© As are all of our blog posts, this story is protected by copyright;
Popular Woodworking Magazine, 2011.
When the debate over table saw safety gets heated, numbers are brought out to bolster one side or the other. This happened the other day when USA Today published the headline: “CPSC wants to stop daily table saw amputations,” followed by a mention in the second paragraph of 10 amputations a day. That sounds pretty serious, maybe somebody should do something about it. Or maybe we should
look at where these numbers come from, how they are used, and how they fit in the big picture of modern life.
The source of these statistics is the National Electronic Injury Surveillance System. NEISS is a database used by the Consumer Product
Safety Commission to track injuries caused by products ranging from abrasive cleaners to zip lines. For any safety issue, we as individuals and as a society need to do a cost/benefit analysis. As much as we would like to keep anyone from ever getting hurt, the value of using anything needs to be compared to the risks involved and the costs of eliminating those risks, and if that isn’t possible the cost of minimizing them. As individuals, we can eliminate risks entirely by choosing not to use something. Beyond the personal level, decisions become far more complicated and difficult.
There are so many things that can hurt us in so many ways that even assessing the risks is an expensive and daunting task. Rather than record every instance of every injury to every individual in the United States, the NEISS uses a small sample of cases from about 100 emergency rooms to project statistics to the population at large. Doctors in these ERs make notes and enter product, injury, body part and diagnostic codes and the results are weighted and projected. It works in a similar way to exit polls during elections; a small sample that’s easy to manage and count is examined and applied to the entire population. It’s a useful tool, but it isn’t perfect and the numbers that come out at the far end are predictions, not facts.
Let’s look at the table saw amputation numbers for 2009, the latest year with available statistics. If you want to follow along at home, the NEISS database is online, and the product code for table saws is 841. The number of samples that mention table saws is 783, and the database suggests that this reflects a total of 35,624 injuries. If the phrase “table saw” appears anywhere in the doctors notes or the product code, the accident is counted as a “table saw accident”. There aren’t many of them, but the numbers include instances where somebody fell off a ladder and hit his head on a table saw, or someone hurt their back moving a table saw.
The same effect occurs if the word “amputation” appears. In 2009 for example, 117 reported cases were projected to arrive at an estimate of 4211 finger amputations. A look at the actual notes reveals that four cases weren’t table saws after all, and six were near amputations. That’s only ten instances, but ten out of 117 is 8.5 percent. The projected numbers are weighted, so you can’t simply multiply, but you can safely say that the total number is overstated.
When the word amputation appears, it’s often followed by a phrase such as “life-altering” that was quoted in the USA Today story. I don’t want to minimize anyone’s pain or suffering, but analyzing injuries and possible solutions is by nature a heartless, dollars and cents exercise. Mangle your hand at work, and your disability claim will be based on how many parts of how many fingers are gone. Losing a thumb, means losing the ability to grip and that loss will make many things more difficult for the rest of your life. That is indeed life-changing.
Of the table saw amputation cases from 2009, about half involved the loss of one or more complete digits. The largest segment of that portion, 37 percent of the total amputations, were one finger or one thumb. Of the remainder, the listings were almost evenly divided between partial amputations, or the loss of a fingertip. In the cases listed as amputations, 23 per cent involved the loss of the tip of one digit. Losing the tip of a pinky will hurt for a while, but unless you’re a musician, it won’t make much difference in your overall abilities or the quality of your life.
With an estimating system such as the NEISS, there is a confidence number assigned that reflects the probable accuracy of the projected numbers. If you were to count every case that occurred in a given year, you could be quite sure of the numbers. The smaller the sample, the less accurate the results, so the confidence number reflects a greater possible variation. The database will reach a point where the number of actual reports is too small to give a reliable projection. When the sample is that small, the NEISS database won’t make an estimate of overall cases. In cases involving commercial woodworking operations, saw operators under 18 years of age or schools, there aren’t enough samples in the database to project a total number of accidents.
