2. Contents
• Introduction
• Criteria for Screening
• Kappa test ratio
• Diagnostic accuracy
• Sensitivity and Specificity
• Bayes Theorem
• Likelihood ratio
• Diagnostic odds ratio
• Diagnostic effectiveness (accuracy)
• Youden's index
• ROC curve
3. Introduction
• Secondary prevention can be defined as "action which
halts the progress of a disease at its incipient stage and
prevents complications".
• Secondary prevention is based on early detection of
disease, through either screening or case finding, followed
by treatment.
4.
5. Iceberg Phenomenon of Diseases
It is a process for evaluating group of people for
asymptomatic disease or risk factor
6. Screening
• Definition:
• In 1951 the United States Commission of Chronic Illness
defined screening as “the presumptive Identification
unrecognized disease or defect by means of rapidly
applied tests, examinations or other procedures in
apparently healthy individuals”.
7. Introduction
• Screening for psychiatric disorders in1917 in United States army as it
is one of the oldest screening programmes.
to exclude subjects with psychological disorders from young men
eligible to join the United States army by administer ingmental tests
• Screening for syphilis as it used one of the earliest screening tests
• Screening for diabetes as one of the first modern forms of mass
screening.
Around 1900, life insurance companies in New York apparently
performed urine glucose tests on 71,729 persons. Prevalence of
glycosuria in men was 2.8%,
8. History
• Screening for cervical cancer using the Pap test as one of
the greatest successes of screening
Papanicolaou and Traut first reported the usefulness of
Papanicolaou (‘‘Pap’’) smear for detecting neoplastic
cervical cells in 1943.
9. COMMON SCREENING TESTS
• BLOOD PRESSURE FOR HYPERTENSION
• PAP SMEAR FOR CERVICAL CANCER
• MAMMOGRAPHY FOR BREAST CANCER
• FECAL OCCULT BLOOD FOR COLON CANCER
• SCREENING OF ANTENATAL MOTHERS
10. Aims and Objective :
• It is to sort out from a large group of apparently healthy
persons those likely to have the disease or at high risk
of the disease under study on voluntary basis often with
little or no direct financial outlay by individuals .
• To bring those who are apparently abnormal under
medical supervision and treatment.
11. DIFFERENCES BETWEEN CASE
FINDING& SCREENING TEST :
• Screening test is usually performed in community setting & is
applied to population, such as residents of a country, students
in school
• Ex- annual blood pressure measurement of employees of
industry
• Case finding – this is use of clinical and or laboratory tests to
detect disease in individuals seeking health care for other
reasons. Clinical setting
• Ex- Chest Radiography for patient admitted for elective surgery
12. Diagnostic test
• Diagnostic test – it is clinical and/ or laboratory procedure to
confirm the existence of disease or true abnormality in
patients with signs and symptoms presumed to be caused by
the disease.
Mammography for Ca breast ( screening test)
positive
additional diagnostic imaging or biopsy( diagnostic test)
13. SCREENING TEST DIGNOSTIC TEST
•Done on apparently healthy
•Applied to groups
•Test results are arbitrary and final
•Based on one criterion or cut-off point
•Less accurate
•Less expensive
•Not a basis for treatment
•The initiative comes from the investigator
or agency providing care
•Done on those with indications or sick
•Applied to single patients, all diseases
are considered
•Diagnosis is not final but modified in
light of new evidence, diagnosis is the
sum of all evidence.
•Based on evaluation of a number of
symptoms, signs (e.g., diabetes) and
laboratory findings
•More accurate.
•More expensive.
•Used as a basis for treatment:
•The initiative comes from a patient with
a complaint
14. Types of Screening :
1. Mass Screening: Screening of whole population
irrespective of particular risk of individual.
E.g. – screening of adults for Hypertension, Diabetes,
visual defects in school children
15. High Risk or Selective Screening
• Screening of high risk groups. Yield will be high.
• Ex:
Ca cervix- in lower socio economic group
Screening of Family member Of Hypertensive,
Diabetic, Ca breast
Occupational diseases.
16. Types of Screening (cont.) :
3. Multiphasic Screening:It has been defined as the application
of two or more screening tests in combination to a large number
of people at one time than to carry out separate screening tests
for single diseases.
It includes:
a health questionnaire,
clinical examination and
a range of measurements and investigations (e.g., chemical and
haematological tests on blood and urine specimens, lung
function assessment, chest X-ray, audiometry and measurement
of visual acuity)
17. Multiphasic Screening
A longshoreman is greeted by staff of the
Kaiser Permanente multiphasic health
screening program in 1961. Photo from the KP
Reporter, May 1961 time
Dr. Breslow had developed the
original multiphasic screening (the
examination of large numbers of
people with a series of tests for
detecting diseases) during the
1940s, and Dr.Collen improved
upon it with new technology. The
first beneficiaries of Collen’s
multiphasic process were members
of the International
Longshoremen’s and
Warehousemen’s Union in 1951.
