Journal Logo

APPLIED SCIENCES

Multiple-Transportable Carbohydrate Effect on Long-Distance Triathlon Performance

ROWLANDS, DAVID S.; HOULTHAM, STUART D.

Author Information
Medicine & Science in Sports & Exercise: August 2017 - Volume 49 - Issue 8 - p 1734-1744
doi: 10.1249/MSS.0000000000001278

Abstract

Practical and empirical observations since the 1920s provide convincing evidence that exogenous carbohydrate ingestion during prolonged intense exercise is ergogenic (17,37). More recent data suggest that further improvement to performance is likely when the ingested carbohydrate is a blend of fructose and glucose, characterized by the property of accessing multiple hexose transport routes across the gastrointestinal epithelia (17,29). Indeed, fructose–glucose (glucose polymer, maltodextrin) composite beverages improved gut comfort, exogenous carbohydrate oxidation rates, and endurance performance, relative to isoenergetic single saccharide (29).

The mechanisms behind improved gut comfort and oxidation rates with fructose–glucose composites probably include faster gastric emptying (18) and mucosal monosaccharide absorption (33), the latter via access to multiple hexose transport routes (i.e., multiple-transportable carbohydrate): sodium-dependent glucose cotransporter 1 for glucose and GLUT5 for fructose (29). Additional synergistic transport of fructose and glucose (31,33) may involve increased membrane recruitment of the facilitated transporter GLUT2 (21), and other intestinal epithelial transporters (e.g., GLUT8 and GLUT12 [4]). Faster intestinal absorption of multiple-transportable carbohydrates may lower gut discomfort because of clearance of residual carbohydrate, which along with increased exogenous carbohydrate availability are the two most likely mechanisms responsible for improved endurance performance (29,33).

All previous experiments examining the efficacy of multiple- versus single-transportable carbohydrate on performance used beverages. However, many endurance athletes ingest carbohydrate during competition in the form of solids (bars, fruit, and bakery items), gels, and beverages (24), a practice that has gained rapid and enthusiastic support from the commercial market. However, limited empirical contrasts of the effect of carbohydrate format suggest that solid and gels may harm performance, relative to drink (5,32), or have negligible impact (2). Accordingly, translational evidence is required to clarify if multiple-transportable carbohydrate ingested in differing formats preferred by endurance athletes is ergogenic in competition.

Therefore, the objective of the current trial was to determine whether the ingestion bar, gel, and beverages containing 2:1 glucose/maltodextrin–fructose would produce an adoption-worthy enhancement to performance during half ironman triathlon competition, relative to the ingestion of standard isocaloric carbohydrate (glucose/maltodextrin only) in double-blind conditions. Carbohydrate was ingested at apparently optimal rates (34) within a large sample of age-heterogeneous well-trained experienced male triathletes.

MATERIALS AND METHODS

Experimental Design

The study design was a double-blind, randomized, crossover comprising competition within two half Ironman triathlons spaced 3 wk apart (Figs. 1 and 2). The study took place in New Zealand in midsummer. The first race was the Taupo Half Ironman (Taupo; http://www.halfironmantaupo.co.nz/) held on December 14, 2013; the second event was the Tauranga Half Ironman (Tauranga; http://mountfestival.kiwi/events/port-of-tauranga-half/) held on January 4, 2014. Each half ironman comprised a 1.9-km swim, a 90-km bike, and a 21.1-km run. Both events were sanctioned and run under the race rules of the national body (Triathlon New Zealand; http://triathlon.kiwi/events/tri-nz-race-rules.html).

F1-27
FIGURE 1:
Consort style flow chart summarizing the study design and progress of the participants through the trial. Shown are number of participants (parentheses) reviewed, enrolled, and transitioning through each component of the study. Also shown within each arm of the study are the number of participants and dropout number and reasons in the full analysis data set (FAS) and PP data set, respectively. Participants who experienced incidents likely to affect race time were listed in the PP variation. Participant numbers under race analysis refer to sample size available for analysis of final time; sample size available for analysis of triathlon leg is provided in Table 3. DNF, did not finish.
F2-27
FIGURE 2:
Experimental design. Shown is the recruitment period, three familiarization sessions, and the period of 5-d standardized diet and training before the controlled prerace diet at the race venue from 24 h before the event, ending with race day. The period reproduced within the crossover design is shown.

Participants and centers

Recruitment occurred between September 1 and November 15, 2013, from the New Zealand triathlon community via e-mail or word of mouth from the Triathlon New Zealand membership, affiliated clubs, and event databases. Inclusion criteria were as follows: male 18–60 yr of age, a history of competitive triathlon performance ≤2 yr, participation in at least four competitive triathlon events within the past 2 yr, and a personal best half-ironman time corresponding to ≤5 h 40 min for age 18–40 yr, ≤5 h 50 min for age 40–50 yr, or ≤6 h 05 min for age 50–60 yr. Exclusion criteria were as follows: failure to meet health requirements (general health questionnaire), prior known fructose or wheat intolerance, precluding food or packaging-related allergies, an existing injury that may affect ability to complete the races, incorrect execution of performance and questionnaire assessment, poor tolerance to the treatment during the familiarization sessions, and medications that may interfere with the study outcomes. Females were excluded because of additional variability to performance introduced by menstrual cycle phase. Triathletes who meet the criteria provided written consent according to the protocol approved by the Massey University Human Ethics Committee (SA: 13/46). Once the study commenced, participant withdrawal criteria included medical reasons, noncompliance with treatment or competition protocol, and injury affecting performance (Fig. 1). Participants that were withdrawn were not replaced.

Protocols

Familiarization trials

Participants undertook three familiarization sessions between 3 and 6 wk before the first triathlon at their home location (Fig. 2). The sessions were a 2-h bike immediately followed by a 1-h run while consuming the intervention supplement at the same carbohydrate (1.4 g·min−1) and fluid (700 mL·h−1) intake rate and ratio (25% bar, 35% gel, 40% drink) as would be experienced in the experimental protocol. Immediately after each session, athletes completed a compliance sheet and the gastrointestinal comfort questionnaire that would be used during the experiment, which were returned to the researchers.

