Screening

Dr. Dalia El-ShafeiLecturer, Community Medicine Department,

Zagazig Universityhttp://www.slideshare.net/daliaelshafei

Levels of Prevention:Levels of Prevention:

No Disease No Disease AsymptomaticAsymptomatic

PreventionPrevention 1ry 1ry preventionprevention

2ry 2ry preventionprevention

3ry 3ry preventionprevention

Remove risk Remove risk factors factors

Early detection Early detection Early treatment Early treatment

Reduce Reduce complications complications

Clinical DiseaseClinical Disease

Screening :Screening :

Application of a test or a procedure to Application of a test or a procedure to large number of population who have no large number of population who have no symptoms of a particular disease for the symptoms of a particular disease for the purpose of determining their likelihood purpose of determining their likelihood of having the disease. of having the disease.

Asymptomatic

Symptomatic

Possibly diseased

4 outcomes

Diseased individuals +ve by screening

Diseased individuals -ve by screening

Non Diseased individuals -ve by screening

Non Diseased individuals +ve by screening

Goals reduce morbidity or mortality from Goals reduce morbidity or mortality from disease: disease:

Objectives : Objectives :

Early detection of disease among Early detection of disease among subclinical cases. subclinical cases.

Identify at risk individuals. Identify at risk individuals. Identify carriers of disease. Identify carriers of disease.

Screening TestsScreening Tests

They can be in the form of:They can be in the form of: QuestionsQuestions ExaminationsExaminations Laboratory testsLaboratory tests X-Rays ??? (Miniature Mass Radiography)X-Rays ??? (Miniature Mass Radiography)

Screening Program Screening Program

Mass screening. Mass screening. Selective screening. Selective screening. Opportunistic screening. Opportunistic screening.

Advantages :Advantages : Magnitude of disease can be precisely assessed. Magnitude of disease can be precisely assessed. Early detected cases can be controlled. Early detected cases can be controlled.

disadvantages :disadvantages : Not 100% accurate test. Not 100% accurate test. Costly. Costly. Adverse effect. Adverse effect. Anxiety due to false positives. Anxiety due to false positives. Sense of security due to false negatives. Sense of security due to false negatives.

Criteria for screening program Criteria for screening program 1- Disease : 1- Disease : Important health problem. Important health problem. Understood natural history. Understood natural history. Identifiable symptomatic stage.Identifiable symptomatic stage. Long DPCP. Long DPCP. 2- Treatment & Diagnosis: 2- Treatment & Diagnosis: Accepted or useful treatment. Accepted or useful treatment. Available facilities for further diagnosis & treatment. Available facilities for further diagnosis & treatment. Cost/benefit is balanced. Cost/benefit is balanced. Favorably influence prognosis “non-melanotic skin carcinoma: Favorably influence prognosis “non-melanotic skin carcinoma:

completely cureable” “Ca. cervix: good prognosis” “Ca. lung: completely cureable” “Ca. cervix: good prognosis” “Ca. lung: no value”.no value”.

3- Test : 3- Test :

Simple. Simple. Rapid. Rapid. Non-invasive. Non-invasive. Cheap. Cheap. Accepted. Accepted. Valid. Valid. Reliable. Reliable. Can be done by non-medicals. Can be done by non-medicals.

ScreeningScreening versus versus clinical examinationclinical examination

ScreeningClinical examination

Used in population studies

On individuals

Absence of medical indication

Presence of medical indication

Treatment can’t be described upon results

Treatment can be described upon results

Subjects can be classified:

Likely to be illLikely to be free

Subjects can be classified:Diseased

Not diseased

Terms Related to Terms Related to Screening TestsScreening Tests

ValidityValidity - relates to accuracy (correctness) - relates to accuracy (correctness)

ReliabilityReliability – repeatability – repeatability

AccuracyAccuracy -proportion of true test results among -proportion of true test results among all test results all test results

YieldYield - the # of tests that can be done in a time - the # of tests that can be done in a time

periodperiod

A. ReliabilityA. Reliability

The ability of a test or combination of tests to give consistent results in repeated applications, whether correct or incorrect.

This could be a function of the test, for example, one nurse making repeat blood pressure measurements on an individual; or of the person performing the test, for example, ten different nurses measuring the blood pressure of the same individual.

