Effect of disease prevalence on positive predictive value (PPV)

00:01 Which of the following statements concerning positive predictive value (PPV) is correct? Answer choice (A) - positive predictive value is the proportion of tests that are true positives; ...if disease prevalence is low, then positive predictive value will be low.

00:20 Answer choice (B) - positive predictive value is the proportion of tests that are true positives; ...if disease prevalence is high, then positive predictive value will be low.

00:32 Answer choice (C) - positive predictive value is the proportion of tests that are true positives; ...disease prevalence has no effect on positive predictive value.

00:43 Answer choice (D) - positive predictive value is the proportion of tests that are false positives; ...if disease prevalence is low, then positive predictive value will be low.

00:54 Answer choice (E) - positive predictive value is the proportion of tests that are false positives; ...if disease prevalence is high, then positive predictive value will be low.

01:05 Now take a moment to answer this question for yourself.

01:13 Okay. Let's take a look at this question, dissect it and then walk through it together.

01:18 So this question falls under the category of epidemiology. We're doing biostatistics.

01:23 Now this question is actually a 2-step question.

01:27 At first glance you may think woah, this is a 1-step question, they're asking about positive predictive value.

01:33 Well, they're not simply asking about positive predictive value.

01:37 In fact, what they're really asking here is, 'what is the relationship between positive predictive value and prevalence?' So it's a 2-step question and of course the stem is required because we have to know what they're even asking us.

01:51 So let's walk through this question.

01:53 Now step one, we have to know the formula for positive predictive value.

01:59 So what is positive predictive value? It is the proportion of positive tests that are truly positive of some type of diagnostic test or other statistical measure.

02:10 Now that's the way we think about it, so it's the proportion of positive test that are truly positive, and hence the name - positive predictive value.

02:19 The people naming this are pretty smart.

02:21 Now what's the formula for positive predictive value? Now, it's the number of true positives divided by all of the positives, which are the true positives plus the false positives.

02:33 So you gotta know that equation.

02:36 Now I just tell you as a clinical pearl.

02:38 FOR USMLE step 1, step 2 and step 3 - you have to know these equations.

02:44 Biostatistics does not go away, it's always high-yield.

02:48 So memorize it now to make it easier for yourself for this current exam and for future USMLE exams.

02:55 Now, let's look at step 2 of this question.

02:58 We need to determine the influence of prevalence on positive predictive value.

03:04 Now prevalence is the proportion of a population found to be affected by a condition.

03:10 That is, the number of people with the condition divided by the total number at risk.

03:15 Now, let's look at our image here that'll actually help us understand the relationship between positive predictive value and how it changes with prevalence.

03:26 Now we can see on the y-axis we have predictive value, and on the x-axis we have the population prevalence.

03:33 Now if you look at that nice kind of light pink color, that's the line for positive predictive value.

03:40 And you can see, as the positive predictive value gets higher, it clearly is in a same relationship with population prevalence.

03:51 That is, as the positive predictive value increases, the prevalence also increases.

03:56 But that makes a lot of sense.

03:58 If there is more of a certain condition, say in a population, the predictive value of that is gonna go up.

04:05 For example, let's look around that 50% mark of population prevalence.

04:11 The predictive value seems to be a little bit above 80%.

04:15 So if half the population has a certain condition, we should have a pretty high predictive value.

04:20 And if we look at a hundred percent, for a 100% population prevalence, of course, we're gonna have a near-hundred percent predictive value; Everyone has it.

04:28 So that makes a lot of sense and if you look back to the answer choices, when they say disease prevalence being low and positive predictive value being low, that makes sense as well.

04:39 because if the prevalence is low, the chances of us predicting it's going to be low.

04:44 And if the prevalence of a condition is high, the chance of us predicting that prevalence as well is also going to be high.

04:52 So the graph here really does illustrate it well that positive predictive value will increase as the prevalence increases.

05:00 Reallly well demonstrated here in this image.

05:03 So let's go back to thinking about step 2.

05:05 So what we can see then is that when the prevalence is high, the number of true positives will increase and the positive predictive value will be high.

05:16 That's more of the equation way of thinking about what we're seeing in our graph And the same, in the opposite scenario, when the prevalence of something is very low, the number of true positives will decrease and the positive predictive value will be low.

05:30 Thus, we can see that the correct answer here is answer choice (A) - Positive pedictive value is the proportion of tests that are true positives; If disease prevalence is low, then positive predictive value will be low.

05:45 Now let's discuss some high yield facts concerning this question.

05:50 Now, positive predictive value and negative predictive value - topics we also need to discuss.

05:56 Now we said, positive predictive value is the proportion of true positives of some type of diagnostic tests or other statistical measure and it is really a reflection of how likely a positive test result is going to be a true positive finding.

06:13 Now, what's the equation for positive predictive value? You have to memorize this.

06:19 It's the true positives divided by all the positives which is true positives plus false positives.

06:26 Now what about negative predictive value? We talked about positive predictive value.

06:30 This is something you also have to know- it's extremely important.

06:33 Now negative predictive value is a proportion of negative tests that are truly negative and it is a reflection of how likely a negative test result is to be a true finding.

06:44 Now the beauty of negative predictive value and positive predictive value is that they're really quite the opposite.

06:51 Now the equation for negative predictive value is true negative divided by all your negatives which is true negative plus false negative.

06:59 Really if you look at negative predictive value and positive predictive value, they're just opposites.

07:05 one is all true positives, one is all true negatives - very easy to memorize.

07:10 Now let's talk about prevalence.

07:11 Now prevalence is a proportion of a population found to be affected by a condition at a specific time point or during a given time period.

07:21 Now prevalence is calculated by the total number of people with the condition divided by the total number of people at risk.

07:29 And this is generally expressed as some type of fraction or percentage.

07:34 Now, it's really really important that you don't confuse prevalence with incidence which is the number of NEW cases per population at risk given a time period.

07:45 So prevalence is, well 'what do we have during a time period' and incidence is 'what is the number of NEW cases' - so very different.

07:54 Now, both predictive values being positive predictive value or negative predictive value are influenced by prevalence.

08:03 Now, what we learned from this question is - as positive predictive value increases as the prevalence increases.

08:12 That means that there are more true positives, then you'll have a higher prevalence.

08:16 Now let's think about negative predictive value.

08:19 As negative predictive value decreases, prevalence will then increase.

08:24 So negative predictive value decreases as prevalence increases, and that's because there are fewer true negative tests with a higher prevalence.

08:34 Both really important high-yield concepts to understand Now, let's look at the rest of our chart here.

08:42 If you go back to the image, we see that we have negative predictive value shown in green.

08:47 And that highlights our second point we just discussed that negative predictive value decreases as prevalence increases.

08:53 You can see, at a zero percent prevalence, our negative predictive value is at a hundred percent.

08:59 And as we increase in prevalence, our negative predicitive value will decrease.