18. Multiphasic Screening
• Gives high frequency of false positive
• The probability that at least 1 of screening test yield false
positive expressed as [1-(1-alpha)]n
• Alpha is false positive error rate.
• Ex- If 2 test are performed & Alpha is 5%
Probability of disease free person called for further
testing is [1-(0.95)]2 =[1-(0.9025)]= 10%
19. Correlation between no of screening tests
and persons with false positive result
No of screening tests performed % of persons with at least one false
positive result
1 5 %
2 9.8 %
4 18.5 %
5 22.6 %
10 40.1 %
20 64.2 %
25 72.3 %
20. Minimum requirements for
Community Screening Program
1. Disease requirements
2. Screening test requirements
3. Health care system requirements
21. Criteria for screening a disease
1. Important public health problem.
Ex- Serum cholesterol test for coronary heart disease which is
leading cause of death in developed countries.
2. There should be recognizable early asymptomatic period.
Ex- cervical cancer screening with Pap smear
3. Natural history should be known.
Ex- small polyps in the colon to colon cancer. Surgical removal
of polyps prevent intestinal cancer and morbidity
22. Criteria for screening a disease
4. There should be test to detect it.
Ex- blood sugar estimation in diabetes screening
5. Facilities should be there to confirm diagnosis.
Ex- additional diagnostic imaging or biopsy for Ca breast
6.There should be effective treatment available
Ex- no use in screening for pancreatic cancer because cure by medical
and surgical methods is extremely small
23. Criteria for screening a disease
7. There should be agreed policy as whom to consider as patients.
Ex- BP between 120/80 -139/89
8. There should be good evidence that early detection will reduce
mortality and morbidity.
Ex- high BP education programme reduced mortality of CHD by 50%
9. Expected benefit exceed the risk and cost.
Ex- newborn screening for congenital hypothyroidism
10. The disease must not be too rare or too common
rare disease- many false positive results. Common – not cost effective
24. Screening test requirements
1. Reasonably quick, easy and inexpensive
Ex- sphygmomanometer for hypertension
2. Safe and acceptable to persons screened
Ex- colonoscopy for colon cancer
3. Sensitivity, specificity and positive predictive value must be
known and acceptable
25. Health care system requirements
1. People with positive test must have access to follow up/
treatment
Ex- as follow up tests are expensive, time consuming or painful
2. Treatment must be acceptable
Ex- not want treatment for prostate cancer because of incontinence
and impotence
26. Health care system requirements
3. Population should be clearly defined so that resulting
data are epidemiologically useful
Ex- screening at health fairs and shopping centers
4. Clear that who is responsible for screening &how these
findings become part of participants medical record
at usual place of care
27. Ethical concerns about community
screening
• Public screening should be safe, with minimal side
effects
• Important obligation to show that benefits outweigh
the costs and potential risks
28. Criteria For Screening Test
• Acceptability
• Repeatability
A. Observer Variation
-Intra observer variation
-Inter observer variation
B. Biological Variation
C. Errors related to technical methods
• Validity (Accuracy)
29. Criteria For Screening Test
• Yield- prevalence is more
• Simplicity- sphygmomanometer for hypertension
• Safety- for visual defects in school children
• Rapidity – Rh typing for antenatal mothers
• Ease of administration and cost- sphygmomanometer
for hypertension
30. ACCEPTABILITY :
• Co-operation and acceptability. Ex- Colonoscopy for colon cancer
REPEATABILITY (RELIABILITY, PRECISION OR
REPRODUCIBILITY) :
• The test must give consistent results when repeated more than
once on the same individual or material, under the same conditions.
• Depends on 3 major factors:
Observer variation.
Biological variation.
Errors relating to technical methods.
31. Observer Variation
• Intra observer (within observer ): This is variation
between repeated observations by the same observer on
the same subject or material at the same time.
• Ex- if clinician examines X-ray film several times without
knowing it is same film, there are usually some differences.
• It can be minimized by taking average of several replicate
measurements at the same time
32. Observer Variation
• Inter observer variation (between observer) : This is
variation between different observers on the same subject or
material.
• Ex- if 2 clinicians examine same X-ray film independently,
there are usually some differences.
