Effects of competition and its outcome on serum testosterone, cortisol and prolactin
Àrea de Psicobiologia. Universitat de València
Psychoneuroendocrinology 24, 551-556
The outcome of an aggressive/competitive encounter, victory or defeat, is a main cause of stress, which affects hormonal responses. In natural environments, the importance of the outcome is determined by its consequences: winners usually obtain compensations (food, territory or sexual mates) that increase their biological fitness. Experiments carried out with rodent show that repeated defeats lead to a hyperactivation of the hypothalamic-pituitary-adrenal (HPA) axis (Blanchard et al., 1993) with increases in ACTH (Skutella et al., 1994) and corticosterone (Blanchard et al., 1995). Corticosterone increases begin some 10 minutes after the defeat, and basal levels are not recovered until 240 minutes later. As well as this HPA activation, decreases in testosterone (T) (Blanchard et al., 1993, 1995; Huhman et al., 1991) and prolactin (PRL) (Huhman et al., 1995) have been reported. Although some authors have not found such T decreases (Dijkstra et al., 1992). At a behavioural level, social stress produced by defeat leads to anxiety (Avgustinovich et al., 1996; Menzaghi et al., 1996) and decreased aggressive behaviour (Potegal et al., 1993; Albonetti and Farabollini, 1994).
Hormonal changes resulting from social interactions may modify the future behaviour in order to maximise the possibilities of survival (Leshner, 1983; Bernstein et al., 1983; Mazur, 1985). From an evolutionary point of view, the biosocial hypothesis (Mazur, 1985) suggests that there is a feedback loop between an individual’s T level and his/her efforts to improve or maintain the social status. It states that aggressive behaviour leads to more or less stable endocrine adaptations, which can determine the organism’s future behaviour. Thus, winning would lead to an increase in the levels of T, which, in turn, stimulates competitivity. Conversely, defeat would involve a decrease in T that should reduce the possibilities of engaging in new potentially damaging encounters.
The study of the effect of victory and defeat on an animal’s hormonal levels can be approached by means of experimental aggression (Martínez, Calvo and Pico, 1998), but also by watching spontaneous aggressions which occur in natural social environments (Sapolsky, 1991). Tests on primates show that defeats lead to drastic decreases in the blood levels of T that affect the hierarchical organisation in a social group (Bernstein et al., 1974; 1983). In humans, some researchers have studied sports contests considering them socially acceptable environments where individuals compete in such a way that the outcome affects their sports status. Mazur and Lamb (1980) found that there were increases in the T levels of winners and decreases in losers of tennis matches. They concluded that these androgenic variations depended on the individual’s mood. Elias (1981) compared blood samples obtained ten minutes before and after wrestling bouts and reported that all the contenders presented increases in the levels of T and cortisol (C). The concentrations of T and C increased consistently during the combat, whereas the level of Sex Hormone Binding Globuline (SHBG) decreased. Even though the average titers of winners and losers did not show significant variations, the winners experienced percentual variations, which were significantly higher than those of the losers. Furthermore, only the winners showed a positive correlation between the increases in T and C. Hence, Elias concluded that the increases in the production of both hormones could be due to the same control mechanism that was victory-sensitive. Booth et al. (1989) also found significant differences in the levels of T but only when comparing samples obtained 15 minutes before the combat and samples obtained between 24 and 48 hours later. However, these authors considered their 36 cases (six samples for each of the six players) as independent cases, which is not methodologically acceptable (Archer, 1991). Studies on judo have shown non significant differences between winners and losers after a combat (Salvador et al., 1987; Salvador et al., 1990). However, the contenders showed hormonal changes that depended on their different sports category; with slight T increases in the judoists of higher ranks, whereas those of lower ranks showed decreases independently of the outcome of the combat. In addition, the score obtained from their sports records was positive significantly related to the percentual changes in T. This result coincides with some of the conclusions drawn in studies with primates (Coe et al., 1979; Bernstein et al., 1983) which made special emphasis on the relevance of status, in this case the hierarchical rank of each animal within the group. Moreover, it has to be considered that the stress generated by competition is produced even before the competition starts, due to uncertainty and threat, two main characteristics of competitive situations (Sapolsky, 1995). In humans, anticipatory rises of T and C have been observed in athletic competitions before the physical effort starts. Likewise, it does not seem to be caused by the specific characteristics of the activity, as has been reported in chess tournaments (Mazur et al., 1992), basketball games (Gonzalez-Bono et al., 1998) and tennis matches (Booth et al., 1989). As it has been shown, two main components of competitive stress affect hormonal responses: the previous uncertainty and the outcome. In the studies using physical sports competitions, the hormonal variations observed could also be due to the physiological stress caused by physical effort during the competition. To analyse this hormonal response without the interference of the physiological variation caused by effort, we have compared T, C and PRL responses to a judo combat with those after a non-competitive physical effort involving a similar energetic consumption. The design will allow us to compare individual hormonal responses in three situations in order to determine if there are differences between winners and losers, and what those differences are.
