– The K Zone –

The Trouble with Tommy John, by Ian Joffe

December 24th, 2018

Up through 2015, baseball was noticing a troubling trend: Tommy John surgeries – in the major leagues, minor leagues, and even among youth – were on the rise. In more recent years, the number of torn UCL’s has started to fall back, at least among professionals, but the concern is still ever-present, especially given the 12-16 month recovery time and far-from-perfect success rate. The rise in Tommy Johns has led a lot of doctors and baseball analysts to chime in with theories on why so many more players are needing the surgery. In this article, I wanted to test a few of the leading theories on which risk factors are significant in increasing the odds of needing Tommy John.

I’ll get to my own research later, but I wanted to start with a few theories that have already been tested by others. The first have to do with pitch selection, and these theories are, to say the least, contradictory. Some hypothesize that an increase in fastballs thrown has led to the spike in Tommy Johns, but at the same time others argue that breaking ball usage ultimately does pitchers in. I had the hardest time finding research to back up the curveball theory. An entry to the American Sports Medicine Institute’s journal found that there is no correlation between throwing curves and needing Tommy John. In terms of the fastball theory, one study from the Journal of Shoulder and Elbow Surgery argues that there is a correlation between fastball usage and torn UCL risk. However, a later study (I couldn’t find the original link) from the American Sports Medicine Institute says that there is no correlation between pitch selection and Tommy John surgery. There is a potential lead here, but it’s not conclusive. High fastball selection may or may not be a Tommy John risk.

One theory that seems to have more widespread backup is that higher velocity can risk Tommy John surgery. This article, by the American Journal of Sports Medicine, suggests that higher velocity may very well lead to higher risk of elbow injury. Another piece, also from the AJSM, makes the same case, and goes as far as to say that pitch velocity is the most predictive element of Tommy Surgery, but it still limits r^2 to 0.07. Specifically, it suggests that peak pitch velocity, as opposed to mean velocity, is a risk factor. These findings are corroborated by this Fangraphs community research article, which details exactly how the data was found.

Based on all of that, it seems that while pitch selection is not a fully proven theory, there is evidence that high velocity leads to heightened Tommy John risk. That begs the question, “What can be done about it?” The obvious answer is “throw less hard,” but it’s very unlikely that pitchers will be willing to sacrifice an essential part of their game to reduce health risks. That especially goes for younger pitchers who are being judged for their tools rather than a career’s worth of stats. In today’s game, when draft signing bonuses are so large, and initial free agent contracts are even more massive, it is borderline unreasonable to ask a young pitcher to risk all their value to improve their health. Additionally, the players most at risk are high-velocity pitchers, and high-velocity pitchers are the ones who depend most on their speed (when combined with other tools), and are therefore least likely to be able to make a change without taking a potential hit to their value. In my research, I wanted to look at changes that I thought could be made in pitchers without really hurting their value.

One proposed theory for the increase in Tommy Johns is sports specialization. This theory is not only a logical causation, but is heavily respected in orthopedic circles, and seems scientifically sound. Unlike the other theories, there are pages after pages on Google that champion this one, but here are the first three. As the theory goes, high school baseball players, especially those looking for scholarships, are always looking to gain a competitive edge. So, a few decide to do baseball year-round, in order to get better. Then, to catch up, others had to do the same thing. Soon enough, every serious baseball player was practicing baseball all year in high school. I’ve seen a few different colloquial explanations as to why this is bad – “the UCL needs rest;” “an arm only has to many bullets;” “one needs to strengthen different muscles” – and I’m not sure which is the closest to the real scientific explanation, but either way the negative aspects of specialization seems like a widely accepted theory among doctors and casual baseball fans alike.

To test the theory, I grabbed a data set of pitchers who had Tommy John surgery between 2015 and 2018. I then built a Python webscraper to sort through MaxPreps data, which keeps tracks of all high school athletes and their statistics. The program searched for the player on MaxPreps, and then checked how many sports he had played in high school. Unfortunately, I was only able to get data on about a third of the players who had Tommy John surgery during the given period. Some players had unusual last name configurations (I’m talking to you, Jose de Leon), others did not go to high school in the United States, and some went to high school before MaxPreps was founded in 2002 and later popularized. The biggest issue, though, was that several high school players had the same name as those who I was looking for. I was able to further filter my search using state, but if two people had the same name and played high school baseball in the same state, which happens more often than one would imagine, I had to remove them from my data set. In total, I was left with a sample size of 28, which while small, is still reasonable enough to mean something.