SensitivitySensitivityIt is the proportion of true positives It is the proportion of true positives among all cases: a/(a+c)among all cases: a/(a+c)

The ability of the test to detect true The ability of the test to detect true positives from all those who are positives from all those who are diseased.diseased.

B. ValidityB. ValidityMeasured by test’s ability to do what it’s Measured by test’s ability to do what it’s

supposed to do.supposed to do.

SpecificitySpecificityIt is the proportion of true negatives It is the proportion of true negatives

among all noncases : d/(b+d)among all noncases : d/(b+d)

The ability of the test to detect true The ability of the test to detect true negatives from all those not negatives from all those not diseased.diseased.

SENSETIVITY (+ve)

SPECIFICITY (-ve)

Important penalty for missing a disease

- Serious disease + definite ttt exist “TB,

Hodgkin’s dis.”- Spread “gonorrhea,

syphilis”

Subsequent diagnostic evaluation of +ves

associated with minimal risks & costs “BP for

HPN”

False +ve results harm pts. Physically,

emotionally, financially.- Cancer “chemotherapy ”

- HIV “stigma”

Subsequent diagnostic evaluation of +ves

associated with high risks & costs “biopsy for

breast cancer”Trade-offs between sensitivity & specificity:

Inverse relationship

Predictive valuesPredictive values Positive Predictive Value “PPV”:

- Probability of disease in a patient with +ve test result- Proportion of a +ve test that are truly +ve (truly diseased)

= a/a+b

Negative Predictive Value “NPV”: - Probability of disease in a patient with +ve test result

- Proportion of a -ve test that are truly -ve (truly non-diseased)= d/c+d

PPredictive value (PV) of a positive PPredictive value (PV) of a positive test:test:The proportion of a positive test that are The proportion of a positive test that are truly positive (truly diseased) : a/(a+b)truly positive (truly diseased) : a/(a+b)

The PV of a positive test increases with The PV of a positive test increases with increasing sensitivity and specificity.increasing sensitivity and specificity.

If the prevalence of a disease in the If the prevalence of a disease in the population increases the PV also increases population increases the PV also increases and the reverse is true.and the reverse is true.

High risk population are frequently chosen High risk population are frequently chosen for screening thus increasing the yield and for screening thus increasing the yield and PV of a positive test.PV of a positive test.

The predictive value of a positive test increases as

the prevalence of diseases increases even with the same sensitivity

& specificity of the screening test.

(See the following example)

Disease Non Diseased 50 50

100

50 50 100 100

100200

True Diagnosis

Total

Test Result

Positive

Negative

Disease Non Diseased 60 40

100

60 40 120 80

100200 Disease Non

Diseased 40 60

100

40 60 80 120

100200

Application:If the target condition is sufficiently rare, even tests with excellent sensitivity & specificity can have low positive predictive value (PPV) generating more false positives than true positive results.

C. AccuracyC. Accuracy

It is the proportion of true test It is the proportion of true test results among all test results:results among all test results:

(a+d)/(a+b+c+d)(a+d)/(a+b+c+d)

Gold standard test +ve-ve Total

+ve

a(true+ve)

b(false+v

e)

a +bPVP= a/a+b PVP= a/a+b x100x100

–vec(false -

ve)

d(true -

ve)

c+dPVN=d/c+d PVN=d/c+d x 100x 100

Total a+cb+da+b+c+d

Sensitivity = Sensitivity = a/a+c x 100a/a+c x 100

Specificity = Specificity = d/b+d x 100d/b+d x 100

Accuracy= a+d/a+b+c+d

Scre

enin

g te

st

Example: Example:

+ve+ve-ve-veTotal Total +ve +ve

––veve

15(true+15(true+ve) ve)

10(false -10(false -ve)ve)

30(false+30(false+ve)ve)

45(true 45(true ––ve)ve)

45455555

Total Total 25 (dis.)25 (dis.)75 (free)75 (free)100100 Sensitivity = 15/25 x 100 = 60%Sensitivity = 15/25 x 100 = 60% Specificity = 45/75 x 100 = 60%Specificity = 45/75 x 100 = 60% Predictive value +ve = 15/45x 100= 33.3%Predictive value +ve = 15/45x 100= 33.3% Predictive value –ve =45/55x 100= 81.8%Predictive value –ve =45/55x 100= 81.8% Accuracy= 15+45/100 x 100=60%Accuracy= 15+45/100 x 100=60%

