Roch confirms no MRI for Gonzalez prior to signing.

[square634]

The way I look at it is, the O's paid Gonzalez $12 million, plus they lost a 2nd round pick, so call it a total investment of $15 million. If an MRI costs $1,000, then if there's better than a 1 in 15,000 chance that the MRI will spot an existing problem, you need to have one done.

Again, that's a simplified analysis. Let's say there is a 1 in 15,000 chance that the MRI will correctly spot an existing problem. You still need to know what the chances are that the MRI will erroneously spot something that is not a problem. And what is the relative cost of losing out on a free agent due to that erroneous information compared to the benefit of avoiding a pitcher with a higher injury risk?

[Nigel Tufnel]

Again, that's a simplified analysis. Let's say there is a 1 in 15,000 chance that the MRI will correctly spot an existing problem. You still need to know what the chances are that the MRI will erroneously spot something that is not a problem. And what is the relative cost of losing out on a free agent due to that erroneous information compared to the benefit of avoiding a pitcher with a higher injury risk?

But to have a better idea of which MRI results are typical and which are season/career threatening, don't you need to gather more information? And how do you gather more information? By giving MRI's to as many pitchers as possible.

[square634]

But to have a better idea of which MRI results are typical and which are season/career threatening, don't you need to gather more information? And how do you gather more information? By giving MRI's to as many pitchers as possible.

Fair enough. I certainly think it would be a great idea for the Orioles to do a prospective study by giving every free agent signing an MRI after signing and an MRI after each completed season -- but of course, to really make it an airtight study, the results should be blinded from the organization so that no action can be taken based on the MRI results alone.* In which case, the lack of an MRI would have had no immediate relevance to signing Gonzalez.

*I am of course assuming that the relevant data don't already exist and that such a study has never been done.

[El Gordo]

FWIW MRI costs can vary as much $400-$3600+ depending on type and location in the US. The average cost of a shoulder MRI in Baltimore is $1600. http://www.comparemricost.com/600.

[Sports Guy]

According to Roch his MRI showed that there was fraying and and two small tears. Now, the question is how normal is that? If you gave Cliff Lee an MRI right now what would his shoulder look like?

If the MRI shows those things, but all other information points to him being fine then what does the MRI really tell you?

I am not sure it is normal or not..I would guess it is. They say that most pitchers have small tears in their rotator cuffs and things like that.

But let's look at this from 2 angles:

Angle 1:

His MRI shows what most pitcher's MRI would show...At that point, do you or do you not believe that doctors and others who are reading the MRI would be able to tell you if it is normal or not? Do you trust those doctors to have an informed opinion?

Angel 2:

The MRI shows something not normal that would cause you to not sign the pitcher...Would that have been a good or bad thing?

[Sports Guy]

Again, that's a simplified analysis. Let's say there is a 1 in 15,000 chance that the MRI will correctly spot an existing problem. You still need to know what the chances are that the MRI will erroneously spot something that is not a problem. And what is the relative cost of losing out on a free agent due to that erroneous information compared to the benefit of avoiding a pitcher with a higher injury risk?

Do you or do you not think the orioles team doctors will be able to figure this out?

[El Gordo]

The question is, do pitchers regularly agree to MRI's and if they don't should that automatically rule them out as an O FA choice?

[sangar]

Can you show me something that 100% guarantees that Gonzo's injury wouldn't have popped up in an MRI?

This is it right here. People keep saying 'It may have not shown anything' and 'all pitchers will show something'. What I've been wondering throughout this thread is how does anyone know it wouldn't have shown a blatantly obvious injury? Maybe it would have shown everything.

SG is right. To not give this guy an MRI prior to signing him was foolish.

[square634]

Do you or do you not think the orioles team doctors will be able to figure this out?

Let me make sure I understand what you are asking. Say the Orioles doctors do an MRI scan on a potential free agent signing, and they find something abnormal. Are you asking whether the doctor would then be able to figure out whether the pitcher will have a serious injury or whether the abnormality will not have any detrimental effects over the life of the pitcher's contract? Because that's the entire point: no doctor can do that without a crystal ball. Furthermore, no doctor can even give you a risk assessment unless studies have been done to tell you the relative injury risk of an abnormal MRI.

[Sports Guy]

Let me make sure I understand what you are asking. Say the Orioles doctors do an MRI scan on a potential free agent signing, and they find something abnormal. Are you asking whether the doctor would then be able to figure out whether the pitcher will have a serious injury or whether the abnormality will not have any detrimental effects over the life of the pitcher's contract? Because that's the entire point: no doctor can do that without a crystal ball. Furthermore, no doctor can even give you a relative risk assessment unless studies have been done to tell you the relative injury risk of an abnormal MRI.

