THE CAUSES OF SCI VARY depending on age, race and

Transcription

1 1350 ORIGINAL ARTICLE Spinal Cord Injury and Co-Occurring Traumatic Brain Injury: Assessment and Incidence Stephen Macciocchi, PhD, ABPP, Ronald T. Seel, PhD, Nicole Thompson, MPH, Rashida Byams, MS, Brock Bowman, MD ABSTRACT. Macciocchi S, Seel RT, Thompson N, Byams R, Bowman B. Spinal cord injury and co-occurring traumatic brain injury: assessment and incidence. Arch Phys Med Rehabil 2008;89: Objectives: To examine prospectively the incidence and severity of co-occurring traumatic brain injury (TBI) in persons with traumatic spinal cord injury (SCI) and to describe a TBI assessment process for SCI rehabilitation professionals. Design: A prospective, cohort design to collect and analyze clinical variables relevant for diagnosing co-occurring TBI. Setting: An urban, single-center National Institute of Disability and Rehabilitation Research Model Spinal Cord Injury System in the Southeastern United States. Participants: People (N 198) who met inclusion criteria and provided consent within an 18-month recruitment window. Interventions: Not applicable. Main Outcome Measure: FIM cognitive scale. Results: Based on participants presence and duration of posttraumatic amnesia, initial Glasgow Coma Scale total score, and presence of cerebral lesion documented by neuroimaging, 60% of our traumatic SCI sample also sustained a TBI (n 118). Most co-occurring TBIs were mild (34%). Co-occurring mild complicated (10%), moderate (6%), and severe TBI (10%) were less common but still occurred in a significant percentage (26%) of persons with traumatic SCI. Persons with traumatic SCI who were injured in motor vehicle collisions and falls were more likely to sustain a co-occurring TBI. Cervical level traumatic SCI was associated with greater rates of TBI but not more severe injuries. Tree analyses established a practical algorithm for classifying TBI severity associated with traumatic SCI. Analysis of variance established criterion validity for the algorithm s TBI severity classifications. Conclusions: Findings from our prospective study provide strong support that TBI is a common co-occurring injury with traumatic SCI. Incomplete acute care medical record documentation of TBI in the traumatic SCI population remains a considerable issue, and there is a significant need to educate emergency department and acute care personnel on the TBI clinical data needs of acute rehabilitation providers. A systematic algorithm for reviewing acute care medical records can yield valid estimates of TBI severity in the traumatic SCI population. From the Shepherd Center, Atlanta, GA (Macciocchi, Seel, Thompson, Bowman); University of Georgia, Athens, GA (Macciocchi); Georgia State University, Atlanta, GA (Byams); and Emory University, Atlanta, GA (Bowman). Supported by the National Institute on Disability and Rehabilitation Research, U.S. Department of Education (grant no. H113G030004). No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit on the authors or on any organization with which the authors are associated. Reprint requests to Stephen Macciocchi, PhD, ABPP, Shepherd Center; 2020 Peachtree Rd, Atlanta, GA 30309, stephen_macciocchi@shepherd.org /08/ $34.00/0 doi: /j.apmr Key Words: Brain injuries; Diagnosis; Incidence; Rehabilitation; Spinal cord injuries by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation THE CAUSES OF SCI VARY depending on age, race and ethnicity, and sex, but most SCIs are caused by trauma sustained during MVCs, falls, assaults, and sports participation. Epidemiologic studies indicate approximately 10,000 persons sustain a traumatic SCI each year. 1 Although some persons who sustain a traumatic SCI do not survive, the overwhelming majority of persons with traumatic SCI who do survive require extended inpatient rehabilitation. Similar to traumatic SCI, TBI can be a significantly disabling event. Not surprisingly, depending on injury etiology, many persons sustain both a traumatic SCI and TBI. 2,3 A co-occurring TBI would be expected to significantly impact traumatic SCI rehabilitation outcome. 4,5 Consequently, diagnosing a co-occurring TBI early in the traumatic SCI rehabilitation process may facilitate treatment planning and enhance eventual rehabilitation outcome. Investigators began to examine the incidence of traumatic SCI and co-occurring TBI 4 decades ago. Published incidence rates vary widely, ranging from 16% to 59% (table 1) In studies in which nosology was provided, at least 50% and as many as 82% of TBIs were considered minor or mild. Most investigations used retrospective clinical metrics to identify co-occurring TBI in persons hospitalized for traumatic SCI rehabilitation. By using acute care medical records, investigators typically retrieved LOC, TBI ICD-9 code, neuroimaging, and less frequently PTA data to establish a diagnosis. Studies that exclusively used TBI ICD-9 codes or positive neuroimaging for TBI diagnosis had lower incidence rates (16% 34%) than studies that included measures of altered consciousness (28% 59%). In many studies, investigators found clinical information critical for diagnosing TBI after traumatic SCI was absent or embedded in acute care records and difficult to ANOVA ASIA CT GCS ICD-9 LOC MVC NIDRR PTA SCI TBI List of Abbreviations analysis of variance American Spinal Injury Association computed tomography Glasgow Coma Scale International Classification of Disease 9th Revision loss of consciousness motor vehicle collision National Institute of Disability and Rehabilitation Research posttraumatic amnesia spinal cord injury traumatic brain injury

