Long-distance running may be associated with compulsive behavior(33) and has been used as therapy for depression and alcoholism (20,28). However, little is known about drinking habits, problem drinking behavior, and the prevalence (or history of) alcohol abuse in recreational long-distance runners.
In the general population. Harburg and colleagues found problems associated with drinking occurred in 16% of males and 5% of females (mean age 39 ± 11.5 yr) (9). The Epidemiologic Catchment Area Study(10) revealed alcohol abuse and/or dependence in 24% of males and 5% of females, but this varied by age, with younger age groups(<45 yr) having higher rates than older adults. There is a strong familial component, with 15-16% of current drinkers reporting a family history of alcoholism among first-, second-, or third-degree relatives(3). Binge drinking (five or more drinks at least once in 2 wk) is associated with problem behavior (drunk driving, violence, risky sexual behavior) (30) and is common in college students(13). Although alcohol may be used to reduce anxiety prior to competition, it has no ergogenic properties, and it may have deleterious effects on athletic performance (1).
Studies relating alcohol consumption to athletic participation have had mixed results. Collegiate athletes report a greater quantity of alcohol consumed and are more likely to drive while intoxicated than their nonathlete peers (17). Others found athletes have more negative attitudes toward alcohol consumption (19) or greater anabolic steroid use than nonathlete controls, but without distinct differences in drinking behaviors or other substance use(4). In contrast, young Italian athletes drank less and reported fewer heavy drinking episodes than nonathletes(5), and older, more active men reported less alcohol consumption than their sedentary counterparts (8).
Because of reported deleterious effects of alcohol on athletic performance and therapeutic benefits of running in alcoholism therapy, we were interested in the relationships between running and drinking habits, parental history of problem drinking, and responses to alcoholism screening tests. We hypothesized that there would be less drinking, especially excessive drinking, in serious recreational racers, and that running might be a substitute for drinking in those at risk, either because of a parental history of problem drinking or because of self-report of problem behaviors on an alcoholism screening test.
Questionnaire and Data Collection
The mailed questionnaire contained items relating to marital and educational status, running and exercise habits, and drinking behaviors. These included self-report of alcoholism and/or alcoholism in recovery, and a quantity-frequency scale of alcohol use in the previous 2 wk. A drink was defined as a 12-oz (360-ml) can of beer, a 4-oz (120-ml) glass of wine, or a shot (1.25-oz (37-ml) of liquor straight or in a mixed drink, or a 12-oz(360-ml) wine cooler. Self-reports of alcohol use have been found to be reliable and consistent (31).
Modified versions of the Michigan Alcoholism Screening Test, the 10-item BMAST and a 12-item SMAST (eight of the BMAST questions and four questions dealing with guilt about drinking, ability to stop drinking, problems with relatives because of drinking, and arrest for drunken behavior) were included(21,24,25). Questions in the BMAST(21) are weighted and explore a subject's perception of his or her drinking problem, such as “Do you feel you are a normal drinker (by normal, we mean do you drink less or the same as most other people)?” as well as the consequences of problem drinking, such as drunk driving arrests. A score of ≥6 has been reported as a cutoff score for subjects with behaviors suggestive of problem drinking(21). Scoring of the SMAST is unweighted, and a score of≥3 is thought to be associated with problem drinking(25). Though primarily developed for and used in clinical settings, these instruments have demonstrated potential usefulness in screening populations (9). Participants were also asked if they had a biologic parent and/or sibling who had a problem with alcohol(3). To enhance the response rate, the questionnaire was kept short and a postage-stamped return envelope (26) was included with the mailing. In addition, registrants could reply anonymously to enhance the response rate and as an attempt to increase the accuracy of drinks and drinking habits reported (31).
Questionnaires were mailed 1 month post-race (mid-June) to all 1578 people registering for the 1993 Syttende Mai, a 20-mile run/walk from Madison to Stoughton, WI. This race attracts serious recreational runners from the immediate Madison-Dane County area, other parts of Wisconsin, and the surrounding states. Approximately half of the participants are local. Questionnaires were also mailed (mid-September) to 1000 randomly selected gender and age-matched clients enrolled in a family practice health maintenance organization (Dean Medical Center, Madison, WI). Written informed consent from the subjects was obtained by their return of the questionnaire.
