Guidelines

A Delphi consensus-based guideline for the implementation of therapeutic carbohydrate reduction in metabolic disorders

Sarah M. Rice, Laura A. Buchanan, Matthew W. Calkins, Mark T. Cucuzzella, Robert Cywes, Pamela L. Devine, Miriam Kalamian, Tro Kalayjian, Douglas B. Reynolds, Eric C. Westman

Received: 30 Apr. 2026; Accepted: 20 July 2026; Published: 02 Oct. 2026

Copyright: © 2026. The Authors. Licensee: AOSIS.
This work is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license (https://creativecommons.org/licenses/by/4.0/).

Abstract

Metabolic disorders represent a growing global health burden, with a high percentage of adults demonstrating some features of metabolic dysfunction. The Society of Metabolic Health Practitioners recognised the need for a consensus guideline for the implementation of therapeutic carbohydrate reduction (TCR), which is increasingly used in clinical practice. This document updates a 2019 expert panel-derived statement by incorporating new evidence and applying a formal consensus methodology. This guideline was developed using a structured Delphi consensus process with predefined agreement thresholds. The steering committee reviewed and expanded the 2019 document during an initial round (pre-Delphi, round 0), after which panel feedback informed substantial structural revision and integration of comments, resulting in a redrafted version that formed the basis for subsequent consensus rounds. Statement development was informed by a targeted, non-systematic review of human studies from research databases. Given the heterogeneous and evolving evidence base, no formal evidence grading framework was applied. After three rounds of feedback and iterations an anonymous survey was conducted. This Delphi consensus-based guideline provides practical, evidence-informed recommendations for the implementation of TCR in metabolic disorders. Consensus was achieved across key domains relating to the implementation of TCR in metabolic disorders. A whole foods TCR approach is nutritionally sufficient and may be tailored to a wide range of cultural and personal preferences, focusing on protein, healthy fats from whole foods, and vegetables. Therapeutic carbohydrate reduction is safe and effective for a range of metabolic disorders. Adaptation effects are generally mild, short-lived, and easily managed, and long-term adherence is both possible and enjoyable. Further research is required to refine application across diverse populations and clinical contexts.

Keywords: therapeutic carbohydrate reduction; insulin resistance; metabolic syndrome; low-carbohydrate diet; ketogenic diet; type 2 diabetes; obesity.

Introduction

Non-communicable diseases (NCDs) are at epidemic levels, with 2021 data suggesting at least 43 million people succumbed that year, with cardiovascular diseases (CVDs) accounting for the most deaths.¹ The leading contributors to these deaths are CVD, cancers, respiratory diseases and diabetes, accounting for 80% of all premature NCD deaths. The metabolic risk factors for the development of NCD are primarily hypertension, dysglycaemia, overweight and obesity and dyslipidaemia.¹

Obesity and type 2 diabetes (T2D), also considered epidemics, come under the umbrella of metabolic diseases, having other clinical features such as insulin resistance, dysglycaemia, hypertension and metabolic-associated steatotic liver disease (MASLD).² Metabolic syndrome, with a cluster of these features, affects approximately 25% of the global population, and in the US, just over 90% of the population have suboptimal cardiometabolic metrics.^3,4 These data reflect alarming trends, which are also emerging in our paediatric population and place significant burdens on our current and future healthcare infrastructure.

Therapeutic carbohydrate reduction (TCR) has resurfaced as a viable option for the treatment of metabolic disorders, with guidelines from major diabetes and nutrition organisations recognising its potential and incorporating it as an option with the appropriate clinical oversight and individualised approach.^5,6,7,8,9

Randomised controlled trials, systematic reviews, meta-analyses and consensus statements support the use of TCR in obesity, metabolic syndrome and T2D to improve glycaemic control, promote weight loss, reduce medication requirements and improve cardiometabolic risk factors.^10,11,12,13 A recent American Diabetes Association standards of care update (2025) states, ‘For people with T2D, low-carbohydrate and very-low-carbohydrate eating patterns have been found to reduce A1C and the need for glucose-lowering medication’.⁵ Other organisations that acknowledge TCR for T2D include Diabetes Australia,⁶ Diabetes UK,⁷ Diabetes Canada⁸ and the European Association for the Study of Diabetes.⁹

There is also a strong body of evidence for the use of tightly managed ketogenic diets for the treatment of epilepsy, which has paved the way for emerging applications in other neurological conditions (neurodegenerative, psychiatric and migraine disorders), where therapeutic levels of ketosis (up to 3 mmol/L – 5 mmol/L) may be more critical for success.^14,15,16 Therapeutic carbohydrate reduction exists on a continuum where nutritional ketosis may or may not be necessary to achieve therapeutic goals in metabolic disorders.

In 2019, a group of clinicians, academics and dietitians working in TCR identified a need for an implementation guideline to support clinicians using TCR in practice. A panel was convened, and after several iterations, the resulting document was developed and subsequently became available on the Society of Metabolic Health Practitioners’ (SMHP™) website (2020).¹⁷ This current consensus guideline has been developed as an update to the original document using a panel of experts and a modified Delphi consensus methodology, with document development and oversight provided by the SMHP steering committee.

Methods

Overview

The Delphi method is widely accepted for consensus development in healthcare for complex or underinvestigated clinical issues and to inform best practices where high-quality evidence is lacking.¹⁸ A classic Delphi methodology involves a panel of experts giving feedback via multiple rounds of anonymous questionnaires and iterations using response statistics. Our modified cloud-based approach allowed anonymous, real-time feedback and interactions to maximise efficiency and opportunities for anonymous comments in each round. The panel rounds began in June 2025, with the final round consisting of an anonymous survey issued in January 2026.

The SMHP steering committee of 10 individuals (which included the SMHP board members), consisting of clinicians, an independent researcher, scientists, a certified nutrition specialist and patient advocates, guided the process and panel recruitment and served as points of contact in the early rounds.

Statements were developed using a structured Delphi consensus process with predefined agreement thresholds. Guideline update and statement development were supported by a non-systematic literature search of human studies guided by the authors’ clinical expertise and familiarity with established and emerging evidence. Studies were selected from PubMed, Google Scholar and a private database using Boolean operators to combine search terms such as ‘low-carbohydrate’, ‘ketogenic’ and ‘reduced carbohydrate’ with topics of interest. The abstracts were screened according to relevance specific to factors affecting TCR implementation. The evidence base was heterogeneous with a risk of bias; hence, no formal evidence grading framework was applied, and consensus methodology was used to generate implementation guidance.

The conditions for consensus were pre-specified and communicated to the panel at the start of the project according to SMHP consensus criteria previously published.¹⁹ The number of rounds depended on achieving stability (minimal or no edits required) before proceeding to an anonymous survey, with consensus determined by the following criteria¹⁹:

• An 80% survey response rate is required.
• Consensus declared at 70% agreement (of those who responded).
• Two similar suggestions for a change or addition warrant a re-evaluation of the statement point.

Expert panel selection

The SMHP approached the original panel members, as well as others with relevant clinical, dietetic or nutritional and/or research experience in the use of therapeutic carbohydrate reduction. The aim was to gather multidisciplinary representation with geographical diversity to best reflect opinions across a broad range of clinical populations. Clinical specialities represented include obesity medicine, family medicine (general practice), dietetics and nutrition, endocrinology, cardiology, surgical specialities, orthopaedics, psychiatry, obstetrics and gynaecology, sports medicine and academia.

At inception, the panel consisted of 18 members, with two withdrawals during the rounds, giving a final panel of 16 members. All members gave their consent for participation, which was voluntary.

Study design

The steering committee reviewed and added to the original implementation guideline, which served as the base document. The draft was circulated to the panel, whose feedback informed structural revisions and the integration of comments into a fully redrafted version. Two further rounds were required before it was determined the document had reached stability (few minor comments were incorporated, leaving core statements unchanged). This iterative process allowed an opportunity for all panel members to review the document and comment anonymously, which included interacting with other comments. Finally, an anonymous online survey was compiled, which consisted of:

• Introduction and instructions – consensus criteria protocol. Agreement with summary statements reflects agreement with the full document.
• Participation consent – agree (complete survey) or disagree (submit name for withdrawal).
• 18 summary statements reflecting the core features of each section – select agree or disagree.
• An opportunity for comment.
• A separate participation log to reconcile response names and numbers and follow up with non-responders.

See Table 1 for a summary of our Delphi rounds process. This article represents a consensus-based guideline, developed using structured expert agreement methods in areas where high-certainty evidence is limited.

Limitations

All types of studies were included, drawing on a wide evidence base with a risk of bias. Participation in rounds 1–3 was not established because of anonymity, which was a limitation of the study design; however, participation in the final survey was established via consent and a second participation survey log. Non-responders at the target date were followed up, and there was consensus agreement regarding the final product.

Ethical considerations

This article followed all ethical standards for research without direct contact with human or animal subjects.

