Development Of An Algorithm To Assess Risk Behavior Toward Third-party Tested Supplement Use

PURPOSE: To develop a questionnaire-based algorithm to predict non-compliance toward using third-party tested (TPT) supplements. METHODS: A total of n = 410 collegiate athletes out of n = 3,580 athletes filled out a questionnaire about nutritional supplement behavior and self-reported TPT supplement use. Logistic regression models, with P ≤ 0.10, revealed binary outcome variables (0 = no/1 = yes) contributing to an algorithm predicting use or non-use of TPT nutritional supplements, resulting in a risk behavior score (0-100). Finally, a best fit to determine low vs. high risk for using TPT was based on risk quadrants using Chi-square and calculating area under the curve (AUC). RESULTS: A total of n = 320 student-athletes (age: 21.5 ± 1.6 years, 52% female, from >20 sports) were randomly selected from fully completed responses. All athletes reported the use of at least one or more supplements or sports foods during the last year, but only 34% (n = 109) of users consistently reported that all supplements they used were TPT. Selected algorithm items associated with use or non-use of TPT (reporting a positive or negative “Estimate”) are shown in Table 1, along with their prevalence (%), and other modeling outcomes. Risk quadrants best fit for classifying low vs. high risk toward inconsistent TPT-use resulted in a cut-off ≥60% (χ2 (4) = 61.26, P < 0.001), with reasonable AUC 0.78. There was a significant association for TPT use (yes/no) and risk behavior (low vs. high) defined from the algorithm (χ2 (1) = 58.59, P < 0.001). The algorithm had a high sensitivity, classifying 89% of non-TPT users correctly, while having a low specificity classifying 49% of TPT-users correctly. CONCLUSIONS: The algorithm accurately classifies high-risk inconsistent TPT-users, but lacks the specificity to classify consistent users at low risk. Still, the algorithm can help to inform sports nutritionists towards the future direction for education and counseling athletes on the individual level. Table 1. - Variables identified as predictors based on logistic regression including the prevalence per outcome for n=320 collegiate-athletes. Variable Prevalence (% yes) DF Estimate Standard error Wald Chi-Square Pr > Chi-Square Intercept 1 0.54 0.36 2.23 0.14 WADA: Are you familiar with banned substances that may occur in nutritional supplements listed on the WADA (world anti-doping agency) list? (Select only one - yes/no) 52% 1 -0.48 0.28 3.02 0.08 FIND&ORDER: I know where to find and order third-party tested supplements. (Only select one - yes/no) 50% 1 -0.64 0.30 4.57 0.03 SEARCH: Where do you go to look for information on nutritional supplements and sports foods?(Check all that apply - checked score: I do not search for information on my own) 22% 1 0.81 0.36 5.15 0.02 DISCUSSWITHRD: I discuss all my supplement choices with the Athletic Departmental Sports RD.(Only select one - yes/no) 58% 1 -0.81 0.28 8.24 0.004 PURCHASEOUTSIDE: Do you purchase or use nutritional supplements outside what is provided by your Athletic Department? (Only select one - yes/no) 49% 1 0.53 0.29 3.33 0.07 ADVICE: I”ve decided to purchase one or more supplements as a result of the advice of family, friends, or teammates. (Only select one - yes/no) 53% 1 0.57 0.28 4.19 0.04 Please check all of the nutritional supplements you have used during the last 12 months. (Check all that apply) MULTIVITAMIN: (Multivitamin and mineral supplement checked) 51% 1 -0.43 0.27 2.53 0.11 WEIGHTGAINER: (Weight gainer checked) 6% 1 -0.96 0.56 2.89 0.09 CAFFEINE: (Caffeine checked) 62% 1 1.34 0.28 22.89 <0.0001 CREATINE: (Creatine checked) 25% 1 -0.62 0.33 3.55 0.06 Wald Chi Square was calculated to indicate significance, with P≤0.10. The multivitamin variable was borderline significant, but because of its prevalence, the supplement was still discriminatory and adding in a meaningful way to the algorithm. The variables listed in the table delivered the following algorithm predicting TPT supplement use: 2.72 ^ (0.54 + -0.48*WADA + -0.64*FIND&ORDER + 0.81*SEARCH + -0.81*DISCUSSWITHRD + 0.53*PURCHASEOUTSIDE + 0.57*FRIENDADVICE + -0.43*MULTIVITAMIN + -0.96*WEIGHTGAINER + 1.34*CAFFEINE + -0.62*CREATINE) / 1 + 2.71828 ^ (0.54 + -0.48*WADA + -0.64*FIND&ORDER + 0.81*SEARCH + -0.81*DISCUSSWITHRD + 0.53*PURCHASEOUTSIDE + 0.57*FRIENDADVICE + -0.43*MULTIVITAMIN + -0.96*WEIGHTGAINER + 1.34*CAFFEINE + -0.62*CREATINE) Supported by: Collegiate and Professional Sports Dietitians Association (CPSDA)