FormalPara Key Points

According to the robust, meta-regression model, a maltodextrin-based carbohydrate oral rinse shows some evidence of improving exercise performance.

According to a conventional meta-analytic approach, rinsing a maltodextrin-based carbohydrate oral rinse for 10 s with a concentration between 6 and 6.5% is effective at improving exercise performance.

1 Introduction

In the 1920s, it was recognised that carbohydrates were a crucial fuel source for exercise [1] with improved exercise capacity and general exercise performance being linked to carbohydrate consumption [2]. In subsequent research, it was demonstrated that a glucose polymer solution improved exercise capacity during cycling time to exhaustion tests compared with a placebo solution [3, 4]. Similar findings indicated that time to fatigue was significantly longer after ingesting a glucose polymer solution and glucose infusion in comparison with a placebo [3]. Currently, it is standard practice for individuals to ingest carbohydrates prior to or during sustained high intensity or endurance exercise [5,6,7,8]. Based on literature by Burke et al. [9], the American College of Sports Medicine (ACSM) recommends the total daily intake of 5–7 g per kilogram per day of carbohydrates for a moderate daily exercise programme (i.e. ~ 1 h/day); which can be ingested either prior to exercise, during exercise or in recovery from a previous exercise session [9,10,11]. The ACSM details further recommendations for acute fuelling strategies for carbohydrate loading, pre-event fuelling and for during brief exercise (< 45 min), sustained high-intensity exercise (i.e. 45–75 min) and endurance exercise (1–2.5 h). During sustained high-intensity exercise, it is currently recommended that small amounts of carbohydrates be ingested (including mouth rinsing) for optimal carbohydrate intake, whereas during endurance exercise, it is recommended that 30–60 g of carbohydrates be ingested per hour [9, 10]. A potential disadvantage associated with carbohydrate ingestion, however, is the possible occurrence of gastrointestinal discomfort [6, 12,13,14,15], which can subsequently negatively affect exercise performance [16,17,18].

An alternative method to utilise carbohydrates during exercise is a carbohydrate oral rinse. Previous research indicates that a maltodextrin rinse comprising of a 6–6.4% maltodextrin-based solution [19,20,21] during moderate- to high-intensity exercise with a time span ranging from 30 to 75 min can facilitate improvements in exercise [6, 8]. The exact mechanism that facilitates improvements in exercise performance after a carbohydrate oral rinse remains unknown. It is proposed that alterations in exercise performance may be influenced by a ‘Central Governor’ mechanism to maintain homeostasis during exercise [22]. The ‘Central Governor’ is thought to modify power output through the use of afferent signals from peripheral physiological receptors and systems that detect changes in the external and internal environment [22]. Therefore, it could be interpreted that during exercise, the positive central responses to a carbohydrate oral rinse could possibly counteract the negative physical, metabolic and thermal afferent signals [23]. An alternative theory is that improved exercise performance is a result of enhanced brain activation in higher brain regions. It is thought that these higher brain regions link the corresponding cognitive, behavioural and emotional response and the gustatory pathways [24, 25]. Furthermore, these regions have been found to be activated by oral exposure to carbohydrates but not by non-nutritive sweeteners [19, 26, 27], which may assist in explaining the positive effects of carbohydrate rinsing on exercise performance.

Carter et al. [28] first investigated the effects of a carbohydrate oral rinse on performance during a cycling time trial. During the time trial, participants were instructed to complete a certain amount of work (kJ) as quickly as possible. This amount of work was based on a formula including each participant’s maximum power output value (Wmax) [28]. During the time trial either a 6.4% (w/v) maltodextrin or water (placebo) sample was rinsed in the mouth for 5 s prior to expectoration. With the carbohydrate oral rinse, performance time was significantly faster (2.9%) in comparison with the water rinse (placebo) [28]. Additionally, improvements in exercise performance after an oral carbohydrate rinse in comparison with a placebo rinse have been found with cycling [29,30,31,32,33], running [34,35,36,37] and resistance exercise [38]. In contrast, some studies have reported no significant improvements in exercise performance [39,40,41,42,43,44,45,46]. This lack of significant improvements may be due to the study design (i.e. mode of exercise, concentration and/or composition of the rinse and rinsing duration) or lack of statistical power to detect changes. Due to the inconsistent results in the pool of literature, further analysis is required to investigate if a maltodextrin-based oral rinse does improve exercise performance.

Furthermore, research in this area has also discussed a possible placebo effect in conjunction with carbohydrate oral rinsing. As previous research has demonstrated that placebo effects may have a significant impact on physical performance [47], it is common practice for at least two oral rinses to be trialled: a carbohydrate oral rinse and a placebo oral rinse [28, 29]. To minimise possible placebo effects between the rinsing conditions, previous research has also blinded participants to the composition of the rinses and also to the true objective of the experiment [45].

