Perfectionism and the Cortisol Response to Psychosocial Stress in Men

LI, MS, SARA GROESSBAUER, MS, PETRA H. WIRTZ, PHD, SIGRID ELSENBRUCH, PHD, LULJETA EMINI, MS, KATHARINA R UDISU AND ULRIKE EHLERT, PHD
Objective: To test the hypothesis that perfectionism is an important moderator of the neuroendocrine stress response, with higher perfectionism predicting increased neuroendocrine activation. Methods: A total of 50 middle-aged men underwent an acute standardized psychosocial stress task (Trier Social Stress Test). Perfectionism, cognitive appraisal of the stressful situation, trait anxiety, and various personality characteristics were assessed with questionnaires. Salivary cortisol, plasma epinephrine and norepinephrine, blood pressure, and heart rate were analyzed before and after stress. Circadian profiles of cortisol secretion during the day and in response to awakening were analyzed to assess basal activity of the hypothalamus-pituitary-adrenal (HPA) axis. Multiple regression analyses were conducted to identify predictors of the neuroendocrine stress response. Results: Perfectionism was significantly associated with area under the total response curve with respect to increase (AUCi) of cortisol (r 0.322, p .046), but not with AUCi of norepinephrine (r 0.217, p .152) or AUCi of epinephrine (r 0.116, p .477). Hence, AUCi of cortisol was the main criterion. As possible predictors, trait anxiety, neuroticism, vital exhaustion, secondary appraisal, depression, and openness were considered. Regression analyses demonstrated that only perfectionism ( 0.45, p .002) and secondary appraisal ( 0.50, p .001) were independent predictors of AUCi of cortisol, the final model explaining 45% of the total variance in cortisol response (R2 0.45, shrunken R2 [sR2] 0.38); perfectionism alone accounted for 18% of this variance (R2 0.18, sR2 0.19). Conclusion: The typical cognitions, and presumably the associated emotions, of perfectionists seem to contribute independently to stress-induced bodily responses, including HPA axis activation, in response to psychosocial stress. Key words: perfectionism, psychosocial stress, stress hormones, cortisol, personality, public speaking stress. HPA hypothalamic-pituitary-adrenal; TSST Trier Social Stress Test; AUC area under the curve; HR heart rate; BP blood pressure; PASA primary appraisal secondary appraisal; MPS Frost Multidimensional Perfectionism Scale; CMD concern over mistakes and doubts.

INTRODUCTION he hypothalamic-pituitary-adrenal (HPA) axis, one of the physiological systems that regulate metabolism and energy production, is centrally involved in the regulation of physiological responses to stress. A large body of human and animal literature indicates increased levels of HPA axis hormones under physically and psychologically threatening conditions, particularly in situations characterized by novelty, uncontrollability, unpredictability, and ego involvement (1 4). Recently, it has been noted that situations that threaten to demean ones social image and thereby elicit feelings of low social worth (e.g., shame or humiliation) and decrease social self-esteem are prone to elicit marked HPA axis activation (5). Emotions evoked by the psychosocial situational aspects likely play a pivotal role in public speaking stress (6), a commonly applied psychosocial stress paradigm included in the Trier Social Stress Test (TSST) (7), which combines a speech in front of an audience and a mental arithmetic task. In a recent study, Gruenewald et al. applied a modified version of the TSST and documented stress-induced elevations in cortisol only in a social evaluation condition, experimentally de-

From the Department of Clinical Psychology and Psychotherapy (P.H.W., L.E., K.R., S.G., U.E.), Psychological Institute, University of Zurich, Zurich, Switzerland; Department of Medical Psychology (S.E.), University Clinic of Essen, University of Duisburg-Essen, Essen, Germany. Address correspondence and reprint requests to Petra H. Wirtz, Department of Clinical Psychology and Psychotherapy, University of Zurich, Binzmu hlestrasse 14, Box 26, CH-8050 Zurich, Switzerland. E-mail: p.wirtz@ psychologie.unizh.ch Received for publication Jan 17, 2006; revision received November 20, 2006. The study was supported by Grant 56233203 from the University of Zurich (P.H.W.). DOI: 10.1097/PSY.0b013e318042589e Psychosomatic Medicine 69:249 255 (2007)
0033-3174/07/6903-0249 Copyright 2007 by the American Psychosomatic Society

signed to increase threat to social self (5). Interestingly, the magnitude of stress-induced cortisol increase was related to the extent to which the subjects experienced shame and loss of self-esteem (5). This is consistent with our recent finding that supports the role of anticipatory cognitive appraisal in mediating cortisol stress responses (8). Interindividual differences in HPA axis responses to psychological stressors have been attributed to personality traits. For example, hostile subjects demonstrate more pronounced neuroendocrine responses to psychological stress (9,10). Other personality traits which have been linked to neuroendocrine as well as immune stress responsiveness include agreeableness, extroversion and neuroticism (10), optimism (11), self-efficacy (12), neuroticism (13), and anger expression style and defensiveness (14). Perfectionism is a personality characteristic that has thus far received little attention in the context of experimental stress research, although it may play an important role in modifying psychophysiological responses to psychosocial stress. Perfectionists are characterized by the setting of excessively high standards for performance accompanied by overly critical self-evaluation (15,16). This critical self-evaluation is strongly associated with a negative cognitive pattern of perfectionism mainly characterized by doubts about actions and concern about mistakes, especially evoked in stressful situations (1517). At the emotional level, excessively high personal standards for performance are likely to predispose perfectionists to fear failure and also to perceive failure subjectively when facing a demanding task, especially in a social context (1719). Little is known about stress reactivity in perfectionists. Relative to the findings of Gruenewald et al. (5) and Gaab et al. (8), we propose that based on the characteristic cognitive appraisal of perfectionism, particularly the high expectations for ones own performance, individuals with higher perfectionism may be particularly prone to fear failure in a socially demanding situation such as the public speaking stress paradigm. Accordingly, we hypothesize that perfectionism is an important mediator of the cor249

