DSX 17 (2023) 102905

Contents lists available at ScienceDirect

Diabetes & Metabolic Syndrome:

Clinical Research & Reviews
journal homepage: www.elsevier.com/locate/dsx

The effects of the ketogenic diet for the management of type 2 diabetes
mellitus: A systematic review and meta-analysis of recent studies
Kimberley Yu Ching Choy a, Jimmy Chun Yu Louie a, b, *
a
School of Biological Sciences, Faculty of Science, The University of Hong Kong, Hong Kong SAR, China
b
Department of Nursing and Allied Health, School of Health Sciences, Swinburne University of Technology, Hawthorn, VIC, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: Objective: To systematically review the effects of the ketogenic diet on glycaemic control, body weight, cardio­
Ketogenic diet vascular risk factors, and liver and kidney function in patients with type 2 diabetes.
Low carbohydrate diet Methods: PubMed, MEDLINE, Embase, Cochrane Library and CINAHL were searched for randomised controlled
Type 2 diabetes
trials published between 2001 and 2021 that compared the ketogenic diet to a control diet for effects on gly­
Meta-analysis
Randomised trial
caemic control, body weight, cardiovascular risk factors, liver and renal function markers in adults with type 2
diabetes for >14 days. Meta-analyses using fixed or random effects models were conducted.
Results: Nineteen reports from 11 randomised controlled trials were included. Compared to the control, the
ketogenic diet showed no significant difference in changes in glycaemic control or body weight, but greater
increases in HDL (standardised mean difference 0.19; 95%CI 0.02–0.37; I2 = 0 %; moderate-quality evidence)
and greater reductions in triglycerides (standardised mean difference − 0.41; 95%CI − 0.64 to − 0.18; I2 = 0 %;
low-quality evidence).
Conclusions: The ketogenic diet may improve lipid profiles but showed no additional benefits for glycaemic
control or weight loss compared to control diets in type 2 diabetes patients over two years.

1. Introduction medication regimens, resulting in inadequate glycaemic control and

increased cardiovascular disease risk factors [4]. This further un­
Type 2 diabetes (T2D) is the most prevalent form of diabetes, ac­ derscores the clinical significance of non-pharmacological approaches,
counting for over 90 % of all diagnosed cases. Its aetiology involves a such as dietary intervention.
combination of insulin resistance and inadequate compensatory insulin In recent years, there has been a growing interest in utilising a
secretion, often linked to obesity [1]. In 2017, the global prevalence of ketogenic diet (KD) to manage obesity and diabetes [5–7]. The KD is
T2D was estimated at approximately 462 million individuals (equivalent characterised as a low-carbohydrate diet with moderate protein intake
to 6059 per 100,000). Projections indicate that this number will rise to and unrestricted fat consumption, aiming to induce nutritional ketosis
552 million by 2030 (i.e., around 7000 per 100,000) [2]. [8,9]. Initially developed as a therapeutic approach for managing re­
The primary objectives of current T2D treatment are to prevent or fractory seizures in children with epilepsy during the 1920s, the KD has
delay complications and to optimise patients’ quality of life. These goals evolved [8]. The original therapeutic KD had a 4:1 fat consumption ratio
are achieved by effectively managing blood glucose levels, cardiovas­ to the combined carbohydrate and protein intake. However, the
cular risk factors (such as blood pressure and lipids) and body weight contemporary standard KD has become less restrictive, with fat consti­
[3]. The consensus guidelines endorsed by the American Diabetes As­ tuting approximately 70 % of daily energy intake, while carbohydrates
sociation and the European Association for the Study of Diabetes and protein account for around 10 % and 20 %, respectively [10].
recommend lifestyle interventions, including medical nutrition therapy Various versions of KD exist [11], including the Atkins diet and
and physical activity, as the initial treatment options upon diagnosis. medium-chain triglyceride KD (Table 1), each differing in aspects such
These interventions are also recommended as adjunct therapy for in­ as the number of phases and specific types of triglycerides or carbohy­
dividuals receiving antidiabetic medications or metabolic surgery [4]. drates consumed. However, they all share a common characteristic of
Notably, many T2D patients struggle with suboptimal adherence to severely limiting carbohydrate intake to induce ketosis, typically to <15

* Corresponding author. Swinburne University of Technology, Level 2 SPW Building, 1 John St, Hawthorn, VIC, 3122, Australia.
E-mail address: jimmylouie@swin.edu.au (J.C.Y. Louie).

https://doi.org/10.1016/j.dsx.2023.102905
Received 31 July 2023; Received in revised form 30 October 2023; Accepted 1 November 2023
Available online 15 November 2023
1871-4021/© 2023 The Author(s). Published by Elsevier Ltd on behalf of Research Trust of DiabetesIndia (DiabetesIndia) and National Diabetes Obesity and
Cholesterol Foundation (N-DOC). This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

Table 1 impact on liver and renal functions [25].

Types of ketogenic diets [7,8]. Other systematic reviews examining the ketogenic diet for diabetes
Type Main characteristics management have had several key limitations: 1) some focused broadly
on low-carb diets rather than ketogenic diets specifically [26–28]; 2)
Therapeutic KD (For - Fat to protein & carbohydrate ratio = 4:1
epilepsy) - Approximately 80–90 % energy from fat, the remaining several did not include meta-analyses to synthesise the results quanti­
10 % from protein & carbohydrate tatively [29,30]; and 3) none performed meta-analyses stratified by
Standard KD - Carbohydrate <50 g per day length of follow-up. Therefore, this systematic review and meta-analysis
- Approximately 70 % energy from fat, 10–15 % from aims to investigate and summarise the effects of the ketogenic diet,
carbohydrate, 20 % from protein
Atkins diet - Divided into 4 phases:
specifically in adults with T2D, while addressing the limitations of
• Phase 1: consume <20 g carbohydrate per day previous reviews. Through meta-analyses, we focused on synthesising
• Phase 2: consume 25–50 g carbohydrate per day the evidence for effects on glycaemic control, body weight, cardiovas­
• Phase 3: consume 80 g carbohydrate per day; remain in cular risk factors, and liver and kidney function. Analyses were stratified
the phase until the desirable body weight is reached
by follow-up duration to elucidate potential differences in short-vs
• Phase 4: consume up to 100 g carbohydrate per day
- Ketosis is observed in phase 1 only long-term effects. Additionally, the review analysed the effects of KD
Modified Atkins diet - Maintain phase 1 of Atkins diet indefinitely and do not on liver and renal functions in T2D patients to assess its safety.
shift to less restrictive phase, so as to sustain the ketosis We hypothesised that the ketogenic diet would improve glycaemic
state control and weight loss but potentially negatively impact cardiovascular
- Restrict carbohydrate and energy intake only, no
limitation is placed on fluid and protein consumption
risk factors and liver/kidney function markers compared to conven­
MCT KD − 40–65 % of energy from medium-chain triglycerides tional diabetes diets. Elucidating the risk-benefit balance is critical to
- Based on the rationale that medium-chain triglyceride is inform clinical recommendations regarding the ketogenic diet for pa­
more effective for production of ketone bodies than long- tients with T2D.
chain triglyceride
Low glycemia index - Allow consumption of 40–60 g carbohydrate per day
treatment - Restricted to carbohydrate with a low glycaemic index 2. Materials and methods
(<50)
- Recommended foods: vegetables, legumes, seeds, dairy, 2.1. Search strategy
and meat

