The American Journal of Clinical Nutrition 117 (2023) 870–882

journal homepage: www.journals.elsevier.com/the-american-journal-of-clinical-nutrition

Original Research Article

Effect of calorie restriction in comparison to usual diet or usual care on

A B S T R A C T

Background: Limited evidence is available on the dose-dependent effects of calorie restriction in patients with type 2 diabetes.
Objectives: We aimed to gather available evidence on the effect of calorie restriction on the management of type 2 diabetes.
Methods: We systematically searched PubMed, Scopus, CENTRAL, Web of Science, and gray literature until November 2022 for randomized trials >12
wk looking at the effect of a prespecified calorie-restricted diet on remission of type 2 diabetes. We performed random-effects meta-analyses to estimate
the absolute effect (risk difference) at 6-mo (6
3 mo) and 12-mo (12
3 mo) follow-ups. Then, we performed dose–response meta-analyses to estimate
the mean difference (MD) for the effects of calorie restriction on cardiometabolic outcomes. We used the Grading of Recommendations Assessment,
Development and Evaluation (GRADE) approach to judge the certainty of evidence.
Results: Twenty-eight randomized trials with 6281 participants were included. Using a remission definition of an HbA1c level of <6.5% without
antidiabetic medication use, calorie-restricted diets increased remission by 38 more per 100 patients (95% CI: 9 more, 67 more; n ¼ 5 trials; GRADE ¼
moderate) at 6 mo and by 13 more per 100 patients (95% CI: 10 more, 18 more; n ¼ 4; GRADE ¼ moderate) at 12 mo in comparison to usual diet or
usual care. Using a definition of HbA1c of <6.5% after at least 2-mo cessation of antidiabetic medications, remission increased by 34 more per 100
patients (95% CI: 15 more, 53 more; n ¼ 1; GRADE ¼ very low) at 6 mo and by 16 more per 100 patients (95% CI: 4 more, 49 more; n ¼ 2; GRADE ¼
low) at 12 mo. At 6 mo, each 500-kcal/d decrease in energy intake resulted in clinically meaningful reductions in body weight (MD: 6.33 kg; 95% CI:
7.76, 4.90; n ¼ 22; GRADE ¼ high) and HbA1c (MD: 0.82%; 95% CI: 1.05, 0.59; n ¼ 18; GRADE ¼ high), which attenuated remarkably at
12 mo.
Conclusions: Calorie-restricted diets may be effective intervention for type 2 diabetes remission, especially when coupled with an intensive lifestyle
modification program.
This systematic review was registered in PROSPERO as CRD42022300875 (https://www.crd.york.ac.uk/prospero/display_record.php?RecordID ¼
300875).
Am J Clin Nutr 2023;xxx:xx–xx.

Keywords: calorie restriction, diet therapy, type 2 diabetes, diabetes remission, randomized trials

Introduction increase to 690 million by the year 2045 [2]. People with type 2 dia-
betes are at a 2–3-fold higher risk of developing cardiovascular disease
The global prevalence of type 2 diabetes epidemic has now and premature death [3], and are at a 5 increased risk of renal failure
increased for 3 consecutive decades [1]. Currently, approximately 450 and a 10–20 increased risk of amputation and blindness [4].
million adults live with type 1 or type 2 diabetes, which is projected to It is established that improving glycemic control can reduce the
risks of microvascular complications and cardiovascular disease events

Abbreviations: FPG, fasting plasma glucose; GRADE, Grading of Recommendations Assessment, Development and Evaluation; ICEMAN, Instrument to assess the Credibility of
Effect Modification Analyses; MCID, minimal clinically important difference; RCT, randomized controlled trial; SBP, systolic blood pressure.
* Corresponding author.
E-mail address: s_shabbidar@tums.ac.ir (S. Shab-Bidar).

https://doi.org/10.1016/j.ajcnut.2023.03.018
Received 27 July 2022; Received in revised form 7 March 2023; Accepted 17 March 2023
Available online 25 March 2023
0002-9165/© 2023 American Society for Nutrition. Published by Elsevier Inc. All rights reserved.
A. Jayedi et al. The American Journal of Clinical Nutrition 117 (2023) 870–882

[5] and that diet therapy is a key component in type 2 diabetes man- existing type 2 diabetes, with or without cardiovascular conditions and
agement programs [6]. This has led to recommendations that, with the regardless of medication use or glucose concentration and HbA1c level
help of a healthcare team, people with type 2 diabetes should adopt an (aged 18 y); 2) the effects of a prespecified calorie-restricted diet,
individualized healthy eating plan to optimize glycemic control and regardless of macronutrient composition, against a control diet were
long-term health [6]. More recently, more intensive caloric restriction evaluated; 3) considered any of these outcomes including type 2 dia-
and structured lifestyle interventions have been shown to achieve betes remission, change in body weight, HbA1c, fasting plasma
remission from type 2 diabetes [7]. However, the effect of calorie re- glucose (FPG), systolic blood pressure (SBP), total cholesterol, LDL
striction on the remission and management of type 2 diabetes has not cholesterol, HDL cholesterol, or triglyceride as the outcome of interest;
been comprehensively evaluated. 4) provided mean and SD of change in the aforementioned outcomes or
There are several meta-analyses of randomized trials demonstrating reported sufficient information to estimate those values; and 5) pro-
the efficacy of different structured dietary programs, including vided the amount of calorie intake in the intervention group. In general,
carbohydrate-restricted, Mediterranean, or vegetarian dietary patterns 2500 kcal/d and 2000 kcal/d are considered the average daily calorie
for type 2 diabetes management [8–10]. However, limited evidence is intake for weight maintenance in men and women, respectively [18].
available regarding the dose-dependent effects of calorie restriction on Therefore, trials that implemented a very low-calorie diet (500–800
type 2 diabetes management. In addition, evidence is lacking regarding kcal/d) [19], low-calorie diet (800–1200 kcal/d) [18], and moderately
the effects of dietary interventions on type 2 diabetes remission. Two restricted calorie diet (1200–1500 kcal/d) [18] and trials that imple-
previous reviews, without meta-analysis, have assessed the efficacy of mented a slight calorie restriction of between 1500 and 1800 kcal/d
calorie restriction on the remission of diabetes; however, their results were considered eligible for the present meta-analysis [18].
are limited by the inclusion of single-arm trials and lack of a Eligible control groups were waitlist controls, usual or conventional
comprehensive systematic search [11,12]. A recent meta-analysis of 23 diets, routine care, dietary guidelines, calorie-restricted diets, or life-
trials indicated that low and very low carbohydrate diets may be style and behavioral recommendations. Trials that implemented 2
effective interventions for the remission of diabetes at 6 mo but not at calorie-restricted diets with different degrees of calorie restriction as
12 mo follow-up [13]. intervention and comparator (e.g., a very low-calorie diet compared
Limited evidence is available regarding the effect of calorie re- with a low-calorie diet) were considered eligible. We excluded trials
striction on type 2 diabetes remission. In addition, the optimum degree that compared the effects of 2 isocaloric dietary interventions with
of calorie restriction to choose in patients with type 2 diabetes has not different macronutrient compositions.
been ascertained. Therefore, we aimed to perform a systematic review
and meta-analysis of randomized trials to assess the effect of calorie- Outcomes
restricted diets on type 2 diabetes remission and to determine how Our primary outcome was type 2 diabetes remission defined as
well the values of cardiometabolic risk factors change along with the follows: 1) type 2 diabetes remission defined according to the recent
decrease in caloric intake in patients with type 2 diabetes. definition proposed by the American Diabetes Association as an
HbA1c level of <6.5% or a fasting glucose level of <126 mg/dL at
Methods least 6 mo after beginning the intervention and 3 mo after cessation of
antidiabetic medications [20]; 2) an HbA1c level of <6.5% or a fasting
To conduct the present systematic review, we followed the glucose level of < 126 mg/dL over any time frame without diabetes
Cochrane Handbook for Systematic Reviews of Interventions [14] and medication use [21]; and 3) type 2 diabetes remission defined as an
the Grading of Recommendations Assessment, Development, and HbA1c level of <6.5% or a fasting glucose level of < 126 mg/dL over
Evaluation (GRADE) Handbook [15]. We registered the protocol of the any time frame and irrespective of medication change.
systematic review in PROSPERO as CRD42022300875 [16]. Secondary outcomes included change in body weight (kg), HbA1c
(%), and adverse events after implementing calorie-restricted diets.
Other secondary outcomes included changes in FPG, serum total, LDL
Systematic search and HDL cholesterol concentrations, serum triglycerides, SBP, anti-
We performed a systematic search in PubMed, Scopus, CENTRAL, diabetic medication usage, and quality of life. All outcomes were
and Web of Science up to January 2022, and an updated search up to assessed separately across 2 time periods, including 6
3 mo (3 to
November 2022. We also searched trial registries (clinicaltrials.gov), 9 mo; 6-mo follow-up) and 12
3 mo (<9 to 15 mo; 12-mo
gray literature sources (ProQuest Dissertations & Theses Global), and follow-up). We also performed an additional post-hoc analysis to test
the reference lists of included articles and relevant reviews. There was the effect of calorie restriction on diabetes remission at a follow-up of
no restriction on language, date, and publication status. Titles and >15 mo (2-y follow-up).
abstracts were screened independently by 2 reviewers (AJ and SSB)
according to the predefined inclusion and exclusion criteria to identify
Data extraction
eligible studies. Full texts were retrieved and independently assessed
After the study selection process, 2 reviewers (HSF and SZM)
for eligibility by the same reviewers. Any disagreements were resolved
independently and in duplicate extracted the general characteristics
by consensus. The complete search strategy, including the Cochrane
from each trial. Disagreements were resolved by consulting a third
recommended filter for the identification of randomized controlled
reviewer (AJ). Detailed information extracted from each trial included
trials in PubMed [17], is provided in Supplemental Table 1.
the last name of the first author, year of publication, sample size, mean
age, baseline weight status, intervention duration, description of
Study selection and eligibility criteria intervention/control arms, the average daily calorie intake in the
Published human intervention studies were considered eligible if intervention and control groups (if reported), exercise and/or physical
they had the following criteria: 1) randomized controlled trials, with a activity, behavioral support (yes/no; if yes, behavioral support details),
minimum intervention period of 12 wk, conducted in adults with intensity of lifestyle modification, and mean and corresponding SD of

