Journal of Clinical and Experimental Neuropsychology
ISSN: 1380-3395 (Print) 1744-411X (Online) Journal homepage: http://www.tandfonline.com/loi/ncen20
An index predictive of cognitive outcome in retired professional American Football players with a history of sports concussion Mathew J. Wright, Ellen Woo, J. Brandon Birath, Craig A. Siders, Daniel F. Kelly, Christina Wang, Ronald Swerdloff, Elizabeth Romero, Claudia Kernan, Robert C. Cantu & Kevin Guskiewicz To cite this article: Mathew J. Wright, Ellen Woo, J. Brandon Birath, Craig A. Siders, Daniel F. Kelly, Christina Wang, Ronald Swerdloff, Elizabeth Romero, Claudia Kernan, Robert C. Cantu & Kevin Guskiewicz (2016): An index predictive of cognitive outcome in retired professional American Football players with a history of sports concussion, Journal of Clinical and Experimental Neuropsychology, DOI: 10.1080/13803395.2016.1139057 To link to this article: http://dx.doi.org/10.1080/13803395.2016.1139057
Published online: 22 Feb 2016.
Submit your article to this journal
Article views: 21
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ncen20 Download by: [RMIT University]
Date: 29 February 2016, At: 00:57
JOURNAL OF CLINICAL AND EXPERIMENTAL NEUROPSYCHOLOGY, 2016 http://dx.doi.org/10.1080/13803395.2016.1139057
An index predictive of cognitive outcome in retired professional American Football players with a history of sports concussion Mathew J. Wrighta,b,c, Ellen Wood, J. Brandon Biratha,c, Craig A. Sidersa,c, Daniel F. Kellye, Christina Wanga,f, Ronald Swerdloffa,f, Elizabeth Romeroa, Claudia Kernanb, Robert C. Cantug and Kevin Guskiewiczh
Downloaded by [RMIT University] at 00:57 29 February 2016
a Los Angeles Biomedical Research Institute at Harbor–UCLA Medical Center, Torrance, CA, USA; bDepartment of Psychiatry and Biobehavioral Sciences, UCLA School of Medicine, Los Angeles, CA, USA; cDivision of Psychology, Department of Psychiatry, Harbor–UCLA Medical Center, Torrance, CA, USA; dEaston Center for Alzheimer’s Disease Research, Department of Neurology, UCLA School of Medicine, Los Angeles, CA, USA; eBrain Tumor Center & Pituitary Disorders Program, John Wayne Cancer Institute at Saint John’s Health Center, Santa Monica, CA, USA; fDivision of Endocrinology, Department of Medicine, Harbor–UCLA Medical Center, Torrance, CA, USA; gBoston University School of Medicine, Boston, MA, USA; hCenter for the Study of Retired Athletes of the University of North Carolina, Chapel Hill, NC, USA
ABSTRACT
ARTICLE HISTORY
Objective: Various concussion characteristics and personal factors are associated with cognitive recovery in athletes. We developed an index based on concussion frequency, severity, and timeframe, as well as cognitive reserve (CR), and we assessed its predictive power regarding cognitive ability in retired professional football players. Method: Data from 40 retired professional American football players were used in the current study. On average, participants had been retired from football for 20 years. Current neuropsychological performances, indicators of CR, concussion history, and play data were used to create an index for predicting cognitive outcome. Results: The sample displayed a range of concussions, concussion severities, seasons played, CR, and cognitive ability. Many of the participants demonstrated cognitive deficits. The index strongly predicted global cognitive ability (R2 = .31). The index also predicted the number of areas of neuropsychological deficit, which varied as a function of the deficit classification system used (Heaton: R2 = .15; Wechsler: R2 = .28). Conclusions: The current study demonstrated that a unique combination of CR, sports concussion, and game-related data can predict cognitive outcomes in participants who had been retired from professional American football for an average of 20 years. Such indices may prove to be useful for clinical decision making and research.
Received 29 April 2015 Accepted 2 January 2016
Sports-related injuries are the second most common cause of traumatic brain injury (TBI) after motor vehicle accidents (Zilmer & Spiers, 2001). It is estimated that between 1.6 million and 3.8 million sports-related TBIs occur yearly in the United States (Langlois, Rutland-Brown, & Wald, 2006; McCrea, Hammeke, Olsen, Leo, & Guskiewicz, 2004), although data from the Centers for Disease Control and Prevention suggest that not all persons suffering from sport-related TBIs present for medical care; 248,418 emergency room visits for sports-related concussion were reordered in 2009 (Gilchrist, Thomas, Xu, McGuire, & Coronado, 2011). Young males (10–19 years of age) tend to CONTACT Matthew J. Wright 90502, USA. © 2016 Taylor & Francis
[email protected]
KEYWORDS
Cerebral concussion; Traumatic brain injury; Cognition; Sports; Cognitive reserve
have the highest rates of sports concussion (Gilchrist et al., 2011). High school and college American football players have an annual risk of concussion of 5–10%; the risk is greatest during game play in contrast to practice (Macciocchi, 2006). A large prospective study of collegiate football players found that 6.5% of concussed athletes had a repeat concussion within the same season, and players with a history of three or more concussions were three times more likely to experience additional concussions (Guskiewicz et al., 2003). Concussions are associated with acute neuropsychological impairments (Collie et al., 2006). Typically, acute neuropsychological impairments