KD, ketogenic diet; MCT, medium-chain triglyceride. This review was planned and conducted per the Preferred Reporting
Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines
% of daily energy intake or <50 g per day. Table 2 provides examples of [31], and the protocol was registered at PROSPERO (www.crd.york.ac.
the recommended and restricted foods for KD. uk/PROSPERO) as CRD42021282221. A systematic literature search
KD has been shown to be effective in promoting weight loss [12–14]. was conducted across multiple databases, including PubMed, Medline,
The hypothetical mechanisms by which KD may facilitate weight control Embase, Cochrane Library and Cumulative Index to Nursing and Allied
include reduced appetite attributed to the satiating effects of proteins, Health Literature (CINAHL) to identify relevant studies. The complete
appetite-regulating hormones, and ketone bodies. Additionally, KD has search strategy and the utilised search terms are provided in Online
been associated with enhanced metabolic efficiency in utilising fat, as Supplemental File 1. A manual search of the reference lists of the
evidenced by decreased resting respiratory quotient. Furthermore, KD is identified articles was performed to ensure the comprehensive inclusion
suggested to increase energy expenditure through gluconeogenesis’s of relevant studies.
energy-demanding process and protein utilisation’s thermic effect [5].
The suppression of pancreatic insulin secretion resulting from low car­ 2.2. Eligibility criteria
bohydrate intake and subsequent low blood glucose levels is also pro­
posed to contribute to the weight control effects of KD [15]. The retrieved studies were screened by title, abstract and full text by
Weight loss has effectively improved glycaemic control [16–20] and both authors to determine their eligibility for inclusion. In this review,
cardiovascular risk factors [21] among patients with T2D. However, the only randomised-controlled trials (RCTs) in English, which assessed the
impact of KD on lowering HbA1c, a marker of average blood glucose effects of KD on human adults with T2D, were included. Other types of
control in the past three months, has yielded inconsistent results in articles, such as observational studies, non-randomised or single-arm
studies [22–24]. Furthermore, controversies exist regarding the effect of clinical trials, case studies, conference reports, commentary, guide­
KD on cardiovascular risk factors and safety, particularly regarding its lines, trial protocols and reviews, were excluded. Studies were also
excluded if participants were not adults with T2D or were suffering from
other chronic diseases in addition to T2D (e.g., chronic renal failure).
Table 2 Trials with intervention durations shorter than 14 days were excluded to
Recommended and restricted foods for ketogenic diet [8,44,70]. ensure sufficient time for the KD to induce ketosis. Articles were further
Recommended Restricted excluded if they did not report any relevant outcome of interest. The
Protein: Tuna, sardine, prawns, Grains: Flour, macaroni, spaghetti, publication year was limited to the period between 2001 and 2021.
shrimps, lobster, salmon noodles, bread, rice
Meat: Kababs, sausages, minced, ham Vegetables and fruits: Starchy 2.3. Data extraction
Poultry: Chicken, eggs vegetables (e.g., potatoes, sweet
Cheese: Full-fat cheese, Mozzarella potatoes, corn, taro), fruits with high
cheese, Cheddar cheese sugar/carbohydrate content (e.g., apple,
Both authors extracted the following information into predetermined
Vegetables & fruits: Non-starchy and banana, grape) data extraction tables: 1) basic information of the articles (author, year
green-leafy vegetables (e.g. Spinach, Beverages: Fruit juices, soft drinks, and of publication, country of relevance); 2) baseline characteristics of
Watercress, Eggplant, Parsley, any sugary drinks participants (sample size, mean age, weight, BMI and HbA1c level); 3)
Mulberry, Coriander, Mint, Artichoke, Sweets: Cakes, cookies, brownies,
details of intervention (macronutrient constituents of the dietary inter­
Okra, Cabbage, Mushroom, Avocado, candy, chocolate, donuts, cinnamon
Leek, Carrot, Radish, Celery, rolls, ice cream, pancakes, waffles, and vention and any co-interventions, time of follow-ups); and 4) results of
Cauliflower, Green pepper, Lettuce, cereal outcomes of interest (mean ± SD changes and p-value of between-group
Cucumber, Tomato, 10–15 olives/day), difference). The outcomes of interest include glycaemic control (HbA1c,
Lemon, Strawberry: 6/day, Avocado, fasting blood glucose level, fasting insulin level and HOMA2IR),
Berries: 10/day
anthropometry (BMI, body weight and waist circumference),

2
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

cardiovascular risk factors (flow-mediated dilatation, systolic and dia­ were created to present the study-specific SMDs and their 95%CIs. The
stolic blood pressure, triglyceride and total-, LDL- and HDL-cholesterol), influence of the included studies on the results in the “overall” meta-
as well as renal functions indicators [estimated glomerular filtration rate analyses was examined using the InfluenceAnalysis function of the dme­
(eGFR), albuminuria and serum creatinine]. Outcome data were tar package in R. Leave-one-out analyses were also performed to assess
extracted at each follow-up time point provided by each trial. the impact of potential outliers on the pooled estimates (Supplemental
Fig. 2). A two-sided p < 0.05 was considered statistically significant.
2.4. Certainty of evidence assessment
3. Results
Both authors assessed the certainty of the evidence based on the
Grading of Recommendations Assessment, Development, and Evalua­ 3.1. Studies identified
tion (GRADE) approach, using the GRADEpro Guideline Development
Tool (Evidence Prime, Inc. 2020) [32]. The GRADE approach consisted Our search strategy returned a total of 612 articles, and three addi­
of four domains: study limitations, consistency of effect, indirectness, tional articles were obtained through hand search. After screening with
and publication bias. The studies were categorised into levels of cer­ the eligibility criteria, a total of 19 reports from 11 studies were deter­
tainty, including high, moderate, low and very low. Factors considered mined as eligible in this review. These included seven reports [35–41]
for upgrading the quality of the evidence included a substantial effect from the study by Tay et al.‘s group; two reports [23,42] from the study
size and the potential influence of plausible residual confounding. The by Breukelman et al.‘s group; two reports [43,44] from the study by
results of the certainty assessment can be found in Online Supplemental Saslow et al.‘s group; and one report for each of the following studies:
Table 1 (for the “overall” meta-analysis) and 2 (for the meta-analyses Barbosa-Yañez et al. [45], Saslow et al. [46], Goday et al. [47], Mayer
stratified by time point). et al. [48], Goldstein et al. [49], Iqbal et al. [50], Westman et al. [51]
and Dyson et al. [52]. The complete search process, selection of relevant
2.5. Risk of bias assessment studies and exclusion reasons are presented in Fig. 1.

The risk of bias for the included RCTs was evaluated using version 2 3.2. Study characteristics
of the Cochrane risk-of-bias tool (RoB2) [32]. The tool assessed various
aspects of the included articles, such as bias arising from the random­ Table 3 summarises the general characteristics of the included
isation process, bias resulting from deviations from intended in­ studies. Overall, this review captured 541 T2D patients originated from
terventions, bias due to missing outcome data, bias related to the these populations: Australian (n = 115, 1 study, seven reports) [35–41],
method of outcome measurement, and bias stemming from the selection American (n = 173, 4 studies, 5 reports) [43,44,46,50,51], British (n =
of reported results. The quality of the identified articles was classified as 58, 2 studies, 2 reports) [48,52], South African (n = 30, 1 study, 2 re­
overall ‘low risk’, ‘high risk’ or ‘some concerns’. ports) [23,42], and the following (1 report each): Spanish (n = 85) [47],
Israeli (n = 44) [49], and German (n = 36) [45]. The number of T2D
2.6. Meta-analysis patients who underwent KD intervention in the individual trials ranged
from 6 to 58, with a mean age at the baseline between 51.2 [51] and
The meta-analyses were performed in the R (version 4.1.3, R Foun­ 64.8 [43] years. Participants were, on average, obese, with baseline
dation), using the meta, metafor and dmetar packages. For studies where mean BMIs ranging from 33.1 [49] to 38.9 [42] kg/m2 and body weight
the SD of the mean difference was not given in the article, we imputed between 91.5 and 118.6 kg. Baseline HbA1c levels, reported in 15 re­
that based on the SDs of the mean at baseline (SDbaseline) and study ports from 9 studies [35–39,41,43–47,49–52], ranged from 6.5 [45] to
endpoint (SDfinal) using this formula: 9.0 % [49]. Follow-up periods ranged from 3 weeks [45] to 2 years [43,
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ 50].
( )̅ The included articles had substantial heterogeneity in the energy
SDchange = SD2baseline + SD2final - 2 × Corr × SDbaseline × SDfinal
intake and co-interventions. Three studies [36,38–42,47] restricted the
Assuming a correlation coefficient (Corr) of 0.5 [33]. If no SDbaseline carbohydrate intake to <50 g/day or 14 % of daily energy intake; two
and/or SDfinal were available, we used the average SDchange from other allowed 20–50 g/day [35,44,46]; two ≤40 g/day or 10 % of daily energy
included studies for the corresponding outcome [33]. Heterogeneity [45,52], two <20 g/day [37,51] and one each for <30 g/day [50] and
between the included studies was assessed by the Cochran Q statistic, ≤25 g/day [49]. In addition, one study [35–41] co-intervened with
with p < 0.1 indicating significant heterogeneity; the Baujat plots [34] supervised exercise programs; four encouraged increasing levels of
(Supplemental Fig. 1); and the I2 test. Potential publication bias was physical activity and provided other advice on lifestyle recommenda­
visually assessed using funnel plots based on the symmetry of the plot tions, including behavioural strategies for adherence, mindful eating
and confirmed using Egger’s linear regression test. and the importance of sleep in both KD and comparator groups [43,44,
A meta-analysis was conducted when two or more studies were 47,51,52]; and one study provided these lifestyle recommendations only
available for a specific outcome. Given that some studies included in the to the KD group [46].
analysis reported outcomes at multiple time points, stratified meta-
analyses were performed based on the duration of follow-up (0–3 3.3. Risk of bias
months, 3–6 months, 6–12 months, 12–24 months). The relevant strat­
ified analyses included the outcome data corresponding to each Appraisal of individual RCT’s risk of bias is presented in Fig. 2. Nine
respective time point. Where an outcome at the same time point was reports from six studies had some concerns about their randomisation
reported multiple times, we included only one report in the stratified process due to the absence of information on allocation sequence
meta-analysis. Only the data collected at the last follow-up of the study concealment [35,41,42,44,45,48–51], and four reports from three
were included in the “overall” summary statistics to avoid duplication. studies had a high risk of bias [43,44,46,51] as there were imbalances in
The effects of the KD on glycaemic control, anthropometry, cardio­ the baseline characteristics of participants between intervention and
vascular risk factors, and renal functions in patients with T2D were comparator groups. Concerning deviations from intended interventions,
meta-analysed and reported as standardised mean differences (SMDs) most studies had a low risk of bias, but 2 had high risks as there was
after the intervention with their 95 % CIs. When I2 < 50 %, indicating a insufficient information to determine whether appropriate analysis had
low heterogeneity, results from the fixed-effect model were used; been used to estimate the effect of assignment to intervention [48,51]. In
otherwise, results from the random-effect model were used. Forest plots addition, three studies had high risks of bias from missing outcome data,

3
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

Fig. 1. PRISMA flow diagram of identification, screening and inclusion of articles.

