Friday, May 5, 2017

The Diet Sodas and Strokes Study - Junk Science?

Back in late April, the "scientists say" study of the week became a study from the American Heart Association's journal Stroke which said drinking as little as one diet soda a day led to a nearly 3x relative risk of ischemic strokes (loss of blood flow) or Alzheimer's disease.  (The exact numbers were 2.96x for stroke and 2.89x for dementia)  The conclusion of the piece was that artificially sweetened soft drinks were associated with the increased risk while sugar-sweetened soft drinks were not.  The press immediately jumped into action saying "one diet coke a day triples your risk!". 

So what's going on here?  Is this meaningful?  Is it presumptuous of me to call a study in a prestigious journal junk?  Stay tuned...

To begin with, the study used nice-sized groups; sure I'd like to see bigger (by at least 10 or 20x and preferably 100x) but they weren't really small groups of < 100 like you sometimes see.  They used 2888 participants over age 45 in the stroke portion and 1484 over 60 in the dementia portion, and the study was conducted over a long time, from 1991 to 2011.  That sort of makes me wonder why it took them until 2017 to announce this, but perhaps they had "bigger fish to fry" and were looking for other associations first.  The biggest disadvantage is that it was a Prospective Cohort study, which attempts to find associations.  How the study was completed was by food-frequency questionnaire at 3 to 4 year intervals,   So, real quick, how many diet sodas did you drink per day in 2014?  This sort of recall by questionnaire stuff just isn't very accurate.

No study of this kind can prove cause and effect, it can only show that two things are associated with each other.  The cliche' example is that firemen are associated with house fires, but nobody thinks firemen are the cause of house fires.  Just like the firemen being at house fires, it may be that they're associated for a different reason.  A prospective cohort study can't prove cause and effect but is useful for finding associations to study with double blind, controlled studies.  Those, unfortunately, are ferociously expensive, and therefore completely impossible to conduct for 20 years, like this one.  There some minor disclaimers for really strong effects, if a plausible mechanism is postulated. 

The thing about this study and its reporting that immediately gave me grief is that it said, "artificial sweeteners" as if it was one compound or class of compounds.  Not even close.  It implies that they're all involved in some sort of mechanism, but no such mechanism is known.  Artificial sweeteners?  Aspartame is a dipeptide,  which is chemically more like a protein (extremely simple...) than a sugar; saccharin is an aromatic compound, unlike sugar or aspartame; in contrast, sucralose is a sugar, just not metabolizable and acesulfame (or Ace K) is another aromatic.  I think those are the sweeteners mostly found in diet sodas although my "artificial sweetener recall questionnaire" doesn't show what was in diet drinks in the 1990s.  The sugar alcohols like sorbitol, xylitol, mannitol and so on are, well, sugar alcohols and are probably more like a sugar than the others mentioned but they're used more often in other foods than sodas.  All of those compounds are handled differently by the body, so it makes me ponder what could they have in common.  If this disparate group of compounds is doing something, it can't be chemical. 

The authors didn't actually put forward an explanation, just the usual refrain: "more research is needed", which is fine; "more research" is what researchers do.

I have a suggestion.  What they've found is that people on diets drink diet soda.  They've found a correlation with something else; people who may have been told they were at risk for Something by their doctors are trying to lose weight, or "get healthy" and are cutting out sugars because of that.  Or they're dieting because they think they should for some reason.  The people in the over-45 stroke cohort or the over-60 dementia cohort who are still drinking sugary drinks are the exception, not the norm.  To say the strokes or dementia associate with drinking diet sodas is looking at the problem backwards.  The risk didn't come from the diet sodas, it came from whatever caused them to start drinking diet soda in the first place.  Find out what the people who were drinking sugary drinks and did not have strokes or dementia were doing that allowed them to do that.  Other than having chosen better parents.