Results

A total of 28 participants (18 panel members and 10 steering committee members) were included in the Delphi process. In the final consensus round, 26 participants completed the survey (16/18 panel members and 10/10 steering committee members), yielding an overall response rate of 92.9%, exceeding the predefined threshold of 80%.

The final round utilised a binary (agree or disagree) response format. All 18 statements met the predefined consensus criterion of ≥ 70% agreement. Fifteen statements achieved unanimous agreement (26/26; 100%), while three statements achieved 96% agreement (25/26).

Primary analysis was conducted using the combined participant group (panel members and steering committee; n = 26). A secondary analysis for panel members (n = 16) was performed to confirm findings. Consensus outcomes were unchanged under this restriction, with all statements continuing to meet the predefined agreement threshold. These findings indicate that consensus was achieved across all statements, with a high level of agreement observed.

Final survey feedback resulted in the inclusion of minor clarifications only; no modifications to the core survey statements were required, and the consensus outcomes remained unchanged. The panel was notified. The consensus statements for the implementation of therapeutic carbohydrate reduction in metabolic disorders are presented in the following sections.

Background and rationale for the implementation of therapeutic carbohydrate reduction in metabolic disorders

History

While the historical development of ketogenic diets for epilepsy is well known, TCR (using a more flexible approach) also has a long history of controlling metabolic conditions such as diabetes and obesity, with manuscripts dating back to the 1800s. Prior to the discovery of insulin, TCR was the primary intervention for type 1 diabetes and appeared in textbooks for the treatment of obesity.^20,21 Even after the discovery of insulin in the 1920s, TCR remained the focus of dietary advice until the diet-heart hypothesis of the 1950s saw TCR fall out of favour as low-fat recommendations took hold.²⁰ Recent interest in TCR has seen a surge of publications supporting the safety and efficacy of TCR for metabolic conditions,^10,11,12 with other applications under investigation (as an adjunct) in neurology, psychiatry and oncology.^22,23,24

Pathophysiology

Insulin resistance

In 1988, Gerald Reaven delivered his seminal lecture on the Role of Insulin Resistance in Human Disease, where he defined insulin resistance as reduced sensitivity to insulin-stimulated glucose uptake and pronounced it a key feature of T2D, hypertension and coronary artery disease.²⁵ He recognised the cluster of features, which he named Syndrome X, now called metabolic syndrome, as downstream features of insulin resistance and identified glucose intolerance, elevated triglycerides, low high-density lipoprotein cholesterol (HDL-c) and hypertension to be associated with chronic disease (see Table 2).

Reaven understood that insulin resistance alone did not determine the degree of glucose intolerance but that the ability of the pancreatic β-cells to compensate for insulin resistance by increasing insulin secretion is crucial in maintaining normal glucose tolerance. Individuals with severe insulin resistance but robust β-cell compensation can remain normoglycaemic, while those who cannot sustain hyperinsulinaemia develop hyperglycaemia and eventually T2D. In T2D, there is resistance to insulin-stimulated glucose uptake as well as to insulin suppression of plasma free fatty acid (FFA) concentrations. Elevated FFAs can stimulate hepatic glucose production, contributing to fasting hyperglycaemia.

Reaven also noted insulin resistance was common in those with hypertension and that antihypertensive treatment did not address the metabolic abnormalities, positing insulin resistance, via its action on the sympathetic nervous system and enhanced renal sodium reabsorption, could be the mechanism driving hypertension.²⁵ Other consequences of insulin resistance include proliferation of vascular smooth muscle, increased sympathetic nervous system activity and diminished release of nitric oxide from the endothelium.^28,29

Numerous studies now support hyperinsulinaemia as an independent risk factor for CVD, with associated hypertriglyceridaemia and low HDL-c.^30,31 In addition, we now know there exists heterogeneity in insulin resistance and β-cell dysfunction between individuals and deficiencies in incretin action and hepatic glucose regulation also occur in patients who develop T2D. Metwally et al. recently identified four phenotypes of metabolic dysregulation.³² These may be loosely grouped as an insulin-resistant dominant subphenotype (muscle and liver insulin resistance – highly intercorrelated) and an insulin secretion subphenotype (β-cell function and incretin deficiency). These phenotypes may not correlate with standard measures of glucose regulation such as haemoglobin A1c (HbA1c).³²

The carbohydrate-insulin model of obesity

Insulin resistance is a common feature of overweight and obesity, and academic discussion continues as to the precise mechanisms contributing to weight gain.^33,34

The carbohydrate-insulin model (CIM) of obesity posits that a high-carbohydrate diet (particularly rapidly digestible carbohydrates in the form of starchy foods and sugar) promotes homeostatic dysregulation, driven primarily by postprandial hyperinsulinaemia.³³ While weight gain can result from various mechanisms, insulin primarily drives anabolic pathways. It stimulates glucose uptake into tissues and inhibits the release of fatty acids from adipose tissue and ketone production, driving fat and glycogen storage.³³ In this model, overconsumption is a consequence of increasing adiposity rather than the driving factor of weight gain. The CIM acknowledges that the insulin response to carbohydrate varies with individuals and that influences individual responses to different diets, where those who exhibit an intrinsically higher insulin response to carbohydrates may be more vulnerable to weight gain on a higher (particularly high-glycaemic load) carbohydrate diet.³³ A key tenet of the CIM is that the metabolic effects of reducing carbohydrates in the diet drive weight loss via increasing the metabolic rate, and that this has a greater influence than reducing dietary fat or energy intake.^35,36 This has been demonstrated by observing an increase in energy expenditure during a reduced-carbohydrate weight loss maintenance period.³⁷ Reduced carbohydrate approaches specifically target insulin signalling, which is a key component of this model.

Others hypothesise that the brain is primarily responsible for excessive weight gain through external and internal signals in response to the food environment. They posit that various hormones impacting neuroregulatory pathways drive imbalances in food intake and fuel partitioning, promoting fat storage. This is also known as the energy balance model (EBM).³⁴

Reduced carbohydrate approaches also target neurobiology via various mechanisms, and over time, both hypotheses have expanded their definitions, and an integrated model was proposed by Ludwig and Sørensen in 2022.³⁸ The competing hypotheses agree that^33,34,39:

• Fundamental changes to the food environment have driven the obesity pandemic. Food quality and composition are important.
• Genetic and epigenetic factors can influence the tendency to weight gain in combination with environmental, social and behavioural factors, e.g. social determinants of health, night shifts, limited access to nutritionally dense whole foods, etc.
• Differences in fuel partitioning among individuals can result in different degrees of adiposity despite isocaloric intake.
• Brain signalling, endocrine regulation, peripheral energy sensing and fuel partitioning contribute to metabolic homeostasis (or lack thereof). May include gut microbiome interactions and influence.⁴⁰
• Overeating (particularly of highly processed carbohydrates and/or ultra-processed foods high in fat and sugar), irrespective of drivers (metabolic or neuroregulatory, or both), contributes to weight gain.
• TCR can be beneficial for the management of obesity and cardiometabolic disorders (see ‘Mechanisms of action of therapeutic carbohydrate reduction for metabolic disorders’ section).

Mechanisms influencing satiety

Macronutrient composition impacts hormonal signalling and thus influences satiety.⁴¹ Targeting adequate protein intake is a feature of TCR, and increasing protein in the diet has been associated with increased satiety when carbohydrate content has been kept stable.⁴² A study from Darcey et al. has shown that the metabolic and neurological effects of fat-restricted diets (low-fat) influence subsequent ad libitum food intake towards foods higher in carbohydrate and fat, suggesting macronutrient effects on brain-reward regions.⁴³ Reducing fat in the diet reduced dopamine tone in brain-reward regions, making adherence more challenging. This effect was not observed in the reduced-carbohydrate arm. They posit, ‘Results suggest that “a calorie is not a calorie,” when it comes to macronutrient effects on brain-reward regions in humans’.⁴³

In real-world settings, via various potential mechanisms, a reduction in energy intake usually occurs spontaneously with carbohydrate reduction.^44,45 Therefore, this, along with increased energy expenditure, may play a role in the weight loss commonly seen with TCR. Therapeutic carbohydrate reduction addresses the core features implicated in obesity and other metabolic disorders, as discussed in ‘Mechanisms of action of therapeutic carbohydrate reduction for metabolic disorders’ section.

Therapeutic carbohydrate reduction

A well-formulated carbohydrate-reduced diet includes adequate intake of protein, fat, vitamins and minerals, and it can be personalised according to a range of therapeutic targets. This encompasses adjusting macronutrient ratios, excluding problematic foods and accounting for personal or cultural preferences and may include increasing, decreasing or maintaining energy intake. The focus should be on whole foods where possible.

Definition of therapeutic carbohydrate reduction

In general, TCR meets nutritional requirements within the context of carbohydrate reduction, typically defined as less than 130 g of dietary carbohydrate per day (< 26% total energy intake [TEI]).⁴⁶ There is no standardised TCR approach in the literature; rather, different protocols have been utilised across study types. The definitions in Table 3 reflect levels of carbohydrate reduction found in academic literature.

