The Diabetes Code: Prevent and Reverse Type 2 Diabetes Naturally

Overall score

Scientific accuracy

Reference accuracy

Healthfulness

How hard would it be to apply the book's advice? Fairly difficult

Book published in 2018

Published by Greystone Books

First Edition, Paperback

Review posted September 1, 2024

Primary reviewer: Shaun Ward

Peer reviewer: Seth Yoder

Table of contents

1. Introduction
2. Scientific accuracy
3. Reference accuracy
4. Healthfulness
5. Most unusual claim
6. Other
7. Conclusion
8. Updates

Claim 1

“Medications do not help to reverse type 2 diabetes”

Supporting quote(s) and page number(s)

[Page xv] “Since weight loss is the key to reversing type 2 diabetes, medications don’t help.”

[Page xv] “We’ve avoided facing an inconvenient truth: drugs won’t cure a dietary disease”

[Page xiv] “We’ve intuitively sensed this truth all along. But only diet and lifestyle changes—not medications—will reverse this disease, simply because type 2 diabetes is largely a dietary disease.”

[Page xiv] “The most important determinant, of course, is weight loss. Most of the medications used to treat type 2 diabetes do not cause weight loss. Quite the contrary…”

[Page xix] “We have overlooked a singular truth: you can’t use drugs to cure a dietary disease.”

[Page 140] “Lowering blood glucose with medications, as opposed to diet and lifestyle, is not necessarily beneficial.”

[Page 150] “The classic medical treatment, which relies almost exclusively on pharmaceuticals to reduce blood glucose, can therefore best be described as how not to treat type 2 diabetes. By contrast, newer agents, which can reduce both blood glucose and insulin levels, show proven benefits to reduce heart and kidney complications of type 2 diabetes. Nevertheless, these medications, while an important step forward, are clearly not the answer; they do not reverse the root cause of type 2 diabetes—our diet.”

[Page 206] “But here’s what is important. A dietary disease requires a dietary treatment.”

[Page 206] In Figure 15.4, drugs are put in the ‘No cure’ category for type 2 diabetes.

Criterion 1.1. How well is the claim supported by current evidence?

1 out of 4

In The Diabetes Code, there are numerous mentions that medications do not and cannot help to reverse T2D, as it is a “dietary disease”. We scored this item a 1, meaning that relevant evidence is intrinsically convincing but partially undermines the author’s claim.

For the sake of reviewing this claim, we decided to treat “reverse T2D” as evidence for remission, which has become more common terminology since the book’s publication in 2018. As defined by the American Diabetes Association in 2021, sufficient changes in blood glucose and medication use can achieve T2D remission; specifically, HbA1c <6.5% (<48 mmol/mol) for at least 3 months after stopping glucose-lowering drugs.

Although it is outside the scope of this review to address all T2D medications, particular attention should be drawn to two medications that clearly increase the likelihood of achieving HbA1c <6.5% (48 mmol/mol) compared with lifestyle-only (diet and exercise) interventions. Both of these medications, tirzepatide and semaglutide, activate the GLP-1 receptor and have been approved by the FDA for the treatment of T2D as an adjunct to diet and exercise (tirzepatide also activates the GIP receptor). The reason for focusing on these medications is that their benefits are mostly a result of modifying eating behavior (reducing appetite and the desire to eat) which directly opposes the book’s claim that medications cannot help to reverse T2D as it is a dietary disease.

The strongest evidence for semaglutide helping with T2D remission to the STEP 2 trial. This was a double-blind, randomized, placebo-controlled trial that assessed weight loss and glycemic control in 1,210 adults with overweight or obese and T2D. Participants received either 2.4 mg semaglutide, 1.0 mg semaglutide, or placebo for 68 weeks. At the end of the trial, HbA1c <6.5% was achieved in more than 60% of participants in both semaglutide groups, compared with only 15.5% in the placebo group—despite all groups receiving a reduced-energy diet and increasing physical activity.

The results of STEP 2 aligned with an earlier 30-week double-blind trial in 388 patients with T2D, that were randomized to semaglutide 0.5 mg or 1 mg once weekly or placebo. At least 70% of patients in both semaglutide groups achieved HbA1c <7.0% at week 30, compared with only 28% in the placebo group. Very similar results were found in three other Phase 3 clinical trials that compared semaglutide with other T2D medications, rather than placebo. They each found that over 50% of participants randomly assigned to semaglutide treatment achieved HbA1c <7.0% at 30–56 weeks, compared with no more than 40% of participants in the control groups using either insulin glargine, exanitide, or sitagliptin. We acknowledge that all these studies other than STEP-2 used a HbA1c cut-off of <7.0%, not <6.5%, but the high rates of overall HbA1c lowering with semaglutide supports the benefits of T2D medication.

To complement the results of these semaglutide studies, there is strong evidence, perhaps even stronger, that tirzapetide medication can help to achieve T2D remission. In the five phase 3 or 4 clinical trials known as SURPASS (1–5), which each included three tirzapetide treatment groups (5mg, 10mg, 15mg), significantly more patients in all tirzapatide groups achieved the HbA1c targets of <7.0%, <6.5%, and <5.7% compared with placebo or active drug comparators. Specifically, at least 66% of participants in every tirzapetide group achieved HbA1c <6.5% at the end of the trial (which ranged from 40–52 weeks). This increased to at least 76% of participants in the 10 mg treatment groups, and at least 81% of participants in the 15 mg treatment groups. In contrast, the proportion of participants that achieved HbA1c <6.5% in placebo groups (SURPASS-1 and SURPASS-5) did not exceed 17%.