Harbor–UCLA Medical Center, 1124 W. Carson St. B-4 South (Box 490), Torrance
Downloaded by [RMIT University] at 00:57 29 February 2016
2
M. J. WRIGHT ET AL.
associated with concussion resolve within a week or so (McCrea et al., 2003). That said, multiple concussions can contribute to greater impairment and longer recovery times (McCrory et al., 2013). Athletes with three or more concussions scored lower on memory tests two days post injury and were 7.7 times more likely to experience major memory decline. Moreover, retired athletes with greater concussion histories tend to show chronic cognitive and neurofunctional declines (De Beaumont, Beauchemin, Beaulieu, & Jolicoeur, 2013; De Beaumont et al., 2009; De Beaumont, Tremblay, et al., 2013). Professional football players with histories of three or more concussions have a fivefold higher prevalence of mild cognitive impairment (Guskiewicz et al., 2005). Thus, multiple concussions have been shown to have cumulative acute and chronic effects on cognitive abilities. The overt cognitive and behavioral signs of concussion typically resolve within one week (McCrea et al., 2003), and the current return to play guidelines are reflective of this timeline (McCrory et al., 2013). However, electrophysiological data suggest that the brain may be in recovery from sports-related concussion for up to 45 days or longer post injury (Prichep, McCrea, Barr, Powell, & Chabot, 2013). Moreover, data suggest that repeat concussions that are relatively close together in time have a greater impact on the brain and cognitive ability (Giza & Hovda, 2001; Laurer et al., 2001). Thus, concussed players who successfully complete the recommended return to play protocol may still be in the process of neurological recovery and may be more vulnerable to subsequent head injuries. Therefore, time, especially between concussive events or concussive and nonconcussive head injuries, may be an important factor in outcome from sports-related concussion. Additionally, not all sport-related concussions have an equal impact. Research suggests that athletes with concussions that resulted in posttraumatic amnesia (PTA) and retrograde amnesia, but not those that were associated with loss of consciousness or disorientation alone, demonstrated poorer recovery two days post injury on computer-administered tests of cognition (Collins et al., 2003). Thus, the literature suggests that the impact of sports-related concussion on cognitive ability is related to aspects of concussion severity, frequency, and the timeframe during which the concussions occurred.
Moreover, while a host of player factors likely influence concussion outcomes, cognitive reserve (CR) and brain reserve (BR), protective mechanisms often cited in the neuropsychological literature (Bigler & Stern, 2015; Satz, Cole, Hardy, & Rassovsky, 2011; Stern, 2002), likely have a great influence over the cognitive impact of athletes’ concussive histories. CR refers to a threshold for the emergence of cognitive decline related to neuropathology. This threshold is set by multiple mentally enriching experiences and abilities, such that CR acts as a protective mechanism against brain injury or disease (Bigler & Stern, 2015; Stern, 2002). Indeed, socioeconomic status (occupation and income), premorbid intelligence, and enriching leisure activities (e.g., writing, socializing, and exercise) independently contribute to CR in moderate-to-severe TBI survivors (Levi, Rassovsky, Agranov, Sela-Kaufman, & Vakil, 2013; Rassovsky et al., 2015). Enriching experiences and abilities are thought to mediate CR through the establishment of more efficient neural networks and, by some accounts, the ability to recruit new, less injured, networks for task completion under conditions of increased task difficulty. Neuroimaging data from healthy older adults and individuals with Alzheimer’s disease support these assertions (Bigler & Stern, 2015; Stern et al., 2000). Like CR, BR also increases the threshold for neuropathology to result in cognitive impairment (Bigler & Stern, 2015; Satz et al., 2011). Simply put, greater brain size is related to better cognitive function following brain injury (Bigler & Stern, 2015). It is likely that CR and BR are interrelated to some degree (Satz et al., 2011). It seems reasonable to assume that CR and BR influence the cognitive signature of sports concussion. A recent meta-analysis provided evidence for the protective influence of CR in sportrelated concussion, as athletes with more years of education demonstrated better neuropsychological outcomes (Dougan, Horswill, & Geffen, 2014). In the current study, our primary aim was to determine whether concussion frequency, severity, and timeframe, as well as CR, could be used to create an index that would predict cognitive outcomes in retired American professional football players. It was hypothesized that a combination of these factors would predict cognitive ability better than concussion frequency alone. Other factors can also have an impact on the outcomes of professional football players. Sports-related concussions have been linked to depression in retired
Downloaded by [RMIT University] at 00:57 29 February 2016
JOURNAL OF CLINICAL AND EXPERIMENTAL NEUROPSYCHOLOGY
professional football players (Chen, Johnston, Petrides, & Ptito, 2008; Guskiewicz et al., 2007). Additionally, obesity, which is often linked to metabolic syndrome, is associated with cognitive decline and higher mortality rates in retired football players (Baron, Hein, Lehman, & Gersic, 2012; Miller et al., 2008; Willeumier, Taylor, & Amen, 2012). Finally, anabolic–androgenic steroid use has been linked to increased musculoskeletal injuries in retired professional football players (Horn, Gregory, & Guskiewicz, 2009). Thus, as a secondary aim, we explored the relationship between mood ratings, body mass index (BMI), and self-reported steroid use as they related to concussion, CR, and cognitive ability, as any or all of these factors could potentially act as confounding variables that might account for potential associations between the proposed index and cognitive outcome.
