Introduction

Cardiovascular diseases (CVD) are a leading cause of morbidity and mortality worldwide, affecting individuals in high-income as well as low-and middle-income countries [1]. The main contributors to the major cardiovascular diseases (ischemic heart disease and stroke) include overweight or obesity, high blood pressure, high blood glucose, and dyslipidaemia [1, 2]. Behavioural modification including dietary intake and physical activity is an important approach to mitigate cardiometabolic risk factors such as overweight or obesity, high blood pressure, elevated cholesterol levels and blood glucose [3, 4]. Weight control through energy restriction has been shown to improve cardiovascular risks including insulin resistance, blood glucose, and blood pressure [5].

Intermittent fasting, which includes a range of approaches to achieve overall energy restriction, has emerged as an appealing alternative to continuous energy restriction (CER) for managing obesity and its related comorbidities due to its relative ease of maintaining long-term adherence [6, 7]. Intermittent fasting refers to dietary patterns that involve cycling between periods of eating and periods of fasting [8]. This creates periods of energy deficit, and metabolic change which can potentially leading to health benefits, including weight loss, improved insulin sensitivity, and better overall metabolic health [6, 9].

Among the many methods of intermittent fasting, some of the most adopted include alternate-day fasting (ADF), modified alternate day fasting (mADF), periodic fasting (PF), and time-restricted eating (TRE) [6, 10]. ADF is cyclic eating approach involves a 24-hour period of complete fasting (no calorie intake) followed by a 24-hour period of normal eating [8, 11]. The mADF is like ADF but allows for some calorie intake on fasting days (25% or less intake of energy) [8, 11]. PF is a cyclical weekly eating pattern with fasting for one or two days per week (consumption of 25% or less of required calories or restricting calorie intake to around 500–600 kcal/day) and then eating normally for the remaining five or six days a week. The 5:2 diet is a popular form of PF [8, 12].TRE involves complete fasting (no calorie intake) for at least 12 h per day and eating freely the rest of the time [8, 12]. TRE involves limiting the daily eating window to a specific period, for example, an individual might eat all meals within an 8-hour window (e.g., 12:00 pm to 8:00 pm) and fast for the remaining 16 h each day (16/8 method). The most common TRE methods are the 16/8 and 14/10 method [8, 12].

Previous pairwise meta-analysis studies have shown some promise for intermittent fasting in reducing risk factors for cardiovascular disease. However, the results are not consistent [8, 11, 13]. Some meta-analyses suggest that intermittent fasting is more effective than usual eating pattern in reducing weight and waist circumference [11,12,13,14]. However, others showed no significant difference between intermittent fasting and CER for these measures [15, 16]. Regarding fat-free mass, there is no clear conclusion on whether intermittent fasting leads to undesirable loss of muscle mass. Some studies found no effect [13, 16], while others showed an increase [17] or decrease [15] compared to usual diet. Findings on blood pressure are also inconsistent. Some meta-analyses suggest intermittent fasting reduces systolic blood pressure (SBP) and diastolic blood pressure (DBP) compared to usual eating [11, 12], while others found no significant difference [13, 14]. Similarly, some studies showed intermittent fasting reduced fasting blood sugar [11, 14] and low-density lipoprotein (LDL) cholesterol [18], but others found no significant difference compared to usual eating on fasting plasma glucose (FPG) [13] and LDL reduction [11, 12, 14]. The inconsistencies of results across the previous meta-analyses could be due to differences in terms of the population, the intervention duration (some included short duration studies) [12,13,14] and number of studies included [12,13,14]. Further, some conducted the analyses by combining all intermittent fasting methods together [11, 12, 16].

