Nutrition Therapy

CHO broadly include available CHO from starches and sugars and unavailable CHO from fibre. The dietary reference intakes (DRIs) specify a recommended dietary allowance (RDA) for available CHO of no less than 130 g/day for adult women and men >18 years of age, to provide glucose to the brain (). The DRIs also recommended that the percentage of total daily energy from CHO should be ≥45% to prevent high intake of saturated fatty acids as it has been associated with reduced risk of chronic disease for adults (). If CHO is derived from low glycemic index (GI) and high-fibre foods, it may contribute up to 60% of total energy, with improvements in glycemic and lipid control in adults with type 2 diabetes ().

Systematic reviews and meta-analyses of controlled trials of CHO-restricted diets (mean CHO of 4% to 45% of total energy per day) for people with type 2 diabetes have not shown consistent improvements in A1C compared to control diets (, , ). Similarly, inconsistent improvements in lipids and blood pressure (BP) have been reported when comparing low-CHO to higher-CHO diets (, , ). As for weight loss, low-CHO diets for people with type 2 diabetes have not shown significant advantages for weight loss over the short term (, ). There also do not appear to be any longer-term advantages. Although a network systematic review and meta-analysis of randomized controlled trials of popular weight loss diets showed that low-CHO diets (defined as ≤40% energy from CHO) resulted in greater weight loss compared with high-CHO, low-fat diets (defined as ≥60% energy from CHO) at 6 months, there was no difference at 12 months in individuals with overweight or obesity with a range of metabolic phenotypes, including type 2 diabetes (). Of note, very-low-CHO diets have ketogenic effects that may be a concern for those at risk of diabetic ketoacidosis taking insulin or SGLT2 inhibitors () (see Pharmacologic Glycemic Management of Type 2 Diabetes in Adults chapter, p. S88).

A limited number of small, short-term studies conducted on the use of low-CHO diets (target <75 g/day) in people with type 1 diabetes have demonstrated modest adherence to the prescribed diets with improved A1C for those who can adhere. This style of diet can be an option for those motivated to be so restrictive (, ). Of concern for those following a low-CHO diet is the effectiveness of glucagon in the treatment of hypoglycemia. In a small study, people with type 1 diabetes treated with continuous subcutaneous insulin infusion (CSII) therapy following a low-CHO diet for 1 week had a blunted response to a glucagon bolus (, ). The long-term sustainability and safety of these diets remains uncertain.

The glycemic index (GI) provides an assessment of the quality of CHO-containing foods based on their ability to raise blood glucose (BG) (). To decrease the glycemic response to dietary intake, low-GI CHO foods are exchanged for high-GI CHO foods. Detailed lists can be found in the International Tables of Glycemic Index and Glycemic Load Values ().

Systematic reviews and meta-analyses of randomized trials and large individual randomized trials of interventions replacing high-GI foods with low-GI foods have shown clinically significant improvements in glycemic control over 2 weeks to 6 months in people with type 1 or type 2 diabetes (, , , , , , , ). This dietary strategy has also been shown to improve postprandial glycemia and reduce high-sensitivity C-reactive protein (hsCRP) over 1 year in people with type 2 diabetes (), reduce the number of hypoglycemic events over 24 to 52 weeks in adults and children with type 1 diabetes () and improve total cholesterol (TC) over 2 to 24 weeks in people with and without diabetes (). Irrespective of the comparator, recent systematic reviews and meta-analyses have confirmed the beneficial effect of low-GI diets on glycemic control and blood lipids in people with diabetes (, , ). Other lines of evidence extend these benefits. A systematic review and meta-analysis of prospective cohort studies inclusive of people with diabetes showed that high GI and high glycemic load (GL) diets are associated with increased incidence of cardiovascular disease (CVD), when comparing the highest with the lowest exposures of GI and GL in women more than men over 6 to 25 years ().