Method The current study was approved by the institutional review boards of the Los Angeles Biomedical Research Institute at Harbor–UCLA Medical Center, the John Wayne Cancer Institute, and the University of North Carolina at Chapel Hill. All participants consented to voluntary participation. Participants Participants in the current analysis were recruited for a project examining rates of hormonal dysfunction in retired professional football players from the Center for the Study of Retired Athletes (CSRA) in collaboration with the National Football League (NFL) Players’ Association (see Kelly et al., 2014). From the CSRA database, 430 retired NFL players who played at least one year in the NFL and who completed the CSRA General Health Questionnaire (GHQ; Horn et al., 2009; Kerr, Marshall, & Guskiewicz, 2012), the Short Form 36 (Brazier et al., 1992) with a Mental Component Score of .30, ps ≤ .05). Scores for each domain were averaged prior to the averaging of all of the domain scores to produce a global cognition T-score: {[(CVLT–II
Downloaded by [RMIT University] at 00:57 29 February 2016
JOURNAL OF CLINICAL AND EXPERIMENTAL NEUROPSYCHOLOGY
learning trial recall + long-delay free recall + recognition discriminability)/3] + [(ROFT immediate recall + ROCFT delayed recall)/2] + [(SDMT oral trial + written trial + TMT–A)/3] + [(F–A–S + Animals + TMT–B)/3]}/4. We did not combine measures of verbal and nonverbal memory, as these test scores were generally uncorrelated (rs = .23 to .30, ps > .05), although CVLT–II recognition discriminability did correlate with delayed recall on the ROFT (r = .34, p = .03). Given our sample size, we opted to use the global cognition score as our primary outcome variable, rather than using individual cognitive domain scores. In addition to the neuropsychological assessment, we collected data regarding demographics, concussion history, game play, and steroid use via questionnaires and interviews, as well as mood, via the Hospital Anxiety and Depression Scale (HADS; Zigmond & Snaith, 1983), and obesity, via the BMI at the time of testing. Historical data regarding concussions and other TBIs were collected via interview questions based on the OSU TBI method (Corrigan & Bogner, 2007) and the PTA-Q (McMillan et al., 1996). Both the OSU TBI method and the PTA-Q have been shown to be reliable and valid methods for assessing remote TBI. Also, all concussion data for the current study were restricted to the period that participants played professional football, as the TBI reports for this period were most reliable and verifiable; we verified TBI reports against previous reports on the GHQ. That said, some might question the ability of retired professional football players to recall instances of concussion that occurred many years ago (e.g., 20 years ago), especially since there appeared to be a culture of downplaying concussive impacts in professional football and only a relatively recent widespread recognition of the long-term consequences of sports-related concussions. However, concussive impacts in athletes have been considered important injuries (e.g., Daly, 1958; Hugenholtz & Richard, 1982), recognized to cause marked cognitive changes (e.g., Gronwall & Wrightson, 1975; Lynch & Yarnell, 1973), and thought to require specialized treatment for more than 20 years (Daly, 1958; Hugenholtz & Richard, 1982). And while these data may have been somewhat restricted to researchers and healthcare workers, we doubt that the acute effects of concussion were lost on our participants, as the sequelae of concussions very likely affected their
5
athletic (and other) performances; regardless of the culture around concussion reporting, athletic performances were surely at the forefront of the minds of individuals involved in professional football (i.e., players, coaches, etc.). Fairly recent data seem to corroborate this position, as retired professional football players showed moderately reliable reports of concussion over a nine-year period (Кw= 0.48; see Kerr et al., 2012). CR was based on participants’ AMNART-estimated verbal intelligence, reports of occupational attainment and academic achievement (see details below). Also, we utilized highest degree obtained rather than years of education as these variables were strongly, but not perfectly, correlated (r = .77, p < .001), and we reasoned that degree completion was more likely reflective of mastery of academic material than years of exposure to academic material.
Statistical analyses Hierarchal linear regression was employed to determine whether CR and concussion data (frequency, severity, and period of concussive risk) could predict cognitive ability in our sample. Specifically, an index based on the following equation was evaluated to determine whether it could predict global cognitive ability in a sample of retired professional American football players above and beyond mere concussion frequency or concussion dose [(ΣS + Σf)/T] (see below). Cognitive Outcome from Sports Concussion Index ðCOSCIÞ ¼ CR ½ðS þ f Þ=T where CR (cognitive reserve) was represented by the sum of the centered values for highest degree obtained (1 = high school, 2 = associate’s degree, 3 = bachelor’s degree, 4 = master’s degree, 5 = doctoral degree or greater), highest occupational obtainment (1 = unemployed, 2 = laborer/unskilled, 3 = operator/domestic worker, 4 = craftsman/foreman, 5 = manager/administrative/clerical/sales, 6 = professional/technical; this coding was adapted from a method for estimating intelligence from demographic variables; Scott, Krull, Williamson, Adams, & Iverson, 1997), and premorbid estimated intelligence (based on the AMNART; Grober & Sliwinski, 1991, Verbal IQ estimate); ΣS (concussion severity) was represented by the sum of PTA ratings
Downloaded by [RMIT University] at 00:57 29 February 2016
6
M. J. WRIGHT ET AL.
(0 = none, 1 = less than 30 min, 2 = 30 min to 24 hours, 3 = over 24 hours)1 for concussions experienced during professional football; Σf (concussion frequency) was represented by the number of concussions experienced during professional football; and T (time period of concussive risk) was represented by the number of professional seasons played. ΣS and Σf were summed in order to reflect the magnitude of the concussive experience, as previous data have suggested that concussions with PTA appear to result in worse cognitive outcomes than concussions without PTA (Collins et al., 2003). The sum of ΣS and Σf was divided by T to reflect the relative dosing of concussion, as repeated concussions occurring closer together were thought to have a greater impact on brain and cognitive function (Giza & Hovda, 2001; Laurer et al., 2001). Finally, this ratio was subtracted from CR, as CR has been shown to be protective against brain injury and disease (Stern, 2002). In general, our CR values tended to be greater than concussion dose values. Thus, higher index scores should be related to better cognitive outcomes (see Appendix for additional detail). We also evaluated the COSCI’s ability to predict the number of neuropsychological domains with at least one deficit score on two commonly used deficit classification systems (Heaton: T < 40; Wechsler: T < 30; see Brooks, Sherman, Iverson, Slick, & Strauss, 2011) with simple linear regressions. Finally, we explored the relationship between mood ratings, BMI, and reported steroid use and CR, concussion history, and cognitive ability with Pearson correlations.
