Prep for USMLEPrep for USMLE Forum
   Forum    Step 1  Step 2 CK Step 2 CS Step 3  Match  IMGs Resources Search






Previous Topic | Next Topic  kapQ 2x2 table 




 



Author10 Posts
  #1

The blood supply used by a local hospital for transfusions during surgical procedures is discovered to be infected with hepatitis C. To protect the blood supply from future contamination, all new donors are required to be screened for hepatitis C. The test used for this screening has a sensitivity of 95%, and a specificity of 90%.If this test is used on a sample of donors in which 10% are known to have hepatitis C, what is the chance that a donor who test negative is actually free from the disease?

A. about 99%
B. about 95%
C. about 90%
D. about 85%
E. about 50%
F. about 45%



___________________
The secret is never give up

  #2

..........hep+....hep-
test+...95.......10......105
test-......5.......90......95
..........100......100

pv- = 90/95
B

that 10% having hep c I guess is useless

Edited by equinoxe on 09/10/06 - 12:41 PM

  #3

here he is asking about the negative predictive value(NPV)
1-sensitivity is NPV
so that is 1-0.95=0.05
that is less than 1
so answer is A about 99%

___________________
Time is God!

  #4

This question asks NPV

........disease...............no disease
--------------------------------------------
+.........9.5...................9

_.........0.5...................81

NPV= 81/81.5 = 99%
Answer is A

___________________
The Key to Succeed is Patience.

  #5

robbin - thank you, still puzzled by your calculation. Where do I go wrong if I do

........disease...............no disease
--------------------------------------------
+.........95...................9

_...........5...................81

PV- = 81/86 = 94.18 % -> B ?

have you made use of the information "10% are known to have hepatitis C"?
===

suv, sorry I am also confused by your approach, could you explain it please?

1-sensitivity = false negative rate

1 - TP/(TP+FN) = (TP + FN - TP) / (TP + FN) = FN / (TP + FN)

whereas PV- = TN / (TN + FN)



Edited by equinoxe on 09/10/06 - 12:57 PM

___________________
waiting gives the devil time

  #6

robin082006 wrote:
This question asks NPV

........disease...............no disease
--------------------------------------------
+.........9.5...................9

_.........0.5...................81

NPV= 81/81.5 = 99%
Answer is A





I, 100% agree.

Let us go for detail.

Prevalence is 10%

It means if we assume total population is 100 ,then
  • No. of diseased presons = 10
  • No of healthy persons = 90

Since sensitivity is 95%
  • True posititive = 95% of 10 = 9.5
  • False Negative = 10 - 9.5 = 0.5

Since specificity is 90%
  • True negative = 90% of 90 = 81
  • False positive = 90-81 = 9

NPV = True negative /all negatives = 81/81+ 0.5 =81/81.5= 0.99

Best of luck


___________________
FA is just a good revision book.It is not a "real" learning tool.

  #7

equixode,

this is a easy question about biostatistics.

You should use the prevalence to calculate NPV.

See my post here.

http://www.prep4usmle.com/forum/comments.php?id=3...


___________________
The Key to Succeed is Patience.

  #8

thank you leopard and thank you robin for excelent explication, I really did not understand until now this type of problems. Now I do. Thank you!

___________________
waiting gives the devil time

  #9

equinoxe wrote:
thank you leopard and thank you robin for excelent explication, I really did not understand until now this type of problems. Now I do. Thank you!



You are welcome

Actually the real problem in such questions is "filling of 2x2 table".Once you fill it then you can easily calculate
  1. PPV
  2. NPV
  3. Accuracy

Best of luck


___________________
FA is just a good revision book.It is not a "real" learning tool.

  #10

thank u sooooo much for ur explanations !!!!!!!!


___________________
Time is God!







You don't have permission to post.




Login or Register to post messages in this topic





















Contact | Leaders | Disclaimer | Privacy

Copyright @ Prep for USMLE. All rights reserved.