Overuse injury and affects in competitive sport: A prospective longitudinal study

Overuse injuries, which have a high prevalence in sport, are suggested to result in different affective responses in comparison to traumatic injuries. Affects may also reciprocally act as risk factors for overuse injury. The aim of this study was to examine the associations between overuse injury and affects within a longitudinal follow‐up design. Competitive athletes (N = 149) of various sports and levels of competition completed the Positive and Negative Affect Schedule (PANAS) and the Oslo Sports Trauma Research Centre Overuse injury questionnaire (OSTRC‐O) once a week over 10 consecutive weeks. Bivariate unconditional latent curve model analyses with structured residuals were performed to evaluate the associations within and across weeks between OSTRC‐O severity score and affects. Results indicated that OSTRC‐O severity score and positive affects (PA) had a statistically significant negative within‐week relation (r = −24.51, 95% CI = [−33.9, −15.1], p < 0.001). Higher scores of overuse injury were significantly related to lower levels of PA across weeks (ß = −0.02, 95% CI = [−0.04, −0.001], p = 0.044), while the reciprocal effect of PA on overuse injury was not significant (ß = −0.13, 95% CI = [−0.52, 0.26], p = 0.51). No statistically significant association was observed between OSTRC‐O severity score and negative affects, neither within nor across weeks. Our findings suggest that overuse injury may have adverse psychological consequences on the long run through lessened PA and address the need for providing sustainable psychological support focusing upon such PA when working with athletes experiencing overuse injury.


| INTRODUCTION
The epidemiology, risk factors and prevention of overuse injuries are increasingly examined, and more-developed methods of data reporting are being implemented (e.g., Oslo Sports Trauma Research Centre Overuse injury questionnaire; [OSTRC-O]); findings show a high prevalence across different sports. 1 Overuse injuries (e.g., stress fracture or Achilles tendinopathy) are a type of sportrelated injuries that result from a mechanism of repetitive micro-trauma accumulating over time. 2 They are often chronic with periods of remission and exacerbation. 3Such temporality can explain the presence of particular psychological responses in athletes sustaining overuse injury that may differ from those following sudden-onset traumatic injury. 4,5Psychological responses to injury are suggested to involve cognitive, affective, and behavioral processes that are inter-related, which dynamically influence short-and long-term outcomes through biopsychosocial pathways. 5,6eciprocally, psychological factors predisposing to overuse injuries are also well documented. 7This suggests the existence of a bi-directional influence, which is supported by complex approaches of sport injuries that refer to the principle of recursivity. 8ffects, which are thought to act as a pivotal mechanism in this complex process, 6 refer to the mental states involving evaluative feelings. 9Unlike emotional episodes, affective feelings are not necessarily directed at specific objects per se, 10 but are consciously accessible as the simplest raw-nonreflective-feelings. 11Affects are primarily characterized by the combination of two dimensions, 11 pleasure-displeasure (pleasure or valence) and activation-deactivation (arousal or energy).Based on these two dimensions, conceptualisations of affects have been proposed (for a review, see 12 ).One of these approaches considers that activated affects account for most of the everyday life experiences and represent the subjective components of two broader biobehavioural systems. 13Hence, focusing on positive and negative activated affects can provide a "highly parsimonious explanation of certain affectrelated phenomena", 13 p.828, including sport injury. 14hese two largely independent categories of affects can be labeled as positive affects (PA), that is the "extent to which individuals feel enthusiastic, active, and alert", 15 p. 1063, and negative affects (NA), that is "a general dimension of subjective distress" subsuming a variety of aversive states such as anger and anxiety, 15 p. 1063.
In line with this conceptual stance, research examining the links between overuse injury and affects indicate that such relationship might be, indeed, recursive.Firstly, NA such as frustration, fear or anger were reported as psychological responses to overuse injury. 16In relation to the progressive onset and chronicity of most overuse injuries, athletes may particularly experience prolonged episodes of emotional distress, whereas sudden-onset traumatic injuries may elicit more prominent, but less durable NA. 5,16 Other research dovetails this suggestion, as shown in athletes with chronic-overuse injury who appeared to be suffering from a type of prolonged stress-response syndrome. 17However, athletes' affective responses that are specific to overuse injury must be further clarified. 16econdly, affects are suggested to act, in turn, as predisposing factors for overuse injury. 18More specifically, Messier et al. 19 reported that NA such as being jittery, irritable and nervous predicted overuse injury in recreational runners.
Thus, a dynamic interplay may develop between overuse injury and affects, creating a downward spiral that would contribute to explain the persistence of overuse symptoms.However, this hypothesis of a recursive relation has yet to be empirically tested; to date, no studies have attempted to capture the bidirectional relations between overuse injury and affects as time unfolds.
In addition, affects are likely to change rapidly over time, 20 especially if resulting from the perception of fluctuating overuse symptoms.Consequently, their associations require to be investigated by using intensive repeated measures at the within-person level and through a longitudinal research design, thus allowing "to examine how people change over time, not just in relation to other people but also in relation to themselves" (, 21 p. 235).Importantly, this type of analysis examining how two different variables fluctuate rapidly in relation to themselves and to each other over time was described as a first step in a causal analysis of change, assuming that the variables are ordered in time. 21,22More precisely, the use of withinperson analyses where both autoregressive (i.e., withinconstruct) and cross-lagged (i.e., between-constructs) effects are specified in the same model has been recently deemed as necessary; 22 it should be considered when studying the bidirectional relation affects-overuse injury.
Based upon the abovementioned conceptual and methodological considerations, the aim of the present study was to examine whether: (1) overuse injury severity at the within-person level was related to PA within the same week; (2) changes in overuse injury severity predicted subsequent changes in PA, and vice-versa (i.e., acrossweek relations); (3) overuse injury severity at the withinperson level was related to NA within the same week; and (4) changes in overuse injury severity predicted subsequent changes in NA, and vice-versa.More precisely, we hypothesized that there would be: (H1) a negative association between weekly overuse injury severity and PA; (H2) a negative bi-directional relation between overuse injury severity and PA across weeks; (H3) a positive association between weekly overuse injury severity and NA; and (H4) a positive bi-directional relation between overuse injury severity and NA across weeks.