Since conventional pairwise meta-analysis is often limited by comparing two intervention at a time and cannot incorporate indirect evidence, there remains considerable uncertainty about which intermittent fasting methods are the most effective for improving cardiovascular health [19]. An alternative approach is network meta-analysis (NMA) which allows statistical comparison of three or more interventions that have not been directly compared in randomised controlled trials (RCTs) (19). Furthermore, NMA has the potential to enhance the precision of effect estimates derived from RCTs and traditional pairwise meta-analyses by integrating both direct and indirect evidence (19). This method offers a more thorough understanding of relative effectiveness and allows for the ranking of intermittent fasting methods, which is not possible with conventional pairwise meta-analysis. The aim of this systematic review and network meta-analysis was to assess the relative effectiveness of different intermittent fasting methods in improving key cardiovascular risk factors, including body weight, waist circumference, fat free mass, elevated blood pressure, FPG, low density lipoprotein cholesterol.

Methods

The protocol was registered at PROSPERO (CRD42023475279), and the NMA was reported in accordance with the Preferred Reporting Items for Systematic Review and Meta-analysis for Network Meta-Analyses (PRISMA-NMA) guidelines [20] (see supplementary material (S1).

Search Strategy

We searched four databases: Medline, Embase, Cochrane Library, and Global Health—from inception to November 09, 2023, and the search was updated up to December 11, 2024. We also performed manual searches of references from relevant reviews and eligible studies. The key search terms include a combination of “intermittent fasting” or “alternate day fasting” or “periodic fasting” or “time restricted eating /feeding” or “intermittent energy restriction” and body weight or waist circumference or fat-free mass or blood pressure or SBP or DBP or LDL or fasting plasma/blood glucose. The full search strategy is presented in the supplementary material (S2). The search was limited to RCTs, published in English. There was no limitation on publication date or location. Search results were exported to Covidence for duplicate removal, screening and data extraction.

Eligibility Criteria

We developed the eligibility criteria based on the PICOS framework (Participants, Interventions, Comparisons, Outcomes, and Study design). All inclusion and exclusion criteria are summarised in Table 1. This systematic review and network meta-analysis included only RCTs.

Table 1 PICOS criteria for inclusion and exclusion of studies
Full size table

Screening and Data Extraction

Three independent reviewers conducted the title and abstract screening: KTK screened all, TKT screened 69%, and YMM screened 31%. KTK performed the full-text review, with TKT double-checking 20%, applying the inclusion and exclusion criteria. Any discrepancies between the reviewers were resolved through discussion and consensus. The step-by-step procedure of identifying, screening, and incorporating or excluding studies presented using the PRISMA 2020 flow diagram (Fig. 1). Data were extracted using a pretested data abstraction form. The following information was extracted from each eligible study: first author, publication year, country, the intervention duration, sample size, participant characteristics (sex, age, BMI) and outcomes measured, intervention or intermittent fasting type (s), control group diet, number of participants in each group (treatment and control group). If intermittent fasting outcomes were reported at multiple time points, we extracted data from the last reported time point or the end of the intervention.

For studies reporting pre- and post-intervention measures, we calculated mean differences and standard deviations using Cochrane Handbook methods [21]. Missing standard deviations were estimated from standard errors or confidence intervals. For studies that reported only medians and interquartile ranges, means were estimated using the Wan method [22]. In cases the data were only available in figures, numerical data was obtained using Plot Digitiser (https://plotdigitizer.com/app).

Risk of Bias Assessment

We assessed the risk of bias using the Cochrane Collaboration’s Risk of Bias 2 (Rob 2) tool for RCTs [23]. This tool comprises five bias components: bias in the randomization process, bias resulting from deviations in intended interventions, bias due to missing outcome data, bias in the measurement of outcomes, and bias in the selection of reported results. Each study was assessed and categorised according to its risk of bias into three levels (low risk of bias, some concerns, or high risk of bias), for each domain evaluated. A study was deemed to have a low overall risk of bias if all domains were rated as having a low risk of bias. Conversely, a study was considered to have a high risk of bias if at least one domain is rated as high risk, or if three and more domains were categorised as having ‘some concerns’. A study would fall into the ‘some concerns’ category overall if one or two of the domains are rated as having some concerns, but none were classified as high risk of bias [23].