Dietary fibre includes the edible components of plant material that are resistant to digestion by human enzymes (nonstarch polysaccharides and lignin, as well as associated substances). They include fibres from commonly consumed foods as well as accepted novel fibres that have been synthesized or derived from agricultural by-products (). DRIs specify an adequate intake (AI) for total fibre of 25 g/day and 38 g/day for women and men 19–50 years of age, respectively, and 21 g/day and 30 g/day for women and men ≥51 years of age, respectively (). Although these recommendations do not differentiate between insoluble and soluble fibres or viscous and nonviscous fibres within soluble fibre, the evidence supporting metabolic benefit is greatest for viscous soluble fibre from different plant sources (e.g. beta-glucan from oats and barley, mucilage from psyllium, glucomannan from konjac mannan, pectin from dietary pulses, eggplant, okra and temperate climate fruits (apples, citrus fruits, berries, etc.). The addition of viscous soluble fibre has been shown to slow gastric emptying and delay the absorption of glucose in the small intestine, thereby improving postprandial glycemic control (, ).

Systematic reviews, meta-analyses of randomized controlled trials and individual randomized controlled trials have shown that different sources of viscous soluble fibre result in improvements in glycemic control assessed as A1C or fasting blood glucose (FBG) (, , ) and blood lipids (, , ). A lipid-lowering advantage is supported by Health Canada-approved cholesterol-lowering health claims for the viscous soluble fibres from oats, barley and psyllium (, , ).

Despite contributing to stool bulking (), insoluble fibre has failed to show similar metabolic advantages in randomized controlled trials in people with diabetes (, , ). These differences between soluble and insoluble fibre are reflected in the EURODIAB prospective complications study, which demonstrated a protective association of soluble fibre that was stronger than that for insoluble fibre in relation to nonfatal CVD, cardiovascular (CV) mortality and all-cause mortality in people with type 1 diabetes (). However, this difference in the metabolic effects between soluble and insoluble fibre is not a consistent finding. A recent systematic review and meta-analysis of prospective cohort studies in people with and without diabetes did not show a difference in risk reduction between fibre types (insoluble, soluble) or fibre source (cereal, fruit, vegetable) (). Given this inconsistency, mixed sources of fibre may be the ideal strategy. Interventions emphasizing high intakes of dietary fibre (≥20 g/1,000 kcal per day) from a combination of types and sources with a third or more provided by viscous soluble fibre (10 to 20 g/day) have shown important advantages for postprandial BG control and blood lipids, including the established therapeutic lipid target low-density lipoprotein cholesterol (LDL-C) (, , ) and, therefore, emphasizing fibre from mixed sources may help to ensure benefit.

Added sugars, especially from fructose-containing sugars (high fructose corn syrup [HFCS], sucrose and fructose), have become a focus of intense public health concern. The main metabolic disturbance of fructose and sucrose in people with diabetes is an elevation of fasting triglycerides (TG) at doses >10% of total daily energy. A systematic review and meta-analysis of randomized controlled trials ≥2-weeks duration showed that added sugars from sucrose, fructose and honey in isocaloric substitution for starch have a modest fasting TG-raising effect in people with diabetes, which was not seen at doses ≤10% of total energy (). Fructose-containing sugars either in isocaloric substitution for starch or under ad libitum conditions have not demonstrated an adverse effect on lipoproteins (LDL-C, TC, high-density lipoprotein cholesterol [HDL-C]), body weight or markers of glycemic control (A1C, FBG or fasting blood insulin) (, , ). Similar results have been seen for added fructose. Consumption of added fructose alone, in place of equal amounts of other sources of CHO (mainly starch), does not have adverse effects on body weight (, ), BP (), fasting TG (, ), postprandial TG (), markers of fatty liver () or uric acid (, ). In fact, it may even lower A1C (, , ) in most people with diabetes. Although HFCS has not been formally tested in controlled trials involving people with diabetes, there is no reason to expect that it would give different results than sucrose. Randomized controlled trials of head-to-head comparisons of HFCS vs. sucrose at doses from the 5th to 95th percentile of United States population intake have shown no differences between HFCS and sucrose over a wide range of cardiometabolic outcomes in participants with overweight or obesity without diabetes (, , , ).

Food sources of sugars may be a more important consideration than the type of sugar per se. A wide range of studies including people with and without diabetes have shown an adverse association of sugar-sweetened beverages (SSBs) with risk of hypertension and coronary heart disease when comparing the highest with the lowest levels of intake (, ). These differences are most apparent when SSBs account for more than 10% of total energy and are likely mediated by the excess calories (, ). This adverse relationship may be specific to SSBs as the same adverse relationship has not been shown for total sugars, sucrose, or fructose (, , , , , , , ), fructose-containing sugars from fruit (, ) or food sources of added sugars, such as whole grains and dairy products (yogurt) (, , , ).

