Psychosocial factors as predictors of ballet injuries: interactive effects of life stress and social support.

Patterson, Elizabeth L. ; Smith, Ronald E. ; Everett, John J. 等

Like athletes in competitive sports, ballet dancers are an "athletic" population that performs in a highly demanding environment and that is at high risk for physical injury. Because the physical and psychological skills required for success in this profession rival and in some cases exceed those of other athletic populations, ballet dancers have attracted recent empirical attention by sports medicine researchers (Bergfeld, 1982; Garrick & Requa, 1993; Hamilton, Hamilton, Meltzer, Marshall, & Molnar, 1989). One topic of interest has been injuries sustained in ballet. A recent epidemiological study of ballet injuries by Garrick and Requa (1993) revealed that over a 3-year period, 104 dancers in a professional company sustained a total of 309 injuries of sufficient severity to result in medical costs of nearly $400,000. Of particular interest was the fact that 23% of the dancers accounted for 52% of ali injuries, suggesting potential injury vulnerability factors. If vulnerability factors do exist, many of them are undoubtedly physical and biomechanical in nature, but some might also be psychosocial. Garrick and Requa highlighted the need for additional research to identify factors that might place certain subgroups of dancers at increased risk of injury. This study focuses on several psychosocial factors that appear to serve as injury vulnerability factors in this population.

Over the past two decades, life stress has been studied in relation to a host of social, psychological, and medical outcomes, including injuries (Hamilton et al., 1989, Passer & Seese, 1983; Petrie, 1992; Schroeder & Costa, 1984; Smith, Smoll, & Ptacek, 1990; Zautra, Guarnacchia, Reich, & Dohrenwend, 1988). Research has also focused on other psychosocial factors that might serve as stress-vulnerability and stress-resiliency factors and thereby exacerbate or attenuate the impact of life stress on physical and psychological well-being. Identification of such variables not only can help to identify psychological processes that may mediate event-outcome relations, but also can provide a basis for identifying at-risk subgroups toward whom intervention programs might be targeted.

A moderator variable is one that influences the nature, the direction, or the strength of a relation between a predictor variable, such as life stress, and a criterion variable, such as physical well-being (Baron & Kenny, 1986). In the study of life stress and its impact on well-being, social support (the subjective belief that there exist sources of support, help, and caring within one's social network) has emerged as a significant moderator variable in a number of studies. It appears that high social support is capable of buffering the impact of stressful life events, whereas low social support may exacerbate their impact (Cohen & Syme, 1985; Isaacson & Janzon, 1986; Petrie, 1992; Sarason, Sarason, & Pierce, 1990; Sarason, Sarason, Potter, & Antoni, 1985). Several recent prospective studies of athletic populations suggest that social support can influence the degree to which life stress is related to athletic injuries (Petrie, 1992; Smith et al., 1990). In these studies, self-report measures of recent life stressors were unrelated to subsequent injuries until social support levels were also taken into account. Petrie (1992) found that for starting football players low in social support, life stress was positively related to number of severe injuries, injury time loss, and number of games missed. In a study of male and female high school athletes in a variety of sports, Smith, Smoll, and Ptacek (1990) found that social support operated in combination with psychological coping skills and that in athletes low in both social support and coping skills, differences in negative life events accounted for nearly 30% of the injury time loss variance.

The stressful nature of the ballet environment has been well documented (Bergfeld, 1982; Hamilton et al., 1989). Ballet dancers would therefore appear to be an appropriate population in which to assess life event-injury relations, as well as factors that serve as protective or vulnerability factors, such as social support. Social support might be of particular importance in this highly competitive and demanding profession, where the greatest rewards are often peer acceptance and accolades from dance masters and critics (Hamilton eta!, 1989). Likewise, the competitive nature of the dance company may increase the importance of support received from one's social network. Finally, from a practical perspective, an understanding of psychosocial factors that influence ballet injuries could help guide attempts to reduce injuries through environmental interventions.

Given the previously documented importance of social support as a buffer against the impact of life stress, it was predicted that life stress (defined as life events appraised as negative in nature) would be predictive of ballet injuries and that social support would moderate the nature of this relation. Specifically, it was predicted that high levels of social support would negate the stress-injury relation, but that a significant stress-injury relation would be found for dancers who reported low social support in their lives.