2 Study and Year Table 1: Chronological Review of Studies Reporting Incidence of Co-Occurring TBI in Persons With SCI Sample Size SCI Level/Etiology Design and Setting TBI Diagnostic Criteria Incidence TBI (%) TBI Nosology Harris 6 (1968) 150 Tetra 45%, NR Retrospective, TC Minor LOC for minutes and 33 60% minor; 40% major PTA 12h Meinecke 7 (1968) 595 NR, MVC 8% Retrospective, TC NR 25 74% concussion; 15% skull fx or brain contusion Shrago 8 (1973) 50 Tetra 100%, NR Retrospective, TC Contusions, abrasions, lacerations, 34 NR or skull fracture Silver et al 9 (1980) 100 Tetra 51%, MVC 41% Retrospective, TC LOC, PTA, contusions 50 82% minor; 18% serious Rimel 10 (1981) 253 NR, MVC 46% Prospective, CNS registry NR 47 NR Young et al 11 (1982) 1615 Tetra 54%, MVC 46% Prospective, SCI registry ICD codes 16 60% concussion; 20% skull fx; 20% brain injury Davidoff et al 12 (1985) 88 C1-T6 76%, MVC 38% Retrospective, TC, AR LOC or PTA 42 NR Richards et al 13 (1988) 150 Tetra 51%, MVC 58% Prospective, AR LOC 59 NR Steudel et al 14 (1988) 59 Tetra 100%, MVC 63% Retrospective, TC Tunnis and Loew classification; GCS 56 45% mild; 30% moderate/severe; 25% NR Davidoff et al 15 (1988) 82 Tetra 44%, MVC 56% Prospective, TC, AR PTA (LOC, 44%) 49 PTA duration: 40% 1h; 18% 1 12h; 16% 12 72h; 24% 72h Michael et al 16 (1989) 92 Tetra 100%, NR Retrospective, TC ICD code for TBI or positive head 24 81% mild/moderate; 19% severe CT scan Saboe et al 17 (1991) 508 Tetra 24%, multilevel Retrospective, TC NR 26 NR 50%, MVC 56% Pagni and Massaro NR, NR Retrospective, TC Mild LOC 1h; small fx or 54 60% mild; 40% severe (1991) contusion; severe LOC 1h Go et al 19 (1995) 4107 NR, MVC 45% Prospective, SCI registry LOC; head injury significant dysfunction secondary to TBI % LOC; 12%; head injury (some group overlap) Strubreither et al 20 (1997) 322 NR, NR Retrospective, AR Cerebral lesion 20 27% none; 42% minor/moderate; 31% severe Abbreviations: AR, acute rehabilitation center; CNS, central nervous system; fx, fracture; NR, not reported or could not be determined; TC, trauma center; Tetra, tetraplegia. SPINAL CORD INJURY AND TRAUMATIC BRAIN INJURY, Macciocchi 1351

3 1352 SPINAL CORD INJURY AND TRAUMATIC BRAIN INJURY, Macciocchi extract. Not surprisingly, investigators also found that in many studies traumatic SCI persons rehabilitation records failed to document any TBI-related information. Sample differences related to injury etiology and level likely influence incidence rates and are factors in whether someone sustains a traumatic SCI with co-occurring TBI as opposed to a traumatic SCI alone. For instance, persons involved in MVCs have been shown to have much greater rates of traumatic SCI with co-occurring TBI ( 50%) compared with persons injured in another manner ,17 In samples in which the rates of MVCs are low, the incidence of co-occurring TBI is also low. 7,12 MVCs may also be more highly associated with severe TBI. 21 Cervical level traumatic SCI also appears related to increased rates of co-occurring TBI in persons with traumatic SCI, 8,17 but these findings have not been universal. 11 Although not directly addressing the base rate of TBI in the traumatic SCI population, a number of studies have examined cognitive functioning in persons with traumatic SCI. Typically, these investigations did not document the presence or severity of co-occurring TBI. In addition, the number of traumatic SCI persons studied, the time frame of examinations, the neuropsychologic tests used, and the type of SCI also varied considerably. 13,22-25 For instance, in 1 study, the only variable used to define brain injury was a dichotomous (yes, no) rating for LOC. 13 Another investigation used various criteria to identify and classify brain injuries, including LOC, ratings by neuropsychologists, and scores on certain neuropsychologic instruments (eg, category test). The investigators found LOC to be an insensitive predictor of TBI and agreement between neuropsychologist raters was low (29%), in part because of a lack of information needed to render a TBI diagnosis. 24 Nonetheless, these studies indicate that persons with traumatic SCI often evidence cognitive impairment relative to normative expectations and/or controls. The lack of TBI diagnostic information in these studies limits the extent that poor cognitive functioning can be attributed to TBI in persons with traumatic SCI. In many studies, authors 24,26 have commented on the lack of precision in diagnosing brain injury presence and severity in traumatic SCI persons. TBI diagnosis in the traumatic SCI rehabilitation setting is dependent on extracting clinical metrics from the acute care record. Although there is no clear consensus on TBI nosology, basic criteria have been established for some time. The NIDRR Traumatic Brain Injury Model System 27 collects data on 4 diagnostic criteria: LOC, PTA, neuroimaging results, and GCS score. Williams et al 28 distinguished a 4-tier classification system for mild, mild complicated, moderate, and severe by using these criteria. Mild TBI was defined as a TBI with or without LOC, but some evidence of disrupted brain function such as confusion and a GCS of 13 through 15. Complicated mild TBI was defined similarly in terms of clinical indicators except persons with complicated mild TBI also were required to have cerebral pathophysiology such as a contusion, hemorrhage, or skull fracture. Moderate TBI was defined as a GCS score of 9 to 12 with or without identifiable cerebral pathology. Finally, persons with severe TBI were required to have GCS scores of 8 or below and experience coma with or without identifiable intracerebral pathophysiology. The presence and duration of PTA are other common indicators of altered consciousness secondary to TBI. In TBI sample studies, PTA has been consistently found to be a better predictor of rehabilitation outcomes and long-term disability than the GCS. The most widely referenced classification system for PTA duration was proposed by Russell and Smith 35 in which persons with greater than 24 hours PTA are classified as severe. However, more recent studies 36,37 have suggested that persons with less than 10 to 14 days of PTA generally have good outcomes at 1 year postinjury. In these studies, good outcome either included persons with moderate disability 37 or was assessed at 6 months postinjury. 