Are Low-Carb Diets Underutilised for Type 2 Diabetes?

Low Carb Meal

As a dietitian, I have held back on talking publicly about a lot of things I strongly believe in. Deep down, I think this is based on being concerned with how other dietitians view me.

I know that I am not alone in this and a lot of dietitians feel this way as well. My recent article on why I’m not a non-diet dietitian made that point clear to me. The inflow of private messages I got from people sharing their stories and why they weren’t speaking up on the topic was incredible.

This is exactly why I am writing about low-carb diets for Type 2 Diabetes (referred to simply as diabetes for the rest of the post). It’s another topic that I personally have avoided speaking up about. Maybe other dietitians feel the same way?

My interpretation of the current evidence is that lower carbohydrate diets are superior to the traditional method (based on carb exchanges + a few other factors) of dietary intervention to manage diabetes. There are a lot of variables in this and it is not a one-size-fits-all approach.

I feel that a lot of dietitians are so afraid of being “wrong” that we may delay being more open towards an effective approach until it is universally accepted.

Interestingly, Diabetes UK recently put out nutrition guidelines for diabetes which are leaning more towards reducing carbohydrate intake than before. In my eyes, these guidelines are a step towards it becoming more accepted.

The Evidence for Low Carb Diets

There are a few research papers that stand out to me, so I will go through them individually.

1) CSIRO Low-Carb Diet 12-month study:


CSIRO Low Carb Diet Study Results
This is the least controversial study on the topic that I can think of. They utilised <50g carbs per day at the start and then increased it to <70g per day by the end. This was compared with a more traditional diabetes diet that was matched for energy intake.

Matching for energy intake is crucial in a study like this, because if one group loses more weight (particularly body fat) it may increase their insulin sensitivity, which is a major variable.

Another reason I like this study is that if you look on the Diabetes NSW website, they have a statement on this study which mentions that it “shows some exciting potential for well-balanced lower carbohydrate diets as one possible intervention for people with type 2 diabetes.”

Emphasis on well-balanced, which is much more feasible with a 50-70g per day carb intake than it is on <20g carbs.

On average, weight decreased by 10kg in both groups. HbA1C decreased by 1% in both groups and fasting BGL’s decreased by 0.7mmol/L in the low carb group and 1.5mmol/L. The HbA1C and Fasting BGL changes are misleading if looked at in isolation since the medication was reduced significantly more in the low-carb group.

In the low carb group, the participants reduced their diabetes medication by 40%. That is a big result. I got that number from their book since the study uses a different unit of measurement for medication changes, which is more difficult to interpret. I estimated that the medication change in the higher-carb group was a <20% reduction.

Either way, since there was a larger reduction in the low-carb group, while also reducing HbA1C the same amount, it was more effective. The fasting BGL reduction being smaller is hard to explain succinctly, but it is related to the greater control of BGL’s during the day leading to fewer fluctuations. That way they could reduce the medication even more, although it meant the fasting BGL would be slightly higher.

Protein was higher in the low carb group, which is a theoretical flaw. There was 28% of energy coming from protein in that group and only 17% coming from protein in the higher-carb group. It potentially would have been better theoretically if the protein was matched and only carb/fat intake was adjusted.

From a practical standpoint though, if you tell people to reduce their carb intake, they typically naturally increase their protein intake to compensate. Based on that, I would say it was fair to perform the study in this manner, but worth noting that protein could also be a major variable.

Overall both groups did well and made significant improvements. This then leads to the conclusion that the best approach would be the one you can stick to.

Diabetes NSW noted that “previous studies have indicated that restrictive diets can be difficult to adhere to.” This is a critical observation and potentially a major factor in why lower-carb diets aren’t the first option recommended.

The most interesting aspect of the study for me is that both groups had similar completion rates (LC diet: 71%; HC diet: 65%). Technically, more people completed the 12 months in the low carb group than in the high carb group.

The fact that 71% of people stuck close enough to the low-carb diet to be included in the study results absolutely shocks me, to be honest.

Maybe people are more compliant with a strict guide to follow? Is it easier to follow because it is so different from their original diet?

As dietitians, a lot of us try to keep things similar to people’s current diets and lifestyles and just make small changes, but perhaps that is harder than big changes when people have motivation.

There is a lot to think about from that result. Another point to ponder though… The CSIRO had a financial interest and a book to sell, so it was in their best interest to have more people complete the low carb diet. Not a massive flaw in such a quality study, but one worth considering.