The limitation to current studies of semaglutide and tirzapetide is that they rarely assess rates of T2D remission when the therapies are stopped—after all, T2D remission is only achieved when HbA1c remains low without glucose-lowering therapy. Therefore, we cannot say these patients are bound to achieve sustained T2D remission when medication stops. In fact, since these medications achieve lower HbA1c primarily via their weight loss effects (often exceeding 10% body weight), and substantial weight regain has been reported when these therapies are stopped, they are unlikely to achieve sustained remission independently. For example, one year after the withdrawal of semaglutide and lifestyle intervention, participants regained approximately two‐thirds of their lost body weight. Similarly, with tirzapetide, cessation at 36 weeks led to regaining approximately half of lost body weight at week 88. We can only hypothesize how the weight regain affects the rates of T2D remission, but it is plausible that lifestyle changes that prevent weight regain are necessary to sustain T2D remission.

Still, the possibility for weight regain should not detract from the benefits of T2D medication. Evidence showing success with T2D remission without medication or surgery is lacking. The DiRECT and VIRTA health studies provide the strongest support for T2D remission without medication or surgery, yet still present the clear limitations of lifestyle-only interventions. The DiRECT study comprised an intervention of total diet replacement and intensive professional support, reporting T2D remission rates of 46% at 12 months. This is impressive and only slightly inferior to the results of semaglutide and tirzapetide interventions, as previously stated, but most people with T2D still did not achieve remission. More importantly, all subgroups of DiRECT participants (control, intervention, extension, no extension) with 5-year data had less than a 20% T2D remission rate despite the initially promising results. Similar conclusions apply to the VIRTA health study, which compared the effect of a low carbohydrate diet and professional lifestyle support versus usual diabetes care in 349 patients with T2DM. At 1 year, 25.5% of participants achieved T2D remission, declining to 17.6% at 2 years.

Both the DiRECT and VIRTA studies provide evidence that most people with T2D supported with lifestyle-only interventions do not achieve remission, and that achieving T2D remission is temporary for many people that do. For this reason, new FDA-approved medications targeting HbA1c reduction via weight loss are helpful therapeutic options that can increase the likelihood of T2D remission. This position directly contradicts the claim in The Diabetes Code; however, we acknowledge that many trials supporting the use of semaglutide and tirzapatide were published after this book, in 2018. The claim that medications do not help to reverse type 2 diabetes would have been easier to support at the time of the book’s publication.

Overall (average) score for claim 1

1 out of 4

Claim 2

“Type 2 diabetes is caused by eating too much sugar”

Supporting quote(s) and page number(s)

[Page xv] “FACT: TYPE 2 DIABETES IS CAUSED BY TOO MUCH SUGAR”

[Page xv] “AT ITS VERY core, type 2 diabetes can be understood as a disease caused by too much insulin, which our bodies secrete when we eat too much sugar.”

[Page xx] “ONCE WE UNDERSTAND that type 2 diabetes is simply too much sugar in the body, the solution becomes obvious. Get rid of the sugar. Don’t hide it away. Get rid of it. There are really only two ways to accomplish this.

1. Put less sugar in.
2. Burn off remaining sugar.

That’s it. That’s all we need to do.”

[Page 101] “DATA FROM MORE than 175 nations links sugar intake inextricably to diabetes, independent of obesity.”

[Page 102] “Diabetes correlates strongly to sugar, not other sources of calories.”

[Page 116] “Metabolic syndrome, of which obesity and type 2 diabetes are a key part, are ultimately caused by—you guessed it—too much sugar.”

[Page 116] “All the conditions we thought were problems—obesity, insulin resistance, and beta cell dysfunction—are actually the body’s solutions to a single root cause—too much sugar. ”

[Page 117] “And when we understand the root cause, the answer to all of these problems—and to type 2 diabetes—becomes immediately obvious. We need to get rid of the sugar and lower insulin.”

Criterion 1.1. How well is the claim supported by current evidence?

1 out of 4

We scored this item a 1, meaning that relevant evidence is intrinsically convincing and partially undermines the author’s claim. The explanation of what causes T2D in The Diabetes Code was confusing and, at times, contradictory. In the opening chapter, the key to reversing T2D was described as weight loss, with barely any mention of dietary sugar or insulin levels. Yet in the following chapters, the cause of T2D was stated as too much sugar, leading to high insulin levels and T2D, with barely any mention of weight gain. We are therefore unsure how the author factors in body weight as a cause of T2D, although we know from his last book, The Obesity Code, that he argues sugar and carbohydrates are the cause of obesity too.

If we speak directly to the claim that T2D is mainly caused by eating too much sugar, there is only weak evidence to support this, with perhaps stronger evidence opposing the claim.

First, trends in dietary sugar intake and the prevalence of T2D do not appear to strongly correlate. For example, in the United States, there is evidence of significant declines in sugar intake among all age groups over the years 1999 to 2018, despite total diabetes rates increasing from 10% to 13% of the adult population from 2001 to 2020. Specifically, NHANES data in over 50,000 individuals showed that total daily grams of dietary sugar intake decreased by approximately 17–27% from 2003 to 2016, and sugar intake as a proportion of total energy intake dropped by 5–13%. (largely driven by reduced consumption of added sugars from sugar-sweetened beverages.)