Method

Participants

The participants were 31 female and 15 male dancers affiliated with a major ballet company in the western United States. The dancers, who participated in the study on a voluntary basis, had a mean age of 26.23 years (SD = 4.14), and 39 of the 46 dancers were Caucasian. All of the dancers who began the study were involved with the ballet company throughout the 8 month duration of the study, and all of them were participating fully in company activities at the time the study began (i.e., were uninjured at the time). Currently injured dancers were excluded from the sample.

Procedure

A prospective research design was employed. The dancers completed measures of recent life events and perceived social support in 30-minute group sessions administered by the first author 11 weeks after the beginning of the season. By this time, they had had sufficient opportunity to experience work-related life events and to become involved in the social network that existed within the ballet company. The dancers were told that the study involved an exploration of the experiences and characteristics of ballet dancers and of the frequency with which they sustained injuries. The dancers were assured of the confidentiality of data and were told that no information on individuals would be reported to the dance company unless a written request was received from a dancer (none were). As soon as the injury data were collected, the questionnaires they had completed were destroyed and the dancers' names were replaced by code numbers.

Following the collection of the predictor measures, daily injury data were collected for the next 8 months of the ballet season. Such information are collected in accordance with Workmen's Compensation procedures already employed by the ballet company, but special procedures were instituted for the purposes of this study.

Measures

Life events. Positive and negative life events were assessed using an adult version of the Perceived Events Scale (Compas, Davis, Forsythe, & Wagner, 1987). The questionnaire contained 207 events covering a wide range of life domains. For each item, the participants indicated whether the event had occurred in the past 6 months. If so, they indicated whether the event was a positive or negative one, and they rated on a 3-point scale whether the emotional impact of the event was slight, moderate or strong at the time it occurred. Finally, they indicated whether they considered it to be a major life event that had long-term consequences for them or a minor ("day to day") event that did not. By summing the emotional impact scores within each category of subject-classified events, we derived separate total impact scores for positive minor and major events and for negative minor and major events, as well as total positive impact and total negative impact scores. Such impact scores are considered to be the most sensitive indices of how life events have affected the individual psychologically (Compas et al., 1987; Zautra et al., 1988), and they served as the life event predictor variables.

Social support. The social support measure was adapted from one used in previous injury research (Smith et al., 1990) to measure the perceived amount of caring and emotional support available from one's social network. The participants rated how helpful each of 20 individuals (e.g., parents, best friend, spouse, dance instructor) and groups (e.g., dance colleagues, church groups, stage crew) were in terms of providing them with caring and emotional support when needed. Each social network element was rated on a Likert scale ranging from 1, not at all helpful to 5, very helpful. The social support index was the sum of the individual item scores. Smith et al. (1990) reported a one-week test-retest reliability coefficient of .87 for this scale.

Injuries. On each working day, the company's dance mistresses identified on an injury report form all dancers who were unable to participate fully in rehearsals or performances because of physical injury. The mistresses were aware that injuries were being studied in relation to psychosocial characteristics, but they were kept unaware of the nature of the psychosocial variables being studied in order to avoid potential bias of any kind. Complete data were collected for every dancer during the course of the study.

Following Garrick and Requa (1978), we defined an injury as any medical problem resulting from dance participation that restricted subsequent participation for at least one day beyond the day the injury occurred. We did not distinguish between injuries and reinjuries, but all recorded injuries followed at least one day (and typically, numerous days) of full physical participation in dance company activities. The completed injury report forms were collected on a weekly basis throughout the 8-month duration of the study. As in Garrick and Requa's (1993) epidemiological study of ballet injuries, the total number of separate injuries (i.e., injury days following days on which full participation occurred) over the course of the season served as the injury measure.

Results

Overall, 28 (60.9%) of the 46 dancers sustained at least one injury during the 8-month period of the study. The mean number of injuries was 1.07 (SD = 1.24). This injury rate is quite similar to that reported by Garrick and Requa (1993), whose dancers sustained a mean of .99 injuries per season over 3 seasons. Time loss due to injury ranged from 1 to 87 days in our sample, with a median time loss of 10.5 days.

On the life events measure, the dancers reported a greater number of positive minor (M = 37.70) and positive major (M = 13.46) life events than negative minor (M = 19.20) and negative major (M = 8.71) life events over the previous 6 months. Repeated measures contrasts revealed that the valence differences were significant for both minor events, t(45) = 9.19, p [less than] .001, and major events, t(45) = 3.35, p [less than] .002.