Training and diet

During the screening process, participants completed a survey capturing training history, including years spent in the sport; racing history and best triathlon performance times; a qualitative estimate of average training time swimming, cycling, running, and other per week; and history of gastrointestinal distress. In the week before the first race, participants recorded all swimming, cycling, and running training duration spent in low, moderate, and high training zones undertaken in the period beginning 5 d before the first triathlon via a qualitative survey sheet with intensity categorized by perceive exertion, and session specifics were noted (e.g., speed, power, and number of intervals) (Fig. 2). Participants were instructed to replicate this training in the same period before the second half Ironman. Participants were lodged in hotel accommodation to control prerace dietary intake and travel stress and to coordinate race logistics. Accommodation, travel, and race entry costs were incorporated within the research design.

Participants were authorized to eat as usual and to take their usual medications except the controlled diet the day before racing, where all meals, drinks, and snack items were catered to participants (Fig. 2). Participants were instructed to ingest food and beverages providing at least 8.5 g of carbohydrate per kilogram body weight and at least 50 mL·kg−1 body weight of fluid. The purpose of the controlled diet was to obtain high muscle and liver glycogen stores and to tightly control prerace dietary conditions. Ingested items were recorded via a serving-size method record booklet that included serving size carbohydrate weight and fluid volume information. Participants could choose their food and drink items from a range of catered or prepackaged options. No other food or drink items were permitted. All servings were homogeneous and weighed. Items and servings on a diet record sheet were replicated on the day before the second triathlon.

Experimental Trial

After awakening and toileting, participants reported to the study headquarters between 04:00 and 04:40 h of race day, for pre-event body weight measurement (Tanita BC532, Tanita Health Equipment H.K. Ltd., Hong Kong, China), supplement condition body-color marking, transition supplement pickup, and breakfast. Breakfast was ingested between 04:40 and 05:10 h and consisted of 700 mL of 7% carbohydrate sports drink and a study bar specific to the random allocation of treatment. Together, the preexercise drink and bar contained a total of 94 g of carbohydrate. To standardize the effect of caffeine on performance, two 1.8-g sachets of instant coffee (Nescafe, Nestle, Auckland, New Zealand) were mixed with hot water and consumed with breakfast. Participants then traveled to the event village to load bottles containing the experimental supplement into the respective bottle holders in the race bicycle and running belts. The cohort competed with all other athletes entered who were not part of the experiment under the race rules and conditions set out by the event organizers. Participants were instructed to stop only at the experimental feed stations and to adhere strictly to the study feeding protocol and event regulations. Air temperature and relative humidity were obtained from local weather stations (Metservice, Wellington, New Zealand).

Nutrition Intervention

The carbohydrate content of the intervention supplements was glucose/maltodextrin–fructose (GLUFRU) in a 2:1 ratio. The control supplements were formulated from glucose/maltodextrin only (GLU). We used a 2:1 supplement ratio to permit comparisons when possible with previous laboratory and field-based research investigating multiple-transportable carbohydrate on metabolism and performance (1,14–16,18,28,30,36,39).

The intervention supplements were adaptations of commercially available solid bar, semiliquid gel, and drink powder products that were prepared by the study funder. The ingredients in the bar were fructose–glucose syrup, maltodextrin, oat flakes, milk protein, rice crisps, fruit preparation (fructose syrup, lactose, apple, gelling agent pectin, flavoring, vegetable fat, and citric acid), sodium citrate, vegetable oil, magnesium carbonate, citric acid, flavoring, and salt. The gel was 65% carbohydrate and comprised fruit juice concentrates, maltodextrin, water, fructose, sodium citrate, sodium chloride, preservatives (sodium benzoate and potassium sorbate), and flavoring. The drink mix powder contained glucose, fructose, maltodextrin, minerals (sodium citrates, sodium chloride, calcium gluconate, potassium chloride, magnesium citrates, and calcium lactate), citric acid, flavoring, safflower concentrate, and anticaking agent (silicon dioxide). The control glucose/maltodextrin bar, gel, and drink were isocaloric produced in matching formats, flavors, and texture in which the total fructose component was replaced by glucose. All supplements were stored at room temperature. Bars and gels were plain packed and identified by a unique code specific to the GLUFRU or GLU formulation, with the identity of the code held by third party.

Blinding technique, randomization, and coding

Double-blind condition was achieved by using a four-color code randomization system. The colors were green, blue, yellow, and black. The color sequences were randomly allocated (even weighting) to participant number before the first race by a third party (Nestle Research Centre) independent to the experimental processes and contact with the participants. The color sequence was rerandomized for the second race. Allocation of treatment code to the correctly randomized color was checked and signed off by a third party (at research site) to ensure correct treatment allocation. Code break occurred before statistical analysis on June 6, 2014, after evaluation of dropouts, athletes who did not finish, time adjustments, and locking of database.

Feeding protocol

Participants ingested sports drinks, water, sports bars, and sports gels according a schedule based on per unit distance covered between feed stations. The proportional quantity of carbohydrate from 7% carbohydrate beverages (40%), bars (25%), and gels (35%) and the rate of total carbohydrate ingestion were established based approximately on the observed supplement intake during a previous recent half ironman competition (24) and 84 g·h−1 (1.4 g·min−1) that was within the upper range for an estimate of the mean optimal total carbohydrate ingestion rate for performance of 78 g·h−1 (95% confidence interval [CI] = 68–88 g·h−1) (34) and that used in our previous field trial (30). Total fluid intake was 700 mL·h−1 chosen from previous laboratory and field studies as a quantity likely to limit body weight loss less than 3%, which was unlikely to impair performance (24,30,38). The total carbohydrate (392.5 g) and fluid to be ingested (2927 mL) was subsequently calculated from the consumption rate and the average finishing time of the present experimental cohort's last three previous half ironman event swim (37 min 51 s), bike (2 h 30 min 4 s), run (1 h 40 min 50 s), and final (4 h 48 min 45 s) times. The total fluid was divided evenly between a 7% carbohydrate beverage (2007 mL) and water (920 mL), with both the carbohydrate beverage and water ingested throughout the triathlons in a 2.18:1 ratio.