871
A. Jayedi et al. The American Journal of Clinical Nutrition 117 (2023) 870–882

change in outcomes from baseline for each arm. For trials reporting mean data, we converted the former to mean data using standard
different effect sizes across one time period (e.g., 3 and 6 mo), the methods [29,30].
results for the longest duration in each period were included. Second, we used the method introduced by Crippa and Orsini [26]
to calculate the MD and its corresponding standard error of change in
Risk of bias assessment outcomes for each 500-kcal decrease in daily calorie intake in the
Two authors (AJ and SS-B) independently assessed the risk of bias intervention group relative to the control group in each trial.
of the trials using guidance outlined in the version 2 of the Cochrane Trial-specific results were pooled using a random-effects model. This
risk-of-bias tool (RoB 2) [14], with disagreements resolved by method requires a value of dose (average calorie) in each study arm, the
consensus. mean and its corresponding SD of change in each study arm, and the
number of participants in each arm. For trials that reported calorie
Statistical analysis intake as a range (e.g., 1000–1200 kcal/d), the midpoint of the lower
We performed separate meta-analyses for each outcome in each and upper bounds was used. For trials that presented calorie intake
time period (6
3 mo and 12
3 mo) using the random-effects meta- across different periods, we used period-specific calorie intake for the
analyses with restricted maximum likelihood assumption [22]. Due to corresponding meta-analysis. For trials that reported calorie intake at
the low number of trials included in the analyses, the restricted different intervals in 1 time period (e.g., reported calorie intake at 3-mo
maximum likelihood assumption may be better than the traditional and 6-mo follow-ups), we used the average calorie intake over each
DerSimonian and Laird’s random-effects meta-analysis. For the ana- time period for the analyses. Nonlinear dose–response associations
lyses of binary outcomes, we performed pairwise meta-analyses to were assessed with restricted cubic splines with 3 knots at Harrell’s
calculate both absolute and relative effects with their 95% CIs using the recommended centiles (10%, 50%, and 90%) [26,31]. STATA syntax
number of events and the total number of participants in each study arm that was used for the analyses is provided in Supplemental Method 1.
in each trial. To calculate the relative effect, we estimated the RR and As per our a priori protocol [16], we performed prespecified sub-
its 95% CI in each trial by dividing the probability of each binary group analyses based on the risk of bias (low risk versus high risk/some
outcome in the intervention arm by the probability of that outcome in concerns), presence of physical activity, behavioral support, intensive
the control arm. Trial-specific effect estimates were pooled using a lifestyle modification in the intervention program (yes/no), and base-
random-effects meta-analysis. However, for the analyses of remission line HbA1c (<7, 7–8, 8%). We defined intensive lifestyle interven-
at the 12-mo and 2-y endpoints, 4 and 3 trials were correspondingly tion as an intensive weight loss program that included a combination of
included, one of which had at least 20 times more patients than the a calorie-restricted diet (500–750-kcal/d energy deficit, which in most
second largest trial. In such cases, random-effects meta-analysis may cases is ~1200–1500 kcal/d for females and ~1500–1800 kcal/d for
not be the most appropriate approach for meta-analysis because of the males), exercise program, and a high frequency of counseling [27]. Due
role that the large tau2 plays in the calculation of weights in a to a lack of information, we did not perform subgroup analyses based
random-effects model [23]. Therefore, we performed a secondary on age and sex. We considered subgroup differences credible based on
fixed-effects meta-analysis for diabetes remission at the 12-mo and 2-y 8 criteria introduced by the Instrument to assess the Credibility of
endpoints. To calculate the absolute effect, we estimated the risk dif- Effect Modification Analyses (ICEMAN) [32]. We also performed
ference (RD) and its 95% CI using the pooled RR. We calculated post-hoc subgroup analyses based on insulin use (yes/no), baseline
pooled RR and then converted the pooled RR to RD using baseline risk weight status (overweight/obese versus mixed), duration of diabetes
[24]. We used the following formula to convert the pooled RR to RD: (<5 versus 5 y), type of comparator, degree of calorie restriction
RD ¼ baseline risk (RR  1) [25]. Baseline risk was estimated using (very low-, low-, moderate-calorie diet), degree of adherence to the
the remission rate in the control group in the largest trial [24]. We did intervention program (good versus not reported), and macronutrient
not calculate study-specific RDs because baseline risk varied consid- composition of the diet in the intervention program (moderate-to-low
erably across trials, and in such cases, this method may result in a carbohydrate, low fat, balanced). We defined the intervention program
misleading result [24]. as moderate-to-low carbohydrate (45% calorie from carbohydrate)
For the analyses of continuous outcomes, we performed random- [33], low fat (<30% calorie from fat), or balanced (>45% calorie from
effects dose–response meta-analyses with studies with sufficient in- carbohydrate and >30% calorie from fat). P values for subgroup dif-
formation to estimate the mean difference (MD) and 95% CI of change ferences were generated using meta-regression analysis. We also per-
in primary and secondary outcomes per 500-kcal decrease in daily formed sensitivity analysis using the Hartung–Knapp adjustment when
calorie intake in the intervention group relative to the control group [8, a small number of studies (n5) were available for the analysis [34,35].
26]. We selected 500 kcal/d since this threshold (500–750 kcal/d) was This method estimates a more conservative point estimate and a wider
recommended by the American Diabetes Association to achieve and CI [36].
maintain 5% weight loss for most overwight or obese people with Publication bias was tested using Egger’s test (P < 0.05) [37], and
type 2 diabetes [27]. inspection of the funnel plot was performed when 10 trials were
To be included in dose–response meta-analyses, studies must report available for the analysis. We evaluated heterogeneity using the I2
the average calorie intake in both the intervention and control groups. statistic and performed a χ2 test for homogeneity (Pheterogeneity > 0.10)
First, we calculated changes from baseline values in each study arm. If [14]. Statistical analyses were conducted using STATA software
the mean values and SDs of changes were not available in text or version 17.0. A two-tailed P value of <0.05 was considered significant.
graphs, we calculated these values using data from measures before and
after the intervention based on the Cochrane Handbook guidance [14]. Grading the evidence
For trials that reported standard error instead of SD, the former was We applied the GRADE approach to rate the certainty of the evi-
converted to SD [14]. If neither SD nor standard error was reported in dence for each outcome in each time period [38]. Two authors (AJ and
the trials, we used the average SDs obtained from other trials included SSB) independently performed GRADE assessment. GRADE rates the
in the meta-analysis [28]. For trials that reported median data instead of certainty of evidence as high, moderate, low, and very low.

