Understanding Injuries in the NFL Part 3

This is the third and final installment of a series exploring what actually correlates to injuries in the NFL, based on a series of articles by Zach Binney of Football Outsiders.

---

If you haven’t read them already, check out part 1 and part 2 of the series.

---

Although this is a look at NFL injuries in general, it may contain nuggets useful to understanding the Redskins injury situation. This installment will have several sections, each attempting to answer one of the following questions:

• Are certain players more prone to injury based on injury history?
• Does size play a role in injury and are smaller players more likely to be injured?
• How can we roll up this knowledge and really understand the importance of all the factors examined in this series?

As always, this will be a bit of a long article with lots of information, so there will be a TL;DR summary at the bottom for people who want the highlights.

Are certain players more prone to injury based on history?

The first thing we’ll look at is how well injuries in previous years predict injuries in the upcoming year for a given player. The goal here is not to develop an operational definition of “injury prone”, but just to establish that there is a clear link between injury history and future likelihood of injury, as described in the original study here. In the words of Mr. Binney:

To see if a prior injury history puts a player at greater future risk, we calculated the one-season risk of missing one or more regular season games due to injury for player-seasons with various two-year injury histories. All injuries occurring in the offseason, preseason, or regular season were counted towards a player’s two-year history. We used injuries from only the previous two years to ensure all players had the same amount of time to accumulate an injury history, and because injuries in the distant past may not be as indicative of future risk. We excluded a player’s first two NFL seasons because he would not have yet accumulated a two-year NFL injury history. We also excluded player-seasons where the player did not play in the team’s Week 1 game but was not on injured reserve for all 16 weeks, as these players may not have been at risk for a full season.

On average, 40 percent of players who played or would have played in Week 1 missed at least one game due to injury that season. Among those with no reported injuries in the prior two seasons, though, that figure is just 26 percent.

So overall, 40% of players will miss at least 1 game due to injury in a given season. But if a player did not have any reported injuries in the previous 2 seasons, his likelihood of missing a game in the upcoming season drops by almost half to 26%. That certainly indicates that perhaps some players are “iron men” who are less likely to get injured than the average player, and the Redskins own Ryan Kerrigan would certainly seem to be one of those players.

But we don’t just want to see if some players are iron men, but rather to see if there is a continuous relationship between number of injuries and likelihood to get injured. To do this, we can directly plot the fractional chance a player misses at least 1 game due to injury in the upcoming season against the number of injuries he was reported to have in the previous 2 seasons, shown below.

As shown in the figure, there is a clear increasing trend in a player’s likelihood to get injured based on the number of injuries they’ve had in the past. Even one reported injury increases the likelihood a player will get injured from 26% to 35%, and it is a continuous increase in likelihood after that, plateauing at around 60%. So there is a clear connection between number of past injuries and likelihood of missing a game in the future. However, it is worth noting the flipside of this, that even the most injured players have a 40% or so chance of playing all 16 games the very next year.

The results shown so far don’t differentiate between different types of injuries, even though some types of injuries are probably more serious than others, and persistent injuries in the same part of the body (eg, the Todd Gurley situation) might be more serious than multiple, unrelated injuries to different parts of the body (eg, Trent Williams over the last few years). So next we will plot the risk of missing at least 1 game vs the number of injuries of a specific type (knee, ankle, etc) reported in the last 2 years.

In general, we can see from this plot that multiple injuries in the same part of the body carry a steadily increasing risk of missing games, although the pattern is rather consistent with the previous plot of general injuries, just shifted up by about 10-15% (greater risk of future injury). The main exceptions are multiple injuries to the face/eye or upper extremity (UE) bone and joints (broken wrist, dislocated thumb, etc), which had a decreasing chance of missing games after multiple injuries. This is understandable as these injuries are most likely due to random chance and violent collisions, and thus less likely to be indicative of an underlying problem with the player. Also note that multiple Achilles injuries are more likely than others to result in missed time.

Overall, I would say my takeaways from this analysis are the following:

• A minority of players are “iron men” who are much less prone to injury than the typical player. In addition to that, there is a continuously increasing risk of future injury to players based on how many injuries they have had in the past. Thus, injury history is a rough predictor of future injury risk.
• For the most part, multiple injuries to a specific part of the body puts a player at a slightly (around 10-15%) increased injury risk compared to the same number of unrelated injuries. The exception is upper body bone and joint injuries, which are less predictive of future injury.
• The opposite is also true though, even players with many injuries over several years have a decent (around 40%) chance to play 16 games. The tradeoff of injury risk must be balanced with talent, as 12 games of a player like Jordan Reed or Rob Gronkowski might still be worth more than 16 games of a lesser player.

Are smaller players more injury-prone?

There are many ideas about what makes a player more injury prone, and many people point to size as a possible culprit. Namely, some people claim that smaller players are more likely to get injured playing football because their bodies are less suited to handle NFL collisions. This is at least a fairly straightforward question to study, as originally done here. However, we should separate this by position group, as we saw in Part 1 of this series that different position groups have different rates of injury, and more mobile positions (RB, WR, DB, LB) tend to have the highest rate of injury, while less mobile positions (OL, QB) tend to have the lowest rate of injury. The overall injury rate by position is summarized in the table below. Remember than an Athlete Exposure (AE) is defined as a game played or a practice session in which a player was listed as a full participant.

