The National Electronic Injury Surveillance System is a database that collects emergency room records from a wide variety of US hospitals. More importantly, it makes this data freely available to researchers, who mine the data in order to draw conclusions intended to make our lives safer. Even more importantly, it also makes this data freely available to people like me, who are about as likely to use this data for its intended purpose as to use glue for its intended purpose, which is not very likely at all. The database is a goldmine of amusing and not-so-amusing snippets of life in the ER. You can read some choice excerpts elsewhere on this blog.
But beyond individual episodes of human drama, we can step back and look at the larger trends in this dataset. For example, about 56% of ER patients are men, which makes sense given that men tend to hold riskier jobs, engage in more dangerous activities, and tend to say, “Hey, watch this!” a lot more than women. If we look specifically at ER patients between 12 and 25, the percentage of men rises to 66%, which just shows you that younger men are especially stupid:
Interesting exceptions in this concerted effort by men to kill themselves off is in the area of scalding burns, where women outnumber men as patients, comprising 63% of cases over the age of 18:
If you are a woman, you might wonder what is responsible for this interesting reversal. You might also wonder what that burning smell from the kitchen is. Get back to that stove, you little floozy.
But you did not come here to read about some boring old data arrived at by rigorous statistical analysis. No, you came here for the same reason everyone else comes here — to find out, once and for all: which is responsible for more visits to the ER — ninjas or pirates? Thanks to the magic of data mining we have our answer: the word “ninja” features in 16 case reports, while the word “pirate” features in 17. Now, some people out there might start complaining about my methodology here, or talking about things like statistical significance, or getting out their nunchucks and heading to my house. But, in my defense, I have charts. Here you go:
Now, perhaps you are not satisfied with that chart. Perhaps you want data that can make a better soundbite on an evening news program. Well, here you go: what is more likely to send you to the emergency room, a gun or a puppy?
Being the smart reader that you are, you know that I wouldn’t be asking this question unless the answer was at least somewhat unintuitive, and you’re right. The answer is a gun-wielding puppy with a cute little Rambo bandana on its forehead and ammo belts slung over its shoulders. There’s one behind you right now. But, no, seriously, as to the original question, it really is a puppy.
Over the next few posts, I hope to probe deeper into the mysteries of the ER database in order to bring you even more wisdom, which may potentially even save your life someday. That is, unless you run into a puppy first.
Methodology note: 7 years worth of NEISS data (from January 1, 2002 through December 31, 2008) was used in this report. All statistics obtained through SQL and grep. The command to filter gunshot wound data was
grep -E 'GUN\s*SHOT|FIRE\s*ARM|\bGSW\b' | grep -v NAIL | grep -v PELLET | grep -v STAPLE | grep -v -E "\bB[.-]?\s*B\b"