The ketogenic diet (in earlier iterations known as the Atkins diet and the Low Carb High Fat diet) has been vilified for decades by members of the nutrition establishment, because it is pretty much telling people to do the exact opposite of what the establishment is telling them to do. While official diet guidelines recommend cutting down on fat intake, the ketogenic diet instead recommends cutting down on carbohydrate intake. Since cutting down on one of the sources of energy virtually requires increasing intake of the other, the two diets are completely incompatible.
The reason proponents of a low fat diet are so against the ketogenic diet is because they believe that fat, and in particular saturated fat, causes heart disease. But this belief has become increasingly hard to maintain over the last couple of decades. In spite of the fact that populations around the world have been dutifully following advice from nutritionists to cut down on fat intake, they’ve actually gotten fatter and more diabetic. The prevalence of metabolic syndrome (a combination of central obesity, type 2 diabetes, and high blood pressure) has exploded since governments started recommending that people cut down on fat. And metabolic syndrome is one of the leading risk factors for heart disease.
This suggests that the nutrition establishment has got everything terribly wrong. It suggests that it’s a high intake of carbohydrates that is responsible for the current poor metabolic health of western populations, rather than a high fat/saturated fat intake. Of course, the data suggesting this is all observational, and observational data can show correlation, but they cannot prove causation.
So, let’s look at what the randomized controlled trials of the ketogenic diet have to say on the matter. A systematic review and meta-analysis was published in Nutrients in July 2020 that gathered together all the trials of a ketogenic diet. The review was funded by the South Korean government.
Fourteen trials were identified, with a total of only 700 participants. This small number shows how under-researched the ketogenic diet is, likely due to strong biases against the diet among those who fund nutrition research. However, eight of the fourteen trials were carried out in the last decade, suggesting that it is gradually becoming easier to finance this sort of research.
Eight of the trials were looking specifically at people with type 2 diabetes, while the other six were looking at non-diabetics. Of the studies not looking specifically at diabetics, all were looking at people with overweight or obesity, so all twelve were studying people who were metabolically unhealthy in one way or another at the beginning of the trials.
The study with the shortest duration lasted two hours(!) while the study with the longest duration lasted two years. Most of the trials lasted for at least a few months. The smallest individual study had 11 participants, while the largest had 146 participants.
Twelve of the trials were comparing a ketogenic diet with a low fat and/or a low calorie diet. Two compared a ketogenic diet with letting people go on eating whatever they were eating to begin with. So most of the studies were comparing two different types active diet modification, rather than comparing a ketogenic diet with just continuing on as usual. This is worth remembering as we get to the conclusions.
The reviewers determined that the overall risk of biases skewing the results was low. The size of the effect of a ketogenic on different parameters was measured using a statistical technique called Standard Mean Difference (SMD). This is a technique that is useful when different studies are measuring the same thing in different ways, because it converts all the different measurements in to one shared format, which allows for meta-analysis. An SMD of zero means that there is no difference whatsoever between groups. An SMD of 0,2 is usually considered small, while 0,5 is considered moderate, and 0,8 is considered large. The disadvantage with converting results to SMD is that they become more abstract and thereby harder to interpret in terms of real world outcomes.
Anyway, let’s get to the results.
Among type 2 diabetics, a ketogenic diet resulted in a reduction in HbA1c (a measure of average blood sugar levels over the last few months) of 0,62 SMD, an effect that was highly statistically significant (p-value 0,001), when compared with a low fat or calorie restricted diet.
Among non-diabetics, the ketogenic diet had no statistically significant effect on HbA1c. This is largely as we would expect, since non-diabetics have functioning homeostatic mechanisms that are able to maintain blood sugar within relatively strict limits regardless of what they eat.
The big effect on HbA1c in type 2 diabetics is important, and aligns with other evidence that a ketogenic diet can be used to effectively push type 2 diabetics in to remission. Back before the advent of medications to treat type 2 diabetes, the main treatment used by doctors was a low carbohydrate diet (a fact which most modern day doctors are completely unaware of). So it makes sense that a ketogenic diet would have this large effect on HbA1c.
People on a ketogenic diet lost more weight than people on low-fat/low-calorie diets, with an SMD for weight loss of 0,46. The statistical significant was pretty weak (p-value 0,03), but 7 out of 8 trials reporting data on weight change showed that the ketogenic diet performed better than the low-fat/low-calorie diet, which suggests that the effect is real. The diabetics benefited more than the non-diabetics, with an average weight loss over the course of the studies of 7,8 kilograms among the diabetics, as compared to 3,8 kilograms among the non-diabetics.
People on a ketogenic diet also did better in terms of reductions in waist circumference than people on low-fat/low-calorie diets, with four out of the five studies that measured waist circumference favoring the ketogenic diet. However, the effect was not statistically significant (p-value 0,09). This could mean that the effect isn’t real, but more likely means that there were too few participants to show a significant difference. In other words, the studies were statistically underpowered for this end point.
More importantly, however, there is no signal that a ketogenic diet results in less waist reduction than the low-fat/calorie-restricted diet. Waist reduction is particularly important, because it is the fat that accumulates in the abdomen that most strongly correlates with heart disease and diabetes. Subcutaneous fat that is more evenly spread out around the body doesn’t show anywhere near as strong a correlation.
