Abstract
Offenders in justice system settings have high rates of traumatic brain injury (TBI) in comparison with the general population. Consequently, justice systems are using screening tools to identify and manage these individuals. Currently, that includes screening for TBI history and gross cognitive impairment. The present study attempted to determine whether the modified Ohio State University Traumatic Brain Injury Identification Method (OSU TBI-ID) was predictive of ongoing cognitive impairment as measured by the Automated Neuropsychological Assessment Metrics (ANAM) Core Battery. If so, the OSU TBI-ID could be used as a stand-alone measure of TBI history and impairment. This study had 223 participants (male = 160, female = 62). Sensitivity and specificity results revealed poor (.65) to very poor (.36) estimates for all OSU TBI-ID indices across all ANAM subtests. This study suggests that screening for lifetime history of TBI does not identify cognitive impairment. Implications for screening policy and future research are discussed.
Keywords
Among working age adults, traumatic brain injury (TBI) is a leading cause of death and permanent disability (Fleminger & Ponsford, 2005). In 2010, the Centers for Disease Control and Prevention (2016) estimated that TBIs accounted for approximately 2.5 million emergency department visits, hospitalizations, and deaths. However, the number of TBIs is likely higher, as this statistic does not account for persons who received care outside an emergency department (e.g., an outpatient provider), those who received care at a federal facility such as a Veteran’s hospital, or those who do not seek treatment.
Not only is TBI a concern among the general population, a growing body of research has identified that the prevalence of lifetime TBI history among incarcerated individuals is far greater than the general population. In 2014, approximately 1.51 million individuals were held in state and federal custody (Carson, 2014). Studies have found self-reported TBI history base rates to range from 65% to 96% across various correctional settings (Ferguson, Pickelsimer, Corrigan, Bogner, & Wald, 2012; Schofield et al., 2006; Slaughter, Fann, & Ehde, 2003; Williams et al., 2010). These studies ask inmates, “Have you ever had an injury to the head which knocked you out or left you dazed, confused, or disoriented?” In an initial study investigating the incidence of TBI in a county jail setting using the OSU TBI-ID, Gafford, McMillan, Gorgens, Dettmer, and Glover (2015) found the incidence of TBI among a sample of offenders with a comorbid mental illness to be 96%. These data are especially shocking in light of 2016 research (McKinlay, Horwood, & Fergusson, 2016) that showed that self-reports of TBI underrepresent the true frequency of injury. All told, the prevalence of TBI in correctional settings varies, but consistently exceeds the figures for the general population.
People with a history of TBI not only have more involvement with the criminal justice system, but they also more frequently utilize specialized services. They enter the custodial system earlier (age 16) than those without history of TBI (age 20; Williams et al., 2010). Furthermore, they also spend an average of seven more months in custody, and are more frequently in custody, relative to those without a TBI. Piccolino and Solberg (2014) suggested that persons with a history of TBI are more likely to use medical and psychological services while in custody. In addition, their results suggested that individuals with a history of TBI have higher rates of prison rule violations and lower chemical dependency treatment completion rates. In their study, they also demonstrate higher recidivism rates (51%) than those without a history of TBI (33%). As such, addressing criminogenic factors using the risk need responsivity (RNR) model may be warranted.
The RNR matches a person’s risk to the appropriate level of treatment based on static and dynamic risk factors (Andrews, Bonta, & Hoge, 1990). In short, those with higher risk are targeted with more intensive programming. Historically, intervention efforts have been focused on addressing dynamic risk factors, but more recently there have been efforts to expand our understanding of criminal behavior to include biological processes (Newsome & Cullen, 2017; Wright & Cullen, 2012). A history of TBI is a static, biological factor that conveys a greater risk for recidivism.
