Screening for Diseases

Presenter: Dr Archana S
Guide: Dr Shalini C Nooyi

03-09-2014 1
 Contents:

• Introduction

• Definition

• Objectives

• Concept of screening

• Screening tests vs Diagnostic tests

• Uses of screening

• Disadvantages of screening programmes

2
03-09-2014
 Contents – contd.

• Types of screening

• Criteria for screening

• ROC curve

• Biases in evaluation of screening

• Evaluation of screening tests

• Conclusion

• References
3
03-09-2014
 Introduction

• Screening is an important aspect of

prevention

• Annual health examinations - early detection of "hidden" disease

• Alternative approach - conserving physician-time for diagnosis &

treatment and administering simple, inexpensive lab tests

4
03-09-2014
History

• Benefits of screening - first demonstrated by use of mass

miniature radiography (MMR) for tuberculosis

• With reduction in burden of TB - application of screening to

other chronic conditions began

• Original screening programmes - TB, syphilis or selected groups

such as antenatal mothers, school children & occupational
groups

• Preventive care function & logical extension of health care

5
03-09-2014
 Definition of Screening

“The presumptive identification of unrecognized defect or disease by

application of tests, examinations or procedures which can be applied
rapidly,

to sort out apparently well persons who probably have a disease, from
those who probably do not.

A screening test is not intended to be diagnostic.

Persons with positive or suspicious findings must be referred to the

physicians for diagnosis and necessary treatment”

Source: 1. United States commission on Chronic Illness. Final Report 1951.

6
2.Morrison AS. Screening in chronic Disease. Oxford University Press, New York, Ist Ed 1985.
03-09-2014
 Definition of Screening

“The search for unrecognized disease or defect by means of

rapidly applied tests, examinations or other procedures in
apparently healthy individuals."

7
03-09-2014
 Objectives

1. To sort out from apparently healthy persons those likely to

have disease & to bring “apparently abnormal” under medical
supervision and treatment

2. To detect disease early before it becomes symptomatic

8
03-09-2014
Concept of screening

Disease First Final Usual Outcome

Onset possible critical of time
detection point diagnosis diagnosis
    A

B
Screening Time

Lead Time

9
03-09-2014
 Concept of screening – contd..

Screening Periodic health examinations

Capable of wide application Not capable of wide application

Relatively inexpensive Expensive

Directed toward multiple

Directed towards a specific
diseases
disease
Requires little physician-time Consumes more physician time

10
03-09-2014
Possible outcomes of screening

Apparently healthy (Screening tests)

Apparently normal Apparently abnormal

Periodic 1.Normal-
rescreening periodic rescreening

2.Intermediate -
surveillance

3.Abnormal -
treatment
11
03-09-2014
Screening tests vs Diagnostic tests
Screening test Diagnostic test
1. Done on apparently healthy 1. Done on those with indications or sick
2. Applied to groups 2. Applied to single patients
3. Diagnosis is not final.
3. Test results are arbitrary and final
Many evidences guide the diagnosis.

4. Based on one criterion or cut-off 4. Based on evaluation of a number of

point symptoms, signs and laboratory findings

5. Less accurate, Less expensive 5. More accurate, More expensive

6. Not a basis for treatment 6. Used as a basis for treatment.

7. Initiative comes from the 7. Initiative comes from a patient with a

investigator or agency providing care complaint

12
03-09-2014
 Uses of screening

a. Case detection

b. Control of disease

c. Research purposes

d. Educational opportunities

13
03-09-2014
 Disadvantages of screening programmes

1. Over-treatment of questionable abnormalities

2. False reassurance for those with false negative results

3. Anxiety and sometimes morbidity for those with false positive

results

4. Resource costs

5. Hazards of screening test itself

14
03-09-2014
 Types of screening

a. Mass screening

b. Opportunistic screening

c. High risk or selective screening

d. Multiphasic screening

15
03-09-2014
 a. Mass screening
• Screening of a whole population or a sub-group