2) A low-carbohydrate, ketogenic diet to treat type 2 diabetes

Ketogenic diet HbA1C resultsHere is where things start getting more controversial. It is difficult to have a balanced diet in ketosis and it is common for there to be a lot of micronutrient inadequacies.

This study was smaller and only had 28 participants for 16 weeks, without a control group. Out of those 28 participants, 21 people finished the study. HbA1C decreased from an average of 7.4% down to 6.3%. Average body weight decreased 6.6% over this timeframe as well.

Even if you only looked at HbA1C, that is a solid result. It becomes an incredible result once you factor in that diabetes medications were discontinued in 7 participants, reduced in 10 participants, and unchanged in 4 participants.

That means that 7/21 people (or 7/28 if you want to be particular) completely got off medications in 16 weeks and 17/21 people significantly reduced their medication requirements while improving their control of diabetes at the same time. Those other 4 participants likely contributed to the improvements in HbA1C as well, so I would wager even they got benefits.

Although there are a lot of flaws in this study (e.g. no control group, short time-frame, 95% men) this is still a study worthwhile factoring in as part of the overall evidence.

One key line of this study that is vital if you were to implement a ketogenic diet is that “on the day the diet was initiated, diabetes medications were reduced – generally, insulin doses were halved, and sulfonylurea doses were halved or discontinued.”

If this part isn’t factored in, it is pretty much-guaranteeing hypoglycaemia. Potentially this is a barrier to dietitians recommending this style of diet, since it means the medical team needs to be on board to adjust insulin and medication dosages accordingly.

3) The Diabetes Remission Clinical Trial (DiRECT)

DiRECT Trial Study Results

And now for the most controversial one: 12-20weeks on a ~850kcal/day diet. The study went for a full year and after that 850kcal diet (using meal replacement shakes, soups and non-starchy vegetables), there were food reintroduction and weight-loss maintenance phases. This was compared to a more traditional control group, similar to how the CSIRO diet study was set up.

At 12 months 24% of participants had lost >15kg in the intervention group and nobody in the control group had achieved that level of weight-loss. Diabetes remission (HbA1C <6.5% with at least 2 months without medication) was achieved in 46% of participants in the intervention group and 4% of participants in the control group. Of the people who lost 15kg or more, 86% went into remission.

Going back to the problem of the medication from the ketosis study, this study involved a drastic approach as well. The wording used was “withdrawal of antidiabetic and antihypertensive drugs” at the start of the trial. 

This is an approach I have no desire to put in place with clients, but it is still interesting nonetheless. Almost 50% of people went into remission from diabetes. Most people with diabetes would absolutely love to no longer need medications or worry about their BGL’s. It would be fascinating to see the long-term results though. What happens to these participants in 5-10 years’ time?

For those who are also interested in seeing how this affected the participants long-term, there are follow-ups to this study that can be found here.

An Alternative Approach

1) A Low-Fat Vegan Diet in Individuals with Type 2 Diabetes

Low fat vegan diet for type 2 diabetes
This study was an eye-opener for me since my intuition was that a higher carb vegan diet would be less effective for managing diabetes. It compared a low-fat vegan diet with recommendations based on the American Diabetes Association Guidelines from 2003.

There were 99 participants and it went for 72-weeks, which is the longest study included in this post. Participants lost 4.4kg in the vegan group and 3kg in the conventional group. HbA1C changes were -0.34% and -0.14% for the vegan and conventional groups respectively. Once again it is hard to notice the real difference until medication changes are accounted for.

The authors summarised “in an analysis controlling for medication changes, a low-fat vegan diet appeared to improve glycemia and plasma lipids more than the conventional diabetes diet recommendations.”

I’m not digging into the details on this to discuss the changes in medications. All I wanted to highlight was that a vegan diet was more effective than recommending what the American Guidelines proposed at the time, based on the compliance of the individuals in the study.

This diet was higher in carbs than the traditional guidelines, meaning that only looking at carbohydrate intake would be overly simplistic. The key point of this is to stay open-minded and notice that there are MANY ways to get results and improve health outcomes with diabetes.

If Low-Carb Diets Are Effective, Why Aren’t They Recommended More Often?