872
A. Jayedi et al. The American Journal of Clinical Nutrition 117 (2023) 870–882

Randomized trials start as high certainty evidence that can then be 51–54,57,58,64–66]; 4 trials implemented a calorie-restricted diet [38,
downgraded based on prespecified criteria. The criteria used to 39,60,67]; 2 trials implemented a standard behavioral therapy [59,61];
downgrade evidence include the risk of bias, inconsistency, impreci- 1 trial implemented a carbohydrate-restricted diet [42]; 1 trial imple-
sion, indirectness, and publication bias. The evidence can also be mented educational sessions about a healthful diet during monthly
upgraded due to the large effect size and dose–response gradient. visits [62]; 1 trial implemented a usual diet [63]; and another trial
Detailed information on the domains of the GRADE tool and how to implemented an education control condition including 3 group sessions
judge each domain is provided Supplemental Method 2. per year on diet, physical activity, and social support [45].
Ten trials excluded patients who used insulin treatment [44,46–48,
Results 51,54,57–59,66]; 2 trials included only patients who used insulin
treatment [39,50], and the other 16 trials included mixed patients
[40–43,45,49,52,53,55,56,60–65]. All but 2 trials [42,47] were con-
Systematic search ducted in overweight or obese patients with type 2 diabetes (BMI 25
Our systematic search in the literature found 10,078 records (Sup-
kg/m2). The average duration of diabetes was 5 y in 9 trials [40,46,48,
plemental Figure 1). We excluded 2547 duplicates and an additional
49,57,58,64–66], and more than 5 y in the other 12 trials [39,41–43,45,
7391 records by screening the title and abstract. We read the full text of
47,51,54,55,60–62]. Seven trials did not report sufficient information
140 articles, and finally, 28 randomized trials with 6281 participants
on the duration of diabetes [44,50,52,53,56,59,63]. The average
were included in the analyses [39–66]. We included 2 publications
HbA1c concentration at the baseline of the trials was <7% in 8 trials
from the same trial [64,65], of which the first publication was included
[47,53,57,60,61,63–65], between 7 and 8% in 15 trials [40,42–46,48,
in the analyses of continuous outcomes [64] and the second publication
49,51,54–56,58,62,66], and 8% in 4 trials [39,41,50,59]. One trial
that was a secondary analysis and reported information on diabetes
did not report sufficient information on the average HbA1c concen-
remission was included in the analysis of diabetes remission [65]. We
tration at baseline [52]. With regard to the macronutrient composition
also included 2 publications from the DiRECT trial; of those, one
of the diet in the intervention program, 8 trials implemented a low fat
publication was included in the analysis of diabetes remission at the
diet (<30% calorie from fat) [39,40,42,45,48,50,57,66], 8 trials
12-mo follow-up [48], and the second publication was included in the
implemented a moderate-to-low carbohydrate diet (<45% calorie from
analysis of remission at 2-y follow-up [66]. Five trials did not report
carbohydrate) [43,44,47,51,54–56,63], and 6 trials a balanced
sufficient information on dose–response meta-analyses (the average
calorie-restricted diet (>45% carbohydrate and >30 fat) [41,46,53,62,
calorie intake in the control group) [45,46,48,57,66]. We contacted the
64,65]. Six trials did not report information on the macronutrient
authors to obtain additional information; and among those, the authors
composition of the diet in their intervention program [49,52,58–61].
of the DiRECT trial stated that they did not quantitate calorie intake in
Fourteen trials (50%) were rated to have a low risk of bias [41,44–46,
the control group [48,66], one study author provided the requested
48,53–55,57,62–66], 2 trials were rated to have some concerns [43,56],
information [46], and another trial did not respond [57]. We read other
and the other 12 trials (43%) were judged to have a high risk of bias
publications of the Look AHEAD trial to find the average calorie intake
[39,40,42,47,49–52,58–61]. Supplemental Table 4 presents the risk of
in the control group, but we did not find any information [45]. For the
bias assessment of the trials included in the present review.
analysis of diabetes remission at 6 mo, 3 trials did not report the
number of patients with diabetes remission [44,47,55], and thus, we
used the data (number of patients with diabetes remission) presented in Primary outcomes
a previous meta-analysis [13] to calculate the RR in those trials. The list At 6 mo, 10 trials reported information on type 2 diabetes remission
of studies excluded via full-text assessment and reasons for exclusion [43,44,46,47,50,51,54,55,58,59]. One trial defined diabetes remission
are indicated in Supplemental Table 2. as an HbA1c level of <6.5% after at least 2 mo off all antidiabetic
medications [51]. The remission rate was significantly higher in the
intervention group as compared to the usual diet (11/21 [52.3%]
Characteristics of the original trials compared with 0/12 [0%]; RD: 34 more per 100; 95% CI: 15 more, 53
Supplemental Table 3 presents the general characteristics of the more; n ¼ 1 trial; GRADE ¼ very low). When remission was defined
trials included in the present meta-analysis. In brief, the included trials as an HbA1c level of <6.5% without medication use [43,46,50,51,58],
were published between 1991 and 2022. Eight trials implemented a calorie-restricted diets increased remission by 38 more per 100 patients
very low-calorie diet (500–800 kcal/d) [39,46,48,52,57,59,60,66], 5 (54/166 [32.5%] compared with 0/153 [0%]; RD: 38 more per 100
trials implemented a low-calorie diet (800–1200 kcal/d) [43,49,51,58, patients; 95% CI: 9 more, 67 more; n ¼ 5; GRADE ¼ moderate) in
61], 8 trials implemented a moderately restricted calorie diet comparison to the usual diet or usual care (Supplemental Figure 2,
(1200–1500 kcal/d) [40,44,45,54–56,62,63], and 7 trials implemented Table 1).
a calorie-restricted diet with an average calorie intake of between 1500 When remission was defined as an HbA1c level of <6.5% irre-
and 1800 kcal/d [41,42,47,50,53,64,65]. Of the trials, 21 trials imple- spective of medication use [44,47,54,55,59], calorie-restricted diets
mented behavioral support alongside calorie restriction [39,41,43,45, increased remission of diabetes by 6 more per 100 patients (26/63
48,50–52,54–66], 14 trials implemented an exercise program [39,41, [40.6%] compared with 5/64 [7.8%]; RD: 6 more per 100 patients; 95%
44,45,48,49,56,59–62,64–66], and 11 trials implemented an intensive CI: 1 more, 17 more; n ¼ 5 trials; GRADE ¼ moderate) in comparison
lifestyle modification program [39,45,48,57,59–62,64–66]. Nineteen to usual diet or usual care (Supplemental Figure 3; Table 1).
trials reported good, high, considerable, or excellent adherence to the Four trials reported remission of diabetes at 12 mo [45,48,57,65].
intervention program [42,43,46–48,51–55,57–59,61–66]; 1 trial re- All trials were at a low risk of bias, implemented an intensive lifestyle
ported suboptimal adherence [56]; and the other 8 trials did not provide modification including a calorie-restricted diet, and defined remission
information regarding the degree of adherence to the intervention of diabetes as an HbA1c level of <6.5% in the absence of antidiabetic
[39–41,44,45,49,50,60]. With regard to comparators, 18 trials used medication use. In comparison to the usual diet or usual care,
standard of diabetic care in the control group [39–41,43,44,46–49, calorie-restricted diets increased remission of diabetes by 13 more per

873
A. Jayedi et al. The American Journal of Clinical Nutrition 117 (2023) 870–882

TABLE 1
Summary of the effects of calorie restriction on primary and secondary outcomes in patients with type 2 diabetes1.
Outcomes 6 mo (3–9 mo) 12 mo (9–15 mo)
Participants Effect size (95% Certainty of Participants Effect size (95% Certainty of
(trials) CI) evidence (trials) CI) evidence
(GRADE) (GRADE)
Primary outcomes
Type 2 diabetes remission (no diabetes 33 (1) RD ¼ 34 more per Very low 399 (2) RD ¼ 16 more per Low
medication þ HbA1c <6.5% after at least 2 100 (15 more, 53 100 (4 more, 49
mo off all antidiabetic medications) more) more)
RR ¼ 13.59 (6.67, RR ¼ 6.94 (2.55,
20.59) 18.85)
Type 2 diabetes remission (no diabetes 319 (5) RD ¼ 38 more per Moderate 5049 (4) RD ¼ 13 more per Moderate
medication þ HbA1c <6.5%) 100 (9 more, 67 100 (10 more, 18
more) more)
RR ¼ 14.84 (4.18, RR ¼ 5.94 (4.59,
52.71) 7.69)
Type 2 diabetes remission (HbA1c <6.5% 127 (5) RD ¼ 6 more per Moderate — — —
regardless of medication use) 100 (1 more, 17
more)
RR ¼ 3.15 (1.39, —
7.13)
Secondary outcomes
Weight loss (kg; per 500-kcal/d decrease in 1450 (22) MD ¼ 6.33 High 504 (6) MD ¼ 4.41 Moderate
energy intake) (7.76, 4.90) (6.57, 2.26)
HbA1c (%; per 500-kcal/d decrease in energy 1152 (18) MD ¼ 0.82 High 504 (6) MD ¼ 0.65 Low
intake) (1.05, 0.59) (1.07, 0.24)
Serious adverse events 424 (4) RD ¼ 1 more per Low 741 (5) RD ¼ 3 more per Low
100 (1 fewer, 3 100 (0 fewer, 6
more) more)
RR ¼ 1.69 (0.61, RR ¼ 1.37 (0.49,
4.72) 1.85)
Hypoglycemic reactions 255 (3) RD ¼ 5 more per Low 288 (3) RD ¼ 17 more per Low
100 (3 fewer, 13 100 (8 more, 26
more) more)
RR ¼ 2.20 (0.47, RR ¼ 1.56 (1.20,
10.34) 2.02)
Fasting plasma glucose (mmol/L; per 500-kcal/ 1038 (15) MD ¼ 1.23 High 504 (6) MD ¼ 1.16 Moderate
d decrease in energy intake) (1.62, 0.85) (1.69, 0.64)
Total cholesterol (mmol/L; per 500-kcal/ 825 (15) MD ¼ 0.06 Moderate 314 (4) MD ¼ 0.27 (0.04, Low
d decrease in energy intake) (0.22, 0.09) 0.57)
LDL cholesterol (mmol/L; per 500-kcal/ 915 (14) MD ¼ 0.02 Low 368 (4) MD ¼ 0.22 (0.05, Low
d decrease in energy intake) (0.11, 0.15) 0.40)
HDL cholesterol (mmol/L; per 500-kcal/ 1016 (16) MD ¼ 0.05 (0.02, High 404 (5) MD ¼ 0.09 (0.05, Moderate
d decrease in energy intake) 0.09) 0.12)
Triglycerides (mmol/L; per 500-kcal/d decrease 912 (17) MD ¼ 0.20 High 404 (5) MD ¼ 0.19 Very low
in energy intake) (0.34, 0.05) (0.61, 0.23)
Systolic blood pressure (mmHg; per 500-kcal/ 887 (14) MD ¼ 5.69 High 368 (4) MD ¼ 0.88 Very low
d decrease in energy intake) (8.53, 2.85) (4.04, 2.27)
Medication reduction (no. of patients who 480 (6) RD ¼ 21 more per Low 262 (3) RD ¼ 26 more per Low
reduced antidiabetic medications) 100 (9 more, 34 10000 (6 more, 47
more) more)
RR ¼ 3.36 (1.49, RR ¼ 1.87 (1.33,
7.60) 2.63)
Medication discontinuing (no. of patients who 263 (3) RD ¼ 24 more per Low 484 (3) RD ¼ 42 more per Low
discontinued antidiabetic medications) 100 (5 more, 43 1000 (15 more, 69
more) more)
RR ¼ 6.10 (1.23, RR ¼ 6.47 (2.34,
59.35) 17.92)
Health-related quality of life 214 (2) SMD ¼ 1.00 (0.71, Low 519 (3) SMD ¼ 0.76 (0.10, Low
1.28) 1.42)

Abbreviations: MD, mean difference; RD, risk difference; SMD, standardized mean difference.
1
Results are from random-effects meta-analyses.