The long-term consequences associated with a TBI are reflected in that heightened risk. After a TBI, cognitive changes are among the most common complaints. There are several predictable areas of impairment including working memory (Christodoulou, 2001), attention, and processing speed (Olver, Ponsford, & Curran, 1996). Carlozzi, Kirsch, Kisala, and Tulsky (2015) reported that people with a history of TBI performed more poorly than matched controls on the Working Memory Index (WMI), Processing Speed Index (PSI), and Full Scale IQ (FSIQ) sections of the Weschler Adult Intelligence Scales, 4th Edition (WAIS-IV). Gafford et al. (2015) found that 92% of inmates with comorbid TBI and mental illness history also showed gross cognitive impairments on neuropsychological screening tests. All told, the high prevalence of TBI in the correctional system, together with the increased risk for cognitive impairment and poor correctional outcomes, indicates that it is imperative to screen for a history of TBI and related deficits in these settings.
The gold standard for the identification of TBI is a medical records review and a neuropsychological examination. Because this approach is not always possible, identifying a TBI history often relies on self-report (Williams et al., 2010). The best TBI screening tools correctly identify those who have the condition (sensitivity) and those who do not have the condition (specificity). A sensitive TBI screening measure correctly identifies those with a history of TBI, or those at risk for a TBI-related cognitive deficit and who are likely to have an increased need for services. A measure with good specificity correctly identifies those who do not have a history of TBI, or those who have no likelihood of TBI-related cognitive deficits and who qualify for standard intervention approaches.
Structured Interview About TBI History: The Ohio State University TBI Identification Method (OSU TBI-ID)
The OSU TBI-ID was developed to meet the need for a retrospective, systematic method to identify TBI in populations thought to be at risk for that injury and its associated complications (Corrigan & Bogner, 2007). It is a three-step interview that guides the interviewer to elicit key information about an interviewee’s history of injury. The interview begins with five items, which branch into additional query questions, allowing the interviewer to probe specific content if the interviewee answers yes to one of the prompts. Additional questions allow the interviewer to gather more detailed information on the cause of injury, age of injury, loss of consciousness (LOC), and memory gaps after an injury.
In addition to widespread clinical use, the OSU TBI-ID has also been validated in correctional settings. Bogner and Corrigan (2009) found that the OSU TBI-ID’s test–retest reliability ranges from acceptable to high (≥.60). That research includes five summary indices including (a) Worst: one moderate or severe TBI; (b) First: TBI with LOC before age 15; (c) Multiple: two or more TBIs close together, including a period of multiple blows to the head; (d) Recent: a mild TBI in the last 4 weeks, or a more severe TBI in the last months; and (e) Other Sources: any TBI combined with another way that brain function has been impaired (Bogner & Corrigan, 2009). Summary indices of the measures are described in the following paragraphs.
Age of First Injury
Age of first injury is important in predicting long-term consequences for TBI. Individuals with pediatric brain injury history have an even greater risk for long-term sequelae (Benz, Ritz, & Kiesow, 1999). For example, Bardoni et al. (2013) reported that children who sustained moderate-to-severe TBI during school age were also documented to have persistent postinjury cognitive, academic, and behavioral deficits. Further research shows that children who experience those deficits demonstrate gaps in skills development and age-appropriate social interactions (Bardoni et al., 2013). That research found that those impairments persist despite access to rehabilitative services. Among offenders with a TBI history, those with an early head injury tend to enter the custodial system during adolescence, which is earlier than offenders with no TBI history (Williams et al., 2010). In addition, Williams et al. (2010) reported that during a 5-year period, persons with an early TBI history spent more time in custody and were more likely to report reoffending, relative to those without early TBI histories (Bardoni et al., 2013).
Worst Injury
TBI severity has also been shown to predict functional outcomes. For example, moderate-to-severe injuries characterize the history of up to 70% of reported TBIs in prison populations (de Souza, 2003). Bazarian, Cernak, Noble-Haeusslein, Potolicchio, and Temkin (2009) reported that injury severity indices predicted functional outcome, and Rassovsky et al. (2015) reported that TBI severity predicted long-term cognitive and social outcomes.
Multiple Injuries
After a single mild TBI (MTBI), individuals are not likely to have long-term consequences, but multiple mild TBIs have been associated with neuropsychological dysfunction (Wall et al., 2006). Belanger, Spiegel, and Vanderploeg (2010) reported that multiple injuries also lead to increasing levels of impairment, and that people who have experienced multiple injuries often suffer from greater memory and executive functioning impairment relative to individuals who have experienced only a single injury. Multiple injuries are also related to significant impairments in social functioning and to behavioral problems not present before injury (Creeley, 2004).