The response rate for the race participants was 56% with 486 men (407 runners, 79 walkers) and 398 women (149 runners, 249 walkers) returning questionnaires (one mailing). The control group's response rate was 48%, with 241 men and 237 women returning information. Total number of responders was 884 racers and 478 controls, with women race participants and controls having higher response rates than men. Among the control population, there were 19 men and 9 women who ran (but not in the 1993 Syttende Mai) and 64 subjects (47 women) who did fitness walking. In addition, there were 118 control respondents who reported some other endurance activity such as biking or aerobics. More than half of the control population reported no endurance activity.
Since our research question primarily addressed the differences in drinking habits between those who run and those who do not, subsequent analyses were limited to those race participants who ran the race and those control subjects who reported no endurance activity. Lifelong abstainers were also excluded from data analysis (men: 13 racers, 3 controls; women: 10 racers, 9 controls). Thus, the study population consisted of 397 male runners, 138 male controls, 144 female runners, and 119 female controls (total, 798). The racers' running habits are listed in Table 1.
Because more runners had attended college and were single and younger(Table 2), we analyzed for the effects of level of education, marital status, and age as well as gender. Single, separated, and divorced were grouped together as one category ('single'); and married and those with partners another ('married'). Respondents' educational level was classified as 'high-school graduate or less' and 'some college experience or more.' In addition, we randomly selected a subset of both runners and controls matched for gender, age, educational level, and marital status. This primary subset consisted of 104 matched pairs of men and 84 pairs of women. A second subset of these 188 pairs was obtained by matching by parental history (89 pairs).
Computations and Statistical Procedures Used in Data Analysis
An estimate of the total drinks consumed during the 2 wk was calculated based on the number of drinks reported per occasion and the number of occasions of drinking during these 2 wk. For those occasions that were collapsed on the questionnaire (i.e., 3-5, 6-9) we chose the interval midpoint as the number of occasions (e.g., 4, 7.5). This total is an underestimate of the amount consumed, since the highest number of occasions a subject could report was 10 or more; we chose 10 occasions for this category. Binge drinking was defined as consuming 5 or more drinks in a row at least once in the past 2 wk.
Weighted BMAST scores were computed (7 of the 10 questions are worth 2 points e.g., [“Have you ever been arrested for drunk driving?”]; 3 questions are worth 5 points if answered affirmatively e.g., [“Have you ever gone to anyone for help about your drinking?”]) with a weighted score of ≥6 suggesting problem drinking behavior(21). Qeustions from the SMAST were scored unweighted and a total score computed (a score of ≥3 reflecting problem drinking behavior (25).
Frequency counts and chi-square analysis using the cross-tabulation procedure were performed for nominal data. Means and standard deviations were calculated for continuous variables. Analysis of variance (ANOVA) was used to compare total drinks, occasions of drinking, and number of binge drinking occasions among runners versus controls by gender, parental history, and sibling history, as well as scores on the BMAST and SMAST (with marital status, educational level, and age as covariates). We analyzed the differences in total number of drinks, number of episodes of drinking, and number of occasions of binge drinking in both subsets by paired t-test. We also used ANOVA to study the effect of parental history of problem drinking and the effect of an elevated score on the alcoholism screening tests on these three dependent variables in the primary subset. We used the Statistical Package for the Social Sciences (SPSS, Inc., Chicago, IL, 1990). An alpha level of 0.05 was chosen for statistical significance for all research questions.
Main Analysis: Responses to BMAST and SMAST Items
Table 3 compares the response rates for problem drinking behaviors for runners versus controls. Male runners were almost twice as likely to report feeling guilty about their drinking than male controls(26.6% vs 13.8%, chi-square = 9.41, P = 0.002). Control women reported drinking caused more problems with their spouse, parent, or other relative as compared with the women runners (chi-square = 4.1, P = 0.04). There were no differences between runners versus controls in the proportion of those with traditional BMAST or SMAST scores suggestive of alcoholism. In the total sample population, 8% of men and 4% of women had BMAST scores ≥6 (chi-square = 4.37, P = 0.04); 26.3% of men and 13.2% of women, SMAST scores ≥3 (chi-square = 16.83, P < 0.001). Among men, 7.4% of runners and 9.7% of controls had a BMAST score of≥6; 25.3% of runners and 29.3% of controls had a SMAST score of ≥3. In women, 2.2% of runners, 6.1% of controls had BMAST scores ≥6; 9.6% of runners and 17.5% of controls had SMAST scores ≥3. Of the 36 subjects who reported that they were alcoholics (19 male runners [4.8%], 9 male controls[6.5%], 4 female runners [2.8%], 4 female controls [3.4%]), 80.6% stated they were in recovery. Self-report of alcoholism or alcoholism in recovery was not different for runners versus controls. The mean SMAST score for these 36 individuals was 7.2 ± 2.8; mean BMAST score, 17.9 ± 8.4; no differences were noted between runners and controls.
Overall, 29.1% of the men and 35.5% of the women reported they had a biologic parent with problem drinking; 25% of men and 28.2% of women reported having a biologic sibling with problem drinking. There were no statistically significant differences by chi-square analyses, between runners and controls, subjects with different educational levels, or single versus married status, in probability of reporting either a parent or sibling with problem drinking.
Table 4 compares estimated total number of drinks, number of occasions of drinking, and number of occasions of binge drinking in a 2-wk period for runners versus controls, stratified by gender and parental history of problem drinking with educational level, marital status, and age as covariates. There were no differences in total drinks consumed between runners and controls. Runners were more likely to have a greater number of occasions of drinking than the control population (F = 4.94, P = 0.027). Men consumed more drinks and reported a greater number of drinking occasions, including binge occasions, than women.
Single or student status were the most likely predictors of binge drinking. One-third of single men reported binge drinking as contrasted with 23.4% of married men (chi-square = 5.56, P = 0.018); 19.8% of single women, 9.2% of married women (chi-square = 5.46, P = 0.019). Current students reported similar numbers: 32.5% vs 20.7% in nonstudents (chi-square = 5.497, P = 0.019). Race participation versus controls, educational level, or age did not make a difference in reporting binge drinking.
Alcohol and Sports Performance
In men, controls were more likely to report that alcohol consumption made no difference in their regular physical activity and sports performance. This contrasted with runners who were more likely to report not knowing what the effect of alcohol was on their running and race performance (chi-square = 8.77, P = 0.03). There were no differences in women.
Associations among Variables
For all participants, the presence of a parental history of problem drinking was associated with more total drinks consumed and number of occasions of drinking (Table 4). Men with a parental or sibling history of problem drinking, regardless of race status, reported at least twice as many occasions of binge drinking than either group of women(F = 6.67, P = 0.01).
Higher BMAST or SMAST scores were associated with parental history of problem drinking in men (runners: BMAST, chi-square = 14.4, P < 0.001; SMAST, chi-square = 16.2, P < 0.001; controls: BMAST, chi-square = 7.96, P = 0.019). Only in female runners was an elevated BMAST score associated with parental history of problem drinking(chi-square = 6.34, P = 0.04). A history of sibling problem drinking was similarly associated with higher BMAST or SMAST scores in men (runners: BMAST, chi-square = 18.1, P < 0.001; SMAST, chi-square = 14.3,P < 0.001; controls: BMAST, chi-square = 6.19, P = 0.04) but not in women.
When we divided the population of runners and controls into those with a BMAST score of <6 and those with a score of ≥6 (Table 5), we noted a significant interaction (F = 6.20, P = 0.013) between runners and controls and level of BMAST score. Runners with higher scores reported fewer occasions of drinking than controls; those runners with a score of <6 reported more occasions of drinking than the controls. Using the SMAST score of 3 as a cutoff point, this interaction was seen only in the women runners in reporting total alcohol consumed(Table 6). All subjects with elevated SMAST scores ≥3 reported a greater number of occasions of binge drinking than those with SMAST scores ≤2 (1.2 ± 2.4 vs 0.4 ± 1.0, F = 23.65,P < 0.001) (gender effect, NS; runner vs control, NS).