• It can be minimized by
a. Standardization of procedures
b. Intensive training of all observers
c. Making use of 2 or more observers for independent
assessment
33. Measuring agreement
•Type of inter-observer agreement
• ex- radiologists frequently disagreed about interpretation
specific mammogram
• Two observers can report the same reading, but both
observers could be wrong.
• Considerable agreement would be expected by chance alone
• Measuring the extent to which agreement exceeds that
expected by chance requires a measurement called Kappa
34. 2*2 TABLE for KAPPA STATISTIC
Observer 1
Observer 2 Yes No Total
Yes a b a+b
No c d c+d
Total a+c b+d a+b+c+d
• a +d= observed agreement (Ao),
•a+b+c+d= maximum possible agreement
•(a +d )/ (a+b+c+d)= overall percent agreement
•[(a+b) (a+c)]/(a+b+c+d)= cell a agreement expected by chance
•[(c+d) (b+d)]/(a+b+c+d)= cell d agreement expected by chance
•cell a agreement expected by chance+ cell d agreement expected by chance= total
agreement expected by chance(Ac)
•(Ao- Ao )/ (N- Ac )= kappa
35. Agreement between 2 clinicians on cardiac murmur by
physical examination
Observer 2 Observer 1
Murmur Yes No Total
Yes 30 7 37
No 3 60 63
Total 33 67 100
• 30 +60= 90= Observed agreement (Ao),
•30+7+3+60= 100= Maximum possible agreement
•(30+60 )/ (30+7+3+60)= 90/100=90% Overall percent agreement
•[(30+7) (30+3)]/(30+7+3+60)= 12.2 cell a agreement expected by chance
•[(3+60) (7+60)]/(30+7+3+60)= 42.3 cell d agreement expected by chance
•12.2+ 42.3= 54.4=Total agreement expected by chance(Ac)
•(90-54.4 ) (100-54.4)= 35.6/45.6= 0.78=78% Kappa
36. KAPPA TEST RATIO
• The numerator is observed improvement over chance agreement
(Ao- Ao) & denominator is maximum possible agreement over
chance agreement (N- Ac )
• Interpretation of Kappa Test Ratio
• < 20%= negligible improvement
• 20-40%= minimal
• 40-60%= fair
• 60-80%= good
• >80%= excellent
37. Biological (Subject) Variation
• There is biological variability with many physiological variables such as
BP, blood sugar
• It may be due to
1. Changes in the parameter observed
Ex- cervical smear same women on 2 days
2. Variations in the way patients perceive their symptoms
Ex- when subject is aware that he is being probed
3. Regression to mean- tendency for values at the extreme of distribution
to regress towards the mean
Ex- stool frequency in ulcerative colitis
38. Errors in technical method: variations inherent in the
method
– Defective instruments.
– Errors in calibration.
– Faulty reagents.
– Test itself may be inappropriate or unreliable.
39. GOLD STANDARD
• GOLD STANDARD: an external source of truth regarding the
disease status of each individual in the population.
• In real life, when we use a test, we clearly do not know who has
the disease and who does not.
• But to quantitatively assess the sensitivity and specificity of a
test, we must have another source of truth with which to
compare the test results
• Ex- tissue biopsy, cardiac angiogram
40. Accuracy and Precision
• Accuracy- ability of measurement to be correct on
average. If it deviates then it is biased
• Precision/ Reproducibility/ Reliability- ability of
measurement to give same result/ similar result with
repeated measurement of same factor
44. DETERMINING THE CUTOFF
POINT:
The factors to be considered are :
• Disease prevalence : when prevalence of the disease
is high in the community the cut-off point is set at low
level.
• The disease : if the disease is very lethal and early
intervention markedly increase the prognosis, cut off
point is set at lower level.
45. 3 2
7 8
7 5
3 5
DIABETIC
DIABETIC DIABETICNON-DIABETIC
NON-DIABETIC
NON-DIABETIC
HIGH
HIGH
HIGH
DIABETIC
DIABETIC NON-DIABETIC
NON-DIABETIC
BLOOD
SUGAR BLOOD
SUGAR
BLOOD
SUGAR
LOW
LOW
LOW
10 10
10 10
+
+
-
-
SCREENING FOR DIABETES IN HYPOTHETICAL POPULATION WITH A PREVALENCE OF 50
%.
EFFECTS OF CHOOSING DIFFERENT CUTOFF LEVELS FOR A POSITIVE TEST:
A B
C
Sensitivity=3/10=30%
Specificity= 8/10=80%
Sensitivity=7/10= 70%
Specificity= 5/10=50%
46. • In either distribution—unimodal or bimodal— it is
relatively easy to distinguish between the extreme
values of abnormal and normal.
