November 2015

Sun Mon Tue Wed Thu Fri Sat
1 2 3 4 5 6 7
8 9 10 11 12 13 14
15 16 17 18 19 20 21
22 23 24 25 26 27 28
29 30          

« Lifesaving Devices Can Cause Havoc at Life’s End (RT @nytimes) | Main | Allergic Contact Dermatitis - Case Report »

June 28, 2010


michel g soete

The concepts of SnOut and SpIn have been softly criticized by some authors (Pewnsner et al., BMJ, 2004; Hegedus et al., j Man Ther. 2009). One of their critics is that for instance, both specificity and sensitivity have to be considered together in assessing the validity of a test.
This critic is so deeply true that it even proofs that SnNouts and SpIns do not exist except as mighty dragons in a fairy-tale.
Indeed a test with a specificity of 99% and a sensitivity of 1% does not allow to rule a diagnosis in and the same must be said for all test for wich the sum of the sensitivity and the specificity equals 100%. Such tests do contribute not at all to a diagnosis. Thus even if their specificity should be extremely high it cannot be tests ruling a diagnosis in given a positive test result.

michel g soete

Yesterday I wrote some comment. Perhaps it should be preceeded by following:
'The article 'SpIn and SnOut' is very well and clearly written and represents by far the most widely spread and common opinion but....'

What I mean is perhaps best illustrated by an example

disease present disease absent
positive 5 5
negative 95 95

The test is highly specific (95%) but the pre-test probability was 50% and the post-test probability given a positive result is 50%. Thus a high specificity is no guaranty that a test is a SpIn.

I am not very happy with it but it is a fact.

The comments to this entry are closed.

Enter your email address:

Delivered by FeedBurner

TFMR Communique

Blog powered by Typepad