Previous reviews have focused on the effects of carbohydrate oral rinsing on exercise performance across running and cycling performance [48], cycling performance [49] and sprinting performance [50], and carbohydrate oral rinsing alongside ingestion and loading on exercise performance [51]. However, no reviews have specifically discussed the intricacies of the maltodextrin used in the carbohydrate oral rinse. Maltodextrin is a variable starch-based structure [52] that is a widely used product in foods and food manufacturing [53]. Maltodextrin can vary depending on its physical and chemical properties, which can in turn affect the overall flavour and appearance [54]. Additionally, as maltodextrin can vary widely in terms of structure and origin, these may be important factors to investigate as this variation may impact exercise performance. For example, starches are composed of two types of glucose polymers: amylose and amylopectin [55]. The ratio between amylose and amylopectin can affect the physical properties of starches including their retrogradation tendencies, viscosity and pasting properties [56,57,58]. Other important structural factors include dextrose equivalent (DE) and degree of polymerisation (DP). For example, a shorter-chain maltodextrin has a higher DE and lower DP and therefore has a sweeter taste in comparison with a longer-chain maltodextrin [54]. The origin of the maltodextrin may also be an important factor to consider as maltodextrin can be made from corn, rice, manioc, oat or potato starch [59]. Depending on the source of the starch, the ratio of amylose to amylopectin changes. For example, high-amylose corn starch has an amylose content of 50–70%, whereas potato, tapioca and wheat starches have an amylose content close to 20% [58]. Without reporting on the type of maltodextrin used in the carbohydrate oral rinse, the information concerning origin and structure is unknown, potentially prohibiting informed observations and mechanistic insights. Furthermore, oral rinse protocol (concentration and duration) is an important factor to investigate as dose response or time/duration response with a carbohydrate oral rinse and exercise performance response may exist. Exercise protocol, fasting and participant characteristics are additional factors that can also vary across the literature and are important to investigate as there may be an optimum level or conditions at which exercise performance can be improved. The primary aim of this systematic review and meta-analysis is to comprehensively examine the isolated effect of maltodextrin-based rinsing on exercise performance. Furthermore, the secondary aim of this review is to investigate the effect of the concentration and composition of the rinse, duration of the rinse and the impact of participant characteristics (i.e. sex), fasting and exercise protocol on exercise performance.

2 Methods

The systematic review and meta-analysis was completed according the Cochrane Handbook for Systematic Reviews of Interventions [60] and following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement [61].

2.1 Eligibility Criteria

For this systematic review and meta-analysis, studies were included if they met the following inclusion criteria: (1) investigated the effect of a maltodextrin oral rinse on exercise performance; (2) randomised, blinded counterbalanced or crossover, control or placebo study design; (3) a maltodextrin-based carbohydrate oral rinse with a concentration of no less than 6% being rinsed for a minimum of 5 s (based on existing literature in the area [28, 49]); (4) original research articles and (5) human participants. Studies were excluded if they (1) did not use maltodextrin in the oral rinse; (2) involved any ingestion of the maltodextrin rinse; (3) were not original research articles (i.e. conference abstracts, review articles); (4) were not written in English and (5) did not have sufficient methodological information to allow a check of the inclusion criteria. The search included articles that were published up to and including February 2021.

2.2 Data Sources and Search

Initially, a small test search using the chosen search terms was conducted to determine the efficacy of the search terms. After the test search confirmed the search strategy was effective, the search terms were finalised. Five separate databases (MEDLINE, PsycINFO, Embase, SPORTDiscus and Global Health) were searched and this was performed initially in June 2020 and updated in March 2021. The following terms were used in the search: Carbohydrate OR CHO OR Carbohydrates AND Oral Rinse OR Oral Rinsing OR Mouth Rinsing OR Mouth Rinse OR Rinsing OR Rinse OR Mouth Wash AND Exercise Performance OR Performance OR Exercise OR Cycling OR Running OR Sprinting OR Resistance.

This search strategy yielded 527 publications. The hand searching technique, which involved searching reference lists of included studies and review articles for relevant studies, found a further ten studies. From the pool of included studies, 288 duplicates were removed which resulted in 239 studies for screening. After screening, a total of 35 studies were included for data extraction and analysis for the systematic review and of these 35 studies, a total of 34 studies were included in the meta-analysis. No relevant studies were found within grey literature. The process of study selection and screening is summarised in a PRISMA flow diagram (Fig. 1).

