Associations between competitive anxiety, athlete characteristics and sport context: evidence from a systematic review and meta-analysis

Background: There is a vast literature investigating the possible associations between competitive anxiety, athlete variables and sports context. As far as we are concerned, there is no study which has compiled such findings to produce more robust evidence on this topic. Objectives: The aim of the study was to conduct an exploratory systematic review of the literature followed by a meta-analysis in order to investigate possible associations between competitive anxiety, social-demographic characteristics, profile of the athlete and sports context. Methods: Systematic searches of PubMed, PsycInfo, Web of Science, Lilacs and SciELO electronic databases were performed to identify studies published between January 2006 and January 2018, including a manual search in the references of the selected studies. Results: A total of 59 studies were included for qualitative synthesis and 27 for meta-analysis. More robust associations were observed between competitive anxiety and female gender, lower age, and less experience time. Discussion: Knowing the variables which exert influence on competitive anxiety can be relevant to plan specific treatment and intervention programs, enabling the athlete´s development beyond technical and physical preparation.


Introduction
Anxiety is an emotional response stimulated by the anticipation of a real or potential threat 1 .In sports, competition can be considered a source of threats as the athlete's image is usually associated with his or her performance, the final result is always uncertain, there is exposure to public opinion and judgement by third parties, among others 2,3 .
Because this condition is potentially anxiogenic and usually experienced by athletes, studies show that high levels of anxiety are conversely associated with sport performance [4][5][6] .In addition, performance can be indirectly affected by different symptoms of anxiety, such as physiological (e.g.energetic expenditure and cardiovascular alterations), motor (e.g.impairment in co-ordination), cognitive (e.g.decrease in attention, concentration and decisionmaking capacity) and relational (e.g. increase in conflicts among staff members) changes 7,8 .
Due to the peculiarities of this psychological phenomenon within the context of sports, the term "competitive anxiety" (CA) was coined to refer to the emotional reaction which regularly emerges before or during sports competitions 3 .This construct is shown to be partially correlated with general anxiety (r = 0.35 to 0.50), which justifies its assessment on a specific basis 9 .
CA was initially understood one-dimensionally, but this approach was surpassed with the introduction of the Multidimensional Theory of Anxiety, which proposes the existence of somatic (SCA) and cognitive (CCA) dimensions correlating to each other positively despite being different 3,10,11 .
SCA is characterised by symptoms such as muscle tension, tachycardia, facial reddening, tremors and sweating, whereas CCA involves preoccupations preoccupations with self-demand, poor performance, negative evaluations, social comparison, expectations and demands by technical commission, team, family and supporters.With regard to the impact on the athlete's performance, CCA tends to have a negative relationship and SCA, in turn, a U-inverted relationship 3 .
In addition to the different CA dimensions from a symptomatic perspective, one should consider its state-trait type.State-CA is experienced transiently, whereas trait-CA is considered a relatively stable trend in individuals who perceive different situations to be threatening 12 .
With regard to the factors influencing CA, a pioneer study by Martens et al. 3 found higher levels in female athletes as well as in solo athletes with little competition experience.Since then, researchers have been investigating the impact of these and other variables such as age, hormones and relationship with coach on the levels of CA because knowing these implications will contribute significantly to the design and guidance of prevention and intervention programs in the specific population.
Although extensive literature on this issue has been produced over the decades, as far as we know there is no study gathering findings to seek robust evidence.Therefore, we have aimed to conduct an exploratory systematic review followed by meta-analysis in order to investigate whether there are differences in the levels of CA between athletes in function of social-demographic characteristics, athlete's profile and sport context.The study of these variables was based on the fact that they were always present or already existing a priori in the athlete and in the context of competition.