36 We identified no empirical classification of PTA severity relative to acute rehabilitation outcomes. Last, diagnosis of PTA in a traumatic SCI population must take into account trauma care use of sedating and paralyzing medications that has the potential to artificially inflate the presence and duration of PTA. In summary, we have a high index of suspicion that traumatic SCI and TBI are common co-occurring disorders, but there is limited prospective research addressing the issue. Unfortunately, existing research has not led to broad-based, systematic investigations of traumatic SCI and TBI comorbidity. For instance, the NIDRR Model Spinal Cord Injury System does not currently collect data on persons with co-occurring traumatic SCI and TBI, despite having access to information widely recognized to be relevant for diagnosing TBI in persons with traumatic SCI. In addition, the most recent SCI state-ofthe-science review did not mention TBI as a risk factor for suboptimal outcomes after acute and/or postacute traumatic SCI rehabilitation. 1 Because of the variability in traumatic SCI and TBI research findings, almost all of which are 15 to 40 years old, our investigation prospectively examines the incidence and severity of TBI in the traumatic SCI population using diagnostic indicators of TBI including both PTA and GCS classification criteria. The primary objectives of our research were to (1) document and describe clinical variables relevant for diagnosing TBI in a traumatic SCI sample, (2) determine the incidence and severity of TBI in a traumatic SCI sample, (3) analyze the impact of traumatic SCI level and etiology on the incidence of co-occurring TBI, (4) identify an efficient algorithm and process for classifying TBI severity in the traumatic SCI population, and (5) evaluate the validity of the TBI classification system by using a criterion standard measure of cognition at acute rehabilitation admission. METHODS Participants All persons between the ages of 16 and 59 admitted for traumatic SCI rehabilitation over an 18-month period between January 2004 and July 2005 were considered for inclusion. Project objectives not reported in this article involved administering multiple measures in addition to documenting TBI incidence and severity. Because of the unavailability of bilingual examiners and culture equivalent, normed, measurement techniques, non English-speaking persons (n 33) were excluded from the study. Another 22 persons with traumatic SCI were discharged or transferred before study recruitment. The total number of eligible persons with traumatic SCI available for recruitment was 266. Of these eligible persons with traumatic SCI, 54 declined to participate, and 14 withdrew before the initiation of data collection. Thus, the final sample for the study was 198. Procedure Before beginning the project, approval was obtained for the investigation from NIDRR and Shepherd Center s Internal Research Review Committee. Potential participants were then interviewed to ascertain interest in the research project. If traumatic SCI persons expressed interest in participating, informed consent was obtained, formal interviews were scheduled, and a detailed record review was undertaken. Data on participants traumatic SCI and TBI were obtained from pri-

4 SPINAL CORD INJURY AND TRAUMATIC BRAIN INJURY, Macciocchi 1353 mary medical records including emergency medical services reports, acute hospital medical, and surgical records including ancillary imaging and medical consultation reports as well as acute rehabilitation records. Patients self-report of LOC and PTA at the time of the research interview was not used for diagnostic purposes. Conversely, emergency and acute care personnel may have used patient self-report of altered mental status at the time of the injury for brain injury diagnostic purposes. Two research coordinators documented participants injury data based on criteria provided during educational sessions before initiating the study. Each record was subsequently reviewed with investigators at 2 additional points in time to ensure clinical variables relevant for identifying the presence and/or absence of TBI were identified. Measures used to identify and classify TBI among traumatic SCI persons included the presence and duration of PTA, the GCS, and neuroimaging findings (eg, CT or magnetic resonance imaging). Brain injuries were classified as none, mild, mild complicated, moderate, and severe by using the categories proposed by Williams et al. 28 The 2 lead investigators used a combination of estimated PTA severity classification criteria and the Williams criteria 28 to establish a diagnosis. We set the estimated PTA severity criteria at longer durations than the Russell and Smith