Screening I

Question 1

Be able to calculate and interpret the risk of being affected with a disorder for an individual based on a quantitative screening marker.

Answer 1

h

Question 2

What is meant by a ‘detection rate’ e.g. a DR of 75%?

Answer 2

Using the example of spina bifida and the maternal serum AFP screening test, a DR of 75% means that 75% of pregnancies with open spina bifida have positive results.

Question 3

How do you calculate detection rate?

Answer 3

number of affected with positive test results/total number of affected

Question 4

How do you calculate false positive rate?

Answer 4

number of unaffected with positive test results/total number of unaffected

Question 5

If the birth prevalence of OSB was 1 per 1000 instead of 2 per 1000. How would the OAPR be affected?

Answer 5

Lower (1:40, about 2.5%)

Question 6

How does the positive predictive value differ from OAPR?

Answer 6

OAPR is a ratio e.g. 1:20. The positive predictive value is a fraction e.g. 1/21.

Question 7

If the birth prevalence of OSB was 4 per 1000 instead of 2 per 1000. How would the OAPR be affected?

Answer 7

Higher (1:10, about 10%)

Question 8

Amniotic fluid AFP measured in screen-positive women has a detection rate of 98% and a false positive rate of 0.5%.
What is the new OAPR?

Answer 8

10:1

about 90%

Question 9

If the amniotic fluid AFP is positive, an AChE test is carried out on the same sample.
The amniotic fluid AChE in screen-positive women with a positive amniotic fluid AFP result has a detection rate of 99% and a false positive rate of 5%.
What is the new OAPR?

Answer 9

200:1

about 99%

Question 10

What is detection rate also called?

Answer 10

Sensitivity

Question 11

What is false positive rate also called?

Answer 11

1-specificity

Question 12

What is the difference between a cervical smear test result and a maternal serum AFP test result?

Answer 12

One is qualitative (result is +ve or -ve), the other is quantitative (result is a number – a cut-off level is needed to separate results into +ve or -ve)

Question 13

What is MoM?

Answer 13

Multiples of the Median

Question 14

If the median (say of MS-AFP) is 30 IU/ml, what is 0.5 MoM?

What about 2 MoM?

Answer 14

15 IU/ml

60 IU/ml

Question 15

Recognise the trade-off between higher detection rate but higher false-positive rate.

Answer 15

Again using the example of the MS-AFP test, the MoM curves for unaffected and OSB overlap. The test needs to set a cut-off level that has a high DR, but if it was 100% then the false positive rate would be very high. Therefore, you set it at a level that the DR is high enough but so you don’t have too many false positives.

Question 16

Recognise the trade-off between higher detection rate but higher false-positive rate.

Answer 16

Again using the example of the MS-AFP test, the MoM curves for unaffected and OSB overlap. The test needs to set a cut-off level that has a high DR, but if it was 100% then the false positive rate would be extremely high. Therefore, you set it at a level that the DR is high enough but so you don’t have too many false positives.
This can be seen on the “receiver operating characteristic curve”.

Question 17

How is the likelihood ratio calculated?

Answer 17

DR/FPR

Question 18

How do you calculate the odds of being affected given a positive test result?

Answer 18

Use the prevalence to create a flow diagram.
E.g. 1000 –> 2 affected
–> 998 unaffected
Then use DR and FPR to calculate how many positive results for each group. Then do the ratio between the two.

OR more simply, the likelihood ratio x prevalence as an odds

Question 19

Calculate the test specific risk of the being affected for an individual woman with an serum AFP of 2.5 MoM.

Answer 19

You calculate the likelihood ratio by looking at the height of the two curves of unaffected and OSB at 2.5 MoM, and dividing the height of OSB curve by the unaffected curve.
You then times the LR by prevalence as a ratio.

Question 20

If a woman is told that her risk of having an open spina bifida pregnancy is quite ‘normal’, that is, the risk is no higher and no lower than the risk for women in general, what is her MS-AFP level?

Answer 20

This means the LR is 1/1 so the risk is unchanged i.e. the normal prevalence (2:1000).
The LR is 1/1 when the height of the curves is the same (where they intersect).
The MoM is just under 2, so the level is just under 60 IU/ml.

Question 21

Define DR.

Answer 21

Proportion of affected individuals with positive results

OAPR =

Question 22

Define FPR.

Answer 22

Proportion of unaffected individuals with positive results

Question 23

Define DR.

Answer 23

Proportion of affected individuals with positive results

Question 24

Which of the following depend on the prevalence of the disorder in the screened population?
- DR, FPR, OAPR

Answer 24

OAPR

Question 25

What are the requirements for a worthwhile screening programme? (10)

Answer 25

1. Well defined disorder
2. Prevalence known
3. Medically important disorder for which there is an effective remedy available
4. Cost-effective
5. Facilities available or easily installed
6. Procedures following a positive result are generally agreed and acceptable both to the screening authorities and to the patients
7. Equal access to screening services
8. Test simple and safe
9. Distributions of test values in affected and unaffected individuals known, extent of overlap sufficiently small, and a suitable cut-off level defined
10. Avoid unnecessary complexity in the screening process

Question 26

What are the requirements for a worthwhile screening programme? (10)

Answer 26

1. Well defined disorder
2. Prevalence known
3. Medically important disorder for which there is an effective remedy available
4. Cost-effective
5. Facilities available or easily installed
6. Procedures following a positive result are generally agreed and acceptable both to the screening authorities and to the patients
7. Equal access to screening services
8. Test simple and safe
   * 9. Distributions of test values in affected and unaffected individuals known, extent of overlap sufficiently small, and a suitable cut-off level defined*
9. Avoid unnecessary complexity in the screening process