Stop The Cycle–Terrible Theory, Terrible Execution
Georgia’s Strong 4 Life campaign seems to be intent on releasing commercials which aim at preventing childhood obesity by presenting a notably one-sided and stereotypical characterization of the obesity problem.
In trying to teach my students how to solve quantitative science problems, or my nieces how to be awesome in life, I keep insisting “Think about what you’re doing. What, in the most precise terms, are you trying to achieve. Then consider how to best achieve those precise things.” It would be nice if this were how the brain automatically worked, but it’s not: it’s actually an intensive process that’s far inferior, 99% of the time, to proceeding on gut instinct. But, for those occasions when you actually want to optimize your results, it’s useful.
So what is this video trying to do? Theoretically, I guess, it wants to show parents the error of their ways (though how feeding a kid French fries made him lazy later in life, I don’t know), and how parental shortcuts early on lead to obesity and an early death. The problem is, they should have paused. Their thought process should have included “We want to communicate to obese people. Therefore, we need to communicate in a way that doesn’t make them immediately turn off the commercial.”
Problems:
1) This seems to be based yet again on the health-belief model. This is the idea that people carefully weigh the health-related pros and cons of their behavior before choosing to undertake it, based on their perceived susceptibility to harm and the perceived severity of that harm. This model is completely broken outside of the clinical setting.
The HBM is a notoriously individualistic model. It says we each make health decisions, in a rational way, and are not susceptible to outside forces (although I think the video might be slightly more sophisticated than that – in that it presents the kid’s access to vending machines and the like – but it strongly implies that it all came down to mom ultimately deciding her convenience was more important than her kid’s health, and everything was just a series of dominos from there).
The problem is that health isn’t that individualistic. It depends on psychosocial factors (i.e., the kid’s peers, mom’s stress level), availability of healthy and unhealthy food (the government now recommends 5-9 servings of fruits and vegetables a day, which for a normal family of 4 implies the purchase and storage of 9 servings/day-person * 4 oz/serving * 4 people * 7 days * (1lb/16 oz) = 63 pounds of fresh produce weekly), financial situations, effects by environmental pollutants on the metabolism, etc.
Oh, also, you know, genetics. Hell, we know - in loose terms - that the bacterial content of your gut is enough to have an effect on your weight.
But let’s ignore all of those things that all those other failed public health interventions have ignored, and pretend it all comes down to mom giving her kid French fries.
As Dr. Freedhoff described the issue of addressing dietary behavior while ignoring environmental problems:
To help illustrate my point, try to imagine childhood obesity as a flooding river with no end in sight. While teaching children how to swim might help temporarily in keeping them afloat, given that the flood isn't abating, chances are, even with the best swimming instructions, the kids are going to get tired and sink. So while swimming lessons certainly can't hurt, what we really need to be shouting about doing is actually changing their environment and building them a levee.
Alright, so the first big problem with the campaign is that it makes completely unwarranted and ignorant assumptions about the cause of the pathology. This is like the good old days when cholera was treated with fresh air, because we assumed it was caused by bad smells.
Even if the campaign is effective in the sense that it makes mothers change their dietary behavior with respect to their children, it won’t actually work too well, since that’s only a single facet of the problem.
Problem 2) There’s no way it’s going to be effective.
First of all, the entire ad is from the point-of-view of a caricature of an obese man; lazy, ravenous, filthy (check out his shirt during the moment he comes into the scene). People tend to receive messages better from people like them or that are otherwise likable. Slim folks are going to curl their noses at disgust; overweight folks are going to do the same, with an added frisson of shame and contempt both for themselves and the idea that they might be perceived that way. No one is going to invest themselves in this commercial: it’s aversive from the get-go.
It then goes on to give us a quick dying-moments-recap of this guy’s life. It only belatedly strikes me that apparently this implies this guy’s entire life is couch-sitting and eating; that’s just a little anger-inspiring. Is that what Strong 4 Life’s public health people think is the requisite for obesity?
Besides being slightly offensive, it’s useless: this isn’t what real obese lives look like, even the lazy and overeating ones, which means it’s not only alienating, but it’s trying to reach a non-existent audience. (hint: if you’re going to try and increase minority higher education enrollment, and going to ignore all the social elements involved in that outcome, at least don’t make a video about a lazy, unemployed, musical, cocaine-snorting, sports enthusiast. That person doesn't exist, and real people are not engaged by denigrating stereotypes).
I actually get the impression that this piece was put together by a staff entirely absent obese people. It has that tone-deafness of 1950s Caucasians discussing problems in the African American community.
After missing all the social complexities, and making a negative and condescending video that will drive everyone away, did these guys do anything right?
I headed on over to Strong4Life’s site and found their “Quick tips” on how to help you be healthy.
Edit: I wrote a scathing and deeply mocking point-by-point dissection of their “quick tips.” I’ve removed them, since I felt it crossed the line in tone from “critical disagreement” to “excoriation.” I strongly suggest you head over there and take a look at them for yourself, though.
All in all, Strong 4 Life seems to have no grasp neither on good marketing, nor as to the myriad of elements that actually contribute to obesity. Their advice is both shallow and cliché, and frankly I hope that anyone that’s serious about undertaking a change in their lifestyle seek a source of information and inspiration diametrically opposite that of Strong4Life.
Edit: A medical student friend of mine wishes me to add that the man in the advert did not look "dirty," but rather sweaty, which is what happens during a heart attack.
I contend that this advert isn't aimed at MDs, it's aimed at the lay public, and as such will be connected to the ad's over-all theme of "dirty pig" far more easily than having people say "Oh, that's just a standard symptom of a heart attack." The medical details aren't relevant; this isn't a simulation, it's an advert, and as such I'm only interested in how and what it actually communicates to its audience.