Grading the Certainty of Evidence

We assessed the certainty of the evidence using Grading of Recommendation Assessment, Development, and Evaluation (GRADE) approach [24]. We classified the certainty of evidence as high, moderate, low, or very low. RCTs initially receive a high grade; however, this grade may be downgraded due to the following specific criteria: the presence of risk of bias (weight assigned to study as assessed by the RoB2 tool); inconsistency (significant unexplained variation among study results, indicated by I2), indirectness (limitations in the generalizability of the results); imprecision (wide 95% confidence intervals for effect estimates or crossing a null value); incoherence (differences between direct and indirect estimates that contribute to a network estimate); and publication bias (significant evidence of small-study effects) [24,25,26].

Statistical Analysis

We performed a random-effects network meta-analysis using a frequentist framework to the compare the effectiveness of different intermittent fasting methods on cardiovascular disease risks. We chose the frequentist approach over a Bayesian framework for its computational efficiency and straightforward implementation using standard statistical software. Additionally, the frequentist method provides robust and interpretable estimates without requiring prior distributions, which were not available for all comparisons in our network. We reported outcomes as mean differences (MD) with their 95% confidence intervals (CI). We created the network geometry diagrams to explore networks of intervention comparisons. The size of the nodes, representing each intervention, reflects total number of participants while the thickness of the lines connecting any two nodes illustrates the number of intervention comparisons. The incoherence assumption was checked by using a statistical test (network node-splitting method). In a closed-loop network, the node-splitting method was used to test incoherence between direct and indirect intervention comparisons [27]. We assessed incoherence by comparing the similarity of point estimates, checking for overlapping 95% confidence intervals, and ensuring non-significant p-values.

Transitivity was ensured by including only RCTs with comparable populations, interventions, and outcomes, and verifying that all included studies could be meaningfully compared based on shared treatment nodes. Multilevel meta-analysis was not conducted due to the primary focus on treatment comparisons across studies rather than variability within individual trials.

The relative rankings of all intermittent fasting methods for each outcome were determined by estimating ranking probabilities using ranking plots and the surface under the cumulative ranking curve (SUCRA) [28, 29].

Classification of Intermittent Fasting Methods as More and Less Effective Intervention

Using a new GRADE approach, we analysed NMA results by classifying intermittent fasting interventions from the most to least effective [30] for each outcome. The new GRDAE approach considers three factors: effect size from the NMA, evidence certainty, and SUCRA (ranking) values [30]. We first categorised evidence quality into high (moderate-to-high) and low (low-to-very-low) certainty. Within each category, intermittent fasting method were divided based on their effect on outcomes: (1) Most Effective: intermittent fasting method with the largest reduction in outcomes compared to the usual diet and superior to at least one moderately effective method; (2) Moderately Effective: intermittent fasting method better than the usual diet but not as effective as the most effective method; (3) Least Effective: intermittent fasting method similar to the usual diet, with confidence intervals crossing zero.

Sensitivity Analysis

We conducted sensitivity analysis to assess the stability or robustness of the pooled effect size by restricting the analysis to studies with medium to long-term intervention durations, some concern or low risk of bias, and studies that did not include participants with diabetes.

Data analysis was conducted using Stata version 18.0 (StataCorp, College Station, TX) [31], and all graphical displays were generated using the tools developed by Chaimani et al. and White [31, 32].

Results

Study Selection and Characteristics

A total of 5993 articles were identified, resulting in the inclusion of 56 studies [33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88] (Fig. 1). These 56 studies were conducted between 2013 and 2024 with a sample size ranging from 18 to 222 and totalling 3,965 participants. The studies were carried out in 16 different countries, including the USA (n = 17), Australia (n = 8), China (n = 6), and Norway (n = 4). The duration of interventions varied from 4 weeks to 104 weeks. Of the 56 studies, seven were three-arm while the rest were two-arm studies. The mean age of participants was 45.0 (SD 10.1) years (see details in Table 2).

Table 2 Characteristics of included stidies
Full size table
Fig. 1
figure 1

The PRISMA study selection flow diagram

Full size image

Risk of Bias

Out of the 56 RCTs, 21 (37.5%) studies were determined to have an overall high risk of bias while 12 (21.4%) studies were rated as overall low risk of bias (Fig. 2). The most common source of bias was related to the randomisation process (high risk, n = 13; some concern, n = 21) followed by bias due to missing outcome data (high risk, n = 5; some concern, n = 13). Detailed risk of bias assessment results is presented in Supplementary Fig. S1.