attributed to low participant retention rates with no reported dropout significant difference in the change in FBGL between the KD and the
reasons [44–46,48]. All RCTs had low risks of bias in measuring the control group. When all 8 studies were combined, no statistically sig­
outcome and selecting reported results. Overall, five studies [43–46,48, nificant difference in changes in FBGL between KD vs control was found
51] were rated as having a high risk of bias, while another two [23,42, (SMDrandom: 0.36; 95%CI: 0.97 to 0.25; I2 = 81 %; very low-quality
47] were rated as having some concerns over their risk of bias. evidence). This remains statistically non-significant when data from
Mayer et al. [43] were removed, substantially reducing the heteroge­
3.4. Glycaemic control neity (SMDrandom: 0.08; 95%CI: 0.37 to 0.22; I2 = 29 %; Supplemental
Fig. 2f).
Changes in HbA1c among 247 subjects were reported in 14 reports Seven reports from four studies [23,37,38,40,43,44,51] reported
from 11 studies [35,37,38,40,41,43–46,48–52]. Of these, four presented changes in fasting insulin among 215 subjects (Fig. 4a). Statistically
changes in HbA1c at 0–3 months follow-up [44,49,51,52]; nine at 3–6 significantly greater reductions in fasting insulin among subjects
months [23,43,45–47,49–51]; six at 6–12 months [38,43,46,48–50]; following the KD were found at 0–3 months (two studies [44,51];
and two at 12–24 months [40,50]. Statistically significantly greater re­ SMDfixed: 0.56; 95%CI: 1.01 to − 0.12; I2 = 0 %; low-quality evidence)
ductions in HbA1c were only observed at 3–6 months (SMDrandom: 0.59; and 3–6 months (four studies [23,37,43,51]; SMDfixed: 0.33; 95%CI:
95%CI: 1.18 to − 0.01; I2 = 79 %; low quality evidence). However, no 0.62 to − 0.04; I2 = 0 %; moderate quality evidence) follow-up, but not
statistically significant differences in HbA1c reductions between the KD at 6–12 months. However, when results from all studies were combined,
and the control diet groups were found at other time points (Fig. 3a). no significant difference in the change in fasting insulin was observed
When combining the results from all 11 studies, there was no significant between KD vs control (four studies [23,40,43,51]; SMDfixed: 0.21; 95%
difference in the change in HbA1c between the KD group and the control CI: 0.48 to 0.06; I2 = 0 %; moderate quality evidence). When the results
group (SMDrandom: 0.33; 95%CI: 0.69 to 0.03; I2 = 64 %; low-quality of Westman et al. [51] were removed, the result became statistically
evidence). None of the studies included in the “overall” meta-analysis significant, although the magnitude was attenuated (SMDrandom: 0.13;
significantly influenced the heterogeneity of the results (Supplemental 95%CI: 0.24 to − 0.02; I2 = 0 %; Supplemental Fig. 2g).
Fig. 2h). For changes in HOMA2-IR (227 subjects; six reports from three
Changes in fasting blood glucose level (FBGL) among 450 subjects studies [37,38,40,43,44,47]; Fig. 4b), no statistically significant differ­
were reported in 10 reports from eight studies [23,37,38,40,44,47–51], ences between KD vs control were found at all follow-up time points, as
with most showing no significant changes (Fig. 3b). Meta-analyses of well as in the “overall” meta-analysis (SMDfixed: 0.22; 95%CI: 0.54 to
data at 0–3 months (three studies [44,49,51]), 3–6 months (six studies 0.10; I2 = 0 %; moderate quality evidence).
[23,37,47,50,51]), 6–12 months (four studies [38,48–50]) and 12–24
months (two studies [40,50]) follow-up showed no statistically

4
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

Table 3
Summary of the included studies.
Study, Country Follow-up Ketogenic diet Control diet Baseline Outcomes
duration characteristics (in assessed
mean ± SD)

Tay et al. [40] 16 [36], 24 Energy-restricted low CHO diet n = 58 Isocaloric high CHO diet n = 57 Ketogenic diet BMI: ✓ LDL: ✓
(and [35–39, [37], 52 [35, CHO: <50 g/d or 14%E CHO: 53%E group BW: ✓ TG: ✓
41]), 38,39,41] and Protein: 28%E Protein: 17%E Age: 58.0 ± 7.0 WC: ✓ SBP: ✓
Australia 104 [40] weeks Fat: 58%E (<10%ESFA, 35 % EMUFA & 13% Fat: 30%E (15%EMUFA, 15%EPUFA) year FBGL: ✓ DBP: ✓
EPUFA) ER: N/A BW: 101.7 ± 14.4 FIns: ✓ FMD: ✓
ER: 500 to 1000 kcal/d CoI: 60 min supervised moderate kg HbA1c: ✓ CRP: ✓
CoI: 60 min supervised moderate intensity intensity aerobic and resistance exercise BMI: 34.2 ± 4.5 HOMA2-IR: SCr: ✓
aerobic and resistance exercise 3 × /week 3 × /week kg/m2 ✓ eGFR:
HbA1c: 7.3 ± 1.1 TC: ✓ ✓
% HDL: ✓ UAlb: ✓
Control diet group
Age: 58.0 ± 7.0
year
BW: 101.6 ± 15.8
kg
BMI: 35.1 ± 4.1
kg/m2
HbA1c: 7.4 ± 1.1
%
Breukelman 16 weeks Low CHO HF diet n = 10 Continue usual diet and activities n = 13 Ketogenic diet BMI: × LDL: ✓
et al. [23] CHO: ≤50 g/d CHO: N/A group BW: × TG: ✓
(and [42]), Protein: No restriction Protein: N/A Age: 54.2 ± 12.7 WC: × SBP: ×
South Africa Fat: No restriction Fat: N/A year FBGL: ✓ DBP: ×
ER: Nil ER: N/A BW: 104.7 ± 14.2 FIns: ✓ FMD:
CoI: Nil CoI: N/A kg HbA1c: ✓ ×
BMI: 38.9 ± 6.1 HOMA2- CRP: ×
kg/m2 IR: × SCr: ×
HbA1c: 5.8 ± 0.7 TC: ✓ eGFR:
% HDL: ✓ ×
Control diet group UAlb: ×
Age: 58.3 ± 5.5
year
BW: 104.9 ± 32.9
kg
BMI: 38.2 ± 10.7
kg/m2
HbA1c: 7.8 ± 1.9
%
Barbosa- Yañez 3 weeks Very low CHO diet n = 16 LF diet n = 20 Both groups BMI: ✓ LDL: ✓
et al. [45], CHO: <40 g/d or 5–10%E CHO: 50%E Age: 63.0 ± 8.0 BW: ✓ TG: ✓
Germany Protein: 20–30%E Protein: 20%E year WC: × SBP: ✓
Fat: 60–70%E Fat: <30%E Ketogenic diet FBGL: × DBP: ✓
ER: reduced to 1200–1500 kcal/d ER: reduced to 1000–1200 kcal/d group FIns: × FMD: ✓
CoI: Nil CoI: Nil BW: 93.9 ± 20.7 HbA1c: ✓ CRP: ✓
kg HOMA2- SCr: ×
BMI: 32.1 ± 4.5 IR: × eGFR:
kg/m2 TC: ✓ ×
HbA1c: 6.7 ± 1.0 HDL: ✓ UAlb: ×
%
Control diet group
BW: 97.6 ± 22.6
kg
BMI: 32.7 ± 4.8
kg/m2
HbA1c: 6.2 ± 0.6
%
Saslow et al. 16 and 32 Very low CHO diet n = 13 ADA’s ‘Create Your Plate’ Diet n = 9 Ketogenic diet BMI: × LDL: ✓
[46], US weeks CHO: 20–50g/d (excluding fibre) All food proportions are based on a 9-inch group BW: ✓ TG: ✓
Protein: No restriction plate: half the plate is filled with non- Age: 53.0 ± 10.2 WC: × SBP: ×
Fat: No restriction starchy vegetables, one-quarter with year FBGL: × DBP: ×
ER: Nil carbohydrates, and one-quarter with lean BW: 109.7 ± 24.9 FIns: × FMD:
CoI: Lifestyle recommendations, including proteins kg HbA1c: ✓ ×
behavioural adherence strategies aimed at CoI: Nil BMI: N/A HOMA2- CRP: ×
increasing positive effect regulation & HbA1c: 7.1 ± 0.4 IR: × SCr: ×
mindful eating & encouragement on increase % TC: × eGFR:
level of physical activity & sleep Control diet group HDL: ✓ ×
Age: 58.2 ± 6.7 UAlb: ×
year
BW: 90.9 ± 16.4
kg
BMI: N/A
(continued on next page)

5
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

Table 3 (continued )
Study, Country Follow-up Ketogenic diet Control diet Baseline Outcomes
duration characteristics (in assessed
mean ± SD)

