its general use?
• The sensitivity of the new test= 8/10×100= 80%.
• its specificity= 9000/10000×100= 90%.
• The new test appears good; a decision on whether to use it in the
general population requires information on its positive predictive
value, which in this case is 8/1008=0.008. This very low value is
related to the low prevalence of the disease. For this reason, it would
not be appropriate to recommend general use of the test.
Natural history and prognosis:
• The term natural history refers to the stages of a disease, which
include:
✓ Pathological onset.
✓ The pre-symptomatic stage, from onset of pathological changes to the
first appearance of symptoms or signs.
✓ The stage when the disease is clinically obvious and may be subject
to remissions and relapses, regress spontaneously or progress to death.
• Detection and treatment at any stage can alter the natural history of
a disease, but the effects of treatment can only be determined if the
natural history of the disease in the absence of treatment is known.
Prognosis:
• Prognosis is the prediction of the course of a disease and is expressed
as the probability that a particular event will occur in the future.
• Predictions are based on defined groups of patients, and the outcome
may be quite different for individual patients. However, knowledge of
the likely prognosis is helpful in determining the most useful
treatment.
• Prognostic factors are characteristics associated with outcome in
patients with the disease in question. For example, in a patient with
acute myocardial infarction, the prognosis is directly related to
residual heart muscle function.
• Epidemiological information from many patients is necessary to
provide sound predictions on prognosis and outcome. Clinical
experience alone is inadequate for this purpose, since it is often based
on a limited set of patients and inadequate follow-up.
• For example, patients who are seen by a doctor are not necessarily
representative of all patients with a particular disease. Patients may be
selected according to severity or other features of their disease, or by
demographic, social or personal characteristics of the patients
themselves.
• Properly designed epidemiological research can produce reliable
information about prognosis.
Quality of life:
• Ideally, the assessment of prognosis should include measurement of all
clinically relevant outcomes and not just death, since patients are
usually as interested in the quality of life as they are in its duration.
• In studies to determine natural history and prognosis, the group of
patients should be randomly selected; otherwise selection bias may
compromise the quality of information obtained.
• For example, the prognosis of patients with chest pain admitted to
hospital is likely to be worse than that of patients with chest pain seen
by health workers in the community.
Quantity of life:
• Prognosis in terms of mortality is measured as case-fatality rate or probability of
survival.
Survival following myocardial infarction (having survived 28 days from the event), Auckland,
1983–84, 1987–88, 1991–92
• Survival analyses may include selected groups, such as patients who
survive the initial month after an event.
• Significantly more people in the later cohort (1991–92) survived three
years after a myocardial infarction than did their counterparts 10 years
earlier, which suggests improvement in secondary prevention of
coronary heart disease.
• Life-table analysis is a more sophisticated method that attempts to
predict the onset of events over time from previous patterns for all
patients at risk.
• In the follow-up of cohorts to determine prognosis, bias is often
introduced by the initial selection strategy and incomplete follow-
up.
Effectiveness of treatment:
• Some treatments are so clearly advantageous that they require no
formal assessment of indication; this is true of antibiotics for
pneumonia and surgery for trauma.
• Specific treatments need be shown to do more good than harm among
patients who actually use them: this is called efficacy.
• The best method for measuring efficacy and effectiveness is by
randomized controlled trial. However, there are many situations in
which such trials cannot be done, and only a small proportion of
current medical interventions have been assessed by such trials.
Use of evidence-based guidelines:
• Guidelines have been defined as systematically developed statements
or recommendations to assist practitioners and patients in making
decisions about appropriate health care for specific clinical
circumstances.
• Putting evidence into practice requires evidence-based guidelines.
While there are many guidelines, they are not necessarily all used in
practice.
• Indeed, there is evidence to suggest that many patients, even in high-
income countries, are not receiving the best evidence-based treatment.
This situation is particularly bad in low-income countries.
Prevention in clinical practice:
• Sound epidemiological knowledge encourages the practice of
prevention in the context of ordinary clinical practice. Much of this
prevention is at the secondary or tertiary level, but primary
prevention can also be implemented on a routine basis.
• Pediatricians have long been involved in child immunization
programs, screening for inborn metabolic defects and the regular
weighing of children and use of standard growth charts.
Reducing risks:
• Doctors, dentists and other health workers are able to convince at least
some of their patients to stop smoking. A controlled trial of different
anti-smoking interventions in general practice showed that routine
advice about tobacco use is useful, and that its effectiveness can be
improved with a variety of techniques.
• Clinicians can improve their efforts to persuade patients to stop
smoking by:
✓ Enhancing the quality of the intervention offered.
✓ Focusing on smokers who are ready to quit.
✓ Increasing frequency of advice to patients.
✓ Linking with other tobacco-control intervention channels.
Stopping works: cumulative risk of lung cancer mortality
Reducing risks in patients with established disease:
• For cardiovascular disease and diabetes, evidence-based approaches
to reducing the risk of adverse outcomes in those with the disease are
very similar to the approaches used to reduce disease onset.
• The major difference is that the risk for future clinical events is much
greater once disease is established. Both behavioral and
pharmacological interventions, amongst others, have been shown to
affect the prognosis of these diseases.
Behavioral interventions:
• Such interventions include promoting tobacco cessation, increased
physical activity, dietary change and weight loss. Together, these
may achieve a risk reduction of over 60% in people with established
heart disease, and contribute to achieving good blood glucose control
in people with diabetes.
Pharmacological interventions:
• For people with established cardiovascular disease, international
guidelines recommend long-term treatment of the risk of coagulation,
high blood pressure and high cholesterol. A combination of aspirin, β-
blockers, angiotensin converting enzyme inhibitors and statins is
expected to reduce the risk of recurrent myocardial infarction by 75%.
• However, there are large treatment gaps in all countries, in part
because of the cost and complexity of multiple drug use and other
barriers to affordable access. Some of these problems can be solved by
the use of fixed-dose combination therapy.
Conclusion:
• What is a clinical epidemiology? and what is its aim?
• What are the concerns of clinical epidemiology?
• What is the purpose of diagnostic test?
• What are the positive and negative predictive values?
• In the national history, what are the stages of a disease ?
• What does involve the management of an epidemic disease?
• What is the prognosis?
• What is the best method for measuring the efficacy and effectiveness?
• What is the guidelines in the epidemiology?