Subjectiv e stress in female elite athletes and non-athletes: Ev idence from cortisol analyses

Stress response can be considered a consequence of psychological or physiological threats to the human organism. Elevated cortisol secretion represents a biological indicator of subjective stress. The extent of subjectively experienced stress depends on individual coping strategies or self-regulation skills. Because of their experience with competitive pressure, athletes might show less pronounced biological stress responses during stressful events compared to non-athletes. In the present study, the short version of the Berlin Intelligence Structure Test, a paper-pencil intelligence test, was used as an experimental stressor. Cortisol responses of 26 female Swiss elite athletes and 26 female non-athlete controls were compared. Salivary free cortisol responses were measured 15 minutes prior to, as well as immediately before and after psychometric testing. In both groups, a significant effect of time was found: High cortisol levels prior to testing decreased significantly during the testing session. Furthermore, athletes exhibited reliably lower cortisol levels than non-athlete controls. No significant interaction effects could be observed. The overall pattern of results supports the idea that elite athletes show a less pronounced cortisol-related stress response due to more efficient coping strategies.


Introduction
Elite athletes are often confronted w ith highly stressful ev ents during contests.Not only because the participation in important competitions can lead to nerv ousness, but also due to critical situations during a contest.Such critical situations leading to acute stress could be a tie-break situation in a tennis game, a referee's w rong decision, experiencing pain or being reprimanded by the coach.I t has been demonstrated that coping abilities hav e a crucial impact on the performance and personal satisfaction of athletes (Anshel, 1990).The inability to cope w ith acute stressors during contests can hav e a negativ e effect on psychological processes such as concentration, attentional focus, and arousal.Low er concentration or attention can lead to low er sport performance.I n case that the athlete is able to cope w ith the stressful demands, the crucial processes and the performance stay unaffected (Anshel, 1990;Smith, 1986).I t is w ell-know n that stress reactions depend on three components: the characteristics of the stressor, the person's appraisal of the situation, and indiv idual coping strategies (Lazarus & Folkman, 1984).Adequate coping strategies lead to less subjectiv ely experienced stress.Consequently, a successful athlete might experience ev ents during the contest generally as less stressful compared to a less successful athlete.The successful athlete might categorize an ev ent as a challenge rather than a threat and is more likely to adopt an adequate and effectiv e coping strategy.
Proceeding from these considerations, elite athletes, compared to non-athletes, may also experience low er subjectiv e stress in performance-related ev eryday-life situations.
From a biological v iew , there is a functional relationship betw een stress reaction and activ ity of tw o physiological systems, namely the sympatho-adreno-medullary (SAM) system and the hypothalamic-pituitary-adrenal (HPA) axis.Experienced stress activ ates the SAM system and inv olv es increased blood pressure, heart rate and the secretion of epinephrine and norepinephrine.The activ ation of the HPA axis represents a mainly endocrine response to stress.During stressful situations, especially during experienced feelings of helplessness as w ell as perceiv ed uncontrollability and inability to cope, the pituitary gland releases adrenocorticotropic hormone (ACTH) into the bloodstream w hich in turn leads to a release of corticosteroids, including cortisol, in the adrenal cortex.Cortisol responses reach their peak approximately 20-30 minutes after exposure to stress.Thus, elev ated cortisol secretion represents a biological indicator of subjectiv ely experienced stress.Therefore it is reasonable to assume that athletes show low er cortisol lev els during threatening or challenging situations compared to non-athletes (Weiner, 1992).

58
I t is necessary to distinguish betw een physical and psychological stressors.A large number of studies show ed, that the HPA-axis can be stimulated by physical activ ity.
For example, healthy participants show ed elev ated cortisol lev els during sporting activ ity on bicycle ergometers (Dav is, Gass, & Bassett, 1981;Mason, et al., 1973), treadmills (Luger, et al., 1987), and long-distance runs (Dessypris, et al., 1980).On the other hand, there is also ev idence for psychological or cognitiv e stimuli affecting the cortisol lev els.Mason (1968) postulated the follow ing basic components of a psychologically prov oked cortisol response: nov elty or unpredictability of a situation, coping w ith stressful and uncontrollable situations, anticipation of future challenging ev ents.For instance, elev ated cortisol lev els w ere found during anticipatory periods prior to exhausting muscle w ork (Mason, et al., 1973), academic or scholar exams (Hellhammer, Heib, Hubert, & Rolf, 1985) or public speeches (Lehnert, et al., 1989).