HbA1c: 7.2 ± 0.3

%
Saslow et al. 12 [44], 26 Low CHO ketogenic diet Moderate CHO calorie-restricted LF diet Ketogenic diet BMI: ✓ LDL: ✓
[43] (and [43] and 52 n = 16 n = 18 group BW: ✓ TG: ✓
[44]), US weeks [43] CHO: 20–50 g/d (excluding fibre) CHO: 45–50%E Age: 64.8 ± 7.7 WC: × SBP: ✓
Protein: As before intervention Protein:/Fat: instructed to lower fat year FBGL: ✓ DBP: ✓
Fat: No restriction consumption BW: 100.1 ± 26.4 FIns: ✓ FMD:
ER: Nil ER: 500 kcal/d kg HbA1c: ✓ ×
CoI: Encouraged to increase sleep and CoI: Encouraged to increase sleep and BMI: 36.2 ± 8.2 HOMA2-IR: CRP: ✓
exercise; provided supportive behavioural exercise; provided supportive kg/m2 ✓ SCr: ×
adherence strategies aimed at increasing behavioural adherence strategies aimed HbA1c: 6.6 ± 0.6 TC: × eGFR:
positive affect and mindful eating at increasing positive affect and mindful % HDL: ✓ ×
eating Control diet group UAlb: ×
Age: 55.1 ± 13.5
year
BW: 99.7 ± 24.2
kg
BMI: 37.4 ± 6.4
kg/m2
HbA1c: 6.9 ± 0.7
%
Goday et al. 16 weeks Very low CHO ketogenic diet n = 45 Low calorie diet n = 40 Ketogenic diet BMI: ✓ LDL: ✓
[47], Spain CHO: <50 g/d (from vegetables) CHO: 45–60 % group BW: ✓ TG: ✓
Protein: HBV protein, 0.8–1.2 g/kg of ideal Protein: 10–20 % Age: 54.9 ± 8.8 WC: ✓ SBP: ×
BW Fat: <30%E year FBGL: ✓ DBP: ×
Fat: 10 g of olive oil/d ER: 500 to 1000 kcal/d BW: 91.5 ± 11.4 FIns: × FMD:
ER: reduced to 600–800 kcal/d CoI: Provided exercise recommendations kg HbA1c: ✓ ×
CoI: Provided exercise recommendations & & individual counselling to support BMI: 33.3 ± 1.5 HOMA2-IR: CRP: ×
individual counselling to support lifestyle lifestyle and behavioural modifications kg/m2 ✓ SCr: ×
and behavioural modifications HbA1c: 6.9 ± 1.1 TC: ✓ eGFR:
% HDL: ✓ ×
Control diet group UAlb: ×
Age: 54.2 ± 8.0
year
BW: 89.5 ± 11.4
kg
BMI: 32.9 ± 1.6
kg/m2
HbA1c: 6.9 ± 1.0
%
Mayer et al. 12 weeks Low CHO diet n = 22 LF diet + orlistat n = 24 Ketogenic diet BMI: ✓ LDL: ✓
[48], UK CHO: ≤20 g/d CHO: No restriction group BW: ✓ TG: ✓
Protein: No restriction Protein: No restriction Age: 56.6 ± 7.3 WC: × SBP: ✓
Fat: No restriction Fat: <30%E (<10 % ESFA) year FBGL: ✓ DBP: ✓
ER: Nil ER: 500 to 1000 kcal/d BW: 118.6 ± 19.2 FIns: × FMD:
CoI: Nil CoI: 120 mg of orlistat for 3 × /d kg HbA1c: ✓ ×
BMI: 38.3 ± 6.5 HOMA2- CRP: ×
kg/m2 IR: × SCr: ✓
HbA1c: Not TC: ✓ eGFR:
reported HDL: ✓ ✓
Control diet group UAlb: ✓
Age: 54.7 ± 8.4
year
BW: 124.2 ± 25.0
kg
BMI: 40.6 ± 6.4
kg/m2
HbA1c: Not
reported
Goldstein et al. 6, 12, 26 and 52 Modified Atkins diet n = 14 ADA’s calorie-restricted diet n = 16 Ketogenic diet BMI: × LDL: ×
[49], Israel weeks CHO: ≤25 g/d for the first 6 weeks, CHO: 80 % divided between CHO and fat group BW: ✓ TG: ✓
thereafter increasing to up to 40 g/d Protein: 10–20%E Age: 57 ± 9 year WC: × SBP: ✓
Protein: No restriction Fat: 18–20%EMUFA, 8–10%EPUFA and BW: 91.7 ± 10.2 FBGL: ✓ DBP: ✓
Fat: No restriction 9–10%ESFA kg FIns: × FMD:
ER: Nil ER: reduced to a maximum of 1500 kcal/ BMI: 33.1 ± 3.6 HbA1c: ✓ ×
CoI: Nil d for men & 1200 kcal/d for women kg/m2 HOMA2- CRP: ×
CoI: Nil HbA1c: 9.0 ± 1.7 IR: × SCr: ×
% TC: ✓ eGFR:
Control diet group HDL: × ×
Age: 57 ± 9 year UAlb: ✓
BW: 92.2 ± 13.7
kg
(continued on next page)

6
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

Table 3 (continued )
Study, Country Follow-up Ketogenic diet Control diet Baseline Outcomes
duration characteristics (in assessed
mean ± SD)

BMI: 33.3 ± 3.0

kg/m2
HbA1c: 8.8 ± 1.2
%
Iqbal et al. [50], 26, 52 and 104 Low CHO diet n = 28 LF diet n = 40 Ketogenic diet BMI: × LDL: ✓
US weeks CHO: <30 g/d CHO: Individualised based on the group BW: ✓ TG: ✓
Protein: No restriction participant’s height & weight Age: 60.0 ± 8.9 WC: × SBP: ✓
Fat: No restriction, select healthy fats (e.g., Protein: No restriction year FBGL: ✓ DBP: ✓
minimise the intake of SFA) Fat: No restrictions, emphasise heart- BW: 118.3 ± 21.3 FIns: × FMD:
ER: No restriction healthy fats, <7%ESFA, < 300 mg of kg HbA1c: ✓ ×
CoI: Nil dietary cholesterol daily BMI: 38.1 ± 5.5 HOMA2- CRP: ×
ER: 500 kcal/d kg/m2 IR: × SCr: ×
CoI: Nil HbA1c: 7.9 ± 1.7 TC: ✓ eGFR:
% HDL: ✓ ×
Control diet group UAlb: ×
Age: 60.0 ± 9.5
year
BW: 115.5 ± 16.7
kg
BMI: 36.9 ± 5.3
kg/m2
HbA1c: 7.9 ± 1.7
%
Westman et al. 12 and 24 Low CHO ketogenic diet n = 21 Low glycaemic index diet n = 29 Ketogenic diet BMI: ✓ LDL: ✓
[51], US weeks CHO: <20 g/d CHO: 55%E group BW: ✓ TG: ✓
Protein: No restriction Protein: No restriction Age: 51.2 ± 6.1 WC: ✓ SBP: ✓
Fat: No restriction Fat: No restriction year FBGL: ✓ DBP: ✓
ER: No restriction ER: 500 kcal/d BW: 108.4 ± 19.5 FIns: ✓ FMD:
CoI: Encouraged to exercise for 30 min at CoI: Encouraged to exercise for 30 min at kg HbA1c: ✓ ×
least 3 times per week; provided nutritional least 3 times per week; provided BMI: 37.8 ± 6.7 HOMA2- CRP: ×
supplements known to have mild lowering nutritional supplements known to have kg/m2 IR: × SCr: ×
on blood glucose levels mild lowering on blood glucose levels HbA1c: 8.8 ± 1.8 TC: ✓ eGFR:
% HDL: ✓ ×
Control diet group UAlb: ×
Age: 50.0 ± 8.4
year
BW: 105.2 ± 19.8
kg
BMI: 37.9 ± 6.0
kg/m2
HbA1c: 8.3 ± 1.9
%
Dyson et al. 12 weeks Low CHO diet n = 6 Healthy eating n = 6 Both groups BMI: ✓ LDL: ✓
[52], UK CHO: ≤40 g/d CHO: No restriction Age: 54.0 ± 9.0 BW: ✓ TG: ✓
Protein: No restriction Protein: No restriction year WC: × SBP: ×
Fat: No restriction Fat: reduce total fat and SFA intakes Ketogenic diet FBGL: × DBP: ×
ER: Nil ER: 500 kcal/d group FIns: × FMD:
CoI: encouraged to increase physical activity CoI: Encouraged to increase physical BW: 99.7 kg HbA1c: ✓ ×
and exercise at moderate intensity for 30 activity and exercise at moderate BMI: 36.5 kg/m2 HOMA2- CRP: ×
min/d on at least 5, preferably 7 days/week intensity for 30 min for at least 5, HbA1c: 7.2 % IR: × SCr: ×
preferably 7 d/week Control diet group TC: ✓ eGFR:
BW: 96.9 kg HDL: ✓ ×
BMI: 33.3 kg/m2 UAlb: ×
HbA1c: 7.5 %

%E, percentage of total energy intake; %EMUFA, percentage energy intake from monounsaturated fats; %EPUFA, percentage energy intake from polyunsaturated fats; %
ESFA, percentage energy intake from saturated fat; ADA, American Diabetes Association; BMI, body mass index; BW, body weight; CHO, carbohydrate; CoI, co-
intervention; CRP, C-reactive protein; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; ER, energy restriction; FBGL, fasting blood glucose
level; FIns, fasting insulin level; FMD, flow-mediated dilatation; HbA1c, glycated haemoglobin; HBV, high biological value; HDL, high density lipoprotein cholesterol;
HF, high fat; HOMA2-IR, homeostatic model assessment of insulin resistance (computer model); LDL, low density lipoprotein cholesterol; LF, low fat; N/A, not
applicable; SBP, systolic blood pressure; SCr, serum creatinine; SD, standard deviation; SFA, saturated fat; TC, total cholesterol; TG, triglyceride; UAlb, urinary al­
bumin; WC, waist circumference.