100 patients (392/2530 [15.5%] compared with 63/2519 [2.5%]; RD: Figure 4; Table 1), and by 12 more per 100 patients (95% CI: 10 more,
13 more per 100 patients; 95% CI: 10 more, 18 more; n ¼ 4; GRADE 18 more) in the fixed-effects meta-analysis (Supplemental Figure 5).
¼ moderate) in the random-effects meta-analysis (Supplemental Two trials defined diabetes remission as an HbA1c level of <6.5% after

874
A. Jayedi et al. The American Journal of Clinical Nutrition 117 (2023) 870–882

at least 2-mo cessation of all antidiabetic medications (RD: 16 more per energy intake significantly reduced HbA1c (MD: 0.82%; 95% CI:
100 patients; 95% CI: 4 more, 49 more; n ¼ 2; GRADE ¼ low; 1.05, 0.59; I2 ¼ 66%; n ¼ 18 trials with 1152 participants; GRADE
Supplemental Figure 6, Table 1) [48,65]. ¼ high; Figure 1; Table 1). The corresponding reduction in HbA1c at
Three trials reported information on diabetes remission at a 2-y 12 mo [39,56,60–62,64] was 0.65% (95% CI: 1.07, 0.24; I2 ¼
follow-up [45,65,66], all of which defined diabetes remission in the 49%; n ¼ 6 trials with 504 participants; GRADE ¼ low; Supplemental
absence of antidiabetic medication use. In comparison to the usual diet Figure 11; Table 1). Calorie-restricted diets did not significantly in-
or usual care, calorie-restricted diets increased the remission rate by 11 crease serious adverse events at the 6-mo and 12-mo follow-ups in
more per 100 patients (286/2298 [12.44%] compared with 56/2278 comparison to usual diet or usual care, but they did increase hypo-
[2.46%]; RD: 11 more per 100 patients; 95% CI: 4 more, 24 more; RR: glycemic reactions at the 12-mo follow-up (RD: 17 more per 100 pa-
5.05; 95% CI: 2.59, 9.85; GRADE ¼ low; Supplemental Figure 7) in tients; 95% CI: 8 more, 26 more; n ¼ 3 trials with 288 participants;
the random-effects meta-analysis, and by 10 more per 100 patients GRADE ¼ low; Supplemental Figures 12–15, Table 1).
(RR: 10 more per 100 patients; 95% CI: 7 more, 15 more; RR: 4.78, Table 1 presents the effects of calorie restriction on other secondary
95% CI: 3.61, 6.33; Supplemental Figure 8) in the fixed-effects outcomes (Supplemental Figures 16–27). In brief, each 500-kcal
meta-analysis. decrease in daily energy intake resulted in a significant and clinically
meaningful reduction in serum FPG at 6 mo (MD: 1.23 mmol/L; 95%
Secondary outcomes CI: 1.62, 0.85; I2 ¼ 71%; n ¼ 15 trials with 1308 participants;
Twenty-two trials reported information on weight loss at 6 mo GRADE ¼ high) and 12 mo (MD: 1.16 mmol/L; 95% CI: 1.69,
[40–44,46,47,49–63]. Each 500 kcal decrease in daily calorie intake 0.64; I2 ¼ 0%; n ¼ 6 trials with 504 participants; GRADE ¼ mod-
reduced body weight (MD: 6.33 kg; 95% CI: 7.76, 4.90; I2 ¼ erate), as well as in SBP at 6 mo (MD: 5.69 mmHg; 95% CI: 8.53,
91%; n ¼ 22 trials with 1450 participants; GRADE ¼ high; Supple- 2.85; I2 ¼ 58%; n ¼ 14 trials with 887 participants; GRADE ¼ high).
mental Figure 9, Table 1). Of note, calorie-restricted diets resulted in a Calorie-restricted diets had little or no effect on serum total and LDL
large weight loss in the subgroup of trials with a low risk of bias (MD: cholesterol at 6 mo; however, they did increase serum LDL cholesterol
8.30 kg; 95% CI: 9.79, 6.81; I2 ¼ 81%; n ¼ 11 trials with 999 concentration at the 12-mo follow-up. The increase in serum LDL
participants; test for subgroup difference < 0.001; ICEMAN credibility cholesterol concentration at 12 mo was larger than the minimal clini-
¼ moderate). At 12-mo [39,56,60–62,64], each 500-kcal decrease in cally important difference (MCID) threshold.
daily energy intake reduced body weight (MD: 4.41 kg; 95% CI: Based on the information obtained from 3 trials with 519 partici-
6.57, 2.26; I2 ¼ 99%; n ¼ 6 trials with 504 participants; GRADE ¼ pants [43,58,62], calorie restriction led to a significant and large
moderate; Supplemental Figure 10, Table 1). (standardized mean difference > 0.80) [68] increase in health-related
Eighteen trials were included in the analysis of HbA1c at 6 mo quality of life in comparison to usual diet or usual care at 6 mo, but
[40–44,46,47,49–51,53–55,58,60–63]. Each 500-kcal/d decrease in the certainty of the evidence was rated low because of a serious risk of

FIGURE 1. The effects of each 500-kcal/d decrease in daily energy intake on HbA1c (%) in patients with type 2 diabetes at 6 mo. Results are from random-
effects meta-analysis. REML, restricted maximum likelihood.

875
A. Jayedi et al. The American Journal of Clinical Nutrition 117 (2023) 870–882

bias due to selective outcome reporting and serious imprecision due to behavioral support led to a larger reduction (MD: 0.27 mmol/L; 95%
a sample size of smaller than the optimal information size (n ¼ 800) CI: 0.42, 0.11; n ¼ 9 trials with 551 participants; test for subgroup
[69]. At the 12 mo, calorie restriction led to a small increase in difference < 0.001) than trials that did not. The results for subgroup
health-related quality of life (Supplemental Figures 28 and 29, Table 1). analyses and the ICEMAN credibility assessment for each subgroup
analysis are presented in Supplemental Tables 5–12 and Supplemental
Dose–response meta-analysis Tables 13–20, respectively.
Dose-dependent effects of calorie restriction at 6-mo follow-up, There were 3 significant subgroup differences based on the degree
obtained from the nonlinear dose–response meta-analyses, are pre- of adherence to the intervention program at 6 mo, where trials with
sented in Table 2. Calorie-restricted diets resulted in clinically impor- good adherence reported stronger effects on body weight (MD: 7.25
tant benefits that were 3–8-fold larger than the MCID thresholds for kg; 95% CI: 9.33, 5.18; n ¼ 15 trials with 1028 participants; test for
most cardiometabolic outcomes. Levels of body weight (Figure 2), subgroup difference ¼ 0.02), HbA1c (MD: 0.96%; 95% CI: 1.26,
HbA1c (Figure 3), FPG (Supplemental Figure 30), and SBP (Supple- 0.65; n ¼ 12 trials with 817 participants; test for subgroup difference
mental Figure 31) decreased linearly with the decrease in calorie intake ¼ 0.03), and serum HDL cholesterol (MD: 0.08 mmol/L; 95% CI:
from 2500 to 750 kcal/d. Dose–response meta-analyses indicated a 0.03, 0.15; n ¼ 11 trials with 699 participants; test for subgroup dif-
modest U-shaped effect for total cholesterol (Figure 4) and LDL ference ¼ 0.02). However, in those cases, chance was a likely expla-
cholesterol (Figure 5), with the greatest reduction at 1500 kcal/d for nation (P for subgroup difference between 0.01 and 0.05) [32] and the
both outcomes. Levels of serum triglycerides decreased linearly up to credibility of subgroup differences was rated low.
1250 kcal/d, with flattening of the curve at a lower energy intake There was also a significant subgroup difference for body weight at
(Figure 6). Dose–response meta-analysis indicated a modest increase in 6 mo, where trials with a balanced low-calorie diet led to a large
serum HDL cholesterol with the decrease in calorie intake (Supple- reduction in body weight (MD: 10.26 kg; 95% CI: 11.11, 9.41; n
mental Figure 32). Due to the limited number of studies, we were ¼ 4 trials with 327 participants) as compared to moderate-to-low
unable to perform nonlinear dose–response meta-analyses at a 12-mo carbohydrate (MD: 6.77 kg; 95% CI: 8.34, 5.21; n ¼ 8 trials
follow-up. with 571 participants) and low fat diets (MD: 5.91 kg; 95% CI:
8.30, 3.52; n ¼ 4 trials with 291 participants, all having a subgroup
Subgroup analyses difference of <0.001); however, the credibility of subgroup difference
As per our a priori protocol [16], we did predefined subgroup an- was rated low because this subgroup was not based on a priori hy-
alyses and applied the ICEMAN criteria to identify credible subgroup pothesis and the number of trials in the smallest subgroup was rather
differences [32]. We did not find any subgroup difference with high small (n ¼ 4 trials; Supplemental Table 5). There was no other sig-
credibility. We found 2 subgroup differences with moderate credibility: nificant or credible difference across subgroups defined by the diet
first, in the analyses of body weight at 6-mo where trials with a low risk composition.
of bias indicated a large weight loss (MD: 8.30 kg; 95% CI: 9.79, We also performed sensitivity analyses using the Hartung–Knapp
6.81; n ¼ 11 trials with 999 participants; test for subgroup difference adjustment when a small number of trials (n 5) were available for the
< 0.001) than those with high risk of bias/some concerns, and second, analyses. The results indicated that the remission rate became weaker
in the analysis of total cholesterol at 6 mo, where trials that involved but remained statistically significant at the 6-mo and 12-mo follow-ups

