A Systematic Review of Athletes’ and Coaches’ Nutrition Knowledge and Reflections on the Quality of Current Nutrition Knowledge Measures

Context: Nutrition knowledge can in uence dietary choices and impact on athletic performance. Valid and reliable measures are needed to assess the nutrition knowledge of athletes and coaches. Objectives: (1) To systematically review the published literature on nutrition knowledge of adult athletes and coaches and (2) to assess the quality of measures used to assess nutrition knowledge. Data Sources: MEDLINE, CINAHL, SPORTDiscuss, Web of Science, and SCOPUS. Study Selection:36 studiesthat provided a quantitative measure of nutrition knowledge and described the measurement tool that was used were included. Data extraction: Participant description, questionnaire description, results (mean correct and responses to individual items), study quality, and questionnaire quality. Data synthesis: All studies were of neutral quality. Tools used to measure knowledge did not consider health literacy, were outdated with regards to consensus recommendations, and lacked appropriate and adequate validation. The current status of nutrition knowledge in athletes and coaches is dif cult to ascertain. Gaps in knowledge also remain unclear, but it is likely that energy density, the need for supplementation, and the role of protein are frequently misunderstood. Conclusions: Previous reports of nutrition knowledge need to be interpreted with caution. A new, universal, up-to-date, validated measure of general and sports nutrition knowledge is required to allow for assessment of nutrition knowledge. Keywords: nutritional knowledge; dietary knowledge; athlete; coach; sport; questionnaire; survey; measure; valid; sports nutrition 1. Introduction A carefully planned nutrition program has signi cant positive effects on athletic performance [1–3]. There has recently been an increase in internationally endorsed dietary guidelines for athletes, re ected by the publication of several consensus statements on optimal intake

for assessment of nutrition knowledge. Keywords: nutritional knowledge; dietary knowledge; athlete; coach; sport; questionnaire; survey; measure; valid; sports nutrition 1. Introduction A carefully planned nutrition program has signi cant positive effects on athletic performance [1–3]. There has recently been an increase in internationally endorsed dietary guidelines for athletes, re ected by the publication of several consensus statements on optimal intake and timing of food, uid, and supplements [4,5]. Despite this, research indicates that many athletes have sub-optimal dietary intakes [6,7], which may be due to lack of time, nances, cooking skills, and access to cooking equipment when attempting to select and prepare appropriate meals and snacks [8]. Food choices may also be driven by factors such as cultural background, taste preferences, appetite, attitude towards nutrition, and nutrition knowledge [8–10]. Nutrition knowledge is one of the few modi able determinants of dietary behaviors. Sports dietitians often center their dietary interventions on nutrition education to improve awareness of and compliance with expert dietary guidelines [10,11]. Nutrition education programs are rarely evaluated. There are a number of cross-sectional studies reporting on the nutrition knowledge of both athletes Nutrients2016,8, 570; doi:10.3390/nu8090570

Nutrients2016,8, 570 2 of 23 and coaches [12–14]. In a 2011 systematic review of the nutrition knowledge of recreational and elite athletes, scores across various nutrition knowledge questionnaires assessing general and sports speci c nutrition were mediocre, with mean scores of approximately 45%–65% [7]. There appeared to be a weak, positive correlation between nutrition knowledge and good quality dietary intake. The review concluded that in order to con rm the nutrition knowledge of athletes, and the relationship between nutrition knowledge and dietary intake, further high-quality research was required [7]. A 2014 review on the relationship between nutrition knowledge and dietary intake in adults also suggested that while the relationship between nutrition knowledge and dietary behavior appears to be moderate at best, results may be affected by the quality of measures used to assess knowledge [6]. Several studies assessing nutrition knowledge in athletes, not included in either of the aforementioned reviews, have been published in recent years [12,15–23]. Despite researchers having raised concerns regarding the validity of current nutrition knowledge measures [6,7,22], a detailed review of their limitations has not been undertaken to date. It is important to consider the comprehensiveness of the tools used. That is, the extent to which they have assessed all the relevant topics of nutrition knowledge, such as knowledge of macronutrients, micronutrients, supplementation, and hydration. In nutrition knowledge measures, questions on each of these topics are often grouped together and referred to as nutrition “sub-sections”. Previous reviews have identi ed concerns with drawing comparison between studies due to the heterogeneity of measures used; however, analysis of related nutrition sub-sections and responses to congruent questions across studies has not been performed. Several reports [9,11] and cross-sectional studies in elite Australian athletes and American College athletes have established that coaches are often a key source of nutrition information for athletes [16,24,25] but there has not been a systematic review of their nutrition knowledge. Considering the importance of nutrition knowledge as a modi able determinant of dietary behavior, the aims of the present review are to determine whether: 1. Athletes (aged 17 years and over) and coaches of adult