For all analyses, an alpha level of .05 was designated for statistical significance. Preliminary analyses indicated that dancers' age was not significantly related to any of the predictor variables or to injury frequency, so that this variable was not considered in subsequent analyses. Pearson product moment correlations between the life event and injury measures are presented in Table 1. The positive event scores were unrelated to injuries. In contrast, the total negative events score was significantly related to number of subsequent injuries. Breaking this total score down into its minor and major event components (which correlated +.39 with one another) revealed that the minor negative events score was significantly related to injury, whereas the major events score, though also positively correlated with injury, did not achieve significance.

The social support score was unrelated to number of subsequent injuries, and it was unrelated to either the positive or negative life event measures. This statistical independence from both the predictor (life events) and criterion (injury) variables satisfied a major statistical requirement for a moderator variable (Baron & Kenny, 1986).

To assess the role of social support as a potential moderator of the life stress-injury relation, a series of hierarchical regression analyses was carried out. In each moderated regression analysis, the life event score was entered first, followed by the social support variable. At the third step, the product score created by multiplying the life event and social support measures was entered. This analysis permitted an assessment of the unique variance accounted for by the product score (the "interaction" between the life event and social support variables) after controlling for the variance attributable to the predictor and moderator variables alone (Baron & Kenny, 1986).

Table I

Correlations Between Positive and Negative Life Event Scores, Social
Support, and Number of Subsequent Injuries.

Minor Positive Events -.03
Major Positive Events .17
Total Positive Events -.08
Minor Negative Events .36(*)
Major Negative Events .27
Total Negative Events .34(*)
Social Support -.07

Note. N = 46; *p [less than] .05

Separate hierarchical analyses were conducted for each of the positive and negative life event measures. These analyses indicated no significant interactive effects of any of the positive event scores with social support in the prediction of injuries. That is, social support exerted no influence on the null relations between positive events and injuries. Likewise, as indicated by the correlational analyses reported above, major negative events did not account for a significant amount of injury variance, and no significant moderator effect was found for social support in relation to major events. On the other hand, social support exhibited significant moderator effects for both minor negative events and total negative events. The results of these regression analyses are presented in Table 2.

Minor negative events accounted for significant injury variance in the first step of the hierarchical analysis, with 13% of the injury variance accounted for by differences in life event scores. The addition of social support scores at the second step of the regression analysis resulted in no significant increment in injury variance. On the other hand, the negative minor events X social support interaction term, entered on the final step, significantly increased the amount of injury variance accounted for to 21%.

Table 2

Hierarchical Regression Analyses of Injuries in Relation to Negative
Life Event Measures, Social Support, and the Life Events-Social
Support Interaction

Variable R [R.sup.2] Beta (F)change

Minor Negative Events .36 .13 1.21 6.35(**)
Social Support .36 .13 .33 .01
Minor Negative Events X
Social Support .45 .21 -.89 4.16(*)

Total Negative Events .34 .11 .58 5.64(**)
Social Support .34 .11 .52 .02
Total Negative Events X
Social Support .47 .22 -1.32 5.63(**)

Note. N = 46; * p [less than] .05, **p [less than] .025

A similar pattern of results was found for total negative events alone, which predicted injury scores at a statistically significant level. Social support, entered next, accounted for no additional variance by itself. Again, however, the events X social support interaction term resulted in a significant increment in injury variance accounted for, doubling the injury variance accounted for by the psychosocial factors from 11% to 22%. The negative beta coefficients for the product scores indicated that in each instance, the combination of high negative event scores and low social support was related to a higher number of injuries.

To assess the extent to which differences in levels of social support influenced the magnitude of the relations between the negative life event measures and injuries, the distribution of social support scores was divided at the median into two groups designated as high and low in social support, each numbering 23 participants. Within these groups, correlations were computed between the minor and total negative event scores (for which moderator effects had been found in the regression analyses) and number of subsequent injuries. The results of these analyses, presented in Table 3, revealed that within the high social support group, no significant relations were found between the negative life events measures and injuries. In contrast, high positive correlations were found in the low social support group between the minor and total negative life events scores and injuries. Within the low social support group, a squaring of the stress-injury correlation coefficients indicated that nearly half of the total subsequent injury variance could be accounted for in terms of differences in the amount of minor or total life stress reported by the dancers.