3.5. Anthropometric changes significant differences in weight loss between KD vs control were found
(SMDrandom: 0.24; 95%CI: 0.57 to 0.09; I2 = 52 %; low-quality evidence).
Changes in body weight in 495 subjects were reported by 13 reports While the removal of data from Goday et al. [47], Saslow et al. [46] or
from 10 studies [37,39,40,43–52] (Fig. 5a), with most reporting similar Goldstein et al. [49] individually reduced the heterogeneity in the
weight loss between KD vs control. Significantly higher weight loss in “overall” meta-analysis, results remain statistically non-significant
the KD group was found at 3–6 months follow-up only (eight studies [37, (Supplemental Fig. 2c). Results were similar for BMI (Fig. 5b and Sup­
43,45–47,49–51]; SMDfixed: 0.29; 95%CI: 0.49 to − 0.10; I2 = 23 %; plemental Fig. 2b; very low-quality evidence). The difference in change
low-quality evidence). When all 10 studies were combined, no in waist circumference (250 subjects; four reports from three studies

7
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

Fig. 2. Results of risk of bias assessment.

Fig. 3. Forest plot for the effect of ketogenic diet on changes in (a) HbA1c; and (b) fasting blood glucose when compared with the control diet. CI, confi­
dence interval.

[12,38,47,51]) between KD vs control was not significant at all 3.6. Cardiovascular risk factors
follow-up time points, and when all studies were combined (Fig. 5c and
Supplemental Fig. 2o very low-quality evidence). In 12 reports from nine studies [37,38,40,42,45,48–52] which re­
ported changes in total cholesterol in 465 subjects (Fig. 6a), there was no
significant difference between KD vs control at all follow-up time points

8
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

Fig. 4. Forest plot for the effect of ketogenic diet on changes in (a) fasting insulin; and (b) HOMA2-IR when compared with the control diet. CI, confidence interval.

Fig. 5. Forest plot for the effect of ketogenic diet on changes in (a) body weight; (b) body mass index; and (c) waist circumference when compared with the control
diet. CI, confidence interval.

and in the overall meta-analysis (moderate quality evidence). No change have a higher increase in LDL at the 6–12 months follow-up time point
in the statistical significance and heterogeneity of the results was (five studies [38,43,46,48,50]; SMDfixed: 0.34; 95%CI: 0.10 to 0.57; I2 =
observed when studies were removed individually (Supplemental 0 %; low-quality evidence). Removing data from Breukelman et al. [42]
Fig. 2m). A significantly higher increase in HDL was found in subjects resulted in a substantial decrease in the heterogeneity but not a change
following a KD when results from all studies (517 subjects; 14 reports in the statistical significance of the results (Supplemental Fig. 2k).
from 11 studies [37,38,40,42–52]; Fig. 6b) were combined (SMDfixed: Significantly greater reductions in triglyceride (518 subjects; 12 reports
0.19; 95%CI: 0.02 to 0.37; I2 = 0 %; moderate quality evidence), but not from 11 studies [38,40,42,43,45–52]) in the KD group were found at the
at the different follow-up time points. In the leave-one-out analysis, only 6–12 months follow-up only (six studies [38,43,46,48–50]; SMDfixed:
the omission of data from Tay et al. [40] rendered the results 0.41; 95%CI: 0.64 to − 0.18; I2 = 1 %; low-quality evidence; Fig. 7b), as
non-significant (Supplemental Fig. 2i). In contrast, the combined anal­ well as when all 11 studies were combined (SMDfixed: 0.40; 95%CI: 0.58
ysis of results from 13 reports from 10 studies [37,38,40,42–48,50–52] to − 0.23; I2 = 31 %; low-quality evidence). The removal of data from
with 488 subjects showed no significant differences in the change in LDL Iqbal et al. [50] substantially reduced the heterogeneity of the results
between KD vs control (SMDrandom: 0.04; 95%CI: 0.33 to 0.40; I2 = 52 %; (Supplemental Fig. 2n).
low-quality evidence; Fig. 7a), but subjects in the KD group was found to Changes in systolic and diastolic blood pressure in 377 subjects were

9
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

Fig. 6. Forest plot for the effect of ketogenic diet on changes in fasting (a) total cholesterol; and (b) HDL-cholesterol when compared with the control diet. CI,
confidence interval.

Fig. 7. Forest plot for the effect of ketogenic diet on changes in fasting (a) LDL-cholesterol; and (b) triglycerides when compared with the control diet. CI, confi­
dence interval.

reported in 10 reports from seven studies [37,39,40,43–45,48–51]. No evidence), but not at other follow-up timepoints or when all seven
significant differences in the change in diastolic blood pressure were studies were combined (Fig. 8b). Removing individual studies one at a
found between KD vs control at all follow-up time points as well as in the time did not change statistical significance or a substantial drop in
overall meta-analysis (Fig. 8a; moderate quality evidence), while for heterogeneity (Supplemental Figs. 2e and 2l).
systolic blood pressure, significantly greater reductions were observed No significant differences in the changes in flow-mediated dilatation
in the KD group at 3–6 months follow-up only (six studies [37,43,45, between KD vs control were observed in the “overall” meta-analysis (two
48–50]; SMDfixed: 0.24; 95%CI: 0.47 to − 0.02; I2 = 0 %; very low-quality studies [40,45]; SMDrandom: 0.25; 95%CI: 3.91 to 3.41; I2 = 56 %; very

10
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

Fig. 8. Forest plot for the effect of ketogenic diet on changes in (a) systolic blood pressure; and (b) diastolic blood pressure when compared with the control diet. CI,
confidence interval.

low-quality evidence; Fig. 9a). Similarly, no difference in change in CRP chronic hyperglycaemia can lead to severe complications, including foot
between KD vs control was observed at all follow-up time points and in ulcers, organ failure and even death [54]. There is strong evidence that a
the “overall” meta-analysis (182 subjects; three studies [40,43,45]; 1%-point decline in HbA1c can significantly reduce the risk of diabetic
moderate quality evidence; Fig. 9b). complications by 14 %–43 % [55]. The KD can achieve approximately
1%-point reductions in HbA1c, comparable to common antidiabetic
medications [56]. Several studies in this review reported significantly
3.7. Liver and renal functions
greater reduction or even discontinuation of diabetic medications
among patients prescribed the KD [37,38,43,44,48]. However, main­
Among all RCTs, only Goday et al. [47] reported the influence of KD
taining strict adherence to the KD’s low carbohydrate intake can be
on liver enzymes in T2D patients. In that trial, KD resulted in a statis­
challenging [38,40,41,43,44,46,50,51], highlighting the need for pro­
tically significant increase in both aspartate aminotransferase (AST) and
fessional support and behavioural strategies to enhance adherence and
alanine aminotransferase (ALT) by 12.58 (p < 0.05) and 12.89 IU/mL (p
optimise benefits.
< 0.001) respectively at 2 weeks; but the changes became insignificant
When examining weight reduction, it becomes evident that while the
at 4 months.
KD successfully induces short-term weight loss, it does not consistently
On the other hand, among studies which reported changes in renal
outperform other diets in the long term. This disparity could be attrib­
function markers, all suggested that KD did not have any significant
uted to the gradual increase in daily energy intake among subjects as
impact on serum creatinine (161 subjects; two studies [40,48]; SMDfixed:
their body weight rebounds, irrespective of the specific dietary inter­
0.00; 95%CI: 0.31 to 0.31; I2 = 0 %; moderate quality evidence;
vention employed in certain studies [38,49]. However, studies
Fig. 10a), urinary albumin (186 subjects; three studies [40,48,49];
comparing the KD with an isocaloric comparator diet or a KD with a
SMDfixed: 0.10; 95%CI: 0.19 to 0.39; I2 = 48 %; moderate quality evi­
smaller energy deficit than the comparator diet reported greater weight
dence; Fig. 10b) and eGFR (161 subjects; two studies [40,48]; SMDfixed:
reduction with the KD [47,49–52], suggesting that factors other than
0.10; 95%CI: 0.41 to 0.21; I2 = 0 %; moderate quality evidence;
energy deficit from dietary intake may contribute to its efficacy. One
Fig. 10c).
proposed mechanism contributing to weight reduction in the KD is the
suppression of appetite induced by β-hydroxybutyrate [57]. Several
4. Discussion
trials have observed a reduction in energy intake without explicit energy
restrictions [48–51], suggesting that the KD may influence appetite
In this systematic review and meta-analysis, we conducted a
control. However, only one report quantitatively analysed β-hydrox­
comprehensive assessment of the existing evidence on the effectiveness
ybutyrate, finding no significant changes in hunger [36]. Further
and safety of the KD for T2D management. Our findings suggest that
research is needed to understand the relationship between ketone levels,
concerns regarding the impact of the KD on cardiovascular disease risk
appetite, and weight loss [58].
and hepatic and renal function may be unfounded. Furthermore, the KD
Concerns about the impact of KD on cardiovascular, liver and renal
appears to be more effective than other diets in improving glycaemic
health have been cited as reasons against its use for obesity and T2D
control and inducing weight loss in the short term (approximately 3–6
management [25]. However, most studies found no significant change in
months). These results are broadly consistent with previous reviews and
blood pressure and cholesterol profile; some even reported improve­
meta-analyses [29,53], although it is worth noting that patients who
ments following the KD intervention. The included trials also generally
adopt new interventions, such as the KD, often demonstrate enhanced
did not show significant changes in relevant markers of liver and renal
adherence to overall medical care, potentially resulting in improved
functions, supporting the safety of following the KD for up to 2 years.
outcomes. Additionally, including physical activity as a co-intervention
Nonetheless, there are some less severe side effects of the KD reported in
in several studies, especially within the KD group, might have influenced
five reports from three studies [37,38,40,47,51]. These include symp­
the observed results.
toms such as asthenia, headache, nausea, vomiting, constipation,
Effective management of T2D relies on optimal glycaemic control, as

11
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

Fig. 9. Forest plot for the effect of ketogenic diet on changes in (a) flow-mediated dilatation; and (b) C-reactive protein when compared with the control diet. CI,
confidence interval.