Second, if we look at studies that follow people over multiple years (prospective observational studies), most large studies do not report significant associations between total sugar intake and the risk of T2D. A systematic review and meta-analysis of 15 prospective cohort studies compared the highest intakes with the lowest intake of total sugars (median 137g vs 65g), fructose (median 35.2g vs 9.7g) and sucrose (median 78g vs 25.8g) with the incidence of T2D over an average 12 years follow-up. No significant association between total sugars or fructose with T2D were reported, and sucrose was associated with a significant 11% reduced risk of T2D. The researchers noted serious inconsistency between studies and imprecision in the pooled estimates, however, which is a limitation.

Similar issues with low study quality were reported with an umbrella review of meta-analyses of prospective observational studies. It found no significant association between T2D and each 100 gram increase in dietary sugar intake per day, and a significant reduction in the risk of T2D per 50 gram increase in sucrose intake per day, with the quality of evidence deemed either low or very low. Most of the meta-analyses included fewer than five studies and had high heterogeneity.

Although observational studies like these are not usually considered strong evidence, if sugar were the main cause of T2D, the effect should be strong and consistent. Therefore it should show up in these types of studies.

One of the reasons why there can be disagreement between studies is because sugar is contained in foods with varying and possibly divergent health effects. For example, there is higher certainty that fruit and dairy, major sources of sugar, are associated with small, significant reductions in the risk of T2D. Whereas each daily serving of sugar-sweetened beverage consumption has been significantly associated with a 27% increase in the risk of T2D. These differences may explain why sucrose is the type of sugar that often associates with a reduced risk of T2D, as it is more concentrated in solid foods (grains, fruit, and dairy) than sugar-sweetened beverages. Therefore, there probably exists sugar-rich food sources that increase the risk of T2D and others that decrease the risk, which may explain the relatively neutral effect when all food sources are combined to calculate ‘total sugar intake’.

Another reason for the disagreement between observational studies is because some control for (i.e. eliminate the effect of) body mass index (BMI) and/or total energy intake and others do not. For example, while some studies that do not adjust for energy intake/BMI will report an association between sugar-sweetened beverages and T2D, other studies that do adjust might not. Therefore, although we agree that dietary sugars can increase the risk of T2D, the effect may only be apparent when this nutrient causes an increase in calorie intake and body fatness.

Supporting the importance of dietary sugars impact on total energy intake are controlled intervention trials, which look at short-term changes in glycemic control rather than long-term risk of T2D. A great example is a meta-analysis of controlled intervention trials that assessed fructose intake on glycaemic control, while stratifying the results based on how the studies controlled for energy intake: substitution studies (sugars exchanged for energy-matched macronutrients); addition studies (excess energy from sugars added to diets); subtraction studies (energy from sugars subtracted from diets); or ad libitum studies (sugars freely replaced by other macronutrients without control for energy). Overall, it found that total fructose-containing sugars had no harmful effect on any measure of glycemic control in substitution or subtraction studies, a small decrease (−0.22%; −25.9 mmol/mol) in HbA1c in substitution studies, but a detrimental effect on fasting insulin in addition studies (+4.68 pmol/L) and ad libitum studies (+7.24 pmol/L).

The researchers used this data to suggest that total energy intake appears to mediate the relationship between fructose and glycaemic control. We note this meta-analysis assessed only fructose-containing food sources, though a similar mediating effect of energy intake has been reported in studies of sucrose. We found four controlled intervention studies that exchanged sucrose with other energy-matched carbohydrate sources in patients with T2D while measuring changes in glycaemic control, and they all reported no significant change, including in HbA1c, even with large intakes of sucrose (220 grams per day).

Overall, the claim in The Obesity Code that too much sugar causes T2D is weakly supported by currently available evidence. Consuming excess amounts of certain sugar-rich sources, such as sugar-sweetened beverages, is likely to promote excess total energy intake and increase the risk of T2D. However, many sugar-rich foods (e.g. fruit and dairy products) have a neutral or protective association with T2D, with no clear evidence that high doses of these foods incur harm. It is likely that all sugar-rich foods can increase the risk of T2D at some high level of intake, but that is true of any food that contains calories.

Overall (average) score for claim 2

1 out of 4

Claim 3

“Low carbohydrate diets reverse type 2 diabetes”

Supporting quote(s) and page number(s)

[Page x] “The groundbreaking idea Dr. Fung presents in these pages is that diabetes is caused by our bodies’ insulin response to chronic overconsumption of carbohydrates and that the best and most natural way to reverse the disease is to reduce consumption of those carbohydrates. A low-carbohydrate diet for treating obesity is not only being practiced now by hundreds of doctors around the world but is supported by more than seventy-five clinical trials, conducted on altogether thousands of people, including several trials of two years’ duration, which establish the diet as safe and effective.”

[Page xi] “In other words, these patients successfully reversed their diabetes simply by restricting carbohydrates—findings that ought to be compared to the official standard of care for diabetics, which states with 100 percent certainty that the condition is “irreversible.””

[Page xii] “For this is the unvarnished truth: the success of carbohydrate restriction directly implies that the last several decades of low-fat, high-carbohydrate nutrition advice has almost certainly fueled the very obesity and diabetes epidemics it was intended to prevent.”

[Page 206] “We know that bariatric surgery, very low-carbohydrate diets, and fasting are well-known treatments for type 2 diabetes and they are proven to cure.”