The DRIs do not specify an AI or RDA for total fat, monounsaturated fatty acids (MUFA), saturated fatty acids (SFA), or dietary cholesterol. AIs have only been set for the essential polyunsaturated fatty acids (PUFA): 12 g and 11 g per day for women and 17 g and 14 g per day for men aged 19-50 years and >51 years, respectively, for the n-6 PUFA linoleic acid and 1.1 g per day for women and 1.6 g per day for men aged >18 years for the n-3 PUFA alpha-linolenic acid (). The quality of fat (type of fatty acids) has been shown to be a more important consideration than the quantity of fat for CV risk reduction. Dietary strategies have tended to focus on the reduction of saturated fatty acids (SFA) and dietary cholesterol. The prototypical diets are the United States National Cholesterol Education Program (NCEP) Step I (≤30% total energy as fat, ≤10% of energy as SFA) and Step II (≤7% of energy as SFA, dietary cholesterol ≤200 mg/day) diets (). These diets have shown improvements in lipids and other CV risk factors compared with higher SFA and cholesterol control diets ().

More recent analyses have assessed the relation of different fatty acids with CV outcomes. A systematic review and meta-analysis of prospective cohort studies inclusive of people with diabetes showed that diets low in trans fatty acids (TFA) are associated with less coronary heart disease (CHD) (). Another systematic review and meta-analysis of randomized controlled clinical outcome trials involving people with and without diabetes showed that diets low in SFA decrease combined CV events (). The benefit, however, was restricted to intakes of SFA <9% total energy and to the replacement of SFA with polyunsaturated fatty acids (PUFA) (). Other analyses of the available clinical outcome trials suggest that the food sources of PUFA may even be more relevant with CV benefit restricted to mixed omega-3/omega-6 PUFA sources, such as soybean oil and canola oil (). Pooled analyses of prospective cohort studies and large individual cohort studies also suggest that replacement of saturated fatty acids with high quality sources of monounsaturated fatty acids (MUFA) from olive oil, canola oil, avocado, nuts and seeds, and high quality sources of carbohydrates from whole grains and low GI index carbohydrate foods is associated with decreased incidence of CHD (, ).

The food source of the saturated fatty acids being replaced, however, is another important consideration. Whereas adverse associations have been reliably established for meat as a food source of saturated fatty acids, the same has not been shown for some other food sources of saturated fatty acids (e.g. such as dairy products and plant fats from palm and coconut) ().

A comprehensive review of long-chain omega-3 fatty acids (LC-PUFAs) eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) from fish oils did not show an effect on glycemic control (). Large randomized clinical outcome trials of supplementation with omega-3 LC-PUFAs do not support their use in people with diabetes (, , ). The Outcome Reduction with Initial Glargine lntervention (ORIGIN) trial failed to show a CV or mortality benefit of supplementation with omega-3 LC-PUFA in 12,536 people with prediabetes or type 2 diabetes (). Subsequent systematic reviews and meta-analyses of randomized trials involving more than 75,000 participants with and without diabetes have failed to show a CV benefit of supplementation with long chain omega-3 PUFAs (). The Study of Cardiovascular Events in Diabetes (ASCEND) in 15,480 people with diabetes free of CV disease (clinicaltrials.gov registration number NCT00135226) will provide more data on the outcomes of supplementation with omega-3 LC-PUFA in people with diabetes.

Although supplementation with omega-3 LC-PUFA has not been shown to be beneficial, consumption of fish may be. Prospective cohort analyses have shown higher consumption of fish, ranging from 1 to 3 servings per month to ≥2 servings/week of oily fish, was associated with reductions in coronary artery disease (CAD) (), diabetic chronic kidney disease (CKD) in type 2 diabetes () and less albuminuria in type 1 diabetes ().

The DRIs specify a recommended dietary allowance (RDA) for protein of 0.8 g per kg body weight for adult men and women >18 years of age (). There is no evidence that the usual protein intake for most individuals (1 to 1.5 g per kg body weight per day), representing 15% to 20% of total energy intake, needs to be modified for people with diabetes (). However, this intake in grams per kg per day should be maintained or increased with energy-reduced diets.