Hence, expecting an upcoming threatening or challenging ev ent can lead to a stress reaction represented by an increased cortisol secretion.
Studies comparing endocrine responses in athletes and non-athletes during psychologically stressful situations are extremely scant.I n a recent study, Rimmele et al. (2007) compared cortisol responses of physically trained and untrained males to psychosocial stress induced by means of the Trier Social Stress Test (TSST - Kirschbaum, Pirke, & Hellhammer, 1993).The trained group consisted of members of Sw iss national or Olympic teams recruited from endurance sports.The TSST consists of an anticipation period and a test period w hich includes a free speech and a mental arithmetic task in front of an audience.The trained group show ed significantly low er heart rate and cortisol responses to the stressor compared to the untrained group.I n a later study, Rimmele et al. (2009) found support for their prev ious findings.A group consisting of amateur sportsmen w as added to the experimental design.Amateur sportsmen and untrained man show ed the same cortisol lev els and only elite sportsmen exhibited reliably low er cortisol lev els after exposure to the TSST.
I n a prev ious study, Sinyor, Schw artz, Peronnet, Brisson, and Seraganian (1983) compared self-reported arousal and anxiety, heart rate, and biochemical measures, such as cortisol and prolactin in trained and untrained subjects prior, during, and after exposure to three different cognitiv e tasks.The three tasks consisted of a mental arithmetic task during exposure to w hite noise, 23 quiz questions of v arying difficulty, and a Stroop task.With this latter task, color-w ords w ere displayed on a screen and participants w ere required to report the color in w hich the w ord w as printed.The font color w as alw ays different from the color-w ord.Trained subjects show ed faster heart rate recov ery.The trained group had marginally higher cortisol v alues than the untrained group but this effect failed to reach statistical significance.
Subjective stress in female elite athletes and non-athletes 59 With regard to cortisol measures, the findings of the mentioned studies are quite inconsistent.There might be sev eral reasons for the discrepancy.A possible reason for these ambiguous results may represent differences in the operationalizations of athleticism.Synior et al. (1983) defined the trained group as subjects w ho participated heav ily in "aerobic activ ities".Rimmele et al. (2007) recruited only sportsmen w ho w ere members of national teams in endurance-trained sports.
Additionally Rimmele et al. (2009) w ere able to show that the lev el of physical activ ity affects cortisol lev els during stressful situations.Thus, athlete samples may hav e differed in the absolute lev el of physical fitness as much as in the efficiency of cognitiv e strategies for coping w ith stressful situations.Another possible reason that could account for div ergent results in the studies by Rimmele et al. (2007Rimmele et al. ( , 2009) ) and Synior et al. (1983) may be the fact, that different psychological stressors w ere applied in the tw o studies.For example, on the one hand, the HPA axis has been show n not to be particularly sensitiv e to mental arithmetic tasks (Biondi & Picardi, 1999).On the other hand, it has been argued that stressors containing a socioev aluativ e and uncontrollable element much more effectiv ely produce an endocrine response compared to other stressors (Dickerson & Kemeny, 2004).This seems to be a crucial point since the TSST, as used by Rimmele et al. (2007Rimmele et al. ( , 2009)), is characterized by a strong socio-ev aluativ e component (Kirschbaum, et al., 1993).
A social-ev aluativ e threat can be defined as a situation in w hich important aspects of self-identity are judged by others (Dickerson & Kemeny, 2004).While certain aspects of self-identity may be indiv idually important in special situations or among certain groups of people -for instance athletic ability among athletes -intelligence is considered a core aspect of self-identity, sensitiv e to social judgment across div erse situations and domains (Crocker & Wolfe, 2001).Consequently, psychometric assessment of intelligence should cause a stress response accompanied by increased cortisol secretion.Furthermore, athletes might exhibit low er cortisol lev els during performing on an intelligence test than non-athletes.