While the different definitions appear challenging to navigate, the appropriate level of carbohydrate reduction to meet therapeutic goals will differ among individuals and may include a very low-carbohydrate ketogenic diet. The data suggest lower levels of carbohydrate intake are associated with more predictable clinical outcomes,⁵⁴ especially if nutritional ketosis is the goal.

Nutritional ketosis is a metabolic state occurring when insulin levels fall sufficiently such that glucagon rises and lipolysis is initiated, denoting a shift from a glucose-based metabolism to one that relies on fatty acid metabolism and ketone production. This typically occurs at a carbohydrate intake of < 50 g of carbohydrate per day (< 10% TEI) but may only occur at lower levels (~20 g/day) in some and may be experienced with less carbohydrate reduction in others (> 70 g/day) when based on ketone measurements.⁴⁶

As blood glucose drops during carbohydrate reduction, FFAs from adipose tissue are converted mainly by the liver to ketone bodies (acetoacetate, beta-hydroxybutyrate and acetone) and distributed to extrahepatic tissues to be used as an alternative fuel source. Ketone bodies are produced predominantly by the liver; however, extrahepatic ketogenesis has been described in specific tissues (e.g. intestinal epithelium and specialised tissues such as the retinal pigment epithelium) and may also occur locally for intercellular signalling in select contexts, though the contribution to circulating ketones outside the liver is generally limited.⁵⁵ Ketosis is a normal physiological state that commonly occurs at low levels after an overnight fast and may be influenced by various dietary changes that affect metabolism.^56,57 Metabolic adaptation may take some time to establish, and while measuring ketones can be a way to inform dietary adjustments and monitor adherence, the presence of ketones does not necessarily reflect a fat-adapted metabolism.⁵⁷

Ketone levels induced by carbohydrate reduction do not approach the levels induced by frank insulin deficiency, as in diabetic ketoacidosis. Pathological diabetic ketoacidosis more commonly affects individuals with type 1 diabetes (T1DM), though TCR may be adopted successfully in this population.^58,59 TCR typically does not induce acidosis in isolation but may rarely occur in combination with other factors such as starvation, drug interactions (e.g. sodium-glucose co-transporter [SGLT2] inhibitors),⁶⁰ lactation, genetic inborn errors of metabolism and alcohol withdrawal. In these situations, the pathways that regulate ketone metabolism may become overwhelmed, leading to acidosis. A recent study found these events to be rare, even in participants using SGLT2 inhibitors, and comparable to those not following a TCR approach⁶⁰ (see Table 4 for an overview).

Mechanisms of action of therapeutic carbohydrate reduction for metabolic disorders

Therapeutic carbohydrate reduction addresses key features of the pathophysiology of insulin resistance, obesity, T2D and CVD risk factors, and its application for these metabolic disorders is robustly supported by a number of systematic reviews and meta-analyses of randomised controlled studies.^10,11,12 Notably, meta-analyses have shown glycaemic and other metabolic improvements independent of energy intake.^10,65 In addition, spontaneous reduction in food intake is a common feature with TCR.^44,45

Various studies point to particular benefits of TCR approaches:

• May demonstrate superior weight loss compared to low-fat diets over the short and longer term.^66,67
• Increased energy expenditure during weight maintenance^35,68 and reduced adaptive thermogenesis.³⁷
• Improved glycaemic control without significant weight loss or caloric restriction.^10,69
• Possibility of remission of prediabetes and T2D.^70,71
• Improved metabolic syndrome independent of weight loss.⁷²
• Decreased appetite and cravings and improved satiety.^44,45
• Improved CVD risk factors, which include body weight, glucose control, triglycerides, HDL-c, improved lipoprotein subclasses and phenotypes and blood pressure.^11,73,74

Therapeutic carbohydrate reduction addresses the core pathophysiology of obesity, insulin resistance and T2D by improving hyperinsulinaemia, reducing visceral adiposity and inflammation and promoting metabolic flexibility (see Table 5). These mechanisms of action also address elements of the four subphenotypes of T2D posited by Metwally et al.³² While nutritional ketosis is not essential for success, it is a key component of TCR and may promote additional benefits. Ketone body metabolism influences multiple pathways and is associated with protection against oxidative stress and inflammation and improved mitochondrial function and energetics. Beta-hydroxybutyrate regulates epigenetics involved in the pathogenesis of disease, exerting therapeutic effects via histone modification and deoxyribonucleic acid (DNA) methylation.⁷⁵ In addition, recent preclinical research has discovered a secondary β-hydroxybutyrate metabolic pathway linked to energy balance that may directly suppress appetite and influence body weight.⁷⁶

See ‘Clinician resources and further reading’ section for additional supporting material and references.

Other potential applications for therapeutic carbohydrate reduction

Therapeutic carbohydrate reduction is being investigated for a wide range of metabolic disorders where hyperinsulinaemia is a core feature, and applications as an adjunct in neurology, psychiatry and oncology are under investigation (see Table 6).

For some conditions, more consistent ketone levels are required for therapeutic effect (in the order of 1.5 mmol/L – 5 mmol/L), and adjustments in macronutrient intake may be required to reach target levels.^22,23 Currently, areas where ketone levels are demonstrating a dose–response effect include neurology, psychiatry and oncology, and the use of the glucose/ketone index may have clinical value.^22,23,93 In this setting, TCR is a ketogenic intervention and may be referred to as ketogenic dietary therapy (KDT)^14,22 or ketogenic metabolic therapy (KMT), and the support of an experienced practitioner is recommended.^15,23 Monitoring ketone levels may be helpful in determining the best macronutrient intake for an individual.

Further discussion on therapeutic ketogenic applications is outside the scope of this article.

Dietary features of therapeutic carbohydrate reduction

For metabolic conditions, TCR comprises the following core features⁴⁶:

• Carbohydrate reduction (< 10% – 25% TEI) depending on therapeutic goals – avoid high-carbohydrate foods (bread, pasta, rice, cereals, etc.), high-sugar fruits and starchy (below-ground) vegetables.
• Carbohydrate sources are allowed from non-starchy vegetables and low-sugar fruits (tomatoes, olives, avocados and berries). Small amounts of carbohydrate are present in dairy and other protein sources.
• Appropriate protein intake is individualised for building or maintaining lean body mass and performance. It is suggested that a higher protein intake during weight loss may be beneficial.⁴⁶ A number of approaches to estimating protein intake exist. Examples include:
  • Based on 1.2 g/kg – 2.0 g/kg reference weight⁴⁶
  • 1.5 g/kg of a medium-frame ideal weight⁴⁷
  • 25 g – 30 g per meal (3 meals)^46,94
  • 15% – 20% TEI or 20g per 100 g of macros eaten (clinician experience)
• In general, fat consumption is to satiety.^41,46 Carbohydrate and protein intake are set first, with fat meeting the remaining nutritional needs.^41,46 Individual response to fat intake varies and may require consideration of differences in fat oxidation and utilisation rates, exercise (training demands and aerobic fitness), body composition goals, dietary preferences and metabolic response.^95,96 These pathways may be influenced by genetics, biological sex, age and hormone profiles.^95,96 Main sources of fat include meats, full-fat dairy (cheese, butter, cream, etc.), olives and avocados, cooking fats (e.g. tallow, butter, coconut oil) and nuts and seeds if tolerated. Caution may be required with energy-dense foods if weight reduction is a goal (examples of foods in this category that are easy to overeat include nuts).
• Emphasise nutrient-dense whole foods; avoid highly processed foods and added sugars. A simple list is meat (including offal), fish, eggs, dairy and above-ground vegetables.
• Monitor for adequate essential vitamin and mineral intake (especially in vegan or vegetarian patterns).
• Tracking energy intake is not generally required unless using TEI to assign % distribution of macronutrients per day, though it may be helpful for troubleshooting or in vulnerable populations (e.g. severe illness to ensure sufficiency). Measuring blood glucose, in the context of food choices, may be helpful because of individual variations in blood glucose responses to foods (see ‘Implementation’ section).

Individualising the approach based on personal preferences, cultures, food sensitivities and therapeutic goals (including carbohydrate content) is assumed. See Table 7 for a summary of food lists.

