In the USA, >300 000 sport-related traumatic brain injuries (TBIs) occur each year and many of the injured are children.1^–5 Between 2001 and 2005, children aged 5–18 years accounted for nearly 135 000 sport-related TBIs that required emergency department (ED) visits, and about 18% of sport-related hospitalisations in this age group were attributed to TBIs.1 A concussion is a type of TBI, defined as a “trauma-induced alteration in mental status that may or may not involve a loss of consciousness”.6 Although death resulting from concussion is uncommon, there is potential for subsequent long-term morbidity.7^–9 This is especially true throughout the years of childhood and adolescence, because the brain is still in a period of active development. Injury to a child’s brain can have many long-term health outcomes, including chronic headaches, emotional liability, memory/recall deficits, and impaired executive function.10^–12 More children are beginning to participate in organised sports at earlier ages, and the total number of child participants is growing annually.13 14 With these trends, paediatric sport-related concussions are becoming an important public health and medical concern.

Much of sport-related concussion research, however, has focused on elite adult athletes, by examining biomechanical, pathophysiological, and neurobehavioral factors to better identify and manage sport-related concussions.15^–20 Little has been reported regarding characteristics of hospitalisations for sport-related concussions. Significantly less is known about the incidence, diagnosis and management of concussions among the paediatric population, especially concussions that occur during sports. These issues are important for healthcare providers because athletic trainers or other medical professionals are usually not present at childhood sports activities. In addition, children may be less able to describe their symptoms accurately and may need special care after concussion.7 21^–24 Understanding the characteristics of paediatric sport-related concussions and their hospitalisation outcomes is crucial for providing optimum care to these patients and developing intervention strategies to prevent future concussions.

In this study, we documented patient and hospital characteristics associated with hospitalisations, including length of stay (LOS) and hospital charges, for non-fatal sport-related TBIs and concussions. We estimated the effect of patient and hospital characteristics on the odds of hospitalisations for sport-related concussions versus other sport-related TBIs.

METHODS

Study design and setting

We conducted a retrospective analysis of 5 years (2000 to 2004) of sport-related TBI resulting in hospitalisation among patients aged 5–18 years, using the Nationwide Inpatient Sample (NIS) of the Healthcare Cost and Utilization Project (HCUP). NIS is the largest all-payer inpatient care database in the USA, provided by the Agency for Healthcare Research and Quality.25

The NIS was originally designed to approximate a 20% sample of American hospitals, defined as “all nonfederal, short-term, general and other specialty hospitals, excluding hospital units of institutions”. Each year, it has provided information on 7–8 million inpatient stays from approximately 1000 hospitals located in 35 states.25 Using a complex algorithm based on five hospital characteristics including ownership/control, bed size, teaching status, urban/rural location and region, NIS data can be weighted to represent the total number of hospital discharges in the USA.25

Subjects

All patients aged 5–18 years with a primary diagnosis of TBI (ICD-9-CM codes 800, 801, 803, 804, 850(.0–.5,.9), 851–854, 955.55, or 950(.1–.3))26 27 and who met one of three E codes for sport-related injury (E886.0, E917.0 or E917.5), were selected from the NIS between 2000 and 2004 and included in the analysis. The three E codes used to define sport-related injury resulting in hospitalisation were: E886.0 (tackles in sports that cause fall on same level from collision, pushing, or shoving, by or with other person); E917.0 (striking against or being struck accidentally by objects or persons in sports without subsequent fall); and E917.5 (striking against or being struck accidentally by objects or persons in sports with subsequent fall). Any cases with missing E codes could not be identified as sports-related. Therefore, 1568 hospitalisations for non-fatal sport-related TBIs were included in the analysis.

Main outcome measures

Concussion hospitalisation was defined on the basis of the primary diagnosis of seven ICD-9-CM codes (850.0–850.5 and 850.9). Patients with any of these seven codes as a primary diagnosis were classified as having concussion. Patients who did not have these codes were classified as having other sport-related TBI. Only the primary diagnosis was used to define concussion in this analysis. In total, 39 (2.5%) patients with a secondary diagnosis (diagnoses 2–15) of one of these seven codes were included as other sport-related TBIs.

