Injuries and illnesses in Swedish Paralympic athletes—A 52-week prospective study of incidence and risk factors

INTRODUCTION
Sports-related injuries and illnesses in Paralympic sport (SRIIPS) is a concern, but knowledge about the aetiology and risk factors is limited. The aim of this study was to describe the annual incidence, type and severity of injuries and illnesses among Swedish Paralympic athletes and to assess risk factors.


METHODS
Swedish Paralympic athletes (n=107) self-reported SRIIPS every week during 52 weeks using an eHealth application. Incidence proportions (IP) and incidence rates (IR) were used as measures of disease burden. Time-to-event methods (Kaplan Meier and Cox regression) were used to identify risk factors.


RESULTS
The annual IP for injury was 68% and for illness 77%. The injury IR was 6.9/1000 hours and the illness IR 9.3/1000 hours. The median time to injury was 19 weeks (95% CI: 10.5-27.4) and to illness 9 weeks (95% CI: 1.4-16.6). Most injuries occurred during training and 34% were classified as severe (≥21 days of time loss). An increased injury risk was observed among athletes in team sports (HR 1.88; 95% CI: 1.19-2.99), athletes with a previous severe injury (HR 2.37; 95% CI: 1.47-3.83) and male athletes (HR 1.76; 95% CI: 1.06-2.93). The most common illness type was infection (84%). Athletes in team sports (HR 1.64; 95% CI: 1.05-2.54) and males with a previous illness (HR=2.13; 95% CI: 1.04-4.36) had a higher illness risk.


CONCLUSION
Paralympic athletes report a high incidence of injuries and illnesses over time. This emphasises the need to develop preventive strategies of SRIIPS and optimise medical services for this heterogeneous athlete population.

athletes. 10However, these studies have been performed during short and intense competitions periods.Studies examining SRIIPS over many months including athletes' training periods are lacking.Thus, there is a need for prospective studies that assess the incidence of SRIIPS over a longer time.In addition, risk factors and mechanisms of SRIIPS specific to Paralympic athletes need to be investigated, as their impairments may influence the risk. 11o increase our knowledge of the health status and risks in this understudied population, we initiated a prospective longitudinal study of SRIIPS. 12In the study protocol, we adapted definitions of SRIIPS to accommodate Paralympic athletes' pre-existing medical conditions. 12To enable weekly data collection of SRIIPS in a heterogeneous and geographically spread population, we developed and evaluated an eHealth-based self-report application adapted to Paralympic athletes with physical impairment (PI), visual impairment (VI), and intellectual impairment (II). 13Selfreports have been shown to be sensitive in monitoring changes in athletes health, and it has been recommended to monitor such changes on a regular basis. 14Subsequently, 107 Swedish Paralympic athletes prospectively self-reported SRIIPS every week during 52 weeks.
The aim of this study was to describe the annual incidence, type, and severity of injuries and illnesses among Swedish Paralympic athletes and to assess risk factors.

| Study design and definition
This was a 52-week prospective cohort study assessing self-reported incidence of SRIIPS, which is part of the epidemiological study "The sports-related injury and illness in Paralympic Sport Study" (SRIIPSS). 12,13,15The SRIIPSS was developed and pursued in collaboration with athletes, coaches, and staff in the Swedish Paralympic Committee.
The study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines, is approved by the Regional Ethical Review Board in Lund, Sweden (Dnr 2016/169), follows the WMA Declaration of Helsinki Ethical Principles for Medical Research Involving Human Subjects, and is registered at ClinicalTrials.gov[NCT02788500].Participation in the study was voluntary, and informed consent was collected from all participants.The definition of SRIIPS was: "Any new musculoskeletal pain, feeling, injury, illness, or psychological complaint that caused changes in normal training or competition to the mode, duration, intensity, or frequency, regardless of whether or not time was lost from training or competition". 12

| Participants
All athletes with PI, VI, and II from the Swedish Paralympic program (N = 150 athletes) were invited to participate.The athletes had been participating at a previous Paralympic Games or were candidates for a future Paralympic Games.The following inclusion criteria were used: (a) being able to communicate in Swedish; (b) age 18-65 years; and (c) having the ability to respond to the eHealth application.In total, 107 (71%) athletes accepted to participate (Figure 1).