This doesn’t mean that cabinetmakers on the job or high school kids in shop class don’t get injured. But it does mean that the number is quite small, compared to the total number of injuries. And where does the number of table saw accidents, particularly amputations of fingers fit in the big picture of the population of table saw users, and in society as a whole? There isn’t a reliable count of table saws in use in the United States. Estimates range between 6 million and 10 million. If we pick from the middle and use 8 million, the estimates suggest that in any particular year, about one table saw in 229 will be in an accident that sends someone to the emergency room, and that one in about 2,076 saws will result in an accident with an amputation as the result. The number of saws in accidents that result in the complete loss of one or more digits is about half that, or one in 4,152.
In the context of the general population of 310 million Americans, about one in 9,000 will go to the emergency room in any given year after tangling with their table saw, one in 80,000 will have a medical report that lists the word “amputation” and one in 160,000 will lose one or more fingers or a thumb. If you, or someone close to you are that “one” you’ll look at this differently than everyone else. For those charged with writing rules and regulations, the task will be to decide if this is a problem that happens often enough for action.
While reviewing the numbers, I wondered about how the figures on table saws compared to other items. In 2009, there was an estimated total of about 22,000 finger amputations due to accidents of all kinds in the United States. There is no doubt that table saws are a significant portion of that number. Once again, the database sample is too small to provide estimates for many other items, but there was one item that caused about the same number of finger amputations as table saws, and about 10 times the number of total accidents. In addition, this product caused enough finger amputations among children under the age of 18 to generate an
estimate of total occurrences, about 45 percent of the total. Five children a day are enduring finger amputations due to this hazard. The
name of this product? The door.
Table saw injuries are a serious matter, and many if not most, could be prevented. The Consumer Product Safety Commission is considering rules that might help in reducing the number and the severity of these injuries. These rules would likely change the number and types of choices we have, and they would almost certainly raise the price of the tool that is central in most of our shops. There are two big questions here: Are table saw injuries prevalent enough to require government action and are there other solutions?
This post is part of a series of articles that will look at this issue in depth. I’ll do my best to put my opinions, emotions and agendas on the shelf so as a group, woodworkers will have better information about the frequency, causes and possible solutions to table saw injuries. In upcoming posts I’ll be reporting on the real costs of these injuries, the history of this issue from all sides, and most important what woodworkers tend to do to injure themselves and how they can prevent those injuries. Stay tuned, and sign up for the RSS feed to receive these articles as they appear. Your comments are welcome, or you can send me email.
Click Here to Read a Previous Post About Table Saw Accidents
Click Here to Read a Previous Post About Our Survey of Table Saw Safety
Click Here to Read Free Articles About Working Safely With Your Table Saw
Here are some supplies and tools we find essential in our everyday work around the shop. We may receive a commission from sales referred by our links; however, we have carefully selected these products for their usefulness and quality.
Mark my words: some day, CPSC, OSHA, or some other organization is going to want blade guards on kitchen knives!
Bob Miller… I think one important element of Bob Lang’s comment relates to the GIGO effect… the provberbial Garbage In => Garbage Out homily from the computer world.
He rightly pointed out that the estimates were, at least in part and probably more than "in part" based on flawed data input right at the source… the Emergency Rooms of the hospitals!
When a doctor or whoever clicks on or writes in "table saw" when the perp was a band saw, the resulting data point is useless in a field of data on table saws!
Unfortunately, it seems that people are more interested in a battle of "my degree makes me more knowledgeable [hence more credible???] than your degree or experience makes you…"
CPSC should clean up its own act by working with the medical profession and ER teams to create better descriptions and categories for the healthcare people to use BEFORE extrapolating from available "data." Or add caveats like, "the source data probably is 60% representative of the reality we’re trying to estimate, so take it all with a grain or more of salt…"
And who am I and what’s my qualification for saying this? I’m "Justin Engineer." 🙂
Just one more quick point: if you want to see what an effective power tool guard looks like, head over to Home Depot and look at the radial arm saw they use for cutting construction lumber.