cramps, myalgia, muscular weakness, heaviness and tiredness of legs, to diabetic medications may be necessary to prevent hypoglycaemic
hair loss, musculoskeletal ailments, gastrointestinal disorders (con­ events [60].
stipation and diverticulitis), and one incident of non-hospitalised It is important to note that there is a common misconception that a
hypoglycaemia. Although the occurrence rates of these symptoms did KD must be high in saturated fat intake, particularly those from animal
not differ significantly from those of the comparator diet in the study by sources. However, multiple studies included in this review demonstrate
Westman et al. [51], they should be considered. that a well-formulated KD can consist of high monounsaturated fatty
acids (35 % of total energy) and polyunsaturated fatty acids (13 %), with
low saturated fat content (<10 %) [36–38,40,43,47]. Studies have also
4.1. Clinical relevance shown that KD meal plans can be designed to provide sufficient micro­
nutrients with the guidance of dietitians [61,62]. By consuming
Based on our meta-analyses, implementing the KD for up to 2 years high-quality dietary lipids, including adequate fibre from green leafy
could be a safe and comparably effective alternative to traditional diets vegetables, water, and micronutrients, individuals on the KD can be
for managing T2D. The results suggest that implementing the KD may be protected against cardiovascular diseases by suppressing atheroscle­
a safe and viable option to help improve lipid profiles, particularly for rosis, modestly lowering blood pressure, improving endothelial function
increasing HDL-cholesterol and lowering triglycerides. However, the KD and reducing inflammation [63,64].
does not appear to confer substantial additional benefits for glycaemic
control or weight loss compared to conventional diabetes diets over the
long term. Although there is limited long-term safety data available, this 4.2. Strengths and limitation
should not discourage the use of the KD for T2D management, as no
evidence suggests any safety concerns associated with following the diet A strength of this study is that we only included RCTs, reducing the
for an extended period. It is recommended to regularly monitor clinical potential bias arising from baseline differences in participant charac­
parameters such as liver enzymes, renal function markers, and blood teristics between intervention groups and allowing for more objective
lipid profiles to minimise potential risks [59]. Additionally, adjustments comparisons with other types of diets. We also conducted meta-analyses

12
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

Fig. 10. Forest plot for the effect of ketogenic diet on changes in (a) serum creatinine; (b) urinary albumin; and (c) estimated glomerular filtration rate when
compared with the control diet. CI, confidence interval.

stratified by the follow-up period, providing greater detail than previous imbalanced baseline characteristics, and reliance on self-reported di­
meta-analyses. etary records to estimate participants’ energy and macronutrient intake,
However, there are several limitations to consider when interpreting should be considered, although the meta-analyses partially addressed
our findings. First, all the identified RCTs included overweight/obese these issues by combining the results.
participants; most included studies were conducted on Caucasian sub­
jects. It is essential to acknowledge that the pathophysiology, pheno­ 5. Conclusion
types, and patterns of diabetic complications in T2D can vary among
populations [65–68], and as such, the generalizability of our findings to Our meta-analyses suggest that KD could be an effective and safe
the broad population of T2D patients may be limited. Second, the alternative to glycaemic control and weight reduction in T2D patients
observed heterogeneity in the control or comparator diets across the for up to 2 years. However, the current body of evidence has limitations
included studies adds complexity to the interpretation of the effects of such as small sample size, co-interventions with exercise programs/
the KD. Future research should aim for standardised control diets and recommendations, and a lack of long-term data. Therefore, medical
carefully consider the dietary quality of comparator interventions to supervision is essential when adopting the KD, and further high-quality
provide a more accurate assessment of the outcomes associated with the research addressing these limitations is necessary before widely rec­
KD. Third, most RCTs included co-intervention, such as supervised ommending KD and promoting it as the first-line treatment option for
moderate exercise programs or other lifestyle recommendations, making T2D.
it difficult to attribute the reported effect solely to the KD. Fourth, it is
crucial to consider the potential impact of publication bias on our
PROSPERO registration
findings. There is a possibility that studies demonstrating positive results
concerning KD and low-carbohydrate diets are more likely to be pub­
This systematic review and meta-analysis is registered as
lished, particularly in prominent journals, resulting in an over­
CRD42021282221 at PROSPERO (https://www.crd.york.ac.uk/prosp
representation of studies showcasing favourable outcomes for these
ero/display_record.php?RecordID=282221).
diets while overlooking studies with less favourable or unfavourable
results [69]. Fifth, our systematic review revealed a lack of long-term
trials assessing the sustainability and safety of the KD’s effects. Poten­ Data share statement
tial long-term side effects such as micronutrient deficiency, hepatic
steatosis, kidney stones and hypoproteinaemia have been suggested [8, Data described in the manuscript, code book, and analytic code will
10], but solid experimental evidence supporting these concerns is be available upon request. Please send your request to A/Prof Jimmy
currently lacking. Sixth, only about half of the included studies moni­ Louie at jimmylouie@swin.edu.au
tored the ketone levels of participants using biomarkers during the
intervention, raising questions about whether the dietary effects, Authors contributions
particularly for weight reduction, were due to ketosis induced by KD.
Therefore, future studies on KD should routinely monitor the ketone KYCC and JCYL identified the review topic and designed the search
level of study participants to allow the observed effects to be attributed strategy. KYCC performed the initial search, and both authors were
to ketosis, a key mechanism of how KD is suggested to exert its beneficial involved in the screening data extraction, risk of bias assessment and
health effects. Seventh, our analysis did not consider the medication certainty of evidence grading. JCYL performed the meta-analyses. KYCC
usage among study participants, which raises the possibility of bias, as wrote the first draft of the manuscript, and both authors provided sub­
those in the intervention group might have higher adherence to phar­ stantial intellectual input to the subsequent edits and have read and
macological treatments, potentially influencing the observed outcomes. approved the final manuscript. JCYL is the guarantor of the work and
Last, common limitations in individual RCTs, such as small sample sizes, has primary responsibility for the content presented in this manuscript.
The corresponding author attests that all listed authors meet authorship