Recent Injury
The OSU TBI-ID defines “Recent Injury” as “a mild TBI in the last weeks or a more severe TBI in the last months.”
Other Sources
“Other Sources” includes “Any TBI combined with another way that brain function has been impaired.” For example, alcohol abuse and mental illness are often comorbid with TBI, and Corrigan and Duetschle (2008) found that persons with substance use disorders and/or mental illnesses are also more likely to have a TBI history.
Neuropsychological Screening Tests: Automated Neuropsychological Assessment Metric
The gold standard for the identification of neuropsychological deficits is a comprehensive neuropsychological evaluation that, due to the expense and need for expertise, is not universally available. As a result, neuropsychological screening batteries are often used to gauge a person’s cognitive functioning following a TBI. The Automated Neuropsychological Assessment Metrics (ANAM; Reeves, Winter, Bleiberg, & Kane, 2007) is a commonly used neuropsychological screening measure.
The ANAM is a computer-based measure of cognitive performance originally developed by the Department of Defense as a “sensitive and repeatable measure of processing speed and cognitive efficiency” (Vincent, Roebuck-Spencer, Gilliland, & Schlegel, 2012, p. 256). It has been used to document impaired functioning in people with TBI (Vincent et al., 2012), takes approximately 20 to 30 min to administer, and measures performance across several functional domains (Reeves et al., 2007). While the ANAM was originally designed for repeated assessment of military personnel, it has also shown utility in a wide variety of populations, including TBI, multiple sclerosis, systemic lupus, Parkinson’s Disease, acquired brain injury, and Alzheimer’s Dementia (Kane, Roebuck-Spencer, Short, Kabat, & Wilken, 2007; Levinson, Reeves, Watson, & Harrison, 2005; Wild, Howieson, Webbe, Seelye, & Kaye, 2008).
The ANAM Core Battery has two subjective scales—the Sleep Scale (SLP) and Mood Scale (MOO)—followed by seven performance subtests. The performance subtests include (a) Simple Reaction Time (SRT)—a 25-trial test that measures attention using visual reaction time; (b) Code Substitution (CDS)—a 36-trial learning subtest consisting of nine symbols paired with digits; (c) Procedural Reaction Time (PRO)—a 32-trial test that assesses reaction time and processing efficiency with a series of single digits that the subject must judge as “low” or “high”; (d) Mathematical Processing (MTH)—a 20-trial item that assesses basic computational skills, concentration, and working memory by asking the subject to solve a series of basic math problems; (e) Matching to Sample (M2S)—a 20-trial test of visuospatial working memory in which examinees are asked to memorize a checkerboard matrix; (f) Code Substitution Delayed (CDD)—a 36-trial delayed memory subtest that serves as a delayed recall of the symbol/digit pairs learned approximately 20 min earlier on the CDS subtest; and (g) Go-no-Go (GNG)—a 120-trial test that assesses response inhibition and executive function by asking the subject to respond as quickly as possible to an “X” and do nothing when an “O” appears.
Cognitive performance is assessed as a composite of Mean Response Time Correct (MeanRTCorr), Accuracy (PercCorr), and Throughput percentiles on each subtest. Results are calculated relative to a general community reference group matched by age and gender strata and classified as follows: Average or Above (≥10th percentile), Below Average (9th-3rd percentile), and Clearly Below Average (2nd percentile or below). These classifications are consistent with acceptable clinical practice standards for classifying groups that would be of concern as compared with reference groups (Vista LifeSciences [VLS], 2014).
Studies have demonstrated the ANAM’s test–retest reliability over both short (daily to weekly) and long (6 months to a year) intervals (VLS, 2014). In addition, studies suggest strong convergent validity between ANAM subtests and several more commonly used and validated cognitive measures, including but not limited to the Hopkins Verbal Learning Test (HVLT), which assesses verbal learning and memory; the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS), which measures cognitive decline or improvement; Trail Making Test Parts A and B, which is used to assess visual attention and task-switching; and the Stroop Task, which assesses the capacity to direct attention (Kabat, Kane, Jefferson, & DiPino, 2001; Wilken et al., 2003; Woodhouse et al., 2013).