P. H. WIRTZ et al.
tisol stress response, with higher perfectionism predicting increased cortisol release. We exploratively studied the psychological and physiological responses to public speaking stress in a group of healthy, middle-aged men. In addition to cortisol, we measured catecholamines to assess a wider spectrum of stress hormone responses. We then conducted regression analyses to identify associations between perfectionism and neuroendocrine stress reactivity. Moreover, we tested whether these associations were confounded by other psychological variables either associated with perfectionism or with hormone stress reactivity, or by their interaction with perfectionism. In addition, we assessed circadian profiles of cortisol secretion in response to awakening and during the day to address if basal HPA axis functions would be associated with perfectionism. METHODS Recruitment of Study Participants
The study was part of a larger project (20). We recruited 50 healthy, middle-aged men who were medication free. Screening was completed by telephone interview using a screening questionnaire. Subjects with any acute or chronic somatic or psychiatric disorders were excluded. Additional exclusion criteria as obtained by subjects self-report included regular heavy exercise, smoking, alcohol and illicit drug abuse, heart disease, increased blood sugar and diabetes, elevated cholesterol, liver and renal diseases, chronic obstructive pulmonary disease, allergies and atopic diathesis, rheumatic diseases, and current infectious diseases. If personal history was not conclusive, the subjects primary care physicians were contacted for clarification. The Ethics Committee of the State of Zurich, Switzerland, formally approved the research protocol. The study was carried out in accordance with the Declaration of Helsinki principles. All subjects provided written informed consent. The study was conducted from April until October 2004. minutes after completion of the TSST for determination of salivary cortisol levels. Blood samples for assessment of catecholamines were taken immediately before and 0, 20, and 60 minutes after completion of the TSST. Heart rate was measured continuously, and blood pressure (BP) was assessed before and 0, 10, and 20 minutes after TSST, and continuously from 5 minutes before stress to 5 minutes after stress. Circadian profiles of cortisol secretion during the day and in response to awakening were also assessed. Subjects were instructed to collect saliva samples within 1 week after the TSST, using the salivette system during a normal work day. Sampling intervals were immediately after awakening and 15, 30, 45, and 60 minutes thereafter as well as at 8:00 AM, 11:00 AM, 4 PM, and 8 PM (23).

Measurements and Data Analysis Personality, Depression, Vital Exhaustion, and Trait Anxiety
To assess the big five personality factors (agreeableness, neuroticism, extroversion, openness, and conscientiousness), we used the shortened German version of the NEO Five Factor Inventory (NEO-FFI) (24,25). The extent of depression was assessed by means of a depression scale, the short version of the Allgemeine Depressionsskala (ADS-K) (26). The ADS-K is the German version of the Center for Epidemiological Studies Depression Scale (27). Vital exhaustion was assessed by the German version of the Shortened Maastricht Exhaustion Questionnaire (MEQ-S) (28). Anxiety was assessed by means of the trait version of the State-Trait-Anxiety-Inventory (STAI) (29).

Anticipatory Cognitive Stress Appraisal

To address anticipatory cognitive appraisal processes relevant for the TSST, we assessed primary appraisal (i.e., judgment about the significance of an event as stressful, positive, controllable, challenging, or irrelevant) as well as secondary appraisal (i.e., assessment of available coping resources and options when faced with a stressor), using a 16-item questionnaire for Primary and Secondary Appraisal (PASA) (8), which is based on the theoretical constructs proposed by Lazarus and Folkman (30). It contains the dimensions threat and challenge (primary appraisal), and self-concept of own abilities and control expectancy (secondary appraisal) (8). Good scores for internal and external statistical validity have been reported (8). The PASA was administered after the preparation time of the TSST.

Assessment of Perfectionism
Perfectionism was assessed with the German Version of the Frost Multidimensional Perfectionism Scale (MPS-d) (15,16). This 35-item questionnaire assesses the dimensions concern over mistakes and doubts (CMD), parental expectations and criticism, personal standards, and organization on 5-point rating scales for each item. For the purpose of this study, we focused on the dimension CMD (13 Items, minimum score 13, maximum score 65) as this scale seems to better capture the classical concept of perfectionism than the other subscales (17). Further, the CMD scale is significantly intercorrelated with the dimension personal standards (r 0.49, p .0001 in our dataset), indicating a strong positive association between these subscales (15). Thus, the CMD scale assesses not only the negative cognitive pattern lying at the heart of the perfectionism characterization but also indirectly reflects personal standards, both key components of perfectionism. Good psychometric properties including internal and external statistical validity of the MPS have been reported elsewhere for the original and the German version (15,16,21,22).