• With either type of curve, however, uncertainty
remains about cases that fall into the gray zone.
47. Choice of cut-off
• Depends on the importance of false positives and false
negatives.
• False positives are associated with costs—emotional and
financial—as well as with the difficulty of “delabeling”
• False positive result- major burden to the health care as large
group of people need to be brought back for a retest
• False negative- will be told they do not have the disease and
will not be followed, so serious disease might possibly be missed
at an early treatable stage.
48. • It is the amount of previously unrecognized disease that
is diagnosed as a result of screening effort.
Yield depends up on :
sensitivity
specificity
prevalence of the disease
49. Validity (accuracy)
• Validity refers to what extent the test accurately measures what it
purports to measure
• Validity is the ability of the test to separate or distinguish those
who have the disease from those who do not.
• For example, glycosuria is a useful screening test for diabetes,
but a more valid or accurate test is the glucose tolerance test.
• There are two components namely
1. Sensitivity
2. specificity
50. Diagnostic accuracy
• Diagnostic accuracy relates to the ability of a test to
discriminate between the target condition and health.
• This can be quantified by the measures of diagnostic accuracy
such as
sensitivity and specificity
predictive values
likelihood ratios
the area under the ROC curve
Youden's index and
diagnostic odds ratio.
51. Diagnostic accuracy
• Measures are not fixed indicators of a test performance,
some are very sensitive to prevalence, while others to the
spectrum and definition of the disease.
• Measures are extremely sensitive to the design of the
study.
• STARD initiative was a very important step toward the
improvement the quality of reporting of studies of
diagnostic accuracy.
52. Sensitivity and specificity
• Unfortunately, perfect test does not exist in real life and
therefore diagnostic procedures can make only partial
distinction between subjects with and without disease.
• Values above the cut-off are not always indicative of a
disease since subjects without disease can also sometimes
have elevated values- false positive values (FP).
• Values below the cut-off are mainly found in subjects
without disease- false negative values (FN).
53. Evaluation of Screening Test :
Screening test
results
Diagnosis Total
Diseased Not diseased
Positive a
( true positive)
b
(false positive)
a + b
Negative c
( False negative )
d
( true negative)
c + d
Total a + c b + d a+b+c+d
54.
55. Sensitivity :
• The term sensitivity was introduced by Yerushalmy in 1940s as a
statistical index of diagnostic accuracy. It is defined as the ability of a test
to identify correctly all those who have the disease, that is "true positive".
=TP/TP+FN =
• A test has 90% sensitivity means: 90 per cent of the diseased people
screened by the test will give a "true positive" result and the remaining 10
per cent a "false negative" result.
(a)
(a + c)
56. Sensitivity
• If the test is not sensitive, it fails to detect disease in some
and these appear in cell C
• The rate at which this occurs is called False negative error
rate & is called as c/a+c
• Sensitivity and False negative error rate add upto
1.0(100%)
57. Specificity :
It is defined as the ability of a test to identify correctly those
who do not have the disease, that is, "true negatives".
= TN/FP+TN =
• 90 per cent specificity means: 90 per cent of the non
diseased persons will give "true negative" result and 10 per
cent of non-diseased people screened by the test will be
wrongly classified as "diseased" when they are not (False
positives).
(d)
(b + d)
58. Specificity
• If the test is not specific, false positive cases appear in
cell B
• The rate at which this occurs is called False positive error
rate & is called as b/b+d
• Specificity and False positive error rate add upto
1.0(100%)
60. Predictive values :
• Useful to know what proportion of patients with abnormal test
results are truly abnormal.
Positive Predictive Value:
The "predictive value of a positive test" indicates the
probability that a patient with a positive test result has, in fact,
the disease in question.
• Proportion of patients with positive test results who are
correctly diagnosed.
• PPV of 90% means that 90% of the patients who are
diagnosed to be positive by the test in fact have the disease in
question.
(a)
(a + b)
100
61. NEGATIVE PREDICTIVE VALUE:
• The "predictive value of a negative test" indicates the
probability of not having the disease, when the test result is
negative.
• NPV of 90% means that 90% of the patients who are
diagnosed to be negative by the test do not have the disease in
question.
• It indicates what proportion of subjects with negative test result
didn’t have the disease
(d)
(c + d)
100
62. PREDICTIVE VALUE OF A TEST
Screening test Diseased Non- Diseased Total
Positive 80 100 180
Negative 20 800 820
Total 100 900 1000
Positive predictive value =80/180= 44%
If the test is positive it is 44% that person has the disease
Negative predictive value =800/820 = 98%
If the test is negative it is 98% that person does not shave the disease
63. Relationship between Positive Predictive
Value and Disease Prevalence
•Higher the prevalence, the higher the predictive value.