Current Study
This study was a partnership between the State of Colorado Brain Injury Program, multiple correctional and justice facilities, and the University of Denver. The research protocol was approved by the Institutional Review Board at the University of Denver. An Administration for Community Living (ACL) grant supports the efforts to identify TBI history in justice-involved populations, provide TBI training and education to staff across multiple disciplines, and facilitate a continuity of care into the community for the individuals identified. In this program, justice-involved persons with both a TBI history and a gross cognitive deficit are identified and provided with assessment feedback, modified therapies, and recommendations. They are also determined to be eligible for jail-based interventions such as self-advocacy education and a referral to community-based services. An interviewer-administered questionnaire (a modified version of the OSU TBI-ID) and a neuropsychological screening battery (ANAM Core Battery) were used to determine TBI history and gross cognitive deficit, respectively.
“Gross cognitive deficit” was defined by an ANAM score 2 or more standard deviations below the mean in any functional domain, or “clearly below average.” In addition, the OSU TBI-ID was modified to reflect only three criteria for significant self-reported TBI history. The “Recent Injury” index was removed because most “non-complicated” concussions or mild injuries will resolve within 3 weeks to 3 months (Dikmen, Machamer, & Temkin, 2001; Ettenhofer & Abeles, 2009) and because the severe injuries will be flagged by the “worst” criterion. Another limitation of the “Recent Injury” index is the absence of an operational definition for “last weeks” and “last months.” The “Other” index was designed to identify individuals whose brain function is impaired by any other mechanism than TBI, such as a serious mental illness or a substance abuse disorder (Corrigan & Bogner, 2007; Corrigan, Bogner, & Holloman, 2012). In a population predominantly composed largely of persons with comorbid mental illness and substance abuse, the sensitivity and specificity of this index to TBI are reduced and it was removed from consideration. The remaining indices—first, worst, and multiple TBIs—are each associated with cognitive deficits in the general population and among prisoners (Bogner & Corrigan, 2009).
Those three retained indices are operationally defined as follows: (a) First injury: individual reports TBI with LOC before age of 15, (b) Worst injury: individual reports one moderate to severe TBI (moderate and severe TBI are indicated by report of LOC greater than 30 min), and (c) Multiple injuries: individual reports a period where three or more blows to the head caused altered consciousness odds ratio (OR) two or more TBIs with LOC within a 3-month period.
Given the high prevalence rates of TBI history in correctional settings and the strong comorbidity between TBI history and cognitive impairment, the present study was developed to determine whether the three OSU TBI-ID indices of “First,” “Worst,” and “Multiple” were predictive of ongoing cognitive impairment as indicated by a score of “clearly below average” on any domain on the ANAM Core Battery. If the OSU TBI-ID acts as a proxy for cognitive function, that is, if OSU TBI-ID criteria predict cognitive impairment, then facilities could deploy only the one screening instrument to identify their highest risk inmates.
Method
Participants
The sample of 223 participants (M = 160, F = 62) consisted of English-speaking, consenting participants from programs in three county jails, juvenile probation, a Veteran’s Court, and a problem-solving court. Ages ranged from 15 to 73 years old (M = 38.4, SD = 11.8), and years of education ranged from 6 to 17 (M = 11.7, SD = 2.1). Race/ethnicity distribution was White/Western European (45.7%), Hispanic (22.0%), Black/African American (17.9%), more than one race (6.7%), American Indian/Alaska Native (4.0%), and Native Hawaiian/Pacific Islander (0.4%). Thirty-five participants (15.7%) were multilingual (one individual needed translation). Due to budget constraints and analysis brevity, the multilingual variable was used rather than native language. This is a limitation of the study. Table 1 displays summary statistics.
Sample
Note. n indicates number of “yes” responses. Overall N = 223.
