The world of nutrition science can be a very murky place. Experts online and on podcasts routinely cite studies that all seem to conflict with one another. To help shed light on what these studies say, and what they don’t, we started “Study series,” a series of posts breaking down some of the most talked about studies, and giving our take on each one.
First up is the new “pro-carbohydrate” study all the plant-based folks will be raving about. The study is particularly interesting because it sheds new light on how to tell a “good carb” from a “bad carb.”
Let’s dive in shall we?
The new “pro-carb” study
Note: for the science grade here we’re not commenting on the quality of the study. Rather, the score relates to the importance we assign to the findings in our nutritional models. This is based on a lot of metrics including the size of the study, the study design and outputs measured.
Published in the Lancet journal in 2019, this large scale meta-analysis (meaning study authors broke down a bunch of studies to find trends) and systematic review attempts to describe how the source of dietary carbohydrate is key in maintaining health. This is occurring in an environment where more and more studies are recognizing that unrefined high-fiber carbohydrates, like yams and potatoes, are much more desirable than rapidly absorbed, often highly refined equivalents, such as bread and processed cereals.
- Data from 185 prospective studies and 58 clinical trials was included in the analysis.
- Clinical trials had to last at least 4 weeks and compare low/high intakes of various quality carbohydrates.
- Studies looking at individuals with pre-existing conditions were excluded so the population included men, women and children without acute or chronic disease. They could however have prediabetes, mild to moderate high cholesterol, mild to moderate hypertension, or metabolic syndrome.
- Studies were required to measure markers of carbohydrate quality such as dietary fiber, dietary glycaemic index or glycaemic load, and whole grain intake.
- Outcomes were all-cause mortality, coronary heart disease mortality, and stroke mortality; and incidence of coronary heart disease, stroke, type 2 diabetes, and colorectal cancer.
The results – what are good carbs?
The results are summarised nicely by four graphs, which split measures into total fiber consumed (in g/day), whole grain intake (g/day) and glycemic index or how fast a carbohydrate spikes blood sugar (GI units per day). This is important as it helps identify what is the best measure of assessing carbohydrate quality, and identifies that GI may not be the best measure. In the first three graphs we see the individual studies plotted on a relative risk graph. The vertical axis shows the relative risk, anything above 1 is a worse effect, whereas anything below is a better effect. The circles represent each study, and the size represents how influential the study was considered, the bigger the circle the more impact that study had on the eventual outcome. This would be determined by the number of participants, the quality of the study, etc.
There are also two lines on each graph, the red line is probably the most accurate to the real world, as the blue line represents a linear effect (as intake increases so does the effect in a linear fashion), whereas in reality there is probably a point where increased intake has no more effect, or even a negative effect, and this is what the red line attempts to model. Finally, the grey area represents the variation observed at each point. Generally speaking it is best if the line and the variation are both below (or above) 1 for the effect to be significant.
So in this first graph the researchers are looking at total fiber intake and how this relates to four outcomes. In graph A looking at overall mortality there is an immediate improvement with increasing fibre intake which was maintained up to the highest dietary intake recorded of 35g/day. A similar strong effect was seen in graph C which is looking at type 2 diabetes (T2D), and graph D which relates to colorectal cancer risk. Interestingly, while there was some effect in relation to coronary heart disease there was significant variation and potentially a suggestion that a more moderate dose of fiber offered the best effect (although importantly the highest dose was still better than a very low intake).
In numerical form for every 8g increase in fiber eaten the relative risk decreases by 0.07 for all mortality, 0.19 for CHD (take with grain of salt due to large variation), 0.15 for T2D and 0.08 for colorectal cancer.
A similar story as for the previous figure was observed when whole grain intake was measured against the various causes of mortality, with perhaps an improved effect observed for CHD. It is important to note the horizontal axis in graph D is significantly longer than the others with a whole grain intake of over 360g in the highest study (which interestingly showed no improvement).
In numerical form for every 15g increase in whole grain eaten the relative risk decreases by 0.06 for all mortality, 0.07 for CHD, 0.12 for T2D and 0.03 for colorectal cancer.
One of the most interesting findings came when looking at GI intake. As you can see in most of the graphs the line remains flat, with no appreciable difference in risk at any GI intake. In graph C, T2D, there is an increase which occurs with a GI intake over 60, but look at the variation observed. No significant effect here!
In numerical form for every 10 GI unit increase in the relative risk increases by 0.16 for all mortality, 0.09 for CHD, 0.10 for T2D and 0.05 for colorectal cancer. But for all the significance is low due to the huge variation.
In this final figure the authors looked at the effect of the three dietary measures on a variety of metabolic outputs, where available from the clinical trials. Effects to the left of the 0 line represent an improvement whereas those to the right reflect a worse effect. It’s difficult to really pull anything meaningful from this graph. There is a trend towards improved measures (bodyweight, total cholesterol and systolic blood pressure) with increased fiber and grain intake but large effects are missing. So high fiber has a protective effect, but these markers might not be what is driving it.