Tip of the hat to Mr. Travis Saunders for finding this video, and to Dr. Michael Siegel for teaching the theory behind these ads.
KEGELES, S. (1980). Marshall H. Becker, Editor, The Health Belief Model and Personal Health Behavior, Slack, London (1974). Social Science & Medicine. Part C: Medical Economics, 14 (3), 227-229 DOI: 10.1016/S0160-7995(80)80012-8
Clinical Significance vs. Statistical Significance
You’d be surprised how often this comes up: I’ve heard several times, now, the comment that “just because it’s not statistically significant doesn’t mean it’s not clinically significant”. Here’s the thing, though: the word “significance” means different things in different circles, and this sounds rather like creationists crowing that a scientist admitted “Evolution is not a fact.” Heck, in one original twist, we’ve got a surgeon crowing about the fact that statistical and clinical significance aren’t the same – as though the statisticians didn’t already know that.
Clinical significance uses the word “significance” in the colloquial sense. It essentially means “important in the treatment of patients.” It generally leads directly to healthier people – symptoms get better, mortality rates fall, etc. Sometimes you can find a detectable effect that doesn’t seem clinically significant: the PSA Test finds something very real, but it’s been found that it vastly over-detects cancer and isn’t really all that important in patient care. A clinically significant effect is usually relative to placebo; where a placebo is the benchmark of no real effect (an oversimplification, I admit, but for most placebos it’s about true), it’s the improvement in symptoms beyond placebo that measures clinically significant advantages of a therapy.
Statistical significance isn’t about therapies at all. It’s a description of the data and its variance. Essentially, it says, compared to some benchmark, what is the probability that the data we’re looking at is just a chance variation on the benchmark? If 95% of normal blood pressures fall between 110 mmHg and 140 mmHg (hypothetically), what’s the chance that our patients with blood pressures 110, 114, 124, and 135 belong to that “normal” group? Maybe there’s a 30% chance. Maybe 11%. Maybe 3%.
I can set the cut-off for “wow, these are some non-normal blood pressures” anywhere I like. By tradition, it’s often 5%. So, if there’s a 5% or less chance our patients belong to the “normal” range, we say hey, that’s statistically significant. It doesn’t mean important, it just means “meets our cut-off for some chance that this is just a sample of normal people.” It says nothing about importance. It doesn’t mean the study is important; it doesn’t mean “the numbers are important” (which is how some people seem to interpret it); it just means that the numbers meet our cut-off for some level of (im)probability of our sample coming from the “normal” or benchmark population.
That said, statistical significant usually is clinical significance, which is where picking on the CPAP writers doesn’t look awesome. Fact is, the smaller an effect you’re looking for – say, a 5% reduction in blood pressure rather than 35% – the bigger a group of patients you have to recruit. You can sort of think of the number of patients – sample size – as the power of a microscope. The smaller an effect you’re looking for (i.e., cellular degeneration rather than…a lopped-off hand?) the more power you need.
In practical terms this means that very few studies have the sample size to find tiny, unimportant effects. There are some enormous studies, like the Physician’s Health Study, that can indeed locate statistically significant effects that might not be clinically significant. Most studies, though, aren’t powered to catch relative risks less than 2 to 3 –fold (i.e., “if you have this gene, you’re 3x more likely to get cancer”).
Effects of the size most studies can catch are generally clinically important. How often do you see a paper that says “This treatment reduces symptom X by 30%. Had no effect on health”? It happens, sure, but that usually depends on people choosing a bad X – I’m thinking of things like proxy indicators, like following cholesterol for MI and assuming cutting cholesterol by 30% will cut heart attacks 30%. If X is in any way a meaningful choice – and it usually is, doctors by and large being smart cookies – then reducing X by 30% will have some sort of clinical impact. Cutting cholesterol 30% won’t reduce heart attacks 30%, but it may reduce heart attacks 3%. Whether this is super-useful is questionable, but it’s not clinically insignificant. It’s important. Because we're usually looking at things that matter, finding statistically significant trends in things that matter usually matters.
For example, in the study looked at by the skeptical surgeon, episodes of apnea dropped from 48 to 31 per hour, on average. Since pulmonologists call “30+ episodes” severe, the category doesn’t change, and therefore he concludes “it’s statistically significant, but not clinically so.” And yet, whatever classification we attach, in any condition with a continuous symptom (the number of episodes can range anywhere from 0 to some number, it’s not just Yes/No), reducing the severity of the symptom by 17 – a whopping 35% – is assured to have some impact on the person suffering from apnea. In a disease like apnea, where the symptoms are the disease – episodes of difficulty breathing slowly choke the brain at night – having fewer episodes of choking would generally be a good thing, indicating less brain-choking. Whether or not we’ve had a large enough study to detect the exact dose-response between consequent symptoms and number of brain-choke episodes, it’s still pretty fair to assume that less brain choking will lead to less brain choking problems.
Just because it’s statistically significant doesn’t mean it’s clinically significant – but it usually does. Stay tuned: 50/50 odds I’ll change my mind next week.
Redolfi, S., Arnulf, I., Pottier, M., Lajou, J., Koskas, I., Bradley, T., & Similowski, T. (2011). Attenuation of Obstructive Sleep Apnea by Compression Stockings in Subjects with Venous Insufficiency American Journal of Respiratory and Critical Care Medicine, 184 (9), 1062-1066 DOI: 10.1164/rccm.201102-0350OC
Hello world!
As a public health grad student, I've been growing concerned regarding the dire shortage of smartasses on the Internet. Don't worry; I'm here to help.