What's my interest, other than just curiosity?  I typically have one diet soda per day; I can think of days when I've had two, but I can think of plenty of days when I haven't had any.  Mrs. Graybeard and I refer to studies like this as he-who studies, or just he-whos: he who drinks one diet soda a day triples his risk of a stroke or dementia.  The media loves he-whos because they require no thought and can be dramatic; just present the press release.  Sodas in general and diet sodas in particular are demonized by a certain segment of society so there's a ready audience.  I look at sodas like everything else:  there are trades involved, everything has a cost/benefit ratio.  Don't think changing from sugary soda to diet soda will fix all your problems, but I don't think changing from soda to unsweetened tea or plain water will fix all of your problems either.
Hey, I said I had one per day, didn't I?  Just kidding.  Image stolen from.


  1. There's a lot to say for a tall, cool glass of pure water.

  2. SiG, I've read in one or two places that - supposedly - drinking carbonated beverages can reduce or rob the calcium stored in bones, putting people (elderly, mainly) at risk for osteopenia (susceptible to developing osteoporosis) or osteoporosis (basically weak, fragile bones which can break more easily than normal). Have you read or heard anything about that?

    1. That sounded vaguely familiar. I think the argument was the phosphoric acid in sodas upset the phosphorus/calcium balance and caused some scavenging of calcium in the bones.

      That's a hazy memory, which made me go look up some commentary on WebMD.

      They started out saying that there was a correlation simply because if people are drinking sodas, they're not drinking milk or other "healthier" drinks and so have lower calcium intake. Then followed it up with:

      "Researchers at Tufts University, studying several thousand men and women, found that women who regularly drank cola-based sodas -- three or more a day -- had almost 4% lower bone mineral density in the hip, even though researchers controlled for calcium and vitamin D intake. But women who drank non-cola soft drinks, like Sprite or Mountain Dew, didn't appear to have lower bone density." [emphasis added]

      Then they went on to talk about the phosphoric acid and messing with the phosphorus/calcium ratio.

      On the other other hand, then they said that the amount of phosphoric acid in soda is minimal compared to that found in chicken or cheese, which no one seems to be telling people to avoid. (My favorite in this category is telling people to avoid cured meats because of the nitrates in it, then I found the amount of nitrate in celery is typically higher, and now there's a beet product on the market that sells itself as being high in "dietary nitrates".)

      So it sounds to me as one of those weak correlations that probably indicates something else is really what's being correlated, but they don't know how to ask the question.

  3. The implied precision is also an example of "bad science".

    Three significant figures implies great precision, like being able to measure one's height to the nearest millimeter. That degree of precision, when talking about height, raises questions of detail: barefoot-or-otherwise, hair style, time of day.

    Relying on people's memory is clearly not conducive to three significant digits of precision. I cannot remember what car I was driving in 1991. The best one could hope for is "attribute" conclusions. That is, diet soda probably has negative implications, or the data is inconclusive or the data suggests positive outcomes.

  4. Sounds like a cross study to diabetes would be in order- that must be one of the common reasons for drinking diet soda.

  5. Personally I am doing a multi decade study of beer. Watch the SK column in QST to see the results. :)
    Good post.

  6. Reminds me of a study done years ago about the elevated cancer risks associated with living near large high tension power lines. Around here, there are lines situated on their own right-of-way that run through suburban neighborhoods; lots of families and so forth. The utility companies are compelled to abate the weed growth on the spacious and cleared paths, where the space is not used for nurseries and other purposes. The "unused" space is hit with herbicides on a regular schedule, but the space is used by the neighborhood kids for dirt biking, and other activities where they become exposed to the toxins aimed at the weeds. Lo and behold, sometimes trouble eventually develops. Meanwhile, half the adjoining neighborhood lies sleeping nightly under electric blankets without any problems reported. This in spite of being hit with much higher fields due to the close proximity to the conductors in the blanket, now being reduced to a centimeter or less. Seems like another case of needing to ban erasers because they cause mistakes.

    1. One that made me laugh was one back in the 80s that said watching black and white TV correlated with some cancers.

      Black and white? If anything, color TV should have been more dangerous - the higher voltages in a color CRT lead to more soft x-rays being emitted. This was in a time when color TVs were the norm and barely cost anything more than B&W. So who was watching B&W? Only the very poorest people. Poverty, unfortunately, correlates with a lot of bad health outcomes and that's really what they found.

