Principles of Screening

Question 1

Describe screening tests in general?

Answer 1

Presumptive identification of unrecognised disease or defect by application of tests, examinations or other procedures which can be applied rapidly
* Screening tests 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
* Person with positive or suspicious findings must be referred to physician for diagnosis & necessary treatment”

Question 2

What is epidemiology & what does epidemiologist do?

Answer 2

• Scientific study of factors affecting health & illness of populations
• Epidemiologists employ range of study designs from observational to experimental with purpose of revealing unbiased relationships between exposures e.g.,
  o Nutrition, biological agents, stress, or chemicals to outcomes such as disease, wellness and health indicators

Question 3

Name the 10 screening principles?

Answer 3

1. Condition being sought should be an important health problem, for individual & community
2. There should be an acceptable & effective form of treatment for patients / recognisable disease
3. Natural history of condition including its development from latent to declared disease, should be adequately understood
4. There should be a recognisable latent or early symptomatic stage
5. There should be a suitable screening test or examination for detecting disease at latent or early symptomatic stage, & this test should be acceptable to the population
   o Need people to be up for doing the test & not boycotting it
6. Facilities required for diagnosis & treatment of patients revealed by screening program should be available
7. There should be an agreed policy on whom to refer & treat as patients
8. Treatment at pre-symptomatic, borderline stage of disease should favourably influence its course & prognosis
9. Cost of case-finding (including cost of diagnosis & treatment) needs to be economically balanced in relation to possible expenditure on medical care as a whole
10. Case-finding should be continuous, not a “once & for all” project

Question 4

Whhat are the 4 sub-divisions of screening and describe them?

Answer 4

• Mass Screening – large-scale screening of whole population groups
  o No selection of population groups is made:
   Tuberculosis/Covid-19
   Visual field defects
• Selective Screening – screening of selected groups in population who are at higher risk of whatever screening is testing for
  o Blue colour vision in divers
  o Visual defects in preschool children – looking for misalignment of eyes, mismatched refractive error (amblyopia)which had not been found previously
• Multiphasic Screening – application of 2 or more screening tests in combination to large groups of people
  o Vision & hearing screening at same time
• Monophasic Screening – application of single screening test

Question 5

What are the purposes of screening?

Answer 5

• Saves time & money in long run
• Short test which is easy to administer – for large groups of people
• Should be inexpensive
• Should be accurate
• It cannot be diagnostic test

Question 6

Describe the validity & reliability of screening tests?

Answer 6

• Screening gives 2 results:
  o Positive test – these people are referred for further investigation
  o Negative test – these people are not referred
• Both of these groups will contain people who suffer from the defect & those who do not suffer from the defect
• Validity of test quantifies accuracy of test
  o Ability of test to correctly determine normality & abnormality
• Of those who test +ve there will be:
  o A number a who have defect – true positives & represent correct referrals
  o A number b who do not have defect – false positives & represent over-referrals
• Of those who test -ve there will be:
  o A number c who have the defect – false negatives & represent under-referrals
  o A number d who do not have defect – true negative and represent correct non-referrals
• Validity of test is usually measured in terms of its:
  o Sensitivity
  o Specificity
• Test that has no false +ves and no false -ves would have 100% sensitivity & 100% specificity
  o This NEVER happens
• Always some error which reduces sensitivity & specificity below 100%

Question 7

Describe sensitivity?

Answer 7

• Ability to correctly determine individuals who suffer from defect
• Proportion of those individuals who have defect that test positive
• Highly sensitive test will find nearly all individuals who have defect
• Test that finds v few individuals who have the defect is not a sensitive one
• Sensitivity = a / (a+c)

Question 8

Describe specificity?

Answer 8

• Ability to correctly identify individuals who do not suffer from defect
• Proportion of those individuals who do not have defect that test negative
• Highly specific test will correctly determine nearly all normal population
• Test has many normal individuals misclassified as abnormal due to a positive test result is not a specific one
• Specificity = d / (b+d)

Question 9

Describe reliability of screening test?

Answer 9

• Refers to test’s repeatability
• Repeatability of test is usually expresses as a test-retest correlation co-efficient:
  o 1 – result of 2nd test is always perfectly predictable from result of first test
  o 0 – it is impossible to predict result of second test from result of first test
• For clinical tests a correlation co-efficient of 0.8 or better is considered satisfactory
• A correlation co-efficient of 0.7 may be acceptable
  o E.g. when measuring the IOP in morning vs evening it would be less than 0.7 correlation co-efficient due to diurnal variation

Question 10

What are some reasons why the screening test can give different results?

Answer 10

• Variations in test or test conditions
• Inconsistencies in test procedure or examiner
• Inherent differences in individuals being tests
  o Screening test large groups of people so don’t want these to be so great that they wipe out the positivity/efficacy of the test

Question 11

Describe errors in measurements of screening test?

Answer 11

• Random errors increase standard deviation of measurement, without affecting mean of measurement
• Systematic errors, e.g. zero error in calibration of a Goldmann tonometer, will change mean but will leave standard deviation of measurement unaffected

Question 12

Describe effectiveness of screening tests?

Answer 12

• Reliable & valid tests
• Reliability is expressed in terms of test-retest correlation co-efficient
• Validity is expressed in terms of sensitivity & specificity of test
• A good test will have this info available, so it can be decided whether the test is suitable as a screening test, or not
  Test has to have good validity and reliability to be used on a big population sample

Question 13

Describe signal detection theory?

Answer 13

• Means to quantify ability to distinguish between info-bearing patterns – ‘signals’ or ‘stimuli’ and random patterns or noise
• Signal Recovery is separation from random background of signal or stimulus
• Signal Recovery – trying to get info from a v noisy background
• Signal Recovery is dependent on threshold levels – whether by a machine or by human nervous system
• Human example: wartime personnel on red alert would be able to detect faintest bit of unusual noise that they would be impervious to in a ‘normal’ environment
• Applications of Signal Detection Theory are in radar, psychophysics & psychology

Question 14

Describe Reciever Operating Characteristic (ROC)?

Answer 14

• ROC or ROC curve – plot that illustrates performance of a yes/no or binary classifier system as its discrimination threshold is varied
• Curve is created by plotting true positive rate against false positive rate at various threshold settings
• In biological sciences, true-positive rate is also known as sensitivity
• In machine learning, it is called Recall
• False-positive rate is called fall-out and can be calculated as 1-specificity
• ROC curve is thus the sensitivity as a function of fall-out
• In general, if probability distributions for both detection and false alarm are known, the ROC curve can be calculated

Question 15

Describe ROC Curves?

Answer 15

• A diagram of sensitivity, plotted on a vertical axis, against specificity, plotted backwards on a horizontal axis (i.e. from 100% at left to 0% at right) for various values of the cut-off point
• The top-left corner of the graph represents 100% sensitivity and 100% specificity
• From graph, 21mmHg is almost perfect as it has high sensitivity and high specificity – at 26mmHg, lose on sensitivity so could have a false +ve for a glaucomatous patient/image
• Diagonal line connecting bottom left corner and top right corner is the No information line, because all points lying on this line tell us nothing to help separate normal from abnormal conditions
• Greater the area between the curve and the no information line, the better the test is at discriminating between the 2 populations
• In order to find the optimal cut-off point, move a parallel to the no information line, towards the top left corner of the graph
• The last point on the curve to be touched represents the optimal cut-off value, and can read off the sensitivity and specificity of the test as that cut-off level – 21mmHg at that cut-off point