Quick and Dirty Statistical Analysis of Purdue Injury Frequency

Alright. This has gotten out of hand.

With Rob Henry's latest donation on the altar of ACL's, Purdue athletic injuries have now reached epidemic proportions.

The CDC should be called in to investigate the possibility of a collagen-destroying virus/bacteria/prion running rampant across West Lafayette. Home Hospital should be re-opened exclusively as a lower-extremity orthopedic surgery unit to deal with the catastrophic ER overflow choking the local health care system. Local authorities should form a special committee to investigate the possibility of nefarious villains striking out at the knees of vulnerable youth sports figures.

At the very least, there should be a mandatory questionnaire for all incoming athletes investigating history of knee injuries, of fluoroquinolone use, and of any demonic pacts they may have made to improve their physical prowess.

These remedies may sound extreme, but the time has come to take drastic measures. Chance alone can hardly account for such a terrible rash of catastrophic injuries.

If you don't believe me, then perhaps you will believe the POWER OF SCIENCE!

(Disclaimer: The severely statistically challenged may not want to make the jump).Wait, what am I worried about? This is a Purdue blog.

After doing a little preliminary research, I stumbled upon a very interesting data set published by the NCAA Injury Surveillance System. (Apparently some folks at the NCAA have been doing more than aggressively pursuing secondary cream cheese violations and turning a blind eye to serious institutional corruption) This research group has been tracking the incidence of injury to NCAA athletes in Div I-III in most, if not all, major collegiate sports for the past several decades.

If you check out the papers that Randall Dick et. al. have published, there is a wealth of information available about the rates, types, mechanisms, and severity of injuries incurred by athletes during both practice and competition. If that is the type of thing that gets you interested, by all means check out the primary publications.

If that sounds terribly boring, don't worry. I have pulled some of the the most relevant data from these papers that presents an interesting background upon which to project Purdue's recent injury woes. Some of this data has been simplified for purposes of brevity, so don't slam me on prefect accuracy if you go check out the original figures.

ACL Injuries:

In the U.S., somewhere between 100,000 and 200,000 people tear their ACL each year (about 1 in 3500). When it comes to athletes, Football players sustain the highest amount of ACL tears (53% of total). Internal derangement of the knee accounts for 17.8% of all football injuries.

The actual rates of knee injury (ACL tears account of the vast majority of these) for Football are:

6.17 / 1000 Athlete-exposures during games

So if Purdue Pete played in 1000 football games, he would get 6 knee injuries, or if our unstoppable army of 1000 Pete's played in 1 game, about 6 would get a knee injury.

The stats say that an average team would send 51 players in to 11 games (561 exposures) for an average rate of about 3.5 knee injuries/year from games.

Of these injuries, 27% are expected to be serious (out for >10 days), for an expected total of about 1 serious knee injury/year, with a standard deviation of about 0.3.

Purdue has had 4 serious knee injuries over the past year that I can recall (Bolden, Marve, Siller, Henry), putting us well outside of normal. The statistics are slightly less bleak if we account for the rate of knee injuries during practice, but still put us more than 3 standard deviations outside the expected norm (2.2 +/- 0.5 for Purdue as I figure it with my available data - and that would be all full contact practices, all players, not just starters) which translates into less than a 0.13% chance of such a streak of injuries occurring.

For basketball, the knee injury rate is:

0.66 / 1000 athlete-exposures

Which is a rate of  0.26 knee injuries /basketball season. You can guess where 2 lands us.

All Injuries:

The rate for football is 36 / 1000 A-E in games, so the predicted value is 20 injuries/year, with 5.5 +/- 0.5 being serious, as defined above.

According to T-Mill's post "Purdue in 2010: The Year of the Injury", we had 13 serious injuries that cost us at least one game. Even if we assume that a few of those improved between days 7-10 that still puts us an astronomically large 9 - 15 standard deviations above normal for Football.

For basketball, the rate is 9.9/ 1000 A-E in games, so we are allotted about 4 injuries/basketball season, which is exactly what we got (Robbie Hummel, Lewis Jackson, Sandi Marcius, John Hart). So that worked out, which makes sense. If you think about it, the season went fairly normally aside from Hummel's re-torn ACL. And since he was already injured, that obviously increases the chance of re-injury, and makes him significantly different from a theoretically "normal" athlete.

Conclusions:

Yep, we have a ridiculous number of injuries on the football team. We have a problem across the board where we tear ACL's all the time. Chance alone has a difficult time accounting for this. But you knew that already.

What you didn't know before was that there was at least one guy at the NCAA who wasn't busy ignoring the fact that college athletics is mostly a wretched hive of Nevin Shapiro's scum and villainy.

-------------

So, there are some problems with this analysis, like the issue of previous injury as discussed above. The granularity of data is also an issue (no such thing as half of an ACL tear, etc.). Lots of other stuff too, but I know that, you know that. Let's not get bent out of shape about it. Also, if you like this kind of stuff, check out this older post.