As with sensitivity and specificity the LR+ and LR- should always be
presented together to describe the test's ability to rule in (LR+/Spin)
or rule out (LR-/Snout) disease. One (sensitivity or LR+ ) alone is not
informative without the other. Only when the purpose of the test is
already known, one can concentrate on one seperately.
Regards,
Jeroen Lijmer
-----Original Message-----
From: Toby Lipman 7, Collingwood Terrace, Jesmond, Newcastle upon Tyne.
Tel 0191-2811060 (home), 0191-2869178 (surgery)
[mailto:[log in to unmask]]
Sent: maandag 20 september 1999 19:32
To: Atle Klovning
Cc: [log in to unmask]
Subject: Re: Likelihood ratios (and a CAUTION)
In message <[log in to unmask]>, Atle Klovning
<[log in to unmask]> writes
>The holy LR is a vast step forwars in understanding tests- BUT when
>combining sens and spec into ONE measure- we lose information. E.g. at
>times having a test with high specificity is very important, whereas at
>other times high sensitivity is the thing you want.
>
>Also, and quite disheartening:
>
>A sens 40% and spec 98% => LR+ =20 and
>B sens 80% and spec 96% => LR+ =20 and
>C sens 100% and spec 95% => LR+ = 20
>
>A, B and C should generate very different test interpretations
>
>
And that is why I have stopped using LRs except as a shorthand for the
powerof the test. If you want to understand the implications of a test
(especially before doing it), then I think it's best to sketch out the
2x2 table and work out predictive values, false/true +ve/-ve rates etc
from that using the sensitivity and specificity and best guesses of the
prior probability. It only takes a few minutes unless you have lost your
calculator! The other reason to know sens and spec is to recognise
SpPIns and SnNOuts (in Atle's example A is a SpPIn and C is a SnNOut
despite the highish specificity)
You can be misled by the general rule that LR+ under 10 shows that a
test is not powerful - if sens=0.99, spec=0.50 then LR+ is around 2 and
you might dismiss it as not much of a test - but it's a SnNOut - (mind
LR- is 0.02)
Cheers
Toby
--
Toby Lipman
General practitioner, Newcastle upon Tyne
Northern and Yorkshire research training fellow
Tel 0191-2811060 (home), 0191-2437000 (surgery)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|