Results Depressive and anxious symptoms (as assessed by the HADS; Anxiety score: M = 8.35, SD = 4.48; 1
Depression score: M = 5.75, SD = 3.89) along with reported frequency of steroid use (M = .24, SD = .45) did not correlate with global cognition, the COSCI, or the CR and concussion dose aspects of the COSCI (rs = –.26 to .04, ps = .11 to .99). That said, BMI scores (M = 33.19, SD = 5.36) were significantly associated with the COSCI (r = –.43, p = .006) and the CR component of the COSCI (r= –.41 p = .01), but not concussion dose or global cognition (rs = –.20 to .04, ps = .16 to .81). Since self-reported steroid use, depression, anxiety, and BMI scores did not correlate with global cognition, they were not included in our analysis of the index. It should be noted that the study sample demonstrated a range of concussions, PTA ratings, seasons played, years of education, and cognitive ability (see Table 1). With regard to cognition, our sample displayed deficits in attention and processing speed (Heaton system: 20.0%; Wechsler system: 5.0%),2 verbal memory (Heaton system: 22.5%; Wechsler system: 7.5%), nonverbal memory (Heaton system: 37.5%; Wechsler system: 2.5%), and executive ability (Heaton system: 37.5%; Wechsler system: 20.0%). Also, a number of our participants displayed deficit scores in two or more cognitive domains (Heaton system: 35.0%; Wechsler system: 12.5%). As can be seen in Table 2, professional football concussion frequency (Σf) did not significantly predict global cognitive ability, nor did concussive dose [(ΣS + Σf)/T], although concussive dose improved the variance accounted for by 9%. The inclusion of cognitive reserve yielded significant results and enhanced the ability of the index to predict global cognition, accounting for an additional 25% of the variance; the total model accounted for 34% of the
Average PTA duration was also considered, but it did not correlate with neuropsychological outcome as strongly as the sum of PTA durations. Also, loss of consciousness (LOC) was considered (LOC average and LOC sum), but it was not significantly correlated with neuropsychological outcome. 2 While the Heaton and Wechsler deficit classification systems (deficits scores < 1 SD and < 2 SDs, respectively) are commonly used in clinical neuropsychology, the Wechsler is the more common of the two (Brooks et al., 2011). When choosing a classification system, it is important to consider normal score variability, level of education, and intellectual functioning (Iverson & Brooks, 2011). Specifically, and beyond the impact of age and sex, healthy individuals commonly obtain one or more low test scores when administered a neuropsychological test battery, and the number of low test scores varies as a function of the classification system used (lower thresholds for deficit result in more low scores), the number of tests administered (more tests will result in a greater number of low scores), education (higher levels of education are inversely related to the number of low scores), and intelligence (greater intelligence is related to fewer low scores; see Iverson & Brooks, 2011). We chose to display sample performances using both systems as our participants displayed a broad range of years of education (15–20 years, M = 16.35, SD = 1.21) and premorbid intelligence estimates (91.11–126.46, M = 113.72, SD = 9.21).
JOURNAL OF CLINICAL AND EXPERIMENTAL NEUROPSYCHOLOGY
Table 2. Regression models for the prediction of global cognition by concussion frequency, concussion dose, and the Cognitive Outcome from Sports-Concussion Index. Model 1 Concussion frequency Value
Downloaded by [RMIT University] at 00:57 29 February 2016
R2 Adjusted R2 F
Model 2 Concussion dose
Model 3 COSCI
β B SE B β B SE B β B 0.06 0.08 0.21 –0.40 –0.59 0.32 0.52 0.27 .00 .09 .34 –.02 .04 .28 0.14
1.75
SE B 0.07
6.15*
Note. Table displays regression models for the prediction of global cognitive ability by concussion data and an index including concussion and cognitive reserve data. All concussion and game play data included in the regressions were restricted to professional football. COSCI = Cognitive Outcome from Sports-Concussion Index. Model 1 included the number of concussions sustained as the predictor. Model 2 added concussion dose [the sum of posttraumatic amnesia (PTA, a measure of severity) and concussion frequency divided by seasons of play; see Method section] as a predictor. Model 3 added the complete COSCI as a predictor. The threshold for statistical significance was set at p < .05. *p = .002.
variance in the prediction of global cognitive ability. With the addition of CR, beta became positive, reflecting the protective nature of CR with regard to global cognitive ability. It should be noted that loss of consciousness was also considered as a predictor, but it did not contribute to the prediction of cognitive outcome. We followed up with a simple linear regression for the prediction of global cognition by the COSCI. The COSCI significantly predicted global cognition, β = 0.56, t(39) = 4.17, p < .001, and accounted for 31% of the variance, R2 = .31, F(1, 39) = 17.38, p < .001. Additionally, the COSCI significantly predicted the number of cognitive domains with one or more deficit scores [Heaton system: β = –0.39, t(39) = –2.57, p = .014, R2 = .15, F(1, 39) = 6.63, p = .014; Wechsler system: β = – 0.53, t(39)= –3.82, p < .001, R2= .28, F(1, 39) = 14.60, p < .001].
Discussion The current data support the notion that protective (e.g., CR) and harmful (e.g., elements of the concussive history) factors can be considered together to create an index that is predictive of cognitive outcomes of athletes with concussive histories. Specifically, the COSCI, an index based on the difference between CR (represented by the sum of centered values for highest degree obtained, highest occupational obtainment, and estimated premorbid intellectual ability) and the
7
ratio of the sum of concussion frequency and severity (indexed by the sum of PTA durations) to the time period of concussive risk (indexed by seasons played) predicted global cognitive ability in retired professional football players, accounting for 31% of the variance. Additionally, COSCI also predicted the number of neuropsychological domains with deficit scores, accounting for 15–28% of the variance, depending on the classification system used. These findings underscore the importance of considering player-specific factors when attempting to predict long-term cognitive outcomes related to sports concussion. Moreover, our data suggest that both protective factors, such as CR (the increased threshold for the emergence of cognitive decline set by mental abilities and cognitively enriching experiences), and harmful factors such as various concussion variables (concussion frequency, severity, and the time period of concussive risk) should be considered in predictions of cognitive outcome. It should be noted that we excluded participants with significant performance validity failure, histories of brain injury or disease (other than sports-related concussion), and complicated TBI related to football play. Additionally, our sample had been retired for an average of 20 years, suggesting that the index was fairly robust to the passage of time. Our findings open the door to the potential use of player-specific indices suggestive of cognitive outcome to aid in return-to-play decisions and planning of potential healthcare needs (e.g., comprehensive neuropsychological assessment, neuroimaging, cognitive rehabilitation, etc.) of athletes in contact sports (see Appendix). That said, it would not be accurate to assume that all components of the index held the same weight in predicting cognitive outcome, as concussive dose accounted for more variance (9%) than concussion frequency alone (
ISSN: 1380-3395 (Print) 1744-411X (Online) Journal homepage: http://www.tandfonline.com/loi/ncen20
An index predictive of cognitive outcome in retired professional American Football players with a history of sports concussion Mathew J. Wright, Ellen Woo, J. Brandon Birath, Craig A. Siders, Daniel F. Kelly, Christina Wang, Ronald Swerdloff, Elizabeth Romero, Claudia Kernan, Robert C. Cantu & Kevin Guskiewicz To cite this article: Mathew J. Wright, Ellen Woo, J. Brandon Birath, Craig A. Siders, Daniel F. Kelly, Christina Wang, Ronald Swerdloff, Elizabeth Romero, Claudia Kernan, Robert C. Cantu & Kevin Guskiewicz (2016): An index predictive of cognitive outcome in retired professional American Football players with a history of sports concussion, Journal of Clinical and Experimental Neuropsychology, DOI: 10.1080/13803395.2016.1139057 To link to this article: http://dx.doi.org/10.1080/13803395.2016.1139057
Published online: 22 Feb 2016.