Fibre

As fibre constitutes the main carbohydrate component of a whole-food TCR approach, a total carbohydrate approach ensures the most consistent results. Net carbohydrate counting (total carbohydrate minus fibre and/or sugar alcohols) could theoretically be applied to fruit and vegetables, as fibre is not typically metabolised to glucose; however, assessing patient responses to total carbohydrate intake and adjusting upwards if tolerated appears to be the most straightforward approach. Fibre intake may be adjusted according to patient responses and therapeutic goals. While general TCR provides adequate fibre and has been shown to favourably influence the gut microbiome,⁹⁷ diets lower in fibre in the context of TCR may be considered because of the adaptability of the gut microbiome and ketone metabolites.^98,99 Emerging evidence has demonstrated that a zero-carbohydrate exclusion protocol (no fibre) may have benefits for select cases of gastrointestinal disorders and suggests fibre is not essential for bowel health.⁵¹ An individualised approach that accounts for patient physiology and preferences is recommended. For example, a low-fibre diet may benefit cases of chronic constipation.¹⁰⁰

Protein

Adequate dietary protein may be achieved with a range of dietary patterns. Animal-sourced foods contain the full complement of amino acids (20) and all nine essential amino acids (EAAs). Plant proteins are combined with the plant fibre matrix, reducing bioavailability,¹⁰¹ which may need to be accounted for when using whole-food plant protein sources, as plant-based protein powders are typically stripped of their fibre content. Vegetarian or vegan carb-restricted dietary patterns should be well formulated to meet protein requirements. It may be necessary to consider the higher carbohydrate content of protein sources, and supplementation should be considered if needed. Vulnerable patients (e.g. those with sarcopenia or cachexia) may benefit from more precise protein goals. An estimated intake of at least 25 g – 30 g of high-quality protein per meal is suggested for maximum stimulus of muscle synthesis across all ages.^46,94

Fat

A well-formulated carbohydrate-reduced diet includes full-fat dairy and other whole foods containing a variety of fatty acids (i.e. unprocessed meats, avocados and olives), which are not typically restricted.

Saturated fats, containing different types of saturated fatty acids (SFA), are required for a range of key biological functions and are required for the absorption of fat-soluble vitamins (A, D, E and K). Despite concern in recent decades, current evidence builds on previous analyses, which do not support the restriction of saturated fat for CVD prevention.^102,103,104

Industrial trans fats, produced by the partial hydrogenation of vegetable oils, should be avoided because of their adverse impact on metabolic health via inflammation, endoplasmic reticulum stress and lipid metabolism.^105,106 Many countries have banned or capped the use of trans fats, and these fats are generally contained in processed foods, which are not recommended in TCR. Naturally occurring trans fats are produced in the stomachs of ruminant animals and found in dairy and meat. These subtypes do not demonstrate harmful effects at levels typically consumed (trace amounts) and appear neutral or even mildly beneficial.^107,108

Contraindications and cautions

Absolute contraindications

The list of absolute contraindications is straightforward and related to inborn errors of metabolism (see Table 8).

Cautions: Specialist medical or dietary management required

The list of relative contraindications and cautions is broad and nuanced. Some concerns in this category are based on a lack of data rather than on evidence supporting the risk of adverse events.¹¹⁰ As such, medical management, medication adjustments and specialist dietary oversight may be required. While ketogenic-level carbohydrate reductions provoke metabolic changes that may be more challenging for some disease states, less severe carbohydrate reductions (e.g. low- or moderate-carbohydrate approaches) may still confer metabolic benefits while allowing a more cautious approach. A general overview, agreed by consensus, can be found in Table 9.

The following sections review the emerging evidence base for some of the conditional categories found in Table 9.

Kidney disease

A complicating comorbidity because of historical data and differences with emerging TCR data is chronic kidney disease (CKD). Historically, there have been concerns about protein consumption in the context of low-carbohydrate diets and the potential for harm in patients with CKD.¹¹⁸ However, a recent Cochrane review compared low protein (0.6 g/kg/day – 0.8 g/kg/day) with unrestricted (≥ 1.0 g/kg/day) protein for at least 12 months and found uncertainty regarding the benefits of protein restriction.¹¹⁹

Practically, target protein levels for reduced carbohydrate diets often fall between 1.2 and 1.5 g/kg body weight, which is comparable to average protein intakes when consuming a standard American diet (at approximately 15% – 20% of daily energy expenditure) and can be below the ‘high protein’ threshold thought to be between 1.5 and 2.0 g/kg.^46,89,120

In patients with kidney disease, there is uncertainty about optimal levels of protein intake, especially where nutritional sufficiency is compromised.¹²¹ A meta-analysis from 2018 examining the effect of protein restriction on CKD progression concluded:

Based on the findings of our meta-analysis, protein-restricted diet may reduce the rate of decline in renal function and the risk of kidney failure for CKD populations, but did not produce a clear beneficial effect for all-cause death events … the optimal level of protein intake in different participants is left unanswered, and the nutritional status should be regarded with caution.¹²¹

A prudent approach in individuals with CKD is to aim for 0.8 grams of protein per kg per day, which is feasible with TCR and aligns with a clinical practice guideline (KDIGO 2024).¹²² The guideline suggests avoiding high protein intake (> 1.3 g/kg body weight/d) in adults with CKD at risk of progression. They make allowance for clinical judgement in cases of frailty and sarcopenia where higher protein targets may be considered, and protein restriction is not recommended in children with CKD.¹²²

Table 10 summarises data on several studies using various forms of therapeutic carbohydrate reduction in patients with mixed grades of CKD (see Table 10).

Tirosh et al. conducted a 2-year randomised controlled trial in individuals with obesity, with or without T2D, comparing low-carbohydrate (20 g/day for 2 months, increasing gradually up to a maximum of 120 g/day), Mediterranean and low-fat diets. They found that a low-carbohydrate diet did not adversely affect renal function, as measured by estimated glomerular filtration rate (eGFR) and urinary microalbumin-to-creatinine ratio, in moderately obese participants with or without T2D.¹²³ Friedman et al. conducted a secondary analysis of a long-term trial that compared a low-carbohydrate diet (20 g/day for 12 weeks, followed by a gradual increase in carbohydrate intake to maintain weight stability) with low-fat diets in obese adults. The study found no significant harmful effects on renal filtration indices, including serum creatinine, cystatin C and creatinine clearance, over a 2-year period.¹²⁴ Tay et al. conducted a randomised trial comparing a very low carbohydrate, high-protein (14% energy as carbohydrate [CHO < 50 g/day], 28% protein [PRO], 58% fat [< 10% saturated fat]) diet to a high-carbohydrate diet (53% CHO, 17% PRO, 30% fat [< 10% saturated fat]) in individuals with T2D.¹²⁵ The study found no decrement or difference between groups on clinical markers of renal function, such as eGFR and albumin excretion rate, over 12 months.

Recent evidence has shown TCR to be safe for patients with mild CKD, and a number of studies suggest TCR may result in improved kidney function.^{89,127,128,129} In a recent study of adults with T2D, a ketogenic diet (initial CHO < 30 g/day, protein target 1.5 g/kg of reference body weight) was linked to improved (from stage 3 to stage 2 or stage 4 to stage 3) or stable kidney function and reduced inflammation over 2 years, with greater benefits observed in those achieving higher levels of sustained ketosis.⁹⁰ A small study using a very low carbohydrate diet (CHO ≤ 30 g/day, protein not specified) has demonstrated improvements in metabolic health with reduced medication burden in patients with grade 3–4 kidney disease and stabilised renal function (15/18 participants improved eGFR score).¹²⁶ Further studies are needed in the setting of advanced CKD, but there is growing evidence that TCR may be beneficial in those with CKD whose renal function is less compromised.

For autosomal-dominant polycystic kidney disease (ADPKD), a ketogenic diet has been shown to be feasible^130,131 and has demonstrated early, promising results, with the potential to improve kidney function.^132,133 Cukoski et al. recommended 0.8 g of protein/kg of body weight per day, and maximum carbohydrate intake was set at 30 g.¹³¹

Liver disease

There is an emerging body of evidence supporting the use of TCR for MASLD, previously referred to as non-alcoholic fatty liver disease (NAFLD). This term acknowledges metabolic dysfunction as a driving cause of the disease and describes a range of liver diseases from hepatic steatosis to steatohepatitis (inflammatory), which may lead to fibrosis and cirrhosis.⁷⁷ Watanabe et al. reviewed evidence and concluded a ketogenic diet to be safe and therapeutic for MASLD (framed as NASH in the review) and suggested it could be considered until end-stage liver disease under the evaluation and care of a skilled hepatologist.^110,134 A very low-calorie ketogenic diet was initiated for two patients with end-stage liver disease to help them meet eligibility for a liver transplant. Both patients lost weight without adverse effects.¹³⁴ A number of studies have shown the benefit of TCR for MASLD,⁷⁷ with some evidence that more severe steatosis demonstrates a greater treatment response compared to milder disease.¹³⁵

Though more research is needed, overall, TCR has demonstrated a benefit for MASLD, which should be considered on a case-by-case basis by an experienced clinician with appropriate monitoring.

Pregnancy and breastfeeding

There is limited data on the safety of prolonged ketosis for foetal development; glucose availability is essential for growth. Healthy pregnant women undergo metabolic fluctuations throughout pregnancy to support foetal growth and development. Studies show that pregnant women appear to enter ketosis more readily, and this is accentuated in the third trimester.¹³⁶ This may contribute to the rare cases of ketoacidosis occurring in pregnancy, usually provoked by concurrent illness.⁶² This susceptibility extends to lactation, where cases of lactation ketoacidosis have occurred.¹³⁷ In addition, pregnant women with abnormal glucose metabolism are more likely to enter ketosis.