Protein quality has been shown to be another important consideration. A systematic review and meta-analysis of randomized controlled trials showed that replacement of animal protein with sources of plant protein improved A1C, FPG and fasting insulin in people with type 1 and type 2 diabetes over a median follow up of 8 weeks ().

People with diabetes who have CKD should target a level of intake that does not exceed the RDA of 0.8 g per kilogram body weight per day (), which has been shown to reduce end stage renal disease and mortality in people with type 1 diabetes and CKD () and improve albuminuria and A1C in people with CKD in diabetes (). When using a low-protein diet, harm due to malnutrition should not be ignored (). Both the quantity and quality (high biological value) of protein intake must be optimized to meet requirements for essential amino acids, necessitating adequate clinical and laboratory monitoring of nutritional status in the individual with diabetes and CKD. Greater incorporation of plant sources of protein may also require closer monitoring of potassium as CKD progresses.

The ideal macronutrient distribution for the management of diabetes can be individualized. Based on evidence for chronic disease prevention and adequacy of essential nutrients, the DRIs recommend acceptable macronutrient distribution ranges (AMDRs) for macronutrients as a percentage of total energy. These include 45% to 65% energy for CHO, 10% to 35% energy for protein and 20% to 35% energy for fat, with 5% to 10% energy derived from linoleic acid and 0.6% to 1.2% energy derived from alpha linolenic acid ().

There may be a benefit of substituting fat as MUFA for carbohydrate (). A systematic review and meta-analysis of randomized controlled trials found that MUFA in isocaloric substitution for CHO (mean replacement of ~14% energy with a dietary macronutrient composition of 40% energy CHO, 33% energy fat, and 17% energy protein) did not reduce A1C but did improve FPG, body weight, systolic BP, TG and HDL-C in people with type 2 diabetes over an average follow up of 19 weeks (). Similarly, the replacement of refined high-GI CHO with MUFA (14.5% total energy) or nuts (5% total energy) to affect a low glycemic load has been shown to improve A1C and lipids, including the established therapeutic lipid target LDL-C in people with type 2 diabetes over 3 months ().

The effect of the replacement of fat with CHO depends on the quality of the CHO and the fat. Whereas the replacement of fat with refined high-GI CHO results in worsening of metabolic parameters in people with type 2 diabetes (), the replacement of saturated fatty acids with low-GI CHO or whole grain sources is associated with decreased incident CHD in people with and without diabetes (, ).

When protein is used to replace CHO, as in a high-protein diet, benefit has only been demonstrated when high-GI CHO are replaced. A 12-month randomized controlled trial in individuals with type 2 diabetes showed improved CV risk profile with a high-protein diet (30% energy protein, 40% energy CHO, 30% energy fat) vs. a high-CHO diet (15% energy protein, 55% energy CHO, 30% energy fat), in which the CHO were high GI. These differences were seen despite similar weight loss with normal renal function being maintained (). In contrast, a 12-month randomized controlled trial comparing a high-protein diet (30% energy protein, 40% energy CHO, 30% energy fat) vs. a high-CHO low-GI diet (15% energy protein, 55% energy CHO, 30% energy fat) failed to show a difference between the diets (). Rather, it was adherence to any 1 diet and the degree of energy restriction, not the variation in diet macronutrient composition, that was associated with the long-term improvement in glycemic control and cardiometabolic risk factors ().

Adjustments in medication type and dosage may be required when embarking on a different macronutrient distribution () or energy reduction () to avoid hypoglycemia.

A Mediterranean diet primarily refers to a plant-based diet first described in the 1960s (). General features include high consumption of fruits, vegetables, legumes, nuts, seeds, cereals and whole grains; moderate-to-high consumption of olive oil (as the principal source of fat); low-to-moderate consumption of dairy products, fish and poultry; low consumption of red meat; and low-to-moderate consumption of wine, mainly during meals (, ). Systematic reviews and meta-analyses of randomized controlled feeding trials have shown that a Mediterranean-style dietary pattern improves glycemic control (, ), and improves systolic BP, TC, HDL-C, TC:HDL-C ratio and TG in type 2 diabetes (, ).

