Dietary intake of total vegetable, fruit, cereal, soluble and insoluble fiber and risk of all-cause, cardiovascular, and cancer mortality: systematic review and dose–response meta-analysis of prospective cohort studies

Objectives To conduct a systematic review and meta-analysis of prospective cohort studies to investigate the association between total, vegetable, fruit, cereal, soluble and insoluble fiber intake and risk of all causes, cardiovascular disease (CVD), and cancer mortality and quantitatively assess the dose–response relation. Methods Eligible studies were identified by searching PubMed, Embase and Web of science before August 2023. Random effects models were used to calculate summary relative risk (RR) and 95% confidence intervals (CI) and restricted cubic splines to model the linear/non-linear association. Results The summary RR for all-cause, CVD and cancer mortality of dietary fiber was 0.90 (95% CI: 0.86,0.93), 0.87 (0.84,0.91), 0.91 (0.88,0.93), respectively. Significant association was observed for all-cause and CVD mortality with fruit, vegetable cereal and soluble fiber intake and cancer mortality with cereal fiber intake. No significant association was found for insoluble fiber, vegetable or fruit fiber intake and cancer mortality. Dose-response analysis showed a significant non-linear relation of dietary fiber intake with all-cause mortality, and linear relation for others. Conclusions Higher dietary fiber including different type and food sources of fiber intake were associated with lower risk of mortality. Our findings provide more comprehensive evidence on dietary fiber intake with mortality. Systematic review registration https://www.crd.york.ac.uk/prospero, identifier: CRD42022338837.


. Introduction
Cardiovascular disease (CVD) and cancer are the leading causes of death globally (1).It has been estimated that global deaths from coronary heart disease, stroke, and cancer will reach up to 18.6 million, 12.2 million, and 10.0 million, respectively in 2019-2020 (1)(2)(3).Poor diet contributed to one of the largest risk factors for death, accounting for 8.3% of all deaths (4).The WHO recommends a daily intake of dietary fiber >25 g/day for adults (5); however, the consumption of dietary fiber remains low in many high-income countries (18.3 g/day in the United States, 14.8 g/day in the United Kingdom, 16.9 g/day in France, and 15.0 g/day in Japan) (6).Major nutrition shifts occur in developing countries with an increase in fat intake and a decrease in whole grain and fiber intake.The dietary fiber consumption level was reported to be even lower in middle-income countries (9.7 g per capita/day in China) (7,8).Accumulating evidence indicated that dietary fiber might decrease the risks of various chronic diseases (9,10), including obesity, diabetes, hypertension, CVD (11)(12)(13)(14), and cancer (15)(16)(17).
Inconsistent results were found in previous studies examining the effect of dietary fiber on mortality.Most of the previous studies detected an inverse association between dietary fiber and all-cause, CVD, or cancer mortality (18)(19)(20), but no association was found in other studies (21,22).Although few systematic reviews and meta-analyses were conducted to analyze the relationship between fiber intake and mortality, some of those meta-analyses focused on specific populations such as patients and cancer survivors (23)(24)(25) and unstable findings have been reported with controversial results in many subgroups.A most recent meta-analysis conducted in 2019 analyzed the relationship between total fiber and a series of health outcomes, which included 68,183 deaths, but did not take into consideration the specific types of dietary fiber (26).More than 10 studies (18-20, 22, 27-33) have been reported since the last meta-analysis, with approximately 424,953 participants and 30,215 deaths that could be further added in this updated metaanalysis.Therefore, it is necessary to conduct an updated metaanalysis to explore the association between dietary fiber intake and all-cause, CVD, or cancer mortality and provide evidence on their dose-response relationship.
Dietary fiber can be classified into insoluble and soluble fibers based on solubility (34).Studies on associations between insoluble or soluble fiber intake and mortality have also been inconclusive.In a cohort study of 92,924 Japanese consumption of both insoluble and soluble fibers was associated with a lower risk of all-cause mortality (20).While some observational studies have not found a significant association between soluble or insoluble fiber intake and all-cause mortality mortality (28,35).Only one previous systematic review and meta-analysis investigated the association between soluble and insoluble fiber intake and CVD mortality (36); however, the study did not assess the association between all-cause between all-cause mortality and cancer mortality.
The levels and sources of dietary fiber intake may be substantially different among countries.For example, grain products are the main source of dietary fiber for the US population (37), while dietary fiber mainly comes from vegetables for the Japanese population (38).Bean, fruit, and vegetable fibers but not cereal fibers are associated with reduced risk of all-cause mortality in a study conducted in Japan (39), whereas others reported no associations of individual food sources of dietary fiber (including fibers from cereals, fruits, or vegetables) with the risk of ischemic heart disease mortality mortality (40).Although a previous metaanalysis investigated the association between cereal fiber intake and all-cause, cardiovascular, and cancer mortality (41), the study included general participants and people with diseases, and several cohort studies with large sample sizes have been published in recent years (20,28).Different from previous meta-analyses, this meta-analysis explored dietary fibers from different sources and cardiovascular or cancer mortality.To the best of our knowledge, most previous meta-analyses (24,26,36,(41)(42)(43)(44)(45) did not analyze the relationship between fibers from different sources and mortality.A meta-analysis (46) was conducted on the association between dietary fibers obtained from different sources including cereal, fruit, legume, and vegetable fibers and cardiovascular mortality.
Hence, our study aimed to conduct an updated systematic review and meta-analysis of prospective cohort studies to investigate the risk of all-cause, cardiovascular, and cancer mortality associated with dietary fiber intake and different food sources and different types (soluble and insoluble fiber) of dietary fiber intake in general populations and further explore the doseresponse relationship.