| Data collection
Data on the incidence of new SRIIPS and sports exposure were collected weekly throughout 2017 using Briteback ® survey tool and an eHealth-based self-report application adapted to Paralympic athletes.Prior to this study, the data collection procedure was evaluated in a four-week pilot feasibility usability study. 12,13Once every week, the athletes received a web survey through email and/or text message with questions regarding their previous training week.If reporting a new injury, the athletes were asked about body location, injury type, and mechanism, involvement of their impairment and diagnosis.For a new illness, questions regarding symptoms, affected body system, contribution of the impairment, and diagnosis were asked.Exposure was reported as the number of training minutes.Data were followed up every week by (KF).Closing reports regarding diagnosis, contact with medical personnel, time loss from sport, and preventive possibilities were sent to those reporting being back in training. 12

| Data categorization
Injuries were categorized in a matrix for classification of musculoskeletal diagnoses according to body location, injury type, and diagnosis from the 10th International Statistical Classification of Disease and Related Health Problems (ICD-10). 12,16Two authors (KF and JJ) independently formed ICD-10 codes.The reported illnesses were categorized into affected body system and ICD-10 diagnoses independently formed by KF and JL. 7,12,17The severity of SRIIPS was determined by time loss (days) from regular sports participation: slight (0-3 days), minor (4-7 days), moderate (8-20 days), severe (≥21 days), and long-term (≥3 months). 18A total training load rank index (TLRI) was calculated for each athlete by multiplying the rate of perceived exertion (RPE) with minutes of training per week throughout the year. 12,19LRI was categorized into low, middle, and high according to percentiles.

| Statistical analysis
Descriptive statistics were used to present data.To estimate the total onset of new SRIIPS, the incidence rate (IR) was calculated by dividing all reported incidents by the total time of exposure in each category. 20The Mann-Whitney U test or the Kruskal-Wallis test 21 were used to compare IR between different variables, such as sex, age, type of sport (team vs individual, summer vs winter), impairment (physical vs intellectual vs visual; central neurological vs intellectual vs les autres vs limb deficiency vs spinal cord injury (SCI) vs visual; and wheelchair users vs ambulatory participants), TLRI (low vs middle vs high), and previous severe injury or illness last year.Chi-square statistics were used to compare any differences between the subgroups in the proportion of incidents by affected body location and body system, respectively.
To determine the probability for an athlete to sustain a new SRIIPS, the incidence proportion (IP) was calculated by dividing the number of athletes that sustained a SRIIPS by the total number of athletes followed. 20Survival analyses using the Kaplan-Meier method were conducted to estimate the cumulative survival probability (SP) and the primary endpoints: median time to the first injury and illness, respectively. 22Log-rank tests assessing the hazard function were used to compare differences in survival times between the subgroups, and Cox proportional hazard regression with corresponding hazard ratios (HR) were performed to analyze the actual risk of sustaining a first SRIIPS. 18Univariate models assessing risks associated with each variable were first tested.To account for covariates and differences in risk between different subgroups, multivariate models with two explanatory variables and their interactions were examined.Significance levels of P < .05 and 95% confidence intervals were used.Throughout, data were analyzed using IBM SPSS version 25.

| RESULTS
The weekly response rate was 72%.The median number of completed weekly reports per athlete was 45 (IQR 25-52, min-max 1-52).The mean time of weekly sports exposure was 6.8 ± 4.8 hours.Four athletes dropped out during the year and were right censored in the survival analysis from the week of drop out.