In my Home Depot, there is a box-like guard made of laminate covered MDF that completely encloses the blade and is wide enough that you really could not get your hand under the guard if you tried. I am sure a clever monkey could defeat it or remove it, but it would take considerable effort to have a traumatic amputation with that guard in place.
I found this very interesting. I am an engineer (who has been trained in undergraduate level statistics) and I did not find Bob Lang’s analysis all that bad–he simply asked questions about the sources and quality of the data collected. His extrapolations are likely not quite right, but his arguments did not seem to be that statistical analysis is bad or wrong, just that when we simplify the data too much it is easy to draw wrong conclusions.
I would prefer to see the data presented with an explicit confidence interval clearly stated in English, such as something like: we can be 90% confident that the number of hospital visits for injuries involving a table saw (in any way) was between xx,xxx and yy,yyy.
The database itself is fascinating. At least one of the sample "table saw" amputations clearly states in the notes that the accident occurred on a band saw, for instance. There are no flesh detecting technologies for bandsaws that I am aware of.
I was also stuck by the age of the patients in the samples that the database showed me. Patients aged 82 and 77 and 76 etc. which makes me think that familiarity and comfort with a tool can be a double-edged sword. Note to self: Don’t get complacent about safety. This also points away from the litigant in the original table saw lawsuit who I believe had zero hours of table saw usage before his accident.
As a country, I think we could have much better bang-for-our-buck by requiring more effective guards, better training, and more safety oriented product materials (as opposed to the current safety warnings that come with power tools which are designed to prevent lawsuits rather than amputations). I would put flesh detecting technology low of my list bang-for-buck mitigation steps.
Bob (we are just bob, bob, bobbing along here)
Thanks for taking a second look. I admit that statistics are beyond my field of expertise, and I accept that the people who can develop something like NEISS do good things. In trying to understand the numbers on the surface, I did some digging into where they came from. While I don’t know how the methodology works, I do know that the limited amount of data on the front end needs to be reliable or the results on the back end aren’t useful.
If you query the NEISS database, you can download the samples and read the ER doctor’s notes for the samples. Many of these are cryptic and unclear, or show that the data going in doesn’t say what is being said about the numbers coming out. If we’re trying to decide if this is a big enough issue for government action, I don’t think we should be counting a fall that lands on a table saw as a table saw accident. And if there aren’t enough accidents in high school shop classes to even register in the projections, I think it’s irresponsible for anyone to say we’re doing this to protect our children.
So far, most of what has been printed and discussed on this subject has been far from factual. I don’t have formal training in journalism either, but I think we need to take a close look at a complicated subject in order to make a decision that does the most good for the most people. Stay tuned, as there is more to come.
Firstly does it feel weird addressing things to other Bob’s, it always throws me off. I may have misread what exactly you were questioning in my haste. If you were looking at how USA Today pick and chose its data points to "prove" a point then I have absolutely no problem with what you did or your questioning of it. The way I had read it, and I admit that I am touchy about it, you were questioning the ability to extrapolate data using statistical methodology. In this case from a 100 hospital sample to national numbers. I think questioning the core of a major field of mathematics without very good reason (more than "it seems wrong") takes the dialog away from where it should be which is "are we looking at the right numbers", which in my second reading is what you were actually writing about, and "what if anything should we be doing about it".
If nobody believed the data that comes out of statistics because they thought it was just a form of guessing, which it is definitely not, then we would loose a very powerful tool for examining the world around us. This was my worry.
Also I am just grumpy today. I am looking forward to your future articles, especially ones on the costs as that is definitely a more subjective measure and I am guessing as a woodworker for many years you know a good number of people who have hurt themselves.