. Methods
The systematic review and meta-analysis were registered in the prospective register of systematic reviews database (PROSPERO) (https://www.crd.york.ac.uk/prospero/index.asp,identifier CRD42022338837) and conducted and reported according to the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (47).

. . Search strategy
We systematically searched the PubMed, Embase, and Web of Science electronic databases from their inception up to 25 August 2023.We used a combination of MeSH terms and freetext terms to identify relevant publications assessing dietary fiber intake and fibers from different food sources in relation to allcause, CVD, and cancer mortality, with restriction to the English language and without date limitation.Moreover, the reference lists from the retrieved articles, systematic reviews, and meta-analyses were searched for further relevant studies.Study authors were contacted, but non-peer-reviewed sources were not considered.Details of the search terms used for querying literature are shown in Supplementary Table 1.The literature search was conducted by two independent investigators (F.Y. and P. Q.).

. . Inclusion and exclusion criteria
The PICOS (participants, interventions/exposures, comparators, outcomes, and study design) criteria were used to identify studies that were eligible for inclusion: (1) the study design was prospective cohort studies; (2) the exposure of interest was dietary fiber intake; (3) the outcome of interest included all-cause, CVD, or cancer mortality; and (4) the risk estimates, including adjusted hazard ratios (HR) or risk ratios (RR), with their corresponding 95% confidence intervals (CIs) were reported.When reports pertained to overlapping participants, we included only the study with a larger population to avoid duplication of data.
Reviews, abstracts, comments, or unpublished results were excluded.Studies on children, adolescents, or patients with chronic kidney disease, or who were undergoing hemodialysis, end-stage cancer, or critical illnesses were excluded.

. . Data extraction and quality assessment
The data extraction and quality assessment were conducted by F. Y. and P. Q., and any discrepancies were discussed with a third investigator (C.H.). The following characteristics from each study finally included in the meta-analysis were extracted using a standardized form: name of first author, publication year, country or region, the name of the study, sample size and number of deaths, follow-up period, types of outcomes, gender, age, types of fibers, amount of intake, measurement of fiber, assessment of interested outcomes, RRs/HRs and 95% CIs, and variables adjusted for in the analysis.When separate risk estimates for men and women were available in a study, their RRs were combined using a fixed-effects model to generate a pooled risk estimate.For dose-response metaanalysis, the risk estimates should be provided for at least three quantitative categories of fiber intake.
We assessed study quality with the Newcastle-Ottawa Scale (NOS) for cohort studies (48).A maximum score of 1 for each question in the checklist can be awarded.Scores were calculated according to three major aspects: selection of participants, adjustment for confounders, and ascertainment of outcomes and nine questions.Scores of 0-2, 3-5, 5-7, and 7-9 were considered poor, fair, good, and high quality, respectively.

