TEALE FENNING

Xerxes I · Posted: Wed Sep 03, 2008 7:48 am Post subject: Statistics

Chose the single best option that describes the test or statistical term:

A Sensitivity
B Specificity
C False positive
D False negative
E Negative predictive value
F Positive predictive value

Can anyone remember the other options? there were a few more but I don't think true positive was one of them.

What statistical term are these scenarios describing?

Question 1:
Out of 10,000 pregnanct women, 250 fell under the high risk group using a down syndrome screening technique. Of these, (did it actually say of these? I think they implied it somehow) 5 were carrying a down syndrome baby.

Question 1:
Of a population of 1000, 200 had endometrial thickness of 10 mm or more. of these, 100 had endometrial cancer.

salihabduallah · Century Club Joined: 04 Sep 2007 Posts: 120

May be A&B , what do you think?

Xerxes I · Posted: Wed Sep 03, 2008 10:16 am Post subject:

I think PPV for both. these are almost identical descriptions aren't they?

nibbs07 · Joined: 03 Dec 2007 Posts: 9 Location: Hampshire

yeah I think PPV as well..
_________________
Nibbs

drmmanish@yahoo.co.in · Joined: 05 Mar 2008 Posts: 42 Location: UK

with whatever nick taught us regarding statistics I thought both were similar descriptions and PPV for both.
_________________
manish

Maud · Joined: 11 Oct 2007 Posts: 73 Location: Bristol

I said PPV for both as well.

rpwalavalkar · Teale Fenning Administrator Joined: 20 Jul 2006 Posts: 966

i would have gone for PPV too. well done there. a different way of asking stats, haven't seen stats feature in this guise before.

r
_________________
Dr Miss. Raj Walavalkar MBBS MRCOG
TealeFenning Administrator
SR O&G Wessex Region

Nick Raine-Fenning · Posted: Sat Sep 06, 2008 8:35 am Post subject:

Yes you are right. PPV is the only correct answer.

You have the proportion of those testing positive and being identified as high risk (this will give us sensitivity and, if we knew how many had been missed, specificity. However we then have how many were truly affected and this is the PPV - those affected in the presence of a positive test.

We do not know how many were truly affected though or how many normals were falsely identified so cannot calculate the sensiitvity, specificity, or NPV.

Just to remind you of the equations and our tips to remember them:

sensitivity = TP/ TP + FN ('spurious' or false negatives)

specificity = TN/ TN + FP ('spurious' or false positives)

positive predictive value = TP/ TP + FP (all the 'positives')

negative predictive value = TN/ TN + FN (all the 'negatives')

where,

F = false
T = true
P = positive
N = negative
_________________
"Teale Fenning Medical Education" delivering evidence-based, exam-orientated learning since 1997

Nick Raine-Fenning · Posted: Sat Sep 06, 2008 8:37 am Post subject:

If you consider my post above and the equations you can see we can only calculate the PPV as we do not know any 'false' figures.

salihabduallah · Century Club Joined: 04 Sep 2007 Posts: 120

But from Q1, we can know that 245 [ 250 - 5 ] were false positive ?

or I am wrong ? Confused

Nick Raine-Fenning · Posted: Sun Sep 07, 2008 9:00 am Post subject:

I'm afraid you are.

We do not know how many 'affected' babies there were in the low risk group. That's the whole point and problem of screening. We onloy only how many there were in the test positive group hence we can calculate PPV.