Ok, so far we know that a ketogenic diet appears to be more effective than the conventionally recommended low-fat/low-calorie diet at both lowering blood sugar and inducing weight loss. But the main concern from establishment nutritionists about the ketogenic diet has always been that it causes heart disease, based on data that suggest it increases on LDL (“bad cholesterol”).
Before we look in to whether the ketogenic diet does in fact raise LDL, let’s think about whether this is something we need to be concerned about or not. Although LDL correlates somewhat with heart disease in observational studies, the evidence from randomized trials strongly suggests that the relationship between LDL and heart disease is not causal. A systematic review published in BMJ Evidence Based Medicine last year could not find any correlation between the amount of LDL lowering achieved with various LDL lowering drugs and the effect on either cardiovascular events or overall mortality. This suggests that we don’t need to be as concerned about LDL as many people believe.
And apart from LDL, there’s also HDL (“good cholesterol”) and there are triglycerides (the amount of fat in the blood stream). All three are considered to be risk factors for heart disease. In general, LDL and triglycerides correlate with increased risk of heart disease, so you want them to be low, and HDL correlates with decreased risk of heart disease, so you want it to be high.
Anyway, let’s see what effect a ketogenic diet has on these markers for heart diease.
Among diabetics, the ketogenic diet had no effect on LDL. It had a modest positive effect on HDL, with the SMD increasing by 0,31. The result was statistically significant. (p-value 0,005). The ketogenic diet also had a negative effect on triglycerides, with the SMD decreasing by 0,45 (p-value 0,01). In other words, in diabetics, the ketogenic diet had no impact on LDL whatsoever, and a beneficial impact on both HDL and triglycerides.
So, for type 2 diabetics, a ketogenic diet appears to protect against heart disease, at least if you go by the traditional risk factors. All the risk factors are either unaffected or affected in a way that correlates with decreased risk. Among non-diabetics, however, the picture isn’t quite so rosy. In this group LDL did increase, with a modest but statistically significant SMD of 0,35 (p-value 0,009), while HDL and triglycerides were unchanged.
So, how shall we interpret this? Does a ketogenic diet decrease your risk of heart disease if you’re diabetic, but increase it if you’re not?
It’s possible. But it seems unlikely that the same dietary intervention could have a beneficial effect on risk of heart disease in diabetics but a harmful effect on risk of heart disease in non-diabetics. Let’s move on and look at the final major risk factor for heart disease, blood pressure. The ketogenic diet did not appear to have any effect on blood pressure, either up or down, as compared to the low fat and calorie restricted diets. This was true in both diabetics and non-diabetics.
Ok, so those are all the results. What can we conclude?
The ketogenic diet appears to be more effective at achieving weight loss than the low fat and calorie restricted diets. Considering how much suffering is involved in the calorie restricted diet, this should be welcome news for people everywhere who want to lose weight without having to feel like they’re starving to death.
In terms of safety, the picture is a little more complex. For type 2 diabetics, the ketogenic diet appears to be more or less like a magic bullet, resulting in them becoming noticeably slimmer, while at the same time pushing their diabetes in to remission, and also decreasing their risk of heart disease. Pharmaceutical companies won’t like it if this gets out, it could put a serious dent in their profits!
For non-diabetics, however, there is one concerning fact, and that is that LDL goes up. Why it does this in non-diabetics but not in diabetics, I have no idea. Could it be a statistical artifact, i.e. not real? Yes, it could. The result is based on just four studies, with a total of 235 people. But all four studies show a similar increase, suggesting that the effect is real. As mentioned, however, LDL only correlates with heart disease in observational studies, and not at all in randomized clinical trials of LDL lowering drugs, so it is somewhat questionable how important it actually is to consider as a risk factor. And people over the age of 60 tend to live longer if they have a high LDL rather than a low LDL, so for elderly people an increased LDL might actually be a good thing.
In other words, the overall evidence to suggest that a ketogenic diet isn’t safe is very weak. It is based entirely on one surrogate marker, LDL, and not on any hard outcome data. Until someone does a longitudinal study that follows people on a ketogenic diet over an extended period of time to see what happens, and compares them with a matched cohort, we won’t know for certain.
Things can often seem to be one way when we look at surrogate markers, and then when we do a proper study, looking at hard outcomes, things turn out to be very different. A good example of this is the Women’s Health Initiative study, a massive randomized trial with 49,000 women that was meant to show that a low fat diet decreases risk of heart disease, but which failed to find any benefit whatsoever. And if a low fat diet doesn’t decrease risk of heart disease, then it’s quite reasonable to think that a high fat diet probably doesn’t increase risk of heart disease.
One last thing worth repeating is that the systematic review discussed here was mostly comparing a ketogenic diet with a low-fat/low-calorie diet. Other studies have, for example, found that the ketogenic diet is effective at lowering blood pressure when compared with a standard western diet. The reason no effect was found here was likely because it was being compared to the low-fat/low-calorie diet, which is also effective at lowering blood pressure. Any diet that causes you to lose weight will cause your blood pressure to drop.
So, to sum up: the ketogenic diet is highly effective as a treatment for type 2 diabetes and as a weight loss strategy. The fact that it modestly raises LDL in non-diabetics is somewhat concerning, but it is questionable whether this actually means anything in terms of real world outcomes.