Neuroendocrine Responses to TSST

For cortisol, a saliva sample was collected using Salivette collection devices (Sarstedt, Rommelsdorf, Germany), stored at 20C until biochemical analysis. Cortisol concentrations were determined with a commercially available competitive chemiluminiscence immunoassay with high sensitivity of 0.16 ng/ml (LIA, IBL Hamburg, Germany). Intra- and interassay variability was 7.7% and 11.5%, respectively. Blood samples for measuring plasma norepinephrine and epinephrine were obtained via an indwelling catheter inserted 45 minutes before obtaining the blood samples. Blood samples were drawn into EDTA (ethylenediaminetetraacetic acid)-coated monovettes (Sarstedt, Numbrecht, Germany), and immediately centrifuged for 10 minutes at 2000 g; plasma was stored at 80C until analysis. Plasma norepinephrine and epinephrine values were determined by high performance liquid chromatography and electrochemical detection after liquid-liquid extraction (31,32). The limit of detection was 10 pg/ml. Intra- and interassay variance was 5% for both epinephrine and norepinephrine. To reduce error variance caused by imprecision of the intraassay, all samples from one subject were analyzed in the same run.

Study Design
All subjects reported to the laboratory on a single study day. The experimental sessions commenced between 2:00 PM and 4:00 PM and lasted for approximately 2 hours. Participants abstained from food and drink (other than water) for 2 hours before the experiment, and also from physical exercise, alcohol, and caffeinated beverages starting the evening before the test day. The TSST was implemented during the sessions (7). Subjects were required to participate in a mock job interview (5 minutes) and mental arithmetic (serial subtraction, 5 minutes) after a 5-minute preparation time in front of a panel of one man and one woman. After task completion, the subjects remained seated in a quiet room for another 60 minutes. Samples of stimulated (by chewing on cotton rolls) whole saliva were taken immediately before the TSST as well as 0, 10, 20, 30, 40, 50, and 60 250

Hemodynamic Measurements
Heart rate was obtained continuously via a portable heart rate monitor (S810, Polar System, Polar, Finland) (3335). BP was measured immediately before and after the stressor as well as 10 and 20 minutes after the stressor by sphygmomanometry, and continuously from 5 minutes before stress to 5 minutes after stress (i.e., average of speech and arithmetic BP) using a Vasotrac APM205A device (Medwave Inc., St. Paul, Minnesota) (36,37). For statistical analyses, average BP values during stress were adjusted to sphygPsychosomatic Medicine 69:249 255 (2007)

PERFECTIONISM AND STRESS REACTIVITY

momanometry readings; for regression analyses, mean arterial pressure (MAP) was calculated by the formula (2/3 diastolic BP) (1/3 systolic BP) to reduce the number of variables that had to be controlled. TABLE 1. Sociodemographic, Medical, and Psychological Characteristics of the Study Subjects n Age, years Body mass index, kg/m2 Systolic blood pressure, mm Hg Diastolic blood pressure, mm Hg High school degree (Swiss Matura), % Full- or part-time job, % Weekly work time, hours 50 50 50 50 45 46 44 Mean SEM 42.5 2.0 25.6 0.4 132.6 2.2 84.4 1.6 28.9 82.6 46.3 2.0

Statistical Analyses
All calculations were performed using SPSS (v11.0.1) software packages (SPSS, Chicago, Illinois). Data are presented as mean standard error of the mean. Results were considered statistically significant at the p .05 level and all tests were two-tailed. The optimal total sample size to detect an expected effect size of 0.35 in regression analyses with a power between 0.75 (maximum of six predictors) and 0.95 (minimum of one predictor) was 42. In case of missing data, cases were excluded listwise. Data were tested for normal distribution and homogeneity of variance using a Kolmogorov-Smirnov test. Using the trapezoid formula, we calculated areas under the total response curves, expressed as area under the measured time points with respect to ground (AUCg) for circadian cortisol secretion, with respect to increase (AUCi) for the responses to TSST of cortisol, epinephrine, and norepinephrine (38). Cortisol area under the curve (AUC) measurements were computed between rest and 60 minutes after stress. To address associations between perfectionism and hormonal stress reactivity and the potential confounding of these associations by other psychological variables or their interactions, we used the following procedure to identify potential predictors and criteria for regression analyses. First, we identified criterion variables (i.e., those stress hormones whose reactivity to the TSST was associated with the perfectionism variable) by calculating partial correlations between the AUCi of the respective stress hormone and perfectionism as we controlled for age, body mass index (BMI), and MAP. Second, we identified potential psychological confounders of these associations. To do this, we calculated Pearsons product-moment correlations between perfectionism and the respective psychological variable, and we calculated partial correlations between the AUCi of the identified stress hormone and the respective psychological parameter. A variable was considered to be a potential confounder when it was associated either with perfectionism or with AUCi of stress hormones associated with perfectionism. To achieve exploratory analyses, we set a liberal criterion of p .30 for identification of potential confounders. Given this exploratory approach, we also abstained from Bonferroni correction at this stage. Third, to avoid multiple testing, we calculated linear regression analyses (enter method) with the identified criterion (AUCi of the stress hormones correlating with perfectionism) and each of the identified explorative predictor variables. In subsequent analyses, we included only those variables that predicted the criterion variable with the regular significance level of p .05. We applied this significance level to minimize model over fitting given our sample size (39). Explained variance of the whole regression model is reflected by R2, whereas sR2 represents adjusted explained variance uninflated by the ratio of predictors to subjects (shrunken R2). To control for age, BMI, and MAP, we entered these variables as predictors in a first step in all analyses. To test if interactions between perfectionism and psychological variables independently predicted stress reactivity of a given physiological factor, interaction terms between perfectionism and significant psychological confounders were entered in a last step. All regression parameters were Z-transformed before analyses, rendering variables centered to the mean with a standard deviation of 1 to allow computation of interaction terms. Not used for modeling and testing, the study groupfor illustrative purposeswas categorized by median split into two groups of subjects based on their CMD perfectionism score: high perfectionism (n 24) or low perfectionism (n 26). Analysis of covariance (ANCOVA) results for repeated measurements for cortisol with age, BMI, and MAP as covariates were then computed.

