Impact of fiber-fortified food consumption on anthropometric measurements and cardiometabolic outcomes: A systematic review, meta-analyses, and meta-regressions of randomized controlled trials

Abstract The consumption of processed and refined food lacking in fiber has led to global prevalence of obesity and cardiometabolic diseases. Fiber-fortification into these foods can yield potential health improvements to reduce disease risk. This meta-analyses aimed to evaluate how fiber-fortified food consumption changes body composition, blood pressure, blood lipid-lipoprotein panel, and glycemic-related markers. Searches were performed from 5 databases, with 31 randomized controlled trial eventually analyzed. Hedges’ g values (95% confidence interval [CI]) attained from outcome change values were calculated using random-effects model. Fiber-fortified food significantly reduced body weight (−0.31 [−0.59, −0.03]), fat mass (−0.49 [−0.72, −0.26]), total cholesterol (−0.54 [−0.71, −0.36]), low-density lipoprotein cholesterol (−0.49 [−0.65, −0.33]), triglycerides (−0.24 [−0.36, −0.12]), fasting glucose (−0.30 [−0.49, −0.12]), and HbA1c (−0.44 [−0.74, −0.13]). Subgroup analysis differentiated soluble fiber as significantly reducing triglycerides and insulin while insoluble fiber significantly reduced body weight, BMI, and HbA1c. Greater outcome improvements were observed with solid/semi-solid food state than liquid state. Additionally, fiber fortification of <15 g/day induced more health outcome benefits compared to ≥15 g/day, although meta-regression found a dose-dependent improvement to waist circumference (p-value = 0.036). Findings from this study suggest that consuming food fortified with dietary fiber can improve anthropometric and cardiometabolic outcomes. Supplemental data for this article is available online at https://doi.org/10.1080/10408398.2022.2053658


Introduction
Cardiometabolic diseases are a primary cause of poor health and a major contributor to global mortality and vast healthcare burden (Bennett et al. 2018;Miranda et al. 2019).It stems from insulin resistance (IR) which develops in early life, before further progression to subsequent metabolic syndrome, type 2 diabetes, and cardiovascular diseases (Guo, Moellering, and Garvey 2014).This is paired with continually growing global obesity trends (Abarca-Gómez et al. 2017;Malik, Willet, and Hu 2020).It is recognized that dietary modifications are a compelling solution to reduce the risk of both cardiometabolic diseases and obesity, such as through greater consumption of dietary fiber (Benziger, Roth, and Moran 2016).
Dietary fiber is the non-digestible component of plant-derived food, serving as a key part of a healthy and balanced diet due to its physicochemical properties (Dhingra et al. 2012).Glycosidic bonds in dietary fiber cannot be broken down in the human digestive tract, rendering it minimal to low calorie.Additionally, it can increase the viscosity of chyme, thereby delaying gastric emptying while promoting greater secretion of appetite-regulating hormones to assist with weight management (Burton-Freeman 2000;Jenkins et al. 2000a).Furthermore, certain types of dietary fiber can offer prebiotic effects, whereby it gets favorably utilized by gut bacteria with enzymes capable of breaking it down during metabolism, generating gut metabolites involved in both lipemic and glycemic control (Dayib, Larson, and Slavin 2020).Entrapment of lipid molecules within the complex matrix of fiber has also been attributed to improvements in certain lipemic blood biomarkers (Anderson et al. 2009).
Previous epidemiological evidence had drawn inverse associations between dietary fiber or whole-grain intake on the incidences of obesity, cardiovascular disease, and type 2 diabetes (Anderson et al. 2009;Liu et al. 1999;Reynolds et al. 2019;Song and Song 2021;Threapleton et al. 2013;Ye et al. 2012).Yet despite advocates for greater consumption of dietary fiber such as via the direct consumption of wholegrains, fruits, and vegetables (Breneman and Tucker 2013;Mayor 2019), the average consumption of dietary fiber remains at less than 20 g/day (Mayor 2019).The American Heart Association recommends a daily total dietary fiber intake between 25 to 30 g/day, yet more than 90% of Americans do not meet this recommended intake (U.S.Department of Agriculture and U.S. Department of Health and Human Services 2020).This is possibly explained by differential socio-economic status or other psychosocial factors such as habitual eating behaviors (Mann et al. 2015;Ruggiero et al. 2019).Concurrently, in today's industrialized modern world, people generally consume more processed and refined food including packaged snacks and instant convenience food (Martínez Steele et al. 2016).Such modifications typically lead to a loss of health-promoting nutrients like dietary fiber or vitamins, rendering the food as energy-dense with minimal to no nutritive value and this can lead to downstream detrimental physiological effects (Mann 2007).
With today's busy modern urban lifestyle comes a growing demand for convenience delivered through processed food.However, a growing generation of health-conscious consumers is increasingly creating a demand for healthier convenience food, which has led to the recent surge in functional and/or fortified food (including those fortified with fiber) consumption as a means to improve diet quality (Transparency Market Research 2021).Furthermore, as increasing processed food consumption raises certain public health concerns, both governments and food companies are turning to fortification to improve the nutritional quality and health benefits of processed food.
Apart from consuming sufficient fiber, it is also understood that the physiological benefits of dietary fiber can vary depending on the fiber type and/or food application.There is a large variety of fibers isolated from a diverse range of food sources used in today's fiber-fortified products (e.g.beta-glucan, inulin, psyllium).Studies have suggested fiber solubility to be a major determinant for the type of health benefits they impart (Dahl and Stewart 2015).Product application also plays a significant role in the physiological benefits, through formulations of a palatable fiber-fortified product and affecting the rate of nutrient digestion or absorption within the food matrix.
Although there are available reviews regarding the effect of dietary fiber consumption on anthropometric measurements and cardiometabolic outcomes (Barrett et al. 2019;Pol et al. 2013;Threapleton et al. 2013), most studies are focused on the consumption of dietary fiber regardless of the type of source -naturally occurring or in the form of a fortified manufactured product.Thus, this study aimed to conduct meta-analyses from systematically searched randomized controlled trials to evaluate the effect of fiber-fortified product consumption on anthropometric measurements and cardiometabolic outcomes.Further subgroup analyses were conducted to investigate the effect of the dietary fiber type, fortification amount, and the state of food provided (solid/semi-solid versus liquid) on anthropometric measurements and cardiometabolic outcomes, along with meta-regressions of fiber fortification dosage, intervention duration, participant age, and participant body mass index on said outcomes.

Methodology
This systematic review and meta-analyses were conducted following Cochrane's PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines (Moher et al. 2009), with the PICOS (population, intervention, comparison, outcome, and setting) statement as described in Table 1.

Search strategy and eligibility criteria
A computerized search of the articles was conducted on 24 June 2020 using 5 online databases: Pubmed, Embase, Scopus, CINAHL Plus with Full Text, and Cochrane.A research librarian was consulted for appropriate database selection and search terms determination to optimize the search process.An updated search was conducted on 25 April 2021.An example of the search string is as detailed in the supplementary material.
The combined searches were exported to EndNote X9 (Clarivate Analytics, London, United Kingdom) to manage the bibliographies.Following duplicate removal, the remaining articles were independently screened based on the title and abstract of articles by the primary reviewer (AP) and secondary reviewer (FT).Any discrepancies or ambiguity were resolved by a third reviewer (JEK) to reach a consensus.

Data extraction and risk of bias assessment
The following data were extracted from the selected articles by the two independent reviewers: author first name, publication year, population characteristics, RCT design, RCT duration, type of fortified intervention food, intervention fiber type, the quantity of fiber in the food product consumed daily in control and intervention groups, and net changes [mean ± standard deviation (SD)] in the pre-and post-intervention outcomes of interest reported for both control and intervention groups.The authors of the respective articles were contacted via email when additional information was required or when the data were presented in graphical format without the numerical data provided.For this study, several authors were contacted for additional information but not all replied.
The risks of biasness were independently assessed by both reviewers using a modified Cochrane risk of bias assessment tool (Higgins et al. 2011).A subjective risk level of high, low, or unclear was assigned to evaluate the risk levels for selection bias (random sequence generation and allocation concealment), performance bias (participant and researcher blinding), and detection bias (blinding of outcome assessor).

Calculations and statistical analysis
Standard errors (SE) reported in articles were converted to SD with the formula SD u n SE while confidence interval (CI) reported in articles were converted to SD with the formula SD u y n Upper CI Lower CI x (where x was determined based on Excel two-tailed inverse of the t-distribution function: TINV n 1 0 95 1 2 u ., ) ), all fol- lowing the Cochrane Handbook (Higgins et al. 2020).Units for TG, TC, HDL-C, LDL-C, and fasting glucose concentrations were standardized to mmol/L.The conversion factors are: TC, HDL-C, and LDL-C conversion = mg/dL ÷ 38.67; TG conversion: mg/dL ÷ 88.57; fasting glucose conversion: mg/dL ÷ 18.0 (Riemsma et al. 2016;Rugge et al. 2011).Units for insulin were standardized to pmol/L, with conversion factor as: μIU/mL × 6.00 (Knopp, Holder-Pearson, and Chase 2019).Imputed correlation coefficients (Corr) derived from the shortlisted articles were used to calculate the SDs of the change values that were not provided in the articles, using the equation 2 (Higgins et al. 2020).Effect sizes of the change values for each parameter were expressed as Hedges' g.Stata/IC 13 software (StataCorp LP, College Station, TX, USA) was used for the statistical analysis of the extracted data.The metan function was utilized for the pooled outcome effect size.Random-effects model was used to account for heterogeneity and variances among studies where each RCT estimates different but related intervention outcomes (Borenstein et al. 2010).Heterogeneity was assessed with I 2 statistics, where I 2 > 50% indicates significant heterogeneity.Sensitivity analysis was assessed via leave-out-one method, through sequential omitting of each set of comparisons.Publication bias was evaluated with the use of metafunnel function, metabias function, and Egger test.Meta-regression was conducted with the metareg function for each outcome on the intervention dosage, intervention duration, participant age, and participant BMI.p-values of the correlation coefficients were determined based on univariate Monte Carlo simulations (20,000 permutations).Statistical significance was noted at p-value < 0.05.
Crossover design articles were analyzed as per parallel design.While the presence of unit-of-analysis error may occur, the error is minimal and gives little underweighting to the comparisons (Higgins, Deeks, and Altman 2008).Multiple arm interventions in studies were compared separately against the control where each intervention arm was defined as a distinct group for multiple pairwise comparisons.The presence of unit-of-analysis error was checked by dividing the control group across the different arms, validated with sensitivity testing.
Subgroup analysis was performed by segregating comparisons based on the type of fiber (soluble fiber versus insoluble fiber), the quantity of fortified fiber consumed per test food per day (<15 g/day versus ≥15 g/day), and the state of food provided (solid/semi-solid versus liquid).The 15 g benchmark for quantity of fiber was determined based on the discrepancy between daily recommended intake for adults by the American Heart Association and the average daily fiber consumption of an adult (King, Mainous, and Lambourne 2012).

Search results
A flow diagram summarizing the search till meta-analyses is shown in Figure 1.A total of 10,829 articles were obtained, with 7,539 articles remaining for screening following duplicate removal of 3,290 articles.Of the remaining, 7,500 articles were excluded for the following reasons: (1) Population was not human or had a mean age of <19 years old, (2) Study design was not an RCT or did not have a control group, (3) Intervention was not fiber-fortified food or included multiple fortifications or parallel interventions like drug or exercise therapy, (4) Outcomes of interest were not analyzed, (5) Article was not written in English, (6) Article could not be retrieved, (7) Irrelevant to the scope of this review.An additional article was added after the updated search.Subsequently, 9 articles were excluded due to the following reasons: (1) 2 articles had incorrect study design with irrelevant control groups, (2) 2 articles had incorrect exposure due to multiple variables, (3) 2 articles were using the same experimental results from another 2 articles, (4) 3 articles had unpublished data which could not be retrieved from the author.A total of 31 articles were utilized for both the systematic review and meta-analyses.

Systematic review and study characteristics
A summary of the study characteristics is tabulated in Table 2 with 31 articles comprising of 45 comparisons (Bays et al. 2011;Biörklund, Holm, and Onning 2008;Biörklund et al. 2005;Cai et al. 2018;Cugnet-Anceau et al. 2010;Dainty et al. 2016;Effertz, Denman, and Slavin 1991;Ferguson et al. 2020;Hashizume et al. 2012;Hu et al. 2013;Ibrügger et al. 2013;Jenkins et al. 2000b;Keenan et al. 2007;Kuang et al. 2020;Lambert et al. 2017;Liatis et al. 2009;Lu et al. 2004;Machado et al. 2019;Naumann et al. 2006;Pedersen, Sandström, and Van Amelsvoort 1997;Penn-Marshall, Holtzman, and Barbeau 2010;Pick et al. 1996;Pol et al. 2018;Reimer et al. 2017;Russo et al. 2010;Schwab et al. 2006;Soltanian and Janghorbani 2019;Tripkovic et al. 2015;Vega-López, Vidal-Quintanar, and Fernandez 2001;Velikonja et al. 2019;Zunft et al. 2003).The RCT duration ranged from 2 weeks to 12 weeks.There were 26 comparisons with a fiber fortification amount of <15 g/day and 19 comparisons with fiber fortification amount of ≥15 g/day.A total of 37 comparisons were provided with food fortified using soluble fiber while the other 8 comparisons were provided with food fortified using insoluble fiber.Solid/semi-solid food form comprised on 27 comparisons, liquid food form made up 14 comparisons, while the remaining 4 comparisons had a mixture of solids and liquids.Each RCT had a corresponding control food for comparison which comprised of either an unfortified version of the test food or a placebo version with a non-fiber component to replace the test fiber (e.g.maltodextrin or starch).The outcomes available from the 31 articles were: body weight, BMI, fat mass, abdominal fat, WC, WHR, TC, HDL-C, LDL-C, TG, SBP, DBP, fasting glucose, insulin, HOMA-IR, and HbA1c.

Quality and risk for bias of included articles
The risk of bias assessment is tabulated in Table S1 of the supplementary material.Under random sequence allocation, 13 RCTs had an adequate random component and were at low risk of bias, while the rest were determined as unclear because the randomization technique was not specified in the article.Randomization was denoted as low risk if there was clear and objective mentioning of its methodology.Allocation concealment procedures were mentioned in only 2 of the RCTs which were assigned low risk but were unspecified for majority of the RCTs.The blinding of participants and investigators was assessed for performance bias.Low performance bias was noted for 16 articles, 11 were marked unclear while 4 had high performance bias.However, it is to note that the high performance bias is due to the difficulty in masking the intervention food product in terms of texture, taste or appearance and this is a difficulty recognized in food products.Majority of the RCTs were blinded with 4 RCTs single-blinded, 15 RCTs double-blinded, and 1 RCT triple blinded, while the remaining 11 RCTs did not specify if blinding was performed.The blinding of outcome assessors was denoted as low for 12 articles, unclear for 18 and high for 1.

Subgroup analyses on fiber type, fiber quantity, and state of food provided
The results of subgroup analyses are summarized in Table 4, with the respective forest plots located in the supplementary material: fiber type (Figures S18-S29), state of intervention food (Figures S30-S43), and quantity of fiber provided (Figures S44-S58).Results of fiber type for TG and insulin are found in Figure 2. Overall, all models were noted with high heterogeneity.Subgroup analysis was not done for WHR and abdominal fat as there was only 1 comparison in at least 1 of the categories.Subgroup for fat percent on food state and for HOMA-IR on fiber type was also not conducted for the same reason.

Sensitivity analyses and publication bias
Robust trends that were stable to sensitivity analysis were observed for >96% of RCTs, whereby there was no significant change in results when the RCTs were excluded.Publication bias, in the form of funnel plot and Egger test, were tabulated in Figure S59

Meta-regressions
Meta-regressions of dosage for each outcome are tabulated in

Discussion
Whilst several systematic reviews have assessed the impact of dietary fiber on cardiovascular diseases or related conditions, they were mainly centered on dietary fiber as a whole food component of the habitual diet (Barrett et al. 2019;Pol et al. 2013;Threapleton et al. 2013).To our knowledge, while some reviews have collated the impact of fiber-fortified food on health outcomes (Evans et al. 2015;McRorie and McKeown 2017;Whelton et al. 2005), none are as comprehensive as our current analysis on anthropometric and cardiometabolic outcomes.Findings from this study support that the consumption of fiber-fortified food can improve anthropometric measurements and certain markers of cardiometabolic outcomes.Our subgroup analyses noted that the type of fiber fortification and the state of the fortified food provided impact the assessed health outcomes differentially.Interestingly, a fortification of <15 g/day of fiber yielded greater improvements in health outcomes compared to ≥15 g/day.Our meta-regressions found no dose-dependent fiber-fortification effect on all health outcomes except for WC.

Effects of fiber-fortified food consumption on anthropometric measurements
Our meta-analyses found improvements in anthropometric measurements with fiber-fortified food consumption by lowering body weight and body fat mass.Two key mechanisms may explain the observed results.Firstly, for the same quantity of food consumed, dietary fiber triggers earlier satiety and appetite reduction compared to other macronutrients (Burton-Freeman 2000; Warrilow et al. 2019).Dietary fiber tends to confer an inherent reduction in palatability and energy density compared to other key nutrients as it adds non-digestible bulk and weight to the food consumed, of which both these two qualities are positively correlated with satiety and satiation (Burton-Freeman 2000).At the same time, the physicochemical properties of dietary fiber explain the appetite and satiety regulating effects.This is contributed by the texture, bulking, and viscosity-promoting characteristics, which alter several digestion processes across the gastrointestinal tract (Burton-Freeman 2000).For one, high fiber-containing food requires a greater degree of mastication along with increased saliva and gastric juice production which in turn cues gastric expansion and induces satiety (Howarth, Saltzman, and Roberts 2009;Slavin 2005).The increased viscosity contributed by dietary fiber also delays the rate of gastric emptying and nutrient absorption, which triggers a greater secretion of gut hormones such as glucagon-like peptide-1 (GLP-1) and peptide YY (PYY), reducing hunger via gut hormone-induced vagal activation (Howarth, Saltzman, and Roberts 2009).Several articles selected in our meta-analyses on body weight and fat mass also reported lower hunger or satiety levels following the intervention of fiber-fortified food with inulin (Reimer et al. 2017), flaxseed (Kuang et al. 2020;Soltanian and Janghorbani 2019), pea fiber (Lambert et al. 2017), and soy fiber (Effertz, Denman, and Slavin 1991;Hu et al. 2013).
A second mechanism is attributed to host homeostasis via regulation of gut metabolites.The prebiotic effects of dietary fiber provide fermentable substrates to the gut microbiota, which in turn generates short-chain fatty acids (SCFA) (Dayib, Larson, and Slavin 2020).SCFA, once absorbed in the bloodstream, can activate the G-proteincoupled receptors known as free fatty acid receptor 2 (FFAR2) and FFAR3, which are involved in the modulation of key metabolic hormones, including the appetite-regulating hormones PYY and GLP-1 within the gastrointestinal tract (Samuel et al., 2008).Furthermore, both FFAR2 and FFAR3 are located within adipocytes, with involvement in the signaling pathways of lipid metabolism regulation (den Besten et al. 2013).In the presence of SCFA, FFAR2 activation can inhibit adipose insulin signaling and reduce subsequent accumulation of fats in adipocytes while upregulating remaining lipid molecules for utilization in muscles (Kimura et al. 2014).
Altogether, the integrated effects of reduced appetite, higher fat oxidation, and lowered fat storage in adipocytes contribute to the lower body weight and fat mass observed in the meta-analyses.

Effects of fiber-fortified food consumption on cardiometabolic outcomes
Our study noted that fiber-fortified food consumption also improved cardiometabolic outcomes in terms of the blood lipid/lipoproteins (TG, TC, LDL-C) and glycemic indicators (fasting glucose, HbA1c).
The hypolipidemic effects of dietary fiber are well recognized across decades of research (Kritchevsky 1987;Soliman 2019).Most prominently, its high water-holding capacity and viscosity-promoting properties contribute to the hindrance in digestion or intestinal absorption of TG, cholesterol, and bile acid, leading to their subsequent excretion (Jenkins et al. 2000a;Sima, Vannucci, and Vetvicka 2018).Dietary fiber can adsorb bile acid during digestion, and this sequesters bile acid from its emulsifying role in lipid digestion, such that less lipids get fully digested and properly absorbed to influence the blood lipid panel overtime (Pasquier et al. 1996).About 95-98% of all bile acid are reabsorbed and recycled after digestion through the enterohepatic circulation yet this is hindered upon binding of dietary fiber, especially viscous soluble fibers whereby its viscosity-promoting activities further delays reabsorption (van de Peppel, Verkade, and Jonker 2020).Resultantly, a lower rate of bile acid reabsorption occurs when bile acid excretion increases.The diminishing supply of bile acid back to the liver induces the catabolism of bile acid from cholesterol, with the aid of cholesterol-7α-hydroxylase and in doing so, lowers circulating cholesterol levels (Gunness and Gidley 2010).Differing mechanisms are at play depending on the fiber's water solubility and viscosity, further elaborated in a later section regarding fiber type on anthropometric and cardiometabolic outcomes.
The mechanism which dietary fiber improves glycemic-related biomarkers is mainly attributed to the delayed gastric emptying and retarded available carbohydrate digestion or absorption (Jenkins et al. 1978).Positive regulation of GLP-1 or PYY production triggered by fiber intake described above also controls glycemic response by promoting pancreatic beta-cell growth and activity, stimulating insulin biosynthesis and sensitivity, and reinforcing insulin-mediated glucose uptake in muscle and adipose tissue (Buteau 2008;Shi et al. 2015;van den Hoek et al. 2004).The observed reduction in HbA1c is consistent with a previous meta-analysis on type 2 diabetes patients (Silva et al. 2013), although our analysis included healthy, at-risk, and diseased populations.Given that HbA1c is a better long-term indicator of blood glucose management, its improvement is of great importance to adults at risk of developing type 2 diabetes and highlights the essential role which fiber can play in glycemic-related disease risk reduction.
Although dietary fiber intake has been associated with lowered risk of hypertension in a previous meta-analysis (Streppel et al. 2005), the consumption of fiber-fortified food did not have a beneficial effect on blood pressure management in our study.This was likely because the said meta-analysis assessed fiber as a pill supplementation, hence factors such as food palatability and the presence of other nutrients were not considered.Another possible explanation may be the type of dietary fiber fortified.A meta-analysis revealed that beta-glucans significantly reduced blood pressures whereas other types of dietary fiber did not show positive effects (Evans et al. 2015).In our meta-analyses, only 3 out of the 12 articles which reported blood pressure results were based on beta-glucans fortification.

Impact of type of fiber on anthropometric and cardiometabolic outcomes
Our results noted that soluble fiber fortification was more beneficial for improving cardiometabolic outcomes (TG and insulin), while insoluble fiber fortification led to greater improvements in anthropometric outcomes (body weight, BMI) and HbA1c.This is possibly a result of differences in physiochemistry which dictates interaction with other nutrients and gastrointestinal components, such as the gut microbiome and gastrointestinal juices.
Soluble fiber can exist as viscous (e.g.beta-glucan, psyllium) or non-viscous fibers (e.g.inulin), whereby viscous fibers can form viscous gels with digesta contents to decrease dietary lipid digestion and absorption.Viscous fibers also entrap and promote the excretion of bile acids, thereby reducing circulating bile acid levels (Morgan et al. 1993).Additionally, soluble fiber, including, non-viscous soluble fibers, can also offer health outcome improvements through the regulation of the rate-limiting enzyme 3-hydroxy-3-methylglutaryl-CoA reductase which governs cholesterol synthesis and bile acid regulation (Juhel et al. 2011).Both mechanisms inadvertently reduce the availability of bile acid, thereby reducing the digestion and absorption of TG.
Most soluble fibers are more readily-fermented compared to insoluble fibers as they are structurally less densely packed in nature and more accessible to gut bacteria for utilization (Gidley and Yakubov 2019), assisting in the correction of obesity-related gut dysbiosis, a known risk factor for IR (Caricilli and Saad 2013).The beneficial effects of SCFA generated in the gut on lipid metabolism are as elaborated in the first section regarding fiber-fortified food consumption on anthropometric measurements.This further relates to the TG lowering effects of soluble fiber through enhanced intracellular lipolysis triggered by SCFA receptors and extracellular lipolysis mediated by lipoprotein lipase (Canfora, Jocken, and Blaak 2015).The insulin and gut hormone regulating effects mediated by SCFA receptors further improve insulin sensitivity and is associated with lowered insulin response (Canfora, Jocken, and Blaak 2015;Naraoka et al. 2018).
On the contrary, whilst most insoluble fibers are substantially less fermentable than soluble fibers, some insoluble fibers can be fermented via much more intricate pathways albeit a lowered generation of SCFA (Liu et al. 2021).As SCFA-driven improvements are more associated with insulin regulation, this may explain why a reduction in insulin and HOMA-IR was only noted with our soluble fiber comparison and not with insoluble fiber due to the fermentability differences (Zhao et al. 2018).Furthermore, insulin signaling also plays a key role in the maintenance of TG metabolism and storage (den Besten et al. 2013), with insulin resistance promoting the development of dyslipidemia.This explains the observed complementary reduction in insulin levels with TG levels following soluble fiber intervention in this meta-analysis.
The benefits of insoluble fiber fortification in this meta-analyses are centered around weight loss, likely a result of the satiety promoting effect of the indigestible roughage which bulk up stools and slow the intestinal transit of the digesta (Anderson et al. 2009).Whilst these results suggest that insoluble fiber may be more effective than soluble fiber in promoting weight loss, this is more likely because there was a greater application of soluble fiber to liquid food than solid/semi-solid food in the RCTs reviewed.Applications to the latter have a less satiating effect and are described in next section regarding fortified food state on health outcomes.

Impact of fortified food state on health outcomes
The fortification of fiber in solid/semi-solid food (baked goods, pasta, spread, and yoghurt) was more effective in improving both anthropometric and cardiometabolic outcomes (body weight, BMI, fat mass, fasting glucose, and HbA1c) than fortification in liquid food (flavored drinks, juice, and soup), where only various improvements in cardiometabolic outcomes (TG and HOMA-IR) were observed.
The observed differences can be due to the degree of satiety and subsequent consumption which solid/semi-solid and liquid food states provide.Food with greater solids requires a greater duration of and effort for mastication, which cues a longer orosensory exposure.This can accentuate post-ingestive responses such as earlier gut hormone secretion to promote satiety (Appleton et al. 2021;Stribiţcaia et al. 2020).A previous study had shown that consumption of isocaloric and fiber-matched apples in solid form yielded a lower subsequent food intake compared to liquid form, thereby highlighting a potential reason for the perceived anthropometric benefits (Flood-Obbagy and Rolls 2009).As mentioned in our earlier explanations, greater satiety and gut hormones secretion can improve anthropometric outcomes as well as glycemic-related regulation.On the other hand, as liquid food spends little time in the oral cavity, it is less likely to observe oral-associated changes to downstream digestion.While improvements were noted for TG and HOMA-IR, it is possibly because the interventions were all soluble fiber.

Quantity of fiber and meta-regression
Since the daily fiber consumption of most adults is approximately 15 g/day (King, Mainous, and Lambourne 2012;Mayor 2019), half the recommended 30 g/day requirements (Mayor 2019), a fortification of up to 15 g/day can help individuals fulfill the daily recommended fiber intake.In a previous meta-analysis on dietary fiber and whole-grain intake on diabetes management, increasing daily fiber intake by 15 g helped reduce the likelihood of premature mortality in diabetic adults (Reynolds, Akerman, and Mann 2020).Our study also showed that fiber fortification of up to 15 g/day elicited favorable effects on anthropometric and cardiometabolic outcomes (body weight, TG, HOMA-IR, and HbA1c).
Provision of fiber-fortified food with ≥15 g/day of fiber mainly yielded anthropometric outcome improvements (fat mass) and meta-regression also revealed that increasing fiber-fortification provision was correlated to a reduction in WC.Our meta-regression results correspond with that of another cohort study which observed a similar correlation with total dietary fiber intake and WC (Du et al. 2010).However, it is surprising to note that greater fiber-fortification provision did not offer greater cardiometabolic improvements.One possible explanation might be due to a potential increase in intake of overall sugar, saturated fat or calorie from the ingredients used to compensate for the reduced palatability in products where fiber content was much higher.Fortification of fiber at high doses or a lack of experience with formulating high fiber products have been shown to reduce overall product palatability due to unwanted changes in taste and texture.For instance, the addition of ingredients of a high insoluble fiber content (e.g.whole grains, bran) may impart grittiness or a coarse texture to products, and the addition of soluble viscous fiber to liquid-based food may change product rheological properties significantly (Alqahtani et al. 2014).This may suggest that to attain the full benefit of fiber-fortified food, a more intricate degree of food formulation is required for better sensorial acceptability without the need for an excess of undesirable nutrients.
Hence, although the meta-analyses showed favorable health outcomes from the consumption of fiber-fortified food, it is advisable to consume fortified food in moderation and avoid fiber-fortified food that are high in calories, sugar or fat which may outweigh the benefits of the added dietary fiber.

