Analysis of Running Back Injuries: Single-Game Carries Are the Problem

Yesterday, I told you that I was going to introduce a study that has been a long time in the works. What I set out to do was try to measure the impact of specific rush totals, to test the hypothesis that higher rush totals lead to greater risk of serious injury in the immediate future.  It is not an easy thing to measure, and this study is by no means perfect, but I’ll try to explain why I did what I did. 

The first step was to settle on a group of running backs to use. I decided to use all running backs who have had at least one season where they had 1200 yards from scrimmage in a single season by age 27, going back to 1978.  There were 174 running backs in the sample, and you can see the full list here. The reason I put the age restriction is that I did not want to include older running backs, who might break down and get injured more frequently at the same workload level.  I then looked at every season of all 174 running backs up until they turned 28.

In this case, when I am trying to measure the impact of a 30 carry game vs. a 22 carry game, a problem arises because lots of backs have high week-to-week variability in their single game rushing attempts. Since I am testing a hypothesis that higher carry games carry greater risk, I settled on assigning a subsequent game (or missed game to injury) to the highest rush total game within 40 touches. I chose 40 touches because I wanted to limit it to the immediate aftermath of a high rush attempt game, and a starting back should generally get to 40 touches within about two games. Let me give you an example that might better illustrate how the games were broken down.

In 1996, Robert Smith of the Vikings played in 8 games. The rush attempts for those eight games were: 22, 30, 18, 26, 17, 27, 18, 4.  Game 2 was assigned to the 22-carry row (because it was the highest carry game prior to game 2).  Games 3 and 4 were assigned to the 30-carry row.  Games 5 and 6 were assigned to the 26-carry row.  Games 7 and 8, and the season ending knee injury on the 4th carry of Game 8, were assigned to the 27-carry row, since the injury occurred within 40 touches of that game (22 touches, to be exact). If you’ll notice, I did not assign any games to the 18-carry game in week 3, because it was sandwiched between the 30-carry game and the 26-carry game.

I used this method for all 174 running backs in the group, through age 27, and assigned all games within 40 touches of a game with 18 or more rush attempts for each running back.  As you will see, this created some fairly large sample sizes within each group.  I measured “injuries” (whether at least one game was missed) and also total consecutive missed games.

Let’s get to the chart.  For ease of viewing, I bunched rush attempt games into groups of three, and list for each the total number of games, total number of rushes and receptions, and total number of injuries occurring within 40 touches of that game type.  “GM” stands for Games Missed, and represents the total number of consecutive games missed for those injuries in each category (I capped it at 16 consecutive games).  The last three columns are the ones of most interest.  “Gm/Inj” stands for the average number of games missed per injury in that category.  “Touch/Inj” represents the total number of touches (rush attempts + receptions) per injury.  “Touch/Gm” is similar, but shows the total touches divided by the total missed games.

rush att games rush rec injuries gm gm/inj touch/inj touch/gm
18-20 1641 25441 4421 75 205 2.7 398.2 145.7
21-23 1700 29792 4481 68 186 2.7 504.0 184.3
24-26 1469 27260 3599 62 212 3.4 497.7 145.6
27-29 878 16979 2127 52 230 4.4 367.4 83.1
30-32 552 10793 1243 20 92 4.6 601.8 130.8
33-35 215 4214 484 8 46 5.8 587.3 102.1
36-45 158 3445 389 6 27 4.5 639.0 142.0

Here are some of the things that stand out to me:

1) The “workhorse” backs that were getting 30+ carries in a game had the highest touches to injury numbers, meaning they actually had fewer injuries per touch.  This might suggest that coaches generally have a sense of who is healthy enough to get higher rush attempts, and these backs are not missing as many games for minor bumps and bruises.  The backs in the 18-20 rush attempt group likely include a fair amount of guys who were already being limited by playing through injury, and some of them ultimately got shut down for a game or two to heal up.

2)  However, you may notice that the games missed per injury increases steadily (well, until the 36+ carry group that has a small sample size), so that when running backs with higher rush attempt games do get hurt soon thereafter, they are suffering more serious injuries that cause more healing time.  The backs in the 18-20 and 21-23 carry groups usually only missed a game or two when they missed time, and the average for both groups was 2.7 games missed, as they had relatively few season-ending injuries.

3)  When we combine the injury frequency with the cost, by looking at touches per game missed, the peak is at the running backs in the 21-23 rush attempt group, with the two groups on either side (18-20 attempts and 24-26 attempts) coming in next.  These three groups all rank ahead of the groups with 27 or more rush attempts, because of the greater increase in serious injuries for backs with higher workload games.

This suggests that high workload games are costly, because I am just measuring the number of consecutive games until the player initially returns.  A player who suffers a significant injury, usually a knee injury, may be far less effective after returning and see their career seriously altered, and the team may have a large investment in the player that is impacted negatively.

However, the “Curse of 370″ is no curse.  The players who comprise that group necessarily had high rush totals all year, including at the end of the season.  Some, like Jamal Anderson in 1999, did get hurt very early the next year. But Natrone Means never came anywhere close to 370, yet suffered multiple serious injuries after high rush attempt games, and was done by the time he turned 28. Cadillac Williams never came close to 370, but what could he have been if he didn’t start his career with an incredible 88 carries in the first three games of his career, and then suffer a costly foot injury in the fourth game. Ricky Williams took a pounding his rookie year, and missed four games right after a 30-carry outing, and then the next year missed six games with an injury right after a 27 carry game. Olandis Gary, never got to 300 carries. One year wonder, right?  He ended that season with a 29-carry game and tore his knee up in the offseason.  The examples are numerous, and these guys didn’t need 370 carries to end a career, all it took was maybe one game where they were pushed too far.

On the other hand, I think teams do need to limit their starters.  No need to do complete platoons, as we can see that backs have been able to handle a healthy diet of 20-23 carries without significant jump in serious injuries, but if I were a team, I would try to more strategically leverage my top back much like baseball teams use starting pitchers, and then be willing to spell him with other backs in late game and short yardage situations that may lead to greater contact and getting hit while fatigued.  I suspect that some teams have privately done research similar to this, probably more advanced, and I don’t know that the recent trend away from the workhorse back model of the late 1990′s and early part of the last decade is going to change.

blog comments powered by Disqus