Screening & Conditional probability

Dr. Unaib Rabbani

Session Objectives:
Describe the concept of screening and its importance Know different types of screening and the criteria for a good screening test Appreciate the concept of lead time and lag time bias in screening test Understand the important Epidemiological terms related to screening test, including Validity (sensitivity& specificity), Reliability, and Yield (positive/negative predictive values) Apply Bayes theorem and conditional probability in computing sensitivity, specificity and predictive values for screening test

What is Screening? Basic Public Health Concepts

Screening is a strategy used in a population to detect a disease in individuals prior to the occurrence of signs/symptoms of that disease

Screening tests are performed on persons without any clinical sign or symptoms of disease
The purpose of screening is to identify disease in a community early, thus enabling earlier intervention and management to reduce disease mortality and morbidity

Screening Test are Concerned with a Functional Definition of Normality versus Abnormality

Difference b/w screening and diagnostic test

Screening
Done on apparently healthy people Applied to groups Usu. one disease considered Based on one criteria

Diagnostic test
Done on sick people with S/S Applied to individual patients Diff. diagnosis is ruled out Based on evaluation of various S/S

Less accurate
Less expensive Not a basis of treatment Initiative comes from investigator

More accurate
More expensive Used as a basis of treatment Initiative comes from pt. with S/S

Types of screening
Several types of screening exist:
Mass screening: involves screening of all individuals in a certain category (for example, all children of a certain age)
Case finding/High risk or selective screening involves screening a smaller group of people based on the presence of risk factors (for example, because a family member has been diagnosed with a hereditary disease) Multi-phasic screening involves 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 i.e questionnaire, blood tests, urine D/R

Key Elements of screening

Disease/disorder/defect

Screening test
Population

When a disease should be screened?

Disease/disorder should be an important public health problem
High prevalence Serious outcome

Early detection in asymptomatic (pre-clinical) individuals is possible Early detection and treatment can affect the course of disease (or affect the public health problem) Effective treatment is available

Criteria for Evaluating a Screening Test

Validity: provide indication of who does and does not have disease
Sensitivity of the test Specificity of the test

Reliability: (precision): gives consistent results when given to same person under the same conditions
Yield: Amount of disease detected in the population, relative to the effort
Prevalence of disease/predictive value

Validity of Screening Test (Accuracy)

Sensitivity: Is the test detecting true cases of disease.(Ideal is

100%: 100% of cases are detected)

Sensitivity=True positives/Total diseased =a/a + c Specificity: Is the test excluding those without disease? (Ideal is 100%: 100% of non-cases are negative) Specificity=True negatives/Total non diseased =d/b+ d

Screening for Glaucoma using IOP

True Cases of Glaucoma Yes No IOP > 22: Yes 50 100

No
(Total)

50
100

1900
2000 False Negative=50% False Positive=5%

Sensitivity = 50% (50/100) Specificity = 95% (1900/2000)

 Screening for categorical variables Screening for continuous variables

Example
520 women were positive for Breast cancer when fine needle aspiration cytology(FNAC) was used on 5,000 women, whom 500 were positive. Out of 520 positive by (FNAC) and 420 were found positive on excision biopsy. What is validity and of (FNAC) for diagnosing the Breast Cancer?

Receiver operator Characteristic ROC Curve

A plot of the true positive rate against the false positive rate for the different possible cut-points of a diagnostic test is called a ROC curve It shows the tradeoff between sensitivity and specificity (any increase in sensitivity will be accompanied by a decrease in specificity) The closer the curve follows the left-hand border and then the top border of the ROC space, the more accurate the test

Example
Cut-point 5 7 9 True Positives 0.56 0.78 0.91 False Positives 0.01 0.19 0.58

Where do we set the cut-off for a screening test?

Consider: The impact of high number of false positives:
anxiety of patient cost of further testing cause burden on health system

Importance of not missing a case: seriousness of disease

Reliability (reproducibility)
Intra-subject variation Intra-observer variation Inter-observer variation Overall % agreement for inter-observer variation=a/a+b+c Kappa statistic = % agreement observed-% agreement expected by chance alone 100%- % agreement expected by chance alone OR

Yield from the Screening Test: Predictive Value

Predictive Value of a Positive Test (PPV):
Likelihood that a person with a positive test has the disease PPV= true positive /test positive= a/a + b

Predictive Value of a Negative Test (NPV):

Likelihood that a person with a negative test does not have the disease NPV=true negative/test negative =d/c + d

Relationship with Sensitivity, Specificity, and Prevalence of Disease

High Prevalence ------ High PPV High specificity & sensitivity--------High Predictive Values

Example
Prevalence of a disease in a community was 15%. A screening test with 85% sensitivity and 50% specificity was applied on 1000 individuals. Calculate predictive values What if prevalence changes to 20%?