Procedure
Correctional staff administered the OSU TBI-ID and subjects who were identified as having had a TBI were flagged to receive a secondary screening using a neuropsychological screening battery. Individuals who consented to participate in that secondary screening completed a clinical interview and the ANAM Core Battery.
Data Analyses
Study data were collected and managed using REDCap (Research Electronic Data Capture; Harris et al., 2009) electronic data capture tools hosted at the University of Denver. REDCap is a secure, web-based application designed to support data capture for research studies, providing (a) an intuitive interface for validated data entry, (b) audit trails for tracking data manipulation and export procedures, (c) automated export procedures for seamless data downloads to common statistical packages, and (d) procedures for importing data from external sources.
A total of 21 cross-tabulation contingency tables were constructed using SPSS 23. Proportions of true positives (TP), true negatives (TN), false positives (FP), and false negatives (FN) were calculated for each OSU TBI-ID summary index (Worst, First, Multiple) and ANAM subtests to assess sensitivity and specificity. Likelihood Ratios (LRs) for positive (+) and negative (–) screening results were then calculated using the following formulas in Excel, and confirmed with an automated Screening Validity Calculator: Sensitivity = TP / (TP + FN), Specificity = TN / (TN + FP), +LR = SEN / (1 – SPEC), –LR = (1 – SEN) / SPEC (LaMorte, 2015; Rigby & Summerton, 2005; Skopp et al., 2012).
Results
Sensitivity, Specificity, and LRs
Overall Results
The sensitivity, specificity, and LR values for the OSU TBI-ID Worst, First, and Multiple indices controlling for gender, age, and years of education are summarized in Table 2, Table 4, and Table 6, respectively. Interestingly, the overall estimates were identical for these three variables. Notably, all three index estimates were slightly different when controlling for the multiple languages variable as displayed in Table 3, Table 5, and Table 7. Gender disaggregated statistics also displayed for continuity and to facilitate visual comparison.
WORST Index Summary Statistics—Gender, Age, Education Controlled
Note. ANAM = Automated Neuropsychological Assessment Metrics; LR = likelihood ratio.
WORST Index Summary Statistics—Multiple Languages Controlled
Note. ANAM = Automated Neuropsychological Assessment Metrics; LR = likelihood ratio.
FIRST Index Summary Statistics—Gender, Age, Education Controlled
Note. ANAM = Automated Neuropsychological Assessment Metrics; LR = likelihood ratio.
FIRST Index Summary Statistics—Multiple Languages Controlled
Note. ANAM = Automated Neuropsychological Assessment Metrics; LR = likelihood ratio.
MULTIPLE Index Summary Statistics—Gender, Age, Education Controlled
Note. ANAM = Automated Neuropsychological Assessment Metrics; LR = likelihood ratio.
MULTIPLE Index Summary Statistics—Multiple Languages Controlled
Note. ANAM = Automated Neuropsychological Assessment Metrics; LR = likelihood ratio.
OSU TBI-ID WORST
“Worst” index overall sensitivity estimates revealed poor to very poor indication of both sensitivity (.65-.51) and specificity (.41-.36) for gender, age, and years of education (Table 2). Poor to very poor results were also indicated for the Multiple languages variable: sensitivity (.64-.50) and specificity (.40-.33) (Table 3).
OSU TBI-ID FIRST
For gender, age, and years of education, Table 4 shows “First” index sensitivity and specificity estimates ranged from .54 to .43, and from .61 to .56, respectively, with +LR and –LR estimates from 0.98 to 1.39, and 0.75 to 1.02, respectively. Results for the multiple languages variable ranged from .58 to .39 (very poor) for sensitivity, and .60 to .52 (poor to very poor) for specificity.
OSU TBI-ID MULTIPLE
“Multiple” index sensitivity estimates for gender, age, and years of education ranged from .56 to .36 (very poor), and specificity estimates ranged from .54 to .46 (very poor) overall (Table 6). Multiple languages sensitivity estimates ranged from .56 to .40 (very poor), and specificity statistics from .55 to .47 (very poor) for this index (Table 7).