13
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

criteria and that no others meeting the criteria have been omitted. [15] Masood W, Annamaraju P, Uppaluri KR. Ketogenic diet. In: StatPearls. StatPearls
publishing; 2021. Treasure Island (F.
[16] Boye KS, Lage MJ, Thieu V, Shinde S, Dhamija S, Bae JP. Obesity and glycemic
Statements and declarations control among people with type 2 diabetes in the United States: a retrospective
cohort study using insurance claims data. J Diabet Complicat 2021;35(9):107975.
https://doi.org/10.1016/j.jdiacomp.2021.107975.
No funding was received for conducting this study. The authors have [17] Churuangsuk C, Hall J, Reynolds A, Griffin SJ, Combet E, Lean MEJ. Diets for
no relevant financial or non-financial interests to disclose. weight management in adults with type 2 diabetes: an umbrella review of
While preparing this work, the author(s) used ChatGPT 3.5 to published meta-analyses and systematic review of trials of diets for diabetes
remission. Diabetologia 2022;65(1):14–36. https://doi.org/10.1007/s00125-021-
improve the flow, clarity, coherence, and grammar of the text drafted by
05577-2.
the authors. After using this tool/service, the senior author (JCYL) [18] Fridman M, Lucas ME, Paprocki Y, Dang-Tan T, Iyer NN. Impact of weight change
reviewed and edited the content as needed and took full responsibility in adults with type 2 diabetes mellitus: a literature review and critical analysis.
Clinicoecon Outcomes Res 2020;12:555–66. https://doi.org/10.2147/CEOR.
for the publication’s content.
S266873.
[19] Huntriss R, Campbell M, Bedwell C. The interpretation and effect of a low-
Funding carbohydrate diet in the management of type 2 diabetes: a systematic review and
meta-analysis of randomised controlled trials. Eur J Clin Nutr 2018;72(3):311–25.
https://doi.org/10.1038/s41430-017-0019-4.
The authors did not receive any funding for the writing of this [20] Wilding JPH. The importance of weight management in type 2 diabetes mellitus.
review. Int J Clin Pract 2014;68(6):682–91. https://doi.org/10.1111/ijcp.12384.
[21] Moazzeni SS, Hizomi Arani R, Deravi N, Hasheminia M, Khalili D, Azizi F,
Hadaegh F. Weight change and risk of cardiovascular disease among adults with
type 2 diabetes: more than 14 years of follow-up in the Tehran Lipid and Glucose
Declaration of competing interest
Study. Cardiovasc Diabetol 2021;20(1):141. https://doi.org/10.1186/s12933-021-
01326-2.
The authors declare that they have no known competing financial [22] Yancy WS, Foy M, Chalecki AM, Vernon MC, Westman EC. A low-carbohydrate,
interests or personal relationships that could have appeared to influence ketogenic diet to treat type 2 diabetes. Nutr Metab 2005;2(1):34. https://doi.org/
10.1186/1743-7075-2-34.
the work reported in this paper. [23] Breukelman GJ, Basson AK, Djarova TG, Du Preez CJ, Shaw I, Malan H, Shaw BS.
Concurrent low-carbohydrate, high-fat diet with/without physical activity does not
improve glycaemic control in type 2 diabetics. S Afr J Clin Nutr 2021;34(1):18–21.
Appendix A. Supplementary data
https://doi.org/10.1080/16070658.2019.1638610.
[24] Sainsbury E, Kizirian NV, Partridge SR, Gill T, Colagiuri S, Gibson AA. Effect of
Supplementary data to this article can be found online at https://doi. dietary carbohydrate restriction on glycemic control in adults with diabetes: a
systematic review and meta-analysis. Diabetes Res Clin Pract 2018;139:239–52.
org/10.1016/j.dsx.2023.102905.
https://doi.org/10.1016/j.diabres.2018.02.026.
[25] Joshi S, Ostfeld RJ, McMacken M. The ketogenic diet for obesity and
References diabetes—enthusiasm outpaces evidence. JAMA Intern Med 2019;179(9):1163–4.
https://doi.org/10.1001/jamainternmed.2019.2633.
[26] Apekey TA, Maynard MJ, Kittana M, Kunutsor SK. Comparison of the effectiveness
[1] American Diabetes Association. Diagnosis and classification of diabetes mellitus.
of low carbohydrate versus low fat diets, in type 2 diabetes: systematic review and
Diabetes Care 2010;33(Supplement_1):S62–9. https://doi.org/10.2337/dc10-
meta-analysis of randomized controlled trials. Nutrients 2022;14(20). https://doi.
S062.
org/10.3390/nu14204391.
[2] Khan MAB, Hashim MJ, King JK, Govender RD, Mustafa H, Al Kaabi J.
[27] Zaki HA, Iftikhar H, Bashir K, Gad H, Samir Fahmy A, Elmoheen A. A comparative
Epidemiology of type 2 diabetes - global burden of disease and forecasted trends.
study evaluating the effectiveness between ketogenic and low-carbohydrate diets
J Epidemiol Glob Health 2020;10(1):107–11. https://doi.org/10.2991/jegh.
on glycemic and weight control in patients with type 2 diabetes mellitus: a
k.191028.001.
systematic review and meta-analysis. Cureus 2022;14(5):e25528. https://doi.org/
[3] Simó R, Hernández C. Treatment of diabetes mellitus: general goals and clinical
10.7759/cureus.25528.
practice management. Rev Esp Cardiol 2002;55(8):845–60.
[28] Jayedi A, Zeraattalab-Motlagh S, Jabbarzadeh B, Hosseini Y, Jibril AT,
[4] Davies MJ, D’Alessio DA, Fradkin J, Kernan WN, Mathieu C, Mingrone G,
Shahinfar H, Mirrafiei A, Hosseini F, Shab-Bidar S. Dose-dependent effect of
Rossing P, Tsapas A, Wexler DJ, Buse JB. A consensus report by the American
carbohydrate restriction for type 2 diabetes management: a systematic review and
diabetes association (ADA) and the European association for the study of diabetes
dose-response meta-analysis of randomized controlled trials. Am J Clin Nutr 2022;
(EASD). Diabetes Care 2018;41(12):2669–701. https://doi.org/10.2337/dci18-
116(1):40–56. https://doi.org/10.1093/ajcn/nqac066.
0033.
[29] Tinguely D, Gross J, Kosinski C. Efficacy of ketogenic diets on type 2 diabetes: a
[5] Paoli A. Ketogenic diet for obesity: friend or foe? Int J Environ Res Publ Health
systematic review. Curr Diabetes Rep 2021;21(9):32. https://doi.org/10.1007/
2014;11(2):2092–107.
s11892-021-01399-z.
[6] Bolla AM, Caretto A, Laurenzi A, Scavini M, Piemonti L. Low-carb and ketogenic
[30] Skow LS, Jha KR. A ketogenic diet is effective in improving insulin sensitivity in
diets in type 1 and type 2 diabetes. Nutrients 2019;11(5):962. https://doi.org/
individuals with type 2 diabetes. Curr Diabetes Rev 2023;19(6):119–29. https://
10.3390/nu11050962.
doi.org/10.2174/1573399818666220425093535.
[7] Drabińska N, Wiczkowski W, Piskuła MK. Recent advances in the application of a
[31] Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD,
ketogenic diet for obesity management. Trends Food Sci Technol 2021;110:28–38.
Shamseer L, Tetzlaff JM, Akl EA, Brennan SE, Chou R, Glanville J, Grimshaw JM,
https://doi.org/10.1016/j.tifs.2021.01.080.
Hróbjartsson A, Lalu MM, Li T, Loder EW, Mayo-Wilson E, McDonald S,
[8] Batch JT, Lamsal SP, Adkins M, Sultan S, Ramirez MN. Advantages and
McGuinness LA, Stewart LA, Thomas J, Tricco AC, Welch VA, Whiting P, Moher D.
disadvantages of the ketogenic diet. A Review Article Cureus 2020;12(8):e9639.
The PRISMA 2020 statement: an updated guideline for reporting systematic
https://doi.org/10.7759/cureus.9639.
reviews. BMJ 2021;372:n71. https://doi.org/10.1136/bmj.n71.
[9] Feinman RD, Pogozelski WK, Astrup A, Bernstein RK, Fine EJ, Westman EC,
[32] Sterne JAC, Savović J, Page MJ, Elbers RG, Blencowe NS, Boutron I, Cates CJ,
Accurso A, Frassetto L, Gower BA, McFarlane SI, Nielsen JV, Krarup T, Saslow L,
Cheng H-Y, Corbett MS, Eldridge SM, Emberson JR, Hernán MA, Hopewell S,
Roth KS, Vernon MC, Volek JS, Wilshire GB, Dahlqvist A, Sundberg R, Childers A,
Hróbjartsson A, Junqueira DR, Jüni P, Kirkham JJ, Lasserson T, Li T, McAleenan A,
Morrison K, Manninen AH, Dashti HM, Wood RJ, Wortman J, Worm N. Dietary
Reeves BC, Shepperd S, Shrier I, Stewart LA, Tilling K, White IR, Whiting PF,
carbohydrate restriction as the first approach in diabetes management: critical
Higgins JPT. RoB 2: a revised tool for assessing risk of bias in randomised trials.
review and evidence base. Nutrition 2015;31(1):1–13. https://doi.org/10.1016/j.
BMJ 2019;366:l4898. https://doi.org/10.1136/bmj.l4898.
nut.2014.06.011.
[33] Higgins JPT, Li T, Deeks JJ. Chapter 6: choosing effect measures and computing
[10] Shilpa J, Mohan V. Ketogenic diets: boon or bane? Indian J Med Res 2018;148(3):
estimates of effect. In: Higgins J, Thomas J, Chandler J, et al., editors. Cochrane
251–3. https://doi.org/10.4103/ijmr.IJMR_1666_18.
handbook for systematic reviews of interventions. 6. 3 edn. Lodon, UK: The
[11] Turton J, Brinkworth GD, Field R, Parker H, Rooney K. An evidence-based
Cochrane Collaboration; 2022.
approach to developing low-carbohydrate diets for type 2 diabetes management: a
[34] Baujat B, Mahé C, Pignon JP, Hill C. A graphical method for exploring
systematic review of interventions and methods. Diabetes Obes Metabol 2019;21
heterogeneity in meta-analyses: application to a meta-analysis of 65 trials. Stat
(11):2513–25. https://doi.org/10.1111/dom.13837.
Med 2002;30:2641–52.
[12] Ting R, Dugré N, Allan GM, Lindblad AJ. Ketogenic diet for weight loss. Can Fam
[35] Brinkworth GD, Luscombe-Marsh ND, Thompson CH, Noakes M, Buckley JD,
Physician 2018;64(12). 906-906.
Wittert G, Wilson CJ. Long-term effects of very low-carbohydrate and high-
[13] Bueno NB, de Melo ISV, de Oliveira SL, da Rocha Ataide T. Very-low-carbohydrate
carbohydrate weight-loss diets on psychological health in obese adults with type 2
ketogenic diet v. low-fat diet for long-term weight loss: a meta-analysis of
diabetes: randomized controlled trial. J Intern Med 2016;280(4):388–97. https://
randomised controlled trials. Br J Nutr 2013;110(7):1178–87. https://doi.org/
doi.org/10.1111/joim.12501.
10.1017/S0007114513000548.
[36] Struik NA, Brinkworth GD, Thompson CH, Buckley JD, Wittert G, Luscombe-
[14] Dowis K, Banga S. The potential health benefits of the ketogenic diet: a narrative
Marsh ND. Very low and higher carbohydrate diets promote differential appetite
review. Nutrients 2021;13(5):1654.