Submit your article to this journal
Article views: 21
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ncen20 Download by: [RMIT University]
Date: 29 February 2016, At: 00:57
JOURNAL OF CLINICAL AND EXPERIMENTAL NEUROPSYCHOLOGY, 2016 http://dx.doi.org/10.1080/13803395.2016.1139057
An index predictive of cognitive outcome in retired professional American Football players with a history of sports concussion Mathew J. Wrighta,b,c, Ellen Wood, J. Brandon Biratha,c, Craig A. Sidersa,c, Daniel F. Kellye, Christina Wanga,f, Ronald Swerdloffa,f, Elizabeth Romeroa, Claudia Kernanb, Robert C. Cantug and Kevin Guskiewiczh
Downloaded by [RMIT University] at 00:57 29 February 2016
a Los Angeles Biomedical Research Institute at Harbor–UCLA Medical Center, Torrance, CA, USA; bDepartment of Psychiatry and Biobehavioral Sciences, UCLA School of Medicine, Los Angeles, CA, USA; cDivision of Psychology, Department of Psychiatry, Harbor–UCLA Medical Center, Torrance, CA, USA; dEaston Center for Alzheimer’s Disease Research, Department of Neurology, UCLA School of Medicine, Los Angeles, CA, USA; eBrain Tumor Center & Pituitary Disorders Program, John Wayne Cancer Institute at Saint John’s Health Center, Santa Monica, CA, USA; fDivision of Endocrinology, Department of Medicine, Harbor–UCLA Medical Center, Torrance, CA, USA; gBoston University School of Medicine, Boston, MA, USA; hCenter for the Study of Retired Athletes of the University of North Carolina, Chapel Hill, NC, USA
ABSTRACT
ARTICLE HISTORY
Objective: Various concussion characteristics and personal factors are associated with cognitive recovery in athletes. We developed an index based on concussion frequency, severity, and timeframe, as well as cognitive reserve (CR), and we assessed its predictive power regarding cognitive ability in retired professional football players. Method: Data from 40 retired professional American football players were used in the current study. On average, participants had been retired from football for 20 years. Current neuropsychological performances, indicators of CR, concussion history, and play data were used to create an index for predicting cognitive outcome. Results: The sample displayed a range of concussions, concussion severities, seasons played, CR, and cognitive ability. Many of the participants demonstrated cognitive deficits. The index strongly predicted global cognitive ability (R2 = .31). The index also predicted the number of areas of neuropsychological deficit, which varied as a function of the deficit classification system used (Heaton: R2 = .15; Wechsler: R2 = .28). Conclusions: The current study demonstrated that a unique combination of CR, sports concussion, and game-related data can predict cognitive outcomes in participants who had been retired from professional American football for an average of 20 years. Such indices may prove to be useful for clinical decision making and research.
Received 29 April 2015 Accepted 2 January 2016
Sports-related injuries are the second most common cause of traumatic brain injury (TBI) after motor vehicle accidents (Zilmer & Spiers, 2001). It is estimated that between 1.6 million and 3.8 million sports-related TBIs occur yearly in the United States (Langlois, Rutland-Brown, & Wald, 2006; McCrea, Hammeke, Olsen, Leo, & Guskiewicz, 2004), although data from the Centers for Disease Control and Prevention suggest that not all persons suffering from sport-related TBIs present for medical care; 248,418 emergency room visits for sports-related concussion were reordered in 2009 (Gilchrist, Thomas, Xu, McGuire, & Coronado, 2011). Young males (10–19 years of age) tend to CONTACT Matthew J. Wright 90502, USA. © 2016 Taylor & Francis
[email protected]
KEYWORDS
Cerebral concussion; Traumatic brain injury; Cognition; Sports; Cognitive reserve
have the highest rates of sports concussion (Gilchrist et al., 2011). High school and college American football players have an annual risk of concussion of 5–10%; the risk is greatest during game play in contrast to practice (Macciocchi, 2006). A large prospective study of collegiate football players found that 6.5% of concussed athletes had a repeat concussion within the same season, and players with a history of three or more concussions were three times more likely to experience additional concussions (Guskiewicz et al., 2003). Concussions are associated with acute neuropsychological impairments (Collie et al., 2006). Typically, acute neuropsychological impairments
