1. Dr. Hannah Jordan Lecturer in Public Health ScHARR
Screening
Dr. Hannah Jordan
Lecturer in Public Health
ScHARR

2. Learning objectives
Define and calculate sensitivity and specificity of a screening test
Define and calculate positive and negative predictive values
Explain the relationship between prevalence and predictive values
Discuss criteria for screening
Describe specific biases in relation to screening (selection bias, lead-time bias, length-time bias)

3. Background: cultural norms
US – annual health check
Screening tests and treatments may be offered by different agencies
UK – rare to be screened
Identification through to treatment all in the NHS
Or…
May be no ‘system’, just a test
Is lots of screening a “good thing”?

4. What is screening?
A process which sorts out apparently well people who probably have a disease (or precursors or susceptibility to a disease) from those who probably do not.
Why?
To make a real difference to health
It is not intended to be diagnostic (diagnostic tests are different).
It is a process, not simply a test

5. The Purpose of screening
Prevention, treatment and information
The Purpose of screening

6. Prevention No disease No symptoms Clinical disease Primary prevention
Secondary prevention
Tertiary prevention

7. Main purpose of screening
Secondary prevention
Aim – detect early disease in order to alter the course of the disease
e.g. screening by mammography for breast cancer in order to treat it early
Primary prevention
Aim - prevent a disease from occurring
screening to identify people with risk factors and reduce risk factor levels

8. What do we want to achieve?
Reduce the risk of developing disease
Provide treatment
Provide information
E.g. pre-natal screening for genetic disorders

9. The sieve Screened population The screening test Further investigation
Discharged from screening
Screened population
The screening test

10. Measuring the effectiveness of screening
Sensitivity and specificity
Measuring the effectiveness of screening

11. What screening does The truth Disease Present Disease Absent
The test result
Positive
True positive a
False positive b
a+b
Negative
False negative c
True negative d
c+d
a+c
b+d
a+b+c+d

12. Definitions: sensitivity
Sensitivity – the proportion of people with the disease who are correctly identified by the screening test
a / a+c
The truth
Disease Present
Disease Absent
The test result
Positive
True positive a
False positive b
a+b
Negative
False negative c
True negative d
c+d
a+c
b+d
a+b+c+d

13. Definitions: specificity
Specificity – the proportion of people without the disease who are correctly excluded by the screening test
d / b+d
The truth
Disease Present
Disease Absent
The test result
Positive
True positive a
False positive b
a+b
Negative
False negative c
True negative d
c+d
a+c
b+d
a+b+c+d

14. Definitions: PPV a / a+b
Positive predictive value – the proportion of people with a positive test result who actually have the disease
a / a+b
The truth
Disease Present
Disease Absent
The test result
Positive
True positive a
False positive b
a+b
Negative
False negative c
True negative d
c+d
a+c
b+d
a+b+c+d

15. Definitions: NPV d / c+d
Negative predictive value – the proportion of people with a negative test result who do not have the disease
d / c+d
The truth
Disease Present
Disease Absent
The test result
Positive
True positive a
False positive b
a+b
Negative
False negative c
True negative d
c+d
a+c
b+d
a+b+c+d

16. Example 1: high prevalence
cancer
present
absent
cytology
positive
14 (a)
2 (b)
negative
6 (c)
78 (d)
Sensitivity = 14/20 = 70%
Specificity = 78/80 = 97.5%
Prevalence = 20/100 = 20%
Prevalence of “no disease” = 80/100 = 80%
Positive predictive value = 14/16 = 87.5%
Negative predictive value = 78/84 = 92.9%
Sensitivity: a/a+c
Specificity: d/b+d
Prevalence: a+c/a+b+c+d
PPV: a/a+b
NPV: d/c+d

17. Example 2: low prevalence
cancer
present
absent
cytology
positive
7 (a)
250 (b)
negative
3 (c)
9740 (d)
Sensitivity = 7/10 = 70%
Specificity = 9740/9990 = 97.5%
Prevalence = 10/10000 = 0.10%
Prev “no disease” = 9990/10000 = 99.9%
PPV = 7/257 = 2.7 %
NPV= 9740/9743 = 99.97%
Sensitivity: a/a+c
Specificity: d/b+d
Prevalence: a+c/a+b+c+d
PPV: a/a+b
NPV: d/c+d

18. Why is the positive predictive value so different?
Predictive values are dependent on underlying prevalence
(Sensitivity and specificity are not)

19. Continuous screening variable – effect of cut-off chosen on sensitivity and specificity

20. Wilson and Jungner (and others)
Screening criteria

21. Criteria for screening
Most based on Wilson and Jungner criteria
The condition
The condition sought should be an important health problem
The natural history of the condition should be well understood
There should be a detectable early stage

22. Criteria for screening
The treatment
There should be an accepted treatment for patients with recognized disease.
Facilities for diagnosis and treatment should be available
Adequate health service provision should be made for the extra clinical workload resulting from screening

23. Criteria for screening
The Test
A suitable test should be devised for the early stage
The test should be acceptable
Intervals for repeating the test should be determined (not a one off)

24. Criteria for screening
Risks and benefits
There should be an agreed policy on whom to treat
The costs should be balanced against the benefits
Additionally
The risks, both physical and psychological, should be less than the benefits

25. Evaluation of screening
Ideally by RCT
individual
cluster
Could use other methods but potential for bias
Well recognised biases
Selection bias
Lead-time bias
Length-time (or length) bias

26. Selection bias
People who choose to participate in screening programmes may be different from those who do not
May be at higher risk
e.g. women with family history of breast cancer more likely to attend
May be at lower risk
e.g. women in higher socioeconomic groups (lower risk of cervical cancer) more likely to attend

27. Lead time bias Disease starts Patient A diagnosed after screening
Death
Patient B diagnosed when symptoms develop

28. Length-time (or length) bias

29. Screening examples

31. e.g. breast cancer screening
Eligible group: women (within an age range)
Test: 2 view mammography
Further tests: mammography, biopsy (tissue), ultrasound, cytology (cells)
Treatment: surgery, radiotherapy, chemotherapy
Avoiding: overall deaths in screened women

32. Types of screening
Population-based screening programmes (“mass” screening)
Thailand, national diabetes and hypertension screening
Opportunistic screening
Prevention and control of substance abuse
Screening for communicable diseases
Heaf test?
Pre-employment and occupational medicals
Vision test for commercial drivers?
Commercially provided screening
Screening is more than a test, it is a programme

33. Pros and cons of screening
Ethics?
Mistakes?
Limitations?
Risk reduction or disease diagnosis?

34. Some summary points
Healthy people are more likely to get screened than less healthy people. This bias means that outcome in screened people looks good even if screening makes no difference.
Screening is more likely to pick up cases that stay symptomless for a long time and less likely to pick up rapidly progressive disease. This means that cases found by screening are a different group compared with cases that present clinically. They have better prognosis, even if screening makes no difference. This bias is called length time bias.
Screening often uncovers cases that would have no clinical impact in the person’s lifetime. These cases are a different group compared with cases that present clinically. They have excellent prognosis, and screening leads to the person receiving diagnosis and treatment that they would not otherwise have. This makes outcome in a screen-detected group appear good. This bias is called overdiagnosis bias.
Test performance can be expressed as sensitivity, specificity and predictive values
To describe all the consequences of screening properly, the evidence about benefits and harms should be summarized as simple frequencies. This can be done using a flow diagram for a screened population.
Development of comprehensive information for potential participants is a relatively new venture, but is progressing

35. Next Two workbooks on Minerva
You have timetabled space to complete these, both this afternoon and Thursday afternoon.