LRs
LRs for all three OSU TBI-ID indices suggested that cases identified by each index (Worst, First, Mult) indicated it is highly unlikely that the indices would yield true positive or true negative results to correctly identify individuals with a cognitive deficit as similarly measured by the ANAM when controlling for all four demographic variables. Overall positive LRs ranged from 0.67 to 1.22 for gender, age, and years of education and from 0.75 to 1.42 for multiple languages, whereas overall negative LRs ranged from 0.71 to 1.38 for gender, age, and years of education and from 0.71 to 1.50 for persons who endorse being multilingual. Overall summary statistics are found in Table 3, Table 5, and Table 7 for multiple languages, and Table 2, Table 4, and Table 6 for gender, age, and years of education.
Gender Results
Disaggregated sensitivity, specificity, and LR results are reported below. Again, the overall estimates were identical when controlling for gender, age, and years of education, and all three index estimates were slightly different when the multiple language variable was controlled. For continuity, gender disaggregated summary statistics are displayed with the overall statistics in Table 2, Table 4, and Table 6 for gender, age, and years of education, and Table 3, Table 5, and Table 7 for multiple languages.
OSU TBI-ID WORST
Estimates indicated poor to very poor results for males (sensitivity = .65-.54, specificity = .43-.38), whereas female estimates indicated fair for MTH to very poor results across all other subtests (sensitivity = .79-.42, specificity = .48-.32) when controlling for gender, age, and years of education (Table 2). Very poor estimates for males (sensitivity = .65-.57, specificity = .44-.35) and fair for MTH only to poor and very poor for all other subtests for females (sensitivity = .73-.25, specificity = .50-.30) were revealed when controlling for multiple languages (Table 3).
OSU TBI-ID FIRST
Gender disaggregated statistics for the “First” index for gender, age, and years of education revealed very poor estimates for males (sensitivity = .58-.35, specificity = .57-.46), poor to very poor estimates for females (sensitivity = .62-.43), and fair for CDD to poor estimates for all other subtests (specificity = .74-.65). When the multiple languages variable was controlled, estimates were very poor for males (sensitivity = .57-.32, specificity = .56-.43), and fair for CDS, M2S, and CDD to very poor for all other subtests for females (sensitivity = .67-.44, specificity = .73-.64).
OSU TBI-ID MULTIPLE
Controlling for the gender, age, and years of education variables on the “Multiple” index revealed very poor estimates for males (sensitivity = .46-.30, specificity = .58-.53). Notably, estimates for females revealed quite good for M2S to fair for GNG to very poor estimates for all other subtests for sensitivity (.87-.46), and very poor estimates for specificity (.50-.30). Results revealed poor estimates for males (sensitivity = .50-.38, specificity = .56-.52) and good for M2S to fair for GNG to poor for all other subtests for females for the controlled multiple languages variable (sensitivity = .88-.44, specificity = .52-.36).
LRs
Summary statistics are displayed in Table 2, Table 4, and Table 6 for gender, age, and years of education, and in Table 3, Table 5, and Table 7 for multiple languages. Positive LRs for males ranged from 0.64 to 1.35 for gender, age, and years of education, and from 0.49 to 1.29 for multiple languages, whereas overall negative LRs for males ranged from 0.77 to 1.39 for gender, age, and years of education and from 0.82 to 1.56 for multiple languages. For females, positive LRs ranged from 0.62 to 2.31 for gender, age, and years of education, and from 0.36 to 2.44 for multiple languages, whereas overall negative LRs for females ranged from 0.27 to 1.81 for gender, age and years of education and from 0.24 to 2.46 for multiple languages. Four positive LRs were within the low acceptable range for females; for the gender, age, and years of education variables, one low acceptable LR was for CDS (+LR = 2.31), and one was for CDD (+LR = 2.17). The other two were for the multiple languages variable with one for CDS (+LR = 2.44), and one for M2S (+LR = 2.05).