  7. In my personal experience, the only fake sweetener that doesn't have apparent side effects is saccharin. I'm allergic to aspartame, which is the most common one used in drinks. Every other one in use in the US causes me intestinal upset. Every one.

    The problem with Aspartame is that it contaminates lots of other foods on the market. If a maker uses it in a sugar-free version of a product, you can be assured that the regular product also contains some amount of it, enough for me to react to it. I suspect that they don't clean equipment when switching between the two versions.

    I had to stop using Sweet-n-Low (saccharin) because they began making their own version of aspartame after the patent expired, and all of their saccharin (including their off-brand versions) is now contaminated.

    The allergy is presented as "dis-hydrotic excema". Hands and feet lose skin, especially fingers and toes. Took 18 months for CHP to get a clear set of prints. A nephew started drinking diet sodas, and developed the problem. He was able to pin it down to the sodas, which revealed the culprit. Doctors saw it for over twenty years, and never figured it out.

    1. While I can probably only barely imagine the impact of your situation, that's not what the article in Stroke was talking about. You have an extreme sensitivity to one chemical, probably microgram quantities, maybe picogram. They're talking about something people consume for years, so they couldn't have the sensitivity you're talking about.

      For the vast majority of people, that doesn't apply and it's safe to consume. Aspartame packages warn people with a rare genetic condition called phenylketonuria to avoid it. They put that warning on there for about 1 in 10,000 people.

    2. What I thought was interesting was the intestinal problems that are fairly common for these various sweeteners. IIRC, it is the most common side-effect for this class of "additive".

      Yeah, it would be somewhat useful to know what my actual reaction dosage level is with aspartame. I have no idea what levels of contamination the other products actually exhibit. Which leads to questions of just how clean a production line may be. Confidence takes a big hit when this sort of situation occurs. Bit frustrating. It's not helped when attempts to contact these companies turns up no practical way to talk to them. THAT raises even more suspicion, that they seem to have no interest in consumer feedback that is not tightly controlled/managed.

    3. Are you sure it's the aspartame and not something else in the packet? The packets have a few other ingredients in there, including some sugars (maltodextrin, I think). The tablets have less stuff in them. Of course, it's hard to recommend doing an experiment where you're going to get hurt.

      My wife has a GI reaction to virtually all sugar alcohols: sorbitol, mannitol, xylitol, and the rest. Rather extreme GI reactions. She's also had it to some of the other things they add to foods that are usually considered pretty benign, like inulin.

      On the other hand, I'm not bothered by them and use sugarless products regularly. There's one or two of them I can definitely tell a threshold. Two teaspoons of xylitol in coffee once a day is fine; four teaspoons within three or four hours is too much.

      The cleanliness of those production lines is very heavily regulated. Not that regulation means a whole lot. It seems every few weeks we hear about some problem on some kind of food production line.

  8. There was a study of eye problems caused by excessive use of computer monitors. The study group was broken down into three groups by age. Something like 18-29, 30-45 and 46-60. Two of the groups had slightly greater risk of needing corrective procedures (glasses or contacts) over time and assumed to have been caused by the computer monitor. The article didn't even discuss the third group although the raw data was in the story. Well as you can imagine the third group saw improvement in vision and that certainly doesn't fit the intended story line. If that was all there was to it that is still quite funny by itself. But if you looked at the raw data the slight increase in required corrective action in the two groups showing that need was when added together the exact number/percentage of the improved vision in the group that seemingly benefitted from the computer usage. How could they not understand that they were merely looking at a random distribution of the result they were searching for and that it was not meaningful. It reminds me of that famous statement an educator (supposedly) once made where he said that even with all the increased effort to improve education 50% of the students were still below average and thus they needed more money to hire more teachers to fix the problem.

  9. This is the sort of thing that drives me crazy about how science is done these days. Weak results, overstated certainty, and "science by Press Release".

    Maybe there's actually something here, but I'm ignoring it until there's something less sensational and more concrete.

    1. As a general rule, ignore any stories in the press that start with "a new study shows" or "scientists say".

      It's safer and less likely to induce stress.