Staffing, Financial, and Administrative Oversight Models and Rates of Injury in Collegiate Athletes

We analyzed injury and exposure data from the NCAA ISP for the school years 2009–2010 through 2013–2014, obtained from its external investigator Data Injury Statistics Clearinghouse program. This dataset includes injury and athlete-exposure data for 25 sports (Table 1). Clinicians at some NCAA schools voluntarily provided data to the NCAA ISP through an interface with the electronic health records used by the school's sports medicine department.³  Additional descriptions of the NCAA ISP data, collection procedures, and quality control are reported elsewhere.³ 

Our 3 primary outcomes were injury incidence, reinjury incidence, and time loss from sport due to injury. We measured 1 secondary outcome: concussion incidence. Our exposure measure was athlete-exposures (AEs), defined as “1 student-athlete participating in 1 NCAA-sanctioned practice or competition in which he or she was exposed to the possibility of athletic injury, regardless of the time associated with that participation.”³  This dataset includes all injuries, regardless of whether they resulted in the athlete's removal from sport participation.³  Injury rates were calculated by summing the total number of injuries for each school-team-year and dividing by the total AEs for that same subgroup. A reinjury was a classification provided by the diagnosing clinician. Reinjury rates were calculated using clinician-classified reinjury at the school-team-year level, divided by total AEs. We used time loss, measured in days and defined as “the time between the original injury and return to play at the level that would allow competition participation.”³  Time loss (in days) was summed at the school-team-year level. Concussion rates were calculated by summing concussions at the school-team-year level and using the same AE denominator as described earlier.

One of our independent variables of interest, clinicians per athlete, came from NCAA research data on sports medicine staffing for the same school years for which we have injury data (2009–2010 through 2013–2014). This information is gathered annually by the NCAA through a required sport sponsorship form filled out by a representative from each school. The ratio of clinicians per athlete was created by taking the total count of sports medicine clinicians at a given school during a given school year and dividing it by the number of participating athletes for that same school year.

Additional variables of interest, the model of financial oversight and the model of administrative oversight, came from a previous telephone survey of sports medicine departments at a subset of NCAA member institutions.¹²  The financial model distinguishes between sports medicine groups that are financed by the athletics department versus another organizational unit (eg, university health services or an academic department). The administrative oversight model distinguishes between clinicians who report to the athletics department versus another organizational unit.

We also analyzed data that were publicly available through Title IX disclosures to the United States Department of Education and its Equity in Athletics Data Analysis.¹⁸  Specifically, we measured athletic department resources by using gross athletic revenues for the same school years for which we had injury and staffing data (2009–2010 through 2013–2014). Participation data on the number of teams across sports were used to extrapolate our findings to the broader NCAA context.

After reporting basic descriptive information about our data, we used regression to quantify the relationships among our 4 outcomes (injury, reinjury, time loss, and concussion) and our independent variable of interest (clinicians per athlete) at the sport-school-year level. We calculated Poisson models for injury and reinjury counts and included AEs as an offset term. We generated a zero-inflated Poisson model for time loss, using total injuries as an offset, because more than half of the injuries were not associated with time loss. We calculated a zero-inflated Poisson model for concussion, using total AEs as an offset, because concussion counts were heavily skewed toward zero. Because zero-inflated models were applied to account for the high proportion of zeros in the data, and not because we suspected 2 mechanisms for zero and nonzero outcomes, we generated an intercept-only model for the zero component of both zero-inflated Poisson models. All regressions controlled for sport, school, division of competition, and academic year. To account for the nonindependence of teams within the same school, in the Poisson models, school was included as a random effect. To assess the relationship between administrative and (separately) financial oversight models on injury outcomes, injury data were restricted to the subset of schools for which model information was available and Poisson regression was conducted. In this case, our independent variables were financial (or administrative) oversight model, clinicians per athlete, sport, and division of competition. The clinicians-per-athlete measure was standardized to a z score for ease of interpretation in the model.

To assess the generalizability of our results, we compared schools that provided data to the NCAA ISP with those that did not by using several structural descriptors. Independent 2-sample t tests compared continuous measures (total athletes, clinicians per athlete, and total athletic revenues) and χ² tests compared categorical measures (public versus private status, NCAA division of competition, financial and administrative oversight models) between groups.

During the course of 3 570 053 games or practices in 25 sports at 141 schools over 5 academic years, 12 296 collegiate athletes experienced 25 203 injuries. Reporting clinicians categorized about 10% of the injuries (n = 2517) as reinjuries. More than half of the injuries (52%, n = 13 025) resulted in no time loss. Of the injuries that resulted in time loss, the median time was 6 days (interquartile range [IQR] = 3–12 days). A total of 1566 concussions were diagnosed.