| Injury incidence rates
In total, 179 injuries were reported, resulting in an overall IR of 6.9 injuries/1000 hours of sports exposure.No significant differences in IR were present between the subgroups.Of all injuries, 41% were primary, 37% new subsequent, and 21% recurrent.The median number of reported injuries per athlete was 2 (IQR 1-3, min-max 0-7) (Table 1).
Fifteen percent of all injuries occurred during competition, 53% during sport-specific training, 17% during other training, and 16% outside sport.The onset of injury was as follows: 32% traumatic, 16% overuse with sudden onset, and 52% overuse with gradual onset.ICD-10 diagnoses related to inflammation,  pain, and soft tissue disorders were most common (47%), followed by sprain, strain, and rupture (15%) (Appendix 1).The most frequently injured body location was the shoulder (23%), followed by the lumbar spine (12%) and the elbow/forearm (11%) (Table 2).The time loss from sport due to injury (severity) was as follows: 0-3 days (33%), 4-7 days (24%), 8-20 days (10%), ≥21 days (23%), and ≥ 3 months (11%).Wheelchair users and athletes with SCI reported more injuries in the upper extremities and the shoulder (P<.001).Ambulatory individuals reported more injuries in the lower extremities (P = .012),with VI athletes reporting more injuries to the lower leg/calf (P = .018).Athletes with VI reported more multiple injuries (P = .004)(Table 1) and also more traumatic injuries (P = .008).For 59% of the injuries, the athletes reported that the impairment was a contributing factor in the injury mechanism.For 68% of all injuries, the athletes sought medical care and the diagnosis was confirmed by a medical professional.The athletes reported that 32% of the injuries could have been prevented.

| Injury incidence proportions and risk factors
In total, 73 (68%) athletes reported a new injury.The median time to first injury was 19 weeks (95% CI 10.5-27.4).Log-rank tests showed statistically significant variations in SP with regard to gender (P = .024),type of sport (P = .005),and previous severe injury (P ≤ 0.001).Men had lower SP (26%) compared to women (43%), and athletes in team sports had a lower SP (19%) compared to individual sports (42%).Twelve percent of the athletes with a previous severe injury remained injury free compared to 39% without a history of severe injury (Table 1; Figure 2).
Results from the Cox regression analyses using univariate models showed that athletes with a previous severe injury had more than twice the risk (HR = 2.37; 95% CI 1.47-3.83) of sustaining a new injury.Also, male athletes (HR = 1.76; 95% CI 1.06-2.93)and athletes participating in team sports (HR = 1.88; 95% CI 1.19-2.99)had a significantly higher risk (Table 3).No multivariate models were statistically significant.

| Illnesses incidence rates
In total, 241 illnesses were reported, resulting in an IR of 9.3 illnesses/1000 h of sports exposure.Athletes with a middle TLRI reported a significantly (P = .019)higher IR (22 illnesses/1000 hours).No other significant differences were found between the subgroups.The median number of reported illnesses per athlete was 2 (IQR 1-4, min-max 0-11) (Table 4).

| Sports-related injuries
4][25] The only comparable study of injuries over time in Paralympic athletes reported an IR of 3.9/1000 hours among wheelchair fencers. 26In the present study, also a high IP was reported (68%).
The studies from the Paralympic Games have reported considerably lower IPs ranging from 11.6% to 23.8%. 2,5,6,8,27,28oteworthy is that 85% of the injuries in the present study occurred outside competition and that one-third were classified as severe.This is a concern as most athletes do not have on-site medical support outside competition.Also, the proportion of overuse injuries was higher compared to the Paralympic Games, 5,27,28 suggesting that overuse injuries and inaccurate training are more common throughout the training season.Risk factors for injuries are dynamic and depend on intrinsic and extrinsic factors as well as the inciting event. 29us, it is recommended to continue conducting athlete health surveillance in different contexts to better understand the etiology of injuries.
In the present study, athletes in team sports had a higher risk for injury.It is possible that athletes in team sports, such as goalball, para ice hockey, wheelchair rugby, and basketball, are more prone to injuries because of high intensities and collisions.Athletes in team sports had high IRs also during the Paralympic Games, 5,8,27,28 and theses sports should be targets for future prevention.Moreover, male athletes had a higher risk of injury, even when adjusting for team sports.Risktaking behavior has been identified as a possible explanation in other areas of injury research. 30More research is needed to establish gender-specific risk factors.Finally, athletes with a previous severe injury had a higher risk of injury, emphasizing the importance of primary and secondary prevention. 31n agreement with previous studies, the shoulder was the most injured body region, 5,27,28 and wheelchair users in particular reported higher proportions of shoulder injuries.These athletes are commonly exposed to overuse of the shoulder both during sport and in daily life. 32Preserving the shoulder in wheelchair athletes may be particularly difficult because of lack of rest, altered seating positions, and configuration of sports equipment. 32,33Given these findings, it is recommended that shoulder injury prevention programs are implemented and that the impact from sports equipment is further addressed.
Notably, athletes with VI reported more multiple injuries and more traumatic injuries.Among persons with VI, the risk of unintentional injuries is generally higher. 34Further studies are needed to assess injury mechanisms among VI athletes.