criteria 35 to minimize false-positives for severity. Similarly, we set the estimated PTA severity criteria at a shorter duration than recommended by Jennett 36 and Ellenberg 37 and colleagues to minimize false-negatives for persons at risk in an acute rehabilitation setting. PTA classifications were set as follows: 24 hours mild, 1 to 6 days moderate, and 7 days severe. In general, participants with no evidence of PTA and no evidence of intracranial lesion were classified as not having sustained TBI. Participants with a PTA duration of less than 24 hours and/or a GCS of 13 to 15 with no evidence of intracranial lesion were classified as mild. Participants with positive evidence of intracranial lesion based on neuroimaging and a PTA duration of less than 24 hours if present or a GCS total score of 13 to 15 if present were classified as mild complicated. Participants with an estimated PTA duration of 1 to 6 days or a GCS total score of 9 to 12 were classified as moderate irrespective of imaging findings. Participants with an estimated PTA duration of 7 days or greater or a GCS total score of 3 to 8 were classified as severe irrespective of imaging findings. To control for the potential of inflated severity classifications resulting from using intubated GCS total scores, an intubated score of 8T was lowered 1 severity classification level from severe to moderate and an 11T score was classified as either mild or mild complicated, depending on imaging findings. Data Analyses SPSS a was used to compute all statistical analyses. Descriptive statistics (ie, means and SDs and/or percentages) were calculated to characterize the sample in terms of demographic variables, injury etiology, level and completeness of SCI, and severity of TBI measures. To examine the impact of demographic, injury etiology, and traumatic SCI injury level and completeness on the presence and severity of TBI, the Cramer V was computed. The Cramer V examines the relationship between qualitative variable pairs and derives a product-moment r value corrected for sample size and number of cells. The Cramer V can range from 0 to 1, with higher values denoting more substantial group differences. Classification tree analysis was used to identify the most efficient algorithm for estimating TBI injury severity. Independent variables considered for establishing a diagnostic pathway were PTA duration, positive CT scan, LOC, and GCS total score. The dependent variable was the 4-tier diagnostic categories established by Williams. 28 The selection of an initial independent variable was left open, and variables with P less than.05 were considered for selection. The validity of TBI injury severity classification groupings was evaluated with ANOVA by using admission FIM cognitive scores as the reference standard. When a main effect for injury severity classification category was found by using an ANOVA procedure, a Tukey post hoc test was calculated to identify significant differences between classification levels. The number of planned comparisons and the potential for nonnormative sampling distributions of means was taken into account when selecting an level. An level of P less than.05 was deemed acceptable. RESULTS Sample Demographics and Traumatic SCI Variables Persons consenting to participate (n 198) totaled 74% of all eligible persons admitted for traumatic SCI rehabilitation during the enrollment period (N 266). Table 2 provides detailed data on participants age, sex, education, ethnicity, injury mechanism, and SCI motor diagnosis. The participants mean age SD was years. The majority of participants were men (79%). Most participants had at least a high school education (72%). Most participants were white (67%). Injuries resulted from MVCs (63%), violence (15%), sporting injuries (13%), and falls (10%). The most common injury levels were C5-8 ASIA grade A to C (30%) and T1-8 ASIA grade A to C (26%). The time from injury to Shepherd Center admission was days (median, 13d; range, 0 141d). TBI Diagnostic Variables and Incidence TBI diagnostic variables including estimated PTA, LOC, GCS scores, and neuroimaging (CT) findings are also presented in table 2. Most participants experienced LOC (54%) and some duration of PTA (58%). In many cases, the duration of PTA was unknown or unable to be estimated based on medical records (34%). GCS total scores ranged from 3 to 15, with the most common scores being 13 to 15 (47%). GCS total scores were unknown or not administered in a high number of cases (42%). The overwhelming majority of participants either had a normal CT scan or no CT scan reported in the medical record (82%). Based on our classification scheme, 60% of participants sustained TBI at the time of their traumatic SCI. Most injuries were mild TBI (34%); 10% sustained mild complicated brain injuries, and 6% and 10% of participants sustained moderate and severe brain injuries, respectively. The Impact of Traumatic SCI Injury Factors on Sustaining a TBI In terms of injury etiology, table 3 shows associated rates relative to any TBI and more severe co-occurring TBI. Persons involved in MVCs and falls were more likely to sustain a TBI than persons who were injured during sporting events or violence (P.001). MVC and falls were also associated with higher rates of more severe co-occurring TBI (P.01). In terms of injury level and completeness, cervical level traumatic SCI was significantly associated with higher rates of co-occurring TBI (P.05). Persons with injuries at the C1 through C4 level, regardless of injury completeness, had the highest rates of TBI.