. . Statistical methods
For studies reporting HRs for fiber consumption, we assumed that the HR was approximately equal to the RR (49).The missing number of cases in each category was calculated by using the reported RRs/HRs and the number of total cases (50).The average or midpoint of each defined quartile was used for the dose amount.If the category dose range was open-ended, we assumed the length of the open-ended interval to be the same as that of the adjacent interval.For studies reporting risk estimates compared to medium or highest dietary fiber intake, the RR was recalculated by setting the lowest category of dietary fiber intake as the reference.
We computed the highest vs. lowest estimates by using a random-effects model (51), which considered variations (heterogeneities) both within and between studies.We calculated summary RRs (95% CIs) of all-cause, CVD, and cancer mortality per 10 g/day increment.We used the generalized least squares regression to estimate study-specific dose-response associations (52) and the random-effects model to pool the study-specific dose-response RR estimates (51).To examine possible linear or non-linear associations, we used restricted cubic splines for each study with more than three categories of exposure, with three fixed knots at 25%, 50%, and 75% of the total distribution of the reported intake, and combined them using multivariable meta-analysis (53).The significance of non-linearity was calculated using null hypothesis testing (53).We combined the study-specific slopes using random-effects models.
Heterogeneity was assessed using Cochran's Q test and I² statistic (54), with a value of I 2 > 50% considered to represent potentially important heterogeneity, and P < 0.1 was considered statistically significant for the Q statistic (55).Publication bias was assessed using Egger's test and funnel plots.When Egger's test indicated bias, a trim and fill method was used to detect the effect of probable missing studies on the overall effect.We further carried out subgroup analyses stratified by study characteristics, including duration of follow-up (>10 vs. ≤10), number of cases (≤1,000 vs. >1,000), geographical location, study quality (>7 vs. ≤7), adjustment for confounding factors (physical activity (PA), comorbidity at baseline, carbohydrate, protein), and dietary assessment methods, and meta-regression to investigate potential sources of heterogeneity.We also conducted sensitivity analyses excluding each study at a time from each analysis to clarify if the results are robust.A two-tailed P < 0.05 was considered significant.The Stata version 15.0 software (Stata Corp., TX) was used for the analyses.

. Results
The flowchart for the selection is presented in Figure 1.We found 7,947 studies through the database search and reference lists.After removing duplicates, 5,355 records remained.After reviewing the title and abstract of these studies, 5,053 studies were subsequently excluded, and 302 full-text studies were then assessed.After full-text screening, a total of 290 publications were excluded because of duplicated data from the same cohort studies (n = 22), reviews (n = 11), or meta-analyses (n = 15), not relevant exposure (n = 110), not relevant outcome (n = 67), not cohort study (n = 16), or not adults or general population (n = 18).Finally, 32 publications were included in the systematic review and meta-analysis.
The results of the highest vs. lowest meta-analyses on the associations between intake of dietary fiber and allcause, CVD, and cancer mortality are shown in Table 1, Supplementary Figures 1-12.
. .Dose-response meta-analysis . . .Dietary fiber A total of 14 studies (18-22, 28, 32, 57, 59, 69, 72, 75-77) with a total of 1,367,285 participants and 97,469 deaths were included in the dose-response meta-analysis of dietary fiber intake and all-cause mortality.The summary RR for a 10-g/day increment of dietary fiber intake was 0.90 (95% CI: 0.86-0.93;I 2 = 86.1%,P heterogeneity < 0.001; Table 2, Supplementary Figure 19).Evidence of heterogeneity between subgroups in stratified analyses was not found (Supplementary Figure 7).A non-linear dose-response association was found between dietary fiber intake and all-cause mortality (P non−linearity = 0.0096, Figure 2).The shape of the nonlinear curve was steeper with a dietary fiber intake of <15 g/day, but the increase was more gradual after 15 g/day.
A sensitivity analysis showed that exclusion of the studies by Eshak et al. (60) or Pietinen et al. (68) resulted in a change from the non-significant association between fruit fiber intake and CVD mortality to a significant inverse association (Supplementary Table 11).
The sensitivity analysis showed that the summary estimate is robust (Supplementary Table 12).
In the sensitivity analysis, the summary estimate is robust for all-cause and CVD mortality.Exclusion of the study by Katagiri et al. (20) resulted in a change from the non-significant association between insoluble fiber intake and cancer mortality to a significant inverse association (Supplementary Table 13).
In the sensitivity analysis, the summary estimate is robust, except that exclusion of the study by Katagiri et al. and Xu et al. (20,77) lead to a non-significant association between soluble fiber intake and all-cause mortality (Supplemental Table 14).

