Dynamic Chiropractic

Dynamic Chiropractic Facebook Twitter
Dynamic Chiropractic
Find
Advanced Search
Wellness Blog
Dynamic Chiropractic
Dynamic Chiropractic PracticeINSIGHTS
Current Graphic
Facebook

Cautions on SpIn and SnOut

Many of us have heard about SpIn and SnOut. The mnemonic helps to remind us that very Specific tests help to rule in a disorder (Sp = specific, In = rule in) and very Sensitive tests help to rule out a disorder (Sn = sensitive, Out = rule out). I usually don’t teach this memory aid because I find it leads to more confusion than clarity.

Now I am not saying there is no value to this approach but here is my take. There is value in SnOut particularly as it relates to Clinical Decision Rules and screening tests but there is very little value to SPIN.

Let me explain.

First of all, sensitivity is the proportion of those who have the disorder who have a positive test. Specificity is proportion of those without the disorder who test negativity.

It is important to remember that each of these relates to different populations. Sensitivity is only about those who have the disorder and Specificity is only about those who don’t.

Now let’s look at some numbers. Suppose we have a test that is very sensitive, say 98%, and not specific at all, say 5%. Now say we have a problem that occurs in 10% of the population. A positive test would identify most people who have the disorder but the test would also be positive in many more who do not. A positive test in this situation is virtually worthless.

If we test 1000 people and have a negative test that would mean we would only miss 2 individuals who actually had the problem. This would be useful in a screening test to see if additional testing was warranted. So SnOut has some value in screening tests and clinical decision rules.

Let’s look at a very specific test for a disorder with a similar prevalence of 10%. Suppose the test is 98% specific and only 5% sensitive. If the test is positive it does not necessarily rule in the disorder. Once again, if we test 1000 people, 5 individuals with the disorder have a positive test and 18 individuals without the disorder will have a positive test. SpIn does not work well in this situation and is even worse when the prevalence of the disorder is lower.

Whenever possible use likelihood ratios. Likelihood ratios will move you from a pre-test to post-test probability. You can use an old fashion nomogram but it is actually easier to use one of the many of the apps there are available. Yes, there is an app for that.

Leave a Reply