Methods
Based on the Systematic Reviews and Meta-Analyses Guidelines 13 , we have conducted systematic searches in the literature by using the electronic databases PubMed, PsycInfo, SciELO, Web of Science and Lilacs for the key-words "competitive anxiety" and "sports performance anxiety", including a manual search from the references cited in the selected studies.
The inclusion criteria for the studies were the following: a) written in English, Spanish or Portuguese; b) human samples with no age-group restrictions; c) quantitative methodology; d) published in the past 12 years (i.e.January 2006 to January 2018); e) objective of investigating the levels of CA in function of social-demographic characteristics, athlete's profile and sport context.
Exclusion criteria as well as selection flowchart for the studies are presented in Figure 1.This stage was conducted by two research psychologists in the field of mental healthcare, with any discrepancy being resolved through consensus.
The qualitative methodology used by the included studies was assessed by means of the using the Strengthening the Reporting of Observational Studies in Epidemiology -STROBE 14 and Consolidated Standarts of Reporting Trials -CONSORT 15 , depending on their design.For each item of the checklist, the following scoring was applied: fully fulfilled (1 point), partially fulfilled (0.5 point) or not fulfilled (0 point).The index of methodological quality was obtained by adding the scores attributed to each item, with the total sum being divided by the total number of items applied to the study and then multiplied by 100.
The following information were extracted from the studies: sample size, age of the participants, sport modality, time of experience in the sport, recruitment source, study design, measurement instruments used, interest variables and outcomes.A global qualitative analysis of the outcomes of each study in function of the independent variables was conducted.
Then, meta-analyses were independently performed in function of the four most studied variables, namely: gender, age, experience and competitive level.We included the studies that described all the necessary gross data in the published material.The missing data were requested to the main authors of the studies.Of the 10 e-mail sent, only one returned with the information.The analyses were conducted   according to the following parameters: a) SCA and CCA dimension, b) state-trait CA types.Statistical analyses were performed using the Stats Direct software package.The heterogeneity was evaluated using Cochran's Q test.The studies were considered homogenous when the significance of the Cochran's Q test was > 0.05, in these cases, a fixed-effects method was used.In the case of heterogeneity among the studies (significance ≤ 0.05), a random-effects method was used 16 .Higgins and Thompson statistics were used for inconsistency (I 2 ), which was interpreted as follows: low (< 25%), moderate (26 to 75%) and high (> 75%) 17 .Egger's regression test was also used to assess the presence of publication bias.The test was rejected when the significance level was p ≥ 0.05 18 .

Results
A total of 3.457 studies have been identified.After application of exclusion criteria, 59 studies were included in the qualitative synthesis, and of these, 27 in the meta-analysis, as shown in Table 1.
With regard to the main characteristics of the studies, one can observe that sample size ranged from 9 to 1.038 athletes (totalizing = 8603 participants; mean = 145.8), the most used methodology was cross-sectional (n = 53; 89.8%), the main source of recruitment was local competition (n = 46; 78.0%), and the majority of the studies used multidimensional scales (n = 45; 76.3%) to evaluate the CA (for example, CSAI-2 and SAS-2).
With regard to methodological quality, the majority of the studies (n = 55; 93.2%) had indices equal or above 50% (mean = 58.9%).The most overlooked items in the studies were specification of measures aimed at preventing potential bias sources and calculation of sample size (Supplementary material available upon request to the author).
The independent variables assessed by the studies were distributed into three categories: a) socio-demographic characteristics (SG); b) athlete's profile (AT); and c) sport context (CO).Table 2 shows briefly the results of the qualitative synthesis regarding the influence of these variables on CA (Supplementary material available upon request to the author).
Qualitatively, higher levels of CA were observed in athletes who were female, younger and less competitively experienced as well as presenting previous poor experience, practicing individual sports, and competing in official events.These results were maintained even when studies using one-dimensional measurement or general anxiety assessment instruments were excluded.
The results of the meta-analysis are available in Table 3 (Supplementary material available upon request to the author).
Female athletes showed higher levels CA compared to males with small effect size in most of the analyses, however, this difference becomes more expressive in function of the trait-CA.There is evidence that younger athletes have higher levels of CA in the somatic dimension, and in the state condition, but the effect size effect is small.Although the difference was small, less experienced athletes have higher levels of CCA and state-CA compared to the more experienced.Finally, there were no differences between the athletes regarding the athlete's competitive level.