Various guidelines acknowledge pregnancy is a ketogenic state, and the emphasis is on avoiding ketoacidosis in the context of gestational diabetes mellitus (GDM).^138,139 Monitoring ketones is suggested by some bodies.¹³⁸ There is a lack of studies specifically addressing mild ketosis during pregnancy. However, it is known that ketone bodies diffuse across the placenta and the placenta contains enzymes capable of converting acetyl-CoA into beta-hydroxybutyrate.¹³⁶ There is some evidence that ketones provide energy for the foetal brain, possibly contributing to brain development, though overall it is difficult to draw conclusions about the effect of elevated ketone levels on the fetus.¹³⁶ Healthy neonates are often in ketosis.¹⁴⁰ A review by Tanner et al. states:

While it is difficult to draw strong conclusions on the safety of ketones and carbohydrate restriction in pregnancy, we believe that there is insufficient evidence to support current recommendations on necessary carbohydrate intake and avoidance of ketones.¹³⁶

If clinically indicated (e.g. diabetes or other conditions being managed by TCR), measuring ketones in this situation could offer a way to safely manage carbohydrate reduction in pregnancy and avoid ketonuria. The American Diabetes Association (ADA) suggests ketones can be measured to identify patients who are ‘severely restricting carbohydrates to manage blood glucose’.¹⁴¹ Including starchy vegetables or low-glycaemic index (GI) fruit in the diet to prevent ketosis is one approach to consider (ensure carbohydrate intake exceeds 50 g/day). The ADA also states that a ‘food plan should provide adequate calorie intake to promote foetal/neonatal and maternal health, achieve glycemic goals and promote appropriate weight gain’,¹⁴¹ which may be achieved using a variety of nutritional approaches. An excellent resource for women and clinicians may be found in Real Food for Pregnancy by Lily Nichols.¹⁴²

Historically, gestational diabetes was managed successfully using reduced carbohydrate approaches (30% – 40% of total calories from carbohydrate), where the goal was to avoid ketonuria but prevent postprandial hyperglycaemia.^143,144 Jovanovic-Peterson et al. studied women with insulin-dependent diabetes with the aim of achieving euglycaemia but avoiding ketonuria.¹⁴⁴ Carbohydrate intake was adjusted to 30% – 40% TEI to keep 1-h postprandial blood glucose < 140 mg/dL (7.8 mmol/L) and avoid ketosis. For example, a 60 kg woman (deemed to be in the ‘normal weight’ range, assigned 30 kcal/kg) would be given 90 g of protein (~1.5 g/kg or 20% TEI) with 30% carbohydrate, estimated at 135 g/day, depending on biofeedback. Their data on 52 women using this approach (with documented euglycaemia) found the rate of complications in the control group (non-diabetic) exceeded that of the intervention group (with insulin-dependent diabetes).¹⁴⁵

More recently, a low GI diet (carbohydrate-reduced diets sit at the lower end of the low GI spectrum) has been associated with less frequent insulin use and less macrosomia than control diets.¹⁴⁶ Ásbjörnsdóttir et al. found that ‘lower daily carbohydrate consumption than recommended by the Institute of Medicine is common among women with T2D in early pregnancy in Denmark’.¹⁴⁷ A small percentage of the women with type 1 and T2D reported consuming less than 100 g of carbohydrates daily (mean 83 and 78 g, respectively), with no ketonuria reported at the first antenatal visit.¹⁴⁷ Mijatovic et al. randomised pregnant women to a moderately carbohydrate-reduced diet (~135 g/d carbohydrate) or routine care (~200 g/d), looking at blood ketone concentration, risk of ketonaemia and pregnancy outcomes. While difficulties with adherence prevented definitive conclusions, these data suggest a modest reduction of carbohydrates is feasible without ketonuria, which may be monitored.¹⁴⁸ An excellent resource for women and clinicians may be found in Real Food for Gestational Diabetes by Lily Nichols.¹⁴⁹

Breastfeeding can increase the risk of lactation ketoacidosis, a rare but serious condition. If TCR is clinically indicated, consider avoiding ketonuria by incorporating starchy vegetables or fruit (> 50 g carbohydrate/day) and monitoring ketone levels. The literature reports the composition of the breast milk may be higher in protein content but otherwise unremarkable.¹⁵⁰ If medically indicated (e.g. mother or infant with epilepsy), reduced carbohydrate approaches may be successful with appropriate monitoring and medical supervision.¹⁵¹

Gout and gallbladder disease

Both gout and gallbladder disease are strongly associated with insulin resistance, and, overall, TCR may be expected to benefit these conditions.

In gallbladder disease, it is thought that low HDL-c and high triglycerides increase the cholesterol saturation index, promoting the development of cholesterol crystals, which are a precursor to gallstone formation. Diets higher in fat have been shown to be protective for the development of gallstones during periods of weight loss,¹⁵² and TCR improves triglyceride and HDL-c markers.¹⁵³ A more gradual transition to increased fat in the diet is advised, especially where a cholecystectomy has been performed.¹⁵⁴

Patients with a history of gout are at a higher risk of flare when transitioning to the diet, although long-term gout flares may improve on low-carb diets.¹¹⁵ Short-term increases in uric acid concentration may occur as ketone bodies and uric acid compete for the same excretion pathways. This resolves over 2–8 weeks, but consistent adherence may be preferable in this setting.⁸⁹ Consider prophylaxis during transition.

Caution because of medication adjustments required: Type 2 diabetes, type 1 diabetes, hypertension

Patients with metabolic disorders can expect rapid improvements in metabolic health and metabolism, which often require medication adjustments in the first few days and fall into these categories:

• Type 2 diabetes
• Type 1 diabetes
• Hypertension

See ‘Medication and monitoring’ section for details on medication adjustments and protocols.

Special populations: Neurology, psychiatry, oncology

Emerging applications for therapeutic ketogenic diets, or ketogenic metabolic therapy, are outside the scope of this article. They may require expert support to meet the therapeutic targets associated with metabolic ketogenic therapy (the ketone target range may be 1.5 mmol/L – 5 mmol/L), and in psychiatry, potentiation effects and temporary worsening of symptoms may necessitate medication adjustments.^22,23,24,116 The use of exogenous ketones is under investigation as an adjunct in these populations, where ketone concentrations in the higher range may be more beneficial.¹⁵⁵

Implementation

Clinical implementation settings

Therapeutic carbohydrate reduction has been successfully implemented across a range of clinical settings, which include inpatient¹⁵⁴ and outpatient settings,¹⁵⁶ and UK National Health Service (NHS) general practice settings^71,157 and a rural Veterans Affairs primary care clinic.¹⁵⁸

Remote settings have been successfully employed using various technologies, including tracking metrics and one-on-one health coaching via text, enabling continued support and personalised dietary advice.⁴⁷ Online low-carb programmes have successfully delivered patient education alongside a clinical team,¹⁵⁹ and recently a team-based metabolic health management programme with real-time biofeedback and remote monitoring has been utilised for an employee health initiative.¹⁶⁰

Continuous glucose monitoring (CGM) provides real-time feedback on carbohydrate reduction approaches and has been used to reduce glycaemic excursions as part of a lifestyle intervention for newly diagnosed T2D.¹⁶¹ Participants highlighted CGM as the most impactful component. In another study, CGM provided no added benefit compared to fingerstick glucose monitoring in ketogenic nutrition counselling for people with uncontrolled diabetes, so inability to obtain or afford CGM should not be seen as a barrier.^162,163

Group coaching in a community setting, with medical supervision, is another method for delivering education and medical support for TCR interventions. This has the advantage of reaching more people and including a patient support structure¹⁶⁴ (see Table 11).

These examples demonstrate that a range of implementation strategies may be successful. Finding the correct approach will depend on your practice and patient population. The best tools are the ones that the patient will actively engage with, which may be a simple meal template or infographic (see ‘Clinician resources and further reading’ section).

Initial evaluation

The initial assessment includes a standard medical history and an evaluation of social support structures, such as:

• Medical history
• Current symptoms
• Medications – consider possible interactions, potentiation effects or deprescribing requirements (see ‘Medication and monitoring’ section)
• Other considerations:
  • Contraindications (see ‘Contraindications and cautions’ section)
  • Co-morbidities
  • Nutritional status
• Social considerations:
  • Existing habits
  • Support structures – family, friends and community
  • Living arrangements
  • Roles – who does food shopping and preparation
  • Other factors – example: economic, cultural, personal
• Patient readiness and facilitating behaviour change:
  • Discuss diet history and current health goals.
  • Address fears about failure related to prior unsuccessful diet attempts.
  • Address concerns regarding carbohydrate reduction.
  • Address specific food fears (orthorexia).
  • Assess readiness to change and provide guidance, reassurance and support, as needed.
  • Consider using the G.R.I.N (Goals, Resources, Increments, Noticing) resource.¹⁶⁵
• Baseline laboratory tests as indicated (see ‘Baseline tests’ section).
• Discuss home testing if necessary, e.g. blood glucose (glucometer or CGM), ketones and blood pressure. There may be clinical utility in measuring ketones for more complex patients who are embarking on TCR (type 1 diabetes, pregnancy, medication interactions, concurrent SGLT2 inhibition), where establishing a normal physiological range for the patient may be of benefit (see ‘Medication and monitoring’ section). Online food trackers may have utility according to clinical needs or patient preferences.