| Study design
This study used a prospective longitudinal design with self-reported weekly measurements of overuse injury and affects over 10 consecutive weeks.The study gained ethical approval (Dnr 2017/1052).No financial compensation was offered for participation.

| Participants
For recruitment purposes, coaches of 37 different clubs were approached by the research team.Following coaches' agreement, email addresses of athletes meeting the inclusion criteria were collected (i.e., eligible athletes; see below).Across the 37 different clubs approached, a total of 493 athletes were identified as eligible to participate.More precisely, of the original 168 athletes who started the weekly follow-up, eight decided to withdraw from the study and 11 were excluded due to insufficient data.These non-respondents, 13 men and six females with a mean age of 28.2 years (SD = 13.3), were competing in athletics (n = 7), in volleyball (n = 5), rugby (n = 3), handball (n = 2), basketball (n = 1), and football (n = 1).
Participation inclusion criteria were: (1) being an active competitive athlete, that is participating in official competitions in either of the countries where data was collected (France and Sweden), and (2) being at least 16 years old.Ultimately, a total of 149 athletes (89%) from 16 different clubs were included in the analysis.Participants were 44 females and 105 males competitive athletes aged between 16 and 60 (M = 27.9,SD = 12.8), and participating in eight different sports: athletics (n = 74), handball (n = 27), rugby (n = 19), football (n = 18), volleyball (n = 9), basketball (n = 4), triathlon (n = 1) and weightlifting (n = 1); some athletes were involved in different sports.At the time of the study, they were actively participating in official competitions that ranged from regional level to international level; competitions were official and held by either of the national federations in France or Sweden.

| Overall procedure and recording methods
A short video presenting the study aim and protocol was sent to each eligible athlete by email.Athletes were redirected to a first questionnaire including informed consent to participate and demographics.Athletes who agreed to participate were then enrolled into a longitudinal follow-up of 10 consecutive weeks.Data were selfreported through a weekly questionnaire.For most of the participants (72%, based in France), we used Briteback®, which is a research tool developed for intensive longitudinal designs available both online and as a mobile application (allowing notifications to be sent).Every Sunday at 4 p.m., participants were contacted to complete three questionnaires (see below) on affects, overuse injury, and sports exposures experienced during the past week.The overall survey necessitated approximately 3 min to be completed.Non-respondents were automatically sent a reminder one, and if needed again 2 days after the first request.If participants failed to answer the weekly questionnaire for two consecutive weeks, they were contacted individually by the first author and encouraged to carry on with participation.For the remaining participants (28%, located in Sweden), data were self-reported manually before the first training session of the following week, and collected by the first author at their respective training sites.

