SDK Forum Fanatic
Topics: 537 Posts: 1,587
| | 04/10/05 - 01:49 AM  
 
   
 
|   #1 |
A pts has new screening test for TB..test indicate that pts is well while pts is infected..... 1)To identify all pts infected with TB ,the cut-off point for this test shud be set at the point of highest?? a)senstivity b)specificty c)+ve predictive value d)-ve PV e)Accuracy 2)If this new screening test has sensitivity of 90% & speficity of 70%..The +ve predictive value of this test is?? a)12.5% b)25% c)30% d)72% e)90% Plz explain the answers in detail???
___________________ Love humans,save humans & live like humans.
|
| mjl1717 Forum Hero

Topics: 955 Posts: 5,451
| | 04/10/05 - 11:33 AM  
 
   
 
|   #2 |
answer-a sensitivity [But it would have low specificity/the false + results would be unacceptably high.]{Just to add more fuel to the fire-there would be few Type I errors and a large number of Type II errors. q2 -Ill return with the second answer, using the punnet -like square! SDK-{You been asking good thought provoking, calculating qs. As you already know no pain no gain} Also congratulations on your 10,000th post 
___________________ Smell the coffee! "Is That an Osler move??"
|
| mjl1717 Forum Hero

Topics: 955 Posts: 5,451
| | 04/10/05 - 12:26 PM  
 
   
 
|   #3 |
Using 1000 people: TP=900 FP=300 FN=100 TN=700 PPV=TP/TP+FP=(900/1200) =about 75% the closesest # is 72%
___________________ Smell the coffee! "Is That an Osler move??"
|
| SDK Forum Fanatic
Topics: 537 Posts: 1,587
| | 04/10/05 - 01:01 PM  
 
   
 
|   #4 |
Thnx mjl1717... But the answer is given as 30/120=25%. BRS page no: 254. How this occur I donot know..... Any one has idea about this??? you give a correct equation....90/120....75%...
___________________ Love humans,save humans & live like humans.
|
| mjl1717 Forum Hero

Topics: 955 Posts: 5,451
| | 04/10/05 - 01:34 PM  
 
   
 
|   #5 |
Just have to say: This type q is what its about. And it seems like there is so MUCH room for error. Hopefully a statistician will respond. If any thing my error MAY be in setting up that punnet-like square. I try to put the disease on the x axis and test results on the y axis. Also I made each column( not row) equal 1000 people using your sensitivity/ specificity results. Also if TP is 90% or 900 the numerator for PPV must be 90% or 900.{As far as I know the the numerators are the same in sensitivity and PPV.} TP is the # who test positive and have the disease.
___________________ Smell the coffee! "Is That an Osler move??"
|
|
| |
| | | | | |