Feed stations and schedule

Supplement ingestion during the triathlons was controlled via treatment-color code allocated collections from six feed station points summarized in Supplementary Digital Content (see Table, Supplemental Digital Content 1, Feed-station location and the quantity of carbohydrate supplement collected at each station for the two half-ironman triathlons, https://links.lww.com/MSS/A905). The first station was in the swim–bike transition area, where participants collected the bar and gel units, with 2 × 700-mL drink bottles loaded on to two frame-mounted bottle holders. The second station was positioned at approximately three-quarter distance into the cycle leg. The third station was the bike–run transition area, where participants collected their preloaded run belts and bar portion. On the run, fuel belts (Profile Design, Long Beach, CA) capable of holding 2 × 225-mL drink bottles were provided to athletes. The final three feed stations were positioned in (traffic management) safe areas evenly spaced throughout the run course. Research assistants wore colored T-shirts to assist in correct collection of allocated supplements. The feed station locations and design are shown in Supplemental Digital Content (see Figure, Supplemental Digital Content 2, Study feed station locations and design, https://links.lww.com/MSS/A906).

Physical location and safety logistics resulted in minor differences in feed station position on the course between each race. Accordingly, fluid volumes were adjusted to standardize the rate of carbohydrate and fluid delivery between each station (see Table, Supplemental Digital Content 1, Feed-station location and the quantity of carbohydrate supplement collected at each station for the two half-ironman triathlons, https://links.lww.com/MSS/A905). Precise control of ingestion rate between feed stations was not possible with the practical boundaries of the current field study. However, participants were instructed to ingest supplement on a regular basis between feed stations and intervals of 10 min.

Compliance monitoring

All supplement ingestion was monitored during both races via inspection and measurement of residual. Breakfast bar wrappers and drink bottles were individually coded and collected upon completion as participants vacated the dining area. Bike transition supplementation was individually coded (two drink bottles, one bar, and two gel packages). Before arrival at all feed stations, the athletes were instructed to have consumed all of the supplement units and volume they were carrying from the prior collection point, which was verified verbally by the participant and by the researchers via visual inspection before handing over the replacement supplements. Drink bottles and wrappers collected at the bike feed station remained with the bike in the transition area, where compliance was checked off after the race (including quantitative measuring of any residual volume). Supplement units ingested between the run transition and the first run feed station were individually labeled and checked for completion at the first run feed station before new supplement was provided. The gels collected at the first, second, and third run feed station were consumed within 500 m. At 500 m after the feed station, the wrapper was collected within a designated discard area by the researchers for compliance recording. At the finish line, fluid collected at the third run feed station was checked for completion immediately after crossing the finish line.

Postrace Questionnaires

Upon completion of the race, participants were chaperoned to the toilet then to the study tent for measurement of postrace body weight and completion of the psychometric survey. A 0–100 Likert scale was used in the postrace survey to assess gut comfort, nausea, and energy level during each leg of the triathlon.

Data Collection, Management, and Validation

All data were checked for accuracy then loaded into a secured Internet-based database. All computerized data were edit checked for any discrepancies by a third party, and amendments made where necessary by the researchers before secondary check and sign off by an on-site third party researcher.

Statistical Methods

Sample size

To have a realistic estimate of the within-subject SD and the effect size of GLUFRU ingestion, data from the results of the Taupo and Tauranga half-Ironman triathlons in 2012 were analyzed. Twenty-seven athletes who participated in both races in 2012 were found. These athletes were 28 min faster at the Tauranga race, which can be interpreted as a period effect. The within-subject SD was calculated to be 14.5 min (95% CI = 11.0–19.0 min). The sample size calculation presented at the synopsis stage was n = 65 subjects for inference by traditional null hypothesis testing using a paired t-test and analysis at 80% power and 5% alpha level. Further analysis found a coefficient of variation (CV) of 1.8% for Olympic distance triathlon (23). Hopkins et al. (8) argued that 0.5 times (90% CI = 0.3–0.7 times; later updated to 0.3 × CV [9]), the CV was an appropriate minimal relevant difference (that improves placing over normal race-to-race error). In case of a crossover design, the smallest relevant difference was therefore 1.8% × 0.5 and the within-subject SD was 1.8% × √2. Using the observed effect of 3% by Currell and Jeukendrup (3) (10 min with respect to average half-ironman finishing time) and using an estimate for the within-subject SD of 14.5 min, analysis using paired t-test on the data of n = 25 subjects can show this effect as statistically significant at an alpha level of 5% with a power of 90%. The sample size of n = 65 provided an a priori power to detect a difference up to 6 min as statistically significant against the null with a power of 90%. A crossover was the only feasible experimental model within an open competition because of participant availability, logistics, and budget. The study cohort comprised a total of 10.0% and 10.9% of race starters in Taupo and Tauranga, respectively.

General method

The effect of GLUFRU ingestion on outcomes was estimated via a mixed model analysis of variance using the nlme package in R 3.0.1. For analysis of the triathlon leg and total performance time, where total finish time was the primary outcome, the treatment, prerace weight, and triathlon (period effect) were fixed effects, and subject was the random effect. Time was transformed into minutes and fractions thereof for the analysis. After analysis of the full data set (FAS), data adjusted for covariate (environmental conditions), dropouts, and moderators (body weight change, adjustments for mechanical events) were run in a per protocol (PP) analysis. Uncertainty was presented as 95% CI. A likelihood (CI)-based approach to inference using clinical effect magnitude thresholds was deployed, with likelihoods calculated from a spreadsheet (www.sportsci.org/resource/stats/xcl.xls). Inferential thresholds were according to performance-based within-subject criteria for the primary outcome (decision threshold relative to smallest worthwhile effect on performance defined earlier) (8,9) and between-subject criteria as the standardized difference for secondary outcomes (gut comfort and energy levels), where effect sizes were trivial 0.0–0.2, small 0.2–0.6, moderate 0.6–1.2, and large 1.2–2.0 (7,9).

RESULTS

Participant Response and Treatment of the Analysis Cohort

The disposition of participants in recruitment and throughout the study is illustrated in a consort diagram (Fig. 1). From the 78 participants enrolled, 4 withdrew before the first race and another 4 either withdrew after the first race, or did not finish, leaving 71 participants starting the second race, with a final dropout rate of 9%. Study withdrawal before competition occurred because of sickness, personal reasons, and injury. Two withdrawals during competition (did not finish) were the result of bicycle mechanical problems and leg injury. All participants who completed the races were included in the full analysis data set (FAS; not presented for brevity). Substantial mechanical events or physical injuries unrelated to study product (leg-muscle strains and asthma attack) affecting performance were excluded from the PP data set. Athlete leg times were adjusted within the PP data set for competition rule infringements leading to known fixed time penalties (e.g., 3-min penalty for drafting on the bike, 15 s for taking off the helmet too soon, and 2 min 30 s for technical assistance with the bike) and mechanicals or punctures requiring stop to change (10 min). Estimated times to repair puncture were not deemed sufficiently objective, so it was decided to treat as 10 min delay for the four cases.

The 74 participants starting the first race had 10.2 ± 7.9 yr triathlon experience with a range of 1 to 33 yr. In addition, these participants previously completed 4.3 ± 2.5 (Range 1–10) half-ironman triathlons. This group's half-ironman best time was (h:m:s) 04:50:50 ± 00:24:13, with a range from 03:52:03 to 05:46:46. Mean ± SD training volume (h·wk−1) over the 6 months before recruitment into the study was as follows: swimming, 2.7 ± 1.3; cycling, 5.8 ± 2.4; and running, 3.6 ± 1.4. The mean training volume (h:min) during the 5 d preceding the Taupo race was as follows: swimming, 01:40 ± 00:56; cycling, 02:58 ± 01:53; and running, 01:36 ± 00:57. Preceding the Tauranga race, training volumes were as follows: swimming, 01:38 ± 00:56; cycling, 02:50 ± 01:44; and running, 01:45 ± 01:01. After exclusion, the baseline demographics for the final 72 athletes completing the first race in the crossover (Taupo) was 179.2 ± 6.3 cm, 77.0 ± 7.8 kg·cm−1, and 41.3 ± 9.3 yr.

Prerace Diet

The mean ± SD 24-h prerace diet macronutrient breakdown before the Taupo race was as follows: total energy intake, 19,391 ± 3110 kJ; carbohydrate, 768 ± 128 g; fats, 125 ± 28 g; protein, 133 ± 28 g; and fluid, 4.1 ± 1.0 l. The same breakdown before the Tauranga race was as follows: total energy intake, 19,207 ± 3068 kJ; carbohydrate, 763 ± 134 g; fats, 121 ± 25 g; protein, 132 ± 29 g; and fluid, 4.1 ± 1.2 l. On average, the diets comprised 66% of energy derived from carbohydrate (9.9 g·kg body weight−1·d−1), 11% from protein, and 23% from fat.

Compliance to Treatment

All of participants ingested all study product units as prescribed, except 25–100 mL of residual carbohydrate beverage in 15 athletes and 30–50 mL residual water in 16 athletes in combined races. The total fluid and carbohydrate ingested in both races was consistent (Table 1).

T1-27
TABLE 1:
Fluids and carbohydrate intake per triathlon.

Primary Outcome

Performance

Analysis of standardized residuals revealed one outlier (>2 standardized residuals on qq-plot) from the Taupo race, who was removed from the PP data set. Summary statistics for the effect of carbohydrate composition on triathlon leg and total finish times for the PP data set and for the fastest and slowest quartile triathletes are presented in Tables 2 and 3. After adjustment for race-order and preexercise body weight, the mean GLUFRU–GLU ingestion effect on final finishing performance time was borderline small at −0.53% (95% CI = −1.30% to 0.24%). This inference was drawn from post hoc likelihood analysis (7,9) of the CI, which allowed for a possible small benefit (where threshold for small benefit was 0.3 times CV: 0.3 × 1.8% [9]) with a very unlikely probability of harm (likelihood of benefit–harm, 48.87%:0.33%; benefit odds, 285:1); under this schema, the smallest benefit–harm ratio supporting adoption of the intervention is 66:1 (7). The effects of GLUFRU–GLU on leg times were as follows: swim, −0.57% (95% CI = −1.34% to 0.19%); bike, −0.42 (95% CI = −0.87% to 0.04%); and run, −0.74% (95% CI = −2.77% to 1.31%) (Table 3). The magnitude of the GLUFRU–GLU effect was similar for the fastest 25% (−0.80%, 95% CI = −1.97% to 0.38%) and the slowest 25% of triathletes (−1.19%, 95% CI = −2.76% to 0.39%). Although the clinical type 1 error (likelihood of harm) chance was marginally elevated (Table 3), the benefit odds were also favorable for the fastest (benefit–harm odds, 191:1) and slowest quartiles (278:1). Performance was faster in the Tauranga race compared with the Taupo race, but the moderating effect of prerace weight was comparatively negligible (Table 3).

T2-27
TABLE 2:
Triathlon leg and final performance times by treatment group and fastest and slowest quartile triathletes.
T3-27
TABLE 3:
Outcomes for the primary statistical analysis and post hoc inferential assessment for the effect of multiple-transportable carbohydrate and the primary covariates race order and prerace body weight on triathlon leg and overall finish time for all and upper and lower quartile triathletes.

