• When good intentions make bad policy – car seat edition

    I recently got into a heated argument with a colleague over the use of car seats on an airplane. She felt that they should be mandatory for small children; I disagreed.

    You see (for those of you who don’t have kids), up until a child is two years old, you can choose to have them ride in your lap instead of buying them their own seat. While this may seem uncomfortable for many, I know of any number of people (including yours truly) who have flown holding an infant in their laps. When our first child was born, I was a fellow and my wife wasn’t working, so travelling to see family was a luxury. We couldn’t easily afford that extra ticket.

    Around that time, the FAA was considering making it mandatory for children less than two years of age to be restrained in their own seats on airplanes. The American Academy of Pediatrics agreed with them, and issued a policy statement in November of 2001:

    Occupant protection policies for children younger than 2 years on aircraft are inconsistent with all other national policies on safe transportation. Children younger than 2 years are not required to be restrained or secured on aircraft during takeoff, landing, and conditions of turbulence. They are permitted to be held on the lap of an adult. Preventable injuries and deaths have occurred in children younger than 2 years who were unrestrained inaircraft during survivable crashes and conditions of turbulence. The American Academy of Pediatrics recommends a mandatory federal requirement for restraint use for children on aircraft. The Academy further recommends that parents ensure that a seat is available for all children during aircraft transport and follow current recommendations for restraint use for all children. Physicians play a significant role in counseling families, advocating for public policy mandates, and encouraging technologic research that will improve protection of children in aircraft.

    Let’s review. First of all, the AAP wanted the FAA to make it a mandatory requirement for children to be restrained on aircraft. They also wanted me, as a parent, to buy a seat for my child. Finally, they they informed me, as a pediatrician, that it was my duty to tell all of my patients’ parents about this.

    I wasn’t too happy about that, for any number of reasons. I got why child safety seats are important in cars. After all, you are much more likely to walk away from a car crash if you are adequately restrained. But can the same be said about a plane crash? Really? There are other considerations as well. How effective would such a policy be? How much would it cost?

    Luckily, some enterprising researchers* crunched the numbers. They performed an econometric analysis to see what would happen under such a policy. Here’s the first bit of awesomeness: the recommended policy would likely prevent about 0.4 child air crash deaths per year in the US. Got that? Less than one child’s life might be saved each year.

    Now you may be thinking that it’s still worth it. After all, that child may be someone you know. There’s a problem, though. If you make people buy seats for their children, they may not be able to afford to fly. Instead, they might choose to drive. Driving is way more unsafe than flying. It’s so much more unsafe that it turned out this policy would increase the number of child deaths if somewhere between 5% to 10% of families decided to make the trip by car instead of plane. Of course, this calculation varied by the distance being driven, with longer trips incurring more car-related danger. But if the average trip was 400 miles, then if just 5% of families chose to drive instead of fly, this policy would result in an increase in child deaths.

    Not so clear anymore is it?

    Moreover, this doesn’t take into account the cost of the policy. Let’s say, for the point of argument, that no families whatsoever would convert to a roadtrip. That’s not realistic, but go with it for a second. This means that we’d actually save lives under the mandatory restraint policy. Let’s also stipulate that the average cost of the policy would be about $200 for a seat for each small child flying. Sound OK?

    Under these simple and reasonable assumptions, the cost per each child death prevented would be $1.3 billion. If you want to be precise, it was calculated to be $1,283,594,063. For comparison, that’s about 33,000 times more per life-year saved than the policies that mandate restraints in cars. It’s also pretty much the most expensive injury prevention policy imaginable.

    If you want to save some children’s lives, I’ve got plenty of ideas for you. None of them will cost in the billions of dollars per prevented death. Here’s a good one from just yesterday over at JAMA’s blog: remember to make sure your child is still using a booster seat when they are being carpooled in other people’s cars.

    *Full disclosure: I know all of the authors of this study. One was even a director of my fellowship. Here’s the link to the paper, but it’s probably gated.

    • You are only addressing the issue of air crashes and deaths. What about injuries cause by turbulence or rough landings which are much more common occurrences.

      Airline seat belts requirements are much more about preventing injuries than they are about preventing death.

      • Again, though, you’re much more likely to get injured in a car. And people will opt to drive. So the same calculations apply.

    • Excellent post. people generally do not spend any time thinking about such risks and policies and so they come up with the wrong conclusion, even some in The American Academy of Pediatrics evidently. That is one of the reason that many of us would like to see very high deductible insurance. People tend to think a little bit more about alternatives if the direct cost to them is high.

      My blog post on a high deductible policy:


    • While I don’t necessarily disagree with Aaron on the issue overall, I do agree that the relevant data is in turbulence related injuries. The vast majority of in flight injuries result from people who are unrestrained during unexpected turbulence. I’m not sure if there’s a good source available for statistics on injuries releated to turbulence, but I wouldn’t be suprised to see in order of magnitude incerase when compared to crash deaths. If you can track down that data, it would make a worthwhile follow-up.