Discussion
The studies included in the present review have had, in general, moderate methodological quality.The weaknesses found are related to the lack of presentation of raw data, measures used to avoid potential bias sources and sample size calculation.Another negative point to be considered, despite the sound literature on the two-dimensionality of this psychological phenomenon 3,10,11 , is the use of one-dimensional measurements of CA (SCAT) and others for measuring the general/global anxiety (STAI and BAI), suggesting that there may be an influence of other anxiety symptoms/characteristics other than competitive.] ♀ >: 12 [ 24,27,34,36,43,51,53,56,62,65,68,74 ] | ♂ >: 2 [ 57,75 ] Age (Younger x Older) 18 7 [ 24,35,42,44,46,58,76 ] Younger >: 9 [ 26,28,29,40,41,45,47,53,62  From a qualitative point of view, it was observed the influence of some variables on the CA levels: female, younger and less competitively experienced as well as presenting previous poor experience, individual sports, and official events.
Meta-analyses have been conducted to seek more robust results by using SCA/CCA dimensions and type of trait-state anxiety as moderating variables.With regard to gender, it was found that female athletes had a higher level of anxiety compared to male counterparts.This fact was corroborated by the meta-analysis, showing significant differences in all dimensions, mainly in the trait-CA.In the neurological perspective of the psychopathology, these findings are supported by the literature showing a higher prevalence of anxiety symptoms and disorders in females due to the fact that the oscillating levels of gonadal hormones occur more often clinically in women than in men, which increases the susceptibility of the former to stress and anxiety [78][79][80][81] .
However, sports specialists argue that female athletes are more anxious than male ones due to the influence of other factors, such as honesty to speak about feelings/emotions 27 , susceptibility to external stimuli 53 , commitment to the sports practicing 62 , susceptibility to pressures from sports environment 82 and greater focus on the risk of failure rather than on achieving success 83 .
With regard to age, the levels of CA tend to be higher in groups of young athletes.Quantitative analysis indicated a small effect size between the groups in the SCA and state-CA.Previous studies associated these findings to the fact that young age is directly related to factors such as feelings of insecurity, emotional dependency and use of less elaborated strategies for coping with physiological responses 28,36,41,47,53 .
Similarly, qualitative analyses indicated that less experienced athletes were those with higher levels of CA, but in meta-analysis, differences with small effect size were found in the CCA and state-CA.Martens et al. 3 believe that experienced athletes have more capacity to control their concerns, whereas other authors support the hypothesis that experienced athletes tend to perceive the competition from a more positive perspective 30 as they are more exposed to competitions and reflect positively on self-effectiveness, which in turn increases self-confidence during the contests 53 and favours the development of more effective coping strategies to deal with criticism from oneself and others 26,84 .
Qualitative analysis has shown that the levels of CA tend to be higher in athletes with low competitive level, but these results were not confirmed in the quantitative analysis as no significant differences were found between the groups assessed.This leads to the questioning of the hypothesis that less competitive athletes tend to use less coping strategies and consequently they are more likely to experience anxiety symptoms 26,30 .
The findings of the qualitative analysis indicate higher levels of CA in athletes who practice individual sports compared to those who practice team sports.Studies reporting that athletes who practice individual sports are more anxious during competitions because of the responsibility for achieving results which depend exclusively on one person only, whereas those practicing team sports would share this responsibility with other members of the team 19 .
With regard to competition, qualitative analysis has shown that the levels of CA tend to increase before and during official competitions and in contests played away from home.The proximity of competition increases the athlete's level of anxiety because of a threatening competitive environment, since important values and goals are at stake 79 .In addition, away contests would be influenced by travel, poor familiarity with venue, away supporters, referee bias and absence of family and social support 20,70 .
Other evidence found has to do with the athlete's previous performance.Qualitative analysis has shown the 75% of the studies reported that poor performance tends to increase the levels of CA, a fact which may be attributed to the low level of self-confidence among athletes following a defeat.Unfortunately, this finding cannot be quantitatively analysed as the selected studies using this variable did not list all raw data needed.
With regard to team status, quantitative analysis has shown no difference in the levels of CA between starting and reserve athletes.However, there are a few studies focusing on this issue, which limits robust conclusions to be drawn.It is believed that the absence of difference may be associated with the fact that starting and reserve athletes are usually submitted to the same routine, level of demand and sports results, which would have a similar impact on their experience of CA.
Despite the reduced number of studies and heterogeneity of the groups studied, the qualitative results indicate influence of nationality/ethnics on the levels of CA.The role of sports and the preference for a specific modality can reveal important characteristics of a country's culture, norms and customs.Experience and manifestation of anxiety may be influenced by ethnical and cultural factors due to beliefs on mental diseases, including social context and norms to which an individual is exposed 85 .
With regard to team position, despite the specificities of each sports modality, the qualitative analysis has shown no differences in the levels of CA in football, futsal and basketball athletes.As for country of birth, education level, income level and opponent's status, one can highlight the presence of exploratory data with heterogeneous and inconclusive results, thus requiring further investigations on their impact on CA.
Despite the efforts made to cover all the scientific studies on the theme available in the past twelve years, we also have to emphasise that there are limitations in the present work: 1) low number of studies included in the quantitative analyses, despite the effort to contact the authors; 2) exclusion of studies written in Korean due to lack of domain of the language, warning that possible specificities of this population should be explored.
For a better understanding of the construct CA, it is suggested that new systematic reviews should be performed in order to seek evidence of associations with other psychological (i.e.personality traits, coping skills, motivation, self-esteem, self-confidence), physiological (i.e.injuries, hormonal changes) and relational (i.e.relationship with coach, family, teammates) variables.Knowing these aspects and those addressed by our study will contribute to the development of a more integrative view of athletes and better management of their symptoms of CA.
In view of the results obtained, one can conclude that CA is a phenomenon occurring in the sports context which deserves attention as it is closely related to the athlete's quality of life and performance, being more significantly associated with female, younger and less experienced athletes.

Figure 1 .
Figure 1.Inclusion and exclusion processes of studies based on PRISMA flowchart.

Table 1 .
Studies included in the systematic review and meta-analysis * = studies included in meta-analysis;

Table 2 .
Qualitative synthesis of the results = Non exclusive category; ** = It was not possible to draw up categories for comparisons due to data's heterogeneity; # = number of the study as mentioned in Table1; ♀ = Female; ♂ = Male; _ = studies that used one-dimensional or general scales to assess CA. *

Table 3 .
Gross results of the meta-analysis