With respect to other covariate effects on performance, the 5-d training log total training volume (pre-Taupo, 372 ± 159 min; pre-Tauranga, 373 ± 148 min) contributed to the PP model, but the effect was trivial (−0.02 min, 95% CI = −0.05 to −0.00) and the moderating impact on the primary outcome negligible (5-d training log adjusted GLUFRU–GLU effect on final performance time, −1.53 min; 95% CI = −4.03 to 0.97).

Secondary Outcomes

Environmental conditions and body weight change

The temperature at the beginning of the swim for Race 1 (Taupo; 06:30 h start time) was cool and humid (12.5°C, 95% relative humidity) but rose sharply during the bike leg and run to 21°C 65% relative humidity by 11:00 h. By contrast, temperature and humidity were higher at swim start in Tauranga (19°C, 90% relative humidity, 06:30 h) but rose to similar levels by mid–late morning where most athletes were finishing. Mean ± SD heat stress index (http://www.hpc.ncep.noaa.gov/html/heatindex_equation.shtml) was 73.3 ± 0.2 and 72.9 ± 0.2 during the Taupo and Tauranga events, respectively. Linear regression analysis revealed that a 1-unit increase in heat stress index slowed race time (7.6%). Heat stress index in Taupo was 0.34 units (95% CI = 0.31–0.37 units) higher than that in Tauranga. Concomitantly, the pre- and postrace reduction in total body weight was 14.4% (95% CI = 6.7 to 22.1) greater at Tauranga compared with the reduction in Taupo. However, both change in body weight and heat stress index had trivial effect on the GLUFRU effect when included as a covariate (not shown).

Gastrointestinal comfort and physical exertion

Descriptive statistics for the perception of gastrointestinal comfort and physical exertion sampled postrace and treatment contrasts are shown in Table 4. The effects of GLUFRU on outcomes were trivial or inconclusive, except possible small lowering effects on the recollection of the perception of nausea during the swim and bike legs. A summary of the qualitative comments from the postrace survey on gastrointestinal comfort and response to supplement is in Supplemental Digital Content (see Table, Supplemental Digital Content 3, Summary of the qualitative comments from the postrace survey on gastrointestinal comfort and response to supplement, https://links.lww.com/MSS/A907). The survey data revealed a heterogeneous gut discomfort response that was common in both carbohydrate conditions. Participants reported from nil to severe instances (five individual reports of vomiting or severe stomach cramps of 34 comments inclusive of both races). Several negative comments about the bars were reported (21 of 127 total postrace survey comments; not shown) pertaining to difficulty in eating (requirement for chewing, mouth dryness) and three comments relating to vomiting proceeding bar ingestion. Three of the 127 comments related ingestion of gels to severe gut discomfort and vomiting during the run. No serious adverse events were recorded.

T4-27
TABLE 4:
Statistical summary of the effect of multiple- and single-transportable carbohydrate on gastrointestinal comfort and physical exertion during the triathlons.

The correlation between gastrointestinal comfort and physical exertion parameter and corresponding triathlon leg times is shown in Supplemental Digital Content (see Table, Supplemental Digital Content 4, Estimate of correlation between the psychometric scales for gastrointestinal comfort and respective leg times, https://links.lww.com/MSS/A908). In both treatment conditions, small–moderate correlations (r) between faster times and higher energy were observed during the swim (GLUFRU −0.26, 95% CI = −0.46 to −0.03; GLU −0.28, 95% CI = −0.49 to −0.05) and run (GLUFRU −0.45, 95% CI = −0.62 to −0.24; GLU −0.49, 95% CI = −0.65 to −0.29). In addition, there was a small correlation between gut discomfort and performance time during the bike when ingesting GLUFRU (−0.23, 95% CI = −0.44 to 0.00); that is, higher discomfort was associated with faster times. By contrast, the pattern was reversed during the run (0.30, 95% CI = 0.06 to 0.50). No multiplicity adjustments were performed, and conservative interpretation is warranted. Moreover, results on run legs might be explained as a cumulative effort, which was not taken into account.

DISCUSSION

The principle finding of the current study was that multiple-transportable carbohydrates ingested in multiple food formats caused a slight improvement in performance with a very low probability of harm, relative to single-transportable carbohydrate. Coupled with the negligible relative financial and practical cost of the intervention, these results suggest that the enhancement with a multiple-transportable carbohydrate feeding strategy is safe and worthwhile adopting by half-ironman competing triathletes.

The present study was the first large-sample controlled trial to estimate the practical effect size of a performance–nutrition invention with an established laboratory case for efficacy on in-competition long-distance triathlon performance. The trial achieved ecological validity and generalizability as an unrestricted in-competition randomized trial of the effect of a widespread commercially adopted intervention for endurance athletes. Blinding was stringent, eliminating participant and investigator bias. Exercise and diet were fully controlled for 24-h preceding sampling. Compliance was recorded and resolved as very high. Consequently, the current unique design and analysis features address major recent criticisms of evidence-based sport nutrition research (6).

At first sight, the magnitude of the average performance enhancement previously indicated from laboratory trials with drinks containing multiple-transportable carbohydrates (i.e., mean benefit 1%–8%) (29) was not immediately translated to the mixed-food format in competition, but examination of the forest plot of standardized differences (refer to Fig. 1B in reference [29]) in six of eight prior studies revealed an overlap of the CI into the trivial small effect size. Although the current estimate is more precise than the previous laboratory studies (narrow CI), together, the results suggest that the real effect of multiple-transportable carbohydrate in the population is small and around the threshold considered worthwhile for adoption, supported by a very low probability of harm. Such real-world findings are important and relevant to triathletes, and other researchers so fairly establish the likely true outcome of an intervention. In other areas of research, promising laboratory data reveal an overinflated estimate of the real-world effect, e.g., fish oil supplements showing improved survival in patients with a previous myocardial infarction or heart failure, but in a large sample (n = 12,513) clinical trial, the oil failed to reduce cardiac death rates in patients with multiple cardiovascular risk factors or atherosclerotic vascular disease (27).