[Page 181] “WE KNOW THAT the very essence of type 2 diabetes is too much sugar in the body, not just in the blood. Once we understand this basic paradigm, the solution is immediately obvious. If the problem is too much sugar (glucose and fructose), two treatments will work. And luckily, neither involves surgery or medications: 1. Stop putting sugar in (low-carbohydrate diets, intermittent fasting).”

[Page 184] “Dr. Elliott Joslin himself successfully treated so-called fatty diabetes (type 2 diabetes) with a diet that contained only 2 percent carbohydrates.”

[Pages 218/219] “The following meal plans provide two sample schedules for a 30- to 36-hour fasting regimen complemented by a low-carbohydrate, healthy-fat diet.”

Criterion 1.1. How well is the claim supported by current evidence?

2 out of 4

We scored this claim a 2 out of 4, indicating that relevant evidence weakly supports the claim. Similarly to our review of claim 1, we decided to treat “reverse T2D” as evidence for T2D remission, defined by the American Diabetes Association in 2021 as HbA1c <6.5% (48 mmol/mol) for at least 3 months after stopping glucose-lowering drugs.

There is strong evidence that low-carbohydrate diets (including very-low carbohydrate and ketogenic diets) can improve glycaemic control, typically measured by changes in fasting glucose or HbA1c. The strongest evidence were four systematic reviews and meta-analyses of randomized controlled trials that all reported that carbohydrate-restricted diet significantly reduced HbA1c in participants with T2D versus controls. One systematic review and meta-analysis of randomized controlled diets also compared low-carbohydrate diet interventions versus higher-carbohydrate diet interventions in participants with T2D, reporting a significant lowering of HbA1c (-0.34%; -3.7 mmol/mol) in favor of low-carbohydrate diets at 6 months but not 12 months.

In addition, most relevant trials found that participants in low‐carbohydrate groups reduced their use of glucose-lowering drugs more than with control diets. For example, in one meta-analysis, seven studies reported a significantly reduced use of glucose-lowering medications with participants on low-carbohydrate at 3 and 6 months but not 12 months (though numerically lower) versus high-carbohydrate diets.

Importantly, however, the improvements in glycaemic control and reduced medication use found in these studies is insufficient to support claims for T2D remission, which is a far less common outcome in diet studies. A narrative review in 2021 reported that across 32 trials that examined carbohydrate restriction as a treatment for T2D, long-term data on T2D remission was limited.

We found five studies with a low carbohydrate diet that almost met the criteria for T2D remission. Three of these studies found that a low-carbohydrate diet for 24–52 weeks in participants with T2D significantly reduced HbA1c levels compared with control or higher carbohydrate diets, and that 78–95% of participants in the low-carbohydrate groups reduced or eliminated their use of glucose-lowering medications (not including metformin). However, none of these three studies reported T2D remission as per ADA criteria.

The other two studies did report on T2D remission. One was a prospective intervention study called the VIRTA health study, which compared the effect of a low carbohydrate diet versus usual diabetes care in 349 patients with T2DM. At 1 year, 52 (26%) participants in the low-carbohydrate group achieved T2D remission, decreasing slightly to 46 participants (18%) at 2 years. The second of these studies was a report from a suburban practice in the United States. In this practice, between the years 2013 to 2021, advice on a lower carbohydrate diet and weight loss was offered to 473 patients with T2D. Of these patients, 186 (39%) chose to follow a low-carbohydrate diet and after an average of 33 months follow-up, 51% achieved T2D remission.

To our knowledge, these two studies are the strongest evidence that a low carbohydrate diet can achieve T2D remission. However, these studies only weakly support the claim, since they were not randomized controlled studies and the interventions consisted of far more than advice to lower carbohydrate intake. In VIRTA, participants that chose to follow the low carbohydrate diet received behavioral coaching from a remote care team (with option for on-site education classes), social support via an online peer group, and advice to consume a multivitamin, vitamin D3 and omega-3 supplements daily.

In the report from the suburban practice, clinicians told patients in advance that average weight loss from low-carbohydrate diets in their clinic was 9kg, sent them an infographic illustrating that reduced carbohydrate intake reverses T2D, and similar to VIRTA, used motivational telephone calls when a patient’s HbA1c began to increase. Patients were presumably paying for the medical services, too, which offers a financial incentive to behavioral change. So considering this, it is difficult to disentangle the effects of a low carbohydrate diet for T2D remission in these studies. It should also be pointed out that despite a lot of professional support, approximately 50–75% of patients that chose to follow a low carbohydrate diet in these studies did not achieve T2D remission.

To close, we want to note that weight loss appears to be a moderating factor in all low-carbohydrate studies that achieved HbA1c <6.5%, as is generally the case for all major studies of T2D remission. For example, in the suburban practice that reported T2D remission with a low carbohydrate diet, participants that achieved T2D remission had a mean weight loss of 12 kg, and no patient achieved remission without weight loss. Similarly, in the VIRTA health study, participants in the low-carbohydrate intervention lost an average of 12% body weight at 1 year versus baseline. There is one randomized controlled trial that found that found a lower-carbohydrate diet lowered HbA1c more than a higher-carbohydrate diet independent of weight loss, but the diets were fairly short term (6 weeks) and differed relatively by 100% in protein intake (30% vs 15% total energy).

Overall (average) score for claim 3

2 out of 4

Overall (average) score for scientific accuracy

1.3 out of 4

Reference 1

Reference

Chapter 13, reference 6. Based on Schauer PR, et al. 2017; N Engl J Med. 376(7):641-651.