• Backed up by suitable treatment.

b. Opportunistic Screening
• Restricted to patients who consult a health practitioner for some
other purpose

Eg: Headache & Giddiness - Hypertension

16
03-09-2014
 c. High risk or selective screening

• Most productive

• Screening for cancer cervix in the lower social groups

• Diabetes, HTN, breast cancer - screening other members of family

• Screening for “risk factors” - elevated serum cholesterol

• Preventive measures can be applied before disease occurs

17
03-09-2014
 d. Multiphasic screening

• Application of two or more screening tests in combination to a

large number of people at one time

Eg., Screening of pregnant women for

• Anemia - Hb%

• Syphilis - VDRL test

• Diabetes - RBS

• RH Status - Rh typing

18
03-09-2014
 Criteria of disease for screening
1. Important health problem
2. Early asymptomatic stage
3. Natural history is adequately understood
4. Test to detect disease in preclinical stage
5. Facilities for confirmation of diagnosis
6. Effective treatment
7. Whom to treat
8. Earlier intervention lead to better outcome
9. Expected benefits exceed risks & costs

19
03-09-2014
 Criteria of Screening test used for screening

• Simple & easy to apply

• Safe & Acceptable to public

• Cost effective

• Yield (amount of previously unrecognized disease that is

diagnosed as a result of the screening effort)

• Repeatability

• Validity

20
03-09-2014
Repeatability
• @ reliability, precision or reproducibility

• Consistent results when repeated more than once on same

individual or material, under same conditions

• Depends upon i. observer variation

ii. biological (or subject) variation

iii. errors relating to technical methods

21
03-09-2014
 Repeatability – contd..
i. Observer variation

a. Intra -observer variation - variation b/w repeated observations

by same observer on same subject or material at same time

minimized by taking the average of several replicate measurements

at the same time.

b. Inter -observer variation

minimized by (a) standardization of procedures (b) intensive

training of all observers (c) making use of two or more observers
for independent assessment
03-09-2014
22
 ii. Biological (subject) variation

(a) Changes in the parameters observed

(b) Variations in the way patients perceive their symptoms & answer

(c) Regression to the mean

Biological variation is tested by repeat measurements over time.

iii. Errors relating to technical methods

• Affected by variations inherent in the method

23
03-09-2014
 Validity (accuracy)

• To what extent the test accurately measures which it purports to

measure

• Closeness with which measured values agree with "true" values

2 components

• Sensitivity(Sn) of the test - proportion of diseased people who

were correctly identified as “positive” by the test

• Specificity(Sp) of the test - proportion of non-diseased people

who are correctly identified as negative by the test
24
03-09-2014
Table: 1
Screening test Diagnosis

Diseased Not diseased

Positive a (True positive) b (False positive)

Negative c (False negative) d (True negative)

Total a+c b+d

Sn=TP/TP+FN
Sp=TN/FP+TN

25
03-09-2014
 Tests with Dichotomous Results
(Positive or Negative)
Eg., Table: 2 Population of 1,000 people, of whom 100 have a disease and
900 do not have disease.
A Screening Test Is Used to Identify the 100 People with the Disease

True characteristics in
population

Screening test Disease No Disease Total

Positive 80 100 180
Negative 20 800 820
Total 100 900 1,000
Sensitivity = 80/100 = 80%
Specificity = 800/900 = 89%
26
03-09-2014
Table 3 four possible groups resulting from screening with a dichotomous test

Test Disease

Present Absent
Positive a (True positive) b (False positive)
Negative c (False negative) d (True negative)

Table 4 grouping all people with positive test results and all people with negative test
results on screening
Test Disease
Present Absent
Positive a + b (All people with positive tests)

Negative c + d (All people with negative tests)

27
03-09-2014
Tests of Continuous Variables
Eg., Figure 1: Screening for DM in hypothetical population of ‘P’ 50%. Effects of choosing cutoff levels for
positive test