Nevertheless, the effect size for the mean difference was less than expected, and several contrasting design parameters may help explain the discrepancy in mean effects between studies. First, the relative mean percent enhancement of multiple-transportable carbohydrate escalates with increasing average ingestion rate (29). The two studies without CI into the trivial standardized difference were published by Currell and Jeukendrup (3) and Triplet et al. (36). These authors reported 8% (95% CI = 4.8%–12.1%) and 7.5% (95% CI = 1.3%–14%) mean power enhancements with concentrated (14.4%) hypertonic (~800 mOsmol·kg−1) glucose–fructose versus glucose solutions ingested on average at 1.8 and 2.4 g·min−1, respectively. These ingestion rates caused slower gastric emptying and induced high severity of gastrointestinal distress (previously linked to impaired performance [3,30,36]), which was more so with glucose only (13,16,36). On the other hand, Baur et al. (1) reported 3.1% (95% CI = −0.2% to 6.4%) better performance with a glucose–fructose versus glucose beverage ingested at 1.6 g·min−1, but there was no benefit of the multiple-transportable carbohydrate versus glucose-only ingested at 1.0 g·min−1, with the higher glucose ingestion rate (1.6 g·min−1) relatively reducing performance time. Although there was a handful of reports of severe gastrointestinal distress (vomiting and gut cramps) in the current study, on the whole, the carbohydrate-type difference was minor, with only a small possible overall reduction in nausea observed (swim–bike–run composite mean GLUFRU–GLU effect −0.4 scale units, 95% CI = −1.2 to 0.3 units). Energy levels on the other hand were not affected by carbohydrate type, which supports similar net levels of carbohydrate absorption. By contrast, glucose–fructose ingested at an average of 1.4–1.5 g·min−1 in drinks had no clear effect on nausea in a mountain bike race, but it lowered nausea (equivalent scale units −0.6, 95% CI = −9.4 to −2.4) during repeated sprints during endurance cycling in the laboratory, relative to glucose only (30). These data revealed a multiple-transportable carbohydrate effect size association between carbohydrate ingestion rate and gastrointestinal distress. Accordingly, it was appropriate to minimize gut discomfort within the current study leading to the selection of the lower average carbohydrate ingestion rate of 1.3 g·min−1 (78 g·h−1) optimal for performance (34). As a result, it is reasonable to speculate that the gastrointestinal distress trouble in previous high-dose laboratory studies arose from the accumulation of glucose within the gut via saturation of the single-monosaccharide absorption capacity through SLGT1 (29). It follows that the moderate–large standardized effect size of multiple-transportable carbohydrate on performance resulting from prior aggressive feeding rate regimens (3,36,40) may be part product of gut carbohydrate accumulation and discomfort (30), in addition to lower exogenous carbohydrate oxidation in the glucose-only condition (13).

Second, negative comments regarding palatability and comfort associated with the bars suggest that carbohydrate format may have modified outcomes. Relative to a beverage, the oxidation rate of the same carbohydrate derived from a bar was 22% less (95% CI = −56% to 12%) (25). Ingestion of carbohydrate in bars relative to the same form and calories of carbohydrate in gels reduced peak power, gut comfort, and ease of exertion during intense cycling, although there was no clear difference between bar versus drink formats (5). The bars also contained fat, protein, and fiber. These other components are known glycemic-response moderators (i.e., glycemic index) known to slow gastric emptying and digestion rate of the carbohydrate (26) responsible for moderating monosaccharide absorption rates across the gut epithelia and, therefore, probably contribute to the current incidence of gastrointestinal distress.

A third possible explanation was partial intrinsic regulation of carbohydrate ingestion rate. Participants received defined supplement units at a specified time where possible (e.g., preexercise meal, bar, and gel ingestion immediately after collection from feed station during the run), but other feeding was instructed at regular intervals (every 10 min), although in-practice was ad libitum between feed station segments. Unlike the experimentally clamped rate differential established within previous laboratory bolus-feeding regimens (1,3,11–13, 28,30,35,40), the current feeding regimen—which represents normal practice—may have attenuated the first purported ergogenic mechanism (gut discomfort) by permitting moderation of the rate of delivery of carbohydrate between feed station segments. Furthermore, most authors, including ourselves, present and describe ingested and oxidized exogenous carbohydrate rate outcomes per minute; this accountancy likely misrepresents conditions at the site of absorption. To illustrate, in all prior studies, drinks were provided at 15- or 20-min intervals as large boli (e.g., 150–250 mL containing 24–36 g carbohydrate). Assuming uninhibited gastric emptying (most test solutions were <500 mOsm·kg−1 [26]), the boli would rapidly distribute and accumulate across the duodenojejunal mucosa creating a carbohydrate-absorption bottleneck associated with saturation of monosaccharide transporters (29). It is reasonable to propose, therefore, that a combination of intrinsic regulation (moderated ingestion rate across the duration of the feed station segment) and mixed-food format digestion may have slowed the rate of delivery of monosaccharide to the gut wall attenuating the absorption bottleneck of glucose, and hence the relative advantage of multiple saccharides of access to alternative and additional transport mechanisms (29).

The maximal physical and mental effort of competition and the multiple-sport nature of the triathlon may both have affected on the efficacy of multiple-transportable carbohydrate relative to previous shorter, less intense, single-mode laboratory trials (1,3,28,30,35,40). Long-distance triathlons are raced at 75%–80% of V·O2max (20), attenuating gastric emptying and splanchnic blood flow (22). Triathlon competition was associated with higher incidence of gastrointestinal distress than running or cycling (19,24,26).