A low-CHO Mediterranean-style diet reduced A1C, delayed the need for antihyperglycemic drug therapy and increased rates of diabetes remission compared with a low-fat diet in overweight individuals with newly diagnosed type 2 diabetes at 8 years (). Compared with a diet based on the American Diabetes Association recommendations, both traditional and low-CHO Mediterranean-style diets were shown to decrease A1C and TG, whereas only the low-CHO Mediterranean-style diet improved LDL-C and HDL-C at 1 year in persons with overweight and type 2 diabetes ().

The Prevencion con Dieta Mediterranea (PREDIMED) study, a Spanish multicentre randomized trial of the effect of a Mediterranean diet supplemented with extra-virgin olive oil or mixed nuts compared with a low-fat American Heart Association (AHA) control diet, was stopped early due to significant benefit with reduction in major CV events in 7,447 participants at high CV risk (including 3,614 participants [49%] with type 2 diabetes) (). Both types of Mediterranean diets were shown to reduce the incidence of major CV events by approximately 30% without any subgroup differences between participants with and without diabetes over a median follow up of 4.8 years () (see Cardiovascular Protection in People with Diabetes chapter, p. S162). Both the extra-virgin olive oil and mixed nuts arms of the PREDIMED trial also reduced risk of incident retinopathy. No effect on nephropathy was detected ().

Vegetarian dietary patterns include lacto-ovovegetarian, lactovegetarian, ovovegetarian and vegan dietary patterns. A low-fat, ad libitum vegan diet has been shown to be just as beneficial as conventional American Diabetes Association dietary guidelines in promoting weight loss and improving fasting BG and lipids over 74 weeks in adults with type 2 diabetes and, when taking medication changes into account, the vegan diet improved glycemia and plasma lipids more than the conventional diet (). On both diets, weekly or biweekly nutrition and cooking instruction was provided by a dietitian or cooking instructor (). Similarly, a calorie-restricted vegetarian diet was shown to improve BMI and LDL-C more than a conventional diet in people with type 2 diabetes (). While both diets were effective in reducing A1C, more participants on the vegetarian diet had a decrease in antihyperglycemic medications compared to those on the conventional diet (43% vs. 5%, respectively). Subsequent systematic reviews and meta-analyses of the available randomized controlled trials have shown that vegetarian and vegan dietary patterns resulted in clinically meaningful improvements in A1C and FBG in people with type 1 and type 2 diabetes over 4 to 74 weeks (, ), as well as body weight () and blood lipids () in people with and without diabetes over 3 to 74 weeks. Although most of these effects have been seen on high-CHO, low-fat vegetarian and vegan dietary patterns, there is evidence from the Eco-Atkins trial that these apply equally to low-CHO vegetarian dietary patterns (130 g/day [26% energy] CHO, 31% energy protein and 43% energy fat) for up to 6 months in individuals with overweight but without diabetes (, ). A systematic review and meta-analysis of prospective cohort and cross-sectional observational studies showed a protective association between vegetarian dietary patterns and incident fatal and nonfatal CHD ().

Dietary approaches to reducing BP have focused on sodium reduction and the Dietary Approaches to Stop Hypertension (DASH) dietary pattern. Although advice to the general population over 1 year of age is to achieve a sodium intake that meets the adequate intake (AI) target of 1,000 to 1,500 mg/day (depending on age, sex, pregnancy and lactation) (), there is recent concern from prospective cohort studies that low-sodium intakes may be associated with increased mortality in people with type 1 () and type 2 diabetes ().

The DASH dietary pattern does not target sodium reduction but rather emphasizes vegetables, fruits and low-fat dairy products, and includes whole grains, poultry, fish and nuts. It contains smaller amounts of red and processed meat, sweets, sugar-containing beverages, total and saturated fat, and cholesterol, and larger amounts of potassium, calcium, magnesium, dietary fibre and protein than typical Western diets (, ). The DASH dietary pattern has been shown to lower systolic and diastolic BP compared with a typical American diet matched for sodium intake in people with and without hypertension, inclusive of people with well-controlled diabetes (, ). These improvements in BP have been shown to hold at high (3,220 mg), medium (2,300 mg), and low (1,495 mg) levels of matched sodium intake (). In addition to BP-lowering benefit, a systematic review and meta-analysis of randomized controlled trials showed that a DASH dietary pattern lowered lipids, including LDL-C in people with and without hypertension, some of whom had metabolic syndrome or diabetes ().