n valid cases, SEM standard error of the mean.

Exploratory Identification of Predictors and Criteria: Correlational Analyses Criteria Partial correlation controlling for age, BMI, and MAP demonstrated that perfectionism was significantly associated with AUCi of cortisol (r 0.322, p .046), but not with AUCi of norepinephrine (r 0.217, p .152) or AUCi of epinephrine (r 0.116, p .477) (Table 2). Thus, only cortisol was considered a dependent variable in subsequent regression analyses. Predictors Perfectionism significantly correlated with trait anxiety (r 0.452, p .002), neuroticism (r 0.308, p .037), vital exhaustion (r 0.311, p .030), and with the secondary PASA scale secondary appraisal (r 0.321, p .023) (Table 3). When applying the more liberal criterion of p .30 for exploratory analyses (40), perfectionism correlated also significantly with depression (r 0.248, p .086) and openness (r 0.236, p .115). Partial correlation controlling for age, BMI, and MAP showed that AUCi of cortisol significantly correlated with openness (r 0.336, p .042) and secondary appraisal (r 0.395, p .013) and on a level of p .30 with neuroticism (r 0.289, p .082) and extroversion (r 0.321, p .052) (Table 3). All of these variables (n 7) were considered as potential predictors in subsequent regression analyses. Regression Analyses AUCi of cortisol was entered as a dependent variable. The predictors identified through our exploratory analyses (secondary
TABLE 2. Stress Hormone Responses to Psychosocial Stress and Correlations With Perfectionism Response to TSST AUCi cortisol AUCi epinephrine AUCi norepinephrine n 42 44 48 Mean SEM 527.7 77.5 212 125 5606 784 ra 0.322* 0.116 0.217

RESULTS Sociodemographic and Medical Characteristics of the Study Groups The final study sample consisted of 50 healthy, middleaged men, whose medical (BMI, basal BPs) and sociodemographic characteristics are detailed in Table 1.
Psychosomatic Medicine 69:249 255 (2007)

n number of valid cases; SEM standard error of the mean; ra partial correlation with perfectionism after correcting for age, body mass index, and mean arterial pressure; AUCi area under the total response curve with respect to increase. * p .05. 251

P. H. WIRTZ et al.
Table 3. Psychological Characteristics and Their Correlations With Perfectionism and AUC of Cortisol Stress Reactivity Mean SEM/n Perfectionism (CMD score) Depression (ADS-K score) Anxiety (STAI) Personality factors (NEO-FFI) 1. Agreeableness 2. Neuroticism 3. Extroversion 4. Openness 5. Conscientiousness PASA dimensions 1. Threat 2. Challenge 3. Self-concept 4. Control expectancy Primary appraisal Secondary appraisal Vital exhaustion (MEQ-s, total score) N (exhausted) 28.6 1.1/50 11.9 0.5/49 34.2 1.1/45 27.9 0.5/46 15.9 0.6/46 22.1 0.8/46 24.8 0.6/46 28.8 0.7/46 4.38 0.16/50 3.03 0.12/50 2.75 0.12/50 2.40 0.11/50 3.70 0.12/50 2.57 0.09/50 4.3 0.5/49 4 r Perf 1 0.248* 0.452*** 0.145 0.308** 0.034 0.236* 0.136 0.200* 0.050 0.182* 0.300** 0.109 0.321** .311** r AUCi Cort 0.322** 0.079 0.163 0.139 0.289* 0.321* 0.336** 0.054

0.095 0.395** 0.008

AUC area under the curve; SEM standard error of the mean; n number of valid cases; r Perf product-moment correlation with perfectionism; r AUCi Cort partial correlation with AUCi of cortisol after correcting for age, body mass index, and mean arterial pressure; CMD perfectionism scale doubts about actions and concern over mistakes; ADS-K Allgemeine Depressionsskala; STAI state-trait-anxiety-inventory; NEO-FFI NEO Five Factory Inventory; PASA Primary Appraisal Secondary Appraisal; Primary appraisal (threat challenge)/2. Secondary appraisal (self-concept control expectancy)/2 (8); MEQ-s Shortened Maastricht Exhaustion Questionnaire; N (exhausted) number of exhausted persons with a vital exhaustion score of 10. * p .30, ** p .05, *** p .01. Potential predictors in italics.

