anp
Forum Newbie
Topics: 6
Posts: 46
|
A 55 yr old man visits his primary care physician with c/o urinary frequency.Examination shows 1-cm nodule in prostate. PSA ordered.By standards-PSA >4 is abnormal.This test has sensitivity-80% and specificity-90%.A cross sectional study showed -10% of men this age have prostate cancer.Patients PSA came-7ng/ml.What is ur best estimate of the likelihood that this man actually has prostate cancer
A-13%,B-25%,C-36%,D-47%,E-58%,F-69%,G-72%,H-81%
Please explain how u get the answer.Im weak i Biostats.
Thanks in advance
|
| dermatology
Forum Elite
Topics: 30
Posts: 293
|
alrite .... answer is 47 %
first write down wat alll is given in question ....
sensitivity =80%
specificity=90%
prevalance=10%.....( cross sectional studies give us prevalance)
now take imaginary number as 1000 as total number of ppl (N)
draw a table with disease and no disease colum and row as + and -
+ is test positive and - is test negative
and total ( N) = 1000
now with prevalnace find out number of ppl diseased .....
prevalance = all diseased / 1000
that comes out to be 10% = all diseased/1000 .........100(TP+FN)
so obviously no diseased number of ppl willl b 900 ( 1000-100)
900=FP+TN
now with sensitivity find out TP ....... TP/TP+FN
80% = TP/100 ............( TP+FN= 100 found out earlier)
so now we get TP= 80 so now start filling gaps in the table u made
TP=80
TP+FN=100
so , FN=20
same for specificity find out TN .....
90%=TN/FP+TN...................TN/900
TN=810.....
FP+TN=900
so, FP=90
now ur table is complete.......
question is What is ur best estimate of the likelihood that this man actually has prostate cancer ????
ie how much percent of ppl are actuallly diseased with positive test.....
take out percentage by number of diseased and test positive ppl *100/total number of ppl with test positive i.e 170
so 80*100/170.........47%
hope this helps .....
wats the right answer????
|
| anp
Forum Newbie
Topics: 6
Posts: 46
|
Thanks a lot- Dermatology, for the good explanation.Ur answer is right, The answer was given but no explanation given,so i dint know how they got it.
Thanks again
|
| dermatology
Forum Elite
Topics: 30
Posts: 293
|
no problem anp .... i know biostats is not my fav subject too.... kaplans have done real good job in clearing doubts and making subject more interesting and understanding ....
refer kaplans notes and videos by Steven Daugherty for epidemiology....he has done real good job explaining ......
best of luck 
Edited by dermatology on 09/25/07 - 04:34 PM
|
| kpmle2
Forum Elite
Topics: 33
Posts: 297
|
Thanks Dermatology for a very good explanation.
|
| starcraftbw
Forum Senior
Topics: 13
Posts: 166
|
dermatology..... you know your stats lol....
|
| gr8doc
Forum Senior
Topics: 8
Posts: 210
|
Good Job Dermat.....
|
|
| |
| | | | | | | |