Harbor–UCLA Medical Center, 1124 W. Carson St. B-4 South (Box 490), Torrance
Downloaded by [RMIT University] at 00:57 29 February 2016
2
M. J. WRIGHT ET AL.
associated with concussion resolve within a week or so (McCrea et al., 2003). That said, multiple concussions can contribute to greater impairment and longer recovery times (McCrory et al., 2013). Athletes with three or more concussions scored lower on memory tests two days post injury and were 7.7 times more likely to experience major memory decline. Moreover, retired athletes with greater concussion histories tend to show chronic cognitive and neurofunctional declines (De Beaumont, Beauchemin, Beaulieu, & Jolicoeur, 2013; De Beaumont et al., 2009; De Beaumont, Tremblay, et al., 2013). Professional football players with histories of three or more concussions have a fivefold higher prevalence of mild cognitive impairment (Guskiewicz et al., 2005). Thus, multiple concussions have been shown to have cumulative acute and chronic effects on cognitive abilities. The overt cognitive and behavioral signs of concussion typically resolve within one week (McCrea et al., 2003), and the current return to play guidelines are reflective of this timeline (McCrory et al., 2013). However, electrophysiological data suggest that the brain may be in recovery from sports-related concussion for up to 45 days or longer post injury (Prichep, McCrea, Barr, Powell, & Chabot, 2013). Moreover, data suggest that repeat concussions that are relatively close together in time have a greater impact on the brain and cognitive ability (Giza & Hovda, 2001; Laurer et al., 2001). Thus, concussed players who successfully complete the recommended return to play protocol may still be in the process of neurological recovery and may be more vulnerable to subsequent head injuries. Therefore, time, especially between concussive events or concussive and nonconcussive head injuries, may be an important factor in outcome from sports-related concussion. Additionally, not all sport-related concussions have an equal impact. Research suggests that athletes with concussions that resulted in posttraumatic amnesia (PTA) and retrograde amnesia, but not those that were associated with loss of consciousness or disorientation alone, demonstrated poorer recovery two days post injury on computer-administered tests of cognition (Collins et al., 2003). Thus, the literature suggests that the impact of sports-related concussion on cognitive ability is related to aspects of concussion severity, frequency, and the timeframe during which the concussions occurred.
Moreover, while a host of player factors likely influence concussion outcomes, cognitive reserve (CR) and brain reserve (BR), protective mechanisms often cited in the neuropsychological literature (Bigler & Stern, 2015; Satz, Cole, Hardy, & Rassovsky, 2011; Stern, 2002), likely have a great influence over the cognitive impact of athletes’ concussive histories. CR refers to a threshold for the emergence of cognitive decline related to neuropathology. This threshold is set by multiple mentally enriching experiences and abilities, such that CR acts as a protective mechanism against brain injury or disease (Bigler & Stern, 2015; Stern, 2002). Indeed, socioeconomic status (occupation and income), premorbid intelligence, and enriching leisure activities (e.g., writing, socializing, and exercise) independently contribute to CR in moderate-to-severe TBI survivors (Levi, Rassovsky, Agranov, Sela-Kaufman, & Vakil, 2013; Rassovsky et al., 2015). Enriching experiences and abilities are thought to mediate CR through the establishment of more efficient neural networks and, by some accounts, the ability to recruit new, less injured, networks for task completion under conditions of increased task difficulty. Neuroimaging data from healthy older adults and individuals with Alzheimer’s disease support these assertions (Bigler & Stern, 2015; Stern et al., 2000). Like CR, BR also increases the threshold for neuropathology to result in cognitive impairment (Bigler & Stern, 2015; Satz et al., 2011). Simply put, greater brain size is related to better cognitive function following brain injury (Bigler & Stern, 2015). It is likely that CR and BR are interrelated to some degree (Satz et al., 2011). It seems reasonable to assume that CR and BR influence the cognitive signature of sports concussion. A recent meta-analysis provided evidence for the protective influence of CR in sportrelated concussion, as athletes with more years of education demonstrated better neuropsychological outcomes (Dougan, Horswill, & Geffen, 2014). In the current study, our primary aim was to determine whether concussion frequency, severity, and timeframe, as well as CR, could be used to create an index that would predict cognitive outcomes in retired American professional football players. It was hypothesized that a combination of these factors would predict cognitive ability better than concussion frequency alone. Other factors can also have an impact on the outcomes of professional football players. Sports-related concussions have been linked to depression in retired
Downloaded by [RMIT University] at 00:57 29 February 2016
JOURNAL OF CLINICAL AND EXPERIMENTAL NEUROPSYCHOLOGY
professional football players (Chen, Johnston, Petrides, & Ptito, 2008; Guskiewicz et al., 2007). Additionally, obesity, which is often linked to metabolic syndrome, is associated with cognitive decline and higher mortality rates in retired football players (Baron, Hein, Lehman, & Gersic, 2012; Miller et al., 2008; Willeumier, Taylor, & Amen, 2012). Finally, anabolic–androgenic steroid use has been linked to increased musculoskeletal injuries in retired professional football players (Horn, Gregory, & Guskiewicz, 2009). Thus, as a secondary aim, we explored the relationship between mood ratings, body mass index (BMI), and self-reported steroid use as they related to concussion, CR, and cognitive ability, as any or all of these factors could potentially act as confounding variables that might account for potential associations between the proposed index and cognitive outcome.