| Sports-related illnesses
The illness IR was higher than the injury IR, emphasizing the need to include illnesses in athlete health surveillance.A majority of the reported illnesses were infections.The basic etiology of infections is transmission of fungal, viral, and bacterial agents.Consequently, the primary prevention strategies are proper hygiene and social distancing. 35Especially, athletes in team sports had a higher risk of illness.Because of close encounters in teams, athletes more easily transmit infections. 36To prevent illnesses in Paralympic team sports, it is recommended to reduce skin contact, not return to sports until complete recovery, and to adopt regular cleaning of equipment, such as wheelchairs, balls, and arena floors. 35gain, male athletes (with a previous illness) had a higher risk, and it is recommended to further assess their increased susceptibility to illnesses.
Noteworthy, athletes with a middle TLRI had a higher IR.In the present study, athletes reported that a common reason for illness was overtraining and stress.In a recent study, we showed that 83% of the Paralympic athletes continued to train unwell and 77% felt guilt when missing training. 3It could be Since middle differed significantly from both low and high, middle was examined using low and high as a combined reference category.b Injury with time loss ≥ 21 days previous year analyzed for first injury and illness with time loss ≥ 21 days previous year analyzed for first illness.c Variables in the simple models were also pairwise combined to test for interactions, that is, models where pairs of variables were tested in combination with their interaction.Finally, the effect of each covariate was tested and presented as above.Only models with significant interactions (P < .05)are presented in this section of the table.Reference categories are shown in italics.
strategies focusing on organizations' policy enforcement, coaches' education, medical staffs' recognition, and athletes' intrapersonal skills. 39

| PERSPECTIVES
The results emphasize the need to develop preventive strategies adapted to Paralympic athletes and to optimize medical services throughout the entire season.As a variety of acute injuries, overuse injuries, and illnesses were reported, there is a need to develop and implement preventive strategies on a primary, secondary, and tertiary levels.
Because of the complex variation of injuries and illnesses, future preventive strategies require both individualized and sport-specific strategies as well as educational interventions involving athletes, coaches, medical staff, and sport organizations.The results from this study can inform athletes, coaches, clinicians, and sports organizations about the epidemiology of sports-related injuries and illnesses in Paralympic athletes.

2
Distribution of injuries and illnesses sustained by Paralympic athletes (n = 107) during one year by body region, body system and impairment /face (ear, eyes, jaw) 13

F 9 FAGHER
I G U R E 2 Kaplan-Meier curves for time to first injury (a-c) and illness (d) during the study period displayed by categories with a significant difference revealed by log-rank tests (P<.05) | Et Al.
Annual injury incidence proportions, time to first new injury, and injury incidence rates among Swedish Paralympic athletes (n = 107) by gender, age, impairment, sport, training load, T A B L E 1

time to injury (weeks) (95% CI)
a Visual, intellectual, and physical impairment compared.b Visual, intellectual, central neurological, les autres, limb deficiency, and spinal cord injury impairment compared.c Kruskal-Wallis test was used instead of Mann-Whitney U test.d ≥ 3 weeks time loss.
Simple and multiple models of risk factors for injury and illness determined by time-to-event analyses (Cox proportional hazard regression models presented with hazard ratios (HR), p-values and 95% confidence intervals (CI)).Simple models show risks associated with variables separately.Multiple models with two explanatory variables and their interactions are reported as categorical variables with the possible subgroups as separate conditions T A B L E 3