Baseline tests

Clinical tests:

• Weight
• Height
• Waist circumference
• Blood pressure

Blood tests:

• The Fasting Comprehensive Metabolic Panel (CMP) comprises 14 metrics, which include blood glucose, liver and renal function tests, as well as electrolyte and fluid balance. A fasting panel is not required if HbA1c is ordered.
• Fasting lipid panel (basic – measures total cholesterol, LDL-c, HDL-c and triglycerides).
• Haemoglobin A1c – screening for diabetes or for baseline in the setting known diabetes.

Urine tests:

• Urine albumin to creatinine ratio (ACR) – in the setting of diabetes.

Other tests

Consider these additional tests depending on the clinical circumstances:

• Body fat percentage on dual-energy X-ray absorptiometry (DXA) scan.
• Fasting total insulin, homeostatic model of insulin resistance (HOMA-IR) or postprandial insulin assay or Kraft protocol (if available and affordable).
• Thyroid-stimulating hormone (TSH), (free T3, free T4, reverse T3 and thyroid-related antibodies if indicated).
• 25-hydroxyvitamin D.
• High-sensitivity C-reactive protein (hsCRP).
• Advanced (NMR) lipid panel including Apolipoprotein B (ApoB) and lipoprotein(a) (LP[a]).
• Glucose tolerance testing (GTT).
• Erythrocyte sedimentation rate (ESR).
• Serum uric acid.
• B12 (if on metformin or vegan or low intake of animal-sourced foods).
• Coronary artery calcium (CAC) score, carotid intima-media thickness (CIMT), coronary computed tomography angiography (CCTA), stress electrocardiogram (ECG) or others as clinically indicated – for risk stratification and ongoing monitoring.
• C-peptide levels – to exclude undiagnosed insulin deficiency (T1DM), latent autoimmune diabetes of adulthood (LADA) and pancreatogenic diabetes. Note: The test should be accompanied by a glucose level in order to interpret the appropriateness of the C-peptide level. This test also helps clinicians determine whether insulin taper is possible.
• Medications potentially interacting with TCR approaches need plasma concentrations measured, e.g. lithium, valproic acid and warfarin (see ‘Medication and monitoring’ section for details).
• Secondary carnitine deficiency may occur in conjunction with drugs such as valproic acid or in patients on long-term ketogenic diet therapy and may need to be monitored¹⁴ (see ‘Epilepsy and psychiatric medications’ section).
• Routine care screenings, e.g. for diabetic retinopathy, are assumed. Note: In individuals with a history of diabetic retinopathy, rapid lowering of HbA1c can cause temporary exacerbation and should be avoided. More frequent assessment of retinopathy may be required.

Patient education and management

Education and support are critical features of successful TCR application, with goals of satiety and simplicity that underpin adherence. Key features include:

• Assess patient knowledge and experience of TCR and provide appropriate levels of education. Explore the reasons for discontinuing TCR if there is prior experience and review the associated challenges.
• Provide simple low-carbohydrate meal planning resources, which could include sample menus, culturally relevant recipes, shopping lists, meal templates, food substitutions and lists of prepared foods.
• Assist in creating a personalised plan that considers food preferences, lifestyles and health conditions while fostering an independent problem-solving approach.
• A gradual approach to TCR can be considered. One approach may be to slowly reduce added sugar and refined starches while increasing the intake of protein and/or fat. (This may be appropriate in patients with mental health diagnoses, particularly bipolar disorder, as mood disturbances may occur during transition.)¹¹⁶
• Schedule routine follow-up according to need (e.g. in person, by cell phone, by platform chat messaging, through Zoom or by email).
• Many examples of patient resources exist. Educational materials can be tailored to the patient population and therapeutic goals. Include family members if appropriate (see ‘Clinician resources and further reading’ section).

Adaptation side effects and treatment

Possible symptoms on initiation of TCR are well described in the literature and are generally short lived, beginning within 2–3 days and lasting a few days, though they may last up to 2–4 weeks.^114,166 These transitional symptoms vary between individuals and reflect an acute response to carbohydrate withdrawal as physiology shifts to a state of nutritional ketosis. The symptoms generally pass quickly, highlighting that it is a metabolic shift that causes them rather than the diet itself. Maintaining adequate energy intake during this phase may improve symptoms.¹⁶⁶ The reduction in carbohydrates, and subsequently, circulating glucose, decreases circulating insulin, promoting natriuresis and diuresis, the major driver of symptoms.

Commonly occurring symptoms are^114,166:

• Headache
• Fatigue
• Decreased exercise capacity
• Light-headedness
• Difficulty concentrating
• Mood changes
• Food cravings
• Muscle cramps and weakness
• Breath odour
• Constipation
• Other gastrointestinal disturbances

Rare side effects do occur, as with any potent intervention (e.g. other diets, weight loss medications and weight loss surgery). They are also features of metabolic and fluid shifts. These include insomnia, temporary hair loss, kidney stones (see Table 9), hypoglycaemia, reduced alcohol tolerance and rash.¹¹⁴ The appearance of a rash in certain individuals (may be referred to as ‘keto rash’ or prurigo pigmentosa) is suggested to be triggered by ketosis-induced neutrophilic and T-cell-mediated skin inflammation and resolves with dietary adjustments (increase carbohydrates to resolution, generally > 50 g/day).¹⁶⁷

Electrolyte balance

Maintaining hydration and sodium intake are key during adaptation, as this manages many of the symptoms. For most patients, 2 g – 3 g of sodium (or 5 g – 7 g of salt) per day is appropriate. This can be accomplished by salting food liberally or adding salt to bone broth.⁴⁶ The dose may be adjusted over time as per individual requirements. Proprietary brand electrolyte solutions may also be useful.

Potassium may also become depleted in the setting of inadequate sodium intake and can present as muscle twitches or weakness, cramps or irregular heartbeats (and constipation), and adequate potassium intake (3 g/day – 4 g/day), especially during adaptation, is recommended.^46,89,114 Consider use of ‘light salt’ containing potassium and sodium. Dietary sources of potassium include spinach, other leafy greens and vine fruits, e.g. tomatoes, cucumbers, zucchini (courgette) and eggplant (aubergine).

Medication interactions and adjustments may occur because of these effects (see ‘Medication and monitoring’ section).

Constipation

Constipation may also be associated with inadequate fluid intake – increase to a minimum of 2 l per day. Fibre consumption (usually sufficient on a whole food TCR approach) can be explored either by increasing it (more vegetables or chia seeds and flax seeds) or decreasing it, depending on the underlying cause of constipation.¹⁰⁰ Other options include milk of magnesia (1 teaspoon) or magnesium citrate (240 mL, adult single dose, mixed with water) at bedtime,¹⁶⁸ additional sodium or a sugar-free fibre supplement (e.g. psyllium). If constipation persists, an individualised approach, examining potential triggers (e.g. an increase in the consumption of certain foods, such as nuts or cheese), may be required. Note: Reduced stool frequency can occur when reducing carbohydrate intake and may not be pathological (constipation, defined as hard, painful stools that require straining to pass, can occur because of a change in sleep rhythm, dehydration, iron supplements or the need to adjust fibre intake).

Muscle cramps

Muscle cramps are a common side effect, especially in the early days and weeks after initiating a low-carbohydrate diet, also influenced by electrolyte imbalance. Magnesium supplementation can be supportive if attention to hydration and sodium is insufficient. Ensure there are no contraindications to use, e.g. impaired renal function, pregnancy or neuromuscular disease.¹⁶⁹ Options include slow-release magnesium chloride or magnesium glycinate (magnesium bisglycinate), supplying 200 mg – 400 mg of elemental magnesium daily. Magnesium glycinate is well absorbed and is less likely to cause gastrointestinal symptoms. Dosage can be decreased to 200 mg/day for maintenance.

Other suggestions include a spoonful of pickle juice or yellow mustard. The mechanism is not fully understood but is thought to occur via an inhibitory oropharyngeal reflex.¹⁷⁰ If muscle cramping increases with exercise or activity, use of a foam roller may benefit.

Medication and monitoring

Carbohydrate reduction initiates changes in metabolism via substrate availability and insulin signalling that impact diabetes medications, electrolyte balance and blood pressure, necessitating medication review, adjustment or discontinuation. Home monitoring of blood glucose and blood pressure is essential for feedback on adaptation and to help guide medication adjustments. The patient should be advised to check blood pressure and, if applicable, blood glucose daily, including some postprandial readings 1–2 h after a meal. These tools greatly assist a safe transition to a low-carbohydrate diet. Paper or digital tools for logging daily readings are available (see ‘Clinician resources and further reading’ section. for an example of a self-monitoring glucose log). The measurement of ketones has clinical utility for patients with metabolic disease in certain situations, e.g. patients using SGLT2 inhibitors.