Mammography (gold standard)Mammography (gold standard)Se

lf ex

am

Self

exam

(s

cree

ning

) (s

cree

ning

)

True positives

True negatives

False positives

False negatives

True positives

True negatives

False positives

False negatives

80

60

40

20

Total 100 100

•Sensitivity: True Positives All Diseased

a/(a+c) = 80%

•Specificity: True Negatives All non diseased

d/(b+d) = 60%

Validity

Screening EthicsScreening Ethics

Informed consent for testing and follow up.Informed consent for testing and follow up. Considerations of the risks of screening.Considerations of the risks of screening. Distributive justice.Distributive justice.

Risks of ScreeningRisks of Screening

A.A. True PositiveTrue Positive““Labeling effect” Person is classified as Labeling effect” Person is classified as

“diseased” from the time of the test forward in “diseased” from the time of the test forward in time.time.

B.B. False PositiveFalse Positive Financial burdenFinancial burden

Harm from confirmatory test (which may be Harm from confirmatory test (which may be invasive)invasive)

-ve psychological impact-ve psychological impact Fear of future screens “phobia”Fear of future screens “phobia”

Risks of ScreeningRisks of Screening

C.C. True NegativesTrue NegativesCosts & risks of screening testsCosts & risks of screening tests

D.D. False NegativesFalse Negatives- False sense of security.- False sense of security.- Delayed interventionDelayed intervention

- Disregard of early signs and symptomsDisregard of early signs and symptoms- Loss of confidence in medical care systemLoss of confidence in medical care system

Exercise :-Exercise :- A medical research team conduct a trial to A medical research team conduct a trial to

find if high plasma level of breast carcinoma find if high plasma level of breast carcinoma promoting factor (BCPF) could be used to promoting factor (BCPF) could be used to diagnose breast cancer.diagnose breast cancer.

Out of 1600 patients included in the Out of 1600 patients included in the study ,600 demonstrated by breast study ,600 demonstrated by breast biopsy(the gold standard) to have breast biopsy(the gold standard) to have breast cancer (D+) and 1000 were found to be cancer (D+) and 1000 were found to be disease –free(D-) disease –free(D-)

Out of the 600 demonstrated to have breast Out of the 600 demonstrated to have breast cancer ,570 were positive by BCPF(T+) and cancer ,570 were positive by BCPF(T+) and Out of the 1000 were found to be disease –Out of the 1000 were found to be disease –free,850 were negative by BCPF(T-) free,850 were negative by BCPF(T-)

It is an example of studying the performance It is an example of studying the performance of a new diagnostic testof a new diagnostic test

PATHOLOGYPATHOLOGY

STUDIED TESTSTUDIED TEST

Breast Breast cancer(D+)cancer(D+)

No breast No breast cancer (D-)cancer (D-)

TotalTotal

Marker (+)Marker (+)(T+)(T+)

570 (TP)570 (TP)150 ( FP)150 ( FP)720720

Marker (-)Marker (-)(T-)(T-)

30 (FN)30 (FN)850 (TN)850 (TN)880880

TotalTotal6006001000100016001600

Sensitivity, specificity, predictive value positive & Sensitivity, specificity, predictive value positive & predictive value negative can be calculatedpredictive value negative can be calculated

Sensitivity=Sensitivity= 570/600 = 0.95 = 95%570/600 = 0.95 = 95%

Specificity=850/1000 = 0.85 = 85%Specificity=850/1000 = 0.85 = 85%

Predictive value positive= Predictive value positive= 570/720 = 0.79=79%570/720 = 0.79=79%

Predictive value negative= Predictive value negative= 850/880 = 0.97=97%850/880 = 0.97=97%

Another Exercise :-Another Exercise :- A medical research team conduct a trial to find A medical research team conduct a trial to find

if a blood marker could be used to diagnose if a blood marker could be used to diagnose breast cancer.breast cancer.

