Exercise and Depression in Midlife: A Prospective Study Lisa Cooper-Patrick, MD, MPH, Daniel E. Ford, MD, MPH, Lucy A. Mead, ScM, Patricia P. Chang, MD, and Michael J. Klag, MD, MPH Introduction Measurements It has been suggested that exercise is associated with better mental health. '-3 However, interventional studies have many methodologic problems, including the nonrandom assignment of participants to experimental conditions and the inability to blind participants to a physical activity intervention.3'4 Prospective observational studies have produced contradictory re- sults.57 We hypothesized that (1) physical activity is protective against depression and psychiatric distress and (2) this protective effect is accentuated in vulner- able individuals, specifically those with an "unstable" temperament or with a paren- tal history of depression. Methods Study Population The Precursors Study is a longitudi- nal cohort study of 1337 former medical students from the Johns Hopkins Univer- sity School of Medicine classes of 1948 through 1964. More than 95% of eligible students were enrolled.8 The study con- ducts annual follow-up by mailed ques- tionnaires, with 5-year response rates ranging from 87% to 93%. The first analysis assessed the rela- tionship between self-reported physical activity in 1978 and the incidence of clinical depression through 1993. Partici- pants who died before 1978 (n = 61), did not respond to the 1978 questionnaire (n = 465), reported an episode of depres- sion in 1978 or earlier (n = 52), or were lost to follow-up after 1978 (n = 7) were excluded from this analysis, leaving 752 eligible participants. The second analysis assessed the relationship between self-reported physi- cal activity in 1986 and psychiatric distress by the General Health Question- naire in 1988. Participants who died before 1986 (n = 106), did not respond to the 1986 questionnaire (n = 290), died in 1987 or 1988 (n = 13), did not complete the questionnaire in 1988 (n = 109), or were lost to follow-up (n = 38) were excluded from this analysis, leaving 781 eligible participants. The main independent variable was the frequency of exercising to a sweat during an average week in either 1978 or 1986, abstracted from the Harvard Alumni physical activity questionnaire.9 This ques- tionnaire has been validated in other studies's'2 and has been shown to predict morbidity and mortality.13-16 We also studied change in physical activity be- tween medical school and midlife, catego- rizing participants as inactive, became active, and remained active. Risk factors for depression were included as covariates in multivariate analyses. Substance abuse was defined as self-reported alcohol/drug abuse or CAGE questionnaire score of 2 or more.'7,18 Socioeconomic status and race are homo- geneous in this cohort; they were not included as covariates. Analyses were stratified by three types oftemperament assessed in medical school and shown to predict mortality in this co- hort: (1) "tension-in," (2) "tension-out," and (3) "stable." 19 Parental history of de- pression, also used as a stratification var- iable, was assessed in medical school and during follow-up by annual questionnaire. Incidence of clinical depression from 1979 through December 31, 1993, was defined by self-report on annual morbidity questionnaires and by review of medical records, with the use of Diagnostic and Statistical Manual of Mental Disorders, 4th edition, (DSM-IV) criteria, when possible, and with grief reactions ex- cluded.20 Psychiatric distress was defined as a score of 4 or more on the 20-item General Health Questionnaire adminis- tered in 1988.21-23 The authors are with the Division of Internal Medicine, Johns Hopkins University School of Medicine, Baltimore, Md. Drs Cooper-Patrick, Ford, and Klag are also with the Department of Health Policy and Management, and Drs Klag and Ford are with the Department of Epidemiol- ogy, Johns Hopkins University School of Hy- giene and Public Health, Baltimore, Md. Requests for reprints should be sent to Lisa Cooper-Patrick, MD, MPH, The Johns Hopkins Precursors Study, 2024 East Monument St, Suite 2-200, Baltimore, MD 21205-2223. This paper was accepted July 9, 1996. April 1997, Vol. 87, No.4 www.cuwai.com Public Health Briefs Analysis To examine the relationship of physi- cal activity in 1978 to the incidence of clinical depression during the subsequent 15 years, Cox proportional hazards mod- els were used. To examine the relationship between physical activity in 1986 and psychiatric distress in 1988, logistic regres- sion analyses were used. For all analyses, two-tailed P values of <.05 indicated statistical significance. Results Characteristics of the study sample for the clinical depression analysis are shown in Table 1. Nonrespondents in 1978 were similar to respondents except that they were more likely to have a physical limitation (35% vs 28%, P < .05) and less likely to have a parental history of depression (8% vs 12%, P < .05). Of the 781 persons in the General Health Ques- tionnaire analysis, 560 (72%) were also included in the clinical depression analy- sis. There were 973 individuals who were included in either the clinical depression or the health questionnaire analysis. The cumulative incidence of depres- sion at a follow-up of 15 years was 6.4%. The prevalence of psychiatric distress (General Health Questionnaire score 2 4) in 1988 was 15%. There was no increased risk of developing depression for those who did not exercise to a sweat relative to those who did, in unadjusted or multivari- ate analyses, whether or not physical activity was considered as a dichotomous or an ordinal variable (Table 2). There was also no increased risk of developing clinical depression for those who were inactive or became active between medi- cal school and 1978 relative to those who remained active. Results for the General Health Questionnaire analysis were simi- lar to those for clinical depression (Table 2). Adjusting for the presence of physical limitations did not influence the results. The risk of developing depression or having psychiatric distress did not differ for exercisers and nonexercisers in any of the strata of temperament or in persons with no parental history of depression (data not shown). Of the 90 individuals with a positive parental history, 9 devel- oped depression, and all were exercisers. The odds of psychiatric distress were higher for nonexercisers compared with exercisers in this stratum (odds ra- tio = 2.68; 95% confidence interval 0.85, 8.48), but the result did not reach statisti- cal significance. Discussion Many interventional studies of vigor- ous exercise training programs have also failed to show a beneficial effect on mental health.2>27 At least two observa- tional studies using population-based male samples have failed to show a protective benefit of exercise for psychiatric dis- tress.5'7 In a sample of medical students, Buchman and colleagues found no rela- tionship between physical activity and depression.28 Most studies showing a relationship between exercise and depres- sion have included such activities as golf and horseback riding, which are strong correlates of higher social status and larger social networks.6'29 The frequency of physical activity in our sample of White male professionals is much higher than in population-based samples. However, the validity of the exercise measure is supported by its association with cardiovascular disease in this cohort.30 The incidence of depression and the prevalence of psychiatric distress in our sample are consistent with other population-based studies.3' Of our clinical depression cases, 86% also reported tak- ing medication or seeing a health profes- sional for their problem (unpublished data by P.P. Chang et al.). Although the long-term effects of physical activity on psychological status have not been established, moderate levels of regular physical activity do American Journal of Public Health 671 TABLE 1-Medical School and Midlife Characteristics, by Exercise in 1978: The Precursors Study of Former Medical Students, Classes of 1948 through 1964 (n = 752) Exercised to a Sweat Weekly in 1978 Total Yes No No. % No. % No. % In medical school Gender** Male 690 92 475 69 215 31 Female 62 8 31 50 31 50 Race White 741 99 