Like the overall results, LRs for all three OSU TBI-ID indices indicated that it is highly unlikely that the indices would yield true positive or true negative results to correctly identify people with a cognitive deficit as similarly measured by the ANAM except for four subtest positive LRs for females alone for the “First” index only. The remaining 80 positive LRs ([2 genders] × [3 indices] × [7 subtests] − 4), and all negative LRs were within the highly unlikely range. Clearly, these statistics provide a strong threat for potential Type I and Type II error.
Cognitive Performance
Cognitive impairment was assessed using 2 standard deviations below the mean on the ANAM as the criterion for positive indication of impairment across the seven subtests. Results for positive cognitive impairment identification revealed a range from 22.5% (n = 36) to 40.0% (n = 64) for males, with a range from 21.0% (n = 13) to 37.1% (n = 23) for females, and an overall range from 25.1% (n = 62) to 38.1% (n = 85). Subtest results indicated relatively equal impairment across all subtests (25.1%-35.5%), except for CDD (M = 36.3%, F = 37.1%, overall = 36.8%) and M2S (M = 40.0%, F = 35.5%, overall = 38.1%). Table 8 displays cognitive performance statistics.
Cognitive Impairment
Note. Cognitive impairment = 2 SD below the mean, male n = 160, female n = 62. All data were self-reported. ANAM = Automated Neuropsychological Assessment Metrics. Overall N = 223.
TBI History
Positive TBI history was determined when one or more of the index conditions were met as operationally defined above. Results indicated positive TBI history for the following indices: Worst (M = 59.4%, F = 60.3%, overall = 59.5%), First (M = 46.9%, F = 37.1%, overall = 43.5%), and Multiple (M = 43.1%, F = 62.9% overall = 49.8%; Table 9).
Positive TBI History
Note. Positive TBI history indicated by any one or more conditions: (a) Worst-one moderate to severe TBI with LOC > 30 min, (b) First-TBI with LOC before age 15, and (c) Multiple—a period where ≥3 blows to the head caused altered consciousness or ≥2 TBIs with LOC within a 3-month period. Overall N = 223, male n = 160, female n = 62. All data self-reported via structured interview. OSU TBI-ID = Ohio State University Traumatic Brain Injury Identification Method; LOC = loss of consciousness.
Discussion
This study examined the sensitivity and specificity of a modified, three-index OSU TBI-ID (Corrigan & Bogner, 2007) to gross neuropsychological impairment as measured by “clearly below average” performance on any domain of the Automated Neuropsychological Assessment Metric (ANAM; Reeves et al., 2007). The results indicate that a lifetime history of TBI is not predictive of a cognitive impairment. Specifically, results show that no single OSU TBI-ID summary index is a good predictor for any single ANAM variable. Sensitivity and specificity findings for overall and gender disaggregated data ranged from poor to extremely poor for all three summary indices across all seven subtests for all demographic variables. This suggests that the OSU TBI-ID is not a good predictor for any ANAM variable. The OSU TBI-ID was designed, constructed, and validated to identify TBI. The ANAM evaluates cognitive impairment. Each instrument has a distinctly different purpose and the absence of OSU TBI-ID sensitivity to cognitive impairment adds to the validity of the instrument for its intended purpose.
Because a positive OSU TBI-ID cannot be used as a proxy for neuropsychological impairment, this study supports the addition of a neuropsychological screening measure (e.g., ANAM) to an interviewer-administered TBI history survey (e.g., OSU TBI-ID) in programs invested in identifying persons who have a lifetime TBI history and a cognitive impairment or those at greatest risk for poor correctional outcomes. Previous research indicates that the prevalence of lifetime TBI history is far greater among those in the correctional system than in the general population. According to those findings, the prevalence of lifetime TBI ranges from 65% to 96% across various correctional settings (Ferguson et al., 2012; Gafford et al., 2015; Schofield et al., 2006; Slaughter et al., 2003; Williams et al., 2010). The higher prevalence of TBI history suggests a greater rate of cognitive impairment; one study found it to be as high as 92% (Gafford et al., 2015), and that impairment confers an increased risk for reoffense. According to the RNR model, these offenders should be identified, assessed, and matched with higher intensity services.