Although sev eral studies analyzed endocrine effects w hile performing different cognitiv e tasks (Dickerson & Kemeny, 2004), to our know ledge, the effect of standardized psychometric intelligence tests on cortisol lev els in athletes compared to non-athletes has not been inv estigated yet.I n the present study, therefore, the short v ersion of the Berlin I ntelligence Structure Test (Jäger, Süss, & Beauducel, 1997), a paper-pencil intelligence test, w as used as an experimental stressor.Cortisol measures prior and after the test hav e been taken as a physiological indicator of the subjectiv e lev el of experienced stress.
We expected that the social-ev aluativ e threat ev oked by the intelligence test w ould elev ate the cortisol lev els significantly in both groups.Furthermore w e predicted, in accordance w ith the explanations abov e, that the athletic sample w ould experience generally less subjectiv e stress, hence show low er cortisol lev els compared to the non-athletic sample.Kirschbaum, Wüst, and Hellhammer (1992) inv estigated sex differences in cortisol responses to psychological stress in four independent studies.Men show ed higher cortisol responses to stress and the anticipation of upcoming psychological stress compared to w omen.The authors suggested that sex differences in cognitiv e and/or emotional reactions to stress exposure may hav e led to the higher cortisol lev els in men compared to w omen.To av oid any possible gender effects on subjectiv ely experienced stress our sample includes solely female participants.A female sample may be of particular interest because prev ious studies on endocrine responses in athletes and non-athletes during psychologically stressful situations primarily inv estigated men.For the measurement of psychometric intelligence, the short v ersion of the Berlin I ntelligence Structure (BI S) test (Jäger, et al., 1997) Bucik and Neubauer (1996) and by Süß, Oberauer, Wittmann, Wilhelm, and Schulze (2002).

Assessment of cortisol lev els
Cortisol concentration w as assessed by collecting saliv a samples using Saliv ette (Sarstedt, Sev elen, Sw itzerland) collection dev ices.Saliv a contains free cortisol w hich is a reliable measure for cortisol concentration in blood plasma (Kirschbaum, 1991).
Assessment took place at three different time points.Participants collected saliv a samples themselv es by chew ing on the cotton sw ab for one minute.All saliv a samples w ere kept frozen at -20 degrees Celsius until biochemical analysis.The samples w ere analyzed as described by Westermann, Demir & Herbst (2004).After preparing the samples by centrifuging at 3000 rounds per minute for fiv e minutes to get a supernatant of low v iscosity, a commercially av ailable immunoassay w ith chemiluminescence detection w as used (CLI A; I BL-Hamburg, Hamburg, Germany).
I n this study all cortisol measures are reported in nmol/l.

Time course of the study
All test sessions started in the afternoon and took 65 minutes.After arriv al, participants w ere briefly informed about the study protocol and instructed how to collect the first saliv a specimen (T1).The second sample (T2) w as obtained 15 minutes later immediately before psychometric assessment of intelligence.During the 15 minutes prior to the second saliv a sampling, participants filled in a Europe's Journal of Psychology 62 questionnaire on personal details, such as demographic information and sportiv e activ ities, and w ere giv en general instructions on the subsequent psychometric assessment of intelligence.After collection of the second saliv a specimen, participants w orked on the short form of the BI S intelligence test for 50 minutes.The final sample (T3) w as collected after completion of the intelligence test.

Results
Descriptiv e statistics for cortisol lev els (three time points) and the operation-related abilities as measured w ith the BI S test (Reasoning, Speed, Memory, and Creativ ity) are presented in Table 1.

Discussion and conclusion
I n the present study, saliv ary cortisol levels prior to and immediately after completion of an intelligence test w ere inv estigated w ith emphasis on differences betw een athletes and non-athletes.The analyses show ed differences in saliv ary cortisol lev els ov er time.After high cortisol measures prior to the intelligence test, cortisol lev els decreased significantly during the testing session in both groups.Furthermore differences in cortisol lev els betw een athletes and non-athletes w ere found.
