Journal Logo

BASIC SCIENCES: Epidemiology

Predictors of Future Anabolic Androgenic Steroid Use


Author Information
Medicine & Science in Sports & Exercise: September 2006 - Volume 38 - Issue 9 - p 1578-1583
doi: 10.1249/01.mss.0000227541.66540.2f
  • Free


The nontherapeutic use of anabolic androgenic steroids (AAS) is high among young people. In the United States, life-time use among high school students varies between 4 and 12% for males (4), and figures from other Western countries fall in the 1-4% range for young males (1). There are between two and six times as many males as females using AAS. AAS for nonmedical purposes have numerous side effects, of which liver problems, cardiovascular problems, and permanent change in secondary sex characteristics (development of female-type breasts in men; growth of facial hair and deepening of voice in women) are among the most serious physical complications. In addition, a wide range of psychological effects are suggested, including increased aggression, irritability, mania, depression, and suicidality (5).

AAS was first known to the public through its use among top athletes in sports requiring explosive strength. After its introduction, use of AAS appears to have spread to subelite levels, including high school and college. Because of its alleged ability to increase muscle size in response to training, there is a comparatively high prevalence of AAS users among body builders and others involved in strength training (6).

Numerous correlates of AAS use have been reported. However, several alleged risk factors for use may be effects of AAS use or AAS-related behavior, such as heavy weight training or associating with a body-building or strength-training milieu (5). Cross-sectionally identified correlates of AAS use can be categorized into three areas (1). At present, there has been no longitudinal study addressing AAS use.


Use of AAS will provide a competitive advantage when muscle mass and explosive strength is important to performance. Thus, high prevalence of AAS has been found in participants of sprint events, throwing events, American football, weight lifting/power lifting, body building (4), and self-defense sports (1). One might hypothesize that the prevalence of AAS use increases as level of performance increases and is highest among those who compete at the international level (29). This might be attributable to higher rewards of winning, which make it more tempting to use AAS. However, it might also be attributable to a selection effect: those whose results have allowed them to compete at national or international levels in power events may have gained these results in part by using AAS.


Because a muscular body fits the present-day masculine body ideal, one should expect body dissatisfaction and possibly also eating problems (4) to be predictive of later AAS use. Appearance improvement is the top motive for AAS use among persons training with heavy weights in gyms and is the second most important motive among adolescent AAS users (12). Indeed, the competitive and appearance motives merge among body builders.


In some countries, AAS use without a prescription and AAS dealing are illegal, and in most societies AAS use is socially condemned. AAS might therefore fit into a wider syndrome of illicit drug use and problem behaviors. AAS dealing is prohibited in Norway, and a law against nonmedical AAS use has been suggested by the government. Positive correlations with illicit drug use have been one of the most consistent relationships across studies (6), but the cooccurrence of delinquency and problem-type behavior (e.g., early sexual debut, conduct problems) have also been reported in those studies addressing this issue (4). These results imply that we do not know whether it is illicit drug use, other types of conduct problems, or the whole syndrome of problem behavior that is decisive for AAS use.

AAS users included in the convenience samples that dominate the AAS literature often report a history of longstanding AAS misuse. However, we do not presently have data concerning the change and stability of AAS use in community samples. Only a small percentage of nonmedical AAS are obtained through legitimate channels such as physicians. As for most other illegal drugs, there are reasons to believe that novice users are offered the drug from a supplier (1). Although most offers are tuned down, those offered the drug are at particular risk of using it. Apart from the physical side effects, there are concerns that AAS use may alter mood and behavior, at least in some individuals. AAS users have been found to have increased levels of aggression (1), mania (2), and eating disturbances (4). Depression and suicidal behaviors, often following AAS withdrawal, have also been reported (1). Concern has also been raised about the possibility of AAS serving as a gateway drug for illicit drug use in otherwise low-risk populations (13).

To address some of the gaps that exist in the current literature, a random sample of Norwegian adolescents was followed prospectively over 5 yr into early adulthood to 1) discern potential risk factors for later AAS use and for being offered AAS, 2) assess change and stability in AAS use, and 3) investigate whether AAS use predicts later mental health problems (conduct problems, depression, attempted suicide, eating problems, frequent alcohol intoxication, cannabis use, and use of hard drugs) when initial levels of such problems are controlled.