Associated quote(s) and page number(s)

Page 168: “Surgery cures diabetes” [the title of Figure 13.2]

Criterion 2.1. Does the reference support the claim?

1 out of 4

We scored this item a 1. The claim that surgery cures diabetes was not mentioned or implied in the study. In addition, the figure is very loosely relevant to claims about diabetes remission. No information about blood glucose levels or the proportion of patients not using glucose-lowering pharmacotherapy is presented, which is necessary to inform on T2D remission or a “cure”.

Reference 2

Reference

Chapter 11, reference 27. Chiasson JL, et al. JAMA. 2003; 290(4):486-494.

Associated quote(s) and page number(s)

Page 147: “The 2003 Study to Prevent Non-Insulin-Dependent Diabetes Mellitus (STOP-NIDDM) trial showed that acarbose, despite relatively unimpressive lowering of blood glucose, reduced the risk of cardiovascular events by a remarkable 49 percent and hypertension by 34 percent. In addition to these unprecedented benefits, acarbose also reduced body weight by 1.41 kilograms and waist circumference by 0.79 cm. These results could have been predicted, since blocking absorption of dietary carbohydrates would be expected to lower insulin levels.”

Criterion 2.1. Does the reference support the claim?

4 out of 4

This reference scored a 4, indicating that it strongly supports the claim. This was a double-blind, placebo-controlled, randomized trial of 1429 patients with impaired glucose tolerance. Patients were randomized to receive either placebo or 100 mg of acarbose 3 times a day for 3 years (mean follow-up time was 3.3 years). All the outcome data stated in the claim is accurate.

Reference 3

Reference

Chapter 2, reference 2. Zhang X, et al. Diabetes Care. 2010; 33(7):1665–1673.

Associated quote(s) and page number(s)

Page 17: “Prediabetes is the in-between stage, where blood glucose levels are abnormally high, but not quite high enough to be considered diabetic. It denotes a state of very high risk of future progression to full-fledged type 2 diabetes. A patient with a baseline A1C of 6.0–6.5 percent (42– 48 mmol/mol) has an estimated 25–50 percent risk of developing diabetes within five years. That’s more than twenty times the risk of a person with an A1C of 5.0 percent (31 mmol/mol)”

Criterion 2.1. Does the reference support the claim?

3 out of 4

We scored this item a 3 as it is intrinsically moderately convincing and mostly consistent with the book’s claim. It was a systematic review of HbA1c level and the risk of developing T2D. In alignment with the claim in the book, among the eight studies that reported HbA1c categories, the HbA1c range of 6.0 to 6.5% was associated with a highly increased risk of incident diabetes, frequently 20 or more times the incidence of HbA1c <5.0% (not Hba1c equal to 5.0%, as stated in The Diabetes Code).

When the researchers modeled HbA1c as a function of T2D incidence, based on data from seven studies, they reported a 0.11% increase in HbA1c per 1.0% increase in T2D incidence. This converted to 5-year incidence of 25–50% across patients with a HbA1c of 6.0–6.5%. However, as the researchers noted, the statistical models were limited by a lack of original data. In turn, they were unable to optimally standardize risk estimates and there was considerable variation in these estimates due to differences in population and/or outcome definitions.

 

Reference 4

Reference

Chapter 4, reference 1. Laitner MH et al. Eat Behav 21:193–197. 2016

Associated quote(s) and page number(s)

Page 38: “That same year, Dr. Willett challenged the conventional thinking, reporting that weight gain after age eighteen was the major determinant of type 2 diabetes”

Criterion 2.1. Does the reference support the claim?

2 out of 4

We scored this reference a 2, indicating that it offers weak support for the claim.

This study assessed the risk of T2D in 113,861 women (from the Nurses Health Study) aged 30–55 years old, who were not initially diagnosed with the disease. The researchers compared the women’s self-reported body weight at 18 years old with their current weight, to retrospectively look for associations between weight gain and the T2D risk.

No part of the manuscript made the generalized claim that weight gain after 18 was “the major determinant” of T2D, i.e. it is more important than other risk factors. The authors did not compare weight gain against other causal risk factors. Rather, the authors spoke to their results based on the data collected. The abstract does mention it as “a major determinant”, which implies that it’s important but there could be other important causes. They said, “Thus, weight gain after age 18 was a major determinant of risk. For an increase of 20-35 kg, the relative risk was 11.3, and for an increase of more than 35 kg, the relative risk was 17.3.”

We are also unsure why The Diabetes Code referred only to Walter Willett as making this claim. The cited study had seven authors and Walter Willett was not the primary or senior investigator (first or last author). These were Graham Colditz and Frank Speizer.

 

Reference 5

Reference

Chapter 12, reference 5. TODAY Study Group. N Engl J Med. 2012; 366(24): 2247–2256.

Associated quote(s) and page number(s)

Page 153: “THE 2012 TREATMENT Options for Type 2 Diabetes in Adolescents and Youths (TODAY) randomized study reduced caloric intake to a miniscule 1200 to 1500 calories per day of a low-fat diet, combined with increased exercise. This followed precisely the recommendations made by the 2008 ADA guidelines. Intensive dietary counseling was provided to ensure compliance in this group of motivated teenagers. Massive effort by both patients and study staff failed to improve blood glucose—and the failure rate was astronomically high. Almost 50 percent of patients required increased doses and numbers of medications. Whether or not patients followed the recommended lifestyle recommendations did not matter at all. Regardless, their diabetes was getting worse, not better.”

Criterion 2.1. Does the reference support the claim?