appraisal, vital exhaustion, depression, anxiety, neuroticism, openness, and extroversion) were entered as independent variables. After controlling for age, BMI, and MAP in a first step, perfectionism significantly predicted total AUCi of cortisol ( 0.31, p .046, n 42). This regression model predicted 24% of the observed variance in cortisol total AUCi (R2 0.24, sR2 0.15), of which perfectionism independently predicted 9% (R2 0.09, sR2 0.07). Next, we entered in a second step each of the exploratory identified predictors alone to test for potential confounding. Secondary appraisal and openness significantly predicted cortisol total AUCi (secondary appraisal: 0.38, p .013, R2 0.28/R2 0.13, sR2 0.20/sR2 0.12, n 2; openness: 0.33, p .042, R2 0.24/R2 0.10, sR2 0.15/sR2 0.08, n 40). The predictive contributions of neuroticism and extroversion were of marginal significance (cortisol total AUCineuroticism: 0.27, p .082, R2 0.21/R2 0.07, sR2 0.12/sR2 0.05, n 40; extroversion: 0.31, p .052, R2 0.23/R2 0.09, sR2 0.14/sR2 0.07, n 40). Neither vital exhaustion nor depression nor anxiety predicted cortisol total AUCi (vital exhaustion: p .96; depression: p .63; anxiety: p .34). To minimize model over fitting given our sample size (40), we next entered only significant predictors of cortisol AUCi, namely, secondary appraisal and openness, and in a third step, perfectionism. Perfectionism independently predicted total cortisol AUCi ( 0.42, p .009, R2 0.48/R2 0.12, sR2 0.39/sR2 0.12, n 40). Only secondary appraisal, but not openness, significantly predicted the total cortisol AUCi (secondary appraisal: 0.49, p .001; openness:
252

0.137, p .35). To ensure that these results were not confounded by predictors of cortisol AUCi with p .30, namely, extroversion and neuroticism, we entered these two variables in a fourth step. Again, only perfectionism ( 0.44, p .008) and secondary appraisal ( 0.47, p .002) remained significant predictors (openness: p .94; neuroticism: p .82; extroversion: p .30). Finally, we tested the independent predictive effects of interactions between perfectionism and the significant and marginally significant predictors of cortisol AUCi. After controlling for perfectionism and secondary appraisal, none of the interaction terms significantly predicted the cortisol stress response (total AUCisecondary appraisal by perfectionism: p .19; neuroticism by perfectionism: p .59; extroversion by perfectionism: p .15; openness by perfectionism: p .95). Thus, the final regression model included perfectionism and secondary appraisal as predictor variables and AUCi of cortisol as criterion at the same time controlling for age, BMI, and MAP, explaining 45% (R2 0.45, sR2 0.38) of cortisol AUCi variance. Perfectionism alone accounted for 18% of this variance (perfectionism: 0.45, p .002, R2 0.18, sR2 0.19; secondary appraisal: 0.50, p .001, R2 0.22, sR2 0.23). This final model has a power of 0.80; hence, the risk of Type II error exists, and we may have missed significant interactions. Figure 1 shows cortisol responses to TSST in two groups of subjects with high perfectionism (n 24) or low perfectionism (n 26), categorized by median-split based on CMD perfectionism scores. ANCOVA analyses with age, BMI, and MAP as covariates showed the high perfectionism group demPsychosomatic Medicine 69:249 255 (2007)

PERFECTIONISM AND STRESS REACTIVITY

DISCUSSION To address the hypothesis that perfectionism is an important mediator of the neuroendocrine stress response, we studied the psychological and neuroendocrine responses to public speaking stress in a group of healthy, middle-aged men. Regression analyses showed that perfectionism as measured by the CMD scale significantly predicted the cortisol, but not the catecholamine response to stress. Hence, higher perfectionism was associated with higher cortisol stress reactivity, and the final regression model explained 45% of the total cortisol stress response (AUCi), of which 18% were explained by perfectionism alone. Although perfectionism significantly correlated with trait anxiety, neuroticism, vital exhaustion, and secondary appraisal, and cortisol response correlated with openness and secondary appraisal, the only other significant predictor in the final model was secondary appraisal. However, perfectionism and secondary appraisal predicted cortisol stress reactivity independent of each other and independent of their interaction. These data are consistent with a previous report that anticipatory cognitive appraisal explained up to 35% of the variance in the salivary cortisol response to TSST (8). The typical cognitions (and presumably the associated emotions) of perfectionists seem to independently contribute to stress-induced bodily responses, including HPA axis activation in response to psychosocial stress. Of note, in contrast to the other perfectionism scales (data not shown), only the scale CMD carried the predictive weight. The following considerations should be taken into account. a) To minimize multiple testing and statistical over controlling given our sample size, all nonsignificant (p .30) variables were not entered into the regression approach. Thus, we cannot completely rule out a potential confounding effect of these variables or of other unmeasured variables on the perfectionism-cortisol relation. b) Although the MPS has repeatedly been described as a psychometrically sound instrument, we cannot exclude the possibility that there may be overlap between the construct of perfectionism and other dispositional characteristics such as shyness, social desirability, and Type A behavior, not assessed in this study. Although Type A behavior may also be characterized by high but ambiguous standards for performance (41 44), we would argue that the items of the MPS, like the dimension CMD, suggest fundamental differences between the constructs perfectionism and Type A behavior pattern. For example, none of the items assessing perfectionism relates to aggressiveness. Moreover, both perfectionists and persons with a Type A behavior pattern strive for achievement but they differ in the underlying motives and cognitions. c) We did not measure a number of characteristics that have been linked to physiologic stress reactivity, including Type A behavior, hostility, optimism, anger expression, and defensiveness, thereby potentially undermining the studys theoretical contribution. d) Despite sufficient external and internal validity scores, it cannot be ruled out that the PASA may reflect emotional reactions and coping responses generated by cognitive appraisal. Nevertheless, the finding of
253