•Screening program is most productive and efficient if it is directed
to a high-risk target population.
•Screening a total population for a relatively infrequent disease can
be very wasteful of resources
65. Determinants of Predictive Value
• It is determined by sensitivity, specificity and prevalence of disease
• As the prevalence of the condition increases positive predictive
value increases and thus more chances of getting true positive
results.
• The more sensitive a test, the better will be its negative predictive
value
• The more specific a test, the better will be its positive predictive
value
66. USE OF MULTIPLE TESTS
1. Sequential Testing
2. Simultaneous Testing
• Sequential Testing (Two-Stage Screening)-
• a less expensive, less invasive, or less uncomfortable test is
generally performed first, and
• those who screen positive are recalled for further testing with a more
expensive, more invasive, or more uncomfortable test, which may
have greater sensitivity and specificity.ti
67. Sequential Testing
• Ex- a population is screened for diabetes using a test with
a sensitivity of 70% and a specificity of 80.
• The disease prevalence is 5%, in the population of 10,000
Test results Diabetes + Diabetes - Total
Positive 350 1,900 2,250
Negative 150 7,600 7,750
Total 500 9,500 10,000
68. • Those 2,250 people are brought back and screened using
GTT with a sensitivity of 90% and a specificity of 90%.
• Net sensitivity= 315/500= 63%
• Net specificity= 7,600=1710/9500=98%
• Use of both tests in sequence has resulted in a gain in net
specificity
Test results Diabetes + Diabetes - Total
Positive 315 190 505
Negative 35 1,710 1,745
Total 350 1,900 2,250
Sequential Testing
69. Simultaneous Testing
• In a population of 1,000 people, the prevalence of a
disease is 20%. Therefore, 200 people have the disease.
• In order to identify 200 people , test A and test B are used
at the same time.
• Test A- Sensitivity = 80%, Specificity = 60%
• Test B- Sensitivity = 90%, Specificity = 90%
70. Test A Test B
Test
results
Diabet
es +
Diabete
s -
Total
Positive 160 320 480
Negative 40 480 520
Total 200 800 1000
Sensitivity = 80%, Specificity = 60%
Test
results
Diabetes
+
Diabetes
-
Total
Positive 180 80 260
Negative 20 720 740
Total 200 800 1000
Sensitivity = 90%, Specificity = 90%
71.
72.
73. • The individual is generally considered to have tested
“positive” if he or she has a positive result on any
one or more of the tests.
• There is a net gain in sensitivity
• The individual is considered “negative” if he or she
tests negative on all of the tests.
• As a result, there is a loss in net specificity.
Simultaneous Testing
74. USE OF MULTIPLE TESTS
• The decision to use either sequential or simultaneous
testing often is based both on the objectives of the testing
• Whether testing is being done for screening or diagnostic
purposes
• Practical considerations -the length of hospital stay, costs,
• and degree of invasiveness of each of the tests
75. Principles of Screening can be
summarized as
1. Screening test, which is used to rule out a diagnosis,
should have high degree of sensitivity
2. Confirmatory test, which is used to rule in a
diagnosis, should have high degree of specificity
Spin –specificity is needed to rule in
Snout – sensitivity is needed to rule out
76. Uses of screening :
1.Case detection :
• “Prescriptive screening” – Presumptive identification of
unrecognized disease which does not arise from patients request.
• People are screened for their own benefit.
• E.g. : Neonatal screening, Deafness in children.
2.Control of disease :
• “Prospective screening” - Early diagnosis and treatment decrease
spread of infections or mortality from it.
• People are screened for others benefit.
• E.g. : immigrants ( tuberculosis, syphilis)
Streptococcal infection to prevent RF.
77. Uses of screening
3.Research Purposes :
• Aid in obtaining more basic knowledge about chronic
diseases for which natural history is not fully known.
e.g.: cancers, hypertension
• Initial screening for prevalence & subsequent screening
for incidence.
• No follow up therapy available.
4.Educational opportunities :
• Creating public awareness
• Educating health professionals.
78. Potential benefits of Screening
Program
• Reduced mortality
• Reduced morbidity
• Reassurance
79. Harms of Screening Program
• Uncomfortable. Ex –mammography
• False positive test results lead to extra time and costs
and cause anxiety and discomfort to individuals
• False positive test results can be even worse. People
with early symptoms to be less concerned
• Over-diagnosis
81. Bias in Screening Program
• Selection bias
• Lead time bias
• Length bias
82. Selection bias
• Individuals may participate – family h/o of disease or
aware of being at higher risk of contracting disease
• Screening Program would find more cases,
exaggerating the apparent utility of Screening.