28
03-09-2014
 Figure 2: Effect of altering cut-off value for an
abnormal test
A Normal
Diseased
TNF
TPF

FNF FPF

Normal
B
TNF Diseased
TPF

FNF FPF

29
03-09-2014
Trade-off between sensitivity and specificity:

• If we increase the sensitivity by lowering the cut-off level,

we decrease the specificity;

• If we increase the specificity by raising the cut-off level, we

decrease the sensitivity

30
03-09-2014
 Receiver Operating Characteristic curve
• Test involves measurement of quantity & interpretation of result
based on level of measurement - ROC curve is used to choose the
cut-off.

• More the area under ROC curve, the better overall discriminating
ability of diagnostic criteria

• Rare condition or risky procedures like radiotherapy or surgery–

strict/lower threshold on curve

• Prevalence is high – lax/higher threshold on curve

31
03-09-2014
ROC Curve

Lax
Mod.

Strict

32
03-09-2014
 Contd…
• Choice of high or low cut-off depends on importance of false
positivity & false negativity for disease

• false positives - burden on the health care system

anxiety and worry

difficulty of delabeling a person

• false negatives – serious disease might be missed at an early

treatable stage

33
03-09-2014
 Use of Multiple Tests

Sequential (Two-stage) Testing

• less expensive, less invasive, or less uncomfortable test is

performed first, and

• those who screen positive are recalled for further testing with a
more expensive, more invasive, or more uncomfortable test,
greater sensitivity and specificity.

• Eg., Diabetes (RBS, GTT)

34
03-09-2014
Eg., Table 5: Hypothetical example of two-stage screening
program: I
Population-10000, ‘P’ 5%, Sn 70% & Sp 80%
Test Diabetes

Present Absent

Positive 350 1900 2250

Negative 150 7600 7750

500 9500 10000

35
03-09-2014
Eg., Table 6:Hypothetical example of a two-stage screening
program: II.

36
03-09-2014
 Sequential (Two-stage) Testing – contd..

• Net sensitivity = 315/500 = 63%

• Net specificity = 7,600 + 1,710 /9500

= 9,310/9,500 = 98%

Gain in net specificity

37
03-09-2014
 Simultaneous Testing

• Population of 1,000 people

• Prevalence of a disease is 20%

Test A Test B
Sensitivity = 80% Sensitivity = 90%

Specificity = 60% Specificity = 90%

38
03-09-2014
Eg., Table 7 - Result of Sensitivity with Test A

Population(1000), ‘P’ 20%

Screening test
Disease No Disease
Positive 160 320
Negative 40 480
Total 200 800
Sensitivity = 80%
Specificity = 60%

39
03-09-2014
Eg., Table 8- Result of Sensitivity with Test B

Population(1000), ‘P’ 20%

Screening test
Disease No Disease
Positive 180 80
Negative 20 720
Total 200 800
Sensitivity = 90%
Specificity = 90%

40
03-09-2014
Eg., Figure 3 Hypothetical example of simultaneous testing:
Net sensitivity

41
03-09-2014
Eg., Table 9- Result of Specificity with Test A

Population(1000), ‘P’ 20%

Screening test Disease No Disease
Positive 160 320
Negative 40 480
Total 200 800
Sensitivity = 80%
Specificity = 60%

42
03-09-2014
 Eg., Table 10- Result of Specificity with Test B

Population(1000), ‘P’ 20%

Screening test Disease No Disease
Positive 180 80
Negative 20 720
Total 200 800
Sensitivity = 90%
Specificity = 90%

43
03-09-2014
 Eg., Figure 4 Hypothetical example of simultaneous
testing: Net specificity

44
03-09-2014
Simultaneous Testing – contd..