Length of stay and total hospital charges were determined at hospital discharge. When computing the mean and median LOS and hospital charges per discharge, we excluded patients who were transferred to another hospital after being admitted to a different hospital (n = 3, 0.4%). We also adjusted the hospital charges to the year 2004 (last quarter) levels to reflect the inflation rate, using the Bureau of Labor Statistics’ consumer price index for inpatient hospital services.28

Patient or hospital characteristics

Patient variables included in the analysis were sex, age group, median household income based on patient’s ZIP code, external cause of injury, admission source, discharge status and insurance type. Hospital variables included in the analysis were hospital region, location, bed size and teaching status.

Statistical analysis

The number and proportion of hospitalisations for sport-related concussions, national estimates, and 95% CI were calculated by patient and hospital characteristics. The discharge-level weights provided by the HCUP were applied in the calculation for national estimates to account for the sample design.25

The top 10 principal diagnoses and the number of principal procedures performed were described for sport-related concussions resulting in hospitalisation. The mean and median LOS and total hospital charges (per discharge) were computed by sex and age group. Logistic regression was used to estimate the effect of each patient or hospital characteristic on the odds of hospitalisations for sport-related concussions versus other sport-related TBIs. Adjusted models were used to evaluate the association between the patient and hospital characteristics and the odds of hospitalisations for concussions. All the analyses were conducted using SAS callable SUDAAN V.9.0, accounting for cluster sampling and sample weights.

RESULTS

Patient and hospital characteristics

In total, 1568 hospitalisations for non-fatal paediatric sport-related TBIs were identified between 2000 and 2004 in the sample of NIS data. Of these, nearly half (n = 755; 48.2%) were sport-related concussions (table 1). When weighted, these represent an estimated total of 3712 hospitalisations for non-fatal sport-related concussions nationwide among children aged 5–18 years between 2000 and 2004, with 742 annual hospitalisations for sport-related concussions. Of all hospitalisations for sport-related concussions, about half occurred among youth 15–18 years of age (50.3%). Boys accounted for 83.7% of all paediatric sport-related concussions.

Similar to all other paediatric sport-related injury hospitalisations,30 the number of concussions admitted was higher in urban hospitals (87.1%), hospitals with high bed volume (62.0%) and teaching hospitals (58.1%). Most hospitalisations for sport-related concussions were admitted through the emergency room (83.8%) and discharged routinely (98.7%). The hospital charges for the majority of patients with concussion were covered by private insurance, including health maintenance organisations (79.3%). In total, 37 sampled hospitalisations for concussion (4.9%), an estimated 184 nationwide, were uninsured, and the patients paid the hospital charges themselves (table 1).

Principal diagnoses and principal procedures

Among all hospitalisations for sport-related concussions, the most common principal diagnosis was concussion with no loss of consciousness, accounting for 308 hospitalisations (40.8%), followed by concussion with brief loss of consciousness (<1 hour) (n = 294; 38.9%) (table 2). Of 755 sport-related concussions, more than half (52.3%) experienced loss of consciousness of variable duration.

In all, 143 principal procedures were performed among sport-related patients with concussion during their hospital stay. The most frequent procedure was CT scanning of the head (n = 85; 59.4%), followed by MRI of the brain and brain stem (n = 15; 10.5%). However, less than one-fifth (18.9%) of patients with concussion received any procedures during their hospital stay and only 12.3% of sport-related patients with concussion had imaging procedures performed(table 2).

Length of stay and hospital charges

Overall, the mean LOS for paediatric sport-related concussions was 1.1 day, with a median of 0.8 day (table 3). The mean hospital charges (per discharge) for sport-related concussions were US$8097, with a median of US$5274. Nationwide, estimated total hospital charges for sport-related concussions between 2000 and 2004 for 5–18-year-olds were >US$29 million, or nearly US$6 million per year (table 3). Nationwide, boys accounted for nearly US$5 million out of the US$6 million annual total hospital charges for sport-related concussions. Patients aged 15–18 years accounted for >US$3 million in total hospital charges per year, and patients aged 10–14 years accounted for >US$2 million in total hospital charges per year for sport-related concussions.