Out of those 28 MLB players who had a torn UCL, 7 played multiple sports and 21 only played baseball. That’s a 25% multiple-sport rate. In a control sample of random baseball players who I could specify on the MaxPreps database, 155 out of 596 played multiple sports in high school, or 26.0%. Based on this, it is safe to conclude that I found no evidence that sports specialization is a Tommy John risk (my chi-square derived P-value was a hardy 0.904). To be clear, I was dealing with a limited sample. My research also says nothing about the very real risk of needing Tommy John surgery while in high school. But, based on that, I see little reason to believe that playing multiple sports in high school leads players to have significantly better odds of staying healthy in the majors.

The most common method of preventing injury in MLB is the pitch count. Every team practices it, and pays special care to number 100. According to common knowledge, high pitch counts risk injury, and managers will take pitchers out when the count gets high because of it. That’s not to say that injury risk is the only reason pitchers are removed in the late game; batters get better multiple times through the order, pitchers get worse as they fatigue, and relievers are often just better than starters. But, injury risk is usually part of the equation, and almost every manager would probably say that high pitch counts do risk injury. So, pitch count is the second factor that I set out to test.

For this test, I gathered data on starters alone, because they are more similar to each other in use (at least for now). Like last time, I was not able to look at every starting pitcher in my data set, so I once again ended up with a very small sample, only 13 starters. So, for the last time, I want to reiterate that because of that, my research is more of a starting point on the subject than an end. Anyways, the first look I took at pitch count had to do with the game of the injury. Here were the results:

The points seem to be scattered rather randomly across the number line. The chunking of data points is a little odd, but I would expect the gaps to fill as n increased. The overall lesson here, though, is that pitch count does not seem to contribute to torn UCL risk. A pitcher is about as likely to tear their ligament on the 40th pitch as on the 90th. In fact, one might note that there are zero data points past the 100th pitch. In a larger sample, there may have been a few, but the point is clear: removing pitchers before or around the 100th pitch does absolutely nothing to decrease injury risk within that game. Tommy John risk does not increase as the game goes on, and players should not be pulled early simply to avoid getting hurt, because that does not work.

While pitch count has no influence on injury odds within a game, it is possible that high pitch counts have a hangover effect, making a pitcher more likely to get hurt in their next start. So, looking at the same lucky 13 pitchers, I charter their pitch count from the start before the one in which they got hurt:

The average among these pitchers was 85.3, 7 pitches below the league average for a start. Even when those lower two outliers are removed, the mean only goes up to 93.5, a pitch and a half above the league average. From this data, it does not appear that lowering pitches in the previous start leads to lowered injury risk overall, for pitchers who got hurt threw about the same number of pitches in their pre-injury start as the average pitcher who will not get hurt. So, that’s one more reason that pitch count should be ignored as a factor for injury risk.

Pitch count on a start-by-start basis appears to be a complete non-factor in the Tommy John question. Still, though, I wanted to give pitch count one last chance and take a look at seasonal trends. Perhaps each individual start is insignificant, but if a player throws too many pitches in one season, they become a higher risk for the surgery. So, to test this, I found the amount of pitches thrown in the season of the injury for Tommy John recipients.

Think of the blue histogram as an extended version of the league average dot, showing how many pitches every pitcher has thrown per season from 2015-2018. The singular red dots exist only on the x-axis, and show the seasonal pitch count of injured pitchers. For the first half of the graph, the density of the red dots seem to match the density shown by the blue histogram. But then, there are no dots at all in the second half on the chart. This shows that seasonal pitch count has no effect on injury risk. Injured pitchers did not throw more pitches than other pitchers. In fact, injured pitchers are completely left out of the upper range of the graph, a range that many healthy pitchers got to. This is to say that there were no pitchers from 2015 to 2018 that got Tommy John surgery because they threw an abnormally high amount of pitches in the year of their surgery.