. . Publication bias
In the highest vs. lowest meta-analysis, Egger's linear regression test and visual inspection of the funnel plots (Supplementary Figure 25) indicated possible publication bias for the association between dietary fiber intake and CVD mortality Frontiers in Nutrition frontiersin.org(P = 0.001), and vegetable fiber intake and all-cause mortality (P = 0.038), but no evidence of publication bias for other outcomes.In the dose-response analyses, Egger's linear regression test and visual inspection of the funnel plots indicated possible publication bias for the association between dietary fiber intake and cancer mortality (P = 0.043) (Supplementary Figures 31,36).
No evidence of significant publication bias was found in other analyses (Supplementary Figures [26][27][28][29][30][32][33][34][35].Application of the trim and fill method did not result in a change in the average effect size, further suggesting that the results were not affected by publication bias.

. Discussion
The present systematic review and meta-analysis investigated the association between dietary fiber intake and different sources and types of fiber intake and all-cause, CVD, and cancer mortality by applying highest vs. lowest, linear, and non-linear dose-response analyses.We found that dietary fiber intake was inversely associated with all-cause, CVD, and cancer mortality.The inverse association was also found for cereal fiber intake.All categories of fibers were inversely associated with CVD mortality.The inverse association of cancer mortality was only detected for cereal fiber and dietary fiber.Significant associations were also found for other fiber intake and all-cause mortality, except for fruit and vegetable fiber intake.Besides, a non-linear relationship was found for allcause mortality.
A large number of longitudinal cohort studies have reported the health benefits of dietary fiber intake (78)(79)(80).Several systematic reviews and meta-analyses suggested that high dietary fiber intake was associated with a reduced risk of all-cause, CVD, and cancer mortality (42,46), which was consistent with the findings from our systematic review and meta-analysis.The subgroup analysis also showed the stability of the findings, which was different from previous meta-analyses meta-analyses (24).This may account for the fact that our study has additionally included more than 14 related studies studies (18-22, 28, 29, 32, 35, 61, 62, 75-77) published in recent years, with than 2,614,294 participants included, compared to the previous meta-analysis.This study found a non-linear relationship between dietary fiber and allcause mortality, showing that the protective effect of dietary fiber is relatively constant when the daily intake is >15 g.A meta-analysis including five papers concluded that risk reduction associated with all-cause mortality was greatest when the daily intake of dietary fiber was between 25 and 29 g, while the dose-response data suggested that amounts >30 g/day confer additional benefits (26).The inconsistent findings might be because of the relatively large number of studies included: publications since 2016 were not included in their dose-response analyses of dietary fiber intake and all-cause mortality (26), and ∼14 more articles updated to 2023 were included in our dose-response meta-analysis.The non-linear relationship of dietary fiber was not found among all-cause and cancer mortality because the of dietary fiber on different health outcomes may have different mechanisms.
Dietary fibers from different food sources have a distinctive mix of different types of compounds and may have a different effect on all-cause and CVD mortality (81, 82).The present systematic review and meta-analysis found the inverse association between vegetable and fruit fiber intake and CVD mortality as well as the significantly inverse association between cereal fiber intake and all-cause, CVD, and cancer mortality, but no association of vegetable fiber with cancer or all-cause mortality.A meta-analysis also found that cereal fiber intake was protectively associated with all-cause, CVD, and cancer mortality, although it included general participants and people with diseases (41).Our study also showed that cereal fiber but not fruit fiber or vegetable fiber was significantly associated with lower total mortality in the dose-response analysis, which was in line with an earlier metaanalysis (45).The recommended level of dietary fiber intake is 25 g for adult women and 38 g for adult men, and the public should consume adequate amounts of dietary fiber from a variety of plant foods (83).Plant foods contain more than just dietary fiber, so any protective properties of plant-based diets may be linked to other dietary components, such as vitamins, minerals, or phytochemicals, and not just isolated dietary fiber (84).The unstable findings in the subgroup analysis suggest that more studies are further needed on the association between fruit fiber and CVD mortality.