Fig. 1
figure 1

PRISMA flow diagram [62]

Full size image

2.3 Study Selection and Data Collection

The studies were imported into a systematic review management tool (Covidence, Veritas Health Innovation, Victoria, Australia) to complete the screening process. The titles, abstracts and full texts of the studies were reviewed separately to check for eligibility criteria. The screening process for abstract screening and full-text screening was completed by three separate reviewers. Screening conflicts were resolved through discussion between the three independent reviewers until a conclusion was reached. Data extraction for the included studies focused on (1) title, year of publication, type of publication and (2) methods and design of the study, participants selected (sample size, sex and age of participants), intervention (composition and concentration of oral rinse), type of exercise performed and outcome measures (i.e. distance covered in the trial [m], time to complete the trial [s], power output [W], speed [km/h]). As the majority of studies reported results as mean ± standard deviation, studies that reported mean ± standard error were converted to mean ± standard deviation for consistency. If additional information was required from a study for data extraction, the corresponding author was contacted. On the occasion where this further information was unable to be obtained, that study was excluded from the meta-analysis.

2.4 Quality Assessment

This quality assessment was completed using a modified version of the Quality Criteria Checklist: Primary Research [63] and previously published formatting [64]. One author (C.H.) performed the quality assessment and a second author (R.S.J.K.) independently cross-checked the quality assessment. Any disagreements were discussed and resolved between the two authors.

2.5 Risk of Bias Assessment

A risk of bias assessment was conducted with the studies that were included in the meta-analysis according to the Cochrane Collaboration’s recommendation for systematic reviews [65]. The categories for assessment included (1) random sequence generation (selection bias); (2) blinding of participants and personnel (performance bias); (3) blinding of outcome assessment (detection bias); (4) incomplete outcome data (attrition bias) and (5) selective reporting (reporting bias). Each category was assessed and assigned either a low risk of bias, high risk of bias or unclear risk of bias. One author (C.H.) performed the risk of bias assessment and a second author (R.S.J.K.) independently cross-checked the risk of bias assessment. Any disagreements were discussed and resolved between the two authors.

2.6 Statistical Analysis

The data extraction process for the meta-analysis focused on primary performance-based outcomes (refer to Table 1 for a complete list). Outcomes that were not deemed to be performance based were excluded from the meta-analysis data set. Studies with multiple results for a single performance outcome (i.e. maximal speed: sprint 1, sprint 2, sprint 3 etc.) were collapsed and averaged together prior to the meta-analysis. From the 34 included articles, the data collection process resulted in 58 data points for analysis.

Table 1 Studies investigating the effect of a maltodextrin-based carbohydrate oral rinse on exercise performance
Full size table

2.6.1 Conventional Meta-Analytic Method—Standardised Mean Differences (SMD)

The effects of oral rinsing were analysed in terms of means and standard deviations comparing CHO and placebo (PLA) treatments at the end of the study time. Thus, the standardised mean difference (SMD) was calculated using the Hedges’ g method for each individual effect (CHO vs PLA) reported in each study. The Hedges’ g method was adjusted using exact computation for the bias-correction factor and Hedges and Oikin were used for standard error for each individual effect size. Conventional meta-analytic techniques rely on the assumption that effect size estimates from different studies are independent and have sampling distributions with known conditional variances [66,

Fig. 2
figure 2

Forest plot comparing the effects of the moderators—sex, rinse concentration (%), rinse duration (s) and fasting—on carbohydrate oral rinsing in comparison with a placebo condition. This forest plot was performed with a conventional random-effects REML (restricted maximum likelihood) model

Full size image

3.3.2 Oral Rinse Protocols—Rinsing Duration

For the individual groups of rinsing  for 5 s (n = 17) (SMD = 0.01, 95% CI − 0.19, 0.21; p = 0.932), 7.5 s (n = 1) (SMD = 0.28, 95% CI − 0.47, 1.03; p = 0.461), 12.5 s (n = 3) (SMD = − 0.07, 95% CI − 0.53, 0.40; p = 0.774), 20 s (n = 2) (SMD = 1.12, 95% CI − 1.19, 3.42; p = 0.342) and 40 s (n = 1) (SMD = − 0.06, 95% CI − 0.78, 0.66; p = 0.864), the mean effect size was not statistically significant (Fig. 2). For the individual group of rinsing for  10 s (n = 34) (SMD = 0.22, 95% CI 0.05, 0.39; p = 0.013), the mean effect size for this group was statistically significant at the 5% level. For this analysis, articles that provided a range for rinsing duration (e.g. 10–15 s), a middle point between the range was used for the analysis. Furthermore, for articles that provided an approximate rinsing time (i.e. ~ 5 s), a value of 4.9 s was used in the analysis.