| Measure of affects
Affects were measured through the French and Swedish validated versions of the Positive and Negative Affect Schedule (PANAS). 15,23,24This self-reported questionnaire consists of a 20-item checklist with a 10-item subscale aiming to measure PA (e.g., active, determined, inspired), and a 10-item subscale aiming to measure NA (e.g., afraid, guilty, jittery).Participants were asked to indicate to what extent they felt a certain way during the past week by using a five-point Likert scale ranging from 1 (very slightly or not at all) to 5 (extremely).The scores for the PA and NA scales were obtained by adding the individual scores for the 10 PA, and for the 10 NA, respectively.A high score on the PA scale reflects an optimal state of energy, concentration, and pleasurable engagement whereas a low score indicates a state of sadness and lethargy.A high score on the NA scale denotes a state of distress and unpleasant engagement whereas a low score suggests a state of calmness and serenity. 15It should be reminded that PANAS subscales "do not measure all pleasant and unpleasant affective states, but rather measure only highactivation states" 10 .The French and Swedish versions of the PANAS have been previously tested and exhibited adequate psychometric properties. 23,24In the present sample, internal consistency was found to be adequate both for the positive affect subscale and the negative affect subscale (Cronbach's α week to week range between 0.84 and 0.87).

| Measure of overuse injury
The first version of the OSTRC-O 25 was used to record overuse injuries.For this study, the instructions, which were originally made to target specific body regions (e.g., knee problems), were reformulated so that information were asked irrespective of the body area.A numerical value ranging from 0 to 25 was allocated to the different answer propositions of the four questions that required information on how potential "physical problems" affected training participation, training volume, performance, and elicited pain, respectively.This allowed to calculate a severity score ranging from 0 to 100 for each week, and represented an ordinal-scale variable.We used two translated versions of the OSTRC-O: the Swedish version, 26 and the French version (Martin et al.,  2021). 27he medical staff of each club was responsible for collecting data regarding traumatic injuries.Indeed, as suggested by Clarsen et al., 1 the OSTRC-O was used to gather information on all types of 'physical problems' (i.e., irrespective of the type of injury).Hence, to differentiate overuse from traumatic injuries, medical staff referred to the definition and criteria of traumatic injuries, described by Fuller et al., 28 which were clarified by the first author prior to the initiation of the study.
Participants were asked to record any new traumatic injury that required medical attention occurring over the data collection period and to report corresponding time-loss when the athlete was unable to train. 28At the end of data collection, injury data from self-reported and medical staff sources were compared.Any injury self-reported by the athlete that matched the data provided by the medical staff was considered as a traumatic injury and removed from the final data set (i.e., not considered as an overuse injury).If data were insufficient to determine the nature of the injury mechanism, athletes were asked by phone at the end of the collection period to make the necessary amendments (i.e., to correctly label the injury as either overuse or traumatic).

| Measure of exposures
.3.5 | Procedure used to increase adherence of participants Following guidelines to improve athletes' response rate and adherence to the study protocol, 3,29 a feedback procedure was implemented with regular reports about overuse injury data sent to the coach or athletic trainers of each club.This procedure was approved by athletes beforehand, as part of the initial consent.To minimize response bias due to social desirability, individual data regarding overuse injuries were pooled and presented as average values of each team/group.

| Analyses
In line with the suggestion of Clarsen et al, 25 we decided to exclude the data of the first week, which seemed to present an artificially high rate of overuse injuries.Also, following these previous recommendations, participants who failed to answer the questionnaire over four consecutive weeks at some point of the study, were excluded from the final data set.The prevalence of athletes reporting overuse injuries and the average severity score (with a 95% CI) were calculated for each week.
We used the Mplus software (version 8.0) to perform all statistical analyses, applying the robust maximum likelihood estimator (MLR).A p-value below 0.05 was used as cut-off for statistical significance in all analyses.As the PANAS consists of two highly independent subscales 10 , two separate bivariate unconditional latent curve models with structured residuals LCM-SR 30 were estimated, one for PA, and one for NA.The LCM-SR has recently been described as a favorable method allowing to estimate more rigorously both between-person and within-person processes of the over-time relations between two constructs simultaneously. 30In other words, this method allows to model the complex (i.e., bidirectional) dynamic relations that link overuse injury and affects over time, and how these relations vary dynamically both within and across individuals in a way that is consistent with our hypotheses.More particularly, as compared to traditional multilevel modeling approaches (e.g., multivariate LCM), the LCM-SR is better suited to specify cross-lagged relationships between the two variables at the within-person level while controlling for the autoregressive effects within each construct, and allow a pure disaggregation of betweenperson and within-person effects (see, 30 for more information on LCMs-SR).
To obtain our two bivariate LCMs-SR (i.e., one for PA and one for NA), we combined the univariate LCM-SR for OSTRC-O severity score with the univariate LCMs-SR for the two PANAS subscales separately.The following goodness-of-fit indices were used to assess the model: root mean square error of approximation (RMSEA) with 90% confidence interval (CI), the comparative fit index (CFI)/Tucker-Lewis index (TLI), and the standardized root mean squared residual (SRMR).The following cut-off criteria were applied to indicate acceptable fit (Marsh, 2007): CFI/TLI >0.90, SRMR and RMSEA <0.08.