TABLE 2
The effects of calorie restriction on cardiometabolic outcomes at 6-mo follow-up from the nonlinear dose–response meta-analysis (mean difference and 95% CI)1.
Calorie 2500 2250 2000 1750 1500 1250 1000 750 MCID
intake (kcal/ (Ref)
d)
Weight (kg) 0 2.42 4.84 (6.84, 7.27 9.80 12.55 15.47 18.46 4.4 kg
(3.42, 2.84) (10.25, (13.45, (16.27, (19.12, (22.44,
1.42) 4.28) 6.15) 8.83) 11.82) 14.48)
HbA1c (%) 0 0.36 0.73 (1.44, 1.09 (2.16, 1.49 (2.78, 1.93 (3.23, 2.41 (3.62, 2.89 (4.03, 0.50%
(0.72, 0.01) 0.02) 0.19) 0.63) 1.20) 1.20)
0.01)
LDL (mmol/ 0 0.11 0.21 (0.43, 0.31 (0.64, 0.35 (0.74, 0.28 (0.66, 0.17 (0.52, 0.06 (0.41, 0.10
L) (0.21, 0.00) 0.01) 0.01) 0.03) 0.09) 0.17) 0.29) mmol/L
TC (mmol/L) 0 0.17 0.34 (0.65, 0.51 (0.98, 0.65 (1.19, 0.63 (1.22, 0.58 (1.14, 0.51 (1.04, 0.26
(0.33, 0.02) 0.04) 0.04) 0.04) 0.01) 0.03) mmol/L
0.01)
FPG (mmol/ 0 0.51 1.01 (2.06, 1.52 (3.13, 2.06 (4.00, 2.65 (4.59, 3.28 (5.03, 3.91 (5.49, 1.60
L) (1.04, 0.03) 0.06) 0.08) 0.12) 0.71) 1.52) 2.34) mmol/L
HDL (mmol/ 0 0.01 (0.04, 0.02 (0.08, 0.03 (0.12, 0.04 (0.13, 0.07 (0.11, 0.10 (0.07, 0.14 (0.02, 0.10
L) 0.06) 0.11) 0.17) 0.22) 0.25) 0.27) 0.30) mmol/L
TG (mmol/L) 0 0.18 0.37 (0.66, 0.55 (0.99, 0.69 (1.22, 0.75 (1.27, 0.77 (1.25, 0.77 (1.26, 0.09
(0.33, 0.07) 0.11) 0.16) 0.24) 0.22) 0.28) mmol/L
0.04)
SBP (mmHg) 0 3.19 6.37 9.56 12.67 15.66 18.59 21.51 2 mmHg
(6.29, (12.55, (18.77, (23.94, (27.41, (30.24, (33.31,
0.10) 0.19) 0.34) 1.40) 3.92) 6.95) 9.71)

Abbreviations: MCID, minimal clinically important difference; SBP, systolic blood pressure; TC, total cholesterol; TG, triglyceride.
1
Results are from random-effects meta-analyses.

876
A. Jayedi et al. The American Journal of Clinical Nutrition 117 (2023) 870–882

FIGURE 2. Dose-dependent effects of calorie restriction on body weight (kg) at 6 mo. Results are from random-effects dose–response meta-analysis.

FIGURE 3. Dose-dependent effects of calorie restriction on HbA1c (%) at 6 mo. Results are from random-effects dose–response meta-analysis.

FIGURE 4. Dose-dependent effects of calorie restriction on serum total cholesterol concentration (in mmol/L) at 6 mo. Results are from random-effects
dose–response meta-analysis.

877
A. Jayedi et al. The American Journal of Clinical Nutrition 117 (2023) 870–882

FIGURE 5. Dose-dependent effects of calorie restriction on serum LDL cholesterol concentration (in mmol/L) at 6 mo. Results are from random-effects
dose–response meta-analysis.

(Supplemental Table 21). The effects of calorie restriction on reduction of cardiometabolic risk factors in patients with type 2 dia-
increasing LDL cholesterol concentrations at the 12-mo follow-up betes. Based on very low to low certainty evidence obtained from a
remained significant but were attenuated after the Hartung–Knapp very small number of trials, calorie-restricted diets that were imple-
adjustment (Supplemental Table 21). mented within intensive lifestyle modification programs, as compared
to usual diet or usual care, increased the remission rate defined as an
Publication bias HbA1c level of <6.5% after 2-mo cessation of all antidiabetic medi-
cations by 34 more per 100 patients at the 6-mo follow-ups and by 16
There was no evidence of publication bias at the 6-mo follow-up
more per 100 patients at the 12-mo follow-ups. Calorie restriction also
(Supplemental Figures 33–40). We did not assess publication bias
increased the remission rate based on an HbA1c level of <6.5% in the
due to the limited number of studies (n < 10) at the 12-mo follow-up.
absence of medication use by 38 more per 100 patients at the 6-mo
follow-ups and by 13 more per 100 patients at the 12-mo follow-ups
Grading the evidence in comparison to the usual diet or usual care. Based on low certainty
Supplemental Table 22 and Table 1 present the details of the evidence, calorie-restricted diets increased the remission rate, defined
GRADE certainty assessment. as an HbA1c level of <6.5% without medication use, by 11 more per
100 patients at 2-y follow-up, suggesting the long-term efficacy and
Discussion durability of such diets for remission of diabetes. Dose–response meta-
analyses at the 6-mo endpoint showed evidence of moderate-to-high
In this meta-analysis of 28 randomized trials, we found that calorie certainty that each 500-kcal decrease in daily energy intake can
restriction can be an effective strategy for diabetes remission and result in clinically meaningful reductions in levels of most

FIGURE 6. Dose-dependent effects of calorie restriction on serum TG concentration (mmol/L) at 6 mo. Results are from random-effects dose–response meta-
analysis. TG, triglyceride.

878
A. Jayedi et al. The American Journal of Clinical Nutrition 117 (2023) 870–882

cardiometabolic risk factors, except for serum total, LDL, and HDL carbohydrate-restricted diets in patients with type 2 diabetes indicated
cholesterol. At the 12-mo endpoint, the effects of calorie restriction similar U-shaped effects on total and LDL cholesterol at 6-mo follow-
remained large and clinically meaningful for HBA1c and FPG, but up, which may be due to an increase in total fat intake [8]. There was
disappeared or diminished markedly for body weight, serum tri- also evidence of low certainty that calorie-restricted diets may lead to
glycerides, and SBP. Based on a low certainty evidence, calorie- an increase in total and LDL cholesterol concentrations larger than the
restricted diets resulted in a clinically meaningful increase in health- MCID thresholds at 12 mo. This may be due to the cessation of
related quality of life at 6 mo. lipid-lowering medications in patients in the intervention arms [57].
Results from meta-analyses of randomized trials suggested that Therefore, it is important to consider the continuing requirement of
structured dietary programs such as carbohydrate-restricted, Mediter- lipid-lowering medications when implementing low-calorie diets in
ranean, or vegetarian dietary patterns are effective dietary interventions patients with type 2 diabetes.
for type 2 diabetes management [8–10]. A recent dose–response
meta-analysis of randomized trials indicated that each 10% decrease in Clinical implications
energy intake from carbohydrates reduced the body weight by 1.44 kg, Indeed, confusion remains on the degree of calorie restriction to
FPG level by 0.34 mmol/L, and HbA1c level by 0.20% in patients with choose in patients with type 2 diabetes for implementing the most
type 2 diabetes [8]; however, these effects disappeared or attenuated at effective lifestyle interventions. Pairwise comparisons used in tradi-
a follow-up of >6 mo. Indeed, limited evidence is available regarding tional meta-analyses have several limitations and cannot present the
the dose-dependent effects of calorie restriction for type 2 diabetes best information needed for decision-making. In this dose–response
management. meta-analysis, we indicated how well the levels of FPG, HbA1c, body
Recently, the American Diabetes Association recommended that weight, blood lipids, and blood pressure change with the decrease in
most overweight or obese people with type 2 diabetes should adopt a calorie intake from 2500 to 750 kcal/d. We also indicated the corre-
lifestyle intervention program that includes a significant weight loss by sponding reductions in cardiovascular risk factors for each 500-kcal/
a 500–750-kcal/d energy deficit, which in most cases is ~1200–1500 d decrease in energy intake. These information can assist decision
kcal/d for females and ~1500–1800 kcal/d for males [27]. Current makers to develop evidence-based recommendations and help patients
evidence suggests that clinical benefits generally appear after 3–5% and clinicians to choose an effective weight-loss diet for improving
weight loss [18,70], and the benefits of weight loss are progressive; glycemia and levels of cardiovascular risk factors.
with more intensive weight loss indicating further health improve-
ments. Our results presented further supportive evidence, indicating Limitations of this study
that the levels of HbA1c, FPG, and body weight decreased linearly Several limitations exist that deserve consideration. We had limited
along with the decrease in daily energy intake. data for the 12-mo endpoint, especially for diabetes remission. For the
Limited evidence is available regarding the effects of different in- analysis of remission at the 12-mo endpoint, 4 trials were available, one
terventions on diabetes remission. Meta-analyses of randomized trials of which had at least 20 times more patients than the second largest
indicated that bariatric surgeries are effective interventions for diabetes trial. However, all trials received approximately the same weight in the
remission [71,72]. However, invasive interventions such as bariatric summary estimate. In such cases, random-effects meta-analysis may
surgeries have several side effects and need comprehensive post- not be the most appropriate approach for meta-analysis because of the
operative care programs, which limit their routine utilization in practice role that the large tau2 plays in the calculation of weights in a random-
[73,74]. effects model [23]. Therefore, we performed a secondary fixed-effects
A recent meta-analysis of randomized trials suggested that low meta-analysis for diabetes remission at the 12-mo and 2-y endpoints.
(11%–26% calories from carbohydrate) and very low carbohydrate The summary estimate at the 12-mo endpoint obtained from the
diets (10% calories from carbohydrate) may result in remission of fixed-effects meta-analysis was far weaker than that of the
diabetes, defined as an HbA1c level of <6.5% regardless of medication random-effects model, and thus, the magnitude of the effect of calorie
use, by 32 more per 100 patients (95% CI: 17 more, 47 more; n ¼ 8 restriction on diabetes remission at the 12-mo endpoint should be
trials) at 3–9 mo follow-up; however, they found a nonsignificant in- interpreted with caution. In addition, only 3 trials defined diabetes
crease in remission rate when remission was defined as an HbA1c level remission based on the recent definition proposed by the American
of <6.5% without medication [13]. Their results indicated that low and Diabetes Association [20]. Thus, more research is needed to test the
very low carbohydrate diets were no longer effective for diabetes long-term effect of calorie restriction on diabetes remission based on a
remission at follow-up longer than 9 mo [13]. Our findings indicated recent definition. We included 28 trials in the analyses, of which only
that calorie-restricted diets were effective lifestyle interventions for 15 trials presented data for diabetes remission. We contacted the au-
diabetes remission, even at a 2-y follow-up. All trials that reported thors to obtain additional information on the number of remissions at
information on diabetes remission at 12-mo (9–15 mo) and 2-y each study arm, but only one study author provided the requested in-
follow-ups implemented an intensive lifestyle weight-loss interven- formation [46]. In addition, we used aggregated data for dose–response
tion program including a calorie-restricted diet. Thus, combining meta-analyses and for investigating the association between calorie
calorie-restricted diets with an exercise program and behavioral support intake and our primary and secondary outcomes. The use of aggregate
may be an effective lifestyle intervention strategy for diabetes remis- data to investigate dose–response associations is subject to aggregation
sion and improving glycemia. bias, and thus, the dose–response associations found in the present
Dose–response analyses suggested some evidence of a modest U- study should be interpreted with caution. Additionally, because base-
shaped effect on total and LDL cholesterol concentrations at 6 mo, in line HbA1c levels and the duration of diabetes are patient-level char-
ways that there was an upward curve at an energy intake of <1500 acteristics based on aggregate data, subgroup analyses according to
kcal/d. This may be due to the fact that most trials included in the baseline HbA1c levels and the duration of diabetes are also subject to
analyses of the 6-mo follow-up implemented a moderate-to-low car- aggregation bias. Therefore, an individual participant data
bohydrate diet. Our previous dose–response meta-analysis of meta-analysis incorporating data from a larger number of trials can