The current field trial was not without operational challenge; it involved substantial logistics, cooperation from event organizers, and risk (e.g., weather events). However, recruitment was relatively straightforward, dropout low, and compliance high. Testing throughput rate, net labor, facility, and equipment costs were substantially less relative to the equivalent sample size in a laboratory setting. Limitations of the study were restricted to control of participant feeding frequency between feed stations (although a positive component of ecological validity) and variability introduced because of mechanical events or technical infringements on the bike leg, although these were adjusted for within the PP analysis. The use of GPS recording devices in future studies, for example, would provide more accurate estimate of time delay because of mechanicals.

Although the effect size was less than expected, statistical precision was greater. Post hoc analysis of the typical error (37) provides a relative measure of model sensitivity. Accordingly, the PP data set typical error was 2.2% (fastest 25%: 1.5%). By contrast, typical error after normalization for inflation error and treatment effect due to preload (10,37) in previous studies on the same intervention but in various cycling models were 1.8% (3), 2.1% (30), 2.8% (1), and 5.1% (36). Therefore, the current in-competition trial provided experimental conditions and sensitivity comparable with the best normalized laboratory tests, but with the substantial advantage of translational validity and clinical inference.

CONCLUSION

Multiple-transportable carbohydrate ingested as a 2:1 ratio of glucose/maltodextrin–fructose within bars, gels, and drinks, and at optimal rates for endurance performance, provided an adoption worthy small benefit to half-ironman competition performance, relative to isocaloric single-transportable carbohydrate (glucose/maltodextrin). Although small reductions in nausea were possible, carbohydrate composition did not clearly affect the magnitude of gut discomfort or perceived energy. High total carbohydrate ingestion rate in prior work and the possibility of moderated carbohydrate digestion and absorption rate with the current mixed-food format may help explain differences between prior work and the current trial. The in-competition clinical trial provided ecological validity and sensitivity for evaluation of nutritional interventions in athletes.

Massey University received funding from Nestec Ltd., Vevey, Switzerland. Statistical power and analysis and randomization schema were provided by Maya Shevlyakova and Dominik Grathwohl, Nestle Research Centre, Lausanne. D. R. designed the research; D. R. and S. H. conducted the research; D. R. analyzed the data. D. R. and S. H. wrote the paper, with D. R. responsible for final content. The authors read and approved the final manuscript. Event organization and logistics: Sport Bay of Plenty, Triathlon New Zealand, Ironman New Zealand Ltd. (Taupo Half Ironman), SMC Events Ltd. (Tauranga Half Ironman), Nikki Scott, Wayne Reardon, and Kate Melville. Research assistance and volunteers: Kim Gaffney, Nick Castro, Brooke Price, Adam Lucero, Anna Hobman, Weijei Lim, Dylan Barrow, Megan McGillivray, Haley Bertelsen, Jo Goudie, Kolonio Tuiravirari, Shaun Higgins, Christina Syratt, Sean Martin, Laura Hardy, Carlene Stark, Brandon Woolley, Steve Stannard, Elizabeth Stannard, Robert Stannard, Thomas Stannard, Catherine Stannard, Kaydi O'Connor, Larissa Simmen, Erin Kouwenhoven, and Lauren Atkinson. Monitoring and databases: Julie Chambard and David Bailey.

Clinical Trial Registration (ClinicalTrials.gov): NCT02031783.

The authors report no conflict of interest. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of the study do not constitute endorsement by the American College of Sports Medicine.