Chris, As for my statement that you should be a professional statistician to do statistics: statistics are not just an opinion, they are a formal field of study that uses specific methodologies to come to conclusions. Without an very large amount of knowledge about how to do statistics you lack the ability to do any meaningful analysis. The same would be the same as saying that building "will/will not" stand without being a mechanical engineer. Yes you can "have an opinion" that their math is wrong but you are just flapping your jaw. There are exceptions usually of the form of the rigging crew saying "this beam should probably be 6" to the left, you know where the bolts actually come through the concrete".
I didn’t see any criticism of NEISS in this article whatsoever; what is implicitly (and I think rightly) criticized is the oversimplification of the NEISS data by other sources – such as, in this case, a media outlet who is inherently interested in conflict, which draws and holds readers more consistently than balanced and factual analysis of dry subjects like complex statistical analysis.
My tablesaw has been in storage for quite a while as it just wasn’t earning the space it occupied in my too-small shop. nonetheless, I am quite discouraged by the continual trend toward protecting people from themselves at the expense of choices. I’ve also noticed a tendency for people who don’t use tablesaws, or have much need for them, to be the ones leading the charge to eliminating them. Never a situation that sits well with me.
Personally, I think the key (as always) is to try to provide education and resources to improve people’s personal safety. However, legislating to the lowest common denominator of caution or education seems to be more and more the strategy du jour in the US.
Personally, I thought it was quite a good writeup Mr Lang.
Bob Miller forgot to add to his signature:
Yes, please, after I have my mandatory table saw
lessons I will also attend my mandatory training
for statistics as well since only certified
experts are allowed to have any opinion on
I am not even sure what Bob Lang said that was so bad. But
the suggestion that he can’t have an opinion because only
certified gurus can tell us what to do or how to think is
Well Bob, I don’t think I expressed any disdain or disagreement with "actual" numbers. It’s the guesses paraded as facts and counting apples as oranges that I have my doubts about. I fail to see how asking questions causes great harm.
I do not have much of an opinion on the matter. I have but do not use a tablesaw, largely because it is an older model that is painfully noisy even with hearing protection when used in my concrete basement.
What I do have an opinion on is this: I believe you do the world a great disservice by attacking modern statistical methods in a very public forum simply because you do not like the numbers they generate. From your article you do not have alternate data that indicate the NEISS data is flawed and from your bio on popularwoodworking.com there is no mention of a background in formal statistics. So I see no grounds for you to be able to disagree with the actual numbers.
Resources like NEISS are designed to take what we feel a number should be and replace it with something founded in reality. Such analyses have brought us modern medicine and engineering and we should not reject them because they run counter to our personal feelings.
I respect you greatly as a woodworker and I would much rather hear about how you plan on reacting to these numbers instead of you attacking them and their methodology. Much like a table saw statistics can seem easy to use at first but you can cause great harm trying to use them without training.
I do have to admit that amateur statistical analysis has been a pet peeve of mine for quite a while and usually sets me off. But I thought it should be stated here.
Thanks for the great article Bob. I appreciate you leading us through the wilderness of table saw injuries by sorting things out for us.
I was wondering though if the database included any information on what brand and model number table saw was being used in each case. That information might be helpful in indicating a particular saw or two that may be more prone to causing injuries than other saws for whatever reason.
Great article! I don’t have a dog in this fight – I don’t use a table saw – but I still find the discussion fascinating. One piece of information I would love to know in this debate, but is probably impossible to assess, is find out how many hours each injury saw operated in an average month. Given how low the injury rate is for professional shops I suspect what we would find would be an 80/20 rule situation – 80% of the injuries are from saws that account for only 20% of the total usage.
If it turns out that saws that run 40 hrs a week are causing fewer injuries than saws that run 4 hours a week I think it would be fair to say that table saws aren’t an inherently unsafe tool. If that’s really the case then I don’t think there is a good solution to the problem of hobbyists cutting off the their fingers on the table saw. I certainly can’t think of one.