83. Concept of lead time:
Disease onset
detection
First
possible
point
Final
critical
diagnosis
OUTCOME
Usual time of
diagnosis
Screening time
Lead time
A
B
87. Repetition of Screening Program
• Initial Screening- prevalent cases
• Repeated Screening- incident cases
• More Screening test done on an individual- more
positive findings, both true positive and false positive
• Recommending frequent repeat examination- burden of
cost and anxiety to rule out disease in individuals with
false positive examination
88. Screening Guidelines and
Recommendations
Organizations that issue Screening Guidelines are:
• Specialty organizations(American Urological Association)
• Organizations representing primary care specialists(American
college of physicians)
• Foundation for treatment and prevention of particular disease
(American Cancer Society)
• Organizations dedicated to evaluating Screening Recommendations
[U.S. Preventive Services Task Force (USPSTF), Canadian task
force on the periodic health examinations]
89. Grades assigned to Screening
Recommendations Screening for ovarian
cancer with CT
Grade Recommendations Net benefit suggestion
A Recommends the service high certainty that the net benefit is substantial.
BP screening in adults
B Recommends the service there is moderate certainty that the net benefit is
moderate to substantial. Breast cancer screening
In >50 years
C Recommends against
routinely providing the
service.
Considered for an individual patient. There is at least
moderate certainty that the net benefit is small.
Breast cancer screening< 50 yrs
D Recommends against the
service
Moderate or high certainty that the service has no
net benefit or that the harms outweigh the benefits.
Screening for ovarian cancer with CT
I The current evidence is
insufficient to assess the
balance of benefits and
harms of the service.
Evidence is lacking, of poor quality, or conflicting, and
the balance of benefits and harms cannot be
determined. Screening for skin cancer, lung cancer
with helical CT
90. Assessing the Effectiveness of Screening
Programs Using Outcome Measures
1. Reduction of mortality in the population screened
2. Reduction of case-fatality in screened individuals
3. Increase in percent of cases detected at earlier stages
4. Reduction in complications
5. Prevention of or reduction in recurrences ormetastases
6. Improvement of quality of life in screened individuals
91. Bayes Theorem
1. If the test results are positive, what is the probability that
a patient has the disease?
2. If the test results are negative, what is the probability that
a patient doesn’t have the disease?
• Bayes Theorem provides answer to these questions
• It is first described by English Clergyman
92. Bayes Theorem Formula
• P(D+ T+)= p(T+ D+ ) p (D+)
p(T+ D+ ) p (D+) + p(T+ D- ) p (D-)
• P = probability
• D+ = patient has the disease in question
• D- = patient doesn’t have the disease in question
• T+ =certain diagnostic test for the disease is positive
• T- = test is negative
• = conditional on what immediately follows
93. Bayes Theorem
• Bayes Theorem is formula for positive predictive value
• Numerator of Bayes Theorem describes cell a (true
positive result)
• Probability of being in cell a is prevalence times the
sensitivity
• p(T+ D+ ) is sensitivity (Probability of being in the top,
test positive, row given the fact of being in diseased
column
94. • Denominator of Bayes Theorem( 2 terms, 1st
describes cell a (true positive result) & 2nd describes
cell b (false positive result)
• 2nd term of denominator, the Probability of false
positive error rate or p(T+ D- ) is multiplied by
prevalence of non- diseased persons or p(D-)
95. Use of Bayes Theorem
• To Determine positive predictive value of tuberculin
screening program
• Sensitivity of tuberculin test= 96%=0.96
• False negative error rate of test= 4%= 0.04
• Specificity of test= 94%=0.94
• False positive error rate of test= 6%= 0.06
• Prevalence of tubeculosis in community= 1%=0.01
96. Use of Bayes Theorem
• P(D+ T+)= p(T+ D+ ) p (D+)
p(T+ D+ ) p (D+) + p(T+ D- ) p (D-)
= (sensitivity) (Prevalence )
(sensitivity) (Prevalence )+(False positive error rate )(1- Prevalence )
= (0.96) (0.01) = 0.0096 = 13.9%
[(0.96) (0.01)+ (0.06)(0.99) 0.0690
97. Use of 2*2 table, study of 10,000 population
Screening test
results
Diagnosis Total
Diseased Not diseased
Positive
96(96%)
594( 6%) 690 (7%)
Negative 4 (4%) 9306 (94%) 9310 (93%)
Total 100 (100%) 9900 (100%) 10,000 (100%)
Positive predictive value= 96/690= 0.139= 13.9%
98. BAYES THEOREM
A prior (pre-test) probability is an initial probability value originally obtained before
any additional information is obtained.