| Survey response rate
The average individual weekly response rate to the PANAS, OSTRC-O and exposure questionnaire was 81.6% (SD = 7.3), ranging between 71.8% and 91.9%.In total, 91% of the athletes (n = 135) reported at least one overuse injury, with an average weekly OSTRC-O severity score over the 9 weeks of 23/100 (95% CI = [22, 25]).Detailed data regarding the prevalence of overuse injuries in this cohort were published elsewhere.

| (H2) Across-week relations
A statistically significant autoregressive effect was observed for OSTRC-O severity score (ß = 0.43, 95% CI = [0.33,0.54], p < 0.001) and for PA (ß = 0.26, 95% CI = [0.14, 0.39], p < 0.001).Namely, the OSTRC-O severity score 1 week was positively related to the OSTRC-O severity score the following week.Similarly, the level of PA 1 week was positively related to the level of PA the following week.At the within-person level, a statistically significant cross-lagged effect of OSTRC-O severity score on PA was observed (ß = −0.02,95% CI = [−0.04,−0.001], p = 0.044).Namely, individuals having reported higher OSTRC-O severity scores relative to their underlying level of OSTRC-O severity score at one point in time were more likely to experience lower levels of PA relative to their underlying level of PA at a subsequent point in time.No statistically significant cross-lagged effect of PA on OSTRC-O severity score was observed (ß = −0.13,95% CI = [−0.52,0.26], p = 0.51).

| DISCUSSION
The present study examined the associations between overuse injury and affects over time.Findings showed support for the first (H1) hypothesis; that is, higher OSTRC-O severity scores were related to lower levels of PA within the same week.Furthermore, findings also showed partial support for the second (H2) hypothesis; that is, an increase in OSTRC-O severity score was related to a subsequent decrease in the level of PA, but no reciprocal effect of the level of PA on OSTRC-O severity score was reported across weeks.Lastly, neither the third (H3) nor the fourth (H4) hypotheses were supported by findings; OSTRC-O severity score and NA were not related, neither within nor across weeks.
Our results of lessened PA in the presence of a worsening overuse injury suggest that overuse injuries may have adverse psychological consequences on the short-term, but also on the long-term run.Our findings corroborate previous results indicating that athletes with chronicoveruse injury were suffering from a form of general and prolonged stress-response 16,17 ; long-term psychological response to overuse injury might be characterized by a decrease in PA more than by the presence of NA.This finding might be of clinical relevance because even minor reductions in the level of PA may cumulatively have a significant effect on mental health, including a higher vulnerability to depressive symptoms. 31Therefore, it seems critical to pay a careful attention to the level of PA over the course of an overuse injury, and to implement strategies allowing to maintain a significant level of PA, to prevent more severe psychological consequences.
One potential explanation for this observation is that different types of injuries may elicit different emotional responses. 4,5More specifically, the mode of onset and chronicity of injury was suggested to influence short-and long-term affective responses. 4It is therefore possible that athletes experiencing progressive onset overuse injuries might be more susceptible to long-term psychological disorders such as depressive mood, frustration, and sadness, 4 whereas the depressive symptoms following traumatic injuries were found to decrease rapidly over time. 32This suggestion is supported by previous research in health psychology, which highlighted that daily hassles (the chronic strains of everyday life) are more predictive of health problems than isolated major stressful life events. 33Indeed, it can be more difficult to cope with daily episodes of stress (e.g., when experiencing a longterm overuse injury) in comparison to coping with a more prominent, but less durable stressful episode (e.g., experiencing a traumatic injury).Overuse injury, thus, deserve attention in relation to the prevention of mental health issues in athletes. 34indings from the present suty, however, showed no week-to-week influence of PA on overuse injury.The fact that this negative relation was observed within the same week but not across weeks suggests that the temporality of these interactions may be important.More precisely, results suggest that the decrease in PA experienced as a prolonged response to overuse injury becomes, with time, too sparse to act as a subsequent risk factor for overuse injury.Notwithstanding, this result might also be due to the strong autoregressive effect obtained for overuse injury, which is a limiting factor for observing a statistically significant cross-lagged effect. 30Similarly, the autoregressive effects for both PA and NA were statistically significant.Because affects, to some extent, are related to personality traits, 13 it is not surprising that a proportion of affects remain somewhat stable over time.This may also explain why no statistically significant cross-lagged effect of overuse injury severity on NA, and vice-versa, were observed.
Moreover, no statistically significant within-week relation between NA and overuse injury was reported.This finding is contrary to previous research findings highlighting that NA may increase the risk of injury through increased autonomic nervous system activation. 18However, according to a recent systematic review of the psychosocial risk factors for overuse injuries in competitive athletes, 7 no other study has investigated the potential contribution of NA in the risk of sustaining overuse injury specifically.Messier et al. 19 found that athletes reporting more NA were more likely to sustain overuse running injuries; however, these athletes were recreational runners and affects were not measured repeatedly with an adequate frequency -as we did in the present study.In a meta-analysis of the psychological factors of traumatic injuries, Ivarsson et al. 35 discussed the potential mediating role of affect and cognition that might explain how stress can generate poor decision-making and reaction-time, which in turn can create a higher vulnerability for injury. 36A potential explanation of a contradictory finding in our study is that this mechanism is only relevant to explain the occurrence of sudden-onset injuries, whereas gradual-onset overuse injuries were suggested to be less dependent on this type of cognitive processing and more likely related to physiological processes affecting training adaptation and recovery. 6Moreover, other findings suggest that overuse injury may elicit avoidance, denial and escape behaviors predominantly. 17This type of cognitive-behavioral reactions allows athletes to reframe the meaning of the problem, for instance by using 'magical thinking' (i.e.meaning that the problem will resolve by itself 7 ;).Consequently, because the meaning that is attached to an experience determine, to a large extent, the threat posed by such experience, 37 athletes using these coping strategies may not report NA in the presence of overuse injury.