Figure 1. Cortisol reactivity to psychosocial stress in subjects with high and low perfectionism. Subjects with high perfectionism, categorized by median split, showed higher cortisol secretion in response to Trier Social Stress Test (TSST) compared with individuals with lower perfectionism. Mean standard error of the mean values.

onstrated a significantly enhanced cortisol response (interaction group by time: F(2.8/100.6) 3.18, p .030). Cortisol Response to Awakening and Circadian Cortisol Profiles Partial correlation controlling for age, BMI, and MAP demonstrated that perfectionism was neither associated with the cortisol response to awakening (AUCg: r 0.0248, p .89) nor with circadian cortisol secretion (AUCg: r 010, p .95). For illustrative purposes, circadian cortisol secretion data are shown as mean values from subjects with high perfectionism (n 24) or low perfectionism (n 26), categorized by median-split based on CMD perfectionism scores (Figure 2). ANCOVA analyses with age, BMI, and MAP as covariates showed no significant differences between groups.

Figure 2. Cortisol response to awakening and circadian cortisol profiles in subjects with high and low perfectionism. Mean standard error of the mean values. Cortisol response to awakening and circadian cortisol profiles did not differ between subjects with high and low perfectionism. Psychosomatic Medicine 69:249 255 (2007)

P. H. WIRTZ et al.
secondary appraisal as a significant and independent predictor is consistent with the concept of perfectionism as the setting of exceedingly high standards for personal performance accompanied by overly critical self-evaluation (15,16), and when faced with a potential threat, a negative cognitive pattern characterized by doubts about actions and concern about mistakes (1517). e) Despite our attempts to reduce the number of variables to enter into the final regression model, this model is still at least a little over fitted. These data call for additional studies. It remains to be determined if higher perfectionism is associated with altered cognitive appraisals and enhanced HPA axis responses only in a socially demanding context, such as the public speaking stress paradigm, or if these associations also hold in the context of other types of stressors. Furthermore, to better understand the possible role of enhanced stress responses in the various long-term risks associated with perfectionism, including vital exhaustion and burnout, it would be important to assess responses to repeated stress because personality factors have previously been found to account for interindividual differences in habituation of adrenocortical stress responses to repeated psychological stress (45). Recent preliminary findings suggest an association between exhaustion and failure to habituate to repeated psychological stress (46). Finally, future studies should incorporate an examination of acutely experienced emotions, particularly fear of failure as well as anxiety, the latter of which may overlap with perfectionism. Fear of failure has been identified as one of the most important emotional consequences of dysfunctional cognitive patterns in perfectionists, and it motivates the various behavioral components of perfectionism (e.g., prolonged overtime work, workaholic behavior), which contribute to the long-term health risks associated with perfectionism, including chronic fatigue, exhaustion, and burnout (17,4750). Perfectionism may contribute to the development of fatigue-related syndromes by affecting the psychological and HPA axis response to acute stressors. However, it is important to note that the men in this study did not have clinically relevant fatiguerelated symptoms, reflected by an average vital exhaustion score of 4.3 0.5, which is markedly below the clinically significant range (10), as well as normal depression scores. Hence, our sample consisted of physically and mentally healthy subjects who did not suffer from clinically relevant feelings of exhaustion or work overload. Therefore, our findings cannot be compared with previous data on stress responses in samples with clinically relevant vital exhaustion or chronic fatigue syndrome, some of which have provided evidence of diminished HPA activity (51). We have recently shown in a prospective study that personality traits (i.e., high level of hostility and low level of self-efficacy) may constitute markers of vulnerability to the development of psychopathological symptoms after trauma exposure (52). These findings are of interest because an association between self-oriented perfectionism and low selfefficacy has been described (53), and (as in perfectionism) low self efficacy has been associated with perceived lack of con254