• Net sensitivity = (16+144+36) / 200 = 98%

• Net specificity = 432/800 = 54%

• Gain in net Sensitivity

45
03-09-2014
Predictive value of a test
• Predictive value of Positive Test: Probability of the person
having disease when the test is positive
TP x 100
TP+FP
• Predictive value of Negative Test: Probability of the person
not having disease when the test is negative
TN x 100
FN+TN
• Affected by two factors: 1. prevalence of disease
2. specificity of the test

46
03-09-2014
 Predictive value of a test

Eg., Table 11- Predictive Value of a Test

Population
Screening test Disease No Disease Total
Positive 80 100 180
Negative 20 800 820
Total 100 900 1,000
Positive predictive value = 80/180 = 44%

Negative predictive value = 800/820 = 98%

47
03-09-2014
 Eg., Table 12- Relationship of Disease Prevalence to Positive
Predictive Value
Example: Sensitivity = 99%, specificity = 95%
Positive
Disease Test Predictive
Prevalence Results Sick Not Sick Totals Value
+ 99 495 594 99/594=17%
1% - 1 9,405 9,406
Totals 100 9,900 10,000
+ 495 475 970 495/970=51%
5% - 5 9,025 9,030
Totals 500 9,500 10,000

Higher the prevalence , increase in positive predictive value

48
03-09-2014
Eg., Table 13- Relationship of Specificity to Predictive Value

EXAMPLE: PREVALENCE = 10%, SENSITIVITY = 100%

Test
Specificity Results Sick Not Sick Totals Predictive Value
+ 1,000 2,700 3,700 1000/3700=27%
70% - 0 6,300 6,300
Totals 1,000 9,000 10,000
+ 1,000 450 1,450 1000/1450=69%
95% - 8,550 8,550 8,550
Totals 1,000 9,000 10,000

49
03-09-2014
Eg., Figure:5 Relationship of specificity to predictive value (PPV)

50
03-09-2014
Biases in evaluation of screening

1. Selection Biases

a. Referral Bias (Volunteer Bias)

b. Length-Biased Sampling (Prognostic Selection)

2. Lead Time Bias

3. Over diagnosis Bias

51
03-09-2014
1. Selection Biases

a. Referral Bias (Volunteer Bias)

• Differences in characteristics of those who participate in screening

& those who do not.

• Disease had a better prognosis – observe lower mortality

• People at high risk volunteer for screening because of anxieties

based on positive family history

• Addressed by comparison with a randomized experimental study

52
03-09-2014
b. Length-Biased Sampling (Prognostic Selection)

• Screening tends to selectively identify those cases that have longer

preclinical phases of illness (false appearance of better prognosis)

53
03-09-2014
2. Lead Time Bias

• Survival seems better because interval

from diagnosis to death is longer,

but the patient is not any better off

because death has not been delayed.

• Problem of an illusion of better survival only because of earlier

detection is called the lead time bias.

54
03-09-2014
Lead Time Bias

55
03-09-2014
3. Over diagnosis Bias

• If normal individuals in screened group are erroneously diagnosed

as positive than in the unscreened group - false impression of
increased rates of detection & diagnosis of disease as a result of
screening

• Diagnostic process be rigorously standardized in order to

minimize over diagnosis

56
03-09-2014
 Evaluation of screening tests
Based on feasibility & effectiveness

1. Randomized controlled trials

2. Uncontrolled trials

• to see if people with disease detected through screening appear to

live longer after diagnosis & treatment eg., cervical cancer screening
3. Other methods

• case control studies & comparison in trends b/w areas with different
degrees of screening coverage.
57
03-09-2014
 Conclusion

• Screening test have been overburdened with problems, many of

which remain unsolved

• Construction of accurate tests that are both sensitive & specific is

a key obstacle to wide application of screening

• Limitation and pit falls are to be borne in mind before

implementing a programme

• In today’s era of evidence based medicine, proper trials should

be done to evaluate screening programme
58
03-09-2014
 References
• Park K. Park’s Textbook of preventive and social medicine. 22 ed.
Jabalpur (India): Banarasidas Bhanot Publishers; 2013. p. 127-131.