Cognitive dysfunction plays a role in behavioral problems in correctional settings (Piccolino & Solberg, 2014; Williams et al., 2010). An enhanced understanding of an individual’s cognitive function can be used to mitigate behavior problems. Iverson, Demarest, and Franzen (1994) recommended the evaluation of specific neuropsychological strengths and weaknesses in the service of individualized treatment planning. Treatment and management strategies can be easily modified to maximize their impact. Research confirms that offenders who receive modified treatments show improved functional outcomes (Voorhis, Spruance, Ritchey, Listwan, & Seabrook, 2004). For example, previous studies found that therapeutic approaches which include adaptations for neuropsychological impairments are more effective in treating those with a history of TBI. Highly structured and repeated interventions like Cognitive Behavioral Therapy (CBT) are most effective (Soo & Tate, 2007). Ponsford et al. (2015) offered a modified CBT psychotherapy to people in the general population with TBI, depression, and anxiety. The modification included the addition of “booster” sessions to promote retention of information. According to their results, anxiety and depression were better alleviated by CBT with booster sessions than by CBT alone. These findings have also been replicated in offender populations. Specifically, Voorhis et al. (2004) found inmates who completed an individualized Cognitive Skills Program had lower recidivism rates than a comparison group. Importantly, the modified interventions appeared to be beneficial to both low- and high-risk offenders.
As to other simple modifications, Sohlberg, Mateer, and Stuss (1993) outlined specific accommodations to address each identified impairment. Environmental modifications such as posting visual cues, labeling cabinets, and providing organizational aids can help those with more severe executive functioning deficits. In correctional settings, these modifications may include posting schedules in easy-to-read formats and offering folders with clearly labeled sections to organize important documents.
In summary, neuropsychological screening is essential; screening for TBI history cannot be used as a proxy for cognitive impairment. By identifying those with TBI and cognitive deficits, rather than those with only a history of TBI, treatment can be modified with minimal or no additional cost to the institution. Screening for cognitive deficits in addition to a lifetime TBI history also identifies justice-involved individuals eligible for special programs (e.g., jail-based interventions such as those described above and referral to specialized community-based services). In that way, offender risk is better managed by matching them to appropriate interventions.
Limitations
The inherent limitations of screening measures are problematic and, in this study that problem is reflected in the over-inclusion (false positives) of offenders in special programming who do not have a true cognitive deficit which taxes already limited resources. Conversely, the false negatives, or under-inclusion of those who do have a true cognitive impairment, denies resources to those who could benefit from modified interventions. When the objective of a program is to match the appropriate, limited resource to each individual, the use of screening measures is inherently problematic.
There are also limitations in screening for TBI history. Self-report history surveys are not intended to replace a review of medical records or a full neuropsychological assessment. Still, in many settings where comprehensive medical records and testing are not available, a structured screening interview is a reasonable alternative. The OSU TBI-ID was developed to minimize the limitations typically associated with self-report, such as bias and faulty recall (Bogner & Corrigan, 2009). According to Corrigan et al. (2012), “a face-to-face interview conducted by an informed professional may be the gold standard one has to accept for determining lifetime history of TBI in most individuals” (p. 141).
Future Research
Overall, research on the identification and more effective management of justice-involved individuals with TBI and neurocognitive deficits is in its infancy. A true cost-benefit analysis would be a particularly useful initial step. As is the case when implementing any new program, identifying and managing justice-involved people with TBI and cognitive deficits will incur costs in advance of benefit. The long-term economic benefits of such programs, however, for overall resource use, reduced risk to correctional staff, and recidivism have yet to be calculated. Every indication is that identifying cognitive deficits and modifying behavioral health treatments with consideration of gross cognitive impairment reduces problem behaviors and improves outcomes in the criminal justice system and the community.
Footnotes
Acknowledgements
A special thank you to Vista Lifesciences for the use of their tool, the ANAM, and John Corrigan, the creator of the OSU TBI-ID, for his participation. Dr. Kim Gorgens is the principal investigator. She is a full-time clinical professor in the Graduate School of Professional Psychology at the University of Denver.
This study was funded by grants from the Health Resources Services Administration/Activities of Community Living and Colorado Office of Behavioral Health.
