A recent study out of Princeton University has the high-fructose corn syrup alarmists out in full force. This study compared the effects of high fructose corn syrup (HFCS) to regular table sugar (sucrose), looking at their effects on body weight, body fat, and triglycerides (fats that float around in your blood). The study found that the rats fed HFCS gained more weight and abdominal fat than the rats fed sucrose. This study has strengthened the belief of some people that HFCS is contributing to obesity in our society, and that it is worse than regular sugar. But is it really?

To answer this question, we need to take a close look at this study. The researchers performed 2 experiments. In the first experiment, male rats were divided into 4 groups. Group 1 (the control group) was fed a regular diet. Group 2 was fed the same diet, with the addition of 24-hour access to water sweetened with HFCS. Group 3 had the regular diet with 12-hour access to the HFCS-sweetened water. Group 4 had the regular diet, with 12-hour access to sucrose-sweetened water. The rats were tracked for 8 weeks; weight was measured, along with food, sucrose, and HFCS intake.

You can see the results for experiment 1 in the following chart:

The rats who got HFCS for 12 hours gained significantly more weight than the other 3 groups. At first glance, this would make you believe that HFCS makes you gain more weight than sucrose, even if you are eating the same number of calories. However, there is a problem with these results. Take a look again at the chart above. If the rats fed HFCS for 12 hours gained more weight, why didn’t the rats fed HFCS for 24 hours also gain more weight? They got HFCS for a full 12 hours more, yet didn’t gain more weight. This is a glaring inconsistency in the results…an inconsistency that the researchers never tried to explain.

Rather than some unique effect of HFCS, a more likely explanation is one of chance. Put on your math hat, because we need to talk about some statistics. Researchers use statistics to get an idea of the probability that their results are due to chance. When the scientists run their stats, they get what is known as a P value. The P value tells you the probability that the results are not due to chance. Usually, if the P value is less than 0.05, a scientist will call the results “significant.” In other words, if you did the experiment 100 times, you would only see these results less than 5 times if there wasn’t a true effect.

The above only holds true if you’re doing a single comparison. If you start comparing a bunch of groups all to each other, the probability of a fluke result dramatically increases. The Princeton study is a perfect example. There are 4 groups all being compared to each other. That makes for 6 total comparisons (group 1 to group 2, 1 to 3, 1 to 4, 2 to 3, 2 to 4, and 3 to 4). Each one of these comparisons is being tested against that 5% level. To calculate the probability of a fluke result in this case, we calculate 1 – (0.95x0.95x0.95x0.95x0.95x0.95) = 26%. In other words, there is a 1 in 4 chance that the greater weight gain in the HFCS-fed rats is a fluke. I don’t know about you, but I wouldn’t put too much faith in results that have a 1 in 4 chance of being wrong. There are ways that scientists can adjust for this, but the Princeton researchers didn’t appear to make those adjustments. Thus, it is not surprising that there was a significant result observed in 1 out of the 4 groups…you would expect this to happen based on random chance alone.

In Experiment 2, the researchers divided male rats into 3 groups: 12-hour HFCS, 24-hour HFCS, and control. They tracked the rats for 6 months. Both HFCS-fed groups gained more weight and fat than the control, and also had higher triglycerides. However, the researchers didn’t compare HFCS to sucrose in this group, so this experiment doesn’t’ say anything about whether HFCS is any worse than sucrose. The researchers also didn’t say anything about food intake and whether the HFCS-fed rats ate more than the control rats.

Experiment 2 also featured female rats on one of the 4 diets used in Experiment 1. These rats were tracked for 7 months. The following chart shows the results of the experiment:

The female rats fed HFCS for 24 hours a day gained significantly more weight than the other groups. Now compare these results to the chart for Experiment 1 earlier. Do you see the disparity? In Experiment 1, the rats fed HFCS for 12 hours per day gained the most weight. However, in Experiment 2, the rats fed HFCS for 24 hours per day gained the most weight, and the female rats fed HFCS for 12 hours didn’t gain any more weight than the other groups. Why did the 12-hour group gain the most weight in one experiment, but the 24-hour group gain the most weight in a nearly identical experiment? This is a glaring contradiction in the results, and a problem which the researchers did not discuss. We also have the same statistical problem that we did with Experiment 1. Since there are 6 comparisons, there is a 1 in 4 chance that the results are wrong (and ironically, we have 1 out of the 4 groups showing a significant result). In fact, when we take both experiments combined, we have at least a 50% chance that the results of one of the experiments are wrong. Out of all the comparisons being made, we would expect to see a couple groups show a significant result based on random chance…*and that’s exactly what happened in this study*.

The bottom line is that there is no valid reason for HFCS to be any different than sucrose in the way that it affects your body. They are both nearly identical in their composition, containing roughly half fructose and half glucose. They are both nearly identical in the way they are metabolized by your body. There is no practical difference between the two as far as your body is concerned. Now, I’m not saying that you should go out and consume all the HFCS that you want. The point is that there is nothing uniquely “bad” about HFCS compared to regular sugar. HFCS is not uniquely responsible for weight gain as some people would have you believe.