Fig. 2
figure 2

Risk of bias (Summary)

Full size image

Certainty of Evidence and Intervention Classifications

The GRADE assessment details for all outcomes are presented in supplementary Tables S1 A-G. Figure 3 and supplementary Table S2 presents the classification of all interventions for each outcome based on the new GRADE certainty of evidence framework.

Fig. 3
figure 3

The summary of results network meta-analysis of intermittent fasting regimes (mean difference with 95% CI) in comparison with usual diet for all outcomes along with ranking by new GRADE certainty of evidence framework. Note: mADF = modified alternate day fasting; ADF = alternate day fasting; CER = continuous energy restriction; PF = periodic fasting; TRE time restricted eating

Full size image

Comparative Effectiveness of Intermittent Fasting

Body Composition

Body Weight

A total of 52 studies reported weight change after intermittent fasting intervention with a total of 3241 participants. Most of the 52 comparisons were between CER vs. PF (n = 14) followed by TRE vs. usual diet (n = 14) (Fig. 4A and Supplementary Table S3). The inconsistency analysis revealed the absence of global inconsistency (Supplementary Fig. S2A) and local inconsistency (Supplementary Table S4). Compared to TRE, mADF (MD= -3.24 kg, 95% CI -5.29 to − 1.20, high certainty evidence) effective intervention in reducing weight.

Fig. 4
figure 4

Network plots of the direct comparisons between intermittent fasting interventions from head-to-head trials for the outcomes: (A) Weight; (B) Fat free mass; (C) Waist circumference; (D) LDL-cholesterol; (E) Systolic blood pressure; (F) Diastolic blood pressure; (G) Fasting plasma glucose. The sizes of nodes correspond to the number of participants randomized to the intermittent fasting methods and the width of line corresponds to the number of studies. Note: mADF = modified alternate day fasting; ADF = alternate day fasting; CER = continuous energy restriction; PF = periodic fasting; TRE time restricted eating

Full size image

When compared to usual diet mADF (MD=-5.18 kg; 95% CI: -7.04 to − 3.22, high certainty evidence), ADF (-4.27 kg; -6.12 to -2.42, high certainty evidence), PF (-3.82 kg; -5.44, -2.21, high certainty evidence), CER (-3.42 kg; -4.73 to -2.11, high certainty evidence), and TRE (-1.93 kg; -3.06, -0.81, moderate certainty evidence) significantly reduced body weight (Fig. 5A, Supplementary Table S1).

Fig. 5
figure 5figure 5

Intermittent fasting network meta-analysis results (mean difference with 95% CI) with corresponding GRADE certainty of evidence for: Weight in kg (A); Fat-free mass in kg (B); Waist circumference in cm (C); Low density lipoprotein-LDL in mg/dL (D); Systolic blood pressure -SBP in mmHg (E); Diastolic blood pressure - DBP in mmHg (F); Fasting plasma glucose– FPG in mg/dL (G). Values in bold indicate a statistically significant effect. Colour coding indicates the GRADE certainty of evidence: green = high certainty, blue = moderate certainty. Note: mADF = modified alternate day fasting; ADF = alternate day fasting; CER = continuous energy restriction; PF = periodic fasting; TRE time restricted eating

Full size image

Among the intermittent fasting methods with high or moderate certainty of evidence, compared to a usual diet, mADF was the most effective, whereas CER, TRE, ADF and PF were among the interventions with intermediate effectiveness in reducing body weight compared to usual diet (Fig. 3 and Supplementary Table S2, Supplementary Fig. S3A).

Fat Free Mass

Change in fat-free mass was reported in 32 studies with a total of 2045 participants. Most comparation were between PF vs. CER (n = 10), followed by TRE vs. usual diet (n = 6) (Fig. 4 and Supplementary Table S3). Both the global inconsistency test (Supplementary Fig. S2B) and the local inconsistency test supported the consistency of the direct and indirect estimates (Supplementary Table S4).