Table I: Physical, training and technical characteristics
15 to 22
157 to 189
51 to 95
0.19 to 0.31
strength (winners: x=152.14; losers: x=142.79),
Reference periods and peak times were of 35 sec and 15 sec, respectively, for protocol 1 and of 40 and 10 sec for protocol 2. In both protocols, subjects pedalled in reference periods at an intensity of 2.28 W/Kg of body weight, which was increased until 5.57 W/Kg during the peaks until completing the 5 min period.
Protocols were carried out in a Jaeger cycloergometer with a load range from 0 to 600 W. Venous blood samples were provided 10 minutes before and 10 minutes after the ergometric effort, whilst capillary blood samples were obtained in the minutes 1 and 3 of recovery. HR was registered continuously during the effort and until minute 3 of recovery.
The maximum value (HRmax) during the fight was estimated by exponential extrapolation (Eclanche, 1988) from the values obtained in seconds 35, 60 and 120 after the end of the fight. The duration of each combat was provided by the referee team. The total and the actual duration of every fight was registered and could be verified in the video-tape. Obvioulsy the duration of each combat was the same for both contenders.
The Vigor/Fatigue Scale included the items from the sub-scales of Vigor and Fatigue of the Profile of Mood States (POMS; McNair et al., 1972), which was used as an estimation of the impact derived from the COMBAT and the ERGOMETRY.
-Do you consider that this is a well-balanced fight?
The Outcome Evaluation Items, fulfilled just after the combat, was elaborated to assess the subject’s cognitive interpretation about the result they had obtained. It was answered on a Lickert scale identical to that of the Expectations Items. The items were:
-Are your expectations fulfilled?
Sport category measures
Before the assay, samples were thawed at ambience temperature and homogenised one-by-one in an automatic shaker. Assays were carried out on the same day to avoid further freezing of blood samples. The samples corresponding to each subject were analysed in duplicate in the same assay with internal controls of the assay-kit and three additional
Bio-Rad Clinical Diagnostic controls for low, medium and high values.
Table II: Hormone levels in the three sessions: Mean and Standard Deviation
Testosterone in ng/ml
Winners and losers showed different androgenic patterns (Table III). Winners experienced a very homogeneous androgenic response, which consisted in increases of T and FAI (in 13 out of 14 judoists). Losers showed a greater variability, including increases (9 for T and 10 for FAI) and decreases (5 and 4, respectively). A Chi2-test showed that the difference between patterns of winners and losers was statistically significant for T (p=.03). Conversely, C and PRL showed a quite uniform response in the whole sample. After the combat, C levels increased in 92.86% of subjects and PRL in 82.14%. No significant differences between winners and losers were found in these hormones.