Table 3

Correlations Between Negative Event Scores and Injuries for Dancers
Reporting High and low Levels of Social Support

 Social Support

Events Low High

Minor Negative .69(***) -.24
Total Negative .71(***) -.02

Note. n = 23 for each social support group ***p [less than] .001

Despite the absence of a significant life events X social support moderator effect in the regression analyses, the major negative event score also correlated significantly with injuries in the low social support group (r = .54, p [less than] .01). In the high social support group, major negative events did not correlate significantly with subsequent injuries (r = . 16, ns). Thus, for dancers low in social support, major stressors are also predictive of injuries and, like the minor negative events, they contribute meaningfully to the correlation between the negative events total score and subsequent injuries.

Discussion

Ballet dancers constitute a human movement population that has begun to attract the attention of injury researchers (Bergfeld, 1982; Garrick & Requa, 1993; Hamilton et al., 1989). Previous epidemiological research has revealed that a relatively small proportion of the dancer population accounts for a disproportionate number of injuries (Garrick & Requa, 1993), suggesting the possible role of physical and/or psychosocial vulnerability factors. Other research has documented the physical and psychological demands placed on dancers in a sport that requires near perfection in performance and fierce competition for professional positions (Hamilton et al., 1989). Given these findings, the role of psychosocial factors in injury vulnerability, which has previously gone virtually unexplored in this population, appears warranted. Our results indicate that in addition to physical vulnerability factors previously identified (Bergfeld, 1982), life stress and social support may be important factors in dancers' physical well-being. In particular, our prospective study suggests that a combination of high life stress and low social support may place dancers at increased risk for subsequent injury.

As in previous prospective studies of life events and injuries in athletic populations (Passer & Seese, 1983; Smith et al., 1990), positive events were not predictive of subsequent injuries. In contrast, significant predictive relations were found for total negative events and for minor negative events, but the positive correlation found for major negative events did not achieve statistical significance. This result is consistent with previous research findings that microstressors or "daily hassles" can sometimes exhibit stronger relations with measures of well-being than do major events, which tend to occur less frequently (Zautra et al, 1988). That such a result should be found in this population is not entirely surprising, given the physically and psychologically demanding conditions under which dancers train on a dally basis (Hamilton et al., 1989). A stressful context involving highly competitive conditions requiring consistently exacting performances may actually magnify the impact of other microstressors, causing them to take a greater toll on well-being than might occur in a more benign environmental context.

In this sample of ballet dancers, correlations between negative events and injuries were substantially larger than those found in prospective studies involving athletes (Passer & Seese, 1983; Petrie, 1992; Smith et al., 1990). They were also larger than those found in a previous cross-sectional and retrospective study of professional ballet dancers in which measures of occupational stress did not distinguish between previously injured and noninjured dancers (Hamilton et al., 1989). Our stress measure was far broader in scope, including many life areas. In this study, the minor negative event measure by itself accounted for 13% of the injury variance, a figure exceeding that typically found in prospective studies involving objective medical outcome measures, where the amount of variance typically accounted for is in the 1-5% range (Schroeder & Costa, 1984; Smith et al., 1990). These results therefore suggest that stressful life events (particularly recurrent "microstressors") constitute a particularly significant injury vulnerability factor in this population.

However, it is clear that not all dancers are vulnerable to the effects of stress; individual differences in social support play a major role. The examination of the social support variable revealed that individual differences in dancers' perceptions of their social resources dramatically affected their vulnerability to the impact of life stress. High levels of perceived social support appeared to serve as a protective factor against the impact of life stress, resulting in nonsignificant relations between negative life events and subsequent injuries. In contrast, dancers who perceived their environment as low in social support appeared highly vulnerable to the impact of stress on physical well-being. Indeed, the magnitudes of the correlations between negative life events and subsequent injuries among low social support dancers, accounting for about half of the injury variance, far exceed those previously found in other physical activity (e.g., athletic) populations (Petrie, 1992; Smith et al., 1990). The results thus affirm the potential stress-buffering effects of social support in this population.

Why would social support be so important in this population? One reason may be the intensely demanding and competitive environment of the ballet world. Previous research has shown that individuals who perceive themselves as lacking in social resources who can provide them with caring and support when they are confronting life stressors fare significantly more poorly on a host of physical and psychological well-being measures (Cohen & Syme, 1985; Isaacson & Janzon, 1986; Sarason et al., 1990). In contrast, perceptions of available social support (even if it is not utilized by the individual) appears to buffer the impact of stress. Social support provides people with reassurance of personal worth, positive relations that help balance threats and negative experiences, feelings of increased resourcefulness, and a positive focus of attention that can prevent self-preoccupation with negative events (Cohen & Syme, 1985; Sarason et al., 1990). Thus, social support may help counteract two of the causal mechanisms that have been suggested as mediators of the stress-injury relation, namely, self-preoccupations that can interfere with attentional processes during performance, and chronic muscular tension produced by emotional arousal that can reduce flexibility and coordination, thereby increasing the risk of injury (Andersen & Williams, 1988; Smith et al., 1990).