The following sections highlight key points that influence medication management during adaptation, including the effects of medication and adjustments for diabetes and hypertension, as well as considerations for epilepsy and psychiatric management. A recent review by Marinescu et al. provides an in-depth discussion of potential interactions between the ketogenic diet and various medications and is recommended for further reading.¹⁷¹

In addition, as metabolic health improves, related conditions may improve, and monitoring for other medication adjustments can be considered, e.g. gastroesophageal reflux disease (GERD), polyendocrine metabolic ovarian syndrome (PMOS), joint pain and migraine.¹⁷²

Adaptation and electrolytes

During adaptation, extra attention should be paid to sodium and hydration status for patients on multiple medications.¹⁷³ Patients with heart failure, CKD or salt-sensitive hypertension will need more careful monitoring. Sodium intake should be kept at baseline until oedema resolves, then increased if the patient has orthostatic symptoms.^115,173

Potassium may also become depleted in the setting of inadequate sodium intake. Medications that interfere with potassium balance (sparing or wasting effects), e.g. thiazides and diuretics, amiloride and spironolactone, should be monitored by the medical team. Attention should be given to adequate dietary potassium and sodium intake, especially in patients at high risk, such as those on digoxin therapy.¹⁷⁴ Supplementation should be considered if hypokalaemia is persistent.^173,175

Hyponatraemia may be exacerbated by many medications (see Table 12 for examples).

Other effects of medications

Depending on the sensitivity of the individual and the therapeutic targets (especially where higher levels of ketones are required), a critical examination of drug composition may reveal added carbohydrates in the pharmaceutical formulations.¹⁷¹

Examples of medications that may elevate blood glucose and interfere with ketosis include¹⁷¹:

• Lactulose
• Inulin
• Corticosteroids (oral, inhaled, injected or topical steroids)
• Statins
• Antipsychotics
• Beta blockers
• Thiazide diuretics

Clinicians should be aware of medications that can interfere with lipolysis and could consider alternatives if clinically reasonable, considering therapeutic responses on an individual basis. Examples include:

• Niacin¹⁷⁸
• Beta blockers¹⁷⁹
• Some antidepressants, e.g. fluoxetine¹⁸⁰
• Some second-generation antipsychotics, example: olanzapine and quetiapine¹⁸¹

Warfarin doses may need to be adjusted, and international normalised ratio (INR) should be monitored more frequently during the diet transition.¹⁷¹ This need is primarily driven by the interaction of vitamin K with warfarin. As TCR includes many foods that are rich in vitamin K (e.g. chicken liver, egg yolks, hard cheeses and leafy greens), clinicians may need to make dietary or medication adjustments.¹⁷¹

Diabetes and antihypertensive medications

Reducing dietary carbohydrates lowers the blood glucose response, and therefore rapid changes to medication may be necessary to avoid hypoglycaemia. As insulin drops, natriuresis and diuresis occur along with changes in the renin-angiotensin-aldosterone system (RAAS) influencing blood pressure.

Home monitoring of blood glucose (or CGM) and blood pressure is essential for guiding medication adjustments. See Box 1 for considerations for dosing adjustments, and see Table 13 for a summary of specific medications and considerations.

If patients are on insulin when beginning a low-carbohydrate diet, it is generally recommended that doses be reduced immediately to prevent hypoglycaemia. Cases of poor glycaemic control may need less aggressive dose reduction initially and should be considered in the context of blood glucose drops commonly seen (50 mg/dL – 100 mg/dL or 3 mmol/L – 6 mmol/L). In well-managed patients, reducing insulin dosing at initiation is necessary to avoid hypoglycaemia at the possible expense of hyperglycaemia. The first 2–3 days, when metabolic shifts are large, carry the greatest risk of severe lows.

Initial medication adjustments need to be individualised and subject to daily review in the initial stages. Precise adjustments are to be made as clinically indicated (See Box 1 and Table 13).

Many patients using TCR and having sufficient C-peptide levels may be able to discontinue insulin. Adult patients with type 2 diabetes with stable blood glucose can often successfully stop insulin once their dose is lower than 10 units per day (use C-peptide to confirm endogenous insulin production).

Before stopping insulin, it is important to assess the risk (clinical assessment and labs) of undiagnosed insulin deficiency (T1DM) or latent autoimmune diabetes of adulthood (LADA) and pancreatogenic diabetes.

The following factors are suspicious for insulin deficiency:

• Younger age at diagnosis.
• Requiring insulin treatment within 5 years of diagnosis of T2D.
• Continued requirement for insulin during periods of weight loss or bariatric surgery.
• Labile blood glucose (standard deviation of ≥ 50 mg/dL [2.8 mmol/L] is suspicious).
• Low body weight, body mass index (BMI) < 30 and/or not overweight at diagnosis (BMI < 25). For Asian populations, the BMI thresholds are lower, that is, on average, BMI < 25 (instead of 30) is the cutoff for obesity and not overweight at diagnosis at the BMI of < 23.
• Normal triglycerides and high HDL-c.
• Personal or family history of autoimmunity.
• History of DKA.
• History of total or partial pancreatectomy or history of pancreatitis (may cause a non-LADA, non-autoimmune form of insulin deficiency).

If there is clinical suspicion of insulin deficiency, it is recommended to check a fasting or stimulated C-peptide level to confirm sufficient endogenous insulin production.

Epilepsy and psychiatric medications

Ketogenic dietary therapy may alter the pharmacokinetics of epilepsy and psychiatric medications as part of its therapeutic effect, where synergistic interactions can be additive or potentiating.^116,171 Metabolic changes can result in decreased or increased serum concentrations of some medications, and the higher fat content of reduced carbohydrate interventions may affect drug absorption.¹⁷¹ A detailed commentary is outside of the scope of this article. Table 14 provides some examples of medications and interaction effects to be aware of. Secondary carnitine deficiency can occur in this population.^14,183

Other considerations

LDL increase

A temporary rise in LDL-c may be observed during weight loss as a physiological response to fat mobilisation and the shift in metabolism. Often LDL-c decreases or returns to baseline.^184,185

In general, low-carbohydrate diets have been shown to be effective at increasing HDL-c and decreasing triglyceride (or triacylglycerol [TAG]) with minimal change in LDL-c or total cholesterol.¹¹ In addition, studies show improvements in low-density lipoprotein (LDL) particle number and atherogenic profile using TCR approaches.^73,186 However, some uncertainty remains until CVD outcome studies are performed.

Recent data have identified the ‘lean mass hyper-responder’ (LMHR) phenotype, which may develop as weight loss occurs or in lean individuals adopting a low-carbohydrate approach for other health reasons.¹⁸⁷ In this population, these data indicate that a lower BMI tends to promote the phenotype.^187,188 Typically, these individuals are metabolically healthy with cholesterol readings defined as LDL-c ≥ 200 mg/dL (≥ 5.2 mmol/L), HDL-c ≥ 80 mg/dL (≥ 2.1 mmol/L) and TAG ≤ 70 mg/dL (0.8 mmol/L). It is still to be determined if these findings represent an increased risk for CVD in the context of good metabolic health and low inflammation, but early data have shown no association between LDL-c and plaque burden.¹⁸⁹ Consider cardiac imaging to monitor and/or stratify risk.^190,191

The cholesterol response to TCR is heterogeneous, and the interpretation of LDL-c changes requires consideration in context with an appropriate clinical assessment (lipid panel, subfractions, imaging and other metabolic health markers). Cholesterol profiles may be altered (LDL-c tends to reduce) by increasing carbohydrates (from starchy vegetables or fruit) to 50 g/day – 100 g/day, which may have a role as determined on a case-by-case basis.¹⁸⁷ TCR may be adjusted as clinically indicated according to individual responses and can accommodate a broad range of dietary patterns.

Nutritional troubleshooting

Carbohydrate reduction for metabolic disorders emphasises a wide variety of plant and animal foods with high nutrient density and will provide good nutrition for the majority of individuals.^192,193 Our current food environment, high in ultra-processed foods, may promote nutritional deficiencies in individuals, and baseline testing for nutritional deficiencies may be required. Globally, deficiencies in iron, vitamin A and iodine are the most common, particularly among children and pregnant women.¹⁹⁴ Individuals following a vegan or vegetarian dietary pattern may be more at risk of iron deficiency. Symptoms suggestive of a nutritional deficiency should be investigated.