Compared to usual diet, TRE (MD= -0.82 kg; 95% CI: -1.46 to -0.17, moderate certainty evidence), PF (-0.80 kg; -1.58 to -0.02, high certainty of evidence) significantly reducing fat-free mass (Fig. 5B and Supplementary Table S1). Among intermittent fasting methods with high or moderate certainty of evidence, compared to a usual diet, TRE, and PF were the most effective for fat free mass reduction, whereas mADF and ADF was not better than usual diet (Fig. 3, Supplementary Table S2, Supplementary Fig. 3B).

Waist Circumference

Most of the 22 comparisons were between CER vs. PF (n = 7), CER VS mADF(n = 3) and TRE vs. usual diet (n = 3) (Fig. 4C and Supplementary Table S3). The global and local inconsistency test indicated no violation of the consistency assumption for direct and indirect estimates (Supplementary Fig. 2C and Supplementary Table S3).

Compared to usual diet with high certainty of evidence, mADF (MD= -3.55 cm; 95% CI: -5.66 to -1.45), CER (-1.78 cm; -3.23, -0.34), PF (-2.77 cm; -4.47, -1.07) and TRE (-3.00 cm; -4.50, -1.51) significantly reduced waist circumference (Fig. 5C and Supplementary Table S1). However, there were no statistically significant differences among the other comparisons (Fig. 5C). Among the intermittent fasting methods with high or moderate certainty of evidence, compared to a usual diet, mADF, CER, TRE, and PF were the most effective for fat free mass reduction, whereas ADF was probably among least effective (not better than usual diet) (Fig. 3, Supplementary Table S2, Supplementary Fig. S3C).

LDL Cholesterol

Change in LDL cholesterol levels were reported in 35 articles with a total of 2488 participants, and most comparisons were TRE vs. usual diet (n = 10) and CER vs. usual diet (n = 9) (Fig. 4D and Supplementary Table S3). With high certainty of the evidence, PF (MD= -6.80 mg/dL; 95% CI: -12.59, -1.00) was associated with a significant reduction in LDL level compared to usual diet; however, there were no significant differences among the other comparisons (Fig. 5D and Supplementary Table S1). Among the intermittent fasting methods with high or moderate certainty of evidence, compared to a usual diet, PF was among the most effective, while mADF, CER, TRE and ADF were not better than usual diet for LDL reduction (Fig. 3 and Supplementary Table S2, Supplementary Fig. S3D).

Blood Pressure

Systolic Blood Pressure (SBP)

SBP was reported in 27 studies, with a total of 1852 participants. Most of the 27 comparisons were CER vs. usual diet (n = 7) and TRE vs. usual diet (n = 6). With high certainty, mADF (-6.08 mmHg; -11.83 to -0.32) was more effective in reducing SBP compared to ADF. Compared to usual diet with high certainty of evidence, mADF (MD= -7.24 mmHg; 95%CI: -11.90 to -2.58), CER (-4.55 mmHg; -6.82 to -2.27), PF (-3.17 mmHg; -6.01 to -0.32) and TRE (-3.18 mmHg; -5.22 to -1.13) significantly reduced SBP (Fig. 5E and Supplementary Table S1). Among the intermittent fasting methods with high or moderate certainty of evidence, compared to a usual diet, mADF, CER, TRE, and PF were the most effective for SBP reduction, whereas ADF was not better than usual diet (Fig. 3, Supplementary Table S2, Supplementary Fig. S3E).

Diastolic Blood Pressure (DBP)

DBP was reported in 27 studies, with a total of 1861 participants, and most compared CER vs. usual diet (n = 7) and TRE vs. usual (n = 6). Compared to ADF, mADF (-5.19 mmHg; -9.61 to -0.78, high certainty evidence), TRE (-3.73 mmHg; -6.49 to -0.98, high certainty evidence), PF (-3.40 mmHg; -6.34 to -0.45, high certainty evidence) are more effective in reducing DBP. Compared to usual diet with high certainty of evidence, mADF (MD= -4.70 mmHg; 95%CI: -8.46 to -0.95), CER (2.66 mmHg; -4.11 to -1.22), PF (-2.90 mmHg; -4.79 to -1.02) and TRE (-3.24 mmHg; -4.69 to -1.79) significantly reduced DBP (Fig. 4F and Supplementary Table S1). Among the intermittent fasting methods with high or moderate certainty of evidence, compared to a usual diet, mADF, CER, TRE, and PF were the most effective for DBP reduction (Fig. 3, Supplementary Table S2, Supplementary Fig. S3F).