A posterior (post-test) probability is a probability value that has been revised by using
additional information that is later obtained.
Example: Road traffic accident patients in a hospital are
listed and one of the patients is randomly selected.
a) what is the probability that he will have
cerebral hemorrhage?
b) during the later analysis of this patient it was
found that he has CT scan positive for cerebral
hemorrhage. Now, what is the probability that he
actually has cerebral hemorrhage?
99. For the first question, we know that
Probability of cerebral hemorrhage in a RTA patient is 28%.
For the second question, we know that
CT scan has a sensitivity of 90% and specificity of 95% for
detecting cerebral hemorrhage.
Bayes theorem gives us a way of calculating probability of an
event in the light of presence of a second event, which itself
occurs in the presence of the first one.
100. Likelihood Ratios
• The likelihood ratio for a particular value of a
diagnostic test is defined as the probability of that
test result in people with the disease divided by
the probability of result in people without the
disease
101. Positive Likelihood Ratio
• It is ratio of sensitivity of a test to the false positive error
rate of the test
LR(+)= TPR = Sensitivity = a/b
FPR 1-Specificity [(a+c)/(b+d)
• Test to be good, ratio should be larger than 1
• Odds of disease among persons in whom the test yielded
positive result, divided by the odds of disease in the entire
population
102. • LR+ indicates how much odds of disease were
increased if the test result was positive
• This ratio is independent of prevalence
• LR+ is the best indicator for ruling-in diagnosis.
• The higher the LR+ the test is more indicative of a
disease. Good diagnostic tests have LR+ > 10
Positive Likelihood Ratio
103. Negative Likelihood Ratio
• It is ratio of false positive error rate divided by the specificity
LR(-) = FNR = 1-Sensitivity = c/d
TNR Specificity [(a+c)/(b+d)]
• Smaller the LR- the better the test is
• Odds of missed disease among persons in whom the test
yielded negative result, divided by the odds of disease in the
entire population
• LR- shows how much odds of disease were decreased if the test
result was negative
104. • LR- is usually less than 1 because it is less likely that
negative test result occurs in subjects with than in
subjects without disease.
• If LR+ of a test is large and LR- is small, then it is a
good test
• LR- is a good indicator for ruling-out the diagnosis.
Negative Likelihood Ratio
105. Use of 2*2 table, study of 80 participants
Screening test
results
Diagnosis Total
Diseased Not diseased
Positive 12 3 15
Negative 8 57 65
Total 20 60 80
Sensitivity =12/20=60%
false positive error rate= 3/60= 5%
LR+= 0.60/.05= 12
LR- =8/20= 0.421
LR+/LR-= 12/0.421=28.5
ad/bc= 12*157/3*8=28.5
106. Likelihood Ratio
• LR+ indicates how much of odds of disease were
increased if the test result was positive
• LR- indicates how much of odds of disease were
decreased if the test result was negative
107. WHY DO WE NEED LIKELIHOOD RATIO?
• Many tests in clinical medicine have continuous results or multiple ordinal
levels. Putting multiple categories into either positive or negative test
causes loss of information. Likelihood ratios enable clinicians to interpret
and use the full range of diagnostic test results.
• While predictive values relate test characteristics to populations, likelihood
ratios can be applied to a specific patient.
• Likelihood ratios refine clinical judgment. Application of a likelihood ratio
to a working diagnosis generally changes the diagnostic probability—
sometimes radically.
108. Receiver Operator Characteristic
(ROC) Curve
• To measure continuous variable- serum calcium,
blood glucose or BP choice of cut off(best) is often
difficult
• To decide on good cut off, investigator construct
ROC Curve
109. ROC Curve
• ROC Curve originated in World War II in evaluating the
performance of radar operator
• True positive- correct early warning of enemy planes
crossing the English channel
• False positive- when radar operator sent out an alarm but
no enemy planes appeared
• False negatives- when enemy planes appeared without
previous warning from the radar operator
110.