trollability of outcomes and low self worth. Therefore, our findings call for further study to address the associations between disturbed stress responsiveness and the long-term health risks associated with perfectionism as well as related personality traits such as self-efficacy. REFERENCES
1. Mason JW. A review of psychoendocrine research on the pituitarythyroid system. Psychosom Med 1968;30(Suppl):666 81. 2. Lovallo WR, Thomas TL. Stress hormones in psychophysiological research: emotional, behavioral, and cognitive implications. In: Cacioppo JT, Tassinary LG, Bernston GG, editors. Handbook of psychophysiology. Cambridge: Cambridge University Press; 2000. 3. Kirschbaum C, Hellhammer DH. Salivary cortisol in psychobiological research: an overview. Neuropsychobiology 1989;22:150 69. 4. Biondi M, Picardi A. Psychological stress and neuroendocrine function in humans: the last two decades of research. Psychother Psychosom 1999; 68:114 50. 5. Gruenewald TL, Kemeny ME, Aziz N, Fahey JL. Acute threat to the social self: shame, social self-esteem, and cortisol activity. Psychosom Med 2004;66:91524. 6. Dickerson SS, Kemeny ME. Acute stressors and cortisol responses: a theoretical integration and synthesis of laboratory research. Psychol Bull 2004;130:35591. 7. Kirschbaum C, Pirke KM, Hellhammer DH. The Trier social stress testa tool for investigating psychobiological stress responses in a laboratory setting. Neuropsychobiology 1993;28:76 81. 8. Gaab J, Rohleder N, Nater UM, Ehlert U. Psychological determinants of the cortisol stress response: the role of anticipatory cognitive appraisal. Psychoneuroendocrinology 2005;30:599 610. 9. Suarez EC, Sherwood A, Hinderliter AL. Hostility and adrenergic receptor responsiveness: evidence of reduced beta-receptor responsiveness in high hostile men. J Psychosom Res 1998;44:2617. 10. Miller GE, Dopp JM, Myers HF, Stevens SY, Fahey JL. Psychosocial predictors of natural killer cell mobilization during marital conflict. Health Psychol 1999;18:26271. 11. Cohen F, Kearney KA, Zegans LS, Kemeny ME, Neuhaus JM, Stites DP. Differential immune system changes with acute and persistent stress for optimists vs pessimists. Brain Behav Immun 1999;13:15574. 12. Butki BD, Rudolph DL, Jacobsen H. Self-efficacy, state anxiety, and cortisol responses to treadmill running. Percept Mot Skills 2001;92: 1129 38. 13. McCleery JM, Goodwin GM. High and low neuroticism predict different cortisol responses to the combined dexamethasoneCRH test. Biol Psychiatry 2001;49:410 5. 14. alAbsi M, Bongard S, Lovallo WR. Adrenocorticotropin responses to interpersonal stress: effects of overt anger expression style and defensiveness. Int J Psychophysiol 2000;37:257 65. 15. Frost RO, Marten P, Lahart CM, Rosenblate P. The dimensions of perfectionism. Cognit Ther Res 1990;14:449 68. 16. Sto ber J. The Frost multidimensional perfectionism scale: more perfect with four (instead of six) dimensions. Personality and Individual Differences 1998;24:48191. 17. Shafran R, Mansell W. Perfectionism and psychopathology: a review of research and treatment. Clin Psychol Rev 2001;21:879 906. 18. Flett GL, Hewitt PL, Blankstein KR, Mosher SW. Perfectionism, selfactualisation, and personal adjustment. Journal of Social Behavior and Personality 1991;6:14750. 19. Thompson T, Davis H, Davidson J. Attributional and affective responses of impostors to academic success and failure outcomes. Personality and Individual differences 1998;25:38196. 20. Wirtz PH, Ehlert U, Emini L, Ruedisueli K, Groessbauer S, Gaab J, Elsenbruch S, Von Kanel R: Anticipatory cognitive stress appraisal and the acute procoagulant stress response in men. Psychosom Med 2006;68: 85158. 21. Parker WD, Kittler Adkins K. A psychometric examination of the multidimensional perfectionism scale. Journal of Psychopathology and Behavioral Assessment 1995;17:323334. 22. Purdon C, Antony MM, Swinson RP. Psychometric properties of the Frost multidimensional perfectionism scale in a clinical anxiety disorder sample. J Clin Psychol 1999;55:1271 86. 23. Gaab J, Baumann S, Budnoik A, Gmunder H, Hottinger N, Ehlert U: Psychosomatic Medicine 69:249 255 (2007)