Schools contributed various amounts of information to the NCAA ISP. Most schools contributed data for only 1 or 2 of the 5 years (57 schools = 1 year, 43 schools = 2 years, 10 schools = 3 years, 12 schools = 4 years, and 19 schools = 5 years). More schools participated in more recent years (2009–2010 = 53 schools, 2010–2011 = 58 schools, 2011–2012 = 51 schools, 2012–2013 = 81 schools, and 2013 and 2014 = 73 schools). A school could contribute data for any or all sports, leading to a range of total sport-seasons of injury data contributed by schools (IQR = 2–14 sport-seasons). Thus, some sports had significantly more seasons of injury information than others (Table 1), some teams had more years of contributed data, and the total number of injuries contributed by a given school ranged widely (IQR = 18–242 injuries) as did the number of AEs (IQR = 3197–13 998 AEs).

The school-level mean rate of injury in the sample overall was 103.5 per 10 000 AEs (Table 2). Rates of reinjury and concussion were 10.5 and 6.1 per 10 000 AEs, respectively. More than half of the injuries (57.7%) resulted in no time loss. Among those that resulted in time loss, the median time was 6 days.

The financial and administrative oversight models for sports medicine were available for 70 of the 141 (49.6%) schools participating in the NCAA ISP. Of the schools with available model information, the majority of sports medicine groups were financed by (n = 55, 78.6%) and administratively overseen by (n = 57, 81.4%) the athletics departments at their institutions.

The average number of athletes per clinician varied substantially, ranging from 24 to 891 (median = 118, SD = 100, IQR = 80–166 athletes). This variation stemmed from large ranges in both the number of total sports medicine clinicians (range = 1–28) and the number of athletes (range = 74–999) across schools.

We found significant associations between clinicians per athlete and injury rates, reinjury rates, time loss, and concussion rates (Table 3). Schools that had 1 SD above the average number of clinicians per athlete had a 9.5% lower rate of injuries (103.6 versus 93.7 per 10 000 AEs; incidence rate ratio [IRR] = 0.905, P < .001), 2.7% lower rate of reinjuries (10.6 versus 10.3 per 10 000 AEs; IRR = 0.973, P = .004), and 6.7% lower rate of concussion (6.1 versus 5.7 per 10 000 AEs; IRR = 0.933, P < .001). Schools that had 1 SD above the average number of clinicians per athlete had 16% more days of time loss (5.0 versus 4.2; IRR = 1.16, P < .001).

We observed significant associations between financial and administrative oversight models, staffing, and overall injury rates (Table 4). Schools in which the sports medicine team was financed by the athletics department had higher rates of injury than those in which financing was from other departments (unadjusted = 140.5 versus 110.6 per 10 000 AEs). Similarly, schools in which the sports medicine team administratively reported to the athletics department had higher rates of injury than schools in which clinicians reported to other departments (unadjusted = 141.5 versus 108.3 per 10 000 AEs). Controlling for staffing, sport, and division of competition, sports medicine groups that were financed by the athletics department were associated with a 31% higher incidence of injury than those financed by other departments (IRR = 1.31, P < .001). Similarly, controlling for staffing, sport, and division of competition, sports medicine groups that reported administratively to the athletics department were associated with a 9% higher incidence of injury than those reporting to other departments (IRR = 1.09, P < .001). The relative effect of staffing on injury increased when examined in conjunction with either the financial model or administrative oversight model. Holding the financial model constant, schools with 1 SD more clinicians per athlete had a 41% lower incidence of sport-related injury than schools with an average number of clinicians per athlete (IRR = 0.59, P < .001). Similarly, holding the administrative model constant, schools with 1 SD more clinicians per athlete had a 42% lower incidence of sport-related injury than schools with an average number of clinicians per athlete (IRR = 0.58, P < .001; Table 4).

Schools that contributed injury information to the NCAA ISP tended to have more athletes, more clinicians per athlete, and higher total athletics revenues and more frequently reported administratively to the athletics department, were less frequently financed through the athletics department, and more often classified as Division 1 (Table 5). Schools for which we had model information had a higher average injury incidence than NCAA ISP schools overall (129.1 versus 103.5 per 10 000 AEs; IRR = 1.25).

Using data from a wide range of men's and women's sports, we determined that the number of clinicians per athlete was associated with the rates of injury, reinjury, concussion, and time loss. We also noted an association between the financial and administrative oversight models of the sports medicine departments and the incidence of injury. With these results, we provide an important initial indication that, as in other areas of health care, structural features in the sports medicine environment may be indicators of health care quality and patient outcomes. Our findings are in line with a previous report¹⁵  of higher rates of injury and recurrent injury in high school girls' basketball and soccer when an athletic trainer was present versus not present. However, they are in contrast with the lower rate of diagnosed concussion when athletic trainers were not present.¹⁵  In our sample of colleges, at least 1 health care provider was always available; thus, this difference may reflect the relative and absolute availability of health care providers and the subsequent effects on athlete health outcomes. Further application of the tools and frameworks from health services research to the sports medicine context will likely advance our understanding of the factors contributing to athlete health outcomes.