111. Plotting and Intrepretating an ROC
Curve
Cut point True positive False positive
5 0.56 0.01
7 0.78 0.19
9 0.91 0.58
Cutpoint Sensitivity Specificity
5 0.56 0.99
7 0.78 0.81
9 0.91 0.42
T4 value Hypothyroid Euthyroid
5 or less 18 1
5.1 - 7 7 17
7.1 - 9 4 36
9 or more 3 39
Totals: 32 93
112. Plotting and Intrepretating an ROC
Curve
Cut point True
positive
False
positive
5 0.56 0.01
7 0.78 0.19
9 0.91 0.58
Cutpoin
t
Sensitivity Specificity
5 0.56 0.99
7 0.78 0.81
9 0.91 0.42
113. ROC Curve
• ROC Curve can be considered a graph of LR+
• Ideal ROC Curve would rise almost vertically from lower
left corner and move horizontally almost along the upper
line i.e excellent curve
• No benefit line- if sensitivity equaled false positive error
rate, result is diagonal straight line from left corner to upper
right corner
• Most clinical test lie between 2 extremes, either good curve
or fair curve
115. An ROC curve demonstrates
1. It shows the tradeoff between sensitivity and specificity (any increase
in sensitivity will be accompanied by a decrease in specificity).
2. The closer the curve follows the left-hand border and then the top
border of the ROC space, the more accurate the test.
3. The closer the curve comes to the 45-degree diagonal of the ROC
space, the less accurate the test.
4. The slope of the tangent line at a cutpoint gives the likelihood ratio
(LR) for that value of the test.
5. The area under the curve is a measure of text accuracy.
116. Diagnostic accuracy
• Diagnostic accuracy relates to the ability of a test to discriminate
between the target condition and health.
• This is measured by
sensitivity and specificity
predictive values
likelihood ratios
the area under the ROC curve
Youden's index and
diagnostic odds ratio.
117. • Measures of diagnostic accuracy are not fixed indicators of
a test performance, some are very sensitive to the disease
prevalence, while others to the spectrum and definition of
the disease.
• It is extremely sensitive to the design of the study.
• STARD initiative was a very important step toward the
improvement the quality of reporting of studies of
diagnostic accuracy
Diagnostic accuracy
118. Diagnostic odds ratio (DOR)
• DOR of a test is the ratio of the odds of positivity in subjects
with disease relative to the odds in subjects without disease.
• DOR = (TP/FN)/(FP/TN).
• DOR depends significantly on the sensitivity and specificity of
a test.
• It depends on criteria used to define disease and its spectrum of
pathological conditions of the examined group (disease
severity, phase, stage, comorbidity etc.).
119. Diagnostic effectiveness (accuracy)
• It is expressed as a proportion of correctly classified
subjects (TP+TN) among all subjects (TP+TN+FP+FN)
• = (TP+TN)/ (TP+TN+FP+FN)= a+d/(a+b+c+d)
• It is affected by the disease prevalence.
• Diagnostic accuracy of a particular test increases as the
disease prevalence decreases.
120. Youden's index
• Youden's index is one of the oldest measures for diagnostic
accuracy. The index was suggested by W.J. Youden in 1950
• Youden's index is calculated by deducting 1 from the sum of
test’s sensitivity and specificity expressed not as percentage
but as a part of a whole number: (sensitivity + specificity) – 1.
• In poor diagnostic accuracy, Youden's index equals 0, and
• In a perfect test Youden's index equals 1
• Youden's index is not sensitive for differences in the sensitivity
and specificity of the test, which is its main disadvantage
121. Design of diagnostic accuracy studies
• Measures of diagnostic accuracy are extremely sensitive to
the design of the study.
• Studies suffering from some major methodological
shortcomings can severely over- or under-estimate the
indicators of test performance as well as they can severely
limit the possible applicability the results of the study
122. Evaluation of Screening Program
1. Randomized controlled trial
2. Un- controlled trial
3. Other methods- case control studies
124. Conclusion
• Medical screening has existed for about 60 years, and has a very rich
history.
• The preclinical identification of disease has been a major component of
modern medicine and public health.
• Screening which is secondary prevention is done in community setting
• Test with high degree of sensitivity are useful for screening
• Test with high degree of specificity are useful for Confirming diagnosis
• Selection, lead time, length bias leads to over-estimate of benefits from
screening
125. References
1. Textbook of Public Health and Community Medicine, Pub by Dept of
Community Medicine, AFMC, Pune in collaboration with WHO, India Office,
New Delhi
2. K. Park. Textbook of Preventive and Social Medicine. Screening for disease
3. A Morabia, F F Zhang. History of medical screening: from concepts to action.
Postgrad Med J 2004;80:463–69.
4. Ana-Maria Šimundić. Measures of diagnostic accuracy: basic definitions
5. Katz, David L.,Jekel, James F., eds. Jekel's Epidemiology, Biostatistics,
Preventive Medicine, And Public Health. Philadelphia, Pa. : Saunders, 2014.
6. Leon gordis. Epidemiology. Fifth edition. Canada. 2014.