| Limitations and future research directions
A first limitation of this study is related to the temporality of measurements of the variables examined.Because affects can fluctuate rapidly over time, 20 their weekly measurement may have overlooked some of their variation.In addition, the relatively short data collection period (10 weeks) may be a limitation as it is certainly not sufficient to fully capture the suggested long-term affective response associated to overuse injury.Future studies are therefore encouraged to implement more frequent measurement points over a longer extended period (e.g., twice a week across several months).A second limitation relates to the fact that information regarding overuse injuries and affects were recorded at the same time every week.More specifically, affects may have been over reported because of the necessary focus on overuse symptoms, and vice-versa, which may not be fully indicative of what has really been experienced over the week.A third limitation is related to the use of PANAS, which does not allow to measure the full spectrum of affective states, as affects are excluded from its structure. 10Other methods that allow both activated and deactivated affects to be captured, such as the list of 20 items proposed by Hyde et al., 38 deserve attention.In line with the Multilevel Model of Sport Injury, 39 which encourage to embrace a more interpersonal approach of sport injury, future studies are finally suggested to investigate the influence of overuse injuries sustained by athletes on the affects experienced by their coaches.Also, it would be of interest to use a similar design in more specific populations (e.g., a given sport in a given country) on the importance of sociocultural factors in sport injury (e.g., the normalization of pain), which were not considered in this study.

| Perspective
Athletes experiencing overuse injury certainly need to be both supported over extended periods of time and screened for potential mental health disorders.Coaches are at the frontline of this process and should be trained to identify early affective signals of overuse injury such as reduced PA, and associated behavioral manifestations.This might be done by spotting athletes' body language or by monitoring the level of PA, for instance weekly.In line with Wadey et al., 39 we also emphasize that enhanced coach-athlete communication (e.g., allowing time for conversation on the sideline, in training sessions) and social support at all stages of the overuse injury are important to consider.Policy makers (especially those setting coaching education programs) should also ensure that coaches know how to help athletes understand what happens to them, and that overuse injuries can have serious adverse physical and psychological consequences in the long run.Similarly, coaches and psychology consultants should strive to promote a multidimensional self in athletes, and to offer sustainable narratives that allow to reduce the affective burden of injuries. 40