Factors associated with sport-related hospitalisations for concussion

We used logistic regression to estimate the effect of each patient or hospital characteristic on the odds of hospitalisations for sport-related concussions versus other sport-related TBIs (table 4). The adjusted OR showed no gender difference. However, compared with patients aged 5–10 years, patients aged 15–18 years were over twice as likely (OR = 2.28; 95% CI 1.64 to 3.18) to be admitted to the hospital for sport-related concussions; patients aged 10–14 years had similarly increased odds (OR = 1.91; 95% CI 1.34 to 2.71). Compared with teaching hospitals, hospitals with non-teaching status had greater odds of admitting sport-related patients with concussion (OR = 1.91; 95% CI 1.44 to 2.53). Rural hospitals also had greater odds than urban hospitals (OR = 1.75; 95% CI 1.11 to 2.77).

DISCUSSION

We found that between 2000 and 2004, there were >3700 hospitalisations for sport-related concussions nationwide, resulting in >US$29 million in total hospital charges for 5 study years, which equals nearly US$6 million per year. Although most sport-related concussions do not require hospitalisation, NIS data demonstrates that sport-related concussions constitute a substantial economic burden to the healthcare system and the patient’s family. We also found that nearly half of sport-related TBI hospitalisations among children aged 5–18 years were due to concussion, and that admissions increased with age. Of patients who sustained sport-related concussions, over half had experienced loss of consciousness. Although sport-related concussions may not directly lead to death, the effect on the overall health and well-being of these young people can be substantial, with potential consequences including cumulative effects of future brain injuries and possible cognitive dysfunction and/or brain damage.7 22 To our knowledge, this is the first study that describes patient and hospital characteristics associated with hospitalisations for paediatric sport-related concussions in the USA, Our findings have implications for the development of age-appropriate concussion management programmes for paediatric sports participants, a population that has often been neglected.7

We found that older age was associated with an increased number of sport-related hospitalisations for concussion and that the proportion of sport-related concussions among all TBI hospitalisations increased with age. Increasingly, competitive sports offered in middle and high schools and a higher level of involvement in organised sports competitions among athletes at this age may be the main reason for this observed difference.23 The increased number of sport-related hospitalisations for concussion in 10–14 year-olds may also imply an increase in children engaging in athletic activities or competitive sports at a younger age.13 14 Our findings suggest that establishing concussion prevention and management programmes for children at an early age is an important step to prevent initial concussions and protect children from repeated concussions. These programmes should be present when children begin involvement in organised sports and include age-appropriate protective equipment, appropriate return-to-play guidelines, consistent enforcement of rules that protect children’s safety, and presence of an individual well trained in recognition and management of concussion symptoms at sporting events.22 30^–32 Recently, the Centers for Disease Control and Prevention developed on-line tools specifically for coaches, including a symptom checklist and an action plan. These can be downloaded and carried on the sidelines during practices and games.33 Given the significant increase in hospitalisations for sport-related concussions occurring in school-age children, concussion education and prevention programmes must target a broader range of people, including players, coaches, parents, athletic trainers and doctors.23

Although boys had a higher number of sport-related hospitalisations for concussion than girls, we found that gender was not associated with increased odds of sport-related concussion versus other sport-related TBIs resulting in hospitalisations. Previous studies examining the role of gender on sport-related concussion incidence have found mixed results. Some reported that high school female athletes were at a higher risk of sustaining concussion than male athletes,4 34 whereas others found that male athletes had higher concussion rates than female athletes.35 36 An overwhelming majority of studies on sport-related concussions has been focused on sports such as football, with primarily male participants.15^–20 Lack of data and research on female athletes has limited our understanding of the role of gender in the risk of sport-related concussions including those severe enough to warrant hospitalisations. With increasing numbers of girls participating in sports and recreational activities, future research efforts should be directed to both boys’ and girls’ sports.

