Screening SGL

Biostatistics Course 2021-2022 / Block 7

Ali Lateef Jasim

MBChB.
 Question 1.

The prevalence of undetected diabetes in a population

to be screened is approximately 1.5% and it is assumed
that 10000 persons will be screened. The screening test
will measure blood serum sugar content. A value of 180
mg or higher is considered positive. The sensitivity and
specificity associated with this screening are 22.9% and
99.8%, respectively.
 Question 1.

A- Set up a four fold table with appropriate numbers in

each cell of the table.
B- Calculate the following values:
1. The percentage of false positives.
2. The percentage of false negatives.
3. The predictive value of positive test.
4. The predictive value of negative test.
Total Blood Serum Sugar test

Negative Positive

C
A

V1
Yes
Diabetics

V2
No

H1

GT
H2
Total
 Question 1. Answer

I. To calculate the total number of diabetics among the

sample we should multiply the prevalence of DM by the
grand total (Sample size).

Total No. of affected = Prevalence X Grand total

= 0.015 X 10000
= 150
 Question 1. Answer

II. To calculate the number of disease and test positive

(diabetics and test positives) from the total affected we
use the equation of the sensitivity.
Sensitivity = Disease & test +ve (A) / Total affected
0.229 = A / 150
A = 150 X 0.229
= 34
 Question 1. Answer

III. To calculate the number of disease positive and test

negative (False negative) from the total affected we
subtract the no. of the affected and test -ve from the
total affected.
C = V1 – A
= 150 – 34
= 116
 Question 1. Answer

IV. To calculate the total number of the non-affected we

subtract the No. of affected from grand total
V2 = GT – V1
= 10000 – 150
= 9850
 Question 1. Answer

V. To calculate the number of the non-affected and test

negative (D) we use the equation of the specificity.
Specificity = Disease & test -ve (D) / Total not affected
0.998 = D / 9850
D = 9850 X 0.998
= 9830
 Question 1. Answer

VI. To calculate the number of disease negative and test

positive (False positive) from the total affected we
subtract the no. of the non-affected and test +ve from the
total affected.
B = V2 – D
= 9850 – 9830
= 20
 Question 1. A. Answer
Diabetics Total
Blood Serum Sugar test

Yes No
Negative Positive

34 20 54

116 9830 9946

Total

150 9850 10000

 Question 1. B. Answer
I. The percentage of false positives:

N who test positive (B)

Either, FP =
All without disease (V2)

OR FP = 1 – specificity

20
FP = = 0.2%
9850

OR FP = 1 – 0.998 = 0.002 = 0.2%

 Question 1. B. Answer
II. The percentage of false negatives:

N who test negative (C)

Either, FN =
All with the disease (V1)

OR FN = 1 – sensitivity

116
FN = = 77.3%
150

OR FN = 1 – 0.229 = 0.771 = 77.1%

 Question 1. B. Answer
III. The predictive value of positive test:

People with disease and test +ve

Either, PPV =
All people with a positive test

34
FP = = 0.629 = 62.9%
54
 Question 1. B. Answer
IV. The predictive value of negative test :

People without disease and test -ve

Either, NPV =
All people with a negative test

9830
FP = = 0.988 = 98.8%
9946
 Question 2.
To observe the effect of increasing sensitivity, assume a
blood sugar screening level 130 mg, with a sensitivity of
44% and specificity 99.0%. Calculate the following values
when the number of persons screening is 10000 and the
prevalence of undetected diabetes is 1.5%:
A. Calculate the percentage
I. False positive
II. False negative
III. Positive predictive value
IV. Negative predictive value
B. How many false positives and false negatives will occur if
100000 persons are screened?
 Questions 2.
Disease Total
Present Absent
Absent Present
Exposure

A B H1

C D H2
Total

V1 V2 GT
 Question 2. Answer
Using the same method as question 1 to create the 2x2
table:
1. Total No. of affected = Prevalence X Grand total
= 0.015 x 10000 = 150
2. Sensitivity = Disease & test +ve (A) / Total affected
0.44 = A / 150
A = 66
3. V1 = A + C < C = V1 – A < C = 150 – 66 = 84
 Question 2. Answer

4. GT = V1 + V2 < V2 = GT – V1
V2 = 10,000 – 150 = 9850
5. Specificity = Disease & test -ve (D) / Total not affected
0.99 = D / 9850
D = 9751
6. V2 = B + D < B = V2 – D < B = 9850 – 9751 = 99
 Questions 2. Answers. A:
Diabetes Total
Present Absent
+ve

66 99 165
Test

-ve

84 9751 9835
Total

150 9850 10000

 Question 1. A. Answer
I. The percentage of false positives:

N who test positive

Either, FP =
All without disease

OR FP = 1 – specificity

99
FP = = 1%
9850

OR FP = 1 – 0.99 = 0.01 = 1%
 Question 2. A. Answer
II. The percentage of false negatives:

N who test negative

Either, FN =
All with the disease

OR FN = 1 – sensitivity

84
FN = = 56%
150

OR FN = 1 – 0.44 = 0.56 = 56%

 Question 2. A. Answer
III. The predictive value of positive test:

People with disease and test +ve

Either, PPV =
All people with a positive test

66
FP = = 0.4 = 40%
165
 Question 2. A. Answer
IV. The predictive value of negative test :

People without disease and test -ve

Either, NPV =
All people with a negative test

9751
FP = = 0.991 = 99.1%
9835
 Question 2. B. Answer
B. How many false positives and false negatives will occur
if 100,000 persons are screened?
All affected = Prevalence x GT
= 0.015 x 100,000 = 1500
All not affected = GT – Affected = 100,000 – 1500
= 98,500
Cell A = 0.44 x 1500 = 660
Cell B = 1500 – 660 = 840
Cell D = 0.99 x 98,500 = 97,515
Cell C = 98,500 – 97,515 = 985
 Questions 2. Answers. B:
Diabetes Total
Present Absent
+ve

660 985 1645

Test

-ve

840 97515 98355

Total

1500 98500 100000

 Question 2. B. Answer

B. So,
1. False positives = 985
2. False negatives = 840
 Question 3.
A trial to screen Hydatid cyst showed:

Hydatid Total
Present Absent
Test Specific IgE

-ve +ve
level

100 29 129
30 121 151
Total

130 150 280

Calculate sensitivity, specificity and predictive values

Is this test of value in screening?
 Question 3. Answer

1. Sensitivity = 100 / 130 = 0.769 = 76.9%

2. Specificity = 121 / 150 = 0.806 = 80.6%
3. PPV = 100 / 129 = 0.775 = 77.5%
4. NPV = 121 / 151 = 0.801 = 80.1%

It is of no screening value.
 Question 4.
A trial to screen Hydatid cyst showed:

Hydatid Total
Present Absent
Test Specific IgE

-ve +ve
level

92 22 114
81 128 209
Total

173 150 323

Calculate sensitivity, specificity and predictive values

Is this test of value in screening?
 Question 4. Answer

1. Sensitivity = 92 / 173 = 0.531 = 53.1%

2. Specificity = 128 / 150 = 0.853 = 85.3%
3. PPV = 92 / 114 = 0.807 = 80.7%
4. NPV = 128 / 209 = 0.612 = 61.2%

It is of no screening value.
Thank You
Any Questions?
Ali Lateef Jasim
MBChB.