879
A. Jayedi et al. The American Journal of Clinical Nutrition 117 (2023) 870–882

provide more robust evidence. Finally, although we evaluated potential analysis of more than 1 million participants, JAMA Netw. Open 2 (4) (2019),
adverse events, only a few studies have reported information on e192696, https://doi.org/10.1001/jamanetworkopen.2019.2696.
[4] J.L. Harding, M.E. Pavkov, D.J. Magliano, J.E. Shaw, E.W. Gregg, Global
adverse events, and thus, more research is needed to assess the safety of trends in diabetes complications: a review of current evidence, Diabetologia 62
these diets. (1) (2019) 3–16, https://doi.org/10.1007/s00125-018-4711-2.
In summary, our findings presented evidence of low-to-moderate [5] I.B. Hirsch, Glycemic variability and diabetes complications: does it matter? Of
course it does, Diabetes Care 38 (8) (2015) 1610–1614, https://doi.org/10.2337/
certainty that calorie-restricted diets, regardless of macronutrient dc14-2898.
composition, can increase the remission rate in patients with type 2 [6] American Diabetes Association Professional Practice Committee, 5, Facilitating
diabetes at 6-mo, 12-mo, and 2-y endpoints, especially when coupled behavior change and well-being to improve health outcomes: standards of
with an intensive lifestyle modification program. Dose–response meta- medical care in diabetes—2022, Diabetes Care 45 (Suppl_1) (2022) S60–S82,
https://doi.org/10.2337/dc22-s005.
analyses at the 6-mo endpoint indicated that the levels of most car- [7] S. Juray, K.V. Axen, S.E. Trasino, Remission of type 2 diabetes with very low-
diometabolic risk factors reduced linearly, larger than the MCID calorie diets-a narrative review, Nutrients 13 (6) (2021) 2086, https://doi.org/
thresholds, with the decrease in calorie intake. A U-shaped effect was 10.3390/nu13062086.
[8] A. Jayedi, S. Zeraattalab-Motlagh, B. Jabbarzadeh, Y. Hosseini, A.T. Jibril,
found for total and LDL cholesterol concentrations at the 6 mo. At the
H. Shahinfar, et al., Dose-dependent effect of carbohydrate restriction for type 2
12-mo endpoint, the effects of calorie restriction remained large and diabetes management: a systematic review and dose-response meta-analysis of
clinically meaningful for HbA1c and FPG, diminished for body randomized controlled trials, Am. J. Clin. Nutr. 116 (1) (2022) 40–56, https://
weight, and disappeared for SBP and serum triglyceride. Calorie- doi.org/10.1093/ajcn/nqac066.
[9] M. Neuenschwander, G. Hoffmann, L. Schwingshackl, S. Schlesinger, Impact
restricted diets may lead to an increase in LDL cholesterol concen- of different dietary approaches on blood lipid control in patients with type 2
tration at the 12-mo endpoint. More large trials are needed to assess the diabetes mellitus: a systematic review and network meta-analysis, Eur. J.
long-term effects of calorie restriction on remission of type 2 diabetes, Epidemiol. 34 (9) (2019) 837–852, https://doi.org/10.1007/s10654-019-00534-
particularly considering the recent definition proposed for diabetes 1.
[10] L. Schwingshackl, A. Chaimani, G. Hoffmann, C. Schwedhelm, H. Boeing,
remission A network meta-analysis on the comparative efficacy of different dietary
approaches on glycaemic control in patients with type 2 diabetes mellitus, Eur.
Acknowledgements J. Epidemiol. 33 (2) (2018) 157–170, https://doi.org/10.1007/s10654-017-
0352-x.
[11] S.J. Hallberg, V.M. Gershuni, T.L. Hazbun, S.J. Athinarayanan, Reversing type
The authors’ responsibilities were as follows – AJ, SSB: conceived 2 diabetes: a narrative review of the evidence, Nutrients 11 (4) (2019) 766,
and designed the study, conducted systematic search, screened articles, https://doi.org/10.3390/nu11040766.
[12] E. Jacob, A. Avery, Energy-restricted interventions are effective for the
read the full texts for eligibility, and performed risk of bias and remission of newly diagnosed type 2 diabetes: a systematic review of the
GRADE assessments; SZM, HS: extracted data from original studies; evidence base, Obes. Sci. Pract. 7 (5) (2021) 606–618, https://doi.org/10.1002/
AJ: performed the analyses; AJ, SZM, HS: contributed to the inter- osp4.504.
pretation of the results and wrote the first draft of the manuscript; EWG, [13] J.Z. Goldenberg, A. Day, G.D. Brinkworth, J. Sato, S. Yamada, T. J€onsson, et
al., Efficacy and safety of low and very low carbohydrate diets for type 2
SSB: contributed to the interpretation of the results and critically diabetes remission: systematic review and meta-analysis of published and
revised the manuscript; SSB: is the guarantor; and all authors: take full unpublished randomized trial data, BMJ 372 (2021) m4743, https://doi.org/
responsibility for the analyses and interpretation of the report, have 10.1136/bmj.m4743.
[14] J.P. Higgins, J. Thomas, J. Chandler, M. Cumpston, T. Li, M.J. Page, et al.,
accessed and verified the data, and read and approved the final
Cochrane handbook for systematic reviews of interventions, John Wiley &
manuscript. AJ, SZM, HSF, EWG, and SS-B report no conflicts of Sons, 2019.
interest. [15] H. Schunemann, GRADE handbook for grading quality of evidence and
strength of recommendation [Internet]. Version 3.2, GRADE Working Group,
2008. Available from: http://www.cc-ims.net/gradepro.
Data Availability [16] [Internet], Dose-dependent effects of calorie restriction for type 2 diabetes
Data described in the manuscript, code book, and analytic code will management: a protocol for a systematic review and dose-response meta-
be made available upon reasonable request to the corresponding author. analysis of randomised controlled trials, Prospero, 2022. Available from: http
s://www.crd.york.ac.uk/prospero/display_record.php?ID¼CRD42022300875.
[17] C. Lefebvre, G.J. Me, Chapter 6: Searching for studies, in: J.P.T. Higgins,
Funding S. Green (Eds.), Cochrane handbook for systematic reviews of interventions.
Version 5.1.0, The Cochrane Collaboration, 2011. Available from: htt
The authors reported no funding received for this study.
p://www.cochrane-handbook.org.
[18] M.D. Jensen, D.H. Ryan, C.M. Apovian, J.D. Ard, A.G. Comuzzie,
Appendix A. Supplementary data K.A. Donato, et al., 2013 AHA/ACC/TOS guideline for the management of
overweight and obesity in adults: a report of the American College of
Supplementary data to this article can be found online at https:// Cardiology/American Heart Association Task Force on Practice Guidelines and
doi.org/10.1016/j.ajcnut.2023.03.018. the Obesity Society, Circulation 129 (25_Suppl_2) (2014) S102–S138, https://
doi.org/10.1161/01.cir.0000437739.71477.ee.
[19] E. Stenberg, L.F. Dos Reis Falc~ao, M. O’Kane, R. Liem, D.J. Pournaras,
References P. Salminen, et al., Guidelines for perioperative care in bariatric surgery:
Enhanced Recovery After Surgery (ERAS) Society recommendations: a 2021
[1] X. Lin, Y. Xu, X. Pan, J. Xu, Y. Ding, X. Sun, et al., Global, regional, and update, World J. Surg. 46 (4) (2022) 729–751, https://doi.org/10.1007/s00268-
national burden and trend of diabetes in 195 countries and territories: an 021-06394-9.
analysis from 1990 to 2025, Sci. Rep. 10 (1) (2020) 14790, https://doi.org/ [20] M.C. Riddle, W.T. Cefalu, P.H. Evans, H.C. Gerstein, M.A. Nauck, W.K. Oh,
10.1038/s41598-020-71908-9. et al., Consensus report: definition and interpretation of remission in type 2
[2] N.H. Cho, J.E. Shaw, S. Karuranga, Y.d. Huang, J.D. da Rocha Fernandes, diabetes, J. Clin. Endocrinol. Metab. 107 (1) (2022) 1–9, https://doi.org/
A.W. Ohlrogge, et al., IDF Diabetes Atlas: global estimates of diabetes 10.1210/clinem/dgab585.
prevalence for 2017 and projections for 2045, Diabetes Res. Clin. Pract. 138 [21] D. Nagi, C. Hambling, R. Taylor, Remission of type 2 diabetes: a position
(2018) 271–281, https://doi.org/10.1016/j.diabres.2018.02.023. statement from the Association of British Clinical Diabetologists (ABCD) and
[3] J.J. Yang, D. Yu, W. Wen, E. Saito, S. Rahman, X.-O. Shu, et al., Association the Primary Care Diabetes Society (PCDS), Br. J. Diabetes 19 (1) (2019)
of diabetes with all-cause and cause-specific mortality in Asia: a pooled 73–76, https://doi.org/10.15277/bjd.2019.221.