To our knowledge, we provide the first empirical evidence relating sports medicine departments' supervisory and financial models with athlete health outcomes. Previous investigators concluded that variations in staffing¹²  and differences in legal risks¹⁹  were associated with the financial model and reporting structure. Beyond the legal risks, we suggest that health risks are associated with structuring a sports medicine group's administrative and financial oversight through the athletics department. Given the nature of our study, we cannot say that this relationship is causal. However, this finding does add to the legal, ethical, staffing, and professional reasons against structuring sports medicine departments administratively and financially under an athletics department.^12,19–21 

Schools with more clinicians per athlete had modestly lower rates of injury. However, the large number of collegiate athletes implies that even a modest relationship could have large overall health effects. For example, a 9.5% reduction in the overall injury count in this sample equates to 2394 fewer injuries, limited only to the teams included in the dataset during the 5-year study period. Extrapolated to the broader NCAA context, this corresponds to an annual reduction of about 14 015 injuries for only the sports included in the data, which represent a fraction of the total NCAA-sanctioned sports. This relationship was strengthened when the financial or administrative oversight model was included in the analysis.

The relationship between the number of clinicians per athlete and injury rates likely reflects several mechanisms. Collegiate sports medicine clinicians participate in injury-prevention programs, which could influence injury rates. Higher staffing levels in a collegiate sports medicine department could also indicate broader positive attitudes toward athlete health at the school, which translates into practices across the athletic environment that help reduce injury rates. For example, researchers¹²  demonstrated that schools whose sports medicine departments were financed by or reported to a medical institution had a greater ratio of clinicians to athletes. Conversely, it is possible that documentation practices vary by staff size. For instance, more clinicians may translate to better reporting or documenting of injuries. Given that our outcomes were all diagnosed injuries (documented in the medical record), if documentation improves as staff size increases, we may be underestimating the effect size of staffing on athlete injury outcomes.

Sports medicine staff may have an even more direct influence on rates of reinjury, which are affected by the clinical management of initial injuries.²²  We showed that schools with more clinicians per athlete had lower rates of reinjury. Reinjury rates may be lower in schools with more clinicians per athlete because the initial injuries are cared for more frequently and attentively. Although we lack data on the treatment of these injuries, this hypothesis is further supported by our finding that athletes at schools with more clinicians per athlete lost more time to injury. Although time loss due to injury can indicate injury severity, greater time loss can also reflect a more conservative approach to injury management. Additional support was provided by researchers⁵  who demonstrated that medical services per high school football injury increased when the athletic trainer was employed full time by the high school as opposed to full or part time by an outreach organization.

Average case loads for clinicians in this sample ranged from 24 to 891 athletes, and in the NCAA staffing data overall, the range was even greater, from 24 to 955 athletes. This signals a significant disparity in access to care for collegiate athletes and workloads for collegiate sports medicine clinicians. This differential workload may have a negative effect on clinicians as well as athletes. Insufficient staffing is associated with more job dissatisfaction, diminished work-life balance, and burnout among sports medicine clinicians.^23–25  This may also affect clinician performance; for example, insufficient staffing of the sports medicine department may impede the implementation of concussion-related care policies.^6,26  Future work to establish benchmarks for appropriate clinician case loads is warranted; until then, institutions whose clinicians care for a relatively large number of athletes (for example, those above the 75th percentile of 166 athletes per clinician) may consider measures to reduce case loads to reduce the risk of injury in athletes.

Our structural measure of the number of clinicians per athlete does not distinguish among levels of credentialing, full-time equivalents, or allocation of clinicians across sports within a school. In addition, injury data are voluntarily contributed to the NCAA ISP. Not all schools provide data, and those that do typically do not provide data for all sports, which may limit the generalizability of our conclusions, especially given the measurable differences between participating and nonparticipating schools we found. This study included only NCAA schools and, therefore, the conclusions may not be generalizable to institutions outside of the NCAA. Injury-reporting practices may vary across schools in unobservable ways. As noted earlier, understanding the mechanisms underlying the relationship between sports medicine staffing and injury outcomes requires additional research.

Based on our data, we suggest that increasing the number of sports medicine clinicians on staff and changing the financial and administrative reporting structures of sports medicine groups may improve the health of collegiate athletes by reducing injury rates. Future research addressing medical services and costs would enhance our understanding of the effectiveness and cost-effectiveness of increasing the number of clinicians in the collegiate sports medicine environment.²⁷  Additional evaluation of features that influence injury rates across sports has the potential to provide widespread positive health outcomes for collegiate athletes. As the relationships among sports medicine staffing, model of care, and athlete health outcomes are clarified, schools, conferences, and leagues will be increasingly well positioned to intervene to improve athlete health.

This work was supported by the National Institute of Mental Health Award Number T32MH019733. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

We thank the many athletic trainers who have volunteered their time and efforts to submit data to the NCAA Injury Surveillance Program. Their efforts are greatly appreciated and have had a tremendously positive effect on the safety of collegiate athletes. We would also like to thank the athletic trainers who participated in the telephone survey on sports medicine staffing. This publication contains materials created, compiled or produced by the Datalys Center for Sports Injury Research and Prevention, Inc. on behalf of the National Collegiate Athletic Association©. All rights reserved. The NCAA Injury Surveillance Program data were provided by the Datalys Center for Sports Injury Research and Prevention. The Injury Surveillance Program was funded by the National College Athletic Association (NCAA). The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the Datalys Center or the NCAA.