To take some quotes from the authors:
“Higher intakes of total dietary fiber or whole grains are associated with reduced incidence and mortality from several NCDs. Less useful markers of carbohydrate quality are glycaemic index, glycaemic load, and sources of dietary fiber, in which inconsistent findings or insufficient data provide evidence of low quality or very low quality.”
“These findings, together with the comparisons of clinical outcomes among individuals with different intakes of dietary fiber, suggest that individual adult intakes of total dietary fiber should be no less than 25–29 g per day with additional benefits likely to accrue with higher intakes.”
“The similar protective effects of higher intakes of whole grain foods and of dietary fiber suggest that the beneficial effects of whole grains could be because of their high dietary fiber content.”
“Until evidence is available, it seems appropriate that dietary advice should emphasize the benefits of naturally occurring dietary fiber in whole grains, vegetables, and fruits that have been minimally processed.”
Total carbohydrate intake
The authors of the study did not incorporate total carbohydrate intake as a measurement in their study, which would have helped address some of the issues relating to specific diet types. For example, does it matter whether the fiber intake is achieved by eating a small total number of carbohydrates, which are all fiber-rich, or is it acceptable to eat a larger amount of carbohydrates, some of which may be fiber-poor, as long as the fiber threshold is reached?
Other studies would suggest that a range of overall intakes are acceptable (R), but further breakdown would have proved very insightful.
Non-grain sources of carbohydrate
Many other fruits and vegetables are also rich in fiber, and may contain other nutrients that provide additional benefits. Other studies have shown a similar benefit to increasing fruit and vegetable intake (R), however the contribution of fiber to this effect remains unknown. It is likely that fiber source is not key, and so a mixed fiber intake will prove most beneficial.
Not a criticism of this study as such, but rather the wider field of dietary research. The lack of easily measurable biomarkers means that assessing carbohydrate intake relies on self-reported intake, which is prone to error and misreporting. By incorporating clinical trials into their study the authors address this somewhat, but the lack of strong effect in metabolic markers may be driven by this effect.
After parsing the study, an important question is how do these intakes relate to the average diet, and to those following particular dietary styles?
Well, in European populations it is thought that the average intake of fiber per day is around 20g (R), with an average of about 15g in the U.S. (R). Looking at those levels on the first graph you can see that they are having a beneficial effect compared to very low fiber intakes, but they could easily be improved. There are no real figures out there for those following paleo or ketogenic diets, but for those following a paleo diet there should be no issue incorporating fiber from non-grain based sources. Ketogenic diets may be more troublesome, with the generally very low carbohydrate intake limiting the potential intake of fiber, so this should be considered when undertaking this diet type. Vegans and vegetarians are obviously at the lowest risk here as a plant-rich diet should be rich in fiber by default (R).
Data for wholegrain intake is more difficult to find, but this study from the UK in 2015 suggests that the typical adult will have an intake of around 20g/day (R), and that this is actually declining as well. Again vegans and vegetarians will likely do well here due to a typically high grain intake (R). Figures from ketogenic and paleo diets are again difficult to find, but we can presume a low intake for both. However it is likely that wholegrain intake is actually just a proxy for fiber intake, and it is the beneficial effect to fiber rather than wholegrain that we’re seeing here, and this is something the authors themselves discuss.
The best carbs
Based on the data in this study, the best carbs are those that are fiber rich and that you can digest effectively. The last thing to do is start eating foods that are hard on your stomach because a blog told you to do so. Having said that, if you can incorporate some of these foods, while avoiding sources of carbohydrate that have very little in the way of fiber, like white bread, processed cereals and the like, all the better.
Note, we also believe the source of carb should be tailored to the individual’s ability to deal with spikes in blood sugar, which is something we emphasize in our custom nutrition plan diet types.
- Yams and sweet potatoes (our overall winner)
- Oats (non-glyphosate)
- Avocado (assuming you do not hyper absorb plant sterol)
- Brown rice (assuming no lectin issues)
- Leafy greens
- Strawberries (assuming no histamine issues)
- Natto (also high histamine)
Does this study end the grain free diet movement?
So is this a study that kills off the low carb/low grain diets?
Not as such, those following such diets may need to carefully watch their dietary intake of fiber to ensure they’re receiving the maximal benefit, but as the authors themselves discuss this is perfectly possible from non-grain, natural unrefined source. A target of 25 – 29 g per day of fiber is entirely possible, and demonstrates significant health improvements, with a larger effect seen at 30 g +.
Finally, the variation in effect again provides evidence that individuals will respond differently to various dietary intakes, an effect which likely has a very significant genetic element, which is not assessed as part of this study. One day we will be able to place people precisely on the graphs you see above. Thanks for reading.
The very latest on genetics, nutrition and supplements delivered to your inbox!
Have a question?
We’re experimenting with QA rather than a comments section.