Method The current study was approved by the institutional review boards of the Los Angeles Biomedical Research Institute at Harbor–UCLA Medical Center, the John Wayne Cancer Institute, and the University of North Carolina at Chapel Hill. All participants consented to voluntary participation. Participants Participants in the current analysis were recruited for a project examining rates of hormonal dysfunction in retired professional football players from the Center for the Study of Retired Athletes (CSRA) in collaboration with the National Football League (NFL) Players’ Association (see Kelly et al., 2014). From the CSRA database, 430 retired NFL players who played at least one year in the NFL and who completed the CSRA General Health Questionnaire (GHQ; Horn et al., 2009; Kerr, Marshall, & Guskiewicz, 2012), the Short Form 36 (Brazier et al., 1992) with a Mental Component Score of .30, ps ≤ .05). Scores for each domain were averaged prior to the averaging of all of the domain scores to produce a global cognition T-score: {[(CVLT–II
Downloaded by [RMIT University] at 00:57 29 February 2016
JOURNAL OF CLINICAL AND EXPERIMENTAL NEUROPSYCHOLOGY
learning trial recall + long-delay free recall + recognition discriminability)/3] + [(ROFT immediate recall + ROCFT delayed recall)/2] + [(SDMT oral trial + written trial + TMT–A)/3] + [(F–A–S + Animals + TMT–B)/3]}/4. We did not combine measures of verbal and nonverbal memory, as these test scores were generally uncorrelated (rs = .23 to .30, ps > .05), although CVLT–II recognition discriminability did correlate with delayed recall on the ROFT (r = .34, p = .03). Given our sample size, we opted to use the global cognition score as our primary outcome variable, rather than using individual cognitive domain scores. In addition to the neuropsychological assessment, we collected data regarding demographics, concussion history, game play, and steroid use via questionnaires and interviews, as well as mood, via the Hospital Anxiety and Depression Scale (HADS; Zigmond & Snaith, 1983), and obesity, via the BMI at the time of testing. Historical data regarding concussions and other TBIs were collected via interview questions based on the OSU TBI method (Corrigan & Bogner, 2007) and the PTA-Q (McMillan et al., 1996). Both the OSU TBI method and the PTA-Q have been shown to be reliable and valid methods for assessing remote TBI. Also, all concussion data for the current study were restricted to the period that participants played professional football, as the TBI reports for this period were most reliable and verifiable; we verified TBI reports against previous reports on the GHQ. That said, some might question the ability of retired professional football players to recall instances of concussion that occurred many years ago (e.g., 20 years ago), especially since there appeared to be a culture of downplaying concussive impacts in professional football and only a relatively recent widespread recognition of the long-term consequences of sports-related concussions. However, concussive impacts in athletes have been considered important injuries (e.g., Daly, 1958; Hugenholtz & Richard, 1982), recognized to cause marked cognitive changes (e.g., Gronwall & Wrightson, 1975; Lynch & Yarnell, 1973), and thought to require specialized treatment for more than 20 years (Daly, 1958; Hugenholtz & Richard, 1982). And while these data may have been somewhat restricted to researchers and healthcare workers, we doubt that the acute effects of concussion were lost on our participants, as the sequelae of concussions very likely affected their
5
athletic (and other) performances; regardless of the culture around concussion reporting, athletic performances were surely at the forefront of the minds of individuals involved in professional football (i.e., players, coaches, etc.). Fairly recent data seem to corroborate this position, as retired professional football players showed moderately reliable reports of concussion over a nine-year period (Кw= 0.48; see Kerr et al., 2012). CR was based on participants’ AMNART-estimated verbal intelligence, reports of occupational attainment and academic achievement (see details below). Also, we utilized highest degree obtained rather than years of education as these variables were strongly, but not perfectly, correlated (r = .77, p < .001), and we reasoned that degree completion was more likely reflective of mastery of academic material than years of exposure to academic material.
Statistical analyses Hierarchal linear regression was employed to determine whether CR and concussion data (frequency, severity, and period of concussive risk) could predict cognitive ability in our sample. Specifically, an index based on the following equation was evaluated to determine whether it could predict global cognitive ability in a sample of retired professional American football players above and beyond mere concussion frequency or concussion dose [(ΣS + Σf)/T] (see below). Cognitive Outcome from Sports Concussion Index ðCOSCIÞ ¼ CR ½ðS þ f Þ=T where CR (cognitive reserve) was represented by the sum of the centered values for highest degree obtained (1 = high school, 2 = associate’s degree, 3 = bachelor’s degree, 4 = master’s degree, 5 = doctoral degree or greater), highest occupational obtainment (1 = unemployed, 2 = laborer/unskilled, 3 = operator/domestic worker, 4 = craftsman/foreman, 5 = manager/administrative/clerical/sales, 6 = professional/technical; this coding was adapted from a method for estimating intelligence from demographic variables; Scott, Krull, Williamson, Adams, & Iverson, 1997), and premorbid estimated intelligence (based on the AMNART; Grober & Sliwinski, 1991, Verbal IQ estimate); ΣS (concussion severity) was represented by the sum of PTA ratings
Downloaded by [RMIT University] at 00:57 29 February 2016
6
M. J. WRIGHT ET AL.
(0 = none, 1 = less than 30 min, 2 = 30 min to 24 hours, 3 = over 24 hours)1 for concussions experienced during professional football; Σf (concussion frequency) was represented by the number of concussions experienced during professional football; and T (time period of concussive risk) was represented by the number of professional seasons played. ΣS and Σf were summed in order to reflect the magnitude of the concussive experience, as previous data have suggested that concussions with PTA appear to result in worse cognitive outcomes than concussions without PTA (Collins et al., 2003). The sum of ΣS and Σf was divided by T to reflect the relative dosing of concussion, as repeated concussions occurring closer together were thought to have a greater impact on brain and cognitive function (Giza & Hovda, 2001; Laurer et al., 2001). Finally, this ratio was subtracted from CR, as CR has been shown to be protective against brain injury and disease (Stern, 2002). In general, our CR values tended to be greater than concussion dose values. Thus, higher index scores should be related to better cognitive outcomes (see Appendix for additional detail). We also evaluated the COSCI’s ability to predict the number of neuropsychological domains with at least one deficit score on two commonly used deficit classification systems (Heaton: T < 40; Wechsler: T < 30; see Brooks, Sherman, Iverson, Slick, & Strauss, 2011) with simple linear regressions. Finally, we explored the relationship between mood ratings, BMI, and reported steroid use and CR, concussion history, and cognitive ability with Pearson correlations.