Clearly, if we just looked at injury rate vs weight, it would appear that smaller players are more likely to be injured, but that is only because the most injured position groups happen to have smaller, more mobile players. Having established this baseline, we can then plot the injury rate vs weight for each of these position groups.

For the most part, it actually looks like smaller players are less likely to be injured, not more likely. However, each graph has a slightly different pattern that is worth describing.

• DB is perhaps the easiest plot to interpret, and shows a clear increasing risk of injury based on size.
• DL shows a similar pattern, but the effect is not as large and it is overall a much flatter plot, indicating size is not strongly correlated to injury for DL. The up-down-up shape of the curve may be due to conflation of DEs and DTs.
• LB actually shows a slightly decreased risk of injury for larger LBs up until the heaviest weight category, which has significantly increased risk.
• OL shows a clearly increasing risk of injury based on size, except for the highest weight category, which has greatly decreased risk of injury (this may be the threshold at which OL can simply win with size).
• QB also shows a clearly increasing injury rate based on size.
• RB shows an increasing injury rate based on size up to a point, then a decreasing injury rate. This pattern may again be because two slightly different position groups (HBs and FBs) are being lumped in the same category together.
• TE also shows an increasing injury rate up to a point, then a decrease. It may be due to larger TEs being used primarily as blockers rather than pass catchers.
• WR shows a strong correlation between increased size and increased rate of injury, although the very smallest weight category of WR also had an increased rate of injury.

Overall, these data seem to indicate that smaller players are actually a bit less prone to injury within a given position group, although usage seems to play a much bigger role than size in determining injury rate. My big takeaway from this is not that teams should target smaller or larger players, but rather just that teams should not assume that smaller players are more prone to injury, as injury history is a much better predictor of future injury than size. Players like Russell Wilson and Christian McCaffrey are perfect examples of undersized players who were healthy in college and managed to stay healthy in the NFL despite being undersized.

Injury risk based on playstyle of defense

When discussing in Part 2 of this series the relative safety of various stadiums, LASkin brought up a great point that the playstyle of the home defense could also be a factor making some stadiums safer or more dangerous than others. In other words, are players more likely to get injured in certain stadiums because the field is worse, or because the home defense is more brutal? To further tease out the playstyle of the defense from the safety of the stadium, we can plot the risk of offensive injury to the opposing team when a defense is on the road, shown below. Since this averages all opposing offenses the team faces in different stadiums, it should only be a reflection of the playstyle of the defense. Also, this is using data from 2012 to 2016.

Well, what do you know? The Redskins have the most injurious defense in the NFL (or at least did from 2012 to 2016). There are 2 major takeaways here for me:

• The Redskins defense is certainly NOT the reason FedExField is so safe.
• Maybe we should look into whether or not we have some bad defensive practices (tackle technique, formations, I don’t know) that are leading to more injuries on offense. It’s a longshot, but worth looking into.

Rolling it all up with a statistical model

Injuries are a very complex phenomenon with many potential causal factors. In this series, we’ve looked at a lot of specific factors that are linked to increased chance of injury in the NFL. But how can we understand how important some of these factors are compared to others? Football Outsiders attempted a massive analysis that separated out the various factors into three levels:

• Player level: age, position, height, weight, years in the NFL, player name
• Team/Coach level: team name, head coach name
• Stadium: the name of the stadium a game was played

They ran a regression analysis to determine weights for each of these factors that reflected how correlated they were to injuries (more details available here). Rolling up all of these factors, they calculated a Median Odds Ratio (MOR) distribution that reflects the level of importance of each of these levels (Player, Team/HC, Stadium) in determining injury likelihood as shown below.

As shown in the plot, Player-level factors are by far the most important in determining injury likelihood, almost 3 times more important than Team/Coach-level factors and Stadium-level factors. That being said, they all matter. According to Football Outsiders:

On median, a higher-risk player (for example, Arian Foster) has nearly three times the odds of injury of a lower-risk player (for example, Frank Gore)... a player moving from a low-risk team/coach combination to a high-risk combination can expect, on median, a 21 percent increase in his odds of injury... on median, a player playing in a higher-risk stadium faces a 13 percent increase in his odds of getting hurt.

Summary

The TL;DR of this information to me is the following:

• Injury history is strongly correlated with future injury risk.
• Multiple injuries to the same part of the body carry a 10-15% increased risk of missing time due to injury in the following season unless they are upper extremity bone/joint injuries.
• Even players with an extensive injury history have a roughly 40% chance of playing 16 games.
• Smaller players are slightly less likely to get injured than larger players, although this effect is not nearly as important as their usage, position group, and injury history in determining injury risk.
• The Redskins defense (according to 2012 - 2016 data) is the most injurious defense in the NFL.
• Overall, player-level factors are far more important in determining injury risk than team/coach-level factors and stadium-level factors.

I think it’s important to reiterate here that this is primarily an attempt to understand injuries in the NFL in general, not a deep examination of the Redskins injury situation in particular. However, there are probably lessons we can learn from this to apply to our team going forward. What (if anything) do you think is the most relevant towards reversing our recent injury trend?