Fasting Plasma Glucose (FPG)

A total of 36 studies reported FPG change after intermittent fasting intervention involving a total of 2428 participants. Most comparison were TRE vs. usual diet (10) and PF vs. CER (n = 9) (Fig. 4 and Supplementary Table S3). The inconsistency examination revealed the absence of global inconsistency and local inconsistency (Supplementary Fig. S2G and Supplementary Table S4). With high certainty, TRE (-3.46 mg/dL; -6.34, -0.57) are more effective than CER in reducing FPG. Similarly, TRE (-3.61 mg/dL; -7.04, -0.19) with high certainty is effective in reducing FPG compared to PF. Relative to usual diet with high certainty of evidence, TRE (-3.74 mg/dL; -6.01, -1.46) significantly reduced FPG (Fig. 5G and Supplementary Table S1). Among the intermittent fasting methods with high or moderate certainty of evidence, compared to a usual diet, TRE was probably the most effective; mADF, PF, and ADF probably among least effective intermittent fasting methods (not better than usual diet) for FPG reduction (Fig. 3 and Supplementary Table S2, Supplementary Fig. S3G).

Sensitivity Analysis

Excluding Studies with Participants with Diabetes

Compared to the main analysis, the effects of intermittent fasting on body weight, FPG, SBP, and DBP remained similar in magnitude and direction. However, the previously significant effects of mADF and CER on waist circumference was no longer observed. Additionally, the positive effects of PF on waist circumference and fat-free mass were no longer statistically significant (Supplementary Fig. S4).

Excluding Studies with High-Risk of Bias

The size and direction of the network estimates for weight, FPG and SBP were consistent with the full analysis in this sensitivity analysis. However, the previously significant effects of PF on waist circumference and LDL, and the effect of mADF on DBP and TRE on fat free mass were no longer significant. Conversely, the effect of CER on fat free mass was statistically significant among this sub-set of higher quality studies (Supplementary Fig. S5).

Excluding Studies with Short Intervention Durations

The size and direction of the network estimates for weight, waist circumference and LDL cholesterol were in line with the full analysis. But the effects of mADF on SBP and DBP, and the effect of TRE on FPG and fat free mass were no longer significant. Conversely, the effect of CER on fat free mass and the effect of mADF on FPG were statistically significant (Supplementary Fig. S6).

Discussion

This systematic review and network meta-analysis synthesised the evidence on the effect of various intermittent fasting methods on cardiovascular disease risk factors using 56 randomised controlled trials conducted between 2013 and 2024. The findings indicated that different intermittent fasting modalities, when compared to a usual diet, significantly reduced body weight, fat-free mass, waist circumference, LDL levels, blood pressure, and FPG. The mADF was found to be the most effective intervention, with high or moderate certainty of the evidence, for the reduction of cardiovascular risk factors including SBP, DBP, weight, and waist circumference. Compared to a usual diet, time-restricted eating was the most effective intermittent fasting regimen for the reduction of fat-free mass and FPG. Moreover, PF was superior to a usual diet in reducing LDL levels. ADF did not show convincing evidence of superiority to a usual diet to reduce cardiovascular risks except for weight. When comparing each other, mADF is more effective than ADF in reducing SBP and DBP. Similarly, TRE and PF are more effective than ADF in reducing DBP. Additionally, TRE is more effective in reducing FPG compared to PF and CER.

The results of this network meta-analysis revealed a significant reduction in body weight across intermittent fasting methods compared to the usual diet, with ADF, mADF, PF, and TRE demonstrating notable effects compared to a usual diet. Likewise, compared to the usual diet, three intermittent fasting methods - mADF, PF, and TRE - significantly reduced waist circumference, a crucial marker of central adiposity. These results align with previous research [11,12,13,14] highlighting the weight management potential of intermittent fasting method. These findings reinforce the potential of intermittent fasting as a viable intervention for weight or waist circumference reduction.