If you see a product with HFCS and a similar product with natural table sugar, don’t assume the product with natural sugar is any better. Rather than worrying about whether something contains HFCS, you should strive to reduce your intake of all types of added sugar and refined carbohydrates in your diet. It is much more important to look at the big picture; keep your physical activity high, manage your overall food intake, make sure most of your food is from minimally refined sources, and keep your protein intake high. This is what will help you lose weight and keep it off, rather than singling out HFCS in your diet. Don’t let the fructose fear-mongerers fool you.

WOW this was very interesting. Nice that someone finally was honest about how research is swayed to the highest bidder. The best part about the article is that it was very easy for the average person to read. Thanks and keep them coming.

Julie

Hi, Julie, thanks for commenting! I’m happy to hear that the article was easy to read and understandable.

James Krieger

James,

You might also want to review and comment on Dr. Lustig’s commentary about HFCS. His position is contrary to what you state. Since over 400,000 have viewed this video – http://www.youtube.com/watch?v=dBnniua6-oM – it might be worth a critical review on your part.

On my end, I try to avoid all processed sugars.

Looking forward to reading your blogs.

Ken Leebow

http://www.FeedYourHeadDiet.com

Ken Leebow

Hi, Ken,

Thanks for your comment. Alan Aragon has written a very good critique of Dr. Lustig’s lecture.

James

James Krieger

You label Lustig as a High Fructose alarmist by linking to Aragon’s critique in your lede. The fact is that Lustig is very clear that HFCS and sucrose are metabolically equal. He says so a number of times in his famous lecture and it should be underlined by the fact that he references the work of John Yudkin’s published in 1972 as the precursor to his own work. Yudkin’s laid out the problems with sugar before the industrial adoption of HFCS.

Because HFCS brought down the cost of sweetening, the fact that it’s adoption tracks with the obesity epidemic and is a sneaky ingredient in so much industrial “food” Lustig talks about HFCS but his critique is of fructose. Unfortunately many observers of Lustig’s lecture seem to have difficulty avoiding conflating HFCS and fructose.

You do this as well in your final statement:

“. . . rather than singling out HFCS in your diet. Don’t let the fructose fear-mongerers fool you.”

You counsel to avoid sucrose and HFCS equally, yet sucrose is half fructose. A little cognitive dissonance is creeping in.

Marc Brazeau

HFCS is also nearly half glucose and half fructose. There can be more variation in HFCS than sucrose but its within 2 – 4% So its nearly equivalent to sucrose in proportion.

Tim

The most common formulation of HFCS is 55% fructose while sucrose is 50%. It seems to me the difference computes to 10%. Besides that some studies indicate that the ratio is often varied by beverage companies so that a higher ratio in favor of fructose over glucose is used. This is a very significant difference, especially over a long period of time.

dubious

Great article, James.

I will echo Julie’s post – I love the way your break the math and science down into something easily digestible for us layfolk. You don’t dumb it down – you make it accessible.

I have been reading your stuff on the BS Detective for some time now, and will happily follow along with you here.

Cord

Thanks for the compliment, Cord! Glad you like the site

James Krieger

I’m sorry, but much of what you state about p-values here is incorrect – for example, “The P value tells you the probability that the results are not due to chance” is wrong – and you’re using a p of 0.05 when the study only says that p < 0.05. Their p value could easily have been 0.00001.

Kevin

Hi, Kevin,

The P value is the probability of obtaining a test statistic at least as extreme as the one that was observed, assuming the null hypothesis is true. It tells us the chance that a experiment of a “null effect” gives us such a result. In this case, it tells us the chance that we would see a greater weight gain in the HFCS-fed rats, if HFCS has no such effect. In other words, if I were to do 100 identical experiments, a P value of 0.04 would mean that I would expect to see such a result only 4 times, if in fact there is no HFCS effect.

My statement was nothing more than putting the above paragraph into laymen’s terms. With a biological experiment such as this, we are trying to determine how probable it is that we would see these results, if there is no biological effect. A layman would consider that a “chance” effect. Technically, I understand what you are saying, and my statement was over-simplified on purely technical grounds, but remember that these articles are meant to be understood by someone with no background in statistics. And in laymen’s terms, statistics are all about trying to get an idea of how likely an experimental result is random.

When we’re talking about the “decision point” to consider something significant, and we’re dealing with multiple comparisons, then this decision point needs to be adjusted downward according to how many comparisons are being made. Otherwise, the probability of a false positive increases. With the number of comparisons performed in this experiment, the decision point is no longer at the 0.05 level. The researchers did not control for the family-wise error rate. Even if the P value was 0.00001 (which it wasn’t….more on this later), the point is that it’s a P value that hasn’t been adjusted for the fact that multiple comparisons were made. So statements such as “P < 0.05" are meaningless because those P values have not been adjusted to control for the family error rate. Second, the P value isn't 0.00001, because throughout the paper, the researchers made statements of "P < 0.01" for P values smaller than 0.01. So it's obvious that the "P < 0.05" values were between 0.01 and 0.05. Third, my article was to explain how much easier it is to see a significant result when doing unadjusted multiple comparisons. The family error rate is the probability of making one or more false discoveries or Type I errors. With this HFCS experiment, we know that the Type I error rate is no longer 5%.....it is much greater than that, which means, in laymen's terms, we are much more likely to take a chance result and consider it real.

James Krieger