Using a sophisticated mathematical approach (machine learning), this project aims to develop a model that can be applied more broadly across military and civilian institutions to inform intervention efforts that improve concussion care seeking.
Our long-term goal is to increase the number of and how quickly military service members and athletes seek medical care after sustaining a concussion. Our immediate objectives are to identify negative outcomes linked with previous history of undiagnosed concussions and delayed medical care and identify individual and institutional level risk factors for undiagnosed concussions and delayed medical care.
Concussions, also referred to as mild traumatic brain injury (mTBI), account for 86.2% of military medical appointments associated with traumatic brain injury, affecting more than 37,292 military service members and upwards of 3.8 million civilians each year. In the absence of outward signs, such as being “knocked out,” medical professionals rely on military service members and athletes to tell someone about the symptoms they are experiencing. However, concussion symptoms are not always observable, resulting in half of concussions going unrecognized. Further, an alarming 64% of those eventually diagnosed with a concussion wait a considerable period of time before seeking medical care.
Seeking medical care quickly decreases the negative impact of concussion. Brain vulnerability leaves those that continue with duty or sport post-concussion exposed to adverse outcomes that may lead to delayed recovery, further injury, or more negative outcomes. Individuals with previous concussions that were not diagnosed experience more severe concussion-related symptoms even when not injured, have more severe psychological symptoms when not injured, are more likely to experience concussion-related symptoms following repeated head impacts, are more likely to be “knocked out” with subsequent concussions, and are more likely to experience symptoms with exercise after another concussion.
To target and tailor interventions that increase the number of and how quickly individuals seek medical care after sustaining a concussion, it is critical to identify those individuals and institutions at risk for previous history of undiagnosed concussions and those that delay seeking medical care. Although military service members and athletes recognize concussion symptoms, they may still choose to conceal the injury for a variety of reasons, such as: fear of letting squadron members or teammates down, fear of losing military qualifications (e.g. pilot slots) or starting positions, and/or not recognizing the seriousness of the injury. Using a sophisticated mathematical approach (machine learning), we aim to develop a model that can be applied more broadly across military and civilian institutions to inform intervention efforts that improve concussion care seeking.
Our long-term goal is to increase the number of and how quickly military service members and athletes seek medical care after sustaining a concussion. Our immediate objectives are to identify negative outcomes linked with previous history of undiagnosed concussions and delayed medical care and identify individual and institutional level risk factors for undiagnosed concussions and delayed medical care. Our proposal addresses the following FITBIR Analysis Award Topic of Interest: correlation of the time between injury and initial treatment with long-term outcomes.
We will address our aims by optimizing the FITBIR CARE Consortium dataset. We will use pre-injury concussion history outcomes from 34,146 military cadets and athletes to identify previous concussion diagnosis status (undiagnosed, diagnosed, no history) and post-injury clinician-reported outcomes from 3,323 concussions to identify concussion care seeking status (immediate care seeker and not exposed, delayed care seeker and exposed, delayed symptom onset and exposed). For aim I, we will compare pre- and post-concussion outcomes between groups at one pre-injury and three post-injury timepoints (24-48 hours post-concussion, no longer symptomatic, and full return to play/duty). For aim II, we will use a sophisticated mathematical approach (machine learning) which will classify individual (e.g., sex, risk-seeking preferences) and institutional risk factors (e.g. military vs. civilian, academic caliber) for identifying distinct previous concussion diagnosis status groups and concussion care seeking groups.
The low and steady concussion reporting rate of ~50% and high prevalence of delayed care seeking make it clear that current concussion education efforts are not effective. Funding and effort can be saved if educational efforts can be targeted and content can be tailored to address the negative outcomes associated with not seeking care and account for the individual and institutional barriers to concussion care seeking. A larger sample that can support a machine learning approach can inform larger scale organizational decision making. We must increase the number and immediacy of those with concussions who seek medical care to ensure access to rapidly improving concussion treatments and rehabilitation. In the absence of this research, 50% of military service member and athlete concussions will remain undiagnosed and untreated, and the remaining majority of those diagnosed will continue to delay seeking care.
Associate Professor, Kinesiology
Assistant Professor, KinesiologyJohna, Register-Mihalik
Assistant Professor, UNC Chapel HillZachary Kerr
Assistant Professor, UNC Chapel HillChristopher D’Lauro
Associate Professor, U.S. Air Force AcademyEmily Kroshus
Adjunct Assistant Professor, University of WashingtonDaniel Leeds
Assistant Professor, Fordham University