One of the concerns surrounding intermittent fasting is its potential undesirable effect on fat-free mass loss which can impair physical function and cardiometabolic health [15, 89]. However, the evidence regarding this effect was not conclusive. Some studies reported no impact on fat free mass [13, 16], while others indicated an increase in fat-free mass [17], and yet other showed intermittent fasting significantly reduced fat-free mass [15]. Our study revealed a significant reduction in fat-free mass in two intermittent fasting methods (TRE and PF), but no significant reduction in other two intermittent fasting methods (mADF, and ADF). But compared to CER, there is no significant difference in fat-free mass reduction in most intermittent fasting methods. It is important to note that reductions in fat free mass are common across various weight loss strategies [90]. This underscores the necessity for a nuanced understanding of the physiological changes associated with different intermittent fasting strategies.

LDL-cholesterol, as a component of lipid profiles, is another important cardiovascular disease risk factor. Our study found variations in effects on LDL-cholesterol among the different intermittent fasting method. Notably, the PF regimen showed a significant reduction in LDL levels. This aligns with a previous study [18]. However, other studies have not found a consistent effect of intermittent fasting on LDL reduction compared to a usual diet [11, 12, 14].

Our study found significant reductions in both SBP and DBP across multiple intermittent fasting methods, including mADF, PF, and TRE. These findings are partially consistent with previous meta-analyses. Some reported a significant decrease in DBP with intermittent fasting [11, 12], while others did not [13]. Similarly, one meta-analysis found a decrease in DBP with intermittent fasting [11], whereas others showed no effect [13, 14]. These variations highlight the need for further research and potentially personalised approaches to intermittent fasting, considering individual health conditions and risk factors. Another potential benefits of intermittent fasting could be for glycemic control (reduction of blood glucose level). Our study found that TRE method significantly reduced FPG levels. However, these findings are not entirely consistent with previous research. While some meta-analyses reported significant FPG reductions with intermittent fasting [11, 14], others did not observe a significant difference compared to usual eating [13]. The discrepancy could potentially be explained by differences in the duration of the intervention (with some having shorter duration studies) [12,13,14] and number of studies (with some having fewer studies) [12,13,14], as well as some analyse lumped different intermittent fasting method together [11, 12].

The underlying mechanisms of the effect of fasting on cardiovascular risk factors are thought to be mediated, at least in part, by the metabolic switch from carbohydrate utilization to fat and ketones oxidation that happens during fasting [9]. Intermittent fasting causes organs to switch between storing and using energy sources [9]. In conventional eating, carbohydrates and fats get stored in the liver, muscles, and fat tissue. But during fasting, the body burns stored glycogen and fat for energy, resulting in more frequent cycling between storing and burning nutrients compared to constant eating and creates metabolic adaptability and weight reduction [91, 92]. This helps the body become more flexible in using energy, leading to various health benefits, including better insulin sensitivity, increased fat burning, and weight loss [93]. However, more research is needed to understand exactly how specific intermittent fasting patterns affect fat breakdown and turnover and how they influence overall calorie burning.

Strengths and Limitations

This comprehensive systematic review and network meta-analysis employed stringent inclusion and exclusion criteria and included only RCTs. A strength of this review is the ability to compare the relative effectiveness of five commonly used intermittent fasting modalities on a range of cardiovascular disease risk factors, and the certainty of evidence was assessed using the revised version of Cochrane risk of bias assessment tool. This provides valid evidence for decision making and the development of guidance on intermittent fasting. This study incorporated both short-term and long-term studies, and sensitivity analysis was done to assess the robustness of the results. Moreover, in this study, the evidence of certainty has been assessed using the newly validated GRADE framework, which helped to grade the intermittent fasting modalities in a more stringent manner based on a combination of criteria, including effect size, certainty of evidence and SUCRA rankings. Our use of randomized trials strengthens the study’s internal validity but may limit generalizability to real-world settings.

