Aaron: We're eating too much, and there are lots of people who think that menu labeling or putting the number of calories in stuff up on the wall will help people to eat more healthily. Sounds great, right? You think there might be some research on that?  Menu labeling is the topic of this week's Healthcare Triage.

(HT Intro plays)

Aaron: Menu labels make sense. We want to know what we're eating, and there's good reason to have those labels. We often have no idea what we're eating. When I was a kid, there was a simple rule in our house when buying cereal: sugar couldn't be the first ingredient in the list. This led to cereals being lumped into two groups: there were sugar cereals and all others. Looking back, this seems laughable, but it kept the peace. My siblings and I became masters, however, at gaming the system. We knew the ingredient list of every cereal and could quote you, at length, lists of cereals that passed the test, yet still seemed awfully sweet to us. We have no such rule in our house today, partially because my wife and I are tyrants and just don't allow some cereals even to be debated, plus, no cereal seems to have sugar as the number one ingredient anymore, and now they all boast "whole grains" so you might assume they're all just healthier. Years ago, though, Consumer Reports did a study. They found that 11 of the most popular brands of cereal are more than 40% sugar by weight. For instance, one cup of Cascadian Farms Organic Oats and Honey Granola has 348 calories. It has the same amount of sugar at 19 grams as a Hershey bar, and if you couple it with half a cup of low-fat milk, it has the same amount of fat as a McDonald's cheeseburger. Honestly? I'd rather have the cheeseburger.  

Anyway, now that I know, I'd like to think I'd avoid the crazy bad cereal. But that's not the same thing as going into a restaurant and ordering food. Let's look at the research there. The American Journal of Preventive Medicine, 2011, "The impact of menu labeling on fast food purchases for children and parents." In this study, researchers gathered a number of families in two counties in the United States and measured their ordering in restaurants. Then, in one of the counties, they implemented menu labeling. Later, they remeasured the ordering in both the county that got labeled and the one that didn't. In the county where menu labeling was implemented, calories ordered for children went from 823 on average to 822. In the county without labeling, calories ordered for children went from 984 to 949. Not an impressive difference. The calories ordered by parents for themselves did drop in the menu labeled restaurants from 823 to 720, but they also dropped in the non-labeled restaurants, just as much, from 895 to 789. Parents said that they did see the nutrition information in the labeling restaurants. Some may hold this up as a good thing, but to be honest, it makes me even more sad. They saw the calorie information, but didn't seem to care that much, or at least they didn't change their behavior that much because of it.  

JAMA, 2011, "Accuracy of stated energy contents of restaurant foods." Okay, in this one, researchers went into restaurants with menu labeling and bought their food. Then they took them to the lab and measured how close the stated caloric content was to the actual caloric content. To be eligible, a restaurant had to be in a chain with sales placing them in the top 400 restaurants in 2008, and had to be either quick-serve or sit-down. They also had to have calorie content reported on their website. So how accurately did they report the calories in their food items? Wait for it. Well, overall, they were pretty accurate, but individual items showed a lot more variation. Of the 269 food items that were measured, almost 20% had 100 or more actual calories than what was stated. The worst offender, a side dish, had more than 1000 calories in a portion that was reported to have only 450 calories.  

Now, it's possible that the researchers got a bad sample. Perhaps there was an overeager server that night who gave them more than they were supposed to get. So in the interest of accuracy, the researchers took some of the worst offenders and went back to get a second sample. They were able to do this for 13 of the 17 foods with the largest discrepancies between the reported and measured calories. In the first pass, these foods had on average 289 more calories than reported. In the second look, they still had 258 more calories than reported. Most concerning for those trying to watch what they eat, the food items with lower reported calories, or the healthy or diet items, were significantly more likely to have higher calories than reported. This was balanced oddly enough by the high reported calorie foods sometimes having fewer calories than reported. 

Now, this isn't to say that I think the restaurants are lying, but if you do look at menu labels, it's gonna screw with your choices. The American Journal of Public Health, 2013, "Supplementing menu labeling with calorie recommendations to test for facilitation effects." Researchers approached 1,121 McDonald's customers both before and after menu labeling began in New York City. Each had a random chance of being handed 1) information that described the recommended calories a man and woman should eat each day, 2) information that described the recommended calories a man or woman should eat each meal, or 3) nothing. The hypothesis was that giving people information about recommended intake would help them to make better choices about how much to order. After all, the whole menu labeling thing is based on the idea that giving people information will reduce obesity. The results are somewhat depressing. Giving people calorie recommendations didn't change what people ordered. In fact, although the result wasn't statistically significant (p=.07), people who were given more calorie information ordered more calories. They also did a lot of sub-analyses. For instance, looking only at those who were ordering full meals instead of just drinks yielded the same non-significant result. They also looked at whether weight changed the results, 'cause if information caused overweight people alone to order fewer calories, that would still be a win. But that model still showed that more information led to more calories ordered, p=.06.  

But let's not cherry-pick. The American Journal of Public Health, 2015, "Systematic review and meta-analysis of the impact of restaurant menu calorie labeling." These researchers looked at all papers in the literature through October of 2013. If they included all 19 qualifying studies, then menu labeling led to an 18 calorie reduction per meal offered, but there was a lot of variation in the studies. 10 of them were in non-restaurant settings. Four were on the Internet, three in labs, two on sidewalks, and one in a hospital waiting room. Only three of them randomized people to order food that they'd otherwise eat from an actual menu. Seven of them, in other words, involved ordering from a hypothetical menu for a pretend meal. All of this is to say that their generalizability should be viewed with a little skepticism. If the researchers included only controlled studies, however, and there were six of them, then there was no significant effect of menu labeling.

That's disappointing, but not unexpected, given what I've already described, but here's the conclusion of the authors, and I'm quoting, "These findings are limited by significant heterogeneity among non-restaurant studies and few studies conducted in restaurant settings." I just have to disagree here. I'm not sure that menu labeling regulations are low-cost. Do they give any evidence for that?  I'm sure it costs businesses quite a bit. We also know, they're often wrong. And without evidence, why would we continue to head down this path? I'm fine with further experiments and research, but it seems odd to do studies, note they don't really seem to work, and say we should keep doing the policy. It sometimes feels that that's the theme of too much nutrition policy in the United States these days.

(HT endscreen)