While I am certainly no fan of observational research, observational research isn't the only bad science out there. An article I found on Science Daily a couple of weeks ago read, "Sugar intake not directly related to liver disease". Since I am familiar with non-alcoholic fatty liver disease, and the title of this article completely flies in the face of what I know, I decided to take a look at the original article and see how the science was conducted. As suspected, it was terrible science at it's best...or worst.
One of the more troubling aspects of the way science is conducted is the fact that you can design an experiment that will show pretty much whatever you want it to show, even if it looks like it's a solid study on the outside. By simply designing a study in the proper way, you can show something that likely isn't true. In the study referenced on Science Daily, this was the case. As you can see, the title makes the assertion that sugar has no direct effect on fatty liver disease, but the study didn't show this for a couple of reasons.
To take care of the low hanging fruit, this study was 2 weeks in length. There is no way that fatty liver disease occurs over that short of a period. Certainly you could see some effects of fatty liver disease in 2 weeks, and this study showed that one of the conditions did see some signs of fatty liver disease in a 2 week period. The problem is, you can't use the way they designed the study to rule out the possibility that sugar intake has something to do with fatty liver disease. Let's take a look at how the study was designed.
This study looked at healthy, overweight men and put them in to either a high fructose (Fructose as 25% of calories) or high glucose (Glucose as 25% of calories) group. The men then ate a diet that was isocaloric(Provided exactly the number of calories they needed) for 2 weeks followed by a diet that was hypercaloric(Provided more calories than they needed) for 2 weeks and measured for signs of fat accumulation in the liver. In the isocaloric condition, neither group showed signs of fat accumulation in the liver, but the high fructose group showed increasing insulin resistance, a risk factor for fatty liver disease. In the hypercaloric period, both groups showed signs of fat accumulation in the liver(1).
(Before I go further, I think most people read high fructose and immediately think of high fructose corn syrup. This is a mistake. While high fructose corn syrup does contain more fructose than sucrose(aka table sugar) does, HFCS is 55% fructose/45% glucose while sucrose is 50% fructose/50% glucose. They are essentially identical for our purposes. If you are really worried about high fructose consumption, and you should be, avoid agave syrup like the plague.)
The authors of this study then came to the conclusion that fatty liver disease is a disease of high caloric consumption, not high sugar consumption. The problem is, the study didn't show this, and here's why. There was no true control group in either scenario. Why wasn't there a group of people who ate a sugar free diet in both an isocaloric and a hypercaloric state? If fatty liver disease is only caused by an increase in calories, you would need to show that people who eat no sugar but that eat too many calories also show signs of fatty liver disease. They didn't even bother testing these conditions. In addition, what if I ate isocalorically and ate 40% of my calories from either fructose or glucose? Certainly if there is no direct link between sugar and fatty liver disease the total dosage of sugar shouldn't matter. Again, they didn't test this condition.
This brings up another point that is probably the most important. In the isocaloric and hypocaloric state, there were different total dosages of each type of sugar. It is just as likely that the total dosage of sugar is the culprit and not whether you are in an isocaloric or a hypercaloric state. They did nothing to control for this so they can't rule it out, yet they do anyway. Things like this are supposed to be caught in the peer review process and studies like this shouldn't be published, but they are, this one in the journal Gastroenterology.
The worst thing of all? From a mechanistic standpoint we believe that non-alcoholic fatty liver disease is caused by too much fructose and/or glucose consumption, which overwhelms the liver's ability to use either sugar, leading to the formation of fat droplets. So, the prevailing theory is that fatty liver disease is caused by a dosage of sugar that exceeds the liver's processing capabilities, but the authors didn't feel the need to design the study in a way that controlled for this? I guess it's somewhat enlightening that one of the authors is on the scientific advisory boards of Mars, inc and Coca Cola.
Bad science aside, there are other problems with using this science to approve sugar consumption. Do people who eat higher amounts of sugar tend to eat at calorie maintenance or do they tend to over-consume calories? What happens if you alternate between over-consuming and consuming the appropriate amount of calories daily or weekly while eating sugar? Since non-alcoholic fatty liver disease progresses over the course of years, a study that is 2 weeks in length is not likely to provide you with any sort of applicable knowledge.
As you can see, you need to dig very deep to see if what is being reported is accurate. The problem is not only with the use of observational research in the nutritional sciences, clinical trials can be manipulated to show whatever you want them to show if you design them in the right way.