ARTICLE Preventing Substance Use and Disordered Eating Initial Outcomes of the ATHENA (Athletes Targeting Healthy Exercise and Nutrition Alternatives) Program Diane L. Elliot, MD; Linn Goldberg, MD; Esther L. Moe, PhD, MPH; Carol A. DeFrancesco, MA, RD; Melissa B. Durham; Hollie Hix-Small, MS Objectives: To implement and to assess the efficacy of Results: Experimental athletes reported significantly a school-based, sport team–centered program to pre- less ongoing and new use of diet pills and less new use vent young female high school athletes’ disordered eat- of athletic-enhancing substances (amphetamines, ana- ing and body-shaping drug use. bolic steroids, and sport supplements) (PϽ.05 for each). Other health-harming actions also were Design and Setting: Prospective controlled trial in 18 reduced (less riding with an alcohol-consuming driver high schools, with balanced random assignment by school [P =.05], more seat belt use [PϽ.05], and less new to the intervention and usual-care control conditions. sexual activity [PϽ.05]). The ATHENA athletes had coincident positive changes in strength-training self- Participants: We enrolled 928 students from 40 par- efficacy (PϽ.005) and healthy eating behaviors ticipating sport teams. Mean age was 15.4 years, 92.2% (PϽ.001). Reductions occurred in intentions toward were white, and follow-up retention was 72%. future use of diet pills (PϽ.05), vomiting to lose weight (PϽ.05), and use of tobacco (PϽ.05) and Intervention: The ATHENA (Athletes Targeting Healthy muscle-building supplements (PϽ.005). The pro- Exercise and Nutrition Alternative) curriculum’s 8 weekly gram’s curriculum components were altered appropri- 45-minute sessions were incorporated into a team’s usual ately (controlling mood [PϽ.005], refusal skills practice activities. Content was gender-specific, peer- [P=.05], belief in the media [PϽ.005], and percep- led, and explicitly scripted. Topics included healthy sport tions of closest friends’ body-shaping drug use nutrition, effective exercise training, drug use and other [PϽ.001]). unhealthy behaviors’ effects on sport performance, me- dia images of females, and depression prevention. Conclusions: Sport teams are effective natural vehicles for gender-specific, peer-led curricula to promote healthy Main Outcome Measures: We assessed participants lifestyles and to deter disordered eating, athletic- by confidential questionnaire prior to and following their enhancing substance use, and other health-harming be- sport season. We determined program effects using an haviors. analysis of covariance–based approach within the Gen- eralized Estimating Equation framework. Arch Pediatr Adolesc Med. 2004;158:1043-1049 Author Affiliations: Department PPROXIMATELY HALF OF criteria for anorexia or bulimia nervosa, of Medicine, Oregon Health & female and male high can be associated with significant mor- Science University (OHSU), school students partici- bidity and mortality.7,8 Unfortunately, Portland (Drs Elliot, Goldberg, 1,2 and Moe and Mss DeFrancesco pate in school sports. programs to prevent these practices 9 and Durham); and the Oregon Contrary to popular largely have been unsuccessful or para- Research Institute, Eugene belief, these young athletes are not pro- doxically, may have increased these det- A 10 (Ms Hix-Small). tected from drug use and other harmful rimental behaviors. Financial Disclosure: Oregon behaviors.3,4 For young females, the Health & Science University and sociocultural pressures toward thinness For editorial comment Drs Elliot and Goldberg have a may be compounded by similar influ- significant financial interest from ences of their sport, resulting in more see page 1084 the commercial sale of prevalent disordered eating behaviors technologies used in this research. This potential conflict and body-shaping drug use (tobacco, The high school years are important of interest has been reviewed diet pills, diuretics, laxatives, amphet- for prevention, as drug use approxi- 5,6 2 and managed by the OHSU amines, and anabolic steroids). These mately doubles, and middle-school pre- Conflict of Interest in Research health-harming actions, even among vention programs’ beneficial effects Committee. individuals lacking the full diagnostic often are attenuated or lost.11 Because (REPRINTED) ARCH PEDIATR ADOLESC MED/ VOL 158, NOV 2004 WWW.ARCHPEDIATRICS.COM 1043 ©2004 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/24/2021 older adolescents’ behaviors are shaped more by peers ians provided written informed consent for study participa- and social influences and less by classroom activities,12 tion. The institutional review board of Oregon Health & Sci- effective high school–based prevention programs are ence University, Portland, approved the study. limited.13 Student involvement was voluntary. Study information high- Although sport participation may increase a young fe- lighted that neither team membership nor playing time would be influenced by ATHENA participation. School personnel were male’s risk for certain unhealthy activities, athletic teams not involved in recruitment, and signed consents were re- also may provide a natural setting for bonded peers to turned to research staff. Research assistants administered con- address gender-specific topics. The sport team’s use as a fidential questionnaires at baseline (immediately preceding the vehicle for health promotion was demonstrated for male sport season) and within 2 weeks of the sport season’s conclu- high school athletes with the ATLAS (Adolescents Train- sion. Coaches, teachers, and other school personnel were not ing and Learning to Avoid Steroids) program, which sig- involved with survey administration. nificantly improved nutrition behaviors and exercise self- efficacy, while reducing use of alcohol, illicit drugs, and performance-enhancing supplements.14 SURVEY INSTRUMENT The ATHENA (Athletes Targeting Healthy Exercise and Nutrition Alternatives) program parallels ATLAS in its school-based, team-centered format and promotion We assessed potential influences on drug use and disordered of healthy nutrition and effective exercise training as al- eating, including those of peers, coaches, the media, and ternatives to harmful behaviors. However, ATHENA’s ob- society.17 In addition, individuals’ knowledge and character- istics were indexed using items from our earlier research14,15 jectives are to reduce the young female athlete’s disor- 17,18 19 dered eating habits and to deter use of body-shaping and surveys concerning disordered eating, depression, self-image,20 and self-esteem.21 For knowledge of specific substances. To evaluate the ATHENA program’s feasi- topics, self-esteem, body image, and perceptions about peers bility and efficacy, we recruited 18 high schools and ran- and teammates, we grouped 3 to 7 survey items into reliable domized them to an intervention and a usual-care con- constructs (standardized ␣ at least .78) and used the items’ trol group. We hypothesized that program participation mean as the score. would result in greater positive change, which would be Acquiring new behaviors typically follows a sequence, mediated by gender-specific influences targeted in the cur- wherein intentions toward an action increase, followed by riculum. experimenting with an activity, and finally, the behavior becomes an established practice.22,23 Higher intentions can characterize those with a greater likelihood of practicing METHODS those habits in the future. We assessed behavioral intent by asking about future actions. These and other items used a 7-point Likert scale, with anchors ranging from strongly dis- PARTICIPANTS agree to strongly agree. Substance abuse questions used a format similar to the Monitoring the Future National Sur- We asked 18 public high schools from northwest Oregon vey24 and the Youth Risk Behavior Surveillance–United and southwest Washington to participate; all enrolled in the States, 2001,2 with similar designs to assess supplement use study. In selecting that number of schools, we used preva- 15,16 and disordered eating behaviors. The final instrument con- lence rates from our prior studies and assumed relatively tained 177 questions and could be completed in approxi- low interdependence among students enrolled at the same mately 30 minutes. school. We anticipated 18 schools would provide a greater than 90% chance of detecting intervention effects accounting for 10% or more of the variance, without covariate adjust- ment, for primary outcomes, such as diet pill use, disordered INTERVENTION eating behaviors, and healthy alternative behaviors. Partici- pating schools had a spectrum of sizes and varied locales to minimize potential cross-contamination. They were matched The ATHENA intervention was delivered during a team’s sport in balanced dyads based on size, average socioeconomic sta- season, and its eight 45-minute classroom sessions were inte- tus, and student demographics. An individual unfamiliar grated into a team’s usual practice activities. During those ses- with the participating schools used a program for generating sions, a team assembled as a group with students organized in random numbers to assign 1 member of the pair to the coach-assigned stable learning clusters of approximately 6 stu- experimental (even) or control (odd) group, with the dents (squads). The coach also designated 1 athlete per small remaining pair member going to the other condition. group as the squad leader. The squad leaders led approximately Because