Throughout the three time points of cortisol measurement the athletes show ed low er cortisol lev els compared to non-athletes.All participants refrained from physical activ ity, eating meals, smoking, and drinking soft drinks w ith low pH at least one hour prior to the experiment.All participants w ere female and of similar age.Controlling these v ariables, the constant difference ov er time in cortisol lev els betw een athletes and non-athletes supports the idea that highly trained and professional athletes use better or more efficient coping strategies compared to untrained people (Anshel, 1990).
To answ er the question of w hether intelligence tests produce an increased cortisol response, it is necessary to distinguish betw een base and reaction lev els in our data.
Base cortisol lev els represent endocrine activ ity w ithout responding to a stressor.Only reaction measures w ould be indicators for subjectiv ely experienced stress.The analyses rev ealed reliably low er saliv ary cortisol lev els in athletes compared to nonathletes at all three time points of cortisol measurement.I f our data consisted only of basal measures in situations w here stressors are absent, highly trained athletes should hav e low er cortisol lev els compared to untrained indiv iduals.According to Luger et al. (1987), repetitiv e high physical activ ity leads to alterations in baseline as w ell as in reaction cortisol lev els.Highly trained runners show ed elev ated basal cortisol concentrations compared to sedentary participants and moderately trained runners.I n the same study, highly trained runners show ed low er cortisol reactions during physical activ ity compared to untrained controls.Based on these findings, our cortisol data do not represent basal cortisol measures because athletes show low er cortisol lev els compared to non-athletes.Both groups did not engage in physical activ ity at least one hour prior to the experiment.This leads to the assumption that the higher saliv ary cortisol secretion in athletes compared to non-athletes in the present study w as due to the anticipation of the psychometric assessment of intelligence.Further support for this assumption prov ides the decrease of s aliv ary cortisol during the test session.Prev ious studies on the circadian activ ity rev ealed a moderate decrease of saliv ary cortisol during the afternoon.The decrease of saliv ary cortisol observ ed from T2 to T3 in the present study is larger than the de crease w hich could hav e been expected due to normal circadian activ ity (Kirschbaum, 1991).A third argument for the assumption that the cortisol lev els in our data represent stress reactions are the absolute v alues for each assessment time.Considering the daytime of assessment our cortisol measures for athletes and non-athletes of 7.02 ± 3.45 (T1), 7.13 ± 3.71 (T2), and 5.04 ± 2.63 nmol/l (T3) w ere higher than baseline Subjective stress in female elite athletes and non-athletes 65 saliv ary cortisol levels as measured by Kirschbaum (1991) w here mean saliv ary cortisol lev els w ere of 4.50 ± 3.5 nmol/l (n = 708) betw een 3 pm and 5 pm.
The first cortisol measure (T1) might represent an anticipatory response prior to the assessment of intelligence.During the first measure of cortisol secretion 15 minutes prior to the assessment of intelligence, participants knew that they w ere supposed to undergo psychometric assessment of intelligence.A reason for a stress reaction prior to the intelligence test might be the uncertainty about the form and content of the upcoming cognitiv e tasks and, of course, the social-ev aluativ e element of such a test.I ntelligence builds a fundamental aspect of self-identity (Crocker & Wolfe, 2001) and w as going to be measured by a test.Consistent w ith earlier studies (Hellhammer, et al., 1985;Lehnert, et al., 1989;Mason, et al., 1973), the anticipation of future threatening or challenging ev ents elev ated cortisol lev els.Both groups, athletes and non-athletes, show ed a stress reaction.The fact that athletes show ed low er cortisol secretion compared to non-athletes at T1 indicates that athletes used better coping strategies prior to the actual stressor.
The next cortisol specimen (T2) w as taken immediately prior to the assessment of intelligence.Again athletes show ed significantly low er stress reactions compared to non-athletes.Although the measures at T2 did not differ significantly from T1 w ithin the groups, there w as a tendency for a slight decrease in the athletes group and an increase in the non-athletic group.This differential pattern of results suggests that athletes not only use better coping strategies but also engage them earlier.