1 out of 4

This reference received a score of 1, indicating that it does not convincingly support the claim. This was a study of 699 participants aged 10–17 years old with recent-onset T2D. Participants were randomly assigned treatment with either metformin, metformin combined with rosiglitazone, or metformin combined with a lifestyle-intervention program focused on weight loss. The primary outcome of interest was the loss of glycemic control, defined as a persistently elevated HbA1c (≥8%) over 6 months and/or sustained metabolic decompensation requiring insulin medication.

The Diabetes Code reported some of the methods and results fairly accurately. The intervention with a lifestyle change did involve 1200–1500kcal diets and professional support. In addition, failure rates were roughly 50%; specifically, 45.6% among all participants, slightly lower with metformin and rosiglitazone (38.6%) and slightly higher with metformin alone (51.7%) or combined with lifestyle intervention (46.6%).

However, we think The Diabetes Code should have mentioned that metformin medication was part of all interventions in this study. Speaking only to the results of diet is potentially misleading, as it was only a part of one of the interventions.

In addition, three parts of the claim in the book are false:

• A low-fat diet was not recommended as part of the lifestyle intervention. The lifestyle intervention was an energy-reduced diet and exercise program based on the Traffic Light Plan, encouraging the consumption of ‘green’ foods (low in energy density) and restricting consumption of ‘amber’ and ‘red’ foods, which were higher in energy density. Foods were not prescribed and the health professionals administering the lifestyle program were told to adapt recommendations to suit individual preferences, traditions and beliefs.
• Dietary intake and adherence were not measured in the study. All that was recorded was attendance to the scheduled visits with a health professional. And by this endpoint, only 53.6% of participants met the preplanned target to attend 75% or more sessions over the 2 year intervention.

• The diet recommendations in this study were not “precisely the recommendations made by the 2008 ADA guidelines”. These guidelines made no reference to the Traffic Light Plan used in this study.

Reference 6

Reference

Chapter 7, reference 26. Hue L, Taegtmeyer H. Am J Physiol Endocrinol Metab. 2009 Sep; 297(3): E 578–E591.

Associated quote(s) and page number(s)

Page 87: “DR. PHILIP RANDLE first described the glucose–fatty acid, or Randle, cycle in 1963. Working with isolated heart and skeletal muscle cell preparations, Randle demonstrated that cells burning glucose could not burn fat and vice versa. Furthermore, this phenomenon did not require the assistance of insulin or any other hormones. Your body simply cannot use both fuels simultaneously. You either burn sugar or fat, but not both.”

Criterion 2.1. Does the reference support the claim?

1 out of 4

This reference received a score of 1, indicating that it weakly supports the claim.

The reference is a retrospective commentary of the Randle Cycle or glucose-fatty acid cycle, published in 1963, which purported that interactions between glucose and fatty acid metabolism in muscle and adipose tissue take the form of a cycle and control blood glucose, fatty acid concentrations and insulin sensitivity. It comments on the findings of the original paper in 1963 and how new evidence has added deeper layers of complexity.

We could not find any statement in either the original paper or this commentary that glucose oxidation stops fatty acid oxidation (or vice versa) or that the body cannot use both fuels simultaneously. What is reported is that “…the Randle cycle draws attention to competition between glucose and fatty acids for their oxidation in muscle and adipose tissue”. Moreover, the commentary states that the original Randle paper “…lacked a convincing molecular explanation for the [control of fatty oxidation by glucose], except for those mediated by insulin.”, which seems to contradict the claim in The Diabetes Code that stopping fat oxidation “…did not require the assistance of insulin…”.

Perhaps more importantly, it would appear that the experimental evidence that led to the initial Randle cycle was in animal models, not humans. Though it is outside the scope of this review, we point toward papers that summarize different tools to measure substrate oxidation at rest or during exercise, which tends to be a mix of both substrates outside of high-intensity exercise. Namely, the respiratory quotient quantifies the relative amounts of glucose and fatty acids being oxidized at any one time, a measurement that was outlined before the Randle Cycle, in 1920.

Reference 7

Reference

Chapter 4, reference 25. Howard BV, et al. JAMA. 2006 Jan 4; 295(1): 39–49.

Associated quote(s) and page number(s)

Page 46: “The Women’s Health Initiative was the most ambitious, important nutrition study ever done. This randomized trial involving almost 50,000 women evaluated the low-fat, low-calorie approach to weight loss. Although it was not specifically a weight-loss trial, one group of women was encouraged through intensive counseling to reduce their daily caloric intake by 342 calories and to increase their level of exercise by 10 percent. These calorie counters expected a weight loss of 32 pounds every single year. When the final results were tallied in 1997, there was only crushing disappointment. Despite good compliance, more than seven years of calorie counting had led to virtually no weight loss. Not even a single pound. This study was a stunning and severe rebuke to the caloric theory of obesity. Reducing calories did not lead to weight loss.”

Criterion 2.1. Does the reference support the claim?

1 out of 4

We scored this item a 1 as the reference does not convincingly support the claim and there are three reporting errors.

The first reporting error is that, in contrast to the description provided in The Diabetes Code, the dietary intervention in this study did not include weight loss or calorie restriction goals. Within the methods section, the researchers state “Participants were informed that the diet was not intended to promote weight loss and were encouraged to maintain usual energy intake by replacing fat calories with calories from other sources, mainly carbohydrate.” There was no expectation to lose 32 pounds per year.