There are many factors in this, but I’ll focus on 3 components that I think are major barriers:


Decreasing carbohydrates is so effective at reducing BGL’s that right off the bat you need to make medication changes if the reduction in carbs is drastic enough. If you don’t adjust insulin levels and then you reduce carbohydrate consumption, there is a risk of having a hypo.

It seems simple: why don’t you just reduce insulin/medication? The reason it is more complicated is that would take other members of the medical team to do this appropriately. As a dietitian, I never tell a client to change their medications. The most I do is advocate for changes.

Unfortunately, it can be hard to communicate with other members of the medical team. Everybody is busy. The most common way is to send a letter, which you will likely receive no response to.

Then that professional will make their judgement and pass it onto the client. What happens if you recommend reducing insulin, so you can start a low-carb diet, but the health professional is against the idea because they aren’t aware of the research or the context that it will be used in. It is a complex thing to communicate, particularly through an individual letter.

What if you get past that first barrier and the medications are changed, but the client doesn’t stick to the diet? Then their BGL’s and HbA1C will increase. The first rule of being a dietitian is to “do no harm.” This is potentially a reason why people avoid going down this route, even if they believe it is more effective.

It’s a safer approach to slowly assist a client to reduce their carbohydrate intake and weight and in turn have their specialist slowly reduce their medication. This would reduce the chances of hypoglycaemia and if their carbohydrate intake is consistent, it would reduce their chances of hyperglycaemia.

Beliefs About Compliance:

As Diabetes NSW stated, restrictive diets have been shown to be hard to follow. This is still a concern, even after the CSIRO’s study showed that slightly more people complied with the low carb diet.

I’m in the middle on this one. With clients the first strategy I generally trial is based on a more traditional approach, unless they have a specific preference.

I often use the carbohydrate exchange system and generally specify <45g carbs at meals and <30g carbs at mid meals. I individualise it to the person, but that is the standard model I use.

If somebody chose to take full advantage of this, it could come out as ~225g of carbohydrates, which is relatively high for somebody with diabetes, but most people won’t end up anywhere near that high.

Mostly, the goal is to put a limit on individual meals and snacks, so that there are no individual points at which BGL’s rise exceptionally high.

The other goal is to ensure that they keep their intake relatively consistent so that it prevents hypos if the insulin dosage remains unchanged. In my opinion, this is a very flexible approach and if somebody struggles with this, it would be extremely difficult for them to do a low-carb diet anyway.

Although this approach is flexible, perhaps that is a flaw itself. For example, people can have pasta/rice still if it is <45g of carbs (aka <1 cup of cooked pasta/rice) and there are no other carb sources in the meal.

Potentially the flexible nature of the system opens the door for people in such a way that it leads to them consuming more than the specified amount of carbs. It could affect their mindset in a way that becomes a barrier to them making significant changes.

For example, since they can still have pasta, maybe a family member cooks Spaghetti Bolognese as normal, yet they end up with >1 cup of cooked pasta on their plate. If they were on a low-carb diet, it would be simpler to avoid the pasta completely.

Potentially the simple rules of a low-carb diet make it just as easy to follow. Intuitively it seems harder, but perhaps the difference in diet style is beneficial from a compliance standpoint.

Both approaches work, so it comes down to what the individual can stick with. The hard part is identifying which approach would be easier to comply with.

Lack of Knowledge of Health Professionals:

There is so much research out there and it is impossible to read every single study that comes out. Dietitians are the experts in nutrition, so I don’t expect other health professionals to be aware of the pros and cons of lower-carb diets for diabetes.

Among dietitians though, it is still not uncommon for people to believe that low-carb diets are ineffective for the management of diabetes. There are dietitians who specialise in diabetes who I have heard definitively claim that low-carb diets “don’t work for managing diabetes.”

Potentially this is related to confirmation bias. Naturally, we seek out evidence to support our beliefs and avoid evidence that is counter to our beliefs.

If somebody learnt about an approach to managing diabetes, it theoretically made sense and it got results with clients, would they be as interested in looking into alternative options? This is exactly why I included the vegan study in this post, since it is the opposite of a low-carb diet, yet still got results.

The Evidence Is There to Support Low-Carb Diets As An Option

Going back to why I wrote this: I am hoping that somebody who is on the fence, reads this and it prompts them to also be more comfortable with the idea of talking about low-carb diets and the evidence. It is still debatable whether low-carb diets are the most effective strategy, but I think the evidence is at a point now where it should be accepted as one of many potential options available.

Share this post: