PIIYSICAL ACTIVITY, PttYSICAL FITNESS, AND PSYCHOLOGICAL ...rmkaplan.bol.ucla.edu/Robert_M._Kaplan/1991_Publications_files/01… · of race/ethnicity for the group was 65% Caucasian,

Printed in (.;re;ll ||litalrl r_ 199! PergamonPressplc

PIIYSICAL ACTIVITY, PttYSICAL FITNESS, AND

PSYCHOLOGICAL CHARACTERISTICS OF MEDICAL

STUDENTS

llltAI) l'. IIU('IIMAN, JAMliS [7. SAI.I.IS, MWrtAI:.[. II. CRtQUI, JOEl. I:.. I)tMst)at.v. andROIII-RT M. KAPt.AN

( R_'_'cil'ed 23 k lar('h 1990; accepted in repised Jbrm 14 if ugust 1990)

Abstract - Exercise habils, c:_rdiovascular limcss, and selected psychological characteristics were assessed

in a sample of over 200 men and women at entrance to medical school. Fitness was measured with a steptest, and oilier variables were me;isured with standardized questionnaires. Anger ('anger-in') showed thestrongest negative correlation with bolh exercise and fitness. Anger suppression, Type A behavior, anddaily stress shtw,,ed significant negative correlations with both exercise and fitness vari:lbles, although thestrengths of these assoeiatiorts were uniformly weak. Associations of exercise/illness with depression andtotal anger invenlory were nonsignificant. Few gender differences were found. Both exercise and fitnessshowed similar patterns of association with psychological variables. Both exercise and fitness wereassociated with a style of anger expression that has been found to be related to cardiovascular risk in otherstudies.

INTRODUCTION

OvEit 1000 articles have been written on the psychological effects of exercise andphysical fitness [I]. The proposed psychological benefits, while widely accepted, havebeen supported by relatively few well-designed studies [i, 2]. Recently, however,improved sttidies have begun to substantiate some of these reported psychologicalbenefits of exercise and fitness [3].

The most convincing studies documenting psychological benefits of exercise haveinvolved the use of aerobic exercise for treatment of patients with mild to moderatedepression [4, 51. In addition, several studies have reported mood elevation withexercise in suhjccts without prior signs of clinical depression [6, 7]. A prospectivestudy fi'ont a large community sample concluded that physical inactivity may be arisk factor for the development of depressive symptoms [8].

lnvcrse associations of exercise with physiological and psychological indices ofcoronary-prone (Type A) behavior have been observed [9, 101. Since recent investi-gations indicate that much of the coronary heart disease (CtlD) risk forlnerlyattributed to Type A behavior may in fact be primarily related to sub-componentsof anger and hostility [I 1-13], it is important to determine the associations of exercisewith anger and hostility as well as with various modes of anger expression.

Nunlcrous studies havc shown that exercise and aerobic fitness influenced the

quantity and quality of responses to psychological stress [14]. Some have demon-

Departments of Conununity and Fanfily Medicine, Pediatrics, and Psychiatry, University of California,San Diego, CA 92182, U.S.A.

Address correspondence to: Janles F. Sallis Ph.l)., Department of Psychology, San Diego SlateUniversity, Still Diego, CA 92182 U.S.A.

This work was suppcPrted h'l part by Nalional Institutes of lle.'illh grants I11.01718, IIL40102, :ind111.36005. and by the IJCSI) (.'linical Research Center grant no. IRR00827.

197

f

stratcd decreased physiologic responses and a more r;qfid returtt to basclitlc _d"physiologic paramcters following psychological stress in high-litness populations[! 5, 16]. Others have shown lower levels of anxiety during recovery periods lbllowingpsychological stress in high-fitness individuals [17, 18]. Ill addition, studies haveshown decreases in trait anxiety after chronic exercise training [19, 20]. Retrospectivestudies have also shown a protective effect of regttlar exercise in maint_tining bothphysical and emotional well-being following periods of increased life-event stress asmeasured by questionnaires [21,22].

There are numerous methodologic problems with previous stt,dics of the relation-ships among exercise, fitness and psychological wtriables. Many previous investi-gations included only men or only wonlen, though sex dill'ercnces have bccn foundin the few studies with both sexes [23-25]. Some stttdics inchtded a small nttmbcr ofpsychological variables, though it is often hypothesized that exercise infhtcnces a widevariety of variables. Some studies measured exercise behavior and some measuredcardiovascular fitness, but no studies measured both. It is, therefore, unclear whetherpsychological variables are more likely to be associated with exercise or fitness. The

present study was designed to address these shortcomings by studying both men andwomen, inclttding measures of several psychological variables, and inchJdingmeasures of both exercise and fitness. The specific hypothesis is that both reported

exercise and cardiovascular fitness are negatively associated with depression, TypeA behavior, various profiles of anger, and perceived stress levels.

The present study was conducted with medical students as sttbjccts. While this isa select sample, the students are relatively homogeneous in _tge and educationalachievement, so these potential confounders are controlled. There is also substantialvariability in exercise habits across medical students [26, 27]. This variability in ahomogeneous population allows a powerful test of the questions under study

191ETIIODS

Subject characteristics

The sample group consisted of the 1986 and 1987 entering first-year medical school classes .'d Univcrsilyof California, San Diego (UCSI)) School of Medicine. Of the 243 possible participants, 207 (85%) clcclcdto participate in a cardiology risk factor assessment as part of a Prcvcntivc C_udiology Acadclnic Awardthat was begun at UCSD in 1986.

Women represented 29.5% of the sample. Mean age for the entire group was 23.'I yr. The brcakdowl}of race/ethnicity for the group was 65% Caucasian, 16% Asiau, 9% ]lispanic, 7% Bl_lck, aJ_d 3% others.Ninety per cenl of the subjecls were single.

Assessment of fitness and exercise t,ariables

Assessment of _dl variables was carried out during one day of orientation wcck for each of the enteringclasses. Sludellts were tested in groups, and every attenll:,t was nlade to standardize conditions and providea uniform testing environment for all subjects.

To assess cardiovascular fitness, _ submaximal 5 tnirt slep test adapted from Sharkey {28] wasadministered to the parlicipauts. After rcslillg comfortably for 5 mi_l, subjccls were illslrucled to slep ill*and down a betlch 15] inches tall for men aml 13 inches tall for women, with each slop cortcspondil_gto the beat of a metronome set at a rate of 90 beats per rain. After 5 rain of exercise, subjects sat on the

_ttC]l and a pulse rale was taken for 15 s by trained technicians. A Fitness Index (FI) was computed using

this heart rate and adjusting for age, sex and body weight [281. The estimate of maximal oxygenconsumption was expressed as ml/kg/min.

Ileight and weight were measured with a secured height anthropometer and balance beam scale withoulshoes. Body mass index (BMI; kg/m 2) was used as a recast, re of obesity.

Exercise was assessed with a number of questionnaire vari_lblcs. We assessed exercise frequc[_cy hyasking: 'In a typical week, how many times do you perform physical aclivily or exercise il_ your leisure

Ihile at least 2() tnitl wilhtitlt sti)pphig, which is vigorous eu,_)ughto make you breathe hard and/or sweat?'l)llralion of exercise ('H physical activiiy was assessed with, 'llow long have you been doing vigorousactivities aI Ihis level ill nat)nths?' In addition to these more quarllitative questions, subjects were also askedto rate Ihcir relative exercise level; 'llow much physical activity do you usually get, compared 1o othersof your same age and sex?' Suhjecls were asked to consider both leisure and work activities, and respondon a live-point scale fiom 'much less' to 'nu|ch nlore'.

/ISS('.L_IHUH[ f?/'/)._V('/I()lOl_i('(l/ rariuhles

I)eprcssiol} w;is n,;sessed using Ihe 20-ilen} Cel)ler h}r Epidcmiologic Studies Depression Scale (CES-I))th;_t is designed to measure depressive symptomalology in the general population 129}. This scale has beenfoulld to be a reliable alltl valid mcasure in botll community and clinical populations [291.

l'sych()logical stress v,'as ;tssesscd by asking Ihe respondents to use a rating scale from zero 'very low,;ihnosl ilouc', tt} li_c "very high. ah)]ost COllStaltt', It) indicate the alnount of stress and lensioll ill theirevery day lives.

"lhe I_ortiler Shot{ Ratil|g Scale was uscd {o measure overt pattern A behavior {30f This test Ires been/otn|d to ll_lvc good correlations wilb other self-reported me;isures of Type A behavior, such as the JenkinsAclivily Survey (r = 0.70), [311 and Ihe P'ranfiugham Type A scale (r = 0,60) 132]. In addition, it has been

shown It) h{ivc excellent test retest reliability in both patients with clinical chest pain (r = 0.84) [331, aswell as ill a population of undergradu,'_te and medical students (r = 0.72--0.74) [34].

Auger was asscsscd usil|g two sclf-reporled r,'tliug scales. The State-Trait Anger Scale (STAS) [35, 36]was used to colnpule the summary variable, anger inventory. The STAS was designed 'to assess both the

iutcusity of angry li-clings alld individual dill_rences in anger-proneness'. Construct validity was supportedI_ygood corrclatiol|s wilh tile I]uss-Dt|rkee I lostility Inventory (r = 0.66 0.73) [37]. The Anger Expression

Self-Analysis Qllcstionnairc 1371 was used Io assess the subjects" propensity to express or suppress angryfeelings aml eul|)lio|_s. The anger expression scores obtained from this survey have shown goodcorrelations (r = 0.46 0.49) with an earlicr questionnaire designed by Harburg [381 to classify subjects asprcdontillanlly expressing anger in an outward fashion or exhibiting suppression of anger in varioussituati_ms [371. Subjects were asked to tale how they generally feel wlten they are angry or furious. Threevariables _erc computed. 'Anger-in' reflects a style of suppressing anger and hostility. 'Anger-our reflectsol|tward cxr, lcssiotl of angry feelings. 'Anger-conlror reflects tile extent to which the subject feels he orshe is able to retaiu control over anger a|ld hostile imptdscs.

All amdyscs were conducted using tile SYSTAT software package.

R F,SU LTS

Sex di[.'[brem'es in .fitm,ss, exercise, or psychological variables

Variable mca_}s, slamlard deviations, and independent sample t-tests grouped bysex arc presented in Table I, Statistically significant differences between men andwomen were observed for only three variables. Women had higher mean Type Ascorcs (p < 0.05) whereas men had higher mean scores on anger-in (p < 0,001). Men

'['AIH E I.---.VARIAIH.I- MFANS, SI)S, ANt) INI)EI'[:NDENT SAMPLES /-TESTS, (;l_.OIJl'l;]) BY SEX

Women Men

(N = 54) (N = 153)Variable Mean (so) Mean (St)) t p

Fitness index (FI) 47.7 (8.7) 47.3 (8.6) 0.33 0.739

Exercise freq. 3.8 (2.9) 4.1 (2.8) 0.617 0.539Exercise duration 52.7 {56,1) 62,9 (55.6) 1.194 0,235

Exercise rating 3.0 (i .0) 3.3 (1.2) 1.603 0.I I IBMI 22.2 (2.6) 23,6 (3.1) 3.250 0.001('I']S-D 10.5 (7.9) 10.3 (7.3) 0,206 0.838l)aily stress 2.5 (0.9) 2.3 (0.9) 1.202 0.232

l}orhler Type A 183.0 (29.9) 172,8 (30,2) 2.240 0,027Anger invenlory 25,5 (4.6) 25,8 (5.3) 0.311 0.756

AJiger-iJ} 13,8 (3.3) 15,6 (3.6) 3.267 0,001Anger-out 15.0 (2.8) 14,7 (2.9) 0.617 0.539Anger-control 8.q (I .7) 9.0 (2.0) 0.261 0.795

Age 23.6 (2.9) 23,2 (2.2) 0.740 0.461

200 B.P. BUCIIMAN et ol.

TABLEII,--F'EARSON CORRELATIONMATRIX

Women: N = 54 (*p < 0,05 if r _ 0.25)FI Frequency Duration Rating BMI

El 1.t)O0Exercise

frequency 0.35* 1.000Exercise

duralion 0.14 0.31* 1.000Exercise

raling 0.41* 0.68* 0.32* 1.000IIM1 --0.35* 0.03 0.33* -0.12 1.000CES-D -0.08 -0.23 -0.21 -0.26* -0.12Slress -0.02 -0.18 0.06 -0.24 0.35*

TypeA 0.04 0.19 -0.05 0.01 0.01Anger inv. -0.12 -0.01 0.11 0.02 0.09Anger-in -0.08 -0.25* -0.I0 -0.36* -0.12Anger-out -0. I I 0.25* 0,07 0.24 0.13Control 0.12 -0.07 -0.09 -0.20 -0.37*

Age 0.03 0. I 1 - 0.00 0.05 0.36*

CE_-D Stress Type A Anger inv. Anger-in Anger-out Control

CES-D 1.000Stress 0.39* 1.000

Type A 0.34* 0.42* 1.000Anger inv. 0.36* 0.01 0.27* 1.000Anger-in 0.44* 0.11 0.12 0.32* 1.000Anger-out 0.1)4 -0.05 0.30* 0.39* -0.36* 1.000Control 0.12 -0.03 -0.05 -0,06 0.22 -0.24 1.000

Age -0.06 -0.12 0.13 0.11 0.03 0.18 0.02

Men: N = 153 (*p <0.05 if r _ 0.16)

FI Frequency Duration Raling BMI

FI 1.000Exercise

freqtlency 0.37" i.000Exerciseduration 0.29* 0.49* 1.000F,xercise

rating 0.47* 0.68* 0.47* 1.000BMI -0.31" 0.04 -0.0t -0.13 1.000

CES-D -0.18" -0.I I -0.19' -0.22* 0.10Stress -0.16" -0.23* -0.08 -0.29* 0.04

Type A 0.04 -0.02 -0.21' -0.08 -0.01Anger inv. -0.1)5 -0.07 -0.18' -0.10 0.08Anger-in -0.20* -0.21' -0.26* -0.26* -(I.01Anger-out 0.08 0.01 -0.08 -11.1)4 0.23*Control --0.1,l --0.07 0.02 --0.09 0.01

Age 0.02 0.08 0.09 0.05 0.28*

CES-I) Stress Typc A Anger inv. Angcr-in Anger-out ('onlrol

(?F,S-D 1.000Stress 0.3(,)* 1.000]'ype A 0.26* 0.33* 1.000Anger inv. 0.40* 0.37* 0.44* 1.000

Anger-in 0.38* 0.36* 0.08 0.45* 1.000Anger-out 0.19" 0.10 0.37* 0.37* -0.13 1.0007"onlrol --0.31)* -- 0.20" --0.25* --0,47" --0.08 - 0,30* [ .()00Age 0.01 0.14 0.07 0.19" --O.l)l .-0.01 -0. t0

Physicalactivityand fitness 201

had a higher mean BMI (p < 0.001), which probably reflects the greater muscle massin nlen.

Uoit,ariate associations

Table II snmmarizes Pearson correlations among fitness, exercise, psychological,and demographic variables separately for women and men.

The FI showed significant positive correlations with exercise frequency per weekin both groups (r = 0.35; 0.37). In men, FI was also positively correlated with exerciseduration in monlhs (r = 0.29). Of the three exercise variables, the exercise rating (ascompared to others) was mos! highly correlated with FI for both groups (r = 0.41;0.47). As expected, very high correlations were found in both groups between theexercise rating score and reported exercise frequency (r= 0.68; 0.68). Body massindex was significantly negatively correlated with FI for both groups (r = -0.35;-0.31) but was significantly positively correlated with exercise duration only forfemales (r = 0.33).

There were signilicant negative correlations between FI and anger-in (r = -0.20),CES-D (r = -0.18), and daily stress (r = -0.16) scores for men, but there were nosignilicant correlations between FI and psychological variables for women. Exercisefrequency showed significant negative corrrelations with anger-in for both groups(r=-0.25; -0.21) and with daily stress for men (r=-0.23), and significant

positive correlations with anger-out for women (r = 0.25). Exercise duration showedsignificant negative correlations with the anger inventory (r =0.18), anger-in(r= -0.26), Type A (r = -0.21), and CES-D scores (r = -0.19) in men, but nosignilicant correlations with psychological variables in women. The exercise rating"showed signilicant negative correlations with anger-in (r =-0.36; -0.26) andCES-D scores for both groups (r= -0.26; -0.22) as well as with daily stress formen (r= -0.29).

There were low to moderate correlations among psychological variables, indicating

that they tapped different domains of psychological functioning.

Mullirariate associaliotts

Slcpwise multiple regression analysis was used to further examine the relationshipof the litncss and exercise variables with the psychological variables. Using multiplercgressiou, the association between independent and dependent variables can bedctermincd aftcr adjusting for multiple other independent variables. The confound-ing wuiables of sex and BMI were forced into all model statements first. This wasdonc bccausc the primary hypotheses did not concern sex or BMI, but both of thesevariablcs wcre significantly associatcd with some fitness, physical activity, andpsychological variables. Forcing scx and BMI into the models first yields associationsthat arc adjusted for these confounders. All remaining independent variables wereentered into the initial ntodel statement with an alpha to enter of 0.05. After a subset

of predictors was identified in the stepwlse model, a final model statement was createdusing only these significant variables. Data are shown for all variables used in thefinal model statements. Two kinds of regressions were run. First, FI was used as thedcpcndcnt variable and related to the physical activity measures. Second, sevenpsychological variablcs were stt_die(I in relation to each fitness and physical activityv_tri_blc in separale regressions.

l|, I'. |]I.J('IIMAN t'l _'l/.

Table'Ill shows the regression of Fitness Index as predicted by exercise frequency.exercise duration, sex, and BMI. Exercise frequency was a sigmlicant positivecorrelate (//=0.376, p_0.001), and BMI a signilicant negative correlale(/_ -- -0.317, p _<0.001) of FI. The signilicance level of the overall regression wasp _ 0.001, with 21.5% of the variance in FI being accounted for by the variables inthe final nlodel stalement. Exercise frequency was the only physical activity variablethat was an imtepcndent correlate of FI.

Table IV shows the results of four regressions. Fih_ess Index, e×ercise frequency,exercise duration, and exercise rating were predicted by all psychological variables,sex, and BMI in separate analyses. Again, data are shown lk_rall significant variablesremaining in each of the final model statements. Looking initially at the regressionof FI as predicted by the psychological variables, BM! was again a significantnegative correlate of FI, (JJ_= -0.305, p _ 0.001). Anger-in was tile only psychologi-cal variable that was a significant negative correlate of F1 ([I = -0.183, p _ 0.01 ).The significance level for tile total regression was p _<0.001, with 10.5% of thevariance in 171being accounted for by linal model statement variables.

The regression with exercise fi'eqnency as the dcpendcnt variable showed angcr-in([I =-0.171, p _<0.025) and daily stress (1/=-0.208, p _<0.009) as signilicantnegative correlates. Type A behavior was a significant positive correlate 1[I= 0.156,p _<0.04). Overall significance level of the regression was p _<0.003, with 6.6% of thevltriance in exercise frequency being accountcd for by the three psychologicalvariables in the final model statement.

The regression of exercise duration as predicted by psychological wu'iablcs alsoshowed anger-in as a significant negative correlate (fl =-0.205, p _<0.005). Incontrast to its reh|tionship with exercise frequency, Type A behavior was a ncarly

•" significant negativc correlate of exercise duration (fl = -0. 138, p _<0.054). Signili-,' cance level for the overall regression was p _<0.006, with 5.4% of file variance in

exercise duration accounted for by the final model statement variables.The final regression in Table IV of exercise rating its predicted by psychological

-, variables again showed anger-in as a signilicant negative correlate (/I = -0.228,p _<0.002). In addition, exercise rating was negatively corrclatcd with daily stress

.,. (//=-0.220, p _<0.002), sex (male= 1, female = 2; [_=-0.176, p _<0.014), and. anger-control ([I = -0.164, p _<0.016). The significance level for the total regression

was p _<0.001, with 14.5% of the variance in exercise rating being accountcd for bysex and the three psychological variables in the final model statelnent.

TAIILE IlI.--.MuLTII'I.E REGRESSION ANALYSIS OF TIlE ASSO('IATI(IN OF I:II'NI!SS INDI-X

WITll EXI_RCISE FREQLCENCY AND I)[JRA'FION, SEX AND BMI

Variable B SE Beta p

Sex -(I.418 1.269 --0.tl22 (I,7,12BMi -0.629 0.131 --0.317 (}.00lExercisefrequency 1.159 0.199 0.376 0.00IConstant 64.072 5J)02 0.l)00 0.001N: 189 SE of cstimate: 7.703Adjusted Squared Multiple R: 0.215 F-ralio: 18.160 p _<0.001

Note:The followingindependentvari;ibleswereentered into Ihe initial regressionstatement: sex, BMi, exercisefrequency,exerciseduralim_.B is Ihe sl:md_lrdizcdregressioncoefficient,and Beta is Ihe unslamtardizedcoelficicnt.

Physical activity and tilness 203

TABI.]! IV.---_|UI.TIPI.I! RI'_(-;R]iSSION ANALYSIk_ OF "l'l|[_ ASSOC'IATION _)F FI'I'NE_ AND

|-X|iRC'lS|i VARIAI|I.ES WI'[II PSY(?IIOLO(;I(?AL VARIAIH.I!S, SEX AND BM ]

Varial)lc B SE Beta p

Fitness h,h,.v t,._'ps)'chofi_gical uarhtblesSex - 1.565 1.389 -0.082 0.262

IIM1 -0.605 0.140 --0.305 0.000

Anger-in -0.,147 0.173 -0.183 0.010('onsta nt 76.192 6.137 0.000 0.(}00

N: 189 SE of estimate: 8.226

Adjusted Stlnared Multiple R: 0.105 /:-ratio: 8.374 p <_0.001

l'_xerctw /i','q,en, 9" t,s p.wchological t'oriablesSex -0.421 0.473 -0.067 0.375BMI 0.059 0.047 0.090 0.217

Daily Slress -1'1.619 0.236 -0.208 0.009TypcA 0.015 0.007 0.156 0.042Anger-in -0.136 1,,I.060 -0.171 0.025('onstanl 3.601 2.329 1,).000 0.124

N: 193 SE of estimate: 2.782

Adjnslcd Squared MultiF, le R: 0.066 F-ratio: 3.715 p _<0.003

Evercise th,,tion rs I,._)'chologicol I'm'iabh'sScx - 11,1.916 9.269 -0.088 0.240BMI 0.709 0.930 0.055 0.447

TypeA -0.258 0.133 -0.138 0.054Angcr-in -3.21t') 1.132 -0.205 0.005Conslalll 145.161 44.932 0.000 0.001

N: 195 SE of estimate: 55.119

Adjuslcd Squared Mulliple R: 1,).054 F-ratio: 3.785 p _<0.006

Evercise r,ti, g t's psychoh_gical t,arh_bh'sScx -0.049 0.181 -0.176 0.014'IIMI -0.027 0.018 -0.101 0.142

I)aily Slrcss -0.267 0.087 -0.220 0.002Anger-in -0.074 0.023 --0.228 0.002Control -0.102 0.042 -0.164 0.016Constant 7.445 0.893 0.000 0.000

N: 189 SE of estimate: 1.081

Adjnsted Squared Multiple R: 0.145 F-ratio: 7.581 p _<0.001

Notc: The following independent variables were entered into the inilial regressionslatcmcnls: sex, BMI, CES-D, daily stress, Type A, anger invenlory, anger-in,

anger-out, anger-control. B is the standardized regression coefficient, and Beta is thet,nslandltrdized coefficient,

DISCUSSION

The primary finding was that both litness and exercise behavior were significantlyassociated with various psychological wtriables, but the associations were uniformlywcak. The coml_ination of psychological variables, sex and BMI accounted for5A- 14.5'¼, of the variance in litness and exercise. While anger-in was consistently

associated with exercise and fitness in the present sample, stress and Type A wereinconsistently associated, and depression and other anger-related variables were notassociated at all with exercise and fitness.

This study also allowed a comparison of whether exercise or fitness was morehighly associated with psychological variables. The findings indicate that the

_'(i,; B. 1_. |_U('IIMAN ¢1 _11.

associations were similar for both litness and exercise. In tile regression analyses,psychological variables accounted for 6.6°/, of exercise frequency, 5.4% of exerciseduration, and 14.5% of the exercise rating variance. While psychological wu'iablcsaccounted for 10.5°,/o of the variance in fitness index, BMI was a signiticantcontributor to the model. The similarity of findings for tit,tess and cxercisc i'l_casurcs

was expected, because cardiovascular fitness is determined primarily by exercise.Fitness index and exercise rating can be expected to be the best indic_ltors oflong-term physical activity practices, and these two wlriablcs l'_roduccd the highestadjusted R 2 with psychological variables. These lindings indicate that exercise andfitness are both associated with important psychological variables in this sample ofyoung men arid women, though the direction of causation cannot bc determined in

this study.The most striking and consistent finding was the association of the fitness and

exercise variables with the psychological variable anger-in. Univariate analysis• showed significant negative correlations of this variable with fitness index, exercise

frequency, exercise duration, and exercise rating in men. Significant negativecorrelations with anger-in were also seen with exercise frequency and exercise ratingin women. Multiv.'u'iate analyses using the entire sample showed anger-in to be asignificant negative correlate of all four measures of litness aml excJ'cise.

This finding not only has importance in relation to the association offitness/exercise and mental health, but it also may have relevance to coronary heartdisease (CHD). Sevcral recent studies have shown an increascd risk of CIII) in mcnand women who suppress anger. In a large prospective study, men and women who

suppressed anger had significantly increased rates of elewded blood pressure andall-cause mortality [40]. Other studies found that suppressed anger was ;issociatedwith elevated resting blood pressure [41,42] and with exaggerated blood pressurereactivity to stress [43]. Thus, it may not be the amount or intensity of anger one

experiences that is related to risk of CHD. Rather several studies indicate it is thenmnner in which this emotion is expressed, with high 'anger-in' scores or anger

suppression being positively associated with CHD. Interestingly, in tire present studythis is the single dimension of anger which showed significant ncgativc correlationswith multiple measures of exercise and fitness in both men and women.

This result might be interpreted in a number of ways. One possibility is that fitness

or regular exercise decreases the quantity of angry feelings that are suppressed orinternalized, or that exercise/fitness in some way alters the nature of anger expression.

This suggests that reduction of anger-in is a possible mechanism by which physicalactivity could protcct one from CHD. Another possibility, however, is that levels ofanger-in somehow affect the amount of exercise that is performed. Suppression ofanger may lead onc to exercise less often. Although the cross-sectional nature of thisstudy precludes any conclusions concerning the causal relationship of these v_Hiablcs,it is none the less a finding that warrants fiH'ther investigation.

Type A was a significant negative correlate of exercise duration. These results arc

consistent with experimental studies finding reductions in indices of Type A behaviorin men [44] and women [45] after periods of regular aerobic exercisc.

Daily stress was a significant negative correlate of exercise frequency in multivari-ate analyses. Since this was the only significant finding for the daily strcss rating, theassociation between exercise and daily stress is considcred weak.

The CI!S-I.) depression score failed to show ally significant correlations willcxcrcisc or litness. This is somewhat surprising in light of numerous studie.,demonstrating beneficial effects of exercise such as mood elewltion in non-clinicalpopulations and improvcment in clinically depressed patienls [4-7]. There wa.,

substantial variability in depression in this sample, so restriction of range is not ;_likely explanation for this failure to replicate earlier studies. Perhaps the associationsbetween cxelcise and depression are stronger in older populations.

Although the currcnt study has generated some interesting lindings, it has alsc3dcmollstratcd Ihat a considerable numbcr of psychological variables had no signili-cant relationships with exercise and titncss in this population. There are severalpossible explanations for the lack of findings with these variables.

Lack g/" t'ariahility _/"p.u'chological rariables

It is possible that in this highly selected homogeneous population, means andstandard deviations of psychological variables would be affected such as to biasagainst linding hypothesized associations. While mean scores on the CES-D scale oldepression were slightly higher than those obtained in several large communitysamples [29, 46], mean scores of anger-in and anger-out were slightly lower than thoseobtained in a popuhHion of high-school students [37]. Because Table I showedsul_stantiat wtriability in all wu'iables, this is unlikely to be an explanation for thefailure to find significant relationships with more psychological variables.

Lack o./"rarial_ility o[ exercise aml filness variables

The mean fitness levels of both men and women fell in the 'very good' to 'excellent'categories [28]. The mean exercise frequency per week reported in this group was fourtimes per week, which is quite high. This sample of young adults was generally veryactive and lit. The general lack of unfit and inactive subjects suggests that a restricted

range on these measures could have suppressed observed associations. Replicationof this study in a pol_ulation with wider variation in fitness and physical activity isreconlnlcnded.

I_[C(ISIII'CIII['II{ I._rl'Ol'

Mcasurcmcnt error is assumed to be an important source of variance in both thesellLreport measures and the step-test. However, the fact that significant relationships

wcrc found argues against measurement error severely limiting the power of thestudy. None the less, it is likely that error has reduced the number and extent ofobserved signilicant associations.

Timing _?/"a.vses.sme, ts

All measurements were made during the first week of medical school. It isreasonable to expect that such a transition could alter both psychological state andphysical activity patterns. Thus, the reported assessments may not have been 'typical'of this group, and non-representative values could have accounted for the lack ofassociations. ,,,

No associations

An eqtmlly likely possibility Ibr the lack of significan t associations of fitness orexercise wilh some of Ihe psychological variables is that the initial hypothesis was

206 IL I'. IJtJCIIMAN el _d.

incorrect in assuming that fitness and exercise would bc negatively associated withall psychological measures being examined.

Thirty-six medical students did not participate in the study, so it is possible thatself-selection could have influenced the results of this study. Unlk_ltunately, it wasnot possible to compare participants and non-participants.

Further limitations of this study include its cross-sectional design and the timitcdage range and highly selected nature of the subject sample. On the other hand, thissample can be expected to provide high quality data for a test of the association of

exercise/fitness and psychological variables in young adults. The availability of bothmen and women allowed us to conchlde that there were few gender diffcrel_ccs i,i

these associations. While all measures of exercise and litncss had importantlimitations, the use of multiple measures permitted the examination of consistencyacross measures. The findings of this study indicate that, among the psychologicalvariables studied, the strongest and most consistent finding was a negative associ-ation of suppression of angry feelings with exercise and fitness. Since angersuppression has predicted CHD in previous studies [I 1-13,40], this finding suggestsanother possible beneficial effect of physical activity and fitness that may be relevantto CHD. Because no other studies have examined the association between auger-inand exercise/fitness, this finding needs to be replicated in other samples.

REFF_RENCES

I. GOFF D, DIMSDALI!JE. The psychologic effects of exercise. J (_:rdiolmhmmary Rehahil 1985; 5:234 -240.

2. lhK;lllkSJR. Psycht_logical effects of habitual aerobic exercise: A critical review. I'ret, Mcd 1984; 13:66-78.

3. TAYLOR Cli, SALLIS JF, NI_EI)LER. The relation of physical activily and exercise to nlenlal heallh.Public lleahh Rep 1985; 100: 195-202.

4. GaIEST JIt, KLEIN Mil, EISCIIENS RR, i:ARIS J, GURMANAS, MORGAN WI). Running ilSa ircatnienl

for depression. Comp Psj,chiat 1979; 211:41--54.5. DOYNE EJ, CIIAMnI.ESSDL, BEUTLERLE. Aerobic exercise as a lrealment for depression in women.

Behav Therapy 1983; 14: 434--440.6. CARTER R. Exercise and happiness. J Sports Med Ph.l'S Fitness 1977; 17:307-313.

7. BERGI_RBG, OWEN DR. Mood alteralion with swimming: swilnmers really do 'feel better'. I'srchos,mMed 1983; 45:425 433.

8. FARMERME, LOCKi.:BZ, MOSCICKIEK, DANNENIII:RGAL, LARSONDB, RADLOI:FLS. Physical activity

and depressive symptoms: the NttANES I epidemiologic follow-up study. Anl J F_lJh&otio11988; 128:1340-1351.

9. LOBITZWC, I_RAMMELLIlL, STOLLS, NICCOLL A. Pllysical exercise and anxiety man:tgcnlent trainingfor cardiac stress management in a nonpalient population. J Cardiac Rehah 1983; 3: 683-688.

10. BLUMENTIIALJA, WILLIAMS RS, WILLIAMSRB, WAI.L^CE AG. Effects of exetcise on lhe Type A

(coronary prone) behavior pattern. P_Tchosom Med 1980; 42: 289--296.• II. Rose MI. Type A behaviour pattern: a concept revisiled. Can Med Assoc J 1987; 136:3,15 350.

12. WILLIAMSRII. Relining the Type A hypothesis: emergence of the hostility complex. Am J Cardiol1987; 60: 27J-32J.

13. IIECKER Mll, CIIESNEY MA, BLACK GW, FRAIJTSCtll N. Coronary-prone bellaviors in the Weslern

Collaborative Group Study. Ps),chosom Med 1988; 5ll: 153-164.14. DIMSI)ALF,JE, ALI'I_P,TBS, SCHNEIDERMAN N. Exercise as a modulator of cardiovascnlar rcaclivity.

In Hamlbook of Stress, Reactivity, and Cardiotlascit/ar Disease (Edited by Ma rrllEWS K, Wl!lSS S.et al.). New York: Wiley, 1986.

15. Cox JP, EVANSJF, JAMIESONJL. Aerobic power and tonic heatl rate responses to psychologicalslressors. Pers Soc Psl,chol Bull 1979; 5: 160-163.

16. Ilot.Mrs DS, Ro_rll I)L. Association of aerobic fitness wilh pulse rale and subjective responses topsychological stress. Psychophysioh_gy 1985; 22:525 529.

I>hysical activity .'tnd Illness 2UI

17. SINY()RD, S(IIWAItTZ SG, I>IiRONNI!T[:, BI_ISSONS, SL!RAGANiANP. Aerobic lilness level and reactivityto psychosocial stress: Physiological, biochemical, and subjective measures. Psychosom Med 1985;47:164 173.

18. I_J_<_)Kt!ST, LON(; BC. Ellicicncy of coping with a real-life stressor: A multimodal comparison ofaerobic litness. Ps?'cholgtrsiology 1987; 24: 173--180.

19. I|I.U_,tI_NTHAI..IA, Wrt.LJA_fS RS, Nt_z_oz_t.sTL, W^LLACI_ AG. Psychological changes accompanyaerobic excrcise in healthy middle-aged adults. P,_l'cho._otll Med 1982; 44: 529-536.

21). S('IIV..'ARIZ (;1:,, I)AVnl)St)N R.I, (;_')I.FMAN l)J. Patterning of cognitive _nd sonmlic processes inIhe scll-icgtota(iotl of atlxiely: I-II_ls of mcditatioll vca'sus exercise. I'srchosam [lh,d 1978; 4(1:321 328.

21. Ilat)',','N JI). I.AW'I'C)NM. Stress ,rod well-being in adolescenlce: The moderating role of physicalcxcteise..1 Iluman Stress 1986; 125 131.

22. l_,_lJn I)1.. I1¢)1r,tl!s I)S. hdhncnce of physical fitness in dctcrminiug the impact ofstressfid life events_n physic;d ;rod psych_logical hcalth. I's.l't'hosoal Med 1985; 47: 164-173,

23. I:t)l _nNS('1 I, L','N('H S, (;ARI>Nn_ntM M. Psychological fitness as a function of physical fitness. ArchI'hys Med Rehah 1972; 53: 503-51)8.

24. JASNt_S_ M I., I l¢>u._,n_si)S, BANKSI)L. Changes in personality associated with changes in aerobic andan;_crobic lilt_css in wolnen and n_cn. J I's.rchosoot Res 1988; 32: 273--276.

25. Sllil'llliNS T. Physic_d ;_ctivily and mental health in the United Slates and Canada: Evidence fromI't_ur poptdatiol_ surveys. Preo Med 1988; 17: 35-47.

26. StoicAl.J. l lcallh protcclive bchaviors in lirst year mcdical students, Soe S_i Med 1986; 22: 593-598.27. (;RII!N SA. l'vllYAI K. Thc impact of mcdical school on the student with respect to interpersonal

rclatio_ships aml li/'c-sty/es ,l Med l'_'d 1986; 61: 177-178.28. SIIAI_KI_YIL I'hl'siological Fitness and llzt, ight Cootrol. Missoula, MT: M_aunlain F'ress Publishing

Co., 1974.

29. I{AI_i._H: LS. The CI;S-D scale: A self-report depression scale for research in the general population...|ppl l'srJu)l Meas 1977; I: 385-401.

30. I]_)P.INI!I_ P,W. A short rating scale its a potential measure of pattern A behavior. J Chron Dis 1969;22:87 91.

31. .I()IINSI()N l)_,V, SFIAI'I!II.AG. Type A behaviour in llritish men: reliability a_d iutercorrela0on of twoine_lsurcs../ ('hron Dis 1983; 36: 203-207.

32. I1'1_(N_iI)G, P,t>S_N_,UANRI I, S('_m.L_!RE, CHI'SN_!VMA. Consistency and variation among instrumentsI)nrl_t)_ling to n)easnre Ihc Type A behavior patlern. Psycho._om Med 1985; 47:242 261.

33. BASS ('. Type A behavior in palients with chest pain: test-retest reliability and psychometriccorrelations of thc Bortner scale. ,I Ps3,chosoot Res 1984; 28: 289-300.

34. Ihu¢'_ K. Reliability of assessment of coronary-prone behavior with special reference to the Bortnerrating scale. J ltwchosom Res 1979; 23: 45-47.

35. SI'nilBERGI!R CI). Prelht_h_ary Manual for the State-Trait Anger Scale (STAS-form x). Tarnpa.Florida: University t_l"South Florida Iluman Resources Institute, 1980.

36. SPn_ _Ht_i_a CD, JA,_'tms G, Rt_ssr_z._.S, (?_taNZ_R. Assessment of_nger: The State-Trait Anger Sc.'de,In ,.fth'anct's h_ l'_'rsomditl' Assessmettt (Edited by I)LJFCIIrR JN, SPIFLIII:.RGI_RC|)), VOl. 2. I lillsdale,N.I: I.EA, 1983.

37. SI,n!_m!a(;r_ CI), JOHNSONEl I, RLJSSH.t.SF, CRANE RJ, JAcon3s GA, Wo_t.>_!NTJ. The experience andexpression of anger: Construction and validation of an anger expression scale. In Anger and Itostility#_ Cardiot'ascukn" aml ltelumioral Disorders (Edited by CH-ESNEVMA, ROS_NMANRIt). New York:

11cmisphere/McG raw-I lill, 1985,38. l IAItl_Ult(;E, Eltl:lrlt'l.lC, |IAUI'NSTEnNLS, CIIAI'E C, SClIULL W J, SCFIORKMA. SocioIecological stress,

suppressed hostility, skin color, and black-white male blood pressure: Detroit. Psychosom Med 1973;35:276 296.

39. II,_'NZ_SS. I:Z'_NZ.t_mM, KAr_z_t. WF3. The relationship of psychosocial factors to coro_mry heart

disease in the Framingham study. II1. Eight-year incidence of coronary heart disease. Am J Epidemiol1_180; II1:37 58.

40. Jtlll_.JS M, II^antm,_; E. Co'r'NNG'tC)N EM, JOIINSON Eli. Anger-coping types, blood pressure, and_dl-catnsc morlalily: A follow-up in Tecumseh, Michigan (1971-1983). Am J Epidemiol 1986; 124:220 233.

41. [)_r,tSt)^t.t! JE, P]t_ntcl!C, SCIIOENI:ELI) D, IqROWNA, ZIJSMAN R, GI_AIIAI_I1(. Suppressedanger andblood prcssure: The effects of race, sex, social class, obesity, and age. Psj,chosom Med 1986; 48:43O 436.

42. (h:N IRV WI). CIH:SNI_YAP, ('iAaV liE, IIM._. RP, i Iaamm_:; E, llabilual anger-coping styles: 1. Effecton mean bh_(_d pressure _nd risk fi_r esse_tial hyperle_sh_,n. I_sychoswn Med 1982; 44: 195-202.

43. JI}RtiI!NSI!NP,S, Uh_m_o_ IlK. Family history of hypertension, personality patterns, anti eardiovas-cul:_r rcaclivity to stress. I'.wchosom Med 1986; 48:102-117.

2ij8 lJ. P. |IUt:IIM^N el al.

44. BLUMENIIIAL JA, Cox DR, WALSll MA, KitltN C, WILt,IAMSRB, WII.LIAMS1{S. Effects ol- exercise on

the Type A behavior pattern (Abstract). Psychosonl Med 1987; 49: 204.45. JASNOSKIML, COrORAY DS, Ilousl'oN IlK, Os/,ll-ss WII. Modilicalion of Type A behavior th,uJgh

aerobic exercise. Motivation Emotion 1987; 11: 1--17.

46. WEISSmAN MM, SIIOLOI_ISKASD, POTTENGER M, PlttJSOFF BA, I.OCK_ IIZ. Assessing depressive

symptoms in five psychiatric populations: A validation study. Am J Epith.miol 1977; 1116:203 21,1.