This is the first study to report that non-teaching hospitals and rural hospitals are more likely to admit a child or adolescent with a diagnosis of sport-related concussion. One possible explanation is that medical care providers in these hospitals may be more conservative in their management of concussion compared with providers in teaching or urban hospitals. The low volume of hospitalisations, coupled with lower access to MRI or other diagnostic tools may have also contributed to a higher proportion of hospitalisations for sport-related concussion in non-teaching and rural hospitals.37 A relative lack of clinical experience with this type of injury and the inability to quickly exclude intracranial pathology with advanced imaging, either as stand-alone or synergistic factors, could logically lead to increased hospitalisations for observation. We also found that, compared with the northeast, all other regions had lower odds of admitted concussion. Although regional differences have been documented for many types of health outcomes and diagnostic patterns, the reasons for our findings are not clear and warrant further study.

Prospective sport-related concussion research has documented that 6.3–8.9% of collegiate athletes have experienced loss of consciousness after a concussion.17 20 38 Although a brain injury can occur without a reported loss of consciousness, the American Academy of Neurology and Colorado Medical Society guidelines both recommend transporting people who have sustained any loss of consciousness to a hospital for further evaluation.39 40 More than half of concussion hospitalised patients in our study experienced loss of consciousness, which may partially reflect the fact that a hospital admission for sport-related concussions would be more likely when loss of consciousness has been reported. However, the Prague agreement statement, a new guideline published in 2005, suggested that loss of consciousness alone is no longer an absolute indication for transport to an emergency department.41

We found that >80% of hospital admitted sport-related concussions were admitted for observations without receiving any procedure and almost half of principal procedures performed among sport-related patients with concussion did not involve imaging studies to exclude the possibilities of other TBIs. Some healthcare providers have questioned whether such hospitalisation is necessary because of the added cost and stress to the patient and their family, as well as the increased burden to the healthcare system.42 However, the lack of standardised concussion diagnosis and management protocols makes it challenging for healthcare providers to care for paediatric sport-related concussions.23 30^–32 In addition, guidelines that require imaging procedures for diagnosis would be difficult to apply in small or rural hospitals lacking particular imaging capabilities.

Discharge without observation also has implications for the patient’s family. We found that more than half of paediatric sport-related hospitalisations for concussion last <24 hours. Therefore, parents, who often play a significant role in sport-related concussion prevention, also bear the primary responsibility for observing and monitoring the symptoms of concussion in their children after the discharge. Educating parents on prevention of a concussion and repeat concussions, and on how to care for concussed children after discharge, should be included routinely in the practice of paediatric healthcare providers.23 43

This study has several limitations. First, our definition of TBI was based on retrospective review of ICD-9-CM codes of primary diagnosis. The absence of clear clinical guidelines for diagnosis and management of TBI may have resulted in inconsistencies among healthcare providers in coding for diagnoses of TBIs, as well as in clinical management that could impact LOS and total hospital charges. These inconsistencies could potentially contribute to biased estimations in our findings. Second, the three E codes we used to define sport-related injury had high specificity but low sensitivity.44 In addition, approximately 16% of NIS cases do not have an E code.27 Therefore, our projection of the number of sport-related hospitalisations for concussion might have been underestimated. Lastly, the financial charge information provided by the NIS is based on hospital charges, not actual costs. Thus, our estimation of total hospital charges may not fully reflect the financial effect on the patients and their families.

CONCLUSIONS

Despite these limitations, this is the first study using national data to estimate morbidity and economic consequences associated with paediatric sport-related hospitalisations for concussion. We found variation in the management of sport-related concussions, both for the patient and hospital. Although guidelines for the identification and early management of sport-related concussions exist, they do not focus on admission decisions or in-patient management.23 31 The development of scientific based guidelines for in-patient care of concussions may be warranted.