PERFECTIONISM AND STRESS REACTIVITY

Reduced reactivity and enhanced negative feedback sensitivity of the hypothalamus-pituitary-adrenal axis in chronic whiplash-associated disorder. Pain 2005;119:219 24. Ostendorf F: Sprache und Perso nlichkeitsstruktur. Zur Validita t des Fu nfFaktoren-Modells der Perso nlichkeit. Regensburg, Germany: Roderer; 1990. Schallberger U, Venetz M: Kurzversion des MRS-Inventars von Ostendorf (1990) zur Erfassung der fu nf grossen Perso nlichkeitsfaktoren. Zurich: University of Zurich; 1999. Hautzinger M, Bailer M: Allgemeine Depressionsskala. Weinheim: Beltz Test GmbH; 1993. Radloff LS. The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement 1977;3: 385 401. Wirtz PH, Von Kanel R, Schnorpfeil P, Ehlert U, Frey K, Fischer JE. Reduced glucocorticoid sensitivity of monocyte interleukin-6 production in male industrial employees who are vitally exhausted. Psychosom Med 2003;65:672 8. Laux L, Glanzmann P, Schaffner CD. Das State-Trait-Angstinventar (Testmappe mit Handanweisung, Fragebogen STAI-G Form X 1 und Fragebogen STAI-G Form X 2. Weinheim: Beltz; 1981. Lazarus RS, Folkman S. Stress, appraisal, and coping. New York: Springer Publishing Company; 1984. Ehrenreich H, Schuck J, Stender N, Pilz J, Gefeller O, Schilling L, Poser W, Kaw S. Endocrine and hemodynamic effects of stress versus systemic CRF in alcoholics during early and medium term abstinence. Alcohol Clin Exp Res 1997;21:128593. Smedes F, Kraak JC, Poppe H. Simple and fast solvent extraction system for selective and quantitative isolation of adrenaline, noradrenaline and dopamine from plasma and urine. J Chromatogr 1982;231:2539. Kudielka BM, Buske-Kirschbaum A, Hellhammer DH, Kirschbaum C. Differential heart rate reactivity and recovery after psychosocial stress (TSST) in healthy children, younger adults, and elderly adults: the impact of age and gender. Int J Behav Med 2004;11:116 21. Kudielka BM, Schommer NC, Hellhammer DH, Kirschbaum C. Acute HPA axis responses, heart rate, and mood changes to psychosocial stress (TSST) in humans at different times of day. Psychoneuroendocrinology 2004;29:98392. Nater UM, Rohleder N, Gaab J, Berger S, Jud A, Kirschbaum C, Ehlert U. Human salivary alpha-amylase reactivity in a psychosocial stress paradigm. Int J Psychophysiol 2005;55:333 42. Belani KG, Buckley JJ, Poliac MO. Accuracy of radial artery blood pressure determination with the Vasotrac. Can J Anaesth 1999;46: 488 96. Belani K, Ozaki M, Hynson J, Hartmann T, Reyford H, Martino JM, Poliac M, Miller R. A new noninvasive method to measure blood pressure: results of a multicenter trial. Anesthesiology 1999;91:686 92. Pruessner JC, Kirschbaum C, Meinlschmid G, Hellhammer DH. Two formulas for computation of the area under the curve represent measures of total hormone concentration versus time-dependent change. Psychoneuroendocrinology 2003;28:916 31. Babyak MA. What you see may not be what you get: a brief, nontechnical introduction to overfitting in regression-type models. Psychosom Med 2004;66:41121. Steyerberg EW, Eijkemans MJC, Harrell FE, Habbema JDF. Prognostic modeling with logistic regression analysis: in search of a sensible strategy in small data sets. Med Decis Making 2001;21:4556. Matthews KA, Haynes SG. Type A behavior pattern and coronary disease risk. Update and critical evaluation. Am J Epidemiol 1986;123:923 60. Delamater AM, Albrecht R, Smith JA, Strube MJ. Cardiovascular correlates of type A behavior components during social interaction. J Psychosom Res 1989;33:64150. Strube MJ, Berry JM, Lott CL, Fogelman R, Steinhart G, Moergen S, Davison L. Self-schematic representation of the type A and B behavior patterns. J Pers Soc Psychol 1986;51:170 80. Strube MJ, Boland SM, Manfredo PA, Al-Falaij A. Type A behavior pattern and self-evaluation of abilities: empirical tests of the selfappraisal model. J Pers Soc Psychol 1987;52:956 74. Kirschbaum C, Prussner JC, Stone AA, Federenko I, Gaab J, Lintz D, Schommer N, Hellhammer DH. Persistent high cortisol responses to repeated psychological stress in a subpopulation of healthy men. Psychosom Med 1995;57:468 74. Kudielka BM, von Kanel R, Preckel D, Zgraggen L, Mischler K, Fischer JE: Exhaustion is associated with reduced habituation of free cortisol responses to repeated acute psychosocial stress. Biol Psychol 2006;72: 14753. White C, Schweitzer R. The role of personality in the development and perpetuation of chronic fatigue syndrome. J Psychosom Res 2000;48: 51524. Magnusson AE, Nias DK, White PD. Is perfectionism associated with fatigue? J Psychosom Res 1996;41:377 83. Mitchelson JK, Burns LR. Career mothers and perfectionism: stress at work and at home. Personality and Individual Differences 1998;25: 477 85. Grosch WN, Olsen DC. Clergy burnout: an integrative approach. J Clin Psychol 2000;56:619 32. Ehlert U, Gaab J, Heinrichs M. Psychoneuroendocrinological contributions to the etiology of depression, posttraumatic stress disorder, and stress-related bodily disorders: the role of the hypothalamus-pituitaryadrenal axis. Biol Psychol 2001;57:14152. Heinrichs M, Wagner D, Schoch W, Soravia LM, Hellhammer DH, Ehlert U. Predicting posttraumatic stress symptoms from pretraumatic risk factors: a 2-year prospective follow-up study in firefighters. Am J Psychiatry 2005;162:2276 86. Hart BA, Gilner FH, Handal PJ, Gfeller JD. The relationship between perfectionism and self-efficacy. Personality and Individual Differences 1998;24:109 13.

39.

24. 25. 26. 27. 28.

40.

41. 42.

43.

44.

29. 30. 31.

45.

46.

32. 33.

47.

48. 49.

34.

50. 51.

35. 36. 37. 38.

52.

53.

Psychosomatic Medicine 69:249 255 (2007)

255