• Gordis L. Epidemiology. 4th ed. Philadelphia (USA): Elsevier

Saunders; 2009. p. 85-106, 316-331.

• Hennekens CH, Buring JE. Epidemiology in medicine. 1st ed.

Boston: Little, Brown and Company; 1987. p. 327-350.

• Detels R, Beaglehole R, Lansang MA, Gulliford M. Oxford

Textbook of Public Health. 5th ed. United Kingdom: Oxford
University press; 2009. p. 1623-1629.
59
03-09-2014
 Contd...
• Bonita R, Beaglehole R, Kjellstrom T. Basic epidemiology. 2 nd ed.
Geneva: WHO; 2006. p. 110-114.

• Balwar R, Vaidya R, Tilak R, Guptha R, Kunte R. Public health

and Preventive Medicine. 8th ed. Department of community
medicine, AFMC, Pune; 2008. p. 57-60.

• Murthy NS, Nandakumar BS, Shalini CN, Shivraj NS, Gautham

MS, Prutvish S. Need for cancer screening program - pitfalls and
solutions. ISMS bulletin July 2010:3-11.

60
03-09-2014
 Bibliography
• Isabel DSS. Cancer Epidemiology: Principles and Methods. France:
WHO International Agency for Research on Cancer; 1999. p. 355-384.

• Andermann A, Blancquaert I , Beauchamp S, Dery V. Revisiting Wilson

and Jungner in the genomic age: a review of screening criteria over the
past 40 years. Bull World Health Organ 2008 Apr; 86(4):317-9.

• Altman DG, Bland JM. Diagnostic tests 1.sensitivity and specificity.

British Medical Journal 1994 Jun; 308:1552.

61
03-09-2014
Bibliography – contd…

• Altman DG, Bland JM. Diagnostic tests 3.receiver operating

characteristics plots. British Medical Journal 1994 Jul;
309:188.

• Altman DG, Bland JM. Diagnostic tests 2.predictive values.

British Medical Journal 1994 Jul; 309:102.

• Barnett AG, Pols JCV, Dobson AJ. Regression to the mean:

what it is and how to deal with. Int. J . Epidemiol. 2005.
34(1):215-220.

62
03-09-2014
 Thank You
Just because you're not sick doesn't mean
you're healthy !

63
03-09-2014
 64
03-09-2014
 Ruling in and ruling out a diagnosis
• To rule a diagnosis out you need a test that has high
sensitivity.
• A perfectly sensitive test will identify all cases of a disease, so
if you get a negative result on a sensitive test, you are fairly
sure that the patient does not have this disease. The mnemonic
is ‘SnNout: sensitive test, Negative result rules out.

• To rule a diagnosis in, you need a test that is high in

specificity.
• A positive result on a specific test, test would only identify
this type of disease. The mnemonic is ‘SpPin’ Specific test +
positive score rules in.
65
03-09-2014
 Likelihood Ratios

• Likelihood Ratio is a ratio of two probabilities

• Likelihood ratio is equal to the probability of the observed test
result among the diseased divided by the probability of the same
test result among the non-diseased
• Uses
• To choose a diagnostic test to rule in or rule or a disease
• To calculate the post test probability

66
03-09-2014
 LR+
The probability of a positive test result when
disease is present=TP/(TP+FN) OR TP/D+
------------------------------------------------------------
The probability of positive test result when
disease is absent =FP/(FP+TN) OR FP/D

LR for POSITIVE test result =[TP/(D+)]/[FP/(D-)]

= Sensitivity / [1-Specificity]

67
03-09-2014
 LR-
The probability of a negative test result when
disease is present=FN/(FN+TP) OR FN/D+
------------------------------------------------------------
The probability of positive test result when
disease is absent =TN/(TN+FP) OR TN/D

LR for NEGATIVE test result =[FN/(D+)]/[TN/(D-)]

= [1-Sensitivity] / Specificity

68
03-09-2014