5 1354 SPINAL CORD INJURY AND TRAUMATIC BRAIN INJURY, Macciocchi Table 2: Demographic and Characteristics of Study Participants (N 198) Characteristics Values Mean age SD (y) Mean days from injury to Shepherd admission SD Sex (men), % 79 Race/ethnicity (%) White 67 Black 31 Other 3 Education (y), % 8th grade 3 9th 11th 25 High school diploma, GED 57 Associate s degree 6 Bachelor s degree 7 Postgraduate degree 2 Etiology (%) MVC 63 Violence 15 Sporting injury 13 Falls/flying object 10 SCI motor level, completeness (%) C1-4, ASIA grade A C 9 C1-4, ASIA grade D 5 C5-8, ASIA grade A C 30 C5-8, ASIA grade D 6 T1-8, ASIA grade A C 26 T1-8, ASIA grade D 2 T9-12, ASIA grade A C 15 T9-12, ASIA grade D 3 L1-S3, ASIA grade A C 5 L1-S3, ASIA grade D 0 Estimated PTA duration (%) No PTA 42 1h h h 1 1 6d 2 7d 5 Present, duration unknown 34 GCS total score (%) 13 15, 11T , 8T 10T 5 3 8, 3T 7T 7 Unknown/not administered 42 Neuroimaging (%) Negative findings, no record 82 Positive findings 18 Cerebral contusion 6 Subdural hematoma 5 Subarachnoid hemorrhage 8 Positive LOC (%) 54 NOTE. Values are mean SD. Abbreviation: GED, General Educational Development diploma. Only persons with an injury level of T1 through S3 and ASIA grade D motor completeness had a significantly lower rate of TBI compared with all other motor level and completeness groups. Last, traumatic SCI injury level and completeness had no significant impact on rates of more severe TBI diagnoses (P.24). Identifying and Validating an Efficient Algorithm for Co-Occurring TBI Diagnosis Classification tree analyses were used to identify the most efficient path to categorize TBI presence and estimate injury severity. PTA duration was empirically selected as the most discriminating variable in establishing a TBI classification. GCS total scores and CT scan findings were also selected as significant discriminating variables. LOC data did not significantly contribute to classification decisions. In some decision branches, analyses considered absent information in addition to available TBI indicators. The final algorithm identified to classify brain injury severity is presented in figure 1. One-way ANOVA revealed that our estimated TBI injury severity classifications had a significant association with acute rehabilitation admission FIM cognitive scores (P.001) (table 4). Tukey post hoc tests revealed that persons with traumatic SCI who were classified as no TBI as well as those who were classified as mild and mild complicated TBI had significantly higher FIM cognitive scores than those classified as having severe injuries. Likewise, persons with traumatic SCI who were classified as no TBI and mild TBI had significantly higher FIM cognitive scores than those classified with moderate injuries. No other significant differences were observed, and all mean scores for TBI severity classification groups were in the expected linear hierarchy. DISCUSSION Investigators have been concerned about the impact of cooccurring TBI in persons with traumatic SCI for some time. Most existing investigations have been retrospective in nature, and variations in diagnostic methods and sample compositions have led to base rates that have ranged from 16% to 59%. Our prospective approach using a variety of sensitive TBI diagnostic data found an incidence of 60%. Most TBIs were mild (57%), but more severe injuries were also common (43%). Consistent with previous research, persons injured in an MVC were significantly more likely to sustain TBI, and this pattern held throughout the severity range. Most persons who sustain traumatic SCI are at fairly high risk ( 50%) for sustaining a TBI, and only persons with T1 through S3, ASIA grade D injuries are at significantly lower risk (11%). Table 3: Incidence of TBI in Persons With SCI by Injury Etiology and SCI Motor Level and Completeness Injury Etiology n % Any TBI* % Mild Complicated/ Moderate/Severe TBI MVC Violence Sporting injury Falls/flying objects SCI Motor Level, Completeness n % Any TBI % Mild Complicated/ Moderate/Severe TBI C1-4, ASIA grade A C C1-4, ASIA grade D C5-8, ASIA grade A C C5-8, ASIA grade D T1-S3, ASIA grade A C T1-S3, ASIA grade D *Cramer V.495, P.001. Cramer s V.245, P.01. Cramer V.241, P.05. Cramer s V.185, P.24.