Table III : Hormonal response patterns in winners and losers
Table IV. Means, (Standard Deviation) and t-test values of pre and post-hormone levels in COMBAT, after Sport Category
Low Sport Category
High Sport Category
Comparison of hormonal responses in COMBAT with those of the other two sessions. Non significant differences between [Lamax] (t=-0.24) or HR (t=-0.04) were found at the end of COMBAT and ERGOMETRY. Both groups showed higher hormone levels in COMBAT than in ERGOMETRY but differences were not always significant (Table V). In the winners, differences were significant for T and C (pre and post levels), previous level of SHBG and post-level of PRL. In the losers, there were significant differences in pre and post levels of T, FAI and PRL and in previous levels of SHBG and C. Two main differences between groups can be signalled: Only in winners, C levels were significantly higher post-COMBAT than post-ERGOMETRY, whilst only in losers, previous levels of PRL and pre and post levels of FAI were higher in COMBAT than in ERGOMETRY. In the winners group, there were no significant differences between changes experienced in both sessions, whilst losers experienced significantly higher percentage changes of T (t=-4.08; p=.001) and SHBG (t= -5.87; p=.00) in ERGOMETRY than in COMBAT (Table V).
Comparing hormonal response in COMBAT and CONTROL also reveals some differences. Only in winners, previous T levels and post FAI levels were significantly higher in COMBAT than in CONTROL. In losers, post-levels of PRL were higher in COMBAT than in CONTROL. For winners, percentage changes of T, FAI and PRL were greater in COMBAT than in CONTROL, whilst changes experienced by losers in both sessions were not significantly different (Table VI).
Table V: t-tests between hormonal levels in CONTROL and COMBAT
Table VI: t-tests between hormonal levels in COMBAT and ERGOMETRY
In winners there are negative correlation of pre (r=-.70, p<.05) and post levels of T (r=-.55, p<.05), previous FAI (r=-.61; p<.05) and percentage changes of C (r=-.53, p<.05) with item 2. Percentage changes of SHBG negatively correlate with item 3 (r=-.58, p<.05) an item 4 (r=-.59, p<.05) and PRL percentage changes are negatively correlated with item 4 (r=-.55, p<.05). Previous levels of C positively correlate with items 3 and 4 (r=.49, p<.05, in both cases) whilst PRL post levels (r=.54, p<.05) and percentage changes (r=.72, p<.001) correlate with item 1.
In losers, there are positive correlation of C post-level with item 4 (r=.54, p<.05) whilst C changes are positively correlated with item 3 (r=.61, p<.05)
In the Total Sample, the post level of T significantly correlates with item 5 of the Outcome Evaluation Test (r=.36; p<.05).
In winners there is to signal the negative correlation of T post levels with items 2 (r=-.74, p<.005) and 3 (r=-.73; p<.005), which respectively refer to usual performance and satisfaction with performance. Percentage changes of C negatively correlate with item 2 (r=-.51, p<.05) and item 3 (r=-.50;
p<.05). Percentage changes of T (r=-.74; p<.001) and FAI (r=-.71; p<.001) negative correlate with item 7. C post levels correlate with item 4 (r=-.70;
In losers, the significant correlation values are also negative: Percentage changes of T (r=-.55, p<.05), FAI (r=-.58, p<.05) and C (r=-.56, p<.05) with item 1. FAI post level (r=-.64; p<.05) also correlates with item 1 whilst T (r=-.58; p<.05) and FAI (r=-.66; p<.05) changes correlates with item 7.
Since our design included two other sessions, we can compare the hormonal responses of winners and losers in the three sessions. The winners' hormonal response was very similar in COMBAT and ERGOM, whilst losers experienced significantly greater androgenic changes in the non-competitive effort. Compared with the non-competitive non-physical stress experienced in CONTROL, winners showed a more pronounced hormonal response to the competition than losers. Androgenic changes of winners were significantly greater in COMBAT, whilst losers experienced similar androgenic changes in both sessions. Considering the hormonal levels of losers, the main difference between competitive and non-competitive stress is that stress hormone levels are higher in the competition.