3.3.3 Oral Rinse Protocols—Rinse Concentration

For the individual groups of rinse concentration of 6–6.5% (group 1) and 8–18% (group 2), the mean effect size for group 1 (n = 46) (SMD = 0.15, 95% CI 0.02, 0.29; p = 0.027) was statistically significant at the 5% level while the mean effect size for group 2 (n = 12) (SMD = 0.16, 95% CI − 0.07, 0.39; p = 0.167) was not statistically significant at the 5% level (Fig. 2). The variable of ungrouped, individual rinse concentrations was examined in a continuous format which is available in Online Resource 2 (see ESM).

3.3.4 Exercise Protocol

For the individual groups of arm cranking (n = 1) (SMD = 0.27, 95% CI − 0.51, 1.04), cycling (n = 18) (SMD = 0.07, 95% CI − 0.13, 0.26), isometric contractions (n = 4) (SMD = 0.58, 95% CI − 0.49, 1.67), resistance exercise (n = 15) (SMD = 0.15, 95% CI − 0.04, 0.35), running (n = 17) (SMD = 0.22, 95% CI − 0.12, 0.55), maximum voluntary contractions (n = 1) (SMD = − 0.02, 95% CI − 0.77, 0.72) and countermovement vertical jump (n = 2) (SMD = 0.01, 95% CI − 0.55, 0.57), the mean effect size for these groups was not statistically significant at the 5% level. The isometric contractions group and running group show some unexplained between-study heterogeneity with estimated I2 of 86.16% and 64.84%, respectively.

3.3.5 Fasting

For the individual groups of no fasting (n = 20) (SMD = 0.15, 95% CI − 0.03, 0.33; p = 0.110), fasting for < 5 h (n = 13) (SMD = 0.20, 95% CI − 0.02, 0.42; p = 0.070) and fasting for > 5 h (n = 25) (SMD = 0.15, 95% CI − 0.08, 0.38; p = 0.211), the mean effect size for these groups was not statistically significant at the 5% level (Fig. 2). The group of fasting for > 5 h shows some unexplained between-study heterogeneity with estimated I2 of 71.05%.

3.4 Subgroup Analysis Results—Meta-Regression Model with Robust Variance Estimation

3.4.1 Characterisation of Participants

There was no significant difference at the 5% level when comparing the individual group of male participants with female participants (difference between SMDs = 0.17, 95% CI − 2.75, 3.08; p = 0.68) or the combination of male and female participants (difference between SMDs = 0.42, 95% CI − 0.98, 1.82; p = 0.48). All data points (n = 58) were included in this analysis.

3.4.2 Oral Rinse Protocols—Rinsing Duration

There was no significant difference at the 5% level when comparing the individual group of rinsing for 5 s with rinsing for 10 s (difference between SMDs = 0.19, 95% CI − 0.07, 0.45; p = 0.15). In a sensitivity analysis, only 51 data points were included in the meta-regression analysis as small sample adjustments could not be done with groups with fewer than four.

3.4.3 Oral Rinse Protocols—Rinse Concentration

There was no significant difference at the 5% level with group 1 (6–6.5%) compared with group 2 (8–18%) (difference between SMDs = − 0.07, 95% CI − 0.36, 0.22; p = 0.58). All data points were included in this analysis.

3.4.4 Exercise Protocol

There was no significant difference at the 5% level when comparing the individual group of cycling with isometric contractions (difference between SMDs = 0.52, 95% CI − 1.15, 2.19; p = 0.42), resistance exercise (difference between SMDs = 0.09, 95% CI − 0.30, 0.48; p = 0.62) or running (difference between SMDs = 0.17, 95% CI − 0.35, 0.69; p = 0.49). In a sensitivity analysis, only 54 data points were included in the meta-regression analysis as small sample adjustments could not be done with groups with fewer than four.

3.4.5 Fasting

There was no significant difference at the 5% level when comparing the individual group of no fasting with fasting group 1 (< 5 h) (difference between SMDs = − 0.04, 95% CI − 0.42, 0.33; p = 0.81) or fasting group 2 (> 5 h) (difference between SMDs = − 0.11, 95% CI − 0.55, 0.34; p = 0.62). All data points were included in this analysis.

3.5 Risk of Bias Assessment Results

Overall, of the 34 total studies (58 data points) included in this meta-analysis, all demonstrated a high level of evidence. The majority of studies reported using random sequence generation (n = 28) and all studies scored a low risk of bias in the categories of incomplete data (n = 34) and selective reporting (n = 34). A proportion of studies had a single-blinded study design (n = 10) and therefore scored an unclear risk of bias for that category. Furthermore, for the category of detection bias, all studies (n = 34) scored an unclear risk of bias. Results of the risk of bias assessment of the studies is presented in Online Resource 3 (see ESM).