Emerging Techniques and Future Development Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
Vulnerability in Speed of Visuomotor Ability Ann B. Shuttleworth-Edwards a · Sarah E. Radloff b · Victoria J. Whitefield-Alexander a
Departments of a Psychology and b Statistics, Rhodes University, Grahamstown, South Africa
Abstract
Concussive brain injury is a common feature characterizing the field contact sports [1, 2]. Alongside the growing popularity of these sports at both child and adult levels, optimal management of the injury has evolved as a major area of concern among the medical and neuropsychological community [3, 4]. It has long been acknowledged that boxing is associated with repeated head insults resulting in structural neurological damage, permanent neurocognitive deficits, and the possible long-term outcome of chronic traumatic encephalopathy [5, 6]. However, it is only more recently that permanent effects of the injury have been raised as an issue of concern in field contact sports. The disquiet has arisen in the wake of multiple neuropsychological studies of American football, soccer, and Rugby Union (hereafter
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
The aim of this report is to illustrate the utility of neurocognitive testing as an investigative method to establish the presence of persistent effects of concussive brain injury amongst players of a contact sport with high risk of such injury. The report reviews the outcomes of three previously published neuropsychological studies on South African Rugby Union (hereafter ‘rugby’) from school through to the national adult level. The diagnostic utility of differential effects per se, as well as differential practice effects on visuomotor function, as a means of distinguishing poorer neurocognitive outcome for rugby versus demographically equivalent noncontact sports players, is described. From various methodological angles, at each level of play, the reviewed studies attest to the presence of long-term vulnerability in visuomotor speed in association with participation in rugby, in turn implicating diffuse frontotemporal dysfunction due to repetitive concussive and subconcussive injury amongst rugby players. As visuomotor speed is a prime function called upon for optimal scholastic and occupational performance, the robustly demonstrated decrement is of immediate clinical importance. Of critical heuristic relevance, however, is evidence of residual brain dysfunction in association with rugby as early as school level, which may ultimately be the precursor of chronic trau© 2014 S. Karger AG, Basel matic encephalopathy.
214
Shuttleworth-Edwards · Radloff · Whitefield-Alexander Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
‘rugby’) that reveal lingering deleterious psychometric test outcome for players of these sports [7]. In light of the interpretive nature of such investigations, which involves measurement of brain output as distinct from direct examination of the brain, as well as the cross-sectional nature of many of the studies, there has been a tenacious reserve, if not skepticism expressed amongst some about the validity of these findings [8]. In very recent times, however, potent cross-validation of the repeated psychometric indications of residual brain dysfunction is being achieved from more direct evidence of structural brain damage from neuroimaging and autopsy studies for older, and even in some instances for younger, participants of these sports [9–12]. In other words, the new relatively ‘hard’ evidence makes it more difficult to ignore or dismiss as alarmist the long history of psychometric test outcome presented over the past two decades similarly in support of residual effects on brain function of cumulative concussive injury [7]. Added impetus in this regard is being derived from the media frenzy around a multimillion dollar legal settlement by the National Football League (NFL) for suppressing the dangers of the injury [13]. Such cross-validation in turn ratifies the use of psychometric testing, when contextualized appropriately within the formal conceptual parameters of clinical neuropsychology, as a critically important investigative method in this area. Importantly, while the imaging and autopsy studies in question have used samples predominantly of players of American football or ice hockey, on logical grounds the disturbing implications can clearly be extrapolated beyond these convenience populations within the USA and Canada to other contact sports including rugby [13]. In other words, there is the compelling indication from these imaging and autopsy studies, supported indirectly by psychometric evidence, for developing neurodegenerative disease in association with long-term participation in any sport where player-to-player and player-to-ground collisions are intrinsic to the sport, thereby resulting in subconcussive (microtrauma) and concussive events being incurred on a repetitive basis. Of the at risk field contact sports, rugby (played extensively in the UK, New Z ealand, Australia, and South Africa) is considered to be the most dangerous, and reveals a higher incidence of reported concussion than American football [2]. Participants of rugby, therefore, are a highly suited population on which to evaluate concussion sequelae per se, and especially the hotly debated issue of whether or not there are damaging long-term effects of repetitive concussive and subconcussive brain injury. To this end, specifically using psychometric testing as an investigative method, a number of studies and/or collation of various studies have been completed and published by the present first author and colleagues [14–16], the key features of which will be reviewed below. Prior to embarking on the review, however, it is important to provide a synthesis of key conceptual parameters relevant to using neuropsychological evaluation in particular as an investigative tool in the concussion arena, as distinct from other investigative techniques, and to define other key concepts as applied in the three research studies.