Since the winners' hormonal response to the competition is similar to the response observed in the non-competitive exercise, the differences observed in losers could be attributed to a negative effect of the experience of being defeated: there are the losers who showed an androgenic response less intense than could be expected attending to the physical effort performed. In previous studies (Salvador et al., 1987, 1990) it has been suggested that the sports category could affect the hormonal response to the competition. If, as has been proposed by Mazur (1985), the outcome of aggressive interactions could produce different hormonal responses which, in their turn, are able to affect the future involvement of individuals in new aggressive encounters, it can be thought that the sports category or status can be a more important difference than merely winning or losing a single combat. In the terms used by Mazur (1983), if the T response is more related to dominant than to strictly aggressive behaviour, hormonal levels would be dependent on the outcome of a fight, only if it contributes to the statement of new dominance-submission relationships. It can be said that, in sports competitions, the hormonal changes would be modulated by the consequences on the sports status or category, which is not always an objective measure since it would depend on the cognitive appraisal of the subject about his-her own situation after the competition. In our competitive event, the subjects did not gain titles or improve their sports category. In spite of that, when we divide the total sample after the sports category of each judoist, only the high category subjects experience significant androgenic increases in the competition. This is a very similar response to the one observed in the winners' group. Conversely, the low category judoists showed an androgenic response more similar to the one observed in losers. There are no differences between sport categories in the other two sessions. It seems to indicate that this variable could just affect the changes produced in competitive situations. It can be interpreted considering the possibility that the high category subjects perceive the combat as a more threatening situation, since defeat could be more aversive for them than for the low category subjects. Nevertheless, comparing winners and losers from both categories, it can be seen that the high category winners did not show different androgenic responses than winners or losers of low category. Only when considering stress hormones there are some significant differences. The low category losers showed higher post-combat PRL levels, and the high category losers showed higher pre-combat PRL levels than the low category winners. So, we should not state that the sports category, at least in this study, can significantly affect his hormonal response to the competition. It is possible that the differences in sports category among the subjects of our sample were not high enough as to elicit different hormonal responses. The judoists had been matched as to ensure that the fights would be equilibrated, which can reduce the differences in sports category.
To study the influence of some cognitive variables on the hormonal response, the scores obtained in the Expectations Items have been correlated with hormonal levels and percentage changes. There was a relationship between androgens and interest in winning. Pre-combat levels were negatively correlated with this item in the total sample and in the winners' group, whilst percentage changes were positively correlated in all groups. For the stress hormones, there is a positive correlation between C levels and self-attributed possibilities and ability to win. The pattern of correlation was very similar in the winners' group. Among losers, the interest to win was positively correlated with androgenic changes. Although in the losers' group there were only four significant correlation values (whilst there were 12 in the total sample and 13 in the winners' group), it can not be stated that this fact constituted an important difference. Analysing the relationships between the Outcome Valuation Items and the hormones studied, there are clearly different correlation patterns in winners and losers. In winners, there are more significant values (17) than in losers (only 6). It must be noted that satisfaction with the outcome negatively correlates with T levels and C changes. Also, the subjects who considered that they could have performed better showed lower C levels. Conversely, in the losers' group, the subjects who considered that their expectations had not been fulfilled, experienced lower T, C and FAI changes. It seems to suggest that the expectations about their own performance, and the cognitive valuation of the outcome are able to influence the hormonal response to the competition, as has been proposed by Mazur et al. (1980, 1992). For C, the winners who viewed themselves as having more possibilities and ability to win, also had higher previous levels and, probably as a consequence of that, experienced less changes. This is not true for losers. The losers who considered to have more possibilities to win, had higher levels and experienced greater C changes. It could be interpreted as if the frustration derived from the defeat was able to elicit a higher anxiety in this subjects than in the judoists who did not view themselves as being able to win. For PRL, the winners who considered the combat as equilibrated showed higher post-combat levels, whilst the judoists who self-attributed less ability to win, experienced less changes. So it is possible to conclude that the relationships between the psychological state variables and the hormones are modulated not only by the outcome but also by the subjects' expectations and valuations, as was suggested by Mazur and Lamb (1980), McCaul et al. (1992) and Mazur et al. (1992). These psychological variables seem to be more important than the ‘objective’ measures of satus, at least, in situations in which sports category differences are not very large.