that coaches are often a key source of nutrition information for athletes [16,24,25] but there has not been a systematic review of their nutrition knowledge. Considering the importance of nutrition knowledge as a modi able determinant of dietary behavior, the aims of the present review are to determine whether: 1. Athletes (aged 17 years and over) and coaches of adult athletes are aware of expert nutrition recommendations 2. 3. The quality (validity, reliability, and comprehensiveness) of measures that have been used to assess nutrition knowledge is acceptable. 2. Methods 2.1. Protocol and Registration Methods for the review were in accordance with PRISMA guidelines and were registered with PROSEPERO [26]. 2.2. Search Terms A systematic search using the strategy nutrition knowledge or diet knowledge and athlete or sports people or sportsman and questionnaire or tool or measure or survey and valid or reliable, was conducted by one researcher (GT) from the earliest record until November 2015. A second search using the terms nutrition knowledge or diet knowledge and coach or questionnaire or tool or measure or survey was also conducted. Searched databases included MEDLINE, CINAHL, SPORTDiscuss, Web of Science, and SCOPUS. To ensure all related texts were captured, the reference lists of included articles were hand-searched. 2.3. Eligibility Criteria Original research (cross-sectional, observational, randomized controlled trials) conducted in adult athletes (17 years and older) or coaches/athletic trainers of adult athletes, and published

Nutrients2016,8, 570 3 of 23 in peer-reviewed journals were included for review. Abstracts, conference posters, reviews, and unpublished theses were excluded. Athletes were de ned as individuals involved in training and playing competitive sport. All `levels' of athletic competition, for example, recreational, college, national, and international were accepted. Only English language studies were included. Studies needed to report an aspect of nutrition knowledge (general, overall sports, or speci c sports nutrition e.g., hydration) using a measure that produced a numerical score. Studies that provided qualitative data only, or stated how many participants answered questions correctly/incorrectly, but failed to report overall quantitative results were excluded. The questionnaires could be in any format including self-administered, researcher-administered, online, or handwritten. To be included, studies also needed to provide a description of the tool used to assess knowledge including number of items, content, and question response categories (Table). Table 1.Eligibility criteria. Included Excluded 1. Original research (cross-sectional, observational, randomized controlled trials) 1. unpublished theses 2. of adult athletes (recreational, elite) 2. than coaches 3. 4. nutrition knowledge that could be converted into a single `score' (% total correct) 4. or intake; studies where a mean nutrition knowledge score could not be determined 5. knowledge including number of items, content and question response-categories 5. the tool used actually measured nutrition knowledge 2.4. Selection Process Duplicate and irrelevant articles were excluded on the basis of abstract and title by two authors (GT and AF). Articles deemed eligible for full-text review were retrieved and screened against the inclusion criteria by two authors (GT and AF) (Figure).Nutrients 2016, 8, 570 3 of 23 international were accepted. Only English language studies were included. Studies needed to report an aspect of nutrition knowledge (general, overall sports, or specific sports nutrition e.g., hydration) using a measure that produced a numerical score. Studies that provided qualitative data only, or stated how many participants answered questions correctly/incorrectly, but failed to report overall quantitative results were excluded. The questionnaires could be in any format including self- administered, researcher-administered, online, or handwritten. To be included, studies also needed to provide a description of the tool

e.g., hydration) using a measure that produced a numerical score. Studies that provided qualitative data only, or stated how many participants answered questions correctly/incorrectly, but failed to report overall quantitative results were excluded. The questionnaires could be in any format including self- administered, researcher-administered, online, or handwritten. To be included, studies also needed to provide a description of the tool used to assess knowledge including number of items, content, and question response categories (Table 1). Table 1. Eligibility criteria. Included Excluded 1. Original research (cross-sectional, observational, randomized controlled trials) 1. Abstracts, conference posters, reviews, and unpublished theses 2. Athletes (aged 17 years and older) and coaches of adult athletes (recreational, elite) 2. Adolescent athletes, all non-athletes other than coaches 3. English language studies 3. Non-English language studies 4. Studies reporting a quantitative measures of nutrition knowledge that could be converted into a single ‘score’ (% total correct) 4. Studies on nutrition attitudes, behavior, habits, or intake; studies where a mean nutrition knowledge score could not be determined 5. Studies that described the tool used to assess knowledge including number of items, content and question response-categories 5. Studies where it was unclear what (and how) the tool used actually measured nutrition knowledge 2.4. Selection Process Duplicate and irrelevant articles were excluded on the basis of abstract and title by two authors (GT and AF). Articles deemed eligible for full-text review were retrieved and screened against the inclusion criteria by two authors (GT and AF) (Figure 1). Figure 1. Flowchart of review process. * Secondary search using the term coach did not yield any additional relevant articles. NK = nutrition knowledge. Figure 1. Flowchart of review process. * Secondary search using the term coach did not yield any additional relevant articles. NK = nutrition knowledge.

Nutrients2016,8, 570 4 of 23 2.5. Data Extraction and Tool Quality Data from eligible studies were extracted by one author (GT). Information retrieved included: country of study, participant description (age, gender, sport played/coached, athletic level), questionnaire description (item generation, number of questions, question-response format), and results(mean nutrition knowledge scores, as well as nutrition sub-sections where participants scored above and below the study's overall mean). All scores were converted into percentage correct for consistency. Athletic level was based on descriptions provided in the paper; if athletic level was not adequately described, judgments on athletic level were based on other available information such as participant recruitment. Where reported, responses to individual items were also extracted then collated and summarized based on congruent themes. If questionnaires were not available, authors were contacted and permission to review a copy of the tool that was used was requested. Detailed data on the quality of the measures reported in the studies reviewed were recorded and used to calculate two separate quality scores: one for validity and reliability and another for questionnaire comprehensiveness. The validity and reliability score was based on a set of guidelines developed by Parmenter and Wardle [27]. Their recommendations are based upon psychometric validation techniques within the classical test theory (CTT) framework and are in line with leaders in the eld of scale development, such as Kline [28] and Nunnally [29]. They outline several methods for the development and evaluation of questionnaires, including: item analysis (item dif culty/item discrimination); homogeneity/”internal consistency” assessed using Cronbach's alpha; face validity assessed using a cohort similar to the target audience; content validity assessed using a panel of experts; construct validity assessed using known-group comparisons; and test–retest reliability using Pearson's correlation. In accordance with these guidelines, a validity score out of six was given. The decision was made to assess face validity, as, although it is similar to content validity, it utilizes different focus groups (target audience, not experts) and has different aims (ensuring readability/tool assesses what it intends to, not ensuring the entire content of the domain is covered). Scales developed under CTT apply only to the

score out of six was given. The decision was made to assess face validity, as, although it is similar to content validity, it utilizes different focus groups (target audience, not experts) and has different aims (ensuring readability/tool assesses what it intends to, not ensuring the entire content of the domain is covered). Scales developed under CTT apply only to the group of people who took the test; therefore, it is necessary to re-run internal consistency calculations for new samples [30]. Accordingly, in instances where an existing measure or modi ed version of an existing tool was used, a point was not awarded for internal consistency unless Cronbach's alpha was reassessed. If this test had been performed in the original sample, a partial point was given (denoted by P). If a tool had been modi ed from a previous tool, or was a composite of various previous questionnaires, validation points were not awarded unless the new version had undergone psychometric testing. For the comprehensiveness score, a point was awarded for each of the following nutrition sub-sections covered: general nutrition knowledge, carbohydrates, proteins, fats, micronutrients, hydration, pre-exercise nutrition, nutrition during competition, recovery nutrition, supplementation, and alcohol. A maximum of 11 points could be awarded. Decisions on whether a questionnaire included adequate coverage on each topic to be included as a nutrition sub-section were made by one author (GT), based on a combination of review of the actual tool (when available) and the description of the measure provided in the article. 2.6. Study Quality The methodological quality of studies was assessed by two reviewers (GT and AF), using the “Academy of Nutrition and Dietetics” “Quality Criteria Checklist for Primary Research” [31]. Disagreements were resolved by a third reviewer (BD). The checklist rates studies as positive, neutral, or negative (poor) on 10 criteria. The criteria addressing study group comparisons (3), methods for handling withdrawals (4), use of blinding (5), and description of interventions/comparisons/description of intervening factors (6) could not be logically applied to cross-sectional or observational studies. All studies awarded positive quality ratings needed to adequately address selection bias, make appropriate study group comparisons,