Cortisol secretion decreased significantly in both groups from immediately before to immediately after the assessment of intelligence.This decrease of cortisol secretion indicates that subjectiv ely experienced stress decreased during the assessment of intelligence.A possible reason for the decreasing stress could be that participants became familiar w ith the nature of the tasks and had the chance to judge or estimate their performance.Such an adaptation to the situation should decrease the social-ev aluativ e threat by the assessment of intelligence.Without any physical or psychological stressors the sole w orking on cognitiv e or arithmetical tasks does not lead to higher cortisol activ ity (Biondi & Picardi, 1999).Because w e w ould not expect athletes to hav e low er cortisol lev els compared to non-athletes in situations w ithout any stressors, the significant difference betw een athletes and non-athletes in T3 might indicate that base lev els might not hav e been reached yet immediately after intelligence testing since it takes cortisol measures around 20 minutes to reach their maximum lev els after exposure to a stressor and about 40 minutes on av erage to return to basal lev els again (Dickerson & Kemeny, 2004;Kirschbaum, 1991).
Mental performance data is show n in Table 1.Possible differences in intelligence betw een athletes and non-athletes are not part of this study and are therefore discussed v ery briefly.Non-athletes show ed a better performance in Reasoning and Memory compared to athletes.One possible explanation could be that athletes took the test session less seriously w hich might hav e an impact on the subjectiv e social-ev aluativ e threat of such an intelligence test.Consequently this idea w ould also explain the low er cortisol lev els.The missing differences in the other operations and the fact, that the athletes and their coaches receiv ed intelligence profiles as feedback do not support this idea.Another explanation could be that the nonathletes group used better memory strategies.Some of the Psychology students might hav e acquired useful strategies in study courses and it is possible that they also hav e more experience in intelligence testing, especially concerning tasks measuring reasoning skills.Also relations betw een cortisol secretion and memory performance should be mentioned here briefly.A large amount of literature documents effects of stress on memory.How ev er, the direction of this effect does not seem to be consistent.While some studies found a negativ e effect of cortisol on memory, others show ed that stress can enhance memory performance (Buchanan, Tranel, & Adolphs, 2010).I t could be that in our non-athletes sample the higher cortisol lev els had a positiv e effect on memory performance, but the fact that enhancing effects w ere mainly found in emotionally arousing memory tasks does not support this assumption (Buchanan & Lov allo, 2001;Cahill, Gorski, & Le, 2010).The examination of differences in intelligence scores betw een athletes and non-athletes and the question in w hat extent cortisol lev els affect psychometric intelligence demand further research.
To av oid any clues w hich could lead to an additional threat of self-esteem and, therefore, to higher stress reactions, no self-ev aluation of the subjectiv ely experienced stress has been conducted.I n future studies information about subjectiv ely experienced stress might help interpret and assure the meaning of cortisol data.Another limitation of the present study can be seen in the fact that no clear baseline measure was obtained.As mentioned above, cortisol levels reach their peak around 20 minutes after exposure to a stressor.It should be noted, however, that several samplings during the period after the completion of the intelligence test would hav e been helpful to better understand the differences between athletes and nonathletes.

