eli5: Sensitivity vs Specificity in testing

143 views

Hey eli5,

I’ve been an RRT for eight years, and I feel like I’m strong clinically; but one thing I’ve struggled with is the concepts of Sensitivity and Specificity. I’ve gargled this time and time again and it never sinks in. I’d like to advance my education but it feels like if I can’t get this why try?

tl;dr Can someone help me cement the concepts of Sensitivity and Specificity by explaining it like I’m 5??

Thank you

In: 2

3 Answers

Anonymous 0 Comments

Sensitivity is “how likely are you to detect X if X is actually present” (few false negatives)

Specificity is “if the test detects X, how likely is it that X is actually present” (few false positives)

So high sensitivity low specificity would mean your test triggers almost guaranteed when the thing is present, but it can also trigger for many other things that aren’t what you’re looking for.

Vice versa a low sensitivity high specificity test would miss the thing you’re looking for often, but when it triggers you can be sure it’s really there.

Anonymous 0 Comments

Suppose you get a dog, to bark at burglars.

A sensitive dog – a dog with sensitivity – will bark at every burglar. A less sensitive dog might bark, or might just sleep through the burglary.

A dog with specificity will *only* bark at burglars. A less specific dog might bark at other things too.

With a given testing technology, there’s usually a tradeoff between sensitivity and specificity: for example, dogs are usually either sensitive but not specific (they bark at burglars, but also Amazon delivery people, cats, leaves, passing cyclists, family members…), or they are specific but not sensitive (they basically don’t bark at anything much. They *might* bark at strangers sneaking through a window at night, but there’s no guarantee)

Anonymous 0 Comments

To put it plainly, sensitivity is how likely a test is to spot the thing its testing for. So let’s just say you have covid-19. Sensitivity of a test is how likely that test is to correctly spot the causative virus. A highly Sensitive test will spot it most of the time, a test with low sensitivity won’t spot it very often and might say you’re covid free when you are actually infected with it.

Specificity is how likely it is to not get confused and spot the wrong thing. So again, if you have covid-19 a test with low specificity might come back and say you’ve got Covid-19 when you don’t, because you *are* carrying a similar virus. A test that is highly specific will only say you have the virus when you do actually have the virus.

So ideally we want a test that is specific and sensitive. As these tests will spot pathogen most of the time and won’t get confused by similar pathogens, so you can trust their accuracy.