It is of interest to note that social support did not relate to injury by itself; it entered into the prediction of injuries only for participants who reported high levels of stress. This supports an interactive, or stress-buffering, model of how social support functions, as opposed to a "main effects" model asserting that social support constitutes a protective factor against injury whether or not stress is present (Baron & Kenny, 1986; Smith et al., 1990).

The results of this study have implications for measures that might be taken to influence psychosocial factors related to ballet injuries. The finding that life stress is related to injuries only in dancers who experience low levels of social support in their lives suggests the potential value of promoting a supportive and caring environment that can help to buffer the impact of life stressors. Certainly, the ballet environment is a highly demanding one that contains endemic stress, but measures can be taken to reduce unnecessary stressors that might impair performance as well as physical well-being. For example, the promotion of positively-toned approaches to dance instruction that emphasize encouragement and positive reinforcement rather than a critical and punitive approach could reduce one potential source of interpersonal stress. Likewise, attempts to build group cohesion within dance companies and encouraging dancers to initiate and foster supportive relationships within their social and family networks could enhance the protective benefits of social support. The present results, which suggest that both life stress and social support are potentially important psychosocial contributors to physical well being, suggest the possibility that such measures might serve to reduce the medical, artistic, and personal costs associated with ballet injuries.

This research was supported by Grant 86-1066-86 from the William T. Grant Foundation to Ronald E. Smith. We express our appreciation to the dancers and instructional staff of the Pacific Northwest Ballet, Seattle, WA, for their participation and cooperation.

References

Andersen, M. B., & Williams, J. M. (1988). A model of stress and athletic injury. Journal of Sport and Exercise Psychology, 10, 294-306.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182.

Bergfeld J. A. (1982). Medical problems in ballet. A round table. The Physician and Sportsmedicine, 10, 98-114.

Cohen, S., & Syme, S. L. (1985). Social support and health. New York: Academic Press.

Compas, B. E., Davis, G. E., Forsythe, C. J., & Wagner, B. (1987). Assessment of major and stressful events during adolescence: The Adolescent Perceived Events Scale. Journal of Consulting and Clinical Psychology, 55, 534-541.

Garrick, J. G., and Requa, R. K. (1978) Injuries in high school sports. Pediatrics, 61, 465-473.

Garrick, J. G., & Requa, R. K. (1993). Ballet injuries: An analysis of epidemiology and financial outcome. American Journal of Sports Medicine, 21, 586-590.

Hamilton, L. H., Hamilton, W. G., Meltzer, J. D., Marshall, P., & Molnar, M. A. (1989). Personality, stress, and injuries in professional ballet dancers. American Journal of Sports Medicine, 17, 263-267.

Isaacson, S. O., and Janzon, L. (1986). Social support: Health and disease. London: Coronet Books.

Passer, M. W., & Seese, M.D. (1983). Life stress and athletic injury: Examination of positive versus negative events and three moderator variables. Journal of Human Stress, 9, 11-16.

Petrie, T. A. (1992). Psychosocial antecedents of athletic injury: The effects of life stress and social support on female collegiate gymnasts. Behavioral Medicine, 18, 127-138.

Sarason, I. G., Sarason, B. R., & Pierce, G. R. (Eds) (1990). Social support: An inter-actional view. New York: Wiley.

Sarason, I. G., Sarason, B. R., Potter, E. H., and Antoni, M. H. (1985). Life events, social support, and illness. Psychosomatic Medicine, 47, 156-163.

Schroeder, D. H., & Costa, P. H. (1984). Influence of life event stress on physical illness: Substantive effects or methodological flaws? Journal of Personality and Social Psychology, 46, 853-863.

Smith, R. E., Smoll, F. L., & Ptacek, J. T. (1990). Conjunctive moderator variables in vulnerability and resiliency research: Life stress, social support and coping skills, and adolescent sport injuries. Journal of Personality and Social Psychology, 58, 360-370.

Zautra, A. J., Guarnacchia, C. A., Reich, J. W., & Dohrenwend, B. P. (1988). The contribution of small events to stress and distress. In L. H. Cohen (Ed.), Life events and psychological functioning: Theoretical and methodological issues (pp. 96-122). Newbury Park, CA: Sage.