It is essential to note that the lack of direct comparisons between specific intermittent fasting modalities, such as ADF, mADF, TRE, and PF, in our study points towards a gap in the existing literature. The observed risk of bias in 37% of the studies included in our analysis is consistent with the challenges faced by many meta-analyses where the quality of individual studies varies, even though the result remains consistent in the sensitivity analysis. Similarly, the short duration of the included studies might limit the findings, even though the results remain consistent in the sensitivity analysis, except for the effects of mADF on SBP and DBP and the effect of TRE on fat-free mass and FPG, which were no longer significant when excluding studies with short intervention durations. This underscores the importance of interpreting the findings with caution and emphasizes the need for further studies. Future studies should aim to directly compare different intermittent fasting modalities, consider longer-term outcomes, and adhere to rigorous methodologies, including randomization and blinding, to enhance the reliability of results.

Conclusions

This network meta-analysis compared various intermittent fasting methods and found that mADF and TRE were associated with greater reductions in SBP and DBP compared to ADF, and TRE showed greater effects on FPG compared to PF and CER. PF was more effective than usual diets in lowering LDL cholesterol. Both mADF and ADF were more effective than usual diets in reducing body weight, while TRE was associated with reductions in waist circumference, DBP, FPG, and fat-free mass. Among the methods assessed, mADF showed relatively greater effects across several cardiovascular risk factors. These findings suggest that certain intermittent fasting approaches may hold promise as part of lifestyle strategies to improve cardiovascular risk profiles. However, the results should be interpreted with caution due to high risk of bias as per reviewer, and other limitations such as short intervention duration in many studies. Further high-quality, long-term randomized controlled trials are needed to establish the sustained efficacy and safety of different intermittent fasting methods.

Key References

  • D. Herz, S. Karl, J. Weiß, P. Zimmermann, S. Haupt, R. T. Zimmer, J. Schierbauer, N. B. Wachsmuth, K. Khoramipour, M. P. Erlmann, T. Niedrist, T. Voit, S. Rilstone, H. Sourij, and O. Moser. “Effects of different types of intermittent fasting interventions on metabolic health in healthy individuals (EDIF): A randomised trial with a controlled-run in phase”, Nutrients. 2024;16(8). https://doi.org/10.3390/nu16081114.

  • This randomised controlled trial investigated the effect of different intermittent fasting on body composition and metabolic and haematological markers in healthy participants. The data suggest that some fasting interventions might be promising for metabolic health. This reference is ‘of importance’.

  • Obermayer, N. J. Tripolt, P. N. Pferschy, H. Kojzar, F. Aziz, A. Muller, M. Schauer, (A) Oulhaj, F. Aberer, C. Sourij, H. Habisch, T. Madl, T. Pieber, (B) Obermayer-Pietsch, V. Stadlbauer, H. Sour. “Efficacy and Safety of Intermittent Fasting in People With Insulin-Treated Type 2 Diabetes (INTERFAST-2)-A Randomized Controlled Trial”, Diabetes Care 2023;46:463–468. https://doi.org/10.2337/dc22-1622.

  • This randomised controlled study elucidates the safety and effectiveness of intermittent fasting in type 2 diabetes. Findings show that intermittent fasting has the potential to become a promising therapy option in people with insulin-treated type 2 diabetes. This reference is of ‘outstanding importance’.

  • S. Lin, S. Cienfuegos, M. Ezpeleta, K. Gabel, V. Pavlou, A. Mulas, K. Chakos, M. McStay J. Wu, L. Tussing-Humphreys. “Time-Restricted Eating Without Calorie Counting for Weight Loss in a Racially Diverse Population”, Ann Intern Med. 2023; 176(7): 885–895. https://doi.org/10.7326/M23-0052.

  • This randomised controlled trial assessed whether time-restricted eating is more effective for weight control and cardiometabolic risk reduction than calorie restriction or control. Time-restricted eating is more effective in producing weight loss when compared with control but not more effective than calorie restriction in a racially diverse population. This reference is ‘of importance’.