Roland Seiler
PhD, holds a position as professor of sport science, mainly sport psychology, at the Participants Participants w ere 26 female athletes ranging in age from 18 to 29 years (mean age ± SD = 21.0 ± 2.9 years) and 26 female non-athletes ranging in age from 18 to 31 years (mean age ± SD = 21.2 ± 2.7 years).The athletes group consisted of 11 members of the Sw iss national foot orienteering team, six members of the Sw iss national judo team, and nine members of top Sw iss floorball teams.All participating athletes had accomplished secondary education and had acquired a general qualification for univ ersity entrance.As a matter of fact, 24 of the 26 athletes w ere enrolled in Univ ersity or adv anced technical college classes.Non-athlete controls w ere undergraduate psychology students from the Univ ersity of Bern w ho neither w ere members of a sports club nor reported w ork out on a regular basis.All participants w ere offered indiv idual intelligence profiles.I n order to av oid interference w ith assessment of indiv idual cortisol lev els, participants w ere instructed to refrain from physical activ ity, drinking soft drinks w ith low pH, eating meals, and smoking at least one hour prior to testing.I nformed consent w as obtained from each participant before the experiment began.Procedure Berlin I ntelligence Structure (BI S) Test w as administered.This test is based onJäger's (1984) BI S model of intelligence.The BI S model classifies cognitiv e abilities w ith respect to the required mental operations and, concur rently, w ith respect to the contents of the processed tasks.Four operations (Reasoning, Speed, Memory, and Creativ ity) and three contents (v erbal, numerical, and figural) are differentiated.The short v ersion of the BI S comprises 15 subtests.Reasoning is assessed by six subtests (tw o v erbal, tw o numerical, and tw o figural subtests), w hile Speed, Memory, and Creativ ity are measured w ith one v erbal, one numerical and one figural subtest, respectiv ely.Scores on each of the 15 subtest w ere z standardized.The four operations w ere quantified by av eraging the indiv idual z standardized scores on the six or three subtests for each operation.A more detailed description of the BI S model and an ev aluation of the BI S test are prov ided by three different time points of assessment (15 minutes before, immediately before and immediately after intelligence testing) in nmol/l, and mean z-scores and standard error of the mean (SEM) for each operation of the Berlin I ntelligence Structure Test in athletes (N = 26) and non-athletes (N = 26).A tw o-w ay analysis of v ariance w as conducted w ith athletes and non-athletes being tw o levels of a group factor and the three measures of cortisol secretion [15 minutes prior to the assessment (T1), immediately before the assessment (T2) and immediately after the assessment (T3)] being three lev els of a repeatedmeasurement factor.This analysis w as computed to inv estigate a possible differential temporal course of cortisol secretion from 15 minutes prior to the psychometric assessment of intelligence until the end of the assessment.The main effect of the Subjective stress in female elite athletes and non-athletes 63 temporal course yielded statistical significance [F(2,100) = 22.20; p < .001;η 2 = .31].After high cortisol measures prior to the intelligence test, cortisol lev els decreased during the testing session in both groups.Post-hoc Scheffé-tests rev ealed that cortisol secretion w as significantly low er at T3 compared to T2 but did not v ary significantly from T1 to Also the group effect w as statistically significant [F(1,50) = 10.50;p < .01;η 2 = .17].Athletes show ed reliably low er absolute cortisol lev els (5.19 ± .53nmol/l) compared to non-athletes (7.61 ± .53nmol/l).Although the interaction effect just failed to reach statistical significance [F(2,100) = 3.06; p = .052;η 2 = .06],there w as a slight increase in cortisol secretion from T1 to T2 in the non-athletic group, w hile the athletic group show ed a slight decrease of cortisol secretion from T1 to T2.

Figure 1 .
Figure 1.Mean cortisol lev els and standard error of mean for each assessment time point (T1: Cortisol sampling on arriv al 15 minutes before intelligence testing; T2: immediately before intelligence testing w hich w as 15 minutes after arriv al; T3: immediately after intelligence testing w hich w as 65 minutes after arriv al).
Institute of Sport Science, University of Berne, Sw itzerland.His main research interest centres around first, the effects of sport involvement on a series of psychological dimensions such as, for example, stress, social anxiety, mental toughness, or attitudes tow ards violence and deviant behaviour, and second, social psychological themes of group action and group performance such as, group collective efficacy, emotions in groups, and coach-athlete and teacher-pupil relation and interaction.Research is embedded into the framework of psychological action theory.A major concern is the quality management of applied sport psychology services, and a specific interest lies in the historical development of sport psychology.Thomas RammsayerIs professor of psychology at the University of Bern, Switzerland.He is trained in experimental, biological, differential, and clinical psychology and received his degrees from the Universities of Tübingen and Giessen, Germany.His research interests lie in experimental and biological psychology, including temporal information processing, pharmacopsychology, the biological basis of individual differences, and research on intelligence.

Table 1 :
Mean and standard error of the mean for the cortisol concentration at the