Results Depressive and anxious symptoms (as assessed by the HADS; Anxiety score: M = 8.35, SD = 4.48; 1
Depression score: M = 5.75, SD = 3.89) along with reported frequency of steroid use (M = .24, SD = .45) did not correlate with global cognition, the COSCI, or the CR and concussion dose aspects of the COSCI (rs = –.26 to .04, ps = .11 to .99). That said, BMI scores (M = 33.19, SD = 5.36) were significantly associated with the COSCI (r = –.43, p = .006) and the CR component of the COSCI (r= –.41 p = .01), but not concussion dose or global cognition (rs = –.20 to .04, ps = .16 to .81). Since self-reported steroid use, depression, anxiety, and BMI scores did not correlate with global cognition, they were not included in our analysis of the index. It should be noted that the study sample demonstrated a range of concussions, PTA ratings, seasons played, years of education, and cognitive ability (see Table 1). With regard to cognition, our sample displayed deficits in attention and processing speed (Heaton system: 20.0%; Wechsler system: 5.0%),2 verbal memory (Heaton system: 22.5%; Wechsler system: 7.5%), nonverbal memory (Heaton system: 37.5%; Wechsler system: 2.5%), and executive ability (Heaton system: 37.5%; Wechsler system: 20.0%). Also, a number of our participants displayed deficit scores in two or more cognitive domains (Heaton system: 35.0%; Wechsler system: 12.5%). As can be seen in Table 2, professional football concussion frequency (Σf) did not significantly predict global cognitive ability, nor did concussive dose [(ΣS + Σf)/T], although concussive dose improved the variance accounted for by 9%. The inclusion of cognitive reserve yielded significant results and enhanced the ability of the index to predict global cognition, accounting for an additional 25% of the variance; the total model accounted for 34% of the
Average PTA duration was also considered, but it did not correlate with neuropsychological outcome as strongly as the sum of PTA durations. Also, loss of consciousness (LOC) was considered (LOC average and LOC sum), but it was not significantly correlated with neuropsychological outcome. 2 While the Heaton and Wechsler deficit classification systems (deficits scores < 1 SD and < 2 SDs, respectively) are commonly used in clinical neuropsychology, the Wechsler is the more common of the two (Brooks et al., 2011). When choosing a classification system, it is important to consider normal score variability, level of education, and intellectual functioning (Iverson & Brooks, 2011). Specifically, and beyond the impact of age and sex, healthy individuals commonly obtain one or more low test scores when administered a neuropsychological test battery, and the number of low test scores varies as a function of the classification system used (lower thresholds for deficit result in more low scores), the number of tests administered (more tests will result in a greater number of low scores), education (higher levels of education are inversely related to the number of low scores), and intelligence (greater intelligence is related to fewer low scores; see Iverson & Brooks, 2011). We chose to display sample performances using both systems as our participants displayed a broad range of years of education (15–20 years, M = 16.35, SD = 1.21) and premorbid intelligence estimates (91.11–126.46, M = 113.72, SD = 9.21).
JOURNAL OF CLINICAL AND EXPERIMENTAL NEUROPSYCHOLOGY
Table 2. Regression models for the prediction of global cognition by concussion frequency, concussion dose, and the Cognitive Outcome from Sports-Concussion Index. Model 1 Concussion frequency Value
Downloaded by [RMIT University] at 00:57 29 February 2016
R2 Adjusted R2 F
Model 2 Concussion dose
Model 3 COSCI
β B SE B β B SE B β B 0.06 0.08 0.21 –0.40 –0.59 0.32 0.52 0.27 .00 .09 .34 –.02 .04 .28 0.14
1.75
SE B 0.07
6.15*
Note. Table displays regression models for the prediction of global cognitive ability by concussion data and an index including concussion and cognitive reserve data. All concussion and game play data included in the regressions were restricted to professional football. COSCI = Cognitive Outcome from Sports-Concussion Index. Model 1 included the number of concussions sustained as the predictor. Model 2 added concussion dose [the sum of posttraumatic amnesia (PTA, a measure of severity) and concussion frequency divided by seasons of play; see Method section] as a predictor. Model 3 added the complete COSCI as a predictor. The threshold for statistical significance was set at p < .05. *p = .002.
variance in the prediction of global cognitive ability. With the addition of CR, beta became positive, reflecting the protective nature of CR with regard to global cognitive ability. It should be noted that loss of consciousness was also considered as a predictor, but it did not contribute to the prediction of cognitive outcome. We followed up with a simple linear regression for the prediction of global cognition by the COSCI. The COSCI significantly predicted global cognition, β = 0.56, t(39) = 4.17, p < .001, and accounted for 31% of the variance, R2 = .31, F(1, 39) = 17.38, p < .001. Additionally, the COSCI significantly predicted the number of cognitive domains with one or more deficit scores [Heaton system: β = –0.39, t(39) = –2.57, p = .014, R2 = .15, F(1, 39) = 6.63, p = .014; Wechsler system: β = – 0.53, t(39)= –3.82, p < .001, R2= .28, F(1, 39) = 14.60, p < .001].
Discussion The current data support the notion that protective (e.g., CR) and harmful (e.g., elements of the concussive history) factors can be considered together to create an index that is predictive of cognitive outcomes of athletes with concussive histories. Specifically, the COSCI, an index based on the difference between CR (represented by the sum of centered values for highest degree obtained, highest occupational obtainment, and estimated premorbid intellectual ability) and the
7
ratio of the sum of concussion frequency and severity (indexed by the sum of PTA durations) to the time period of concussive risk (indexed by seasons played) predicted global cognitive ability in retired professional football players, accounting for 31% of the variance. Additionally, COSCI also predicted the number of neuropsychological domains with deficit scores, accounting for 15–28% of the variance, depending on the classification system used. These findings underscore the importance of considering player-specific factors when attempting to predict long-term cognitive outcomes related to sports concussion. Moreover, our data suggest that both protective factors, such as CR (the increased threshold for the emergence of cognitive decline set by mental abilities and cognitively enriching experiences), and harmful factors such as various concussion variables (concussion frequency, severity, and the time period of concussive risk) should be considered in predictions of cognitive outcome. It should be noted that we excluded participants with significant performance validity failure, histories of brain injury or disease (other than sports-related concussion), and complicated TBI related to football play. Additionally, our sample had been retired for an average of 20 years, suggesting that the index was fairly robust to the passage of time. Our findings open the door to the potential use of player-specific indices suggestive of cognitive outcome to aid in return-to-play decisions and planning of potential healthcare needs (e.g., comprehensive neuropsychological assessment, neuroimaging, cognitive rehabilitation, etc.) of athletes in contact sports (see Appendix). That said, it would not be accurate to assume that all components of the index held the same weight in predicting cognitive outcome, as concussive dose accounted for more variance (9%) than concussion frequency alone (