In the analysis of matched pairs, male runners reported more total drinks consumed than their paired controls (14.2 ± 19.6 vs 5.4 ± 7.6,P = 0.004, N = 49 pairs) (Table 7). This was also true for the male runners when matched for parental history of problem drinking as well (20.3 ± 26.0 vs 6.1 ± 8.8, P= 0.027, N = 21 pairs). Both men and women runners reported more occasions of drinking than their paired controls (2.8 ± 2.7 vs 2.0± 2.3, P = 0.004, N = 178 pairs). Stratification of the matched pairs by level of BMAST score (<6, ≥6) revealed a similar trend in the interaction noted earlier (Tables 5 and 6) in number of occasions reported; i.e., runners with low BMAST scores were more likely to report more occasions than their matched controls, but those runners with high BMAST scores reported less (F = 3.60,P = 0.058). Feelings of guilt about drinking were again reported by 26.2% of male runners, as compared with 13.6% of the matched controls(chi-square = 5.15, P = 0.02).
In this study, runners reported more alcohol consumption than matched controls except in those runners who had BMAST or SMAST scores suggestive of problem drinking and/or alcoholism. These runners drank less. Long-distance running 30 miles·wk-1, 11 months·yr-1, did not appear to diminish reported alcohol consumption in the majority of the runners, especially in those at risk for alcoholism because of parental history of problem drinking. As in other studies(11,12,15) male gender was a predictor of total alcohol consumption. Single or student status predicted binge drinking.
Yates and colleagues (33) describe `obligatory runners' as middle-age men with a voracious desire for activity, inability to moderate involvement even in the face of injury, and who suffer signs of withdrawal when they are prevented from running. It is not clear if'obligatory running' is really an addiction or a compulsion, or either(33), or whether addiction is a sin, disease, or maladaptive behavior (29). However, it is clear from our study that there are long-distance runners who possess certain characteristics similar to those at risk for problem drinking. This may explain the increased alcohol consumption in the male runners in this study. This does not support our hypothesis that there would be less drinking in a sample of long-distance runners.
Conversely, our finding that runners with elevated scores on an alcoholism screening test drink less than the control subjects with similar scores (as well as other runners) may be evidence in support of our hypothesis. That is, running may be a healthy substitute in a person prone to alcoholic behavior and may aid in recovery. We were unable to ascertain if running was the cause of the decreased alcohol consumption or if the decreased intake was present before the subject began to run. In addition, men runners had lower prevalence rates on 12 of the 14 items from the MAST than men controls, and similarly for women (10 of 14) (Table 3). This may suggest that runners have fewer drinking-behavior problems or the control population more.
Parental problem drinking is associated with more severe alcohol abuse on almost all factors of alcohol use (32). This includes amount of alcohol consumed as well as alcohol-related problems (social, vocational, physical, loss of control) independent of age of alcoholism, current age, socioeconomic background, or marital status(32).
The overall prevalence of a score suggestive of problem drinking behavior on one of the screening tests was similar to prevalence rates reported in the general population (9,11). Lower prevalence rates with the BMAST as compared with the SMAST may reflect greater sensitivity of the SMAST questions to detect problem behaviors (25). The difference between our study's prevalence rates of problem drinking, that is, self-report of alcoholism, versus a “score suggestive of problem drinking” on the screening tests, emphasizes the denial frequently found in those who have problems with alcohol (16). Though found not to be significantly different in our study, the lower prevalence rate in women runners (vs women controls) of an elevated score on one of the MAST tests suggests that further study of this issue in a larger group of women might be warranted. Do women attracted to running have fewer problem behaviors associated with alcohol?
Runners reported more occasions of drinking than controls. College-educated people are more likely to drink lightly but more frequently(15). However, our primary subset of 188 pairs was matched for educational level to eliminate this as a confounder. The increased prevalence of binge drinking in students is not surprising, and is similar to other reports (30). Our findings are in contrast to those of Schneider and Greenberg (22) who found that those in endurance sports such as running, jogging, and fast-walking were less likely to consume large quantities of alcohol as compared with those participating in team sports. Similarly, a study of Finnish endurance athletes(long-distance runners, cross-country skiers) revealed less smoking and alcohol consumption (amount of alcohol·wk-1) than the reference sedentary group (6).
Guilt about drinking is a classic characteristic of alcoholism(2). Harburg and colleagues reported 19% of men and 6% of women feeling guilty about drinking (9). The reporting of guilt may predate the development of addiction (18). Guilt about drinking is also a strong predictor of hangovers(10). We found that male runners had a much higher abnormal response rate (26%) on this MAST item. Though increased health awareness (concern about following a healthy lifestyle) is associated with a college education (7), guilt about drinking was also reported by 26% of the men runners in the subset of pairs matched by level of education.
Limitations of this study include nonresponders, time of administration of questionnaire to race participants, and certain questionnaire items. Denial is common among those with drinking problems (16) and may be reflected in those who chose not to respond to the questionnaire. However, the response rates of 56% for race participants and 48% for the control population should ensure that responders were a reasonable representation of the population. The size of the control population was also smaller than anticipated, due to nearly half of them reporting some form of endurance exercise, and thus being eliminated as a contrast group for this study.
The questionnaire was administered approximately one month after the race. Since men runners reported not knowing how alcohol affected their race performance, it is conceivable that runners may have altered their drinking patterns before the race to improve their performance, and returned to regular habits post-race, even though the runners report running nearly year round. Ultra-marathoners of approximately the same age (40 yr) and running 68 miles·wk-1 reported higher alcohol intakes in their usual diets as compared with prerace intakes (27). University athletes use alcohol frequently, but use tends to decrease during the competitive season (23). Respondents may not have used the same 2-wk period for reporting current alcohol consumption. For some respondents this 2-wk period may have included a holiday or other major event associated with a higher alcohol intake and not for others. Lower alcohol use is reported after December as compared with November(14). This kind of bias seems unlikely, since most of the race respondents replied in June and the controls responded in September-October, and no questionnaires were received in January.
The questions on the alcoholism screening tests (BMAST, SMAST) used in our survey reflect lifetime drinking behaviors and not necessarily current ones. It is possible that long-distance running could have had an impact on problem drinking behaviors beginning with the onset of running. We did not ask when the problem drinking behaviors occurred in relation to the onset of regular running. Questions reflecting family history of problem drinking were limited to only one for parents and another for siblings, and may limit our interpretations of such.
In summary, results of the present study suggest that long-distance runners without scores on a screening test suggestive of problem drinking drank more than matched controls, whereas those runners with scores suggestive of problem drinking drank less. The present data do not address whether running helps the alcoholic in recovery maintain a sober state. Further studies comparing runners who are recovering alcoholics with an appropriate control group could provide a meaningful group of subjects to explore this issue.
1. American College of Sports Medicine. Position stand: the use of alcohol in sports. Med. Sci. Sports Exerc.
2. Cloninger, C. R., S. Sigvardsson, S. B. Gilligan, A. L. VON Knorring, T. Reich, and M. Bohman. Genetic heterogeneity and the classification of alcoholism. Adv. Alcohol Subst. Abuse
3. Dawson D. A., T. C. Harford, and B. F. Grant. Family history as a predictor of alcohol dependence. Alcohol. Clin. Exp. Res.
16: 572-575, 1992.
4. Dezelsky, T. L., J. V. Toohey, and R. S. Shaw. Non-medical drug use behavior at five United States universities: a 15-year study. Bull. Narc.
5. Donato, F., D. Assanelli, M. Marconi, C. Corsini, G. Rosa, and S. Monarca. Alcohol-consumption among high-school students and young athletes in North Italy. Rev. Epidemiol. Sante Publique
6. Fogelholm, M., J. Kaprio, and S. Sarna. Healthy lifestyles of former Finnish world class athletes. Med. Sci. Sports Exerc.
26: 224-229, 1994.
7. Frauman, A. C. and B. Nettles-Carlson. Predictors of a health-promoting life-style among well adult clients in a nursing practice.J. Am. Acad. Nurse Pract.
8. Hallfrisch, J., D. T. Drinkwater, D. C. Muller, et al. Physical conditioning status and diet intake in active and sedentary older men. Nutr. Res.
9. Harburg, E., R. Gunn, L. Gleiberman, P. Roeper, W. DiFranceisco, and R. Caplan. Using the Short Michigan Alcoholism Screening Test to study social drinkers: Tecumseh, Michigan. J. Stud. Alcohol
10. Harburg, E., R. Gunn, L. Glieberman, W. DiFranceisco, and A. Schork. Psychosocial factors, alcohol use and hangover signs among social drinkers: a reappraisal. J. Clin. Epidemiol.
11. Helzer, J. E., A. Burnam, and L. T. McEvoy. Alcohol abuse and dependence. In: Psychiatric Disorders in America. The Epidemiologic Catchment Area Study
, L. N. Robins and D. a. Regier (Eds.). New York: The Free Press (Macmillan, Inc.), 1991, pp. 81-115.
12. Hilton, M. E. Drinking patterns and drinking problems in 1984: results from a general population survey. Alcohol. Clin. Exp. Res.
13. Johnston, L. D., P. M. O'Malley, and J. G. Bachman.Drug use among American high school seniors, college students and young adults, 1975-1990. Vol II. College Students and Young Adults.
Rockville, MD: U.S. Department of Health and Human Services, Public Health Service; DHHS publication no. (ADM) 91-1835, 1991.
14. Midanik, L. T. Reliability of self-reported alcohol consumption before and after December. Addict. Behav.
15. Midanik, L. T. and R. Room. The epidemiology of alcohol consumption. Alcohol, Health Res. World
16. Morse, R. M. and D. K. Flavin. The definition of alcoholism. The Joint Committee of the National Council on Alcoholism and Drug Dependence and the American Society of Addiction Medicine to study the definition and criteria for the diagnosis of alcoholism. J.A.M.A.
17. Nattiv, A. and J. C. Puffer. Lifestyles and health risks of collegiate athletes. J. Fam. Pract.
18. O'Connor, L. E. and J. Weiss. Individual psychotherapy for addicted clients: an application of control mastery theory. J. Psychoact. Drugs
19. Overman, S. J. and T. Terry. Alcohol use and attitudes: a comparison of college athletes and non-athletes. J. Drug Educ.
20. Palmer, J., N. Vacc, and J. Epstein. Adult inpatient alcoholics: physical exercise as a treatment intervention. J. Stud. Alcohol
21. Pokorny, A. D., B. A. Miller, and H. B. Kaplan. The brief MAST: a shortened version of the Michigan Alcoholism Screening Test.Am. J. Psychiatry
22. Schneider, D. and M. R. Greenberg. Choice of exercise: a predictor of behavioral risks? Res. Q. Exerc. Sport
23. Selby, R., H. M. Weinstein, and T. S. Bird. The health of university athletes: attitudes, behaviors, and stressors. J. Am. Coll. Health
24. Selzer, M. L. The Michigan Alcoholism Screening Test: the quest for a new diagnostic instrument. Am. J. Psychiatry
25. Selzer, M. L., A. Vinokur, and L. van Rooijen. A self-administered Short Michigan Alcoholism Screening Test (SMAST). J. Stud. Alcohol
26. Shiono, P. H. and M. A. Klebanoff. The effect of two mailing strategies on the response to a survey of physicians. Am. J. Epidemiol.
27. Singh, A., P. Evans, K. L. Gallagher, and P. A. Deuster. Dietary intakes and biochemical profiles of nutritional status of ultramarathoners. Med. Sci. Sports Exerc.
28. Taylor, C. B., J. F. Sallis, and R. Needle. The relation of physical activity and exercise to mental health. Public Health Rep.
29. Thombs, D. L. Introduction to Addictive Behaviors
. New York: Guilford Press, 1994, pp. 1-6.
30. Wechsler, H., A. Davenport, G. Dowdall, B. Moeykens, and S. Castillo. Health and behavioral consequences of binge drinking in college. J.A.M.A.
31. Werch, C. E. Two procedures to reduce response bias in reports of alcohol consumption. J. Stud. Alcohol
32. Worobec, T. G., W. M. Turner, T. J. O'Farrell, H. S. Cutter, R. D. Bayog, and M. T. Tsuang. Alcohol use by alcoholics with, without a history of parental alcoholism. Alcoholism
33. Yates, A., C. M. Shisslak, J. Allender, M. Crago, and K. Leehey. Comparing obligatory to nonobligatory runners.Psychosomatics