499 67 242 33 Asian 11 1 7 64 4 36 Temperament Tension-ina 188 28 119 63 69 37 Tension-outh 117 17 78 67 39 33 Stablec 373 55 257 69 116 31 At midlife Age in 1978** -44 y 243 32 181 74 62 26 45-54 y 425 57 276 65 149 35 >55 y 84 11 49 58 35 42 Parental history of depression* No 662 88 4353 66 227 34 Yes 90 12 71 79 19 21 Substance abuse No 669 89 447 67 222 33 Yes 83 11 59 71 24 29 Smoker in 1978** No 603 83 420 70 183 30 Yes 123 17 69 56 54 44 Physical limitation in 1988 or before* No 502 72 352 70 150 30 Yes 192 28 116 60 76 40 *P < .05; **P < .01; chi-squared test for association with exercise status. aTension-in: uneasy, anxious, with appetite loss and difficulty sleeping when under stress. bTension-out: anxious, angry, with increased activity when under stress. CStable: more self-contained and solid when under stress. April 1997, Vol. 87, No. 4 www.cuwai.com Public Health Briefs TABLE 2-Relative Risk of Depression and Relative Odds of Psychiatric Distress, by Physical Activity: The Precursors Study of Former Medical Students, 1948 through 1964 % Unadjusted Adjusted RR of depression by physical activity in 1978 (n = 752)a Sweat weekly No 33 1.27 (0.67, 2.41) 1.11 (0.56, 2.18) Yes 67 1.00 1.00 Number of sweats* 0 times/week 33 1.45 (0.68, 3.09) 1.18 (0.53, 2.64) 1-2 times/week 30 1.26 (0.56, 2.80) 1.08 (0.48, 2.45) 3 or more times/week 37 1.00 1.00 Change from medical school Inactiveb 32 0.87 (0.34, 2.22) 0.86 (0.33, 2.24) Became activec 52 0.60 (0.24,1.48) 0.66 (0.27,1.65) Remained actived 16 1.00 1.00 RO of psychiatric distress for physical activity in 1986 (n = 781 )e Sweat weekly No 35 1.20 (0.80,1.80) 1.16 (0.72,1.76) Yes 65 1.00 1.00 Number of sweats* 0 times/week 35 1.26 (0.81, 1.98) 1.22 (0.77,1.92) 1-2 times/week 22 1.22 (0.72, 2.06) 1.17 (0.69, 2.00) 3 or more times/week 43 1.00 1.00 Change from medical school Inactiveb 35 1.18 (0.59, 2.40) 1.13 (0.55, 2.32) Became activec 51 1.16 (0.59, 2.29) 1.19 (0.60, 2.36) Remained actived 14 1.00 1.00 Note. RR = relative risk; RO = relative odds. aAdjusted for gender, age in 1978, smoking status in 1978, and substance abuse prior to depression. blnactive: no sweats weekly in 1978 or 1986, regardless of medical school activity. CBecame active: no or little exercise in medical school, sweat weekly in 1978 or 1986. dRemained active: moderate or much exercise in medical school, sweat weekly in 1978 or 1986. eAdjusted for gender, age in 1986, smoking status in 1986, and substance abuse prior to 1988. *Pfor test of trend = .60 (unadjusted) and 0.94 (adjusted). *Pfor test of trend = .15 (unadjusted) and 0.20 (adjusted). benefit physical health status. 1-16 32.33 Exercise has more physical health ben- efits, lower costs, and fewer side effects than psychotherapy and pharmacotherapy. Efforts to determine whether exercise is effective in preventing depression and psychiatric distress should continue, espe- cially in vulnerable subgroups, such as individuals with a family history of depression. Any protective effect of physi- cal activity on depression or psychiatric distress is probably small and will require a larger sample size than the present study to measure with confidence. [1 Acknowledgments This work was partially supported by research grants from the National Institute of Mental Health (MH46967), the National Institute on Aging (AGOI 760-13A1 and AGO1760-14S1), and by a Health of the Public Grant from The Pew Charitable Trust and The Rockefeller Foundations. Dr Klag is an Established Investi- gator of the American Heart Association. This paper was presented in part at the 33rd Annual Conference on Cardiovascular Disease Epidemiology, American Heart Associa- tion, Santa Fe, NM, March 17, 1993, and at the 13th Mid-Atlantic Regional Meeting of the Society of General Internal Medicine, Philadel- phia, PA, March 11, 1994. References 1. Folkins CH, Sime WE. Physical fitness training and mental health. Am Psychol. 1981 ;36:375-389. 2. Plante TG, Rodin J. Physical fitness and enhanced psychological health. Curr Psy- chol. 1990;9:3-24. 3. Byrne A, Byrne DG. The effect of exercise on depression, anxiety and other mood states: a review. J Psychosom Res. 1993;37: 565-574. 4. Hughes JR. Psychological effects of ha- bitual aerobic exercise: a critical review. Prev Med. 1984;13:66-78. 5. Farmer ME, Locke BZ, Moscicki EK, Dannenberg AL, Larson DB, Radloff LS. Physical activity and depressive symp- toms: the NHANES I epidemiologic fol- low-up study. Am J Epidemiol. 1988; 128: 1340-1351. 6. Camacho TC, Roberts RE, Lazarus NB, Kaplan GA, Cohen RD. Physical activity and depression: evidence from the Alame- da County study. Am J Epidemiol. 1991; 134:220-231. 7. Weyerer S. Physical inactivity and depres- sion in the community: evidence from the Upper Bavarian field study. Int J Sports Med. 1992; 13:492-496. 8. Thomas CB. Observations on some pos- sible precursors of essential hypertension and coronary artery disease. Bull Johns Hopkins Hosp. 195 1;89:419-441. 9. Paffenbarger RS Jr, Wing AL, Hyde RT. Physical activity as an index of heart attack risk in college alumni. Am J Epidemiol. 1978;108: 161-175. 10. Albanes D, Conway JM, Taylor PR, Moe PW, Judd J. Validation and comparison of eight physical activity questionnaires. Epi- demiology. 1990;1 :63-71. 11. Washburn RA, Smith KW, Goldfield SRW, McKinlay JB. Reliability and physiologic correlates of the Harvard Alumni Activity Survey in a general population. J Clin Epidemiol. 1991;44:1319-1326. 12. Ainsworth BE, Leon AS, Richardson MT, Jacobs DR, Paffenbarger RS Jr. Accuracy of the College Alumnus physical activity questionnaire. J Clin Epidemiol. 1993;46: 1403-1411. 13. Paffenbarger RS, Hyde RF, Wing AL, Hsieh C. Physical activity, all-cause mortal- ity, and longevity of college alumni. New Engl J Med. 1986;314:605-613. 14. Paffenbarger RS, Wing AL, Hyde RF, Jung DL. Physical activity and incidence of hypertension in college alumni. Am J Epidemiol. 1983; 117:245-257. 15. Lee I, Paffenbarger RS, Hsieh C. Physical activity and risk of developing colorectal cancer among college alumni. J Nati Cancerlnst. 1991;83:1324-1329. 16. Helmrich SP, Ragland DR, Leung RW, Paffenbarger RS. Physical activity and reduced occurrence of non-insulin-depen- dent diabetes mellitus. New Engl J Med. 1991;325: 147-152. 17. Ewing JA. Detecting alcoholism: the CAGE questionnaire. JAMA. 1984;252: 1905- 1907. 18. Bush B, Shaw S, Cleary P. Screening for alcohol abuse using the CAGE question- naire. Am J Med. 1987;82:231-235. 19. Graves PL, Mead LA, Wang N, Liang K, Klag MJ. Temperament as a potential predictor of mortality: evidence from a 41-year prospective study. J Behav Med. 1994;17:111-126. 20. American Psychiatric Association. Diagnos- tic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: Ameri- can Psychiatric Association; 1994. 21. Goldberg D, Williams P. A User's Guide to the General Health Questionnaire. Wind- sor, Berkshire, England: Nfer-Nelson Pub- lishing Co Ltd; 1988. 22. Goldberg DP. The Detection ofPsychiatric Illness by Questionnaire: A Technique for the Identification and Assessment of Non- 672 American Journal of Public Health April 1997, Vol. 87, No. 4 www.cuwai.com Public Health Briefs psychotic Psychiatric Illness. London, En- gland: Oxford University Press; 1972. 23. Ford DE, Anthony JC, Nestadt JR, Ro- manoski AJ. The General Health Question- naire by interview: performance in relation to recent use of health services. Med Care. 1989;27:367-375. 24. De Geus EJC, van Doorien UP, Orlebeke JF. Regular exercise and aerobic fitness in relation to psychological make-up and physiological stress reactivity. Psychosom Med. 1993;55:347-363. 25. Hughes JR, Casal DC, Leon AS. Psycho- logical effects of exercise: a randomized cross-over trial. J Psvchosom Res. 1986;30: 355-360. 26. King AC, Taylor CB, Haskells WL, De- Bush RF. Influence of regular aerobic exercise on psychological health. Health Psychol. 1989;8:305-324. 27. Moses J, Steptoe A, Mathews A, Edwards S. The effects of exercise training on mental well-being in the normal popula- tion: a controlled trial. J Psvchosom Res. 1989;33:47-61. 28. Buchman BP, Sallis JF, Criqui MH, Dims- dale JE, Kaplan RM. Physical activity, physical fitness, and psychological charac- teristics of medical students. J Psychosom Res. 1991 ;35: 197-208. 29. Ross CE, Hayes D. Exercise and psycho- logic well-being in the community. Am J Epidemiol. 1988; 127:762-77 1. 30. Deckers JW, Klag MJ, Seidler AJ, Mead LA. Association of youthful and midlife physical activity with subsequent develop- ment of cardiovascular disease. Circula- tion. 1990;82:258. 31. Robins LN, Regier DA, eds. Psychiatric Disorders in America: The Epidemiologic CatchmentArea Study. New York, NY: The Free Press; 1991. 32. Leon AS, Connett J, Jacobs DR, Rauramaa R. Leisure-time physical activity levels and risk of coronary heart disease and death: the multiple risk factor intervention trial. JAMA. 1987;258:2388-2395. 33. Cummings SR, Kelsey JL, Nevitt MD, O'Dowd KJ. Epidemiology of osteoporo- sis and osteoporotic fractures. Epidemiol Rev. 1985;7:178-208. Rsu"'..Oer 12 month. s ~~~~~~~~~~~~~~~~~~.. on' temlilloitcfction *.ve t jthe ond..i.ions. Sta .. ..... ...... dardized . scores. decreased for * ; ; i & ...............................:.. '''..... ,.....''. behavl6ral counseling relativeto hel *f all ther oaditioiuW Coclsins Bhaioalcon *. *elig proucslare chage in to.coronaly heart disease risk thae olmhy se itrestions' .~~~~~ ~ ~ ~ ~~~~~~~~~~~~~~~~ 7. (Amco Pub.1feaL 1997e;e8i 673-n 676)...:*M}. :1i8F..671. A*.il 1997,Vo.87 No 4 Composite Cardiovascular Risk Outcomes of a Work-Site Intervention Trial Michelle K. Gomel, PhD, Brian Oldenburg, PhD, Judy M. Simpson, PhD, Marilyn Chilvers, MAppStat, and Neville Owen, PhD Introduction Most work-site multiple risk factor intervention trials have used single out- come measures to assess intervention effects. Of six large randomized trials, three have demonstrated significant reduc- tions in at least one risk factor after intervention.'-5 However, single risk fac- tor outcomes may not reflect the overall effect of intervention, because change in one risk factor may be accompanied by negative or positive changes in other risk factors. Furthermore, small positive changes in individual risk factors may only be detected when the information from each is combined into a summary measure. In fact, this may be the reason why some trials failed to find an effect. This paper reports a reanalysis of the main results presented in an earlier paper5; composite outcomes were used to take into account the extent and direction of change occurring for each risk factor outcome. Two composite outcome mea- sures were used, not for the purpose of comparison but because each had particu- lar advantages for combining the single outcome measures used in this trial. Methods Twenty-eight ambulance stations in metropolitan Sydney, Australia, were ran- domly assigned to one of four interven- tions. The methodology and interventions have been described previously in detail.5 All participants received a risk factor assessment and feedback at baseline, as well as at 3, 6, and 12 months after the initial assessment. Participants in the health risk assess- ment condition (n = 130) received the risk factor assessment only. Participants in the risk factor education condition (n = 82) received information on risk factors, an educational resource manual, and videos. At the time this paper was written, Michelle K. Gomel was with the Department of Psychologi- cal Medicine, University of Sydney, Sydney, New South Wales, Australia. She is now with the Division of Mental Health and Prevention of Substance Abuse, World Health Organization, Geneva, Switzerland. Brian Oldenburg is with the School of Public Health, Queensland Univer- sity ofTechnology, Queensland, Brisbane, Austra- lia. Judy M. Simpson is with the Department of Public Health, University of Sydney. Marilyn Chilvers is with the Department of Statistics, Macquarie University, Sydney. Neville Owen is with the School of Human Movement, Deakin University, Melbourne, Victoria, Australia. Requests for reprints should be sent to Michelle K. Gomel, PhD, Division of Mental Health and Prevention of Substance Abuse, World Health Organization, 20 Ave Appia, CH- 121 1 Geneva 27, Switzerland. This paper was accepted September 6, 1996. Editor's Note. See related annotation by Heitjan (p548) in this issue. American Journal of Public Health 673 www.cuwai.com