Yield from a Screening Test for Disease X Predictive Value

Principles for Screening Programs (Wilsons Criteria)

1. The condition sought should be an important health problem. 2. There should be an accepted treatment for patients with recognized disease. 3. Facilities for diagnosis and treatment should be available. 4. There should be a recognizable latent or early symptomatic stage. 5. There should be a suitable test or examination. 6. The test should be acceptable to the population. 7. The natural history of the condition, including development from latent to declared disease, should be adequately understood. 8. There should be an agreed policy on whom to treat as patients. 9. The cost of case-finding (including diagnosis and treatment of patients diagnosed) should be economically balanced in relation to possible expenditure on medical care as a whole. 10. Case-finding should be a continuing process and not a once and for all project.

Advantages and disadvantages of screening

Advantages Screening can detect medical conditions at early stage before symptoms when treatment is more effective than for later detection In the best of cases lives are saved Disadvantages Tests used in screening are not perfect The test result may show false positive, or false negative Screening for low probability condition---absolute number of false +ves high though the % of true +ves is low

Disadvantages of screeningcont
Adverse effects of screening procedure (e.g. stress and anxiety, discomfort, radiation exposure, chemical exposure)

Disadvantages of screeningcont
Lead time bias The intention of screening is to diagnose a disease earlier than it would be without screening Without screening disease discovered when symptoms appear Even if in both cases a person will die at the same time, because we diagnosed the disease earlier with screening the survival time since diagnosis is longer with screening; but life span has not been prolonged, and there will be added anxiety as the patient must live with knowledge of the disease for longer Looking at statistics of survival time since diagnosis, screening will show an increase (this gain is called lead time)

Disadvantages of screeningcont
Lag time bias Many screening tests involve the detection of cancers It is often hypothesized that slower-growing tumors have better prognoses than tumors with high growth rates Screening is more likely to detect slower-growing tumors (due to longer pre-clinical time), which may be less deadly Thus screening may tend to detect cancers that would not have killed the patient or even been detected prior to death from other causes

Conditional probability and its application of in computing sensitivity, specificity and predictive values for screening test

Conditional Probability
Probability of occurrence of an event given that another event B has already occurred Notation: P A B = P(A and B) P(B) provided P(B) is not eq. to zero Applications: Diagnosis of medical conditions (Sensitivity/Specificity) Data Analysis and model comparison

Conditional Probability Example

Diagnosis using a clinical test Sample Space = all patients tested Event A: Subject has disease Event B: Test is positive

Interpret: Probability patient has disease and positive test (correct!) Probability patient has disease BUT negative test (false negative) Probability patient has no disease BUT positive test (false positive) Probability patient has disease given a positive test Probability patient has disease given a negative test

p A B ' p A' B pA B p A B '

p A B

Conditional probability for test validity

Sensitivity=a/a+c
p A B ' p A' B pA B p A B ' p A B

P(T/D)=P(TD)/P(D)

Conditional probability for test validity

Specificity=d/b+d
P(T-/D-)=P(T-D-)/P(D-)

Bayes theorem
A theorem describing how the conditional probability of each of a set of possible causes for a given observed outcome can be computed from knowledge of the probability of each cause and the conditional probability of the outcome of each cause Bayes' theorem shows how to determine inverse probabilities: knowing the conditional probability of B given A, what is the conditional probability of A given B?

Bayes theoremcont
We can use Bayes Rule to find the predictive values if we know the sensitivity and the specificity of the screening instruments Let D and D- denote the events that the disease is actually present and absent respectively T+ is the event that the screening test gives a positive result and T- is the event that the screening test gives a negative result Let P(D)=The probability of disease in the general population Using Bayes Rule we get

Bayes theoremcont
P(D T+)=P(T+ D)P(D) P(T+ D)P(D)+P(T+ D-)P(D-)

Calculating predictive values using Bayes rule

PPV=P(D T)=P(T D)P(D) P(T D)P(D)+P(T D-)P(D-)
= sensitivity*prevalence e

(sensitivity*prevalence)+(1-specificity)*(1-prevalence)

Calculating predictive values using Bayes rule

NPV=P(D- T-)=P(T- D-).P(D-) P(T- D-)P(D-)+P(T- D)P(D) = specificity*(1-prevalence) specificity*(1-prevalence)+ (sensitivity)* prevalence

Statistics vs. Probability In statistics, you put your hand into a black box of marbles of diff color, examine your handful, and try to ans the question ``What is in the box?'' In probability, you look into a transparent box, count the different colored marbles in it, mix them up well, and then blindly take out one handful; without opening your eyes, you predict how many marbles of each kind are in your hand