Not really..

let's say Gonzo was given an MRI and it shows "normal" tears in his elbow, shoulder or wherever. Your argument seems to be that you don't know for sure...My question is this...If the MRI is a "normal" MRI for a pitcher, do you think that a doctor would know that? Simple yes or no is all that is needed.

[sangar]

Let me make sure I understand what you are asking. Say the Orioles doctors do an MRI scan on a potential free agent signing, and they find something abnormal. Are you asking whether the doctor would then be able to figure out whether the pitcher will have a serious injury or whether the abnormality will not have any detrimental effects over the life of the pitcher's contract? Because that's the entire point: no doctor can do that without a crystal ball. Furthermore, no doctor can even give you a relative risk assessment unless studies have been done to tell you the relative injury risk of an abnormal MRI.

So basically you're saying an MRI is useless?

[square634]

So basically you're saying an MRI is useless?

I am saying that without context, blindingly scanning things can be useless or even detrimental. At a very simple level, this involves Bayesian analysis of conditional probabilities of a) the probability that a pitcher's arm is a ticking injury time bomb, b) the probability that a pitcher in category a. will be correctly identified by an MRI scan, and c) the probability that a pitcher who is not in category a. who will be erroneously red-flagged by an MRI scan.

If you are running a test on a pitcher who has arm pain or other telling symptoms, then the conditional probability that the pitcher is in category a. likely skyrockets (in other words, the probability that a pitcher who currently has arm pain is an injury risk is likely much higher than the probability that a pitcher with no arm pain is an injury risk). If the a priori probability of being an injury risk increases, then the potential benefit of an MRI increases, even though the false positive rate remains the same.

If you want, I can give you an actual numerical example of how this works with prostate cancer screening data.

[square634]

Well, you didn't ask, but I'm going to post this excerpt from my paper anyway.

As a hypothetical example, in men under age 40, prostate cancer is rare, occurring in about 1.5 percent of patients. The probability that a man who actually has prostate cancer will be positively identified by the PSA is about 90 percent, and the probability that a man without prostate cancer will have a false positive PSA test is about 7 percent. In other words, in a random sample of 1,000 30-year-old men, about 15 would be expected to have prostate cancer, of which about 13 would be detected by PSA screening. Of the 985 men without prostate cancer, however, about 69 would have a positive PSA test. In this example, there is about a 5.3:1 ratio of false positives to correct diagnoses, so only about 16 percent of men with positive tests truly have prostate cancer.

What is important with regards to preemptive MRI screening is what those specific probabilities are. It is within the realm of possibility that the ratio could be much worse than 5.3:1. In other words, you might have to reject 10 or more free agents on the basis of an MRI in order to avoid one free agent who is actually an injury risk. In order to decide what ratio is acceptable, you have to assign a cost to forgoing a free agent who would not have had an injury and a benefit to avoiding a free agent who would have had an injury.

[El Gordo]

I am saying that without context, blindingly scanning things can be useless or even detrimental. At a very simple level, this involves Bayesian analysis of conditional probabilities of a) the probability that a pitcher's arm is a ticking injury time bomb, b) the probability that a pitcher in category a. will be correctly identified by an MRI scan, and c) the probability that a pitcher who is not in category a. who will be erroneously red-flagged by an MRI scan.

If you are running a test on a pitcher who has arm pain or other telling symptoms, then the conditional probability that the pitcher is in category a. likely skyrockets (in other words, the probability that a pitcher who currently has arm pain is an injury risk is likely much higher than the probability that a pitcher with no arm pain is an injury risk). If the a priori probability of being an injury risk increases, then the potential benefit of an MRI increases, even though the false positive rate remains the same.

If you want, I can give you an actual numerical example of how this works with prostate cancer screening data.

I was recenly given a bone scan which lit up in the area of my ankle. If it were not known that I had an extreme arthritic condition in my ankle, that would have been a strong indication for bone cancer. Context is very important.

[Sports Guy]

Well, you didn't ask, but I'm going to post this excerpt from my paper anyway.

What is important with regards to preemptive MRI screening is what those specific probabilities are. It is within the realm of possibility that the ratio could be much worse than 5.3:1. On other words, you might have to reject 10 or more free agents on the basis of an MRI in order to avoid one free agent who is actually an injury risk. In order to decide what ratio is acceptable, you have to assign a cost to forgoing a free agent who would not have had an injury and a benefit to avoiding a free agent who would have had an injury.

And this is all assuming that the MRI looks like it would with most pitchers...What if they find something major by doing the MRI?