      • Again, though, injuries from cars FAR outstrip injuries from planes. So you need just a small conversion to driving to make the number of overall injuries go up. You’re falling into the same trap of assuming that a bad outcome will go down on one side and nothing will happen on the other.

        • Actually, I think it is a different trap. Lots of folks, in the comments here and in the physical world, take issue with a study or commentary because it only focuses on XYZ and not ABC. That may be true, but it often doesn’t invalidate the main qualitative points, which is the case here as Aaron has stated. I think people are very good at critique but not good at sensing when the critique misses the forest for the trees.

    • The study only looks at the risk of death to infant. But who is actually at risk in an airplane? An unrestrained baby in an airplane places the surrounding passengers at risk. A risk they did not voluntarily assume. That is where comparison to a family traveling in a car falls apart.

      As far as misses the forest for the trees, I maintain that the study focused on one tree and ignored the forest of trees that relate to the issue of unrestrained infants.

      • Couldn’t a child be restrained with a belt-type system while still being in a parent’s lap? Is the purchase of another seat required to protect the safety of other passengers?

        Don’t drivers put other drivers at risk? Don’t those other drivers assume that risk when they take to the road? By the same logic, don’t other passengers assume all the risks of air travel when they fly, including that due to an unrestrained child?

        In short, you’ve lost me.

        • Really? The fault with this study is that they didn’t model the risk of an unrestrained infant hurting others?

          I maintain that there are many, many other things (computers, bottles, drink carts, ipods, ipads, other tablets, shoes, dvd players, bags, etc.) that are far more likely to hit you on a plane than in a car. Should we mandate restraint of all these objects, too?

    • “Preventable injuries and deaths have occurred in children younger than 2 years who were unrestrained in aircraft during survivable crashes and conditions of turbulence.”

      Please note that the AAP recommendations were based on deaths and injuries during survivable crashes and conditions of turbulence. The study you are citing looked only at deaths: “the recommended policy would likely prevent about 0.4 child air crash deaths per year in the US.”

      “I maintain that there are many, many other things (computers, bottles, drink carts, ipods, ipads, other tablets, shoes, dvd players, bags, etc.) that are far more likely to hit you on a plane than in a car. Should we mandate restraint of all these objects, too?”

      They are ready are. Electronic devices are required to be put away during take off and landing. At any time during the flight, the captain may direct that all such devices be put under the seat or in the overhead compartments. I have been on flights where this was done and where the attendants were told to secure all equipment and take their seats. It’s an order that gets your attention.

      BTW, I look forward to reading this blog every date. It is my main source of information on economic and health care matters. I just happen to disagree with your position on this matter. We always purchased seats for our son regardless of his age.

    • There are regulations covering other objects. They must be put away for take offs and landings and the captain may order that they be put away at anytime during the flight. Just as he may order the flight attendants halt all service, secure all equipment and take their seats and all passengers to return to their seats and fasten their seat belts. Except, of course, for infants who are sitting in their parents lab.

      As far as the major faults with this study are concerned, the AAP recommendation was based upon deaths and injuries due to survivable crashes and turbulence. The study did not address injuries and, apparently, did not distinguish between survivable crashes and not survivable crashes.

    • You analysis makes sense to me. A hypothetical (and maybe absurd?) thought question though: from a cost-effectiveness point of view wouldn’t a direct subsidy of airline travel — that encourages people to shift their travel from the more dangerous road trip to the safer flight — pay off in terms of dollars per life saved? Ie, flying is so much safer than driving that encouraging travelers to shift more of their long distance travel from driving to flying would reduce death, and it might actually be cost effective to do so (up to a point). I don’t have the data to crunch it out myself but it seems plausible though not certain.

    • There would seem to be another issue with requiring car seats for air travel that is not addressed: Car seats are designed (quite specifically) for events that do not happen in airplanes. For the under-age-two set in question, the primary consideration is sudden acceleration/deceleration. High transient G loads are a problem in aircraft, but if you experience them, then you’re probably not in a position where a car seat is going to do you any good, i.e., crashing.

      Additionally, the same American Academy of Pediatrics has now recommended (and will probably soon mandate) rear-facing seats for all children under two. Rear-facing seats barely fit in a small car, let alone an airplane seat. Forward-facing children under two in a collision do not receive sufficient protection.

      They must be properly installed with LATCH anchors and installed in the correct angle, to offer protection. Do aircraft seatbelts have the same stretch factor as car seatbelts? That’s part of the design assumption of a car seat.

      How many children under two are on an average flight? Two or three? Would having the airline provide half-a-dozen specially designed, aircraft-specific booster seats be truly onerous? Southwest, as an example, operates 559 Boeing 737s. Even if they had to buy 6,000 seats, this is not a huge expense. $500,000? The entire US passenger fleet is around 7,000 aircraft ). Create a $100 seat, and it’s under $10 million. You have them stacked up at the gate, hand one out as needed.