6 SPINAL CORD INJURY AND TRAUMATIC BRAIN INJURY, Macciocchi 1355 Fig 1. Algorithm to estimate TBI presence and severity. Problems with incomplete acute care documentation of TBI sequelae have been reported in the research literature for over 40 years, and limited documentation complicated TBI assessment in our study. In many cases, emergency medical service and/or acute care medical records did not contain basic information necessary to diagnose the presence and severity of TBI. GCS scores were absent in 42% of cases, despite the widespread use of GCS scores by most emergency medical services and emergency room staff. In addition, although PTA was reported to be present in most cases, PTA duration was often difficult to estimate because of a lack of documentation. PTA duration has been shown to be a reliable index of TBI severity Table 4: Validity Evaluation of TBI Estimated Injury Classification Algorithm TBI Injury Classification Mean Admission FIM Cognitive SD 95% CI for Mean TBI not present (n 79) Mild (n 67) Mild complicated (n 20) * Moderate (n 11) Severe (n 18) ANOVA F P.001 Abbreviation: CI, confidence interval. *Significantly greater than severe injury classification based on Tukey post hoc test. Significantly greater than moderate injury classification based on Tukey post hoc test. and eventual outcome so access to PTA data is critical for traumatic SCI rehabilitation staff. We spent considerable time reviewing each case to extract injury variables that were not readily available in discharge summaries or in tabulated form. Based on our experience, SCI rehabilitation staff should formally review primary medical records for TBI in all traumatic SCI cases, especially when MVCs and falls are involved. Particular attention should be paid to emergency response reports, emergency department admission and discharge records, and daily intensive care documentation. In many cases, clinical evidence of a co-occurring TBI is not easily obtained. Although scale scores and clinical markers are highlighted in some cases, most often, TBI-related changes in mental status must be interpreted from daily acute care documentation in physician, nursing, and therapist notes. The algorithm presented in figure 1 is an approach clinical evaluators and traumatic SCI rehabilitation professionals can use to estimate the presence and severity of TBI in their admissions. Although the algorithm provided will be helpful when screening for a TBI, clinicians should be aware that a number of medical interventions and problems other than TBI can cause changes in mental status. Paralyzing and sedating medications are known to affect cognition and may produce periods of unresponsiveness and PTA. Extracranial trauma may also alter consciousness, which may be mistaken for signs of a TBI. In general, clinicians must review the medical record closely and determine the etiology of changes in mental status. Distinguishing between a TBI and a medically based change in brain function can be difficult, but documentation of persisting disorientation, confusion, and altered mental status often provides a basis for establishing a TBI diagnosis.