14
K.Y.C. Choy and J.C.Y. Louie Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17 (2023) 102905

responses in adults with type 2 diabetes: a randomized trial. J Nutr 2020;150(4): [50] Iqbal N, Vetter ML, Moore RH, Chittams JL, Dalton-Bakes CV, Dowd M, Williams-
800–5. https://doi.org/10.1093/jn/nxz344. Smith C, Cardillo S, Wadden TA. Effects of a low-intensity intervention that
[37] Tay J, Luscombe-Marsh ND, Thompson CH, Noakes M, Buckley JD, Wittert GA, prescribed a low-carbohydrate vs. a low-fat diet in obese, diabetic participants.
Yancy Jr WS, Brinkworth GD. A very low-carbohydrate, low–saturated fat diet for Obesity 2010;18(9):1733–8. https://doi.org/10.1038/oby.2009.460.
type 2 diabetes management: a randomized trial. Diabetes Care 2014;37(11): [51] Westman EC, Yancy WS, Mavropoulos JC, Marquart M, McDuffie JR. The effect of a
2909–18. https://doi.org/10.2337/dc14-0845. low-carbohydrate, ketogenic diet versus a low-glycemic index diet on glycemic
[38] Tay J, Luscombe-Marsh ND, Thompson CH, Noakes M, Buckley JD, Wittert GA, control in type 2 diabetes mellitus. Nutr Metab 2008;5(1):36. https://doi.org/
Yancy Jr WS, Brinkworth GD. Comparison of low- and high-carbohydrate diets for 10.1186/1743-7075-5-36.
type 2 diabetes management: a randomized trial. Am J Clin Nutr 2015;102(4): [52] Dyson PA, Beatty S, Matthews DR. A low-carbohydrate diet is more effective in
780–90. https://doi.org/10.3945/ajcn.115.112581. reducing body weight than healthy eating in both diabetic and non-diabetic
[39] Tay J, Thompson CH, Luscombe-Marsh ND, Noakes M, Buckley JD, Wittert GA, subjects. Diabet Med 2007;24(12):1430–5. https://doi.org/10.1111/j.1464-
Brinkworth GD. Long-term effects of a very low carbohydrate compared with a 5491.2007.02290.x.
high carbohydrate diet on renal function in individuals with type 2 diabetes: a [53] Choi YJ, Jeon S-M, Shin S. Impact of a ketogenic diet on metabolic parameters in
randomized trial. Medicine (Baltim) 2015;94(47):e2181. https://doi.org/10.1097/ patients with obesity or overweight and with or without type 2 diabetes: a meta-
md.0000000000002181. analysis of randomized controlled trials. Nutrients 2020;12(7):2005.
[40] Tay J, Thompson CH, Luscombe-Marsh ND, Wycherley TP, Noakes M, Buckley JD, [54] Fowler MJ. Microvascular and macrovascular complications of diabetes. Clin
Wittert GA, Yancy Jr WS, Brinkworth GD. Effects of an energy-restricted low- Diabetol 2008;26(2):77–82. https://doi.org/10.2337/diaclin.26.2.77.
carbohydrate, high unsaturated fat/low saturated fat diet versus a high- [55] Mohan V, Mapari JA, Karnad PD, Mann JS, Maheshwari VK. Reduced diabetes
carbohydrate, low-fat diet in type 2 diabetes: a 2-year randomized clinical trial. mellitus-related comorbidities by regular self-monitoring of blood glucose:
Diabetes Obes Metabol 2018;20(4):858–71. https://doi.org/10.1111/dom.13164. economic and quality of life implications. Indian J Endocrinol Metab 2018;22(4):
[41] Wycherley TP, Thompson CH, Buckley JD, Luscombe-Marsh ND, Noakes M, 461–5. https://doi.org/10.4103/ijem.IJEM_216_17.
Wittert GA, Brinkworth GD. Long-term effects of weight loss with a very-low [56] Sherifali D, Nerenberg K, Pullenayegum E, Cheng JE, Gerstein HC. The effect of
carbohydrate, low saturated fat diet on flow mediated dilatation in patients with oral antidiabetic agents on A1C levels: a systematic review and meta-analysis.
type 2 diabetes: a randomised controlled trial. Atherosclerosis 2016;252:28–31. Diabetes Care 2010;33(8):1859–64. https://doi.org/10.2337/dc09-1727.
https://doi.org/10.1016/j.atherosclerosis.2016.07.908. [57] Roekenes J, Martins C. Ketogenic diets and appetite regulation. Curr Opin Clin
[42] Breukelman GJ, Basson AK, Djarova TG, Du Preez CJ, Shaw I, Shaw BS. Nutr Metab Care 2021;24(4):359–63. https://doi.org/10.1097/
Combination low carbohydrate, high fat diet and physical activity intervention on mco.0000000000000760.
lipoprotein-lipids in type 2 diabetics. Asian J Sports Med 2019;10(4):e86905. [58] Gibson AA, Seimon RV, Lee CMY, Ayre J, Franklin J, Markovic TP, Caterson ID,
https://doi.org/10.5812/asjsm.86905. Sainsbury A. Do ketogenic diets really suppress appetite? A systematic review and
[43] Saslow LR, Daubenmier JJ, Moskowitz JT, Kim S, Murphy EJ, Phinney SD, Ploutz- meta-analysis. Obes Rev 2015;16(1):64–76. https://doi.org/10.1111/obr.12230.
Snyder R, Goldman V, Cox RM, Mason AE, Moran P, Hecht FM. Twelve-month [59] Anekwe CV, Chandrasekaran P, Stanford FC. Ketogenic diet-induced elevated
outcomes of a randomized trial of a moderate-carbohydrate versus very low- cholesterol, elevated liver enzymes and potential non-alcoholic fatty liver disease.
carbohydrate diet in overweight adults with type 2 diabetes mellitus or Cureus 2020;12(1). https://doi.org/10.7759/cureus.6605. e6605-e6605.
prediabetes. Nutr Diabetes 2017;7(12):304. https://doi.org/10.1038/s41387-017- [60] Cucuzzella M, Riley K, Isaacs D. Adapting medication for type 2 diabetes to a low
0006-9. carbohydrate diet. Front Nutr 2021;8. https://doi.org/10.3389/fnut.2021.688540.
[44] Saslow LR, Kim S, Daubenmier JJ, Moskowitz JT, Phinney SD, Goldman V, [61] Zinn C, Rush A, Johnson R. Assessing the nutrient intake of a low-carbohydrate,
Murphy EJ, Cox RM, Moran P, Hecht FM. A randomized pilot trial of a moderate high-fat (LCHF) diet: a hypothetical case study design. BMJ Open 2018;8(2):
carbohydrate diet compared to a very low carbohydrate diet in overweight or e018846. https://doi.org/10.1136/bmjopen-2017-018846.
obese individuals with type 2 diabetes mellitus or prediabetes. PLoS One 2014;9 [62] Turton JL, Field RJ, Struik NA, Parker HM, Rooney K. Formulating nutritionally
(4):e91027. https://doi.org/10.1371/journal.pone.0091027. adequate low- carbohydrate diets: an analysis of the Australian food composition
[45] Barbosa-Yañez RL, Dambeck U, Li L, Machann J, Kabisch S, Pfeiffer AFH. Acute database. Biomed J Sci Tech Res 2022;44(1):35166–80.
endothelial benefits of fat restriction over carbohydrate restriction in type 2 [63] Mashek DG, Wu C. MUFAs. Adv Nutr 2015;6(3):276–7. https://doi.org/10.3945/
diabetes mellitus: beyond carbs and fats. Nutrients 2018;10(12):1859. an.114.005926.
[46] Saslow LR, Mason AE, Kim S, Goldman V, Ploutz-Snyder R, Bayandorian H, [64] Ander BP, Dupasquier CM, Prociuk MA, Pierce GN. Polyunsaturated fatty acids and
Daubenmier J, Hecht FM, Moskowitz JT. An online intervention comparing a very their effects on cardiovascular disease. Exp Clin Cardiol 2003;8(4):164–72.
low-carbohydrate ketogenic diet and lifestyle recommendations versus a plate [65] Joost H-G. Pathogenesis, risk assessment and prevention of type 2 diabetes
method diet in overweight individuals with type 2 diabetes: a randomized mellitus. Obes Facts 2008;1(3):128–37. https://doi.org/10.1159/000137673.
controlled trial. J Med Internet Res 2017;19(2):e36. https://doi.org/10.2196/ [66] Ke C, Narayan KMV, Chan JCN, Jha P, Shah BR. Pathophysiology, phenotypes and
jmir.5806. management of type 2 diabetes mellitus in Indian and Chinese populations. Nat
[47] Goday A, Bellido D, Sajoux I, Crujeiras AB, Burguera B, García-Luna PP, Oleaga A, Rev Endocrinol 2022. https://doi.org/10.1038/s41574-022-00669-4.
Moreno B, Casanueva FF. Short-term safety, tolerability and efficacy of a very low- [67] Ma RCW, Chan JCN. Type 2 diabetes in East Asians: similarities and differences
calorie-ketogenic diet interventional weight loss program versus hypocaloric diet with populations in Europe and the United States. Ann N Y Acad Sci 2013;1281(1):
in patients with type 2 diabetes mellitus. Nutr Diabetes 2016;6(9). https://doi.org/ 64–91. https://doi.org/10.1111/nyas.12098.
10.1038/nutd.2016.36. e230-e230. [68] Gujral UP, Pradeepa R, Weber MB, Narayan KMV, Mohan V. Type 2 diabetes in
[48] Mayer SB, Jeffreys AS, Olsen MK, McDuffie JR, Feinglos MN, Yancy Jr WS. Two South Asians: similarities and differences with white Caucasian and other
diets with different haemoglobin A1c and antiglycaemic medication effects despite populations. Ann N Y Acad Sci 2013;1281(1):51–63. https://doi.org/10.1111/
similar weight loss in type 2 diabetes. Diabetes Obes Metabol 2014;16(1):90–3. j.1749-6632.2012.06838.x.
https://doi.org/10.1111/dom.12191. [69] Churuangsuk C, Kherouf M, Combet E, Lean M. Low-carbohydrate diets for
[49] Goldstein T, Kark JD, Berry EM, Adler B, Ziv E, Raz I. The effect of a low overweight and obesity: a systematic review of the systematic reviews. Obes Rev
carbohydrate energy-unrestricted diet on weight loss in obese type 2 diabetes 2018;19(12):1700–18. https://doi.org/10.1111/obr.12744.
patients - a randomized controlled trial. Clin Nutr ESPEN 2011;6(4):e178–86. [70] Volek JS, Westman EC. Very-low-carbohydrate weight-loss diets revisited. Cleve
https://doi.org/10.1016/j.eclnm.2011.04.003. Clin J Med 2002;69(11):849–62.

15