880
A. Jayedi et al. The American Journal of Clinical Nutrition 117 (2023) 870–882

[22] S.W. Raudenbush, Analyzing effect sizes: random-effects models, in: [43] C. Durrer, S. McKelvey, J. Singer, A.M. Batterham, J.D. Johnson,
H. Cooper, L.V. Hedges, J.C. Valentine (Eds.), The handbook of research K. Gudmundson, et al., A randomized controlled trial of pharmacist-led
synthesis and meta-analysis, Russell Sage Foundation, 2009, pp. 295–316. therapeutic carbohydrate and energy restriction in type 2 diabetes, Nat.
[23] M. Egger, G.D. Smith, Misleading meta-analysis, BMJ 310 (6982) (1995) Commun. 12 (1) (2021) 5367, https://doi.org/10.1038/s41467-021-25667-4.
752–754, https://doi.org/10.1136/bmj.310.6982.752. [44] P.A. Dyson, S. Beatty, D.R. Matthews, A low-carbohydrate diet is more
[24] L. Smeeth, A. Haines, S. Ebrahim, Numbers needed to treat derived from meta- effective in reducing body weight than healthy eating in both diabetic and non-
analyses—sometimes informative, usually misleading, BMJ 318 (7197) (1999) diabetic subjects, Diabet. Med. 24 (12) (2007) 1430–1435, https://doi.org/
1548–1551, https://doi.org/10.1136/bmj.318.7197.1548. 10.1111/j.1464-5491.2007.02290.x.
[25] R.G. Newcombe, R. Bender, Implementing GRADE: calculating the risk [45] E.W. Gregg, H. Chen, L.E. Wagenknecht, J.M. Clark, L.M. Delahanty,
difference from the baseline risk and the relative risk, BMJ Evid. Based Med. J. Bantle, et al., Association of an intensive lifestyle intervention with remission
19 (1) (2014) 6–8, https://doi.org/10.1136/2Feb-2013-101340. of type 2 diabetes, JAMA 308 (23) (2012) 2489–2496, https://doi.org/10.1001/
[26] A. Crippa, N. Orsini, Dose-response meta-analysis of differences in means, jama.2012.67929.
BMC Med. Res. Methodol. 16 (1) (2016) 91, https://doi.org/10.1186/s12874- [46] G.S. Gulsin, D.J. Swarbrick, L. Athithan, E.M. Brady, J. Henson, E. Baldry, et
016-0189-0. al., Effects of low-energy diet or exercise on cardiovascular function in
[27] American Diabetes Association Professional Practice Committee, 8, Obesity working-age adults with type 2 diabetes: a prospective, randomized, open-label,
and weight management for the prevention and treatment of type 2 diabetes: blinded end point trial, Diabetes Care 43 (6) (2020) 1300–1310, https://doi.org/
standards of medical care in diabetes—2022, Diabetes Care 45 (Suppl_1) 10.2337/dc20-0129.
(2022) S113–S124, https://doi.org/10.2337/dc22-s008. [47] T. J€onsson, Y. Granfeldt, B. Ahren, U.C. Branell, G. Pålsson, A. Hansson, et
[28] T.A. Furukawa, C. Barbui, A. Cipriani, P. Brambilla, N. Watanabe, Imputing al., Beneficial effects of a Paleolithic diet on cardiovascular risk factors in type
missing standard deviations in meta-analyses can provide accurate results, 2 diabetes: a randomized cross-over pilot study, Cardiovasc. Diabetol. 8 (1)
J. Clin. Epidemiol. 59 (1) (2006) 7–10, https://doi.org/10.1016/ (2009) 35, https://doi.org/10.1186/1475-2840-8-35.
j.jclinepi.2005.06.006. [48] M.E. Lean, W.S. Leslie, A.C. Barnes, N. Brosnahan, G. Thom, L. McCombie,
[29] D. Luo, X. Wan, J. Liu, T. Tong, Optimally estimating the sample mean from et al., Primary care-led weight management for remission of type 2 diabetes
the sample size, median, mid-range, and/or mid-quartile range, Stat. Methods (DiRECT): an open-label, cluster-randomised trial, Lancet 391 (10120) (2018)
Med. Res. 27 (6) (2018) 1785–1805, https://doi.org/10.1177/ 541–551, https://doi.org/10.1016/S0140-6736(17)33102-1.
0962280216669183. [49] J. Leroux-Stewart, B. Elisha, S. Tagougui, C. Suppere, S. Bernard,
[30] X. Wan, W. Wang, J. Liu, T. Tong, Estimating the sample mean and standard H. Mircescu, et al., Effect of caloric restriction with or without physical activity
deviation from the sample size, median, range and/or interquartile range, BMC on body composition and epicardial fat in type 2 diabetic patients: a pilot
Med. Res. Methodol. 14 (1) (2014) 135, https://doi.org/10.1186/1471-2288-14- randomized controlled trial, Nutr. Metab. Cardiovasc. Dis. 31 (3) (2021)
135. 921–929, https://doi.org/10.1016/j.numecd.2020.11.005.
[31] F.E. Harrell, Regression modeling strategies: with applications to linear models, [50] W.F. Mollentze, G. Joubert, A. Prins, S. van der Linde, G.M. Marx, K.G. Tsie,
logistic and ordinal regression, and survival analysis, Springer, 2015. The safety and efficacy of a low-energy diet to induce weight loss, improve
[32] S. Schandelmaier, M. Briel, R. Varadhan, C.H. Schmid, N. Devasenapathy, metabolic health, and induce diabetes remission in insulin-treated obese men
R.A. Hayward, et al., Development of the Instrument to assess the Credibility with type 2 diabetes: a pilot RCT, Int. J. Diabetes Dev. Ctries. 39 (4) (2019)
of Effect Modification Analyses (ICEMAN) in randomized controlled trials and 618–625, https://doi.org/10.1007/s13410-019-00734-1.
meta-analyses, CMAJ 192 (32) (2020) E901–E906, https://doi.org/10.1503/ [51] E. Morris, P. Aveyard, P. Dyson, M. Noreik, C. Bailey, R. Fox, et al., A food-
cmaj.200077. based, low-energy, low-carbohydrate diet for people with type 2 diabetes in
[33] R.D. Feinman, W.K. Pogozelski, A. Astrup, R.K. Bernstein, E.J. Fine, primary care: a randomized controlled feasibility trial, Diabetes Obes. Metab.
E.C. Westman, et al., Dietary carbohydrate restriction as the first approach in 22 (4) (2020) 512–520, https://doi.org/10.1111/dom.13915.
diabetes management: critical review and evidence base, Nutrition 31 (1) [52] R. Paisey, P. Harvey, S. Rice, I. Belka, L. Bower, M. Dunn, et al., Short-term
(2015) 1–13, https://doi.org/10.1016/j.nut.2014.06.011. results of an open trial of very low calorie diet or intensive conventional diet in
[34] J. Hartung, An alternative method for meta-analysis, Biom. J. 41 (8) (1999) type 2 diabetes, Pract, Diabetes Int 12 (6) (1995) 263–267, https://doi.org/
901–916. 10.1002/pdi.1960120611.
[35] J. Hartung, G. Knapp, On tests of the overall treatment effect in meta-analysis [53] P. Ruggenenti, M. Abbate, B. Ruggiero, S. Rota, M. Trillini, C. Aparicio, et al.,
with normally distributed responses, Stat. Med. 20 (12) (2001) 1771–1782, Renal and systemic effects of calorie restriction in patients with type 2 diabetes
https://doi.org/10.1002/sim.791. with abdominal obesity: a randomized controlled trial, Diabetes 66 (1) (2017)
[36] J. IntHout, J.P. Ioannidis, G.F. Borm, The Hartung-Knapp-Sidik-Jonkman 75–86, https://doi.org/10.2337/db16-0607.
method for random effects meta-analysis is straightforward and considerably [54] L.R. Saslow, A.E. Mason, S. Kim, V. Goldman, R. Ploutz-Snyder,
outperforms the standard DerSimonian-Laird method, BMC Med. Res. H. Bayandorian, et al., An online intervention comparing a very low-
Methodol. 14 (1) (2014) 25, https://doi.org/10.1186/1471-2288-14-25. carbohydrate ketogenic diet and lifestyle recommendations versus a plate
[37] M. Egger, G.D. Smith, M. Schneider, C. Minder, Bias in meta-analysis detected method diet in overweight individuals with type 2 diabetes: a randomized
by a simple, graphical test, BMJ 315 (7109) (1997) 629–634, https://doi.org/ controlled trial, J. Med. Internet Res. 19 (2) (2017) e36, https://doi.org/10.2196/
10.1136/bmj.315.7109.629. jmir.5806.
[38] G.H. Guyatt, A.D. Oxman, G.E. Vist, R. Kunz, Y. Falck-Ytter, P. Alonso- [55] J. Sato, A. Kanazawa, S. Makita, C. Hatae, K. Komiya, T. Shimizu, et al.,
Coello, et al., GRADE: an emerging consensus on rating quality of evidence A randomized controlled trial of 130 g/day low-carbohydrate diet in type 2
and strength of recommendations, BMJ 336 (7650) (2008) 924–926, https:// diabetes with poor glycemic control, Clin. Nutr. 36 (4) (2017) 992–1000,
doi.org/10.1136/bmj.39489.470347.AD. https://doi.org/10.1016/j.clnu.2016.07.003.
[39] A. Brown, A. Dornhorst, B. McGowan, O. Omar, A.R. Leeds, S. Taheri, et al., [56] L. Stern, N. Iqbal, P. Seshadri, K.L. Chicano, D.A. Daily, J. McGrory, et al.,
Low-energy total diet replacement intervention in patients with type 2 diabetes The effects of low-carbohydrate versus conventional weight loss diets in
mellitus and obesity treated with insulin: a randomized trial, BMJ Open severely obese adults: one-year follow-up of a randomized trial, Ann. Intern.
Diabetes Res. Care. 8 (1) (2020), e001012, https://doi.org/10.1136/bmjdrc- Med. 140 (10) (2004) 778–785, https://doi.org/10.7326/0003-4819-140-10-
2019-001012. 200405180-00007.
[40] K.M. Choi, K.A. Han, H.J. Ahn, S.Y. Lee, S.Y. Hwang, B.H. Kim, et al., The [57] S. Taheri, H. Zaghloul, O. Chagoury, S. Elhadad, S.H. Ahmed, N. El Khatib, et
effects of caloric restriction on Fetuin-A and cardiovascular risk factors in rats al., Effect of intensive lifestyle intervention on bodyweight and glycaemia in
and humans: a randomized controlled trial, Clin. Endocrinol. 79 (3) (2013) early type 2 diabetes (DIADEM-I): an open-label, parallel-group, randomised
356–363, https://doi.org/10.1111/cen.12076. controlled trial, Lancet Diabetes Endocrinol 8 (6) (2020) 477–489, https://
[41] K.J. Coppell, M. Kataoka, S.M. Williams, A.W. Chisholm, S.M. Vorgers, doi.org/10.1016/S2213-8587(20)30117-0.
J.I. Mann, Nutritional intervention in patients with type 2 diabetes who are [58] M. Umphonsathien, P. Rattanasian, S. Lokattachariya, W. Suansawang,
hyperglycaemic despite optimised drug treatment—Lifestyle Over and Above K. Boonyasuppayakorn, W. Khovidhunkit, Effects of intermittent very-low
Drugs in Diabetes (LOADD) study: randomised controlled trial, BMJ 341 calorie diet on glycemic control and cardiovascular risk factors in obese patients
(2010) c3337, https://doi.org/10.1136/bmj.c3337. with type 2 diabetes mellitus: a randomized controlled trial, J. Diabetes
[42] P.M. Dodson, P.J. Pacy, P. Bal, A.J. Kubicki, R.F. Fletcher, K.G. Taylor, Investig. 13 (1) (2022) 156–166, https://doi.org/10.1111/jdi.13619.
A controlled trial of a high fibre, low fat and low sodium diet for mild [59] K.V. Williams, M.L. Mullen, D.E. Kelley, R.R. Wing, The effect of short periods
hypertension in type 2 (non-insulin-dependent) diabetic patients, Diabetologia of caloric restriction on weight loss and glycemic control in type 2 diabetes,
27 (5) (1984) 522–526, https://doi.org/10.1007/BF00290388. Diabetes Care 21 (1) (1998) 2–8, https://doi.org/10.2337/diacare.21.1.2.

