Tylenol: A narrow margin of safety

I’m always surprised at how many people don’t know how dangerous Tylenol (or anything with acetaminophen/paracetamol) is. I can’t remember where I initally learned that just a few pills could cause liver failure, but it was at least 10 years ago. Of course, more than 10 years ago, I remember that Tylenol was the only painkiller allowed to be given out in my high school because it was so “safe.”

Anyway, it’s certainly been in the news lately and I have collected some interesting sources. First off, This American Life devoted an hour to investigating the problem, its history, and ongoing conflict. Absolutely worth a listen. Here is a teaser bit of the transcript:

Will Lee is one of the liver specialists and researchers who’s been calling attention to the dangers of acetaminophen. He is at the University of Texas Southwestern Medical Center and talked to reporter Sean Cole.

Will Lee: In one of the articles, we actually printed up the little coupon that I got with my Zocor prescription that says, $2 off on your next acetaminophen bottle. And one of the things on the coupon, it says safest. Not safer, or not safe, but safest. Well, this is the number-one drug killing Americans every year.

Sean Cole: Over-the-counter drug killing Americans.

Will Lee: Over-the-counter drug.

Sean Cole: Not the safest.

Will Lee: Not the safest, for sure. So I guess that’s– if you think I have a bee in my bonnet, that’s probably where it came from.

There are two human factors issues related to this “narrow margin of safety” that I’d like to highlight:

1. Warning design
2. Counter-intuitive doses for children


If you’d like a good overview of warning design, see Wogalter, Conzola, & Smith-Jackson, 2002.

The warnings are printed in red on the lid of the container and say “Contain acetaminophen – Always read the label.”
I know it took a long time to even have this, but I’d like to point out that this does not follow the best practices for warning design because it does not include the consequences or focus on the “hidden hazard.” Hidden hazards are really the reason warnings are important. Here are some examples:
  1. Does a knife need a warning that it is sharp? No, that’s not a hidden hazard. Now, if the knife is SO sharp it’s unexpectedly dangerous, then it needs a warning that conveys that information. I had a friend in high school who had a habit of testing knives with his fingertip. This never cuts you, just gives a feel for sharpness. Well, one time he had someone sharpen a swiss army knife for him – and what usually would have “felt sharp” went instantly through the skin down to the bone. That’s a hidden hazard.
  2. Does coffee need a warning that it’s hot? No, not if it’s just drink-ably hot. But yes if it is purposefully super-heated enough to cause 3rd degree burns. Watch this video to get new insight on the McDonald’s coffee story.

It is the narrow margin of safety that is the hidden hazard of acetaminophen. Nothing else available over the counter can kill you with such a small increase in dose. You can take 20x the recommended dose of Advil before it becomes threatening. I ignored the dosing for Advil once and took 6 in 3 hours rather than 2 – it made me sick and I regretted it, but would I deserve to die for that mistake? I could have died if it had been Tylenol instead of Advil. This hidden hazard is the most critical part of the warning – even if it’s on the label, I think it should be highlighted on the lid.

Which brings me to… the hidden hidden hazard that was created by the Tylenol company.

Counter-intuitive doses for children
In brief, for over a decade Tylenol provided medication labeled for infants and medication labeled for children (and of course, medication for adults). 99/100 people on the street would assume that these were in order of strength – least for the infants, more for the kids, and most for the adults. That assumption was correct for the kids and the adults. The counter-intuitive dose was for the infants: instead of being weaker because they were smaller, it was much higher. Their reasoning was that it is hard to get medication into an infant, so if it’s stronger you can get an effect with less. An article from Pro-Publica gives a more detailed summary.
An excerpt from Pro-Publica:

The two types of pediatric Tylenol had a counterintuitive difference. Drop for drop, the strength of Infants’ Tylenol far exceeded that of Children’s Tylenol.

In addition, the active ingredient in Tylenol, acetaminophen, has what the FDA deems a narrow margin of safety. The drug is generally safe at recommended doses, but the difference between the dose that helps and the dose that can cause serious harm is one of the smallest for any over-the-counter drug.

By confusing the pediatric products and administering too much of the infants’ version, parents could inadvertently overdose their children. Other manufacturers also made two children’s products with different concentrations of acetaminophen.

Between 2000 and 2009, the FDA received reports of 20 children dying from acetaminophen toxicity – a figure the agency said likely “significantly underestimates” the problem. Three deaths were tied directly to mix-ups involving the two pediatric medicines. Such errors may have caused some of the other deaths, but the agency has acknowledged that its data lacks sufficient detail to determine the precise cause.

This American Life included a heartbreaking case where parents were instructed to give their infant doses measured in terms of the children’s Tylenol (their doctor just said “Tylenol”), but since they had an infant they did what anyone would and used the Tylenol for infants, resulting in the death of their baby.

The big push seems to be for better warnings (with the example in the picture as “better.”) Indeed, this can have an impact – for example, as reported in the NY Times, removing infant versions of medication and saying a drug is “not for children under 2” reduced drug-related emergency room visits for that age range by half.

Excerpt from the NY Times article:

In 2007, amid mounting concern that infant cough and cold medicines were unsafe and misused, manufacturers voluntarily withdrew products intended for children younger than 2. The makers revised the labels on the rest of the medicines, which now warn parents that they should not be given to children younger than 4.