Data for the present research stem from the second and third waves of data collection in the Young in Norway Study (1). In the first wave in 1992 (T1), 12,287 students in grades 7-12 (ages 12-20 yr) from 67 schools representative of high schools in Norway comprised the initial sample. Response rate was 97%. Three schools were included at T1 for nonprospective reasons and were not part of the follow-up. At one other school, there was a burglary in the school's archives, and project ID records were lost. In all, 9679 students from 63 schools were eligible to complete the T2 (1994) questionnaire. Students who were tested at T1 in the spring term of 8th or 11th grade, in 9th grade, or in 12th grade would, in most cases, have completed the 3-yr track at their junior or senior high school 2 yr later (T2) and would have therefore left their original schools. These students received the T2 questionnaire by mail. Students still in their original schools filled out the questionnaire at school according to the same procedure as in the initial survey (T1). Among those who were still at their original school, 92% responded. Only students who completed the questionnaires at school at T2 (N = 3844) were followed up at T3 (1999) because of a comparatively lower response rate among those receiving the questionnaire by mail. Because the study was originally planned as a two-wave study, new informed consent had to be obtained at T2. Those consenting at T2 (N = 3507; 91.2%) received questionnaires by mail at T3. Data were received from 2924 participants (84%). The overall response rate was therefore 68%. It is possible that those who dropped out of the study were different from those who participated, thereby reducing the generalizability of results. A logistic regression analysis with forward inclusion according to likelihood ratio was run to build a best-fit model of attrition. The final model included the following measures at T1 as predictors of attrition at T3: gender (male), age, grade level, poor grades, suburban or urban residency as opposed to rural or small-town residency, and the participant's prediction of manual work for an occupation when 40 yr of age. In all, 75% of the cases were correctly classified using this information, including 59% of the attrition group. Repeating the analyses reported below by correcting for this attrition by means of weighting the sample according to the results from the logistic regression analysis predicting attrition yielded almost identical results to the ones presented here. Hence, although attrition was somewhat selective, it had no impact on the main findings. The mean age of the participants was 14.9 yr (SD = 1.7) at T1, 16.5 yr (SD = 1.9) at T2, and 22.1 yr (SD = 1.9) at T3.


Every student gave his/her consent in writing based both on an oral and written description of the project formulated according to the standards prescribed by the Norwegian data inspectorate. According to these standards, written informed consent was also obtained from parents of students below the age of 15 yr. The study was approved by the regional ethical committee south, Norway. Students were instructed to place the completed questionnaires in an envelope and to seal it themselves. It was made clear to the participants that by using this procedure, the researchers would have access to their questionnaires, but that the researchers could not know their identity (whereas school administrators knew the participants' identities but had no access to their questionnaires). A teacher trained by the liaison officer monitored the students in the class during completion. To prevent students from influencing classmates' responses, all eligible students at each school completed the questionnaire simultaneously. Students who had consented to participate but who were not present in class during school that day completed the questionnaire together on a later occasion. At T3, participants received the questionnaire by mail. Those not responding within 4 wk were mailed another questionnaire with a reminder letter.


AAS involvement.

The participants were asked whether they had ever used anabolic steroids (doping) (yes/no), the number of times they had used AAS during the preceding 12 months (6-point scale; range: "0 times" to "more than 50 times"), and whether they had ever been offered AAS (yes/no). Follow-up questions were used to determine whether they had used or been offered types of doping other than AAS (yes/no). Questions about AAS were included at T2 and T3 only.

Involvement in power sports.

Study participants were asked an open-ended question about whether they had competed in or were currently competing in any sports, and to state the type of sport. Students were also asked whether they had been involved or were currently involved in noncompetitive sports. Those indicating weight lifting, body building, boxing, gymnastics, wrestling, or martial arts were grouped as power sports participants. In addition, hours per week spent training in private gyms was recorded. Perceived athletic competence was measured using a revised version of the Self-Perception Profile for Adolescents (SPPA-R) (1) (α = 0.82).

Appearance and eating problems.

Eating problems were measured using a 12-item version of EAT-26 (4) (α = 0.75). Those scoring above 7 were categorized as having eating problems. Body mass index (BMI) (kg·m−2) was based on self-report. Two measures of physical appearance were included. The Body Areas Satisfaction Scale (BASS) (4) is a 7-item measure that asks for ratings of satisfaction with specific body parts: face, lower torso, midtorso, upper torso, muscle tone, weight, and height (α = 0.81). Perceptions of global physical appearance were measured using the physical appearance subscale in SPPA-R (α = 0.82), and romantic appeal was measured using another subscale in SPPA-R (α = 0.75).

Problem behavior.

Conduct problems were measured by ratings of involvement in 13 different types of antisocial or illegal behaviors. A measure of conduct problems approximating the diagnostic criteria for conduct disorder in the Diagnostic and Statistical Manual of Mental Disorders (3rd edition, revised; DSM-III-R) was computed (4). Sexual involvement was measured by asking the adolescents whether they had ever had sexual intercourse, as well as their debut age. Those indicating a sexual debut before the age of 15 were considered to have had an early debut. Drug use was measured by asking respondents to indicate their use of three substances during the preceding 12 months, on a 6-point scale ranging from "0 times" to "more than 50 times." The three substances were cannabis, "hard" drugs, and alcohol (they were asked whether they had "drunk so much that you felt clearly intoxicated"). An open-ended question about the type of hard drugs used revealed the following substances in order of frequency of use: amphetamine, hallucinogens, barbiturates, opiates, and cocaine.

Emotional problems.

Depressed mood was measured by the Depressive Mood Inventory (1,10) (α = 0.82). Participants were initially asked a gateway question about parasuicide, that is, whether they had "Taken an overdose of pills or otherwise tried to harm yourself on purpose." The response options were "no, never," "yes, once," or "yes, several times." To detect more serious attempts, they were then asked "Have you ever tried to kill yourself?" with the same response options. Responses to the latter question were used in the present study. The subjects were also asked to give the date of the most recent attempt.


Because no specific theoretical models were tested, a stepwise procedure with backward elimination according to the likelihood ratio test was applied for selection of risk factors. The likelihood ratio (−2 of the log of the likelihood that the observed values of the dependent may be predicted from the observed values) test provides a chi-square for the difference in likelihood ratios for the full model compared with a model where the variable in question is withdrawn from the model. T1 and T2 measures of the same variable are expected to correlate highly, which may cause multicollinearity problems. There is no ready method for determining the extent of this problem in logistic regression. As a proxy, an OLS regression was run, and when the square root of the variance inflation factor was greater than 2, these variables were removed from the model one at a time before the main stepwise procedure was run.


Twenty-four persons (0.8%) reported that they had used AAS for the first time between T2 and T3. Another two persons had used AAS before T2 as well, whereas additional 31 persons had discontinued their use after T2. Hence, only 6.5% of the AAS users at T2 continued their use into the observation period. In all, 1.9% of the sample reported AAS use at some time.

As can be seen in Table 1, risk factors sampled at T1 and T2 generally provided similar results. Both sports-specific variables (power sports involvement) and problem behavior (conduct problems, cannabis use, alcohol intoxication, and sexual debut before the age of 15) were predictive of AAS use at T3, but not appearance-related variables. The same variables generally predicted being offered AAS at T3, but in addition, being offered cannabis (but not using it) and hard drug use were predictive as well. Moreover, appearance-related variables (high BMI, high perceived physical appearance, and high satisfaction with body parts) were also predictive of being offered ASS.

Odds for AAS use and being offered AAS at T3 according to exposure to risk factors at T1 and T2, respectively.

Young age, male gender, previous AAS use, and previous AAS offers were predictive of later AAS use and later AAS offers (Table 1), respectively, and were therefore controlled in all analyses by forced entry as covariates. Participating in power sports at T1 and being intoxicated 50+ times the previous year at T1 added to the prediction of AAS at T3 over and beyond the effect of age, gender, and previous AAS use (Table 2). In addition to the effects of age, gender, and previous AAS offers, participating in power sports at T1 also predicted AAS offers between T2 and T3, as did being offered cannabis at T2 (Table 3).