as positive, neutral, or negative (poor) on 10 criteria. The criteria addressing study group comparisons (3), methods for handling withdrawals (4), use of blinding (5), and description of interventions/comparisons/description of intervening factors (6) could not be logically applied to cross-sectional or observational studies. All studies awarded positive quality ratings needed to adequately address selection bias, make appropriate study group comparisons, clearly describe any interventions, and use valid and reliable measurements. To receive a “Yes” for criterion (7), “Were

Nutrients2016,8, 570 5 of 23 outcomes clearly de ned and the measurements valid and reliable?”, the questionnaire needed to undergo a least three different types of expected psychometric validation, outlined by Parmenter and Wardle [27], as above. Meta-analysis was not possible due to heterogeneity in measures used to assess nutrition knowledge. 3. Results 3.1. Study Selection The original search yielded 331 results. After removal of duplicate and irrelevant records, 42 studieswere retained for full-text review. An additional 11 records were identi ed through hand searching reference lists. Thus, a total of 53 full-text articles were screened for inclusion in the nal review. Thirty-six of these met the inclusion criteria. The reasons for excluding the other articles included: the age of the participants being less than 17 years old (n= 5), inability to extract a mean score (n= 9), lack of adequate questionnaire description (n= 2), or failure to assess nutrition knowledge (n= 1) (Figure). A secondary search using the term `coach' did not yield any additional relevant articles. 3.2. Study Characteristics The majority of the studies (n= 34) employed a cross-sectional design, with the remaining two [32,33] using a questionnaire to assess the effectiveness of an education program at two time points. Of the 36 included studies, 15 assessed nutrition knowledge in American college athletes [13,23–25,32–42]; two of these also collected data on coaches and athletic trainers, stratifying the results [23,24]. There were an additional four studies [38,43–45] that assessed the knowledge of coaches alone. Six studies assessed college athletes outside of the USA (three in Iran [15,20,46]; one each in India [17], Malaysia [21], and Nigeria [18]). Five studies [12,14,16,22,47] were conducted with elite athletes and three studies [48–50] assessed knowledge in recreational athletes. Five studies [15,24,33,34,45] did not report what sport the athletes played. Across the remaining studies, the other sports that were represented included: Australian football (AFL) [16], basketball [13,20,23,35,37–40,42,44], baseball [23,25,37,38,42], cross-country [13,35,41,42,44], cycling [50], football [13,20,23,35,37,38,44], golf [13,23,35,37,40], gymnastics [13,35,40,49], hockey [35,40,47], lacrosse [23,35,39], soccer [13,23,32,38,42], softball [13,19,35,37,40,42,44], running and/or track and eld [14,23,25,35,37,42,44,48], rugby [12,22,47,51], swimming [13,18,32,35,37,52], tennis [35,37,38,42,43], and triathlon [50]. Participant numbers

report what sport the athletes played. Across the remaining studies, the other sports that were represented included: Australian football (AFL) [16], basketball [13,20,23,35,37–40,42,44], baseball [23,25,37,38,42], cross-country [13,35,41,42,44], cycling [50], football [13,20,23,35,37,38,44], golf [13,23,35,37,40], gymnastics [13,35,40,49], hockey [35,40,47], lacrosse [23,35,39], soccer [13,23,32,38,42], softball [13,19,35,37,40,42,44], running and/or track and eld [14,23,25,35,37,42,44,48], rugby [12,22,47,51], swimming [13,18,32,35,37,52], tennis [35,37,38,42,43], and triathlon [50]. Participant numbers ranged from ve [17] to 595 [46]. Most studies were mixed-gender (n= 19) [13–15,18,20–24,35–37,39,44,46,47,50,53]. There were a total of 5231 participants: 2307 males, 2170 females, and 754 where gender was not reported. The mean age ranges of coaches and athletes were 33.0 to 43.2 years and 19.0 to 35.2 years, respectively. No studies reported the nutrition knowledge of older athletes (master's level) (Table). 3.3. Nutrition Knowledge Results 3.3.1. Demographic Factors Related to Nutrition Knowledge Scores Seven out of 11 studies that reported on prior nutrition knowledge found that higher levels of (general) education, previously undertaking a nutrition course, or currently majoring in nutrition studies correlated with higher nutrition knowledge scores [14,20,40,41,46,48,50]. Fifteen studies reported on male versus female scores, and 10 of these studies reported no signi cant difference [14,15,21,35,36,39,42,44,49,53]. All studies that assessed for differences between athletes from varying sports reported no signi cant differences in nutrition knowledge scores based on sport played [21,25,39,40]. Where reported, there was no signi cant difference in nutrition knowledge scores across National College Athletic Association (NCAA) divisions I, II, and III (ranked according to level of support and participation) [41,52].

Nutrients2016,8, 570 6 of 23 Table 2.Nutrition knowledge of athletes and coaches. References Participant Characteristics Questionnaire Author, Year, Country Athletic Level Sport Played N(Gender) Mean Age (Years) SD Questionnaire Used/Item Generation and Number of Questions Type of Questions Mean Correct Nutrition Knowledge Score (%) SD Nutrition Sub-Sections with Scores above Average Compared to the Total Mean Score within the Same Study (% SD Where Available) Nutrition Sub-Sections with Scores below Average Compared to the Total Mean Score within the Same Study (% SD Where Available) Quality Rating Abood et al. [32], 2004, USA College Soccer, Swimming n= 30 (F) 19.5 (SD NR) Self-developed; n= 42 True/False 68.5 NR NR Neutral Alaunyte et al. [12], 2015, UK Elite Rugby n=21 (M) 25 5 Existing Questionnaire(A–C of GNKQ);n= 28 Multi-Choice, Open-Ended, Less/More/Not Sure/Same 72.82 6.11 Recommendations made by experts (85.7 13.0) Food groups (71.2 7.2) and Making healthier Food Choices (69.5 14.0) Neutral Arazi and Hosseini [15], 2012, Iran College/`Non-College' NR n= 250 (130 M; 120 F); 121 College, 129 Non-College College M: 24.71 2.3, College F: 23.61 2.10, Non-College M: 23.42 1.8, Non-College F: 21.49 2.8 Modi ed Questionnaire (Zawila et al. 2003);n= 40 True/False 54.0 Vitamins (61.2), Calcium and Iron (56.48), Weight Loss (57.95) Macronutrients (50.7), Fiber (52.3), Sports Nutrition (49.74), General Nutrition (49.97) Neutral Azizi et al. [46], 2010, Iran College Sport Olympiad (range of sports) n= 595 (298 M; 297 F) M: 22.8 1.9, F: 21.8 1.8 Self-developed;n= 15 Strongly agree/agree/neutral/ disagree/strongly disagree *58.9(M: 52.36 6.2; F: 54.3 6.3) NR NR Neutral Barbaros-Tudor et al. [43], 2011, Croatia Coaches Tennis n= 58 (50 M; 8 F) 33 10. 8 Self-developed; n= 40 True/False 68.9 (SD not reported) NR NR Neutral Barr [48], 1987, USA Recreational Marathon runners n= 104 (F) Andn= 105 tness class participants (F) NR; 1.0% <20, 40.8% 20–29, 43.7% 30–39, 14.6% >40 years Self-developed;n= 87 True/False/Don't Know Athletes: 50.1 *; Fitness class participant: 42.6 * (SD not reported) Knowledge about general nutrition (50.9) Knowledge about sports nutrition (48.3) Neutral Botsis and Holden [44], 2015, USA Coaches Volleyball, Softball, Cross-Country, Track and Field, Football, and Basketball

This review assesses the nutrition knowledge of athletes and coaches and the quality of measurement tools.