881
A. Jayedi et al. The American Journal of Clinical Nutrition 117 (2023) 870–882

[60] R.R. Wing, E. Blair, M. Marcus, L.H. Epstein, J. Harvey, Year-long weight [67] G.H. Guyatt, A.D. Oxman, R. Kunz, J. Woodcock, J. Brozek, M. Helfand, et
loss treatment for obese patients with type II diabetes: does including an al., GRADE guidelines: 7. Rating the quality of evidence—inconsistency,
intermittent very-low-calorie diet improve outcome? Am. J. Med. 97 (4) (1994) J. Clin. Epidemiol. 64 (12) (2011) 1294–1302, https://doi.org/10.1016/
354–362, https://doi.org/10.1016/0002-9343(94)90302-6. j.jclinepi.2011.03.017.
[61] R.R. Wing, M.D. Marcus, R. Salata, L.H. Epstein, S. Miaskiewicz, E.H. Blair, [68] L.A. Becker, Effect size (ES) [Internet], 2000. Available from: http://
Effects of a very-low-calorie diet on long-term glycemic control in obese type 2 www.bwgriffin.com/gsu/courses/edur9131/content/EffectSizeBecker.pdf.
diabetic subjects, Arch. Intern. Med. 151 (7) (1991) 1334–1340, https://doi.org/ [69] G. Guyatt, A.D. Oxman, R. Kunz, J. Brozek, P. Alonso-Coello, D. Rind, et al.,
10.1001/archinte.1991.00400070100012. Corrigendum to GRADE Guidelines 6. Rating the quality of evidence-
[62] A. Celli, Y. Barnouin, B. Jiang, D. Blevins, G. Colleluori, S. Mediwala, et al., imprecision, J. Clin. Epidemiol. 137 (2021) 265, https://doi.org/10.1016/
Lifestyle intervention strategy to treat diabetes in older adults: a randomized j.jclinepi.2021.04.014.
controlled trial, Diabetes Care 45 (9) (2022) 1943–1952, https://doi.org/ [70] M.J. Franz, J.L. Boucher, S. Rutten-Ramos, J.J. VanWormer, Lifestyle weight-
10.2337/dc22-0338. loss intervention outcomes in overweight and obese adults with type 2 diabetes:
[63] K.S. Dorans, L.A. Bazzano, L. Qi, H. He, J. Chen, L.J. Appel, et al., Effects of a systematic review and meta-analysis of randomized clinical trials, J. Acad.
a low-carbohydrate dietary intervention on hemoglobin A1c: a randomized Nutr. Diet. 115 (9) (2015) 1447–1463, https://doi.org/10.1016/
clinical trial, JAMA Netw. Open. 5 (10) (2022), e2238645, https://doi.org/ j.jand.2015.02.031.
10.1001/jamanetworkopen.2022.38645. [71] W.M. Admiraal, F. Celik, V.E. Gerdes, R.M. Dallal, J.B. Hoekstra,
[64] M.Y. Johansen, C.S. MacDonald, K.B. Hansen, K. Karstoft, R. Christensen, F. Holleman, Ethnic differences in weight loss and diabetes remission after
M. Pedersen, et al., Effect of an intensive lifestyle intervention on bariatric surgery: a meta-analysis, Diabetes Care 35 (9) (2012) 1951–1958,
glycemic control in patients with type 2 diabetes: a randomized https://doi.org/10.2337/dc12-0260.
clinical trial, JAMA 318 (7) (2017) 637–646, https://doi.org/10.1001/ [72] S. Kodama, K. Fujihara, C. Horikawa, M. Harada, H. Ishiguro, M. Kaneko, et
jama.2017.10169. al., Network meta-analysis of the relative efficacy of bariatric surgeries for
[65] M. Ried-Larsen, M.Y. Johansen, C.S. MacDonald, K.B. Hansen, diabetes remission, Obes. Rev. 19 (12) (2018) 1621–1629, https://doi.org/
R. Christensen, A.S. Wedell-Neergaard, et al., Type 2 diabetes remission 1 year 10.1111/obr.12751.
after an intensive lifestyle intervention: a secondary analysis of a randomized [73] S.H. Chang, C.R. Stoll, J. Song, J.E. Varela, C.J. Eagon, G.A. Colditz, The
clinical trial, Diabetes Obes. Metab. 21 (10) (2019) 2257–2266, https://doi.org/ effectiveness and risks of bariatric surgery: an updated systematic review and
10.1111/dom.13802. meta-analysis, 2003–2012, JAMA Surg 149 (3) (2014) 275–287, https://
[66] M.E. Lean, W.S. Leslie, A.C. Barnes, N. Brosnahan, G. Thom, L. McCombie, doi.org/10.1001/jamasurg.2013.3654.
et al., Durability of a primary care-led weight-management intervention for [74] J. Picot, J. Jones, J.L. Colquitt, E. Gospodarevskaya, E. Loveman, L. Baxter, et
remission of type 2 diabetes: 2-year results of the DiRECT open-label, cluster- al., The clinical effectiveness and cost-effectiveness of bariatric (weight loss)
randomised trial, Lancet Diabetes Endocrinol 7 (5) (2019) 344–355, https:// surgery for obesity: a systematic review and economic evaluation, Health
doi.org/10.1016/S2213-8587(19)30068-3. Technol. Assess. 13 (41) (2009) 1–190, https://doi.org/10.3310/hta13410.

882