Soluble fiber is found in oat bran, barley, beans, lentils, peas, and some fruits and vegetables, while insoluble fiber is rich in foods such as wheat bran, whole grains, nuts, and seeds (77).Although mounting evidence has suggested the protective role of dietary fiber against various chronic diseases (13, 22), the health effect may depend on the dietary fiber type (85,86), and the findings on soluble and insoluble fiber intake and mortality are contradictory (20, 22).Our study found the inverse association between both soluble and insoluble fiber intake and all-cause and CVD mortality.The finding on CVD mortality was in line with one previous systematic review and meta-analysis (87), and our study included eight additional studies (18,20,21,29,32,57,72,77) after 2012 and found a linear relationship.To the best of our knowledge, this is the first study to explore soluble and insoluble fiber intake and allcause and cancer mortality.No significant association was found between insoluble or soluble fiber intake and cancer mortality in the present study, which may be explained by the limited number of studies included.Insoluble fiber is characterized by a fecalbulking ability, which may reduce the risk of cancer mortality (77); however, evidence regarding soluble or insoluble fiber on cancer mortality remains limited and inconsistent, and only three studies (20, 35,77) conducted in Japan and the United States were included in our systematic review and meta-analysis.Further prospective studies on soluble and insoluble fiber intake and cancer mortality are therefore needed.
The mechanism underlying the inverse relationship between dietary fiber and mortality is unclear, but there are several plausible explanations.The protective effect of dietary fiber on cholesterol (88,89), blood pressure (90), insulin sensitivity (85), and blood glucose (91) as well as the anti-inflammatory effects (92) may partly explain the protection from mortality.A study demonstrated that the inclusion of a practical dose of dietary fiber (11.6 g) in a bakery product significantly reduced postprandial glucose and insulin responses in healthy adults (93).Insulin is known to promote the action of the hepatic enzyme 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase (94).Inhibition of HMG-CoA reductase may result in the prevention of excess cholesterol being synthesized and released into circulation by the liver and may thereby reduce the risk of CVD (95).Moreover, alteration of intestinal microbiota composition and function may be an important reason for the potential benefits of dietary fiber (96).Experimental studies also suggested that the reduction of soluble fiber may influence the synthesis of microbial metabolites that are important for regulating metabolic, immune, behavioral, and neurobiological outcomes (97).
This review has some strengths.First, the present study was a comprehensive systematic review and meta-analysis of prospective cohort studies to investigate the association between dietary fiber intake and mortality, using high vs. low analysis and dose-response analysis.Second, the different types and food sources of dietary fiber were also considered, which can provide valuable insight into the mechanisms and evidence for strategies to derive the greatest benefit from balanced consumption of dietary fiber.Furthermore, a large number of participants and deaths have been included and allowed us to quantitatively assess the association between dietary fiber intake and risk of mortality.
In terms of study limitations, first of all, most studies did not consider other nutrients as confounding factors, such as protein, carbohydrate, or fiber from other food sources, which may affect the magnitude of the association between dietary fiber intake and mortality.Besides, comorbidity at baseline was not controlled in a few studies, which could affect the association between dietary fiber and mortality.Second, different dietary fiber assessment tools were used, which might lead to variation in the study results.Third, only three studies (20, 35,77) reported risk estimates on soluble or insoluble fiber intake and cancer mortality, which limit us to conduct the subgroup and sensitivity analyses and suggest the necessity of further studies.Fourth, different diet assessment tools were used (FFQ, 24-h dietary recall, semiquantitative FFQ), and therefore measurement error was unavoidable.Fifth, sensitivity analyses demonstrated a profound lack of robustness among summary estimates for vegetable fiber and fruit fiber intake on mortality in the doseresponse meta-analysis.Sixth, high heterogeneity exists in our meta-analysis of fruit fiber-CVD mortality and dietary fiberall-cause mortality associations, although sensitive and subgroup analyses were conducted to show stable findings.The metaregression analysis was also conducted, and we found that the heterogeneity may come from different levels of study quality for studies included in the meta-analysis of dietary fiber and all-cause mortality and different durations of follow-up for the studies on the association of fruit fiber and CVD mortality.
In conclusion, the present systematic review and metaanalysis found that higher dietary fiber intake was associated with a lower risk of all-cause, CVD, and cancer mortality.For different food sources of dietary fibers, fruit, vegetable, and cereal fiber intake were related to reduced risk of mortality, but there was no association of vegetable or fruit fiber with cancer mortality, showing a significant non-linear relationship between dietary fiber intake and all-cause mortality and a linear relation for other fibers.Our study incorporates different types and food sources of dietary fibers, which provide valuable insight into the mechanisms and may provide evidence for strategies to derive the greatest benefit from a balanced consumption of dietary fiber.The association between insoluble or soluble fiber intake and mortality and the difference between sources of dietary fiber and cancer mortality warrants further investigation.

FIGURE
FIGUREFlowchart of study selection.
TABLE Main characteristics of prospective studies examined the association of dietary fiber intake with all-cause, cardiovascular, cancer mortality.TABLE Dietary fiber intake and risk of all-cause, cardiovascular, and cancer mortality for the highest versus lowest and dose-response meta-analysis.