In this post, I’ll look into the claim that 1 calorie of sugar is less healthy than 1 calorie of other food. For today, I will not look into the unique effects of soda and juice or of adding sugar to your diet without subtracting something else. The results I summarize below are all for constant total calorie intake.
To acknowledge my bias, I expected to find strong evidence for a small effect that sugar increases death from heart diseases and a smaller but still significant effect for all-cause mortality. I came with an open mind about whether added sugar has a different effect than not-so-added sugar.
The evidence is much more jumbled than I expected. I’ll compare the largest, prospective trials I can find. The list is short because I actually found it rather difficult to find studies of sugar from everything instead of just sugar from soda and juice.
Let’s start with the most straightforward result, which will surprise no one. The National Health and Nutrition Examination Survey (NHANES) is one of these projects that has asked a large number of people for decades what they eat and then watches what happens to them. This paper analyzes results from 12,000 people and finds that eating more added sugar will make you die faster from cardiovascular disease (CVD). Their very careful work also calculates a dose-response and finds the harm grows faster than linearly in how much sugar you eat. Viewed the other way around, lowering the fraction of calories you get from added sugar below about 10-12% probably won’t bring you any marginal benefit. (The editorial accompanying the paper actually suggests just getting below 15%) The internet told me a male my age and weight should eat something like 2,100 to 2,400 calories per day. So apparently I can safely eat about 50-75 grams per day of added sugar with no reason to think I’m harming myself. This study did not look into total sugar or all-cause mortality.
This plot is the main result of the NHANES paper on sugar. The solid line is risk of CVD mortality versus percentage of calories from added sugar.
Now things get more confusing. The NIH-AARP Diet and Health Study followed over 350k people for about 13 years. This is more than 25 times the sample size of the study above. This much larger trial sends a mixed message and has gotten less than one tenth as many citations as the NHANES-based paper. With this larger sample size, this paper, published only a month after the NHANES paper, digs into distinctions between men and women, sucrose and fructose, added and total sugar and death from CVD, cancer, or all causes. It total, their main results table tests 48 hypotheses. If all the treatments have no effect, we should expect 2-3 of them to reach p 0.05 by random chance.
Two results stand out with p-values < 0.0001.Total sugar and total fructose are correlated with mortality from all causes in women. In men, the total sugar result disappears but the total fructose result reaches p 0.002. Confusingly, it’s only men in the highest quintile of fructose consumption that are less healthy. In the middle quintiles they’re actually healthier than the men in the lowest quintile of fructose. Directly contradicting the NHANES, the NIH-AARP study finds no impact of added sugar on CVD mortality (in fact, for both men and women more added sugar was insignificantly correlated with lower CVD mortality).
Finally, the Women’s Health Study is a randomized controlled trial of aspirin and Vitamin E in 40,000 women that includes nutritional surveys, giving yet more data for nutritional studies that aren’t interested in aspirin. This paper, plan and simple, finds no connection between any kind of sugar consumption and type 2 diabetes risk.
My verdict: for now, I don’t think we really know what the effect of total sugar on health is. There are almost certainly large prospective trials or meta-analyses I missed. I’ll plan on more posts on this topic in the future.