Clinical experience suggests vitamin D and zinc are commonly deficient. Vitamin B12 and folate testing may be needed (especially if patients take metformin). However, methylmalonic acid and total homocysteine may be important to accurately diagnose a B12 deficiency.¹⁹⁵

Omega-3 long-chain polyunsaturated fatty acids (PUFAs) are essential to our diet, as they cannot be synthesised endogenously and are required for many biological functions associated with reducing inflammation, modulating lipid metabolism, brain function and neuroprotection.¹⁹⁶ Dietary sources of omega-3 PUFAs primarily come from fatty fish, and the Dietary Guidelines for Americans recommend 2 servings of fish per week. Other dietary sources include leafy greens, seeds and nuts, but, in general, omega-3 PUFAs are less common in the diet compared to omega-6 PUFAs, and supplementation may be required for sufficiency.¹⁹⁷

Magnesium is an essential mineral that recent research has implicated as a cofactor for over 600 enzymes involved in metabolism, including protein synthesis, nerve transmission, muscle function and glucose and insulin metabolism.¹⁹⁸ This mineral is commonly underconsumed (impacted by soil depletion and food processing) and is implicated in glucose control. Low-carbohydrate sources of magnesium (if tolerated) include dark green vegetables, nuts, seeds, unprocessed meats and homemade broths.¹⁹⁹

In cases of calcium deficiency, sources include dairy foods, particularly hard cheeses, which are commonly consumed on reduced-carbohydrate diets. Other sources of calcium include small fish with bones (e.g. sardines), nuts and seeds, tofu and homemade bone broths. The recommended dietary allowance for calcium in adults is 1000 mg/day – 1200 mg/day.¹⁹⁹ If patients are at risk, e.g. osteoporosis, and are not getting adequate dietary calcium, a supplement, such as calcium citrate (with vitamin K2), may be considered.²⁰⁰ In this setting, vitamin D adequacy should also be considered, as it increases intestinal absorption of calcium.²⁰¹

Follow-up and maintenance

Clinicians should work with patients to empower them to take charge of their health. Setting health goals together and providing patients with the resources and support to reach them is an important part of patient care. Having the patient record weekly waist circumference and body weight measurements may also be helpful in monitoring progress if changes in those markers are a goal.

Close supervision is important during the transition to TCR to facilitate medication adjustments and help establish new dietary habits. A team approach, comprising various health professionals (which may include health coaches, nurses, dietitians or pharmacists), can provide ongoing support either individually or in a group setting. Including carers or other family members may be appropriate. Changing circumstances may necessitate a re-evaluation (see ‘Initial evaluation’ section), and the food environment, ongoing support structures and social circumstances may require continuous evaluation as needs evolve.

Individual and cultural considerations

Early guidance and support should assist individuals in finding appropriate food substitutions and include tips for navigating social situations. Setbacks and adherence challenges can be successfully navigated, though practitioners should be aware of behaviour patterns suggestive of trigger foods or a tendency towards emotional eating. A discussion on symptoms of processed food addiction (which may be a complicating feature of obesity)^202,203 is outside the scope of this article. In general, cravings diminish when adopting TCR,^45,204 but individual responses may vary, so additional support should be available if needed.

Financial constraints need not be a barrier to TCR approaches, which have been shown to be comparable in cost to standard diets,²⁰⁵ and have been successfully employed in diverse and under-resourced communities.^164,206

Accounting for individual preferences and cultural foods is important for long-term sustainability. Developing a collaborative approach towards food substitutions and considering ancestral foods (which may be more likely to be lower in carbohydrates or have seasonal availability) could help individuals find long-term solutions that align with their culture and preferences.

Therapeutic carbohydrate reduction is an adaptable whole-food approach that can be individualised according to preferences, budget and cultural or religious practices.

Long-term sustainability, adherence and quality of life

While the therapeutic benefits of TCR are evident, the long-term continuation of the diet remains a topic of discussion.

Clinical experience, along with emerging data, is showing that TCR can be sustainable in the long term, with data (5–10 years) from populations with T2D,²⁰⁷, metabolically healthy populations,^189,208 and case studies in individuals with T1DM and GLUT 1 deficiency.^209,210 Improvements in well-being, mood, sleep and sweet cravings are also appreciated in healthy populations who may choose to continue with TCR once adapted.⁴⁵

Nutritional sufficiency using a whole-food TCR approach has been established,^192,193 so any decision to increase carbohydrates should be evaluated according to patient preferences and clinical markers. Other considerations, such as traditional or celebratory foods, may be considered at certain times if tolerated. It may be helpful to redefine traditions through less emphasis on food and more emphasis on social connection. Often individuals can appreciate the effects of overconsuming carbohydrates, and encouraging people to notice any changes, such as mood or energy levels, may be useful. Keeping a food diary specifically for these occasions can help with motivation and enable the connection of food choices to physical effects.

As with any intervention, regular monitoring, collaboration and necessary adjustments guide a personalised approach to TCR.

Conclusion

Approximately 90% of the US population has suboptimal metabolic health, which reflects a global crisis. Many healthcare providers are using TCR to successfully manage a range of conditions, and the evidence base is expanding monthly:

• TCR is safe and effective for a range of metabolic disorders, with growing evidence for MASLD, CKD and emerging evidence for PMOS.
• TCR, especially in the ketogenic range, is under investigation as an adjunct in the fields of neurology, psychiatry and oncology.
• A whole foods TCR approach is nutritionally sufficient and may be tailored to a wide range of cultural and personal preferences, focusing on protein, healthy fats from whole foods and vegetables.
• Adaptation effects are generally mild, short lived and easily managed.
• Even in healthy populations, people readily appreciate the benefits of TCR, and long-term adherence is both possible and enjoyable.

Therapeutic carbohydrate reduction is a validated and practical dietary intervention for metabolic disorders, but further research is necessary to refine approaches and explore limitations in understudied populations, such as different stages of kidney and liver disease, paediatric populations and type 1 diabetes.

Clinician resources and further reading

The following list reflects a selection of resources helpful in supporting the implementation of therapeutic carbohydrate reduction in metabolic disorders:

• Scientific references and research resources²¹¹
• The SMHP™ position statement on therapeutic carbohydrate reduction for type 1 diabetes¹⁹
• Therapeutic carbohydrate reduction in type 1 diabetes: A guide for dietitians and nutritionists⁵⁸
• Nutritional considerations during major weight loss therapy: Focus on optimal protein and a low-carbohydrate dietary pattern⁴⁶
• Ketogenic: The science of therapeutic carbohydrate restriction in human health¹⁷³
• Society of Metabolic Health Practitioners – Practitioner resources²¹²
• Patient self-monitoring glucose log²¹³ (Online Appendix 1)

Acknowledgements

The Steering Committee would like to thank the Delphi panel members for their input in the shaping of this consensus guideline, listed alphabetically:

Nadir Ali, MD; Amy Berger, CNS; Lori Calabrese, MD, MHP; Dr David Cavan MD, FRCP; Dr Gary Fettke M.B.,B.S., F.R.A.C.S. (Orthopaedic Surgery), F.A.Orth.A.; Tony Hampton, MD, MBA, ABOM, CPE, MSHNFM; Lily E. Johnston, MD, MPH, MHP; Brian Lenzkes, MD, MHP; Barbara Levy, MD; Prof. Tim Noakes; Caroline G.P. Roberts, MD; Roshani Sanghani, MD, MHP; Bret Scher, MD; Franziska Spritzler, RD, CDCES; Dr Dave Unwin, FRCGP; William S. Yancy, Jr. MD, MHS.

The steering committee acknowledges the late Adele Hite, PhD, MPH, RD, who coordinated and synthesised expert input for the original 2019 document. That document informed early drafting stages of the present work.

Sarah M. Rice serves as a consultant to the Society for Metabolic Health Practitioners (SMHP™) and receives compensation for these services. Sarah M. Rice received financial compensation for services related to the production of this manuscript. Robert Cywes serves as an editorial board member of this journal. Pamela L. Devine serves as operations manager of LowcarbUSA® and receives compensation for this role. Tro Kalayjian has received a grant from the Baszucki Foundation and has spousal ownership of a food company. Douglas B. Reynolds serves as the president of the SMHP and the CEO of LowcarbUSA® and receives compensation from both. Eric C. Westman is the owner of an Internet teaching company, Adapt Your Life, and is an editorial board member of the Journal of Metabolic Health. The authors have no other competing interests to declare.

Sarah M. Rice: Methodology, Project administration, Writing – original draft, Writing – review & editing. Laura A. Buchanan: Writing – review & editing. Matthew W. Calkins: Writing – review & editing. Mark T. Cucuzzella: Conceptualisation, Methodology, Writing – review & editing. Robert Cywes: Conceptualisation, Writing – review & editing. Pamela L. Devine: Writing – review & editing. Miriam Kalamian: Writing – review & editing. Tro Kalayjian: Conceptualisation, Resources. Douglas B. Reynolds: Conceptualisation, Project administration, Supervision, Visualisation, Writing – review & editing. Eric C. Westman: Writing – review & editing. All authors reviewed the article, contributed to the discussion of results, approved the final version for submission and publication and take responsibility for the integrity of its findings.

The authors received no financial support for the research, authorship and/or publication of this article.

The authors declare that all data that support this research article and findings are available in the article and its references.

The views and opinions expressed in this article are those of the authors and are the product of professional research. They do not necessarily reflect the official policy or position of any affiliated institution, funder, agency, or that of the publisher. The authors are responsible for this article’s results, findings, and content.

References