The second reporting error is that the book described the participants as “calorie counters” despite no self-monitoring of energy/calorie intake—participants self-monitored only total grams of dietary fat and also total servings of vegetables, fruits, and grains (found here). The energy intake of participants was estimated with food-frequency questionnaires at baseline and 1 year, and then once every 3 years.

The third reporting error is that arguably neither the treatment or control group achieved a low-fat diet. Dietary fat intake reduced by 8% of total calories in the low fat diet group but, along with the control group, still remained between 29.8–38.8% at baseline and follow-up. The definition of a low-fat diet tends to be <30% at a maximum but often less.

Reference 8

Reference

Chapter 1, reference 20. Menke A, et al. JAMA. 2015; 314(10): 1021–1029.

Associated quote(s) and page number(s)

Page 12: “It is estimated that, in 2012, diabetes cost $245 billion in the United States due to direct health costs and lost productivity”

Criterion 2.1. Does the reference support the claim?

4 out of 4

We scored this claim a 4, indicating that it strongly supports the claim.

The study does not investigate the socioeconomic cost of the diabetes; however, the $245 billion is mentioned in the first line of the introduction with a citation to a paper from  the American Diabetes Association, titled ‘Economic costs of diabetes in the U.S. in 2012’. Here, we found the reported costs to support the claim in The Diabetes Code.

Reference 9

Reference

Chapter 14, reference 16. Shin JY, et al. Am J Clin Nutr. 2013 July; 98(1): 146–159.

Associated quote(s) and page number(s)

Page 177: “In fact, consuming lots of eggs reduces the risk of diabetes by 42 percent”

Criterion 2.1. Does the reference support the claim?

0 out of 4

This item scored 0, indicating that it does not support the claim and offers contradictory evidence.

The item in question is a systematic review and meta-analysis of egg consumption in relation to cardiovascular disease and T2D. The researchers compared the highest (≥1 egg per day) versus lowest category of egg consumption (<1 egg per day) to measure overall risk (hazard ratio).

One of the main findings directly opposes the direction of effect as claimed in The Diabetes Code. That is, higher egg consumption was associated with a 42% increase (not reduction) in the risk of incident T2D versus the lower egg consumption, which was statistically significant. The researchers concluded, “This meta-analysis suggests that egg consumption is not associated with the risk of CVD and cardiac mortality in the general population. However, egg consumption may be associated with an increased incidence of type 2 diabetes among the general population and CVD comorbidity among diabetic patients.”

Results aside, we also consider this meta-analysis insufficient to comment on the risks associated with eating “lots of eggs”, as the highest category for egg consumption starts with only 1 egg per day.

Reference 10

Reference

Chapter 11, reference 27. Marso SP et al. N Engl J Med. 2016; 375(4): 311–322.

Associated quote(s) and page number(s)

Page 147: “The 2016 LEADER trial of the GLP-1 analog Liraglutide showed that nausea occurred four times more often in the drug group than the placebo group. Patients on the medication averaged 2.3 kg weight loss compared to placebo and lowered their A1C by 0.4 percent.”

Criterion 2.1. Does the reference support the claim?

4 out of 4

This item scored a 4, indicating that the reference strongly supports the claim.

This was a double-blind placebo-controlled trial and the data reported in The Diabetes Code for changes in body weight and HbA1C are correct.

Overall (average) score for reference accuracy

2.1 out of 4

Summary of the health-related intervention promoted in the book

A list of the dietary recommendations in The Diabetes Code are as follows:

• Eliminate dietary sugar, sweeteners, and concentrated protein sources (shakes/ powders) from the diet
• Eat a diet low in carbohydrates, moderate in protein, and high in natural fats
• Eat whole and minimally processed (“real”) food
• Practice intermittent fasting (30—36 hour fasts a few days a week)

In the first sample meal plan, days alternate between 3 meals (breakfast, lunch and dinner) and complete fasting. In the second sample meal plan, days alternate between 2 meals in the morning and afternoon before fasting in the evening, and then 1 meal in the evening before fasting the next morning and afternoon. Both meal plans contain similar foods and meals:

• Breakfast options include bacon-wrapped egg frittatas; omelet with sausage; coconut flour pancakes with whipped cream and berries; chia pudding.
• Lunch options include prosciutto salads; chicken drumsticks wrapped in bacon with celery and carrots; chicken “breaded” in pork rinds; steak fajitas.
• Dinner options include beef stir-fry; BBQ shrimp skewers; pulled pork sliders on almond flour buns; salmon salad; ginger chicken lettuce cups with baby bok choy.

Condition targeted by the book, if applicable

Type 2 diabetes

Apparent target audience of the book

People with or at risk of type 2 diabetes.

Criterion 3.1. Is the intervention likely to improve the target condition?

3 out of 4

This item received a score of 3, indicating the intervention is likely to moderately improve the condition. The dietary recommendations in The Diabetes Code will likely help people with or at risk of T2D to lose weight and control their blood glucose levels.

There is a fair amount of data from interventional trials that low carbohydrate diets for 6+ months can help to improve glycemic control. As per Claim 3 in the Scientific Accuracy section of this review, we discussed four systematic reviews and meta-analyses of randomized controlled trials that all reported a carbohydrate-restricted diet significantly reduced HbA1c in participants with T2D versus controls. This benefit seems to extend to short-term interventions that focus exclusively on reducing either (or a combination of) added sugar, high glycemic index carbohydrates and refined carbohydrates.