Predictors of AAS use between T2 and T3. Multivariate logistic regression.
Predictors of being offered AAS between T2 and T3. Multivariate logistic regression.

After controlling for age and gender, a potential effect of AAS use at T2 on conduct problems, alcohol intoxications, drug use, depressive symptoms, eating problems, and suicide attempts at T3, respectively, was evaluated controlling for the level of these problems at T2, respectively. AAS use at T2 was found to be protective of later high-frequent alcohol intoxications (50+ times), adjusted odd ratio (aOR)= 0.12, 95% CI: 0.015-0.99; otherwise, AAS use was not predictive.


During the present 5-yr follow-up, 0.8% had their AAS debut, whereas 1.9% reported AAS use at some time during adolescence or early adulthood. Apart from male gender, comparatively young age, and prior AAS use, being involved in power sports and frequent alcohol intoxications predicted later AAS use. Being offered AAS during follow-up was predicted by male gender, young age, prior offers of AAS, power sports involvement, and having been offered cannabis. AAS use did not alter the risk for a variety of later problems except lowering the risk for later frequent alcohol intoxications.

This study is the first to present prospective data on the predictors of AAS use. The vast majority of population studies of AAS have examined students, predominantly in high school (1,4,12), although some have studied college students (1). Hence, we know little about the age trend in use beyond the teenage years. This study indicates that the point prevalence of AAS use is diminishing during early adulthood, mimicking the age-crime curve. This is in accordance with viewing AAS use in the general population as a behavior closely related to conduct-type problems including antisocial behavior and illicit drug use, more so than being appearance or performance related (22,28). Previous AAS use did predict future AAS use. However, the stability in AAS use may best be described as modest, with only 6.5% continuing their AAS use. Thus, the majority of AAS users most likely use the drug for a limited period, possibly restricting it to experimental use only. Numerous studies have shown higher levels of AAS use among those involved in power sports and strength training compared with other sports or no sports (6,15). This is not surprising because the main effect of AAS on muscle hypertrophy is most likely associated with increased response to strength training.

Lifetime prevalence of AAS use was lower in the present study than figures reported in U.S. studies, which typically range from 3 to 7%, and is in the lower end of the range of prevalence of AAS among adolescents originating from other Western countries (typically 2-4%) (1). However, the present figure represents lifetime prevalence in young adults, cumulated across measurement points. Most other studies have examined participants once during adolescence only, and lifetime prevalence reported among adults should therefore be expected to be higher than that stemming from adolescents. Lifetime prevalence of AAS use among Norwegian adolescents have been reported to be lower than that of adolescents from other Western countries (1). When the present lifetime figures for adults in Norway are lower than for adolescents in other Western countries (1), there are reasons to believe that the comparatively low lifetime prevalence of AAS use holds for Norwegian adults as well.

The temporal relation between alcohol intoxication and AAS revealed a paradoxical relation. Alcohol intoxication increased the risk of future AAS use, whereas AAS use decreased the risk of future alcohol intoxications. Cross-sectionally, alcohol use has been positively associated with AAS use (1). There is no ready explanation to the seemingly protective effect of AAS use. However, the weight training associated with AAS use might in turn imply a change of milieu for the adolescent. Alcohol use among peers has a causal impact upon one's own alcohol use (1). Body building, as well as high-intensity weight-training milieus, might emphasize self-control, high work ethics, and strict eating regimens. Heavy alcohol use might compromise the strict diet and training regimen involved in this type of training and, thus, the perceived effects of such training. AAS might therefore entail less alcohol intoxication among some AAS users, not attributable to the substance itself but rather to the strict training and diet regimen. Prospective studies of larger samples of AAS users will be necccesary to test this.

This finding should be interpreted with some caution because there were multiple statistical tests involved, and the present finding did border on significance. However, the present prospective study did not support the numerous correlations found between AAS use and psychological and social problems in cross-sectional studies (3,6).

Although the present study had many strengths, among them a large national probability sample with favorable response rates and a prospective design, several limitations should be noted. AAS use is a fairly infrequent phenomenon in Norway. Hence, the power to detect true associations was restricted. The low frequency of AAS use also make the estimates susceptible to unsystematic errors stemming from coding errors on behalf of the respondents or the coders. Several steps were taken to minimize this problem. Questionnaires were checked manually for obviously humorous or incorrect answers, and these were optically read. Moreover, the correlates of AAS were similar to those found in countries where the prevalence of AAS is much higher, supporting the validity of our findings. Although previous studies have found self-report of AAS use to have adequate test-retest reliability and to have acceptable validity compared with urine samples, self-reports have inherent validity and reliability problems. The latter, however, would merely dilute our findings.

It should be mentioned that Norwegian society might differ in several important respects from other Western countries. Major sports are those that place emphasis on technical skills (e.g., soccer, team handball, skiing) or endurance (e.g. cross-county skiing) and less on explosive strength. The rate of cannabis use and the use of hard drugs are much lower than in most other Western countries. Hence, predictors of AAS use might be different in other cultures with higher rates of AAS, a higher prevalence of power sports, and a different pattern of illicit drug use. Moreover, it is important to note that questions about AAS were introduced to the study at T2. The stability and predictors of AAS use were therefore not measured in early adolescence, and risk factors and stability of AAS use may be different from the risk factors and stability during middle and late adolescence, or from late adolescence to early adulthood. Future studies should capture risk factors for AAS use as well as change and stability in AAS use during this early phase of adolescence.

This research was funded by grants from the Norwegian Research Council and Social Science Norway.


1. Amtmann, J. Self-reported training methods of mixed martial artists at a regional reality fighting event. J. Strength Cond. Res. 18:194-196, 2004.
2. American Psychiatric Association. A. P. Diagnostic and Statistical Manual of Mental Disorders, 3rd - Revised ed, Washington DC: American Psychiatric Association, pp. 1-567, 1987.
3. Bahrke, M. S., and E. Yesalis. Weight training - a potential confounding factor in examining the psychological and behavioral-effects of anabolic-androgenic steroids. Sports Med. 18:309-318, 1994.
4. Bahrke, M. S., C. E. Yesalis, and K. J. Brower. Anabolic-androgenic steroid abuse and performance-enhancing drugs among adolescents. Child Adolesc. Psychiatr. Clin. N. Am. 7:821-838, 1998.
5. Bahrke, M. S., C. E. Yesalis, A. N. Kopstein, and J. A. Stephens. Risk factors associated with anabolic-androgenic steroid use among adolescents. Sports Med. 29:397-405, 2000.
6. Beel, A., B. Maycock, and N. McLean. Current perspectives on anabolic steroids. Drug Alcohol Rev. 17:87-103, 1998.
7. Brown, T. A., T. F. Cash, and R. J. Lewis. Body-image disturbances in adolescent female binge purgers - a brief report of the results of a National Survey in the USA. J. Child Psychol. Psychtr. Allied Disc. 30:605-613, 1989.
    8. Cole, J. C., R. Smith, G. Halford, and G. F. Wagstaff. A preliminary investigation into the relationship between anabolic-androgenic steroid use and the symptoms of reverse anorexia in both current and ex-users. Psychopharmacology 166:424-429, 2003.
    9. Copeland, J., R. Peters, and P. Dillon. Anabolic-androgenic steroid use disorders among a sample of Australian competitive and recreational users. Drug Alcohol Depend. 60:91-96, 2000.
    10. Curran, P. J., E. Stice, and L. Chassin. The relation between adolescent alcohol use and peer alcohol use: a longitudinal random coefficients model. J. Consult. Clin. Psychol. 65:130-140, 1997.
    11. Durant, R. H., V. I. Rickert, and C. S. Ashworth. Validity of adolescents' self-report of anabolic steroid use. N. Eng. J. Med. 329:889, 1993.
      12. Durant, R. H., V. I. Rickert, C. S. Ashworth, C. Newman, and G. Slavens. Use of multiple-drugs among adolescents who use anabolic-steroids. N. Eng. J. Med. 328:922-926, 1993.
      13. Garner, D. M., M. P. Olmsted, Y. Bohr, and P. E. Garfinkel. The eating attitudes test - psychometric features and clinical correlates. Psychol. Med. 12:871-878, 1982.
      14. Gottfredson, M., and T. Hirschi. A General Theory of Crime, Stanford, CA: Stanford University Press, pp. 1-297, 1990.
        15. Green, G. A., F. D. Uryasz, T. A. Petr, and C. D. Bray. NCAA study of substance use and abuse habits of college student-athletes. Clin. J. Sport Med. 11:51-56, 2001.
        16. Gridley, D., and S. Hanrahan. Anabolic-androgenic steroid use among male gymnasium participants: dependence, knowledge and motives. Sport Health 12:11-14, 1994.
          17. Gruber, A. J., and H. G. Pope. Psychiatric and medical effects of anabolic-androgenic steroid use in women. Psychother. Psychosom. 69:19-26, 2000.
          18. Hallagan, J. B., L. F. Hallagan, and M. B. Snyder. Anabolic androgenic steroid use by athletes. N. Eng. J. Med. 321:1042-1045, 1989.
            19. Harter, S. Manual for the Self-Perception Profile for Adolescents, Denver, CO: University of Denver, pp. 1-198, 1988.
              20. Hibell, B., B. Andersson, S. Ahlstrom, et al. The 1999 ESPAD Report. Alcohol and other drug use among students in 30 European countries, Stockholm: Swedish Council for Information on Alcohol and Other Drugs, pp. 1-383, 2000.
                21. Kandel, D. B., and M. Davies. Epidemiology of depressive mood in adolescents - an empirical-study. Arch. Gen. Psychiatry 39:1205-1212, 1982.
                22. Kindlundh, A. M. S., D. G. L. Isacson, L. Berglund, and F. Nyberg. Factors associated with adolescent use of doping agents: anabolic-androgenic steroids. Addiction 94:543-553, 1999.
                23. Lavik, N. J., S. E. Clausen, and W. Pedersen. Eating behavior, drug-use, psychopathology and parental bonding in adolescents in Norway. Acta Psychiatr. Scand. 84:387-390, 1991.
                24. Pope, H. G., and D. L. Katz. Psychiatric and medical effects of anabolic-androgenic steroid use - a controlled-study of 160 athletes. Arch. Gen. Psychiatry 51:375-382, 1994.
                25. Pope, H. G., E. M. Kouri, and J. I. Hudson. Effects of supraphysiologic doses of testosterone on mood and aggression in normal men - a randomized controlled trial. Arch. Gen. Psychiatry 57:133-140, 2000.
                26. Riem, K. E., and K. G. Hursey. Using anabolic-androgenic steroids to enhance physique and performance-effects on moods and behavior. Clin. Psychol. Rev. 15:235-256, 1995.
                27. Vanhelder, W. P., E. Kofman, and M. S. Tremblay. Anabolic-steroids in sport. Can. J. Sport Sciences-Revue Canadienne Des Sciences Du Sport 16:248-257, 1991.
                  28. Wichstrom, L. The emergence of gender difference in depressed mood during adolescence: the role of intensified gender socialization. Dev. Psychol. 35:232-245, 1999.
                  29. Wichstrom, L. Harters self-perception profile for adolescents-reliability, validity, and evaluation of the question format. J. Pers. Assess. 65:100-116, 1995.
                  30. Wichstrom, L., and W. Pedersen. Use of anabolic-androgenic steroids in adolescence: winning, looking good or being bad? J. Stud. Alcohol 62:5-13, 2001.
                  31. Wichstrom, L., K. Skogen, and T. Oia. Increased rate of conduct problems in urban areas: what is the mechanism? J. Am. Acad. Child Adolesc. Psych. 35:471-479, 1996.
                  32. Yesalis, C. E., C. K. Barsukiewicz, A. N. Kopstein, and M. S. Bahrke. Trends in anabolic-androgenic steroid use among adolescents. Arch. Pediatr. Adolesc. Med. 151:1197-1206, 1997.
                  33. Yesalis, C. E., R. T. Herrick, W. E. Buckley, K. E. Friedl, D. Brannon, and J. E. Wright. Self-reported use of anabolic-androgenic steroids by elite power lifters. Phys. Sportsmed. 16:90, 1988.


                    ©2006The American College of Sports Medicine