COVID-19 Files: Ep. 5.5

What's up everyone? Here we are, again, we're on episode 5.5, I had to bring back my favorite second favorite co host, Dr. Akerman, because what we're going to talk about today is really interesting. It's the thing that everybody has been discussing in the news. It's how do we test people? Everyone needs to be tested. And then people go, Oh, well, there's these bad tests out there. Oh, with a sensitivity specificity. Nobody's explaining any of it. So as it turns out, Dr. Akerman is really smart at this stuff. And we've been working with a company because we want to introduce rapid testing here in North Texas. And Dr. Akerman has actually helped this company a ton in the type of tests that they're doing, how to explain it. And we're talking sensitivity, specificity, negative predictive value, positive predictive value stuff that I struggle with a ton, well as it turns out, the other guy on the side of the screen here is super smart at it. So Dr. Akerman, can you comment on that real quick?

Yeah. First of all, Make sure you don't burn any bridges with your first favorite co hosts.

Yeah, right.

And, and you're right. I mean, we've pretty much turn on the TV go on the internet, and we're talking about, you know, testing, what kind of testing, is it available, how much do we need? And then you kind of hear from the other side that there may be tons of tests, but too many, and they're not as good. And, you know, you sort of take that all at face value. But what does that mean that the test is no good, what makes it a good or a bad test? And I think that's where, you know, maybe we can do a better job of explaining it to people. And you're right, it's it's super confusing. So...

We did Episode 4.5, where we talked about different the mcas and the antibody testing and things. So we kind of hinted at it. But now we're getting to this point where we're saying, Yeah, but what does all that mean? Nobody's talking about it. You You and I got in a discussion John Oliver. You watched a show on that. 

Yeah. So you know, I don't want to sort of rehash I don't think we're going to do it justice. John Oliver with his extensive medical degree, did an excellent job last week explaining some of the differences between the tests and some of the pitfalls. But I think where we can add a little value to the discussion is to discuss why. Why is it that the test you might go get at your local, you know, minuteclinic, or doctor's office or anything, why that might not offer you the value that you're really looking for.

Yeah, and we can we talked before about the pros and cons of that. We don't know that. But what we're going to talk about today is why that's why getting a test may mean something different if you're in New York or if you're in Plano, Texas,

right? And you want to know the characteristics of the test the characteristics of the disease process, where you are, and what you're trying to get out of the test. And that will determine whether or not it number one makes sense for you to get tested, and how you can get a better result.

Yeah, so we've been seeing that the FDA is saying that you have to apply for this and then you have to achieve a certain. Let's start with the basics, a certain sensitivity and specificity, sensitivity, my friend.

So actually, what I want to do is I'm gonna I'm gonna share my screen with a couple slides here. And, you know, I want to I want to go over some important definitions. There's tons of stuff and you don't need a degree in statistics to figure this out. But it can be confusing. So I think if we sort of write it down, and might it might hit home a little bit better. So

I'm still impressed that you actually left New York I really because you seem like somebody who would be teaching at NYU or Mount Sinai or something like that. So, thank you for joining our group.

I wanted to have more time for podcasts.

There we go.

So a couple six definitions that I want to I want to go over here. And the first you alluded to sensitivity and specificity. The best way I would explain sensitivity is what percentage of patients that actually have the disease will test positive. Right? So the true positives coming out of the, of the test. And the specificity is the opposite. It's the percentage of the patients that don't have the disease and will test negative. So the patients who appropriately test negative for this disease process. Those are intrinsic qualities of the test itself. So you can apply the specificity and the sensitivity to any different population because the test characteristics don't change. What does change and this is the difference is the positive and the negative predictive values. So the positive predictive value, are the percentage of those positive tests that are accurately positive, and the negative predictive value is the percentage of negative tests that are accurately negative to say it another way, when I get a test, if it tells me I'm positive, what are the chances that it's right? That's what I really care about. I want to know If I think I have strep, and I go get a strep test, and it's positive, should I be taking antibiotics? And will I get better? That's what we're really asking. And it is different than the sensitivity and specificity, which I will clarify in a second. The other two really important definitions are the incidence of disease and the prevalence of the disease because these are two different things. The easiest way to differentiate is the incidence are the proportion of people who actively currently have the disease. The prevalence is how many people were affected, not necessarily currently infected. So when you talk about it in terms of the test that we're talking about the PCR test looking for active viral replication, that's looking at the incidence of disease, who's got it now. But the antibody testing, looks at who's got it or who had it. And that's more a test for the prevalence of the disease.

Well explained. Okay, yeah.

All right. So the classic way we look at this is using this, it's called a two by two table or a two by two plot. And you basically up on the, in the in the top bar here, that's the disease or the process that you're looking at whether you have positive or negatives. And then along the margin here, you're talking about the test characteristics, a positive or a negative test. So when you want to calculate the sensitivity, you're looking at this row right here. So the people who have true positives over the total number of positives, so 80, excuse me, if you've got 100% of people ten, let's say people who have a positive test and only eight of them actually have the disease. That means eight out of ten or 80% sensitivity for that test. When you look on the, the the right column here, that's where you're gonna figure out your specificity. That's your negative side. It's the total number of true negatives, people who actually don't have the disease and tested negative over all the negative tests that's going to tell you your specificity.

So I remember...

Sorry?

Go ahead. No, I was just gonna dumb it down to a level that this...I just remember that when I had somebody explained it to me that think of sensitivity and the specificity. If you were to look at these graphs, think of it like a car alarm, that if you make everything so sensitive, so that you could catch the person trying to steal your car. That means that a horn or a loud noise will set it off. If you make it super specific, that it only means that one method of a car being stolen. And so you have this issue of you have this trade off where you have the you want a car alarm that's super sensitive and will catch everything but you're going to get false positives all the time or do you want a specific test that only catches but every once in a while, they're gonna be able to steal your car a ...