Currently, not many studies on intermittent fasting assess long-term outcomes (6+ months) and the available evidence is not conclusive that it improves glycemic control. A recent systematic review and meta-analysis of 11 interventional studies in patients with T2D found no significant differences in HbA1c and fasting blood glucose versus control diets. The confidence intervals in all of the included studies bar one were wide though, meaning that it’s hard to rule out that the intervention was effective (probably due to small sample sizes). Two other systematic reviews and meta-analyses on a similar number of clinical trials have reported benefits to intermittent fasting on fasting blood glucose (-0.05 to -0.15mmol/L) versus control diets, and one reported a benefit to HbA1c (-0.08), but effect sizes were very small and not clinically meaningful.

Similarly, considerable uncertainty remains for the effect of non-nutritive sweeteners and the risk of T2D. Recent systematic reviews and meta-analyses of observational studies reported a positive association between a high intake of artificially sweetened beverages and the risk of T2D, but this may be a result of reverse causality: people with T2D being more likely to replace dietary sugar with sweeteners.

Interventional evidence is more neutral but subject to interpretation. One meta-analysis of 9 randomized controlled trials (4–10 months in duration) that swapped dietary sugar for non-nutritive sweeteners in patients with diabetes summarized that there was inconclusive evidence of very low certainty regarding the effects of non-nutritive sweeteners on HbA1c compared with either sugar, placebo, or caloric sweeteners. There is probably no issue with eliminating artificial sweeteners from the diet, but we note that switching to artificial sweeteners is probably an effective strategy to lower sugar intake, which is another key recommendation in The Diabetes Code.

In terms of The Diabetes Code recommending eating whole and unprocessed (“real”) foods, this closely aligns with the American Diabetes Association recommendation to eat “Less processed foods”. Some prospective studies have reported a 10% increased risk of T2D with each incremental intake of ultra-processed food, and most whole and minimally processed foods (e.g. fruit, legumes, whole grains and dairy) are also linked with a reduced risk of T2D, despite The Diabetes Code saying to restrict or avoid these particular foods as they are not low in carbohydrates.

Overall, some of the dietary recommendations, particularly the low carbohydrate intake and advice to eat mostly whole foods, should help people to control and possibly lower their blood glucose levels. We decided to not score this item a 4 due to the lack of long-term studies that investigate these interventions for lowering HbA1c and achieving T2D remission. There is also uncertainty around the effect of intermittent fasting and avoiding non-nutritive sweeteners on glycemic control.

Criterion 3.2. Is the intervention likely to improve general health in the target audience?

3 out of 4

This intervention received a score of 3, indicating it is likely to moderately improve health in the target audience. The main reason for this fairly high score follows from criterion 3.1, where we note the intervention is likely to moderately improve glycemic control in people with or at risk of T2D. There is a long list of minor and major health consequences of T2D that can be improved either with remission or adequate and sustained control of blood sugar levels, and a low carbohydrate diet is one dietary strategy that may help achieve these targets. Some data from a network meta-analysis even suggests a low carbohydrate diet may be the best dietary approach for reducing blood glucose levels in people with T2D, outperforming a low-fat diet, Mediterranean diet, and paleolithic diet.

In addition, the diet eliminates most foods that we think are the most problematic in typical diets. These are the high fat, sugar and salt foods (HFSS) that are highly palatable and often termed “junk food”. We think these foods are an important driver of chronic health issues such as obesity, metabolic disease, and cardiovascular disease, and limiting them is likely to improve health in most people.

Our main concern is that The Diabetes Code also restricts some whole and minimally processed foods (e.g. fruit, tubers, legumes, whole grains, and dairy) that are consistently linked with improved health outcomes. The extent to which a low intake of these foods will impact health will depend on the rest of the diet; however, we note that the average intake of these foods among the general population is already relatively low, so this recommendation should not worsen health.

Criterion 3.3. Does the diet portion of the intervention promote an adequate nutrient intake for general health in the target audience?

1 out of 4

The diet received a score of 1, indicating that it is likely somewhat nutritionally inadequate.

Many nutrient-rich foods are recommended (eggs, meat, fish, and leafy green vegetables) but whether the diet is nutritionally sufficient will depend on how strictly a reader follows the advice. If a reader chooses to simply reduce carbohydrate intake and focus on consuming nutrient-rich whole foods, their diet quality will probably improve and there are no obvious concerns. However, if a reader decides to implement the 30–36 hour fasts a few days a week, and strictly avoids consuming nutrient-rich foods that are higher in carbohydrates (fruit, legumes, tubers, whole grains and dairy), they may risk nutrient deficiencies, particularly with dietary fiber (current recommended daily amount (RDA): 28+ grams), vitamin C (RDA: 90 mg) and calcium (RDA: 1000mg).

To test this, we decided to use Cronometer to analyze the nutrient content of the Sample Meal Plans in the book. Dr. Jason Fung did not state portion sizes, so we adjusted the standard portions on Cronometer until the meals met the recommended calorie target for the average male adult, 2,500 kcal. On non-fasting days, all of the daily meal plans did not meet at least 7 of the RDA’s (for a male adult) for the following 9 nutrients: dietary fiber, calcium, potassium, manganese, magnesium, vitamin C, vitamin K, folate, and vitamin B5. On the fasting days with no food in Sample Meal Plan 1, deficiencies were expectedly flagged with every essential nutrient. Therefore, the book’s dietary advice may pose an unnecessary nutrient deficiency risk for people with or wanting to prevent T2D.

Overall (average) score for healthfulness

2.3 out of 4