7 1356 SPINAL CORD INJURY AND TRAUMATIC BRAIN INJURY, Macciocchi Table 5: Estimating PTA Presence and Duration Based on Prerehabilitation Medical Records Documentation: PTA Present LOC ( ) No memory for collision ( ) GCS score 15 Opens eyes only when name called Unresponsive Inconsistently responds to questions Inconsistently follows commands Confused Combative Uncooperative and/or inappropriate Agitated Restraints used for behavior: physical, medical Documentation: PTA Resolution GCS score of 15 throughout full day Awake, alert, oriented 3 or 4 Fully oriented Communicates awareness of where he/she is and what has happened Participates in medical decision-making Describes and localizes pain Any set of daily communication that indicates orientation, awareness of circumstances, and lack of confusion Medical Record Source EMS, ED EMS, ED AC Medical Record Source NOTE. Rule out the initiation of sedating and paralyzing medications or extracranial trauma as primary cause of altered mental status; PTA resolution is best determined when supporting evidence is documented over 2 successive days. Abbreviations: AC, daily acute care records: physician, nursing, therapy; ED, emergency department records; EMS, emergency medical services report. PTA may be the most discriminating indicator and best predictor of TBI outcomes, but establishing the presence and duration of PTA can be challenging. SCI rehabilitation staff should take a systematic approach and consider all relevant TBI variables and potential confounds. Staff should carefully review acute care medical records for any physiologic or behavioral evidence of altered brain function. Staff must look for surrogate indicators of TBI embedded in clinical notes when traditional measures like the GCS are unavailable. Table 5 highlights common indicators of the presence and resolution of PTA that can be found in acute care documentation to estimate the severity of TBI in the traumatic SCI population. Importantly, negative or missing imaging findings are not sufficient to rule out the presence of TBI. Studies using neuroimaging to establish TBI diagnosis had incidence rates that were as much as 50% less than studies using measures of altered consciousness. In our sample, 16% of persons with moderate to severe TBI had negative neuroimaging findings. In these cases, failure to access GCS, PTA, or other indicators of altered brain function will likely lead to misdiagnosis. Study Limitations Several limitations of our research should be noted. First, our sample was comprised of a somewhat younger group of traumatic SCI survivors who were recruited from a single rehabilitation (long-term care) hospital. Consequently, samples recruited from multicenter rehabilitation centers that include older traumatically injured persons may show slightly different injury incidence and patterns. Second, the absence of a TBI diagnosis in the acute medical record accompanied by signs of normative cognitive functioning was interpreted as TBI not present. Improved specificity of TBI diagnostic documentation in acute care medical records might alter incidence rates. Third, there is no universally accepted nosology for TBI severity, and using different criteria such as GCS and PTA to establish TBI severity within a sample can be problematic. We developed an efficient TBI severity classification algorithm based on data likely to be found in the acute care medical record. Given the lack of an empirical criterion standard for classifying PTA severity, we identified and validated a classification scheme with an emphasis on minimizing false-negative classifications for more severe injury. CONCLUSIONS Findings from our prospective study and the research literature provide definitive evidence that TBI commonly occurs with traumatic SCI. Incomplete acute care medical record documentation of TBI in the traumatic SCI population has been reported for over 40 years and, nonetheless, remains a considerable issue. There is a significant need to provide education to emergency room and acute care personnel on the TBI diagnostic information requirements of acute rehabilitation providers, particularly formal tracking of PTA, GCS, and neuroimaging assessment. Because past research indicates that moderate and severe co-occurring TBI negatively impacts functional traumatic SCI rehabilitation outcomes, traumatic SCI rehabilitation staff should seek to index co-occurring TBI injury severity in the most efficient and comprehensive manner possible. Finally, future research should assess the impact of co-occurring TBI and premorbid cognitive abilities on traumatic SCI rehabilitation cognitive and functional outcomes as well as long-term disability. References 1. Sipski ML, Richards JS. Spinal cord injury: state of the science. Am J Phys Med Rehabil 2006;85: Elovic E, Kirshblum S. Epidemiology of spinal cord injury and traumatic brain injury: the scope of the problem. Top Spinal Cord Inj Rehabil 1999;5: Zafonte RD, Elovic E. Dual diagnosis: traumatic brain injury in a person with spinal cord injury. In: Kirshblum S, Campagnolo DI, DeLisa JA, editors. Spinal cord medicine. Philadelphia: Lippincott; p Bowman B, Macciocchi SN. Dual diagnosis: diagnosis, management and future trends. Top Spinal Cord Inj