Out of 1600 patients included in the study ,600 Out of 1600 patients included in the study ,600 demonstrated by breast biopsy(the gold demonstrated by breast biopsy(the gold standard) to have breast cancer (D+) and 1000 standard) to have breast cancer (D+) and 1000 were found to be disease –free(D-) were found to be disease –free(D-)

Out of the 600 demonstrated to have breast Out of the 600 demonstrated to have breast cancer ,570 were positive by the marker cancer ,570 were positive by the marker “BCPF”(T+) and Out of the 1000 were found to “BCPF”(T+) and Out of the 1000 were found to be disease –free,850 were negative by be disease –free,850 were negative by “BCPF”(T-) “BCPF”(T-)

Feedback of the another Exercise :-Feedback of the another Exercise :- It is an It is an example of studying the validity of a new example of studying the validity of a new

screening diagnostic testscreening diagnostic test PATHOLOGY

STUDIED TEST

Breast cancer(D+)

No breast cancer (D-)

Total

Marker (+)(T+)

570 (TP)150 ( FP)720

Marker (-)(T-)

30 (FN)850 (TN)880

Total60010001600

Feedback of Exercise (cont.):-Feedback of Exercise (cont.):- Sensitivity, specificity, predictive value positive & Sensitivity, specificity, predictive value positive &

predictive value negative can be calculatedpredictive value negative can be calculated

Sensitivity=Sensitivity= 570/600 = 0.95 = 95%570/600 = 0.95 = 95%

Specificity=850/1000 = 0.85 = 85%Specificity=850/1000 = 0.85 = 85%

Predictive value positive= Predictive value positive= 570/720 = 0.79=79%570/720 = 0.79=79%

Predictive value negative= Predictive value negative= 850/880 = 0.97=97%850/880 = 0.97=97%

Find the validity of testing sugar in urine for Find the validity of testing sugar in urine for detection of diabetes from the following tabledetection of diabetes from the following table

gold standardScreening

Blood sugar curve +veDiabetic

Blood sugar curve –veNon- Diabetic

Total

+ ve diabetes byurine test

251540

- ve diabetes byurine test

204060

Total4555100

Sensitivity= 25/45 = 55.5%Sensitivity= 25/45 = 55.5%

Specificity=40/55 = 72.7%Specificity=40/55 = 72.7%

Predictive value positive= Predictive value positive= 25/40 = 62.3%25/40 = 62.3%

Predictive value negative= Predictive value negative= 40/60 = 66.7%40/60 = 66.7%

Two hundred individuals (80with and 120 Two hundred individuals (80with and 120 without infarction were examined by two without infarction were examined by two laboratory methods (A&B) to find out which of laboratory methods (A&B) to find out which of these lab. Tests is more valid in detection of these lab. Tests is more valid in detection of coronary infarction:coronary infarction:

Test A: the number of detected infraction cases by Test A: the number of detected infraction cases by test were 70.40 of them were truly infracted cases.test were 70.40 of them were truly infracted cases.

Test B: the number of detected infraction cases by Test B: the number of detected infraction cases by test were 100.60 of them were truly infracted cases.test were 100.60 of them were truly infracted cases.

Gold standard

ScreeningTEST

InfarctionNo infarctionTotal

+ ve infarction by test A

40 (TP)30 ( FP)70

- ve infarction byTest A

40 (FN)90 (TN)130

Total80120200

test Atest A

Sensitivity= 40/80= 50%Sensitivity= 40/80= 50%

Specificity=90/55120 = 75%Specificity=90/55120 = 75%

Predictive value positive= Predictive value positive= 40/70 = 57%40/70 = 57%

Predictive value negative= Predictive value negative= 90/130 = 69%90/130 = 69%

))test Btest B((

Gold standard

ScreeningTEST

InfarctionNo infarctionTotal

+ ve infarction by test B

60 (TP)40 ( FP)100

- ve infarction byTest B

20 (FN)80 (TN)100

Total80120200

))test Btest B((

Sensitivity= 60/80 = 75%Sensitivity= 60/80 = 75%

Specificity=80/120 = 66.7%Specificity=80/120 = 66.7%

Predictive value positive= Predictive value positive= 60/100 = 60%60/100 = 60%

Predictive value negative= Predictive value negative= 80/100 = 80%80/100 = 80%

IGTNo IGT

Total

+ve50TP

35FP

85

-ve8FN

103TN

111

Total58138196

Predictive value of positive = 50 =58.8% 85

Predictive value of negative = 103 =92.8% 111

Remember:- Sensitivity = 86.2% Specificity = 74.6%

Predictive value varies with prevalence (pretest probability).