Single-year trends showed no evidence that season pitch count had an effect on Tommy John risk. The very last step of my study was to examine multi-year trends. First, I took a look at changes in pitches from year to year. Often, teams will say that normally they would not cap a pitcher, but because he only threw so many pitches in the previous season, that number could only increase by a certain amount next season. For example, if a pitcher throws 1500 pitches in 2017, the manager may conclude they can only throw 2000 pitches in 2018. To test the theory of pitch increase, I charted the percent change in pitches per season for injured starters. Unfortunately, I did not have access to minors pitch data, so I had to remove rookies from the set, once again shrinking the sample.

Like all graphs before, the pattern on this one shows how small the effect of pitch count is on injury risk. Most pitchers who needed Tommy John threw far fewer pitches in the season of the injury compared to the season before, not far more. Only one pitcher experienced a severe workload increase, and only two had small workload increases. Torn UCL’s had no tendency to occur more often to pitchers with heavy workload increases.

Since two-year trends seemed to have no effect on Tommy John risk, the next multi-year trend to turn to is a player’s career span. I don’t have the pitch count data for the players’ careers, and even if I did it wouldn’t mean much because it wouldn’t account for the bullets on their arm in high school, college, and perhaps most significantly MLB practice. However, I do have player ages. In the career-pitches “the arm only has so many bullets” theory, it is suggested that as players tear their UCL’s after a certain number of career pitches, thus as a pitcher’s total career pitches increases, their odds of hurting themselves increase too. Age increases at the same rate as total pitches, so it would follow that as age increases, likelihood of surgery also would increase. This is not the trend that I found in the real player ages. The average MLB player age in the seasons that I studied was 28.9. Rather tragically, the average of a Tommy John pitcher was a year younger, 27.9. This discredits the theory that injury risk increase is directly proportion, or even somewhat proportional, to career pitches thrown, as players who were younger actually turned out to have the higher injury risk in the data.

From all my research and all the research of my peers, we are left with few clues about the causes of torn UCL’s. The only useful piece of information was the study about fastball velocity, but even that just barely had any predictive power. This inability to find causes does not signify a weakness in modern research, but rather a weakness in the traditional views of health. It’s easy to look at injuries as directly caused events. Just like how I could stub my toe because I jammed it against the wall, a pitcher tears his UCL as a direct effect of more complex causes. Instead, health should be looked at as a skill, like, for example, batting. Everyone is born with some degree of batting skill, whether it be very high or very low. People can improve on that natural ability through techniques like diet and practice, and then bring their total skill level to the plate. But, once at the plate, their chance of getting a hit is rather random. A .333 hitter, for example, gets a hit one in three at bats, and it is more or less random which of those three at bats he got a hit in. Similarly, players are born with good or bad health skills, which they can work on improving through techniques like proper stretching. But, once they bring that health skill to the table, which may be represented by a probability of injury, the likelihood that an injury occurs within that probability is more or less random. It’s impossible to know if an injury is more likely on the 10th or 100th pitch of a game or season because the pitch at which the injury occurs should be viewed as randomized. This skill-based view of health is more accurate to the data, and scientifically assumes the null hypothesis. If teams thought like that, it may lead them to more successful pitcher use.

 

If you liked this article, follow us on Twitter and be the first to know when more original research, articles, and interviews come out on the site.

Image Attributed to Sports Illustrated

8 Replies to “The Trouble with Tommy John”

  1. I think you need to look at the pitches per year of injury data from a different perspective. A pitchers total pitch count during their year of injury is in most circumstances going to be lower than their peers as they are not pitching a full year due to the injury (unless their injury occurs at the very end of the season). It may make more sense to look at pitch count/game averages prior to injury compared to pitch count/game average for their peers.

    Like

  2. My theory is that there is an imbalance between muscle and tendon strength. Tendons thicken slowly over time as exposed to repetitive, demanding work, such as swinging an axe or sledgehammer. Muscles on the other hand respond very quickly to exercises that isolate and stress that particular muscle group. We’re left with 300 pound muscles joined to 200 pound tendons.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: