Introduction
According to the recent estimates, the global prevalence of overweight and obesity will reach approximately 20% by year 2025 if the current trends continue(Reference Kelly, Yang and Chen1). This is alarming because obesity has been linked to a number of adverse health outcomes including diabetes mellitus, cardiovascular issues, fatty liver disease and poor mental health among others, posing a substantial economic burden for the global healthcare systems(Reference Miller and Spencer2). Preventive and therapeutic efforts focus on improving diets and/or promoting greater physical activity. Programmes specifically focused on dietary intakes predominantly target ‘What’(Reference Zobel, Hansen and Rossing3) (food types, macronutrients), ‘How much’(Reference Marteau, Hollands and Shemilt4) (portion size guidance) and ‘When’ people eat(Reference Mielmann and Brunner5) (snacking behaviour). A growing body of evidence shows that eating rate, which characterises ‘How’ people eat(Reference Ohkuma, Hirakawa and Nakamura6), is also an important predictor of body weight and may potentially be a novel avenue for weight management programmes.
Accumulating evidence demonstrates positive associations between eating rate (i.e. the amount of food consumed per unit of time) and weight status across various age groups (from childhood to advanced age)(Reference Fogel, McCrickerd and Goh7,Reference Zhu and Hollis8) , demographics (gender, education and income levels)(Reference Sonoda, Fukuda and Kitamura9,Reference Park and Shin10) and cultural backgrounds (i.e. European, Asian and American samples)(Reference Ketel, de Wijk and de Graaf11,Reference Bellisle12) , in general and clinical (e.g. with obesity or underweight) populations. However, the mechanisms underlying these associations are not well understood. Eating at a slower rate extends the oral exposure time of food, and has been linked with increased glucose response, higher postprandial level of anorexigenic gut peptide YY, greater satiation (earlier meal termination), longer inter-meal satiety(Reference Kokkinos, le Roux and Alexiadou13–Reference Galhardo, Hunt and Lightman15), greater ghrelin suppression, greater reported post-meal fullness, more accurate portion size memory and reduced inter-meal snack consumption(Reference Hawton, Ferriday and Rogers16). Faster eaters may experience less satiety and eat more, which over time can lead to sustained positive energy balance and, in consequence, obesity(Reference Ohkuma, Hirakawa and Nakamura6). Eating rate is considered to be habitual, as it shows good-to-excellent within-individual consistency and stability across the meals, independently of the food type(Reference McCrickerd and Forde17). Still, eating rate changes depending on, for example, food texture(Reference Choy, Goh and Chatonidi18,Reference Koc, Vinyard and Essick19) or eating location(Reference Petty, Melanson and Greene20) and shows individual differences between genders(Reference Shiozawa, Mototani and Suita21), ethnicities(Reference Ketel, Aguayo-Mendoza and de Wijk22) or age groups(Reference Ketel, de Wijk and de Graaf11), pointing to the complex interaction between individual and environmental factors that have not been systematically summarised to date, and are currently poorly understood.
Considering the diversity of factors associated with eating rate, it is necessary to gain a holistic understanding of the variety of individual differences and environmental influences on eating rate, as well as how these factors interact to develop effective and evidence-based interventions that target eating rate. To our knowledge, a systematic review of the factors associated with eating rate has not been conducted to date. The objectives of this systematic review were to: (i) identify factors associated with eating rate; (ii) evaluate the strength and direction of the associations between the identified factors and eating rate; and (iii) to conduct a narrative synthesis of the identified factors, accounting for the strength of the reported associations.
Given the diversity of the factors associated with eating rate, the adapted socio-ecological model (SEM)(Reference Sallis, Owen and Fisher23,Reference Golden and Earp24) was applied to scaffold the identified variables for the purpose of the narrative synthesis. The SEM construct of health posits that internal individual factors interact with the external social and environmental factors to affect health and health-related behaviours(Reference Mahmudiono, Segalita and Rosenkranz25,Reference Kilanowski26) , which was deemed an appropriate framework for the current research question. The adaptation of the SEM for the purpose of the current study included clustering together two levels (community/policy level with group, culture, organisation level to represent ‘Social level’) that were considered separate in the original publication(Reference Sallis, Owen and Fisher23) (Fig. 1). No other adaptations were made. This multi-level integrative framework will allow the transfer of research evidence into translation and implementation guidelines. This study will contribute to identify existing gaps, and guide further multi-disciplinary research directions to develop engaging and well-informed solutions to optimise eating rate.
Socio-ecological model for eating rate. Note: The adaptation includes merging of community/policy level with group, culture, organisation level to represent ‘Social level’. Individual and Environmental levels remain unchanged.
Methodology
Search strategy
A systematic search of eleven databases (CINAHL, EMBASE, IEEExplore, MEDLINE, PAIS, PsycINFO, PubMed, Science Direct, Scopus, Web of Science and ACM Digital Library) was conducted following with Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) standards(Reference Moher, Liberati and Tetzlaff27) and registered with the PROSPERO database(Reference Chien, Khan and Siassakos28) (International Prospective Register of Systematic Reviews: no. CRD42021236498). A list of keywords, including general terms related to eating rate and other related factors were used for each database: (‘eating rate’ OR ‘eating speed’ OR ‘slow eating’ OR ‘eating time’ OR ‘fast eating’ OR ‘quick eating’ OR ‘rapid eating’ OR ‘slow chewing’ OR ‘fast chewing’ OR ‘eating pace’ OR ‘oral-process rate’ OR ‘eating slowly’ OR ‘eating too fast’) AND (‘factor*’ OR ‘cause*’ OR ‘influence*’ OR ‘reason*’ OR ‘determinant*’). For an example search strategy adapted for the Web of Science, see Supplementary 1. Forward and backward reference list searches of all included articles were also conducted through Publish or Perish™ to ensure a comprehensive search. The search covered all the relevant studies in the past five decades (1971–2022) to reflect studies in the contemporary eating environment.
Study eligibility and selection
Following the initial search, Rayyan, an online software system developed for conducting systematic reviews, was used for title and abstract screening(Reference Ouzzani, Hammady and Fedorowicz29). Both abstracts and full texts were independently screened by two reviewers, based on a set of predefined eligibility criteria. Any discrepancies were resolved through discussion until a unanimous decision was reached. Papers were included if the report was a peer-review publication published in English. As this review was predominantly focused on identifying factors associated with eating rate, there was no restriction on study design to ensure a broader coverage of relevant research. Studies were excluded if: (i) the paper was an abstract, dissertation, book, book chapter, demo, review or meta-analysis; (ii) the paper described devices or novel technologies for intake monitoring or eating pattern detection; or the study (iii) reported relevant factors such as bite/sip size and frequency, number of chews and meal time, but did not directly relate to eating rate; (iv) was a randomised control trial that reported eating rate, but did not report factors associated with eating rate; (v) investigated the associations between eating speed and measures of body composition or body weight other than body mass index (BMI)(Reference Seidell and Flegal30,Reference Weir and Jan31) , waist circumference (WC) or waist-to-height ratio (WHtR)(32–Reference Mukai, Doi and Ninomiya34) (these indices of body weight/body composition were selected as the most commonly reported to streamline the search process and subsequent data synthesis); (vi) investigated the factors linked only with drinking behaviour (i.e. alcoholic/non-alcoholic beverages) and not with food-related behaviours; (vii) included subjects that were non-human animals; (viii) investigated the effect of pharmacological therapy on eating rate (e.g. fluoxetine-induced eating behaviour change); (ix) investigated participants with eating disorders including dysphagia, anorexia nervosa, binge-eating disorder, and other eating difficulties or impairments; and (x) investigated participants who need assisted eating, such as infants who require parental feeding and individuals with physical or mental disabilities, who have difficulties in manipulating food in the mouth.
Data extraction and quality assessment
The following essential information was extracted from each eligible study by two researchers, if reported: first author, year of publication, country, abbreviated aim(s) of the study, study characteristic (study design), participant characteristic (age, gender, BMI, ethnicity), sample characteristic (size, sampling method, population), data collection and analysis, outcome and factors identified to be associated with eating rate with statistical evidence. All data were compiled into a standard Microsoft Excel template for further synthesis. Any disagreements were resolved in a discussion. Effect sizes for each identified factor were extracted and/or calculated on the basis of the available data, when the factor of interest was significantly associated with eating rate at the statistical threshold level pre-specified by the study authors (this was p < 0·05 for all of the identified studies) to evaluate the strength of the relationship between eating rate and the individual factors. When the data necessary for the effect size calculation (e.g. standard deviation) were not provided, one of the researchers contacted the corresponding author to obtain the additional information.
The quality of the manuscripts was assessed using standard quality assessment criteria(Reference Kmet, Cook and Lee35). This quality assessment checklist contained fourteen items, including clarification of the research question, study design, method, participants, intervention and random allocation justification, investigator and/or subject blinding (if possible), justification of outcome methods, appropriateness of sample size, control of confounding variables, and sufficient detail in the results, with conclusions supported by results. Each item was scored (‘yes’ = 2, ‘partial’ = 1, ‘no’ = 0) based on the degree to which the specific criterion was met. The summary score for each paper was calculated by dividing the total sum of points by the total possible score and ranged from 0 to 1. The quality assessment was performed by two researchers independently with a strong level of agreement (κ = 0·93). Note that the critical appraisal was used only to assess the quality of studies and no papers were excluded on the basis of their quality score.
Data synthesis
Due to significant heterogeneity in study designs, population characteristics, outcome measurement methods and analytical approaches, a meta-analysis was not considered(Reference Walker, Hernandez and Kattan36). This decision was based on the expert opinion of the Investigator team familiar with the high heterogeneity of methodologies within the eating rate literature and later confirmed by conducting preliminary searches across the databases, prior to commencing the full systematic search and prior to study registration. A narrative synthesis was conducted to provide qualitative evaluations of the available evidence. Two independent reviewers identified factors and synthesised them inductively, and a third reviewer resolved disagreements.
Considering the multifaceted nature of factors associated with eating rate, the socio-ecological model (SEM) was used to synthesise the complex data(Reference Glass and McAtee37). SEM assumes that individual behaviour is shaped and influenced by social and ecological environments, which was considered an appropriate approach given the available evidence on the factors associated with eating rate. Using this approach, a wide range of potential factors were nested under three hierarchical levels: individual (e.g. gender, habits, attitude), social (e.g. family, peers) and environmental (e.g. eating surroundings, food). Social factors, in principle, are also environmental specifically involving other people. Note that, in this review, the differentiation between these factors was highlighted due to the implied degree of their modifiability (social factors are more difficult to modify as they are typically outside of one’s control or deeply engrained). Therefore, in the current study all the environmental factors that involve other people will be referred to as social factors, and all the external factors, excluding the social ones, will be referred to as environmental factors. All the identified factors were described in terms of the direction of the association (i.e. significant and positive, significant and negative, and non-significant), with effect sizes as an indication of their magnitude. Various indices of effect size were either extracted or if not reported, they were calculated by the Investigator team on the basis of the study design, inferential statistics reported and available data. These included Cohen’s d for t-tests, r for correlation coefficients, η 2, R 2, and Cohen’s f for analyses of variance (ANOVAs). As examples, Cohen’s d is based on the difference between observations, divided by the standard deviation of these observations while η 2 was calculated from the sum of squares for the effect divided by the sums of squares for other factors in the design(Reference Higgins, Deeks, Higgins, Thomas, Chandler, Cumpston, Li, Page and Welch38,Reference Rosenthal39) . Three different types of effect size were reported for the studies that conducted ANOVAs, depending on the effect size reported by the authors or, if the effect size was not reported, depending on what data were available. The effect size was interpreted as small, medium or large based on the benchmarks suggested for each effect size measure(Reference Leppink, O’Sullivan and Winston40,Reference Fritz, Morris and Richler41) . Further information is provided in the footnote to Table 1.
A summary of factors and effect sizes# associated with eating rate that emerged from the review, narratively synthesised across the levels of the socio-ecological model
Note: Numbers in the table refer to the relevant references;
*indicates the baseline condition. Any further remarks about the study findings are presented next to the references;
#Effect size annotation: small effect size – black box outline, white background, black font 
; medium effect size – black box outline, grey background, black font 
; large effect size – black box, white font 
; when effect size was not computed, only the reference number is presented without an accompanying box. Effect size calculations based on the following values: small (Pearson r or correlation coefficient = ±0·1; coefficient of determination (r 2 or R 2) = 0·01; η 2 = 0·01; ω 2 = 0·01; Cohen’s f = 0·1; Cohen’s d = 0·2; odds ratio (OR) = 1·5; relative risk or risk ratio (RR) = 2); medium (Pearson r or correlation coefficient = ±0·3; coefficient of determination (r 2 or R 2) = 0·09; η 2 = 0·06; ω 2 = 0·06; Cohen’s f = 0·25; Cohen’s d = 0·5; odds ratio (OR) = 2; relative risk or risk ratio (RR) = 3); large (Pearson r or correlation coefficient = ±0·5; coefficient of determination (r 2 or R 2) = 0·25; η 2 = 0·14; ω 2 = 0·14; Cohen’s f = 0·40; Cohen’s d = 0·8; odds ratio (OR) = 3; relative risk or risk ratio (RR) = 4); NA indicates that there were not enough data provided for a effect size calculation; BMI, body mass index; WC, waist circumference; WHtR, waist-to-height ratio; UPF, ultra-processing food; detailed information about raw effect size is provided in Supplementary Material 2.
Results
General study characteristics
A flowchart outlining the literature search is summarised in Fig. 2. Of the 4932 studies, 337 were included during the initial screening. Major reasons for exclusion were: 109 studies did not investigate eating rate (i.e. most studies focused on eating behaviours other than eating rate), 65 did not identify factors associated with eating rate, 23 were interventions that did not identify factors associated with eating rate, and 41 reported associations body weight/body composition and eating rate but reported measures of body weight/body composition other than BMI, WC or WHtR. To ensure sufficient coverage of all relevant papers, a follow-up citation tracking(Reference Weightman, Morgan and Shepherd42) was employed. Specifically, the reference lists of the 100 eligible papers were scrutinised to identify any further pertinent articles for inclusion. Consequently, this process yielded 4 additional papers that satisfied the predetermined inclusion criteria, resulting in a total of 104 manuscripts that were included in this systematic review.
Flow diagram of the literature search strategy.
An overview of the study characteristic and key findings of the included studies are outlined in Supplementary 2. Among the included studies, forty-five (43%) were conducted in Europe, thirty-nine (38%) in Asia, seventeen (16%) in North America, two (2%) in New Zealand and one (1%) in Australia. The study populations consisted of various ethnicities, and five studies(Reference Ketel, de Wijk and de Graaf11,Reference Ketel, Aguayo-Mendoza and de Wijk22,Reference Fogel, Goh and Fries43–Reference Teo, van Dam and Whitton45) specifically examined ethnic differences in eating rate. Among the different types of study design, the majority (n = 47; 45%) had cross-sectional design, thirty-six (35%) were (crossover) randomised controlled trials, fifteen (14%) came from cohort studies and the remaining six (6%) had case–control design. Most studies (n = 99; 95%) reported effect sizes; however, five studies(Reference Bell, Spruijt-Metz and Vega Yon46–Reference Bellisle, Lucas and Amrani50) (5%) failed to provide sufficient data for computing the effect sizes. Corresponding authors of these studies were contacted to obtain data for effect size computation, though none of the authors replied within the stipulated time period. The sample sizes of the reviewed studies varied greatly, ranging from 10(Reference Bellisle, Lucas and Amrani50) to 197 825(Reference Kudo, Asahi and Satoh51) participants. Similarly, participants from various age groups were included, with age ranging from 4 years(Reference Berkowitz, Moore and Faith52,Reference Costa, Severo and Oliveira53) to 87(Reference Nakamura, Nakamura and Takashima54) years. Most studies (n = 76; 73%) included adults between 18 and 65 years old, a number of studies (n = 21; 20%)(Reference Fogel, McCrickerd and Goh7,Reference Fogel, Goh and Fries43,Reference Bell, Spruijt-Metz and Vega Yon46,Reference Van den Bulck and Eggermont47,Reference Berkowitz, Moore and Faith52,Reference Costa, Severo and Oliveira53,Reference Llewellyn, van Jaarsveld and Boniface55–Reference Ochiai, Shirasawa and Nanri69) explicitly explored eating rate among children or adolescents, and seven studies(Reference Zhu and Hollis8,Reference Ketel, de Wijk and de Graaf11,Reference Ketel, Aguayo-Mendoza and de Wijk22,Reference Ketel, Zhang and Jia44,Reference Nakamura, Nakamura and Takashima54,Reference Aguayo-Mendoza, Santagiuliana and Ong70,Reference Okubo, Murakami and Masayasu71) (7%) investigated older adults. The majority of the studies (n = 82; 79%) included both males and females; however, nine studies (8%)(Reference Hordern72–Reference Hamada, Miyaji and Hayashi80) investigated females only, and six (6%) recruited only males(Reference Choy, Goh and Chatonidi18,Reference Nishitani, Sakakibara and Akiyama81–Reference James, Maher and Biddle85) . Only seven studies (7%)(Reference Fogel, McCrickerd and Goh7,Reference Park and Shin10,Reference Petty, Melanson and Greene20,Reference Barkeling, Rossner and Sjoberg86–Reference Ilic, Tomasevic and Djekic89) had a balanced male/female ratio. Most studies (n = 42; 40%) examined participants with a healthy weight status, and thirty-seven studies (36%) did not provide inclusion or exclusion criteria based on body weight, body composition or adiposity. Four studies (4%) focused only on people with overweight and obesity(Reference Laessle and Lehrke68,Reference Hordern72,Reference Canterini, Gaubil-Kaladjian and Vatin73,Reference Almiron-Roig, Tsiountsioura and Lewis77) defined by ethnicity-specific cut-off values(90–92), whereas two (2%) studies involved lean Asian young men(Reference James, Maher and Biddle85) and lean Asian children(Reference Okubo, Murakami and Masayasu71). The study quality scores are summarised in Supplementary 3, ranging from 0 (poor quality) to 1 (high quality). In general, study quality was moderate (mean 0·78, median 0·82), with scores ranging from 0·41(Reference Azrin, Kellen and Brooks93) to 1(Reference Fogel, McCrickerd and Goh7,Reference Eloranta, Lindi and Schwab67,Reference Ohkuma, Fujii and Iwase94) . Four papers(Reference Buergel, Bergman and Knutson59,Reference Canterini, Gaubil-Kaladjian and Vatin73,Reference Suh and Jung78,Reference Azrin, Kellen and Brooks93) received low ratings (using 0·5 as the cut point(Reference Kmet, Cook and Lee35)).
Definition and measurement of eating rate
There was no consensus on the definition of eating rate across the studies. In 53 of 104 (51%) papers (Table 2), eating rate was defined using a pre-specified formula. Among these, the majority (n = 37; 70%) calculated eating rate by dividing the grams consumed by the total meal duration (i.e. from the start of the first bite to the swallow of the last bite). One study(Reference Berkowitz, Moore and Faith52) (1%) defined eating rate as total energy intake (i.e. kilocalories) divided by the meal duration (referred to as ‘energy intake rate’ in other studies(Reference Teo, Lim and Goh95,Reference Forde, Mars and de Graaf96) ), and five studies(Reference Petty, Melanson and Greene20,Reference Hordern72,Reference Teo, Lim and Goh95,Reference Laessle, Lehrke and Duckers97,Reference Rogers, Drumgoole and Quinlan98) (9%) used both measures. In thirteen studies (25%) total oral exposure time (time food spent in mouth) rather than meal duration was used to compute eating rate (e.g. Fogel et al.(Reference Fogel, Goh and Fries43) investigated the association between faster eating rates and higher BMI among 4·5-year-old children). Two studies(Reference Bell, Spruijt-Metz and Vega Yon46,Reference Llewellyn, van Jaarsveld and Boniface55) (4%) defined eating rate as number of bites per minute. Other studies (n = 32, 31%) measured eating rate using questionnaires (e.g. Sakata’s Eating Behaviour Questionnaire (SEBQ)(Reference Nishitani, Sakakibara and Akiyama81)) and defined it according to the questionnaire definition (e.g. one study(Reference Lee, Kim and Jang99) using mealtime duration as a numerical measure) or did not define eating rate at all (e.g. question(Reference Lee, Mishra and Hayashi100) such as ‘compared to other people, is your eating speed quicker’).
Definition of eating rate provided in reviewed papers (n = 53)
Note: Five studies(Reference Petty, Melanson and Greene20,Reference Hordern72,Reference Teo, Lim and Goh95,Reference Laessle, Lehrke and Duckers97,Reference Rogers, Drumgoole and Quinlan98) used both grams and kilocalories consumed divided by the unit of time (g/min) and (kcal/min) or (g/s) and (kcal/s).
Due to the heterogeneous definitions and study designs, methodologies used to determine eating rate varied between the studies (Table 3). In general, these can be divided into two categories: (1) objective measurements (n = 69; 66%) using video analysis, universal eating monitors (UEM)(Reference Kissileff and Guss101), Mandometer®(102), electromyography (EMG) sensors and others; (2) and subjective measures (n = 32; 31%) from self-reported questionnaires, which were mostly used in studies with large sample sizes. Subjectively measured eating rate often utilised Sakata’s Eating Behaviour Questionnaire (SEBQ)(Reference Nishitani, Sakakibara and Akiyama81) in studies with adults, Children’s Eating Behavior Questionnaire (CEBQ)(Reference Costa, Severo and Oliveira53) for parental reports of children’s eating rate and self-reports measured using direct questions in non-validated surveys (e.g. How fast is your eating speed? How quickly do you eat in comparison with others?), with semi-qualitative responses (e.g. slow, average/medium, fast)(Reference Bellisle, Lucas and Amrani50), visual analogue scales(Reference Van den Bulck and Eggermont47,Reference Slyper, Shenker and Israel60,Reference Canterini, Gaubil-Kaladjian and Vatin73,Reference Eloranta, Lindi and Schwab103,Reference Leong, Gray and Horwath104) or other scoring methods(Reference Yamagishi, Sairenchi and Sawada64). Across the studies which used subjective measures of eating rate there was often poor distribution of scores(Reference Teo, van Dam and Whitton45), and in most cases (eight out of nine studies) the categories had to be combined (e.g. ‘very slow’ and ‘slow’ or ‘fast’ and ‘very fast’) to facilitate data analysis(Reference Yamane, Ekuni and Mizutani105). Only one study(Reference Lee, Kim and Jang99) explicitly benchmarked meal duration in units of time (e.g. <5 min per meal, ≤5 and ≤10 min per meal) as a proxy for asking the participant to determine their eating rate. For objective measurements, studies predominantly (n = 34; 49%) used video recordings to capture the eating process and code eating microstructure from the recordings. The oral processing behaviours such as oral processing time, number of chews, number of bites and number of swallows were subsequently computed from analysis of eating microstructure using behavioural annotation software. Among these, some studies (n = 6; 18%) additionally placed stickers on subjects’ faces (e.g. nose tip, forehead and chin) as a reference point to extract more accurate microstructure parameters (i.e. bites, chews and swallows) during video analysis(Reference Ketel, de Wijk and de Graaf11,Reference Ketel, Aguayo-Mendoza and de Wijk22,Reference Ketel, Zhang and Jia44,Reference Aguayo-Mendoza, Ketel and van der Linden106–Reference Gonzalez-Estanol, Libardi and Biasioli108) . Similarly, participants in the three studies were instructed to indicate every moment of swallowing by raising their hands(Reference Aguayo-Mendoza, Santagiuliana and Ong70,Reference Doyennette, Aguayo-Mendoza and Williamson109,Reference Aguayo-Mendoza, Chatonidi and Piqueras-Fiszman110) . Sixteen studies (23%) only recorded total food consumption and mealtime with an electronic scale and/or stopper for eating rate calculation rather than measuring other eating oral processing behaviours, such as bite size and chewing rate. Other, less common methods, included the Universal Eating Monitor (UEM)(Reference Petty, Melanson and Greene20,Reference Laessle, Uhl and Lindel56,Reference Laessle and Lehrke68,Reference Hordern72,Reference Almiron-Roig, Tsiountsioura and Lewis77,Reference Barkeling, Rossner and Sjoberg86,Reference Laessle, Lehrke and Duckers97,Reference McLeod, James and Witcomb111) , Mandometer(Reference Langlet, Anvret and Maramis57), electromyography (EMG) sensors(Reference Park and Shin10,Reference Bellisle, Lucas and Amrani50,Reference Sun, Ranawana and Tan112) and direct observation(Reference Rosenthal and McSweeney49,Reference Hodgson and Greene75,Reference Westerterp-Plantenga, Wouters and ten Hoor79,Reference James, Maher and Biddle85,Reference Azrin, Kellen and Brooks93,Reference Hill and McCutcheon113,Reference Lin, Lin and Hung114) .
Eating rate detection methods/tools provided in reviewed papers (n = 101)
EMG, electromyography of the masticatory muscles; SEBQ, Sakata Eating Behaviour Questionnaire; CEBQ, Children’s Eating Behavior Questionnaire.
Factors associated with eating rate: socio-ecological model
Individual factors
Of the 104 included studies, 66 (63%) investigated the relationship between eating rate and various individual-level factors. The factors were centred around five key themes including eating microstructure(Reference Pearce, Cevallos and Romano115) (bites, chews, swallows) and/or food oral processing behaviour (e.g. chews per bite), eating habits, demographic factors, body characteristics and cognitive/psychological factors. For simplicity, we will refer only to oral processing behaviours rather than microstructure and/or oral processing behaviours throughout.
All the identified factors have been graphically presented in a word cloud in Fig. 3 created using open access software (Wordart). The word cloud summarises the identified factors using two dimensions: font size and font colour. Font size is a qualitative interpretation of the study findings representing the number of studies that investigated the given factor and their effect size, with larger font representing a larger number of studies, with stronger and/or more consistent effect sizes. Font size has been determined using the following formula created especially for the purpose of the current analysis: number of entries of the factor into the word cloud = ((number of studies with small effect size × 1) + (number of studies with medium effect size × 2) + (number of studies with large effect size × 3)) – (number of studies that found no association). Where the study reported a significant positive/negative association but the effect size was not provided and/or could not be computed, a small effect size was assumed and the factor was entered once. Different colours have been used to differentiate between levels of SEM.
A word cloud diagram depicting all factors associated with eating rate identified in the current study. Note: The font size varies depending on the number of studies that investigated the specific factor and the effect size, with larger font representing more studies and/or with larger effect sizes. Different colours have been used to differentiate between the different levels of socio-ecological model (blue, individual level; orange, environmental level; pink, social level). The following formula was created for the purpose of this analysis to determine the font size: ((number of studies with small effect size × 1) + (number of studies with medium effect size × 2) + (number of studies with large effect size × 3)) – number of studies that found no association) = number of entries of the factor to the word cloud. Where the study reported a significant positive/negative association but effect size could not be computed, a small effect size was assumed.
Food oral processing behaviour
Food oral processing(Reference Stieger and van de Velde116), which is the manipulation and degradation of food inside the mouth before swallowing, has been one of the most investigated (n = 14; 21%) individual factors. Despite the variety of food properties, individuals tend to exhibit consistent habitual oral processing behaviours(Reference McCrickerd and Forde17) such as chews per bite, oral exposure time, bite size or inter-bite interval, all of which have been associated with eating rate. Among these, higher number of chews per bite(Reference Zhu and Hollis8,Reference Fogel, Goh and Fries43,Reference Canterini, Gaubil-Kaladjian and Vatin73,Reference Ferriday, Bosworth and Godinot74,Reference Gonzalez-Estanol, Libardi and Biasioli108,Reference Forde, Leong and Chia-Ming117,Reference Bolhuis, Forde and Cheng118) has shown strong associations with reduced eating rate, except one study showing a small effect size(Reference Ketel, Zhang and Jia44) and another study(Reference Zhu and Hollis48) that failed to provide sufficient data to calculate an effect size. Most studies have also consistently identified larger bite size(Reference Ketel, de Wijk and de Graaf11,Reference Fogel, Goh and Fries43,Reference Ferriday, Bosworth and Godinot74,Reference James, Maher and Biddle85,Reference Hill and McCutcheon113,Reference Forde, Leong and Chia-Ming117–Reference Mosca, Torres and Slob119) as a strong factor linked to faster eating rate among participants who differed in weight status (e.g. people with overweight(Reference Almiron-Roig, Tsiountsioura and Lewis77)) or age groups (e.g. children(Reference Fogel, Goh and Fries43), adults(Reference Hill and McCutcheon113) and elderly(Reference Ketel, de Wijk and de Graaf11)), with only one study showing a small effect size(Reference Ketel, Zhang and Jia44). Other elements of food oral processing, such as oral exposure time per bite showed mixed associations with eating rate, with two studies(Reference Fogel, Goh and Fries43,Reference Mosca, Torres and Slob119) reporting a small effect, while three(Reference Choy, Goh and Chatonidi18,Reference Gonzalez-Estanol, Libardi and Biasioli108,Reference Forde, Leong and Chia-Ming117) reported a large effect size. In only one study, shorter intervals between mouthfuls(Reference Ferriday, Bosworth and Godinot74) were associated with faster eating rate with a large effect size.
Eating habits
Eleven studies (16%) investigated elements of eating habits and eating rate. Various factors were considered, including snacking behaviour(Reference Kudo, Asahi and Satoh51) and irregular diet(Reference Kudo, Asahi and Satoh51), food-to-mealtime context(Reference McLeod, James and Witcomb111), food palatability(Reference Bellisle, Lucas and Amrani50,Reference Hodgson and Greene75,Reference Westerterp-Plantenga, Wouters and ten Hoor79,Reference Yeomans87,Reference Hill and McCutcheon113) and perceived hunger/satiety(Reference Bellisle, Lucas and Amrani50,Reference Slyper, Shenker and Israel60,Reference Hodgson and Greene75,Reference Azrin, Kellen and Brooks93,Reference Hill and McCutcheon113,Reference Hinton, Leary and Comlek120) . Small positive associations were shown between tendency to snack or irregular diet and faster eating rate, though these were investigated only in single studies. McLeod et al.(Reference McLeod, James and Witcomb111) suggested that eating rate was slower and less food was consumed when food was presented at an unusual mealtime (i.e. pasta for breakfast) rather than the usual food-to-mealtime context (i.e. pasta at lunch; small positive effect size), though this has not been consistent across the foods (inconsistent results for a sweet porridge dish).
Food palatability and perceived satiation were two of the most investigated factors (n = 8) pertaining to the eating habits. Individuals who reported having greater food flavour preference or food palatability were more likely to consume food at a faster rate. Among the studies, three indicated a large effect size(Reference Hodgson and Greene75,Reference Westerterp-Plantenga, Wouters and ten Hoor79,Reference Yeomans87) , one indicated a small effect size(Reference Hill and McCutcheon113), while another(Reference Bellisle, Lucas and Amrani50) did not supply adequate data for effect size computation. Similarly, hungry subjects were observed to take less time per bite and had faster eating rate than non-hungry subjects(Reference Slyper, Shenker and Israel60,Reference Hodgson and Greene75,Reference Azrin, Kellen and Brooks93,Reference Hill and McCutcheon113) with small-to-medium effect size, though this association was not very consistent(Reference Bellisle, Lucas and Amrani50,Reference Hinton, Leary and Comlek120) . Studies differed in the methodologies, including duration of fast (15 h versus 4 h), diurnal versus nocturnal fasting, types of food served and meal context (e.g. an ad libitum buffet or fixed-portion meals). Generally, foods perceived to be more palatable were eaten faster, while perceived hunger/satiety showed less consistent associations.
Demographic characteristics
Out of the studies included in this review, twenty-one studies (31%) examined demographic factors in the context of the eating rate, including age, gender and ethnicity. Four(Reference Ketel, de Wijk and de Graaf11,Reference Ketel, Aguayo-Mendoza and de Wijk22,Reference Ketel, Zhang and Jia44,Reference Aguayo-Mendoza, Santagiuliana and Ong70) of the six studies reported a strong negative association, and one reported a weak negative association(Reference Kudo, Asahi and Satoh51), between age and eating rate, suggesting that older adults (age range 70–85 years old) had lower eating rate compared with the younger participants (age range 18–30 years old). One study(Reference Bell, Spruijt-Metz and Vega Yon121) investigating eating rate in families with adolescents indicated a marginally faster eating rate among adolescents compared with their parents (effect size could not be computed). No studies compared eating rate among various age groups in childhood (early, mid-, late childhood), or across various age groups in adulthood (e.g. early, mid-, late adulthood).
Of the sixteen studies that examined gender, twelve(Reference Park and Shin10,Reference Ketel, de Wijk and de Graaf11,Reference Petty, Melanson and Greene20,Reference Ketel, Aguayo-Mendoza and de Wijk22,Reference Kudo, Asahi and Satoh51,Reference Langlet, Anvret and Maramis57,Reference Gomez-Martinez, Martinez-Gomez and Perez de Heredia63,Reference Barkeling, Rossner and Sjoberg86,Reference Yeomans87,Reference Hinton, Leary and Comlek120,Reference Leong, Madden and Gray122,Reference Stubbs, O’Reilly and Whybrow123) reported that females have slower eating rate than males regardless of the type of food consumed and their age groups (e.g. children(Reference Fogel, Goh and Fries43), school-going adolescents(Reference Langlet, Anvret and Maramis57), and adults(Reference Ketel, Aguayo-Mendoza and de Wijk22)), while the rest reported lack of association(Reference Shiozawa, Mototani and Suita21,Reference Bell, Spruijt-Metz and Vega Yon46,Reference Slyper, Shenker and Israel60,Reference Rosenthal and Philippe124) . Gender-based differences could be partially mediated by physical differences in oral physiology(Reference Ketel, de Wijk and de Graaf11) including larger oral cavity among males, as well as larger head height and width. The reported effect sizes varied across the studies, ranging from small(Reference Kudo, Asahi and Satoh51,Reference Gomez-Martinez, Martinez-Gomez and Perez de Heredia63) to medium(Reference Laessle and Lehrke68,Reference Hill and McCutcheon113) to large(Reference Park and Shin10,Reference Ketel, de Wijk and de Graaf11,Reference Petty, Melanson and Greene20,Reference Ketel, Aguayo-Mendoza and de Wijk22,Reference Langlet, Anvret and Maramis57,Reference Barkeling, Rossner and Sjoberg86,Reference Yeomans87,Reference Hinton, Leary and Comlek120) .
Consistent ethnic differences in eating rate were found(Reference Ketel, Aguayo-Mendoza and de Wijk22,Reference Fogel, Goh and Fries43–Reference Teo, van Dam and Whitton45) with medium-to-large effect sizes, though only limited cross-cultural comparions have been reported. Specifically, Asian Chinese participants tended to have better mastication performance(Reference Ketel, Zhang and Jia44) and lower eating rate characterized by more chews per bite and a smaller average bite size compared with Caucasian Dutch participants when consuming chewable foods (e.g. raw carrots, cheese and beef). Moreover, Chinese participants tended to have a relatively slow eating rate regardless of their age (children(Reference Fogel, Goh and Fries43), adults(Reference Teo, van Dam and Whitton45)), when compared with other major Asian ethnic groups (e.g. Malay and Indian(Reference Fogel, Goh and Fries43)). Notably, all the reported studies focused on cross-cultural comparisons between Asians and European Caucasians, and there was lack of evidence pertaining to any other ethno-cultural backgrounds (e.g. African, Middle Eastern ethnicities or Aboriginal cultures).
Body characteristics
The association between body weight status and eating rate has been extensively studied (n = 39 studies) and covered in other reviews(Reference Robinson, Almiron-Roig and Rutters125). In most studies (n = 33; 85%), BMI was used to estimate body weight status using ethnic specific cut-off values, further used to compare eating rates among participants with healthy weight and with overweight/obesity. Thirty-seven out of thirty-nine studies reported a positive association between higher body weight/overweight and eating rate, with effect size ranging from small(Reference Teo, van Dam and Whitton45,Reference Gomez-Martinez, Martinez-Gomez and Perez de Heredia63,Reference Ochiai, Shirasawa and Nanri69,Reference Okubo, Murakami and Masayasu71,Reference Azrin, Kellen and Brooks93,Reference Eloranta, Lindi and Schwab103,Reference Zhu, Haruyama and Muto126,Reference van den Boer, Kranendonk and van de Wiel127) to medium(Reference Kudo, Asahi and Satoh51,Reference Berkowitz, Moore and Faith52,Reference Nakamura, Nakamura and Takashima54,Reference Slyper, Shenker and Israel60,Reference Fagerberg, Charmandari and Diou61,Reference Yamagishi, Sairenchi and Sawada64,Reference Zeng, Cai and Ma66,Reference Shin, Lim and Sung83,Reference Laessle, Lehrke and Duckers97,Reference Yamane, Ekuni and Mizutani105,Reference Hill and McCutcheon113,Reference Fujii, Funakoshi and Maeda128–Reference Nagahama, Kurotani and Pham130) to large(Reference Llewellyn, van Jaarsveld and Boniface55,Reference Potter, Gibson and Ferriday58,Reference Gong, Li and Wang62,Reference Hamada, Miyaji and Hayashi80–Reference Tanihara, Imatoh and Miyazaki82,Reference Barkeling, Rossner and Sjoberg86,Reference Ohkuma, Fujii and Iwase94,Reference Lee, Kim and Jang99,Reference Lee, Mishra and Hayashi100,Reference Bolhuis and Keast131–Reference Wuren and Kuriki134) . Among these, twenty-one were cross-sectional analyses, including eight studies specifically focusing on children and adolescents from various ethnicities, such as Japanese(Reference Ochiai, Shirasawa and Nanri69,Reference Okubo, Murakami and Masayasu71) , Finnish(Reference Eloranta, Lindi and Schwab103), Chinese(Reference Gong, Li and Wang62,Reference Zeng, Cai and Ma66) , Spanish(Reference Gomez-Martinez, Martinez-Gomez and Perez de Heredia63), Swedish or Greek(Reference Fagerberg, Charmandari and Diou61) and Multi-Asian(Reference Fogel, Goh and Fries43). In prospective analyses, five studies(Reference Berkowitz, Moore and Faith52,Reference Llewellyn, van Jaarsveld and Boniface55,Reference Slyper, Shenker and Israel60,Reference Yamagishi, Sairenchi and Sawada64,Reference Okubo, Miyake and Sasaki129) focused only on children, while one study(Reference Tanihara, Imatoh and Miyazaki82) assessed the relationship between 8-year weight change and eating rate for Japanese middle-aged male workers only. These findings support findings from the previous reviews(Reference Ohkuma, Hirakawa and Nakamura6,Reference Garcidueñas-Fimbres, Paz-Graniel and Nishi135,Reference Yuan, Liu and Liang136) demonstrating consistent associations between faster eating and greater body weight. Three studies failed to find an association between body weight and eating rate(Reference Laessle, Uhl and Lindel56,Reference Leong, Gray and Horwath104) . Some mediators of this association have been investigated. One cross-sectional study(Reference Laessle, Uhl and Lindel56) demonstrated that children with overweight ate significantly faster than children with normal weight only when the mother was present in the laboratory, suggesting that parental influence could be an important mediator. Another study(Reference van den Boer, Kranendonk and van de Wiel127) in Dutch adults reported that, while a positive association was found between BMI and eating rate in both genders, the positive association between eating rate and waist circumference was observed only in females. Two other studies investigated the associations between measures of oral physiology and oral cavity volume, demonstrating faster eating rate among participants with a greater number of teeth(Reference Ketel, Zhang and Jia44), smaller oral cavity(Reference Ketel, Zhang and Jia44) and thicker tongue(Reference Ketel, de Wijk and de Graaf11).
Cognitive/psychological factors
Cognitive/psychological factors including individual differences in inhibitory control(Reference Fogel, McCrickerd and Goh7,Reference Shiozawa, Mototani and Suita21) , depressive symptoms(Reference Zhang, Yin and Cai137) and mindfulness(Reference Rogers, Drumgoole and Quinlan98) consistently demonstrated negative associations with eating rate, whereas psychological stress(Reference Nishitani, Sakakibara and Akiyama81) was positively associated with eating rate, all with small effect sizes. In two cross-sectional studies, one in a Japanese population(Reference Ohkuma, Fujii and Iwase94) and one in a multi-Asian population(Reference Teo, van Dam and Whitton45), addictive behaviours (such as smoking, alcohol intake) showed weak associations with rapid eating, while one retrospective longitudinal study(Reference Tanihara, Imatoh and Miyazaki82) conducted over an 8-year period failed to find statistically significant associations between addictive tendencies to exercising, smoking, habitual alcohol intake and eating rate.
The impact of portion size effect on eating rate has been examined in three studies(Reference Fogel, McCrickerd and Goh7,Reference Almiron-Roig, Tsiountsioura and Lewis77,Reference Wilkinson, Ferriday and Bosworth138) , and results consistently indicated that greater meal sizes were associated with faster eating rate, with two of the studies showing a medium effect size. However, rather than a linear relationship between portion size and eating rate, one study(Reference Almiron-Roig, Tsiountsioura and Lewis77) in women with overweight reported a threshold, approximately 15% greater than a reference portion size, beyond which eating rate began to decrease. Wilkinson et al.(Reference Wilkinson, Ferriday and Bosworth138) explored the influence of the ongoing perceptual volume of food remaining (rather than the actual volume of food at the beginning) on eating rate. This interesting study showed that eating rate was faster when participants saw a small portion of soup (300 ml) but actually consumed a large portion (500 ml), compared with when participants saw 500 ml but in fact consumed a 300 ml food portion. Therefore, the eating rate may not be solely influenced by the portion size at the beginning of the meal, but also by the perceived portions of food available. However, this relationship did not persist when comparing the eating rate of custard with large portion size (500 ml) and small portion size (300 ml), indicating that the effect of portion size on eating rate is moderated by the food type and palatability, highlighting the complexity of these associations.
Social factors
Eleven studies (11%) investigated the relationships between eating rate and social factors such as caregiver feeding practices(Reference Berkowitz, Moore and Faith52,Reference Costa, Severo and Oliveira53) , family structure(Reference Potter, Gibson and Ferriday58), eating with companions(Reference Bell, Spruijt-Metz and Vega Yon46,Reference Rosenthal and McSweeney49,Reference Lumeng and Hillman65,Reference Hordern72,Reference Hermans, Lichtwarck-Aschoff and Bevelander76) , parental influence(Reference Laessle, Uhl and Lindel56), cultural and social eating norms(Reference Sun, Ranawana and Tan112) and socially determined time available for food consumption(Reference Buergel, Bergman and Knutson59).
In a longitudinal study(Reference Costa, Severo and Oliveira53) that assessed bidirectional relationships between parental controlling feeding practices and children’s eating rate, it was demonstrated that pressuring children to eat and monitoring of eating at age 4 years were positively associated with eating very slowly at age 7 years, and the same was observed in the opposite direction, with medium effect sizes. This association was not observed for other controlling feeding practices such as food restriction. Parental prompts to eat were also significantly associated with children’s eating rate (kcal/min) in another longitudinal study(Reference Berkowitz, Moore and Faith52) (with a large effect size), though bidirectional associations were not examined. Laessle et al.(Reference Laessle, Uhl and Lindel56) found that children with overweight were particularly susceptible to maternal presence during mealtime and increased eating speed when mothers were present in the room. Similarly, Potter et al. (Reference Potter, Gibson and Ferriday58) found that children with siblings tended to have a faster eating rate and consumed more food than children without siblings. During the sensitive period for development of eating behaviours, parental feeding practices and behaviours around the meal, as well as presence of other models including siblings, can influence the development of children’s habital eating rates. However, the behaviours of models are also, to some extent, a response to spontaneous eating behaviours of children that the models observe.
Contrary to studies conducted in children, in adults companionship during a meal was linked with a slower eating rate compared with eating alone(Reference Hordern72), though this has been investigated in only a single study. Three studies have investigated the phenomenon of behavioural mimicry, where people unconsciously change their eating patterns in social conditions (only one study(Reference Hermans, Lichtwarck-Aschoff and Bevelander76) provided data to compute an effect size). In the family eating context, participants (males and females) were more likely to speed up eating rate within 5 s following a bite by their eating partner(Reference Bell, Spruijt-Metz and Vega Yon46); this was, however, observed only for the subset of family members (nineteen of seventy-eight dyads). Among strangers, it was reported that participants (females) eating with a fast-eating confederate model consumed their lunch significantly faster than participants who ate with slow-eating models(Reference Rosenthal and McSweeney49). Gender differences were explored in experiment II in the same study(Reference Rosenthal and McSweeney49), showing that female subjects ate more crackers in 7 min if the model confederate was a male who ate more crackers, and they ate less if the model was a female or if the model was a male who ate fewer crackers. A crossover study(Reference Lumeng and Hillman65) which examined the influence of a group size (small group of three and large group of nine) on children’s eating behaviour found no associations between eating rate and group size.
Two studies investigated elements that can be attributed to cultural influences on eating rate demonstrating moderate(Reference Buergel, Bergman and Knutson59) to strong(Reference Sun, Ranawana and Tan112) relationships between eating speed and mealtime norms and practices(Reference Sun, Ranawana and Tan112) or socially guided time available for food consumption(Reference Buergel, Bergman and Knutson59). Lunch break times in schools as well as cultural/social practices with regard to lunch provisions affect time that pupils in schools have to consume food, which can further impact eating speed(Reference Buergel, Bergman and Knutson59). Furthermore, in Chinese culture it is typical to share food with others, as platters are always placed in the middle of the table and food is taken on plate individually, and this has been associated with a slower rate of consumption compared with consuming food mixed on the individual plates(Reference Suh and Jung78). Similarly, cultural practices can influence other eating behaviours for example use of cutlery. Chopsticks, a frequently used cutlery in Chinese culinary culture, are known to reduce eating rate compared with other eating methods(Reference Sun, Ranawana and Tan112), such as spoons commonly used in Western cultures and fingers commonly used in South Asian or African cultures.
Environmental factors
Food environment
Thirty-two out of 104 studies investigated the role of food environment on eating rate and four factors emerged: food processing approaches(Reference Ilic, Tomasevic and Djekic89,Reference Lu, Venn and Lu139–Reference Ilic, Tomasevic and Djekic142) , food characteristics including textural and mechanical properties(Reference Choy, Goh and Chatonidi18,Reference Gonzalez-Estanol, Libardi and Biasioli108,Reference Aguayo-Mendoza, Chatonidi and Piqueras-Fiszman110,Reference Ilic, Tomasevic and Djekic141–Reference Wee, Goh and Stieger143) , hardness(Reference Ketel, de Wijk and de Graaf11,Reference Choy, Goh and Chatonidi18,Reference Ketel, Aguayo-Mendoza and de Wijk22,Reference Ketel, Zhang and Jia44,Reference Aguayo-Mendoza, Santagiuliana and Ong70,Reference Teo, Lim and Goh95,Reference Aguayo-Mendoza, Ketel and van der Linden106,Reference Doyennette, Aguayo-Mendoza and Williamson109,Reference Bolhuis, Forde and Cheng118,Reference De Wijk, Kaneko and Dijksterhuis144) and viscosity(Reference Zhu, Hsu and Hollis84,Reference Wilkinson, Ferriday and Bosworth138,Reference McCrickerd, Lim and Leong145,Reference Oladiran, Emmambux and de Kock146) , mixtures of food with various mechanical matrices(Reference Aguayo-Mendoza, Santagiuliana and Ong70,Reference Aguayo-Mendoza, Chatonidi and Piqueras-Fiszman110,Reference Rosenthal and Philippe124) and food presentation(Reference Suh and Jung78,Reference Forde, van Kuijk and Thaler88,Reference van Eck, Wijne and Fogliano107,Reference van Eck, van Stratum and Achlada147) .
Seven studies investigated the association between food processing approaches and eating rate (medium-to-large effect size). Among these, five(Reference Ilic, Tomasevic and Djekic89,Reference Lu, Venn and Lu139–Reference Ilic, Tomasevic and Djekic142) focused on the impact of different cooking methods on altering the mechanical properties of food, which in turn contributed to the changes in eating rate. In general, studies have reported that complex cooking methods (e.g. grilling, steaming, boiling), as compared with simple cooking methods (e.g., raw or sous-vide) could significantly alter the innate structure of food, contributing to less consumption effort and a faster eating rate. This has been observed in various foods, including potatoes(Reference Ilic, Tomasevic and Djekic142), potherb mustard greens (Mizuna; Brassica rapa)(Reference Zhou, Yamanaka-Okumura and Seki140) and wild boar ham(Reference Ilic, Tomasevic and Djekic89). One study(Reference Lu, Venn and Lu139) found that the storage condition of rice affected the eating rate. Specifically, cold-stored parboiled rice was found to have a more ordered structure compared with freshly cooked medium-grain rice, resulting in a longer chewing time and slower eating rate. The degree of food processing and its positive association with eating rate was examined in two studies(Reference Teo, Lim and Goh95,Reference Forde, Mars and de Graaf96) . In particular, highly processed foods defined by NOVA classification (a framework that classified foods into four groups based on the level of processing) as ultra-processing foods (UPFs), on average, were found to be associated with faster eating rate than unprocessed or minimally processed foods(Reference Forde, Mars and de Graaf96). However, there was significant variability within each food category, suggesting that other factors (e.g. composition and texture) may also play a role in mediating the influence of processed foods on eating behaviour.
Six studies demonstrated that the physical–chemical, rheological and mechanical properties of food play a crucial role in influencing eating speed. In the study by Wee et al.(Reference Wee, Goh and Stieger143), across fifty-nine solid commercial food products in Asian and Western cuisines, higher springiness, chewiness and resilience of food were related to slower eating rate as they required more chews to form a swallowable bolus. This has been supported by other studies consistently showing medium-to-large effect sizes(Reference Choy, Goh and Chatonidi18,Reference Gonzalez-Estanol, Libardi and Biasioli108,Reference Aguayo-Mendoza, Chatonidi and Piqueras-Fiszman110,Reference Ilic, Tomasevic and Djekic141,Reference Ilic, Tomasevic and Djekic142) . Other textural properties such as hardness(Reference Ketel, de Wijk and de Graaf11,Reference Choy, Goh and Chatonidi18,Reference Ketel, Aguayo-Mendoza and de Wijk22,Reference Ketel, Zhang and Jia44,Reference Aguayo-Mendoza, Santagiuliana and Ong70,Reference Teo, Lim and Goh95,Reference Aguayo-Mendoza, Ketel and van der Linden106,Reference Doyennette, Aguayo-Mendoza and Williamson109,Reference Bolhuis, Forde and Cheng118,Reference De Wijk, Kaneko and Dijksterhuis144) and viscosity(Reference Zhu, Hsu and Hollis84,Reference Wilkinson, Ferriday and Bosworth138,Reference McCrickerd, Lim and Leong145,Reference Oladiran, Emmambux and de Kock146) were also quite consistently negatively associated with eating rate, with medium-to-large effect sizes. Specifically, chewable foods were consumed at the slowest rate, with the smallest bite size, the greatest chewing rate per bite and the longest consumption time, compared with drinkable and spoonable foods(Reference Aguayo-Mendoza, Ketel and van der Linden106). Additionally, stiffer solid foods and more viscous liquid and semi-solid foods were associated with slower eating rate. These results, however, were not consistent. For instance, in the study of Zijlstra et al.(Reference Zijlstra, Mars and Stafleu148) the eating rate of hard luncheon meat was slower than that of soft luncheon meat, but this was not observed for alternative protein food and candy.
Apart from directly manipulating the texture of a single food, six studies(Reference Aguayo-Mendoza, Santagiuliana and Ong70,Reference van Eck, Wijne and Fogliano107,Reference Aguayo-Mendoza, Chatonidi and Piqueras-Fiszman110,Reference Mosca, Torres and Slob119,Reference Rosenthal and Philippe124,Reference van Eck, van Stratum and Achlada147) investigated the associations between eating rate and composite foods, in which two or more foods are combined. It has been consistently suggested(Reference van Eck, Wijne and Fogliano107,Reference van Eck, van Stratum and Achlada147) that the inclusion of condiments such as mayonnaise or cheese, which are frequently consumed with carrier foods such as bread or crackers, can speed up eating rate. For example, carrots with mayonnaise were consumed faster than plain carrots, presumably due to the condiment’s lubricating effect on bolus particles, which reduces saliva incorporation and increases eating rate(Reference van Eck, Wijne and Fogliano107). Alternatively, particle addition to certain foods(Reference Aguayo-Mendoza, Santagiuliana and Ong70,Reference Aguayo-Mendoza, Chatonidi and Piqueras-Fiszman110,Reference Rosenthal and Philippe124) , including granola and yogurt, may result in complicated textural properties that require greater oral manipulation skills that consequently slow down eating rate. This(Reference Aguayo-Mendoza, Santagiuliana and Ong70) may be exacerbated by particle hardness and particle number, which can slow down the eating rate even further. Another study assessed(Reference Mosca, Torres and Slob119) the influence of particle size on eating rate and reported that yogurt with smaller granola particles (6 mm) was consumed slower than yogurts with large granola particles (12 mm). However, in the study that investigated candy particle size on oral behaviour(Reference Rosenthal and Philippe124), a single large candy led to a longer total oral processing time compared with eight small candy portions. The disparity is likely due to hard candy requiring not only chewing but also sucking to dissolve it with saliva. All the studies reported large effect sizes pointing to the saliency and consistency of this factor on eating rate.
In terms of food presentation, two studies have reported that consuming smaller food pieces can reduce the eating rate compared with the food served in larger sizes with a medium effect size. For instance, carrots presented in large cubes were found to cause lower mastication effort and require lower number of chews than carrots pre-cut into smaller pieces(Reference van Eck, Wijne and Fogliano107). Moreover, carrots cut in julienne shape with a higher surface area triggered slower eating rate in comparison with carrots presented in cubes(Reference van Eck, Wijne and Fogliano107). Similar effects were demonstrated in another study that investigated the shape of crackers on eating rate(Reference van Eck, van Stratum and Achlada147). Other factors that may influence eating rate, including the serving format of Korean foods (separately or together)(Reference Suh and Jung78) and the presentation of potatoes (mashed or whole)(Reference Forde, van Kuijk and Thaler88), were examined in only a single study and warrant further investigation. It should be noted that food presentation can also change the texture of the food even though physical–rheological properties of the food due to, for example, thermal processing are unaffected. Mashed potatoes will therefore have a different texture to whole potatoes despite being cooked in the same way.
Eating environment
Eating environment and eating rate were assessed in twelve studies. Three of these reported consistent associations between eating methods(Reference Sun, Ranawana and Tan112,Reference Bolhuis and Keast131,Reference Hogenkamp, Mars and Stafleu149) and eating rate, with two(Reference Sun, Ranawana and Tan112,Reference Bolhuis and Keast131) indicating a large effect size and one(Reference Hogenkamp, Mars and Stafleu149) indicating a small effect. Specifically, eating with chopsticks was slower than eating with fingers or spoons(Reference Sun, Ranawana and Tan112), and eating with a fork has also been demonstrated to be slower than eating with a spoon(Reference Bolhuis and Keast131). These differences may be driven by the maximum permitted carrying volume. For example, a small spoon (teaspoon) resulted in a smaller bite size, longer mealtime duration and slower eating rate compared with a larger spoon (dessert spoon)(Reference James, Maher and Biddle85). This trend was also observed for straws(Reference Lin, Lo and Liao150) (thin or thick) and chopsticks (long or short)(Reference Lin, Lin and Hung114), with large effect sizes. One recent study(Reference Smith and Dando151) examined the effect of different 3D-printed textured spoons on eating rate, but no significant associations with eating rate were found. Participants’ familiarity with certain eating tools, such as chopsticks, may also play a role pointing to the interaction between socio-cultural elements and physical eating environment.
The association between eating rate and eating location has not been reported consistently. One study(Reference De Wijk, Kaneko and Dijksterhuis144) found that, compared with laboratory settings, eating rate was slower when food was consumed in free-living conditions (a large effect size). Conversely, another study(Reference Hordern72) that explored the effect of test location (lab versus home) on sensory profile and eating behaviour found an opposite relationship, albeit with a small effect size.
The effect of auditory features of the eating environment on eating rate was examined in two studies. The findings showed that, compared with silence, listening to music while eating effectively prolonged the meal duration and thus modulated the eating speed (a large effect size)(Reference Mathiesen, Mielby and Byrne152). People also spent significantly more time eating when music was slower in tempo and with legato articulation. In the other study it was demonstrated that participants could slow down their eating rate without being aware they were doing so by just listening to their slower chewing sounds, although the effect size was small(Reference Chen and Yen153).
Other interesting findings include faster eating speed observed during lunch compared with other meals (breakfast, dinner, snack)(Reference Petty, Melanson and Greene20), with a large effect size. Additionally, the impact of media usage(Reference Van den Bulck and Eggermont47) on eating rate among secondary school children was examined, demonstrating that 25% of children self-reported eating the meal faster at least once a week to be able to watch TV or play a computer game.
General discussion
The objective of the review was to examine and synthesize the current evidence on factors associated with eating rate. This review provides a comprehensive summary of the identified factors and a narrative synthesis with discussion of their effect size. Overall, 104 eligible studies were analysed and 40 factors were extracted and mapped onto the SEM, following qualitative evaluation. The majority were factors pertaining to the individual characteristics (20/39), followed by those related to social factors (11/39) and other environmental factors (8/39). Evidence suggests that individual factors such as food oral processing behaviour, gender, measures of body weight, and environmental factors such as food physical, rheological and mechanical properties are among the most frequently investigated factors with the most consistent associations with eating rate. The overall findings are summarised in Fig. 3.
Looking at the individual factors associated with eating rate, food oral processing behaviours including fewer chews per mouthful(Reference Zhu and Hollis8,Reference Fogel, Goh and Fries43,Reference Canterini, Gaubil-Kaladjian and Vatin73,Reference Ferriday, Bosworth and Godinot74,Reference Gonzalez-Estanol, Libardi and Biasioli108,Reference Forde, Leong and Chia-Ming117,Reference Bolhuis, Forde and Cheng118) , a large bite size(Reference Ketel, de Wijk and de Graaf11,Reference Fogel, Goh and Fries43,Reference Ferriday, Bosworth and Godinot74,Reference James, Maher and Biddle85,Reference Hill and McCutcheon113,Reference Forde, Leong and Chia-Ming117–Reference Mosca, Torres and Slob119) and shorter oral exposure time(Reference Choy, Goh and Chatonidi18,Reference Gonzalez-Estanol, Libardi and Biasioli108,Reference Forde, Leong and Chia-Ming117) have consistently been identified as the primary drivers of rapid eating, with medium-to-large effect size.
Of the demographic factors considered, there was a consistent indication that eating rate decreases with age(Reference Ketel, de Wijk and de Graaf11,Reference Ketel, Aguayo-Mendoza and de Wijk22,Reference Ketel, Zhang and Jia44,Reference Bell, Spruijt-Metz and Vega Yon46,Reference Kudo, Asahi and Satoh51,Reference Aguayo-Mendoza, Santagiuliana and Ong70) , with particularly strong evidence for slower eating among the older adults compared with younger adults. This could be attributed to the degradation of oral functionalities associated with ageing(Reference Mioche, Bourdiol and Monier154,Reference Alsanei and Chen155) , such as lower biting force, less tongue pressure and fewer teeth, which may contribute to the longer consumption times and the increased number of chews before swallowing that the elderly take to compensate for their reduced mastication efficiency. Additionally, age-correlated anatomical alterations, including decrease in salivation(Reference Affoo, Foley and Garrick156) or oral cavity volume(Reference Ketel, Zhang and Jia44), might also impair the effectiveness of food oral processing behaviour, leading to a slower eating rate. Further research across the full age spectrum from childhood through adolescence and early, mid- and late adulthood is required to better understand age-related changes in masticatory efficiency and eating rate.
There was also a large body of evidence suggesting that males eat faster than females, though effect sizes varied considerably and several studies failed to find this association. This potential gender-based difference could be partially attributable to attitudes towards mealtime etiquette(Reference Rolls, Fedoroff and Guthrie157), impression management tactics(Reference Gilmore, Stevens, Harrell-Cook, Eder and Harris158) and oral physiology. For instance, research on consumption stereotypes(Reference Vartanian, Herman and Polivy159) demonstrated that individuals who consume smaller meals and healthy diets are more feminine, less masculine, more physically attractive and more moral. Additionally, males normally have higher muscle strength, larger oral cavity, bite force, and salivary flow rate, resulting in shorter chewing cycle duration, larger bite sizes and faster eating rate(Reference Pereira and Van der Bilt160). The discrepancies in the identified effect sizes may be related to the textures of food served across the identified studies, or the heterogeneity in the methodology and/or sample size. Ketel et al.(Reference Aguayo-Mendoza, Ketel and van der Linden106) reported gender differences in eating rate only for solid foods, with females eating chewable food at a slower rate than males, while that trend was not apparent for semi-solid and liquid foods.
Ethnic differences in eating rate have also been consistently reported(Reference Ketel, Aguayo-Mendoza and de Wijk22,Reference Fogel, Goh and Fries43,Reference Ketel, Zhang and Jia44) , albeit within a very limited number of ethnicities that were investigated (typically European and Asian). Asian Chinese participants tended to eat more slowly than other Asian ethnicities and Caucasian Dutch. This could be attributed to differences in their oral physiology (e.g. oral cavity volume)(Reference Ketel, Zhang and Jia44), cultural/social practices(Reference Sun, Ranawana and Tan112,Reference Bolhuis and Keast131) or other external factors such as different cutlery congruent with various cuisines(Reference Sun, Ranawana and Tan112).
Finally, one of the most researched and consistently reported individual factors was the association between faster eating and greater body weight(Reference Teo, van Dam and Whitton45,Reference Zhu and Hollis48,Reference Kudo, Asahi and Satoh51,Reference Berkowitz, Moore and Faith52,Reference Nakamura, Nakamura and Takashima54,Reference Llewellyn, van Jaarsveld and Boniface55,Reference Potter, Gibson and Ferriday58,Reference Slyper, Shenker and Israel60–Reference Yamagishi, Sairenchi and Sawada64,Reference Zeng, Cai and Ma66,Reference Ochiai, Shirasawa and Nanri69,Reference Okubo, Murakami and Masayasu71,Reference Hamada, Miyaji and Hayashi80–Reference Shin, Lim and Sung83,Reference Barkeling, Rossner and Sjoberg86,Reference Azrin, Kellen and Brooks93,Reference Ohkuma, Fujii and Iwase94,Reference Laessle, Lehrke and Duckers97,Reference Lee, Kim and Jang99,Reference Yamane, Ekuni and Mizutani105,Reference Hill and McCutcheon113,Reference Zhu, Haruyama and Muto126,Reference Fujii, Funakoshi and Maeda128–Reference Nagahama, Kurotani and Pham130,Reference Kang, Joo and Hong133) , in line with other systematic reviews in this area published previously(Reference Ohkuma, Hirakawa and Nakamura6,Reference Garcidueñas-Fimbres, Paz-Graniel and Nishi135,Reference Yuan, Liu and Liang136) . It should be noted that the directionality of this relationship is still under debate as it is at present unclear if higher body weight causes faster eating, or whether faster eating leads to greater body weight. Likely, this association is bidirectional(Reference Henry, Ponnalagu and Bi161).
Few studies examined cognitive/psychological factors associated with eating rate, including inhibition control(Reference Fogel, McCrickerd and Goh7,Reference Shiozawa, Mototani and Suita21) , depressive symptoms(Reference Zhang, Yin and Cai137), psychological stress(Reference Nishitani, Sakakibara and Akiyama81) and mindfulness(Reference Rogers, Drumgoole and Quinlan98), or eating habits such as snacking behaviour or diet irregularity, which were examined in single studies, making it difficult to draw any conclusions regarding the significance, direction or strength of this association. More consistently, it has been reported that larger food portions may be linked to faster eating(Reference Fogel, McCrickerd and Goh7,Reference Almiron-Roig, Tsiountsioura and Lewis77,Reference Wilkinson, Ferriday and Bosworth138) in both children and adults(Reference Fisher, Liu and Birch162–Reference Rolls, Roe and Meengs165). Several mechanisms may be involved including visual references for meal termination(Reference Linné, Barkeling and Rössner166), which provide indirect information to eaters regarding the amount consumed and large portions which stimulate increases in bite size(Reference Fisher, Rolls and Birch167). Further research is needed to gain a clearer understanding of how portion size impacts eating rate. This includes investigating whether the effect is due to inability to accurately judge the amount of food served and/or if it relates to specific alterations in oral processing behaviours (i.e. bite size).
Interestingly, a number of social factors emerged as a result of our synthesis, though the support for these associations is rather weak and is often based on a single study in this area. Nevertheless, this emerging evidence is interesting as it offers insight into specific mechanisms through which eating rate is learnt or acquired during early development and provides context for potential implementation of interventions to reduce eating rate across cultures and populations. Several social factors, including feeding practices(Reference Costa, Severo and Oliveira53) (i.e. pressure to eat, monitoring), family structure(Reference Potter, Gibson and Ferriday58) (i.e. birth order, parity) and socially guided time available to consume foods(Reference Buergel, Bergman and Knutson59), were investigated in single studies, with small-to-medium effect sizes. The influence of other overarching factors such as social and cultural norms(Reference Sun, Ranawana and Tan112,Reference Bolhuis and Keast131) can potentially help explain other individual-level factors such as previously discussed ethnic differences. The impact of social/cultural factors can also influence quantities or types of food consumed, using specific cutlery for certain meals, food-to-mealtime context incongruity or perception of the mealtime etiquette. Research on a wider variety of ethnical backgrounds and cultures is needed to better understand the impact of social/cultural norms on eating behaviours.
Unsurprisingly, of the environmental factors, food characteristics, such as textural and mechanical properties(Reference Choy, Goh and Chatonidi18,Reference Gonzalez-Estanol, Libardi and Biasioli108,Reference Aguayo-Mendoza, Chatonidi and Piqueras-Fiszman110,Reference Ilic, Tomasevic and Djekic141–Reference Wee, Goh and Stieger143) , hardness(Reference Ketel, de Wijk and de Graaf11,Reference Choy, Goh and Chatonidi18,Reference Ketel, Aguayo-Mendoza and de Wijk22,Reference Ketel, Zhang and Jia44,Reference Aguayo-Mendoza, Santagiuliana and Ong70,Reference Teo, Lim and Goh95,Reference Aguayo-Mendoza, Ketel and van der Linden106,Reference Doyennette, Aguayo-Mendoza and Williamson109,Reference Bolhuis, Forde and Cheng118,Reference De Wijk, Kaneko and Dijksterhuis144) and viscosity(Reference Zhu, Hsu and Hollis84,Reference Wilkinson, Ferriday and Bosworth138,Reference McCrickerd, Lim and Leong145,Reference Oladiran, Emmambux and de Kock146) of the food, were the most investigated factors associated with eating rate, with consistent medium-to-large effect sizes. Strong consistent(Reference Forde, Van Kuijk and Thaler168) correlations between instrumental texture properties and oral processing behaviours, which then impact eating rate, have been well established. Importantly, research shows that people adopt their eating behaviour in response to the structural properties of foods they consume. An adhesive, chewy, less lubricated, hard-textured food may enable smaller bites and more chewing, which will result in a slower eating style, a phenomenon that is evident when it comes to foods of liquid, semisolid and solid form(Reference Zijlstra, Mars and de Wijk169). Supported by the previous findings, food texture modification can be accomplished through various food processing (i.e. minimal or ultra-processing(Reference Teo, Lim and Goh95,Reference Forde, Mars and de Graaf96) ) and cooking approaches(Reference Ilic, Tomasevic and Djekic89,Reference Lu, Venn and Lu139–Reference Ilic, Tomasevic and Djekic142) (i.e. grilling, steaming), by destroying the innate structures of foods, which leads to the reformulation of food texture, or by changing food presentation methods (mashed versus whole potatoes). Aside from dramatic changes in the mechanical properties of food, changing food shape(Reference van Eck, Wijne and Fogliano107,Reference van Eck, van Stratum and Achlada147) to a high aspect ratio and large surface area (e.g. cutting vegetables into elongated, julienne particles) may also be an alternative method of prolonging mastication time and slowing down eating rate. Other approaches, such as adding condiments(Reference van Eck, Wijne and Fogliano107,Reference van Eck, van Stratum and Achlada147) or particles(Reference Aguayo-Mendoza, Santagiuliana and Ong70,Reference Aguayo-Mendoza, Chatonidi and Piqueras-Fiszman110,Reference Mosca, Torres and Slob119) to create a mixture of food with a variety of mechanical matrixes, demonstrated great efficacy in decreasing eating rate. Extrinsically introduced food texture manipulation could potentially alter other factors such as sensory perception and acceptability of food, thus impacting palatability(Reference Forde, Van Kuijk and Thaler168). Minor changes (i.e. food shape, addition of condiments or particles) that reduce the alterations to the eating experience may therefore be more desirable than dramatic changes to food texture. Further research is needed to determine whether food texture manipulation approaches could be applied across the foods and whether these would be acceptable to both producers and consumers. Several environmental attributes, such as food forms(Reference Suh and Jung78), meal types(Reference Petty, Melanson and Greene20) (i.e. breakfast, dinner and snack), auditory features(Reference Mathiesen, Mielby and Byrne152,Reference Chen and Yen153) and media usage(Reference Van den Bulck and Eggermont47) were examined in single studies, and as such, the findings, though promising, are still preliminary.
Given the complexity of eating rate, the SEM model was successful in outlining the various known factors associated with eating rate at different levels, differentiating between factors pertaining to the individual, the environment and the cultural–societal elements. Such presentation of individual factors can help steer future research efforts to explore the associated mechanisms or to study the effectiveness in manipulating eating rate in experimental designs. It is important to note that, although the factors here are listed separately for ease of navigation, they should be considered as inter-dependent. For example, food oral processing behaviours have been suggested to be highly modifiable and, as such, should be effective in slowing down or speeding up eating rate(Reference Forde, Van Kuijk and Thaler168,Reference Bolhuis and Forde170) . However, the effectiveness of such manipulation may vary depending on the individual differences such as age, gender, ethnicity or body weight, social factors including presence of siblings, duration of lunch breaks at school or work, cultural appropriateness of cutlery, and other environmental factors (e.g. media usage while eating).
Implication for future research
One interesting finding from this review was the heterogeneity in the definitions of eating rate and methods of measurement across the studies. To enhance replicability and the generalisability of the study findings, it is recommended that the definitions of eating rate and methods of measurement be more standardised across the studies, while accounting for the study design, population and objectives. For example, it is recommended that validated questionnaires be used for studies that rely on self-reports of eating rate, rather than self-reported eating rate based on a single question. Further research to establish and validate briefer versions of currently used questionnaires, appropriate for large population-based studies, is necessary. For laboratory-based studies that measure eating rate in grams per unit of time or kcal per unit of time, we recommend to provide both measures (e.g. g/min and kcal/min). It is worth noting that the discrepancies between self-reported eating rate and objectively measured eating rate have been extensively discussed(Reference Petty, Melanson and Greene20,Reference Fagerberg, Charmandari and Diou61,Reference Stubbs, O’Reilly and Whybrow123,Reference Woodward, Haszard and Worsfold171) . These findings suggest that, while self-reports may be a viable options for population level studies, smaller experimental studies should strive for more objective methods of measurement(Reference van den Boer, Kranendonk and van de Wiel127).
Despite its accuracy, manual video coding of eating microstructure is not cost-effective due to reliance on expert manpower and the time required, limiting its utility in real-life eating environments. Alternative methods such as automated video coding(Reference Lasschuijt, Brouwer-Brolsma and Mars172) may be more cost-effective and efficient; however, this method is prone to errors and may not be easily applied. Modern technological tools, such as audio sensors(Reference Chen, Cui and Yen173,Reference Papapanagiotou, Diou and Zhou174) for capturing chewing and swallowing behaviour, and inertial measurement unit (IMU) sensors(Reference Zhang, Zhao and Nguyen175–Reference Heydarian, Adam and Burrows177) for recognising eating gestures can help improve the objectivity of eating rate assessment, though these methods are in early phases of development and are not yet widely available. We have been working on developing a smart eating tool capable of measuring eating rate in children (work in progress), for whom video-coding of eating rate is more difficult than for adults due to more movement during the meal and less mature eating behaviour. Our technology(Reference Chen, Fennedy and Fogel178) also allows dynamic regulation of bite size during eating and showed promising early results among participants with healthy weight and with overweight.
Based on the factors identified in this systematic review, we suggest that, to optimise eating rate, interventions targeting eating rate should focus on modification of specific oral processing behaviours, that is, bite size, oral processing time and inter-bite interval. This can be achieved by manipulating the texture of food to promote smaller bites and/or more chewing, which has been shown to be effective in small laboratory-based studies(Reference Bolhuis, Forde and Cheng118). Whether translation of such methods to real life is feasible requires further evidence. Using 3D food printing technologies(Reference Zhang, Pandya and McClements179,Reference Mantihal, Kobun and Lee180) it is also possible to re-formulate the microstructure of food, though cost-effectiveness of this process today may be an important barrier. A challenge remains in understanding how to develop food products that reduce the potential for overconsumption of energy without compromising their sensory appeal(Reference Bolhuis and Forde170). Alternatively, or in addition, we can manipulate the type and size of eating cutlery to statically reduce the bite size, though sustainability and social acceptance of such methods may be problematic. It is also important to foster a supportive environment (in the school and workplace) where ample time is provided for eating.
Implementing effective interventions to improve eating behaviour is challenging due to its complex nature. Future interventions should prioritise identifying clear targets that are evidence-based, feasible and accessible for individuals and do not require a radical change in a person’s lifestyle. It is important to note that, to optimise the effectiveness, interventions should not be treated as one-size-fits-all; rather, they may need to be tailored to specific circumstances. For example, food shape control is not feasible for vulnerable populations such as the elderly. In this instance, lubrication of food to accommodate physical conditions as well as to stimulate faster eating and greater energy intakes may be more desirable.
Strengths and limitation
This review provides a comprehensive synthesis of factors associated with eating rate mapped onto the socio-ecological model. This review highlights well-established factors across the individual, environmental and social levels, as well as emerging evidence in this area that shows promising findings that require further replication and/or research. By embracing an ecological perspective, we discuss the links and interactions between different factors providing clear directions for future research. This study had several limitations. Due to the methodological heterogeneity in eating rate definition and assessment methods, population characteristics and study designs, a meta-analysis was not feasible and was not considered. Several of the identified factors were supported by evidence from single studies; thus, their discussion in the broader context was not possible due to insufficient evidence. In addition, conclusions regarding specific factors that were drawn from a small number of studies are not generalisable and will not apply to other populations or circumstances. The discussion was instead focused on the most prevalent and researched factors, which were briefly discussed in the context of this emerging evidence. Moreover, we were unable to discuss the directionality of some of the factors due to the correlational designs of many of the studies, and as such, these findings and recommendations for future research and interventions need to be approached with caution. Finally, many of the factors that were mapped onto SEM could belong to more than one category (e.g. food presentation and texture), but were assigned on the basis of subjective interpretation of the study team.
Conclusions
This review provided an evidence-based synthesis of factors associated with the eating rate from multidisciplinary studies and mapped them to the socio-ecological model framework in a systematic way. The study highlighted heterogeneity in the definitions of eating rate across the studies, as well as a need for more standardised methods of measurement, appropriate for specific study designs, populations and objectives. Overall, we identified forty different factors associated with eating rate that were subsequently mapped onto the socio-ecological model, across individual, social and environmental levels. We identified some well-explored factors, which consistently showed strong associations with eating rate (i.e. oral processing behaviour, food texture, age), as well as many less explored factors such as eating habits, cognitive/psychological factors, feeding practices, eating companions and eating methods. These less explored factors are not yet well understood and require further investigation, but they may complement the development of effective strategies for eating rate modification and can enhance our understanding of the mechanisms involved. Based on the findings of this review, we propose further directions for research in this space as well as recommendations for feasible and evidence-based interventions to optimise eating rate.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S0954422423000239
Financial support
This research is supported by the Ministry of Education, Singapore, under the Academic Research Fund Tier 1 (FY2022).
Competing interests
There are no conflicts of interest.
Open access