Government researchers said on Monday that those moves have had a remarkable effect: a significant decrease in emergency hospital visits by toddlers and infants with suspected medical problems after using these medicines.

Dr. Daniel Frattarelli, a former chairman of the committee on drugs at the American Academy of Pediatrics, praised the study, saying it showed that “the label is a very powerful tool for changing parent behavior.”

In the new study, published in the journal Pediatrics, researchers at the Centers for Disease Control and Prevention reviewed data from 63 hospitals to estimate the number of emergency visits from 2004 to 2011 by young children who had taken cough and cold medicines.

Children under 2 accounted for 4.1 percent of all emergency visits for suspected drug-related effects before the 2007 withdrawal, the researchers found, and accounted for 2.4 percent afterward.

However, even casual readers of the blog have probably noted how often I mention the Hierarchy of Safety: first try to design out the hazard, guard against the hazard, and warn. I’m not saying warnings aren’t important, but if we want to have the biggest impact we should be working on designing out or guarding against the hazard. Some ideas in that realm include bottle design that restricts the flow of liquid (this could prevent a child from dosing him or herself, but not parents from giving the wrong dose), and packaging “single servings” of medication, so that its obvious how much to give at one time. Although the treehugger in me isn’t a fan of more packaging, this also could provide more space for good warnings AND have those warnings in extremely close spatial and temporal proximity to use of the product.


Brilliant guard against accidents in indoor rock climbing

For those who don’t follow news of climbing accidents as closely as I do, there has been a spate of accidents associated with the automatic belay devices (autobelays) installed at climbing gyms.

These devices are handy to have around as they negate the need for a climbing partner, allowing one to exercise and train alone. The climber clips his or her harness into the device at the bottom of the wall, and it automatically retracts (like a seat belt) when you climb upward. At the top, you let go of the wall and the device lowers you slowly back to the ground. You are probably imagining that the accidents had to do with failures of the equipment – while that is not unheard of, the most recent issues have all been with climbers forgetting to clip into the system at all.

The most recent tragedy occurred this past September, where an experienced climber died after a fall in a Texas gym, and it’s been listed as so common it happens at “every gym,” though not always resulting in a fall. Here is the facebook page with members of another gym discussing a similar accident.


If you talk with climbers or read accident forums you will invariably be faced with a large contingent bent on blaming the victim. I’ll grant that it is hard to imagine forgetting to clip into a safety device and climb 30 feet up a wall, but that’s because I hardly ever do it. One characteristics these accidents share is that the victims were experienced and used the auto-belays frequently.

When a procedure becomes automatic, it becomes more accurate and less effortful, but it also becomes less accessible to the conscious mind. When a step is skipped, but all other steps are unaffected, it’s especially hard to notice the skipped step in an automatic process. If caring more or working harder or “being more careful” could actually prevent this type of problem, we wouldn’t have any toddlers left in hot cars, perfectly good airplanes flown into the ground, or climbers falling because they didn’t clip into the autobelay.

That brings me to the device I saw installed at a climbing gym last night.

guardAbove: The guard in place, clipped to the wall and ready to go. Notice how it blocks the footholds of the climbs.

photo 2Above: Nikki shows how to unclip the guard before attaching to her harness.


photo 4Above: Clipped in and safely ready to go. Guard is on the ground and out of the way (it is ok to step on it!)

Let me tell you why I think this is brilliant.

  • It’s highly visible.
  • It functions as a guard. This adheres to the hierarchy of safety: First, try to design out the hazard. Second, guard against the hazard. Last, warn. These are in order of effectiveness. Prior to this device, I had only seen signs on the wall saying “Clip in!” (And a year ago, even those didn’t exist.) This device physically blocks the start of the climbing routes, demanding interaction before one starts climbing.
  • Using it properly does not add any additional time or mess to climbing a route. If it weren’t there, the climber would still have to unclip the autobelay from an anchor close to the ground, etc. With it there, the climber does the same thing and once done, the guard becomes a flat mat that doesn’t get in anyone’s way.

Is it perfect? No. You can also climb with a belayer on the same or nearby routes, and then it’s also blocking your way at the start of the climb. Some adaptation should be made by the route-setters at the gyms to minimize this. But overall, what a great and simple solution.


Usability process not used for ACA website

slideA recently released report, done in March 2013, reveals the process of creating Healthcare.gov. Hindsight is always 20/20, but we’ve also worked hard to establish best practices for considering both engineering and the user in software development. These contributions need to be valued, especially for large scale projects. After looking through the slides, one thing I note is that even this improved approach barely mentions the end users of the website. There is one slide that states “Identify consumer paths; review and modify vignettes.” The two examples of this are users who have more or less complex needs when signing up for insurance. I don’t see any mention of involving actual users prior to release.

The NPR write-up states:

Consultants noted there was no clear leader in charge of this project, which we now know contributed to its disastrous release. And there was no “end-to-end testing” of its full implementation, something we now know never happened.

Some of this may fall on us, for not being convincing enough that human factors methods are worth the investment. How much would the public be willing to pay for a solid usability team to work with the website developers?