Epidemiology: Sensitivity vs. Specificity question

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.
Status
Not open for further replies.

JoMo23

New Member
5+ Year Member
Joined
Mar 24, 2018
Messages
3
Reaction score
0
I have a question in my epidemiology class that I've been going round and round about but need a little more explanation before I feel confident that I understand the concept.

The question is, in terms of being screened (blood pressure, BG, cholesterol), which is more important to the patient and physician (respectively) sensitivity or specificity?

Right now I think specificity to the patient and sensitivity to the physician, but I'm not sure I'm correct. Both are jumbled in my mind.

Thank you!

Members don't see this ad.
 
I'll try to clear it up. What helped me back in the day to get those 2 straight is to think of them in this way.
Sensitivity is true positive rate. Sensitivity = true pos/(true pos + false neg). This means that if you get a negative result in a test with high sensitivity, it is most likely a true negative (because for sensitivity to be high, false negative results should be very rare, check the equation), which makes this great for screening.
Specificity is true negative rate. Specificity = true neg/(true neg + false pos). This means that if you get a positive result in a test with high specificity, then it is most likely a true positive (because for specificity to be high, false positive results should be rare, again, the equation), which makes these great for establishing a diagnosis or confirming a positive screening test.
Does this help at all?
 
I'll try to clear it up. What helped me back in the day to get those 2 straight is to think of them in this way.
Sensitivity is true positive rate. Sensitivity = true pos/(true pos + false neg). This means that if you get a negative result in a test with high sensitivity, it is most likely a true negative (because for sensitivity to be high, false negative results should be very rare, check the equation), which makes this great for screening.
Specificity is true negative rate. Specificity = true neg/(true neg + false pos). This means that if you get a positive result in a test with high specificity, then it is most likely a true positive (because for specificity to be high, false positive results should be rare, again, the equation), which makes these great for establishing a diagnosis or confirming a positive screening test.
Does this help at all?

That does help clear it up. So based off of what your explanation is, patients would be more concerned with sensitivity and physicians would be more concerned with specificity? If that is correct then I think i understand the concepts.
 
Members don't see this ad :)
It kinda depends on what the patient wants. If somebody wants to make sure they DO NOT have the disease, they would want a test with high sensitivity (First Aid makes you remember this with a mnemonic: SNOUT (Sensitivity) rules OUT, SPIN (specificity) rules IN). If the patient wants to make sure that they actually, truly have the diagnosis, they would want a test with high specificity.

Same goes for the physician, the answer to your question depends on whether you wanna make/confirm the diagnosis (high specificity), or rule something out (sensitivity).
 
It kinda depends on what the patient wants. If somebody wants to make sure they DO NOT have the disease, they would want a test with high sensitivity (First Aid makes you remember this with a mnemonic: SNOUT (Sensitivity) rules OUT, SPIN (specificity) rules IN). If the patient wants to make sure that they actually, truly have the diagnosis, they would want a test with high specificity.

Same goes for the physician, the answer to your question depends on whether you wanna make/confirm the diagnosis (high specificity), or rule something out (sensitivity).

Okay thank you that makes more sense. The mnemonic is handy too! Thanks.
 
You really shouldn't see these as "which one does the provider/patient want" because both will want some degree of sensitivity and specificity. Having a 100% sensitive test is no use for a physician since that usually means that the specificity will be pretty close to zero. If you think about it, what that means is that for a screening test (say for cancer), you've set the bar so high that only those who very obviously have cancer (and likely at a later, more progressed stage) will test positive on the test. You want to catch people early with a screening test so setting the threshold that high isn't good.

You also don't want to set the specificity too high because that means the sensitivity will decrease. Again, walk through the logic in your head. If your sensitivity is high, that means that you're very sure that if the test is negative, that means the patient doesn't have the disease. In order to do that, you have to set the threshold for the test to be very low. In that case, you get a lot of false positives and your sensitivity is super low. A lot of false positives makes a test pretty useless (you can do things like doing multiple tests to rescue it, for example).

So patients and providers will both want some degree of sensitivity and specificity to the test (and their interests don't necessarily diverge).
 
Status
Not open for further replies.
Top