Strengths, limitations, and insights
This review extensively covers how fiber-fortified food can bring about beneficial improvements to anthropometric and cardiometabolic outcomes, beyond just looking at fiber in diets in general.In doing so, it eliminates potential confounding nutrients such as various flavonoids, micronutrients or anti-nutrient factors which tend to be present within natural sources of high fiber food and proves that there is value in the consumption of processed food if they are appropriately formulated and fortified.
Our study also highlighted the differential health benefits between different fiber types, dosage, and state of fortified food.This provides valuable insight for government regulation, food manufacturers and consumers.Several countries allow basic content claims (e.g."source of fiber, " "high in fiber") but few have considered permitted health claims (e.g."beta-glucans contribute to the maintenance of normal blood cholesterol levels") (Stephen et al. 2017).Our review also contributes knowledge on the benefits of specific fiber used in processed food and this can help regulators permit similar health claims for other health benefits (e.g.prebiotic effects).Additionally, food manufacturers interested in making specific fiber-related health claims may use this as a foundation to navigate the complexities of fiber-based health benefits.With the growing demand for more indulgent food (e.g.sugar-sweetened beverages, snacks) (Martínez Steele et al. 2016), fiber fortification can also help counter some of the deleterious effects of these ultra-processed food.
In the studies reviewed, participants were mostly compliant with low drop-out rates, indicative that the proposed dietary interventions were practical to follow and can realistically be extended to habitual living for better health maintenance.This is further supported by the fact that our study assessed for long-term effects of dietary fiber consumption beyond its acute effects on lipemic, glycemic control, and appetite regulation.
On the other hand, one clear limitation within our study is the evident high heterogeneity.We performed sensitivity analysis, subgroup analysis, and meta-regression to define the key sources of heterogeneity but did not note anything of major importance.Aside from the presented results, we also conducted additional subgroup analysis on intervention duration, participant sex, participant health status, participant age group, participant BMI group, type of diet during the intervention, type of intervention provision, type of intervention consumption, and fiber constituent which may differ across the various articles to influence the outcome of interest (data not shown).While majority of the I 2 remained > 90%.Across these additional subgroup analyses, intervention duration of 4 weeks and 6 weeks attained I 2 < 30% for some of the health outcomes although the other duration heterogeneity remained high.While this may mean that intervention duration contributed to some degree of heterogeneity, it is likely that other potential sources of heterogeneity also arose from inconsistencies in the type of food provided, laboratory measurement methods, the total amount of dietary fiber consumed from the test food and other food sources in the diet and other nutrient or lifestyle changes during the intervention period.Additionally, there is evident publication bias for some of the outcomes measured and should be interpreted with caution.

Conclusions and future works
To conclude, our findings support that the consumption of fiber-fortified food products improved anthropometric and cardiometabolic outcomes.Subgroup analyses suggest that soluble and insoluble fiber bring differing outcomes due to their differences in physicochemical properties.Fortified food provided in solid/semi-solid food state tend to offer more beneficial effects on health outcomes compared to liquid state.A consumption of <15 g/day of fiber from fortified food had a greater effect on health outcomes than ≥15 g/day, but greater provision was correlated with lowered WC.It also provides confidence that processed food can act as good carriers of dietary fiber, a nutrient which intake is deficient in most populations.Given that fiber used for food-fortification tend to be isolated fibers, future research can venture into the understanding of whether the former or more natural sources of fiber can have differing impact on said health outcomes.This not only creates better comprehension of dietary fiber research but also brings food manufacturers one step closer to developing healthier food for consumers.

Figure 1 .
Figure 1.Prisma flow chart for the systematic review, meta-analyses and meta-regressions.
fonts indicate significant Hedges' g values.Confidence interval (Ci); Body Mass index (BMi); waist Circumference (wC); waist-Hip ratio (wHr); total cholesterol (tC); High-density lipoprotein cholesterol (Hdl-C); low-density lipoprotein cholesterol (ldl-C); triglycerides (tG); systolic Blood Pressure (sBP); diastolic Blood Pressure (dBP); Homeostatic Model assessment of insulin resistance index (HoMa-ir); Glycated Hemoglobin a1c (Hba1c).aCategory of solid and liquid not shown as comparisons were from the same article.bno analysis was conducted as there was at least one category with only 1 comparison.

Figure 2 .
Figure2.Forest plots of subgroup analysis of fiber type on the effect of fiber fortification on tG (top) and insulin (bottom).the whiskers on either side of the data points represent the 95% Cis, gray boxes indicate the % weight of the comparison and the red dashed line indicate the mid-point of the overall effect size diamond.diamonds which cross the black solid line entirely show a significant effect size and the direction the diamonds lean toward expresses whether the outcome improvement is favored with or without fiber-fortification.Comparisons from the same article are indicated with (i-iv) wherever appropriate.

Table 1 .
description of the PiCos criteria used to define the research question.

Table 2 .
study characteristics of the 31 included articles (45 comparisons).

Table 2 .
(Continued).study were denoted unclear for blinding although the article title stated double blind/ single blind.

Table 3 .
overall Hedges' g and i 2 values of the various outcomes.

Table 4 .
subgroup analyses for fiber type, fiber quantity, and state of food provided, across the various outcomes.outcome

Table 5 .
Meta-regressions of dosage on the various outcomes.