Rehabil 2004;10: Macciocchi SN, Bowman B, Coker J, Apple D, Leslie D. Effect of co-morbid traumatic brain injury on functional outcome of persons with spinal cord injuries. Am J Phys Med Rehabil 2004;83: Harris P. Associated injuries in traumatic paraplegia and tetraplegia. Paraplegia 1968;5: Meinecke FW. Frequency and distribution of associated injuries in traumatic paraplegia and tetraplegia. Paraplegia 1968;5: Shrago GG. Cervical spine injuries: association with head trauma. Am J Ther Nucl Med 1973;118: Silver JR, Morris WR, Otfinowski JS. Associated injuries in persons with spinal injury. Injury 1980;12: Rimel R. A prospective study of patients with central nervous system trauma. J Neurosurg Nurs 1981;13: Young JS, Burns PE, Bowen AM, McCutchen R, editors. Spinal cord injury statistics. Phoenix: Good Samaritan Medical Center; p

8 SPINAL CORD INJURY AND TRAUMATIC BRAIN INJURY, Macciocchi Davidoff G, Morris J, Roth E, Bleiberg J. Cognitive dysfunction and mild closed head injury in traumatic spinal cord injury. Arch Phys Med Rehabil 1985;66: Richards JS, Brown L, Hagglund K, Gua G, Reeder K. Spinal cord injury and concomitant traumatic brain injury: results of a longitudinal investigation. Am J Phys Med Rehabil 1988;67: Steudel WI, Rosenthal D, Lorenz R, Merdes W. Prognosis and treatment of cervical spinal injuries with associated head trauma. Acta Neurochir (Wien) 1988;43(Suppl): Davidoff G, Thomas P, Johnson M, Berent S, Dijkers M, Doljanac R. Closed head injury in acute traumatic spinal cord injury: incidence and risk factors. Arch Phys Med Rehabil 1988;69: Michael DB, Guyot DR, Darmody WR. Coincidence of head and cervical spine injury. J Neurotrauma 1989;6: Saboe LA, Reid DC, Davis LA, Warren SA, Grace MG. Spine trauma and associated injuries. J Trauma 1991;31: Pagni CA, Massaro F. Concomitant cranio-cerebral and vertebromedullary injuries: analysis of 121 cases. Acta Neurochir (Wien) 1991;111: Go BK, DeVivo MJ, Richards JS. The epidemiology of spinal cord injury. In: Stover SL, DeLisa JA, Whiteneck GG, editors. Spinal cord injury: clinical outcomes from the model systems. Gaithersburg: Aspen; p Strubreither W, Hackbush B, Hermann-Gruber M, Stahr G, Jonas HP. Neuropsychological aspects of the rehabilitation of patients with paralysis from a spinal cord injury. Spinal Cord 1997;34: Iida H, Tachibana S, Kitahara T, Horiike S, Ohwada T, Fujii K. Association of head trauma with cervical spine injury, spinal cord injury, or both. J Trauma 1999;3: Roth E, Davidoff G, Thomas P, et al. A controlled study of neuropsychological deficits in acute spinal cord injury patients. Paraplegia 1988;27: Davidoff GN, Roth EJ, Haughton JS, Ardner MS. Cognitive dysfunction in spinal cord injury patients: sensitivity of the functional independence measure. Arch Phys Med Rehabil 1990;71; 4: Richards JS, Osuna FJ, Jaworski TM, Novack TA, Leli DA, Boll TJ. The effectiveness of different methods of defining traumatic brain injury in predicting postdischarge adjustment in a spinal cord injury population. Arch Phys Med Rehabil 1991;72: Stutts M, Kreutzer JS, Barth JT, et al. Cognitive impairment in persons with recent spinal cord injury: finding and implications for clinical practice. Neurorehabilitation 1991;3: Davidoff G, Morris J, Roth E, Bleiberg J. Closed head injury in spinal cord injured patients: retrospective study of loss of consciousness and post-traumatic amnesia. Arch Phys Med Rehabil 1985;66: National Data and Statistical Center, Traumatic Brain Injury Model Systems. On-line syllabus. Available at: org/syllabus/live/live.asp. Accessed March 27, Williams DH, Levin HS, Eisenberg HM. Mild head injury classification. Neurosurgery 1990;27: Zafonte RD, Hammond FM, Mann NR, Wood DL, Black KL, Millis SR. Relationship between Glasgow coma scale and functional outcome. Am J Phys Med Rehabil 1996;75: Zafonte RD, Mann NR, Millis SR, Black KL, Wood DL, Hammond F. Posttraumatic amnesia: its relation to functional outcome. Arch Phys Med Rehabil 1997;78: Sherer M, Sander AM, Nick TG, High WM Jr, Malec JF, Rosenthal M. Early cognitive status and productivity outcome after traumatic brain injury: findings from the TBI model systems. Arch Phys Med Rehabil 2002;83: Novack TA, Bush BA, Meythaler JM, Canupp K. Outcome after traumatic brain injury: pathway analysis of contributions from premorbid, injury severity, and recovery variables. Arch Phys Med Rehabil 2001;82: Bush BA, Novack TA, Malec JF, Stringer AY, Millis SR, Madan A. Validation of a model for evaluating outcome after traumatic brain injury. Arch Phys Med Rehabil 2003;84: Brown AW, Malec JF, McClelland RL, Diehl NN, Englander J, Cifu DX. Clinical elements that predict outcome after traumatic brain injury: a prospective multicenter recursive partitioning (decision-tree) analysis. J Neurotrauma 2005;22: Russell WR, Smith A. Post-traumatic amnesia in closed head injury. Arch Neurol 1961;5: Jennett B, Snoek J, Bond MR, Brooks N. Disability after severe head injury: observations on use of the Glasgow Outcome Scale. J Neurol Neurosurg Psychiatry 1981;44: Ellenberg JH, Levin HS, Saydjari C. Posttraumatic amnesia as a predictor of outcome after severe closed head injury. Arch Neurol 1996;53: Supplier a. Version 15.0; SPSS, 233 S Wacker Dr, 11th Fl, Chicago, IL