REFERENCES

1. Baur DA, Schroer AB, Luden ND, Womack CJ, Smyth SA, Saunders MJ. Glucose–fructose enhances performance versus isocaloric, but not moderate, glucose. Med Sci Sports Exerc. 2014;46(9):1778–86.
2. Campbell C, Prince D, Braun M, Applegate E, Casazza GA. Carbohydrate-supplement form and exercise performance. Int J Sport Nutr Exerc Metab. 2008;18(2):179–90.
3. Currell K, Jeukendrup AE. Superior endurance performance with ingestion of multiple transportable carbohydrates. Med Sci Sports Exerc. 2008;40(2):275–81.
4. DeBosch BJ, Chi M, Moley KH. Glucose transporter 8 (GLUT8) regulates enterocyte fructose transport and global mammalian fructose utilization. Endocrinology. 2012;153(9):4181–91.
5. Guillochon M, Rowlands D. Solid, gel, and liquid carbohydrate format effects on gut comfort and performance. Int J Sport Nutr Exerc Metab. 2016;20:1–21.
6. Heneghan C, Perera R, Nunan D, Mahtani K, Gill P. Forty years of sports performance research and little insight gained. BMJ. 2012;345:e4797.
7. Hopkins WG, Batterham AM. Error rates, decisive outcomes and publication bias with several inferential methods. Sports Med. 2016;46(10):1563–73.
8. Hopkins WG, Hawley JA, Burke LM. Design and analysis of research on sport performance enhancement. Med Sci Sports Exerc. 1999;31(3):472–85.
9. Hopkins WG, Marshall SW, Batterham AM, Hanin J. Progressive statistics for studies in sports medicine and exercise science. Med Sci Sports Exerc. 2009;41(1):3–13.
10. Hopkins WG, Schabort EJ, Hawley JA. Reliability of power in physical performance tests. Sports Med. 2001;31(3):211–34.
11. Jentjens RL, Achten J, Jeukendrup AE. High oxidation rates from combined carbohydrates ingested during exercise. Med Sci Sports Exerc. 2004;36(9):1551–8.
12. Jentjens RL, Jeukendrup AE. High rates of exogenous carbohydrate oxidation from a mixture of glucose and fructose ingested during prolonged cycling exercise. Br J Nutr. 2005;93:485–92.
13. Jentjens RL, Moseley L, Waring RH, Harding LK, Jeukendrup AE. Oxidation of combined ingestion of glucose and fructose during exercise. J Appl Physiol. 2004;96(4):1277–84.
14. Jentjens R, Shaw C, Birtles T, Waring R, Harding L, Jeukendrup A. Oxidation of combined ingestion of glucose and sucrose during exercise. Metabolism. 2005;54(5):610–8.
15. Jentjens RL, Venables MC, Jeukendrup AE. Oxidation of exogenous glucose, sucrose, and maltose during prolonged cycling exercise. J Appl Physiol. 2004;96(4):1285–91.
16. Jeukendrup AE, Moseley L. Multiple transportable carbohydrates enhance gastric emptying and fluid delivery. Scand J Med Sci Sports. 2010;20(1):112–21.
17. Jeukendrup AE. Carbohydrate and exercise performance: the role of multiple transportable carbohydrates. Curr Opin Clin Nutr Metab Care. 2010;13(4):452–7.
18. Jeukendrup AE, Moseley L. Multiple transportable carbohydrates enhance gastric emptying and fluid delivery. Scand J Med Sci Sports. 2010;20(1):112–21.
19. Jeukendrup AE, Vet-Joop K, Sturk A, et al. Relationship between gastro-intestinal complaints and endotoxaemia, cytokine release and the acute-phase reaction during and after a long-distance triathlon in highly trained men. Clin Sci (Lond). 2000;98(1): 47–55.
20. Laursen PB, Rhodes EC, Langill RH, McKenzie DC, Taunton JE. Relationship of exercise test variables to cycling performance in an Ironman triathlon. Eur J Appl Physiol. 2002;87:433–40.
21. Leturque A, Brot-Laroche E, Le Gall M, Stolarczyk E, Tobin V. The role of GLUT2 in dietary sugar handling. J Physiol Biochem. 2005;61(4):529–37.
22. Maughan RJ, Leiper JB, McGaw BA. Effects of exercise intensity on absorption of ingested fluids in man. Exp Physiol. 1990;75:419–21.
23. Paton CD, Hopkins WG. Competitive performance of elite olympic-distance triathletes: reliability and smallest worthwhile enhancement. Sportscience. 2005;9:1–5.
24. Pfeiffer B, Stellingwerff T, Hodgson AB, et al. Nutritional intake and gastrointestinal problems during competitive endurance events. Med Sci Sports Exerc. 2012;44(2):344–51.
25. Pfeiffer B, Stellingwerff T, Zaltas E, Jeukendrup AE. Oxidation of solid versus liquid CHO sources during exercise. Med Sci Sports Exerc. 2010;42(11):2030–7.
26. Rehrer NJ, van Kemenade M, Meester W, Brouns F, Saris WH. Gastrointestinal complaints in relation to dietary intake in triathletes. Int J Sport Nutr. 1992;2(1):48–59.
27. Risk and Prevention Study Collaborative Group. n-3 fatty acids in patients with multiple cardiovascular risk factors. N Engl J Med. 2013;368(19):1800–8.
28. Roberts JD, Tarpey MD, Kass LS, Tarpey RJ, Roberts MG. Assessing a commercially available sports drink on exogenous carbohydrate oxidation, fluid delivery and sustained exercise performance. J Int Soc Sports Nutr. 2014;11(1):8.
29. Rowlands DS, Houltham S, Musa-Veloso K, Brown F, Paulionis L, Bailey D. Fructose–glucose composite carbohydrates and endurance performance: critical review and future perspectives. Sports Med. 2015;45(11):1561–76.
30. Rowlands DS, Swift M, Ros M, Green JG. Composite versus single transportable carbohydrate solution enhances race and laboratory cycling performance. Appl Physiol Nutr Metab. 2012;37(3):425–36.
31. Rumessen JJ, Gudmand-Høyer E. Absorption capacity of fructose in healthy adults. Comparison with sucrose and its constituent monosaccharides. Gut. 1986;27(10):1161–8.
32. Sareban M, Zügel D, Koehler K, et al. Carbohydrate intake in form of gel is associated with increased gastrointestinal distress but not with performance differences compared with liquid carbohydrate ingestion during simulated long-distance triathlon. Int J Sport Nutr Exerc Metab. 2016;26(2):114–22.
33. Shi X, Summers RW, Schedl HP, Flanagan SW, Chang R, Gisolfi CV. Effects of carbohydrate type and concentration and solution osmolality on water absorption. Med Sci Sports Exerc. 1995;27:1607–15.
34. Smith JW, Pascoe DD, Passe D, et al. Curvilinear dose–response relationship of carbohydrate (0–120 g·h−1) and performance. Med Sci Sports Exerc. 2013;45(2):336–41.
35. Tarpey MD, Roberts JD, Kass LS, Tarpey RJ, Roberts MG. The ingestion of protein with a maltodextrin and fructose beverage on substrate utilisation and exercise performance. Appl Physiol Nutr Metab. 2013;38(12):1245–53.
36. Triplett D, Doyle JA, Rupp JC, Benardot D. An isocaloric glucose–fructose beverage's effect on simulated 100-km cycling performance compared with a glucose-only beverage. Int J Sport Nutr Exerc Metab. 2010;20(2):122–31.
37. Vandenbogaerde TJ, Hopkins WG. Effects of acute carbohydrate supplementation on endurance performance: a meta-analysis. Sports Med. 2011;41(9):773–92.
38. Wall BA, Watson G, Peiffer JJ, Abbiss CR, Siegel R, Laursen PB. Current hydration guidelines are erroneous: dehydration does not impair exercise performance in the heat. Br J Sports Med. 2015;49(16):1077–83.
39. Wallis GA, Rowlands DS, Shaw C, Jentjens RL, Jeukendrup AE. Oxidation of combined ingestion of maltodextrins and fructose during exercise. Med Sci Sports Exerc. 2005;37(3):426–32.
40. Wilson PB, Ingraham SJ. Glucose–fructose likely improves gastrointestinal comfort and endurance running performance relative to glucose-only. Scand J Med Sci Sports. 2015;25(6):e613–20.
Keywords:

GLUCOSE; FRUCTOSE; SLGT1; MALTODEXTRIN; GUT COMFORT

Supplemental Digital Content

© 2017 American College of Sports Medicine