Clinical neuropsychology is defined as ‘an applied science concerned with the behavioural expression of brain dysfunction’ [17]. Intrinsic to the science is a vast body of research starting in the first half of the 20th century with screening of brain-injured servicemen during the First World War that has linked various alterations and limitations of cognitive function with disturbances in specific parts of the brain and/or more complex neuropathological presentations. The critical point about this form of investigation is that it is an indirect measure of the brain, and therefore is necessarily interpretive and hypothetical to a much greater extent than the more direct examinations of brain structure and function enabled by autopsy and/or neuroimaging investigative methods. Nevertheless, the crucial contribution of neuropsychological research to neurological knowledge in general resides in the need to establish the behavioral effect of brain changes in addition to the concrete indication of brain alteration in and of itself. Specifically, neuropsychological investigation can be defined as a specialist level of inquiry about brain output rather than straight brain observation, which can only be established on the basis of cognitive, emotional, and behavioral testing, clinical observation, and inquiry, as this is a relatively noninvasive and cost-effective mode of investigation. Importantly, expense aside, it is not always possible to identify brain changes by means of other more direct investigative techniques despite massive technological advances in those fields, especially when anomalous brain function is in the early stages, and of a diffuse nature such as with concussive brain injury [4, 17]. In contrast, evidence of behavioral change, especially on probing neurocognitive examination, has the power to identify early signs of critical brain alterations in advance of their subsequent confirmation via other more direct modes of examination. In turn this knowledge may provide crucial clues as to etiological links and stage of onset of a particular pathology. Finally, for the conceptual clarity in this report, it is noted that the term ‘concussive brain injury’ will be alternatively referred to in this report as ‘mild traumatic brain injury’ such as is conventionally applied in the sports literature, although it is acknowledged that some use the terms separately to refer to different types of injury. In their research protocols, all the studies to be reported on below [14–16] use the term concussion to refer to a relatively low-velocity injury that causes brain ‘shaking’, being a subset of traumatic brain injury [4]. Operationally, the research studies in question define a concussion as any event following a head or body collision where there was any loss of consciousness or period of amnesia, or any changes in mental and physical status implicating postconcussion symptoms as typically described in the literature [18]. The focus of the three studies to be reviewed is on the long-term, i.e. chronic, outcome of concussion, as distinct from acute and subacute effects. In these studies, the term ‘chronic’ has been operationally differentiated from more immediate acute sequelae that typically resolve within 3 months after the injury, with effects which persist for longer than this being viewed as chronic, i.e. relatively intractable [19].
Visuomotor Vulnerability Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
215
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
Conceptual Issues
A review of the three designated studies on rugby follows below [14–16], the full methodological details are available in the original reports and will not be replicated here.
Study 1: Shuttleworth-Edwards et al., 2004 [14]
Study Participants Exclusion criteria for the study were a history of substance abuse, neurological or psychiatric condition, prior moderate or severe head injury for any reason (i.e. loss of unconsciousness >30 min), any prior grade failure, and/or a recorded learning problem. The final sample of predominantly white English- and Afrikaans-speaking adults consisted of 26 national open rugby players (15 forwards and 11 backs), 19 national under-21 rugby players (11 forwards and 8 backs), and 47 school rugby players (28 forwards and 19 backs), with control groups being made up, respectively, of 21 national open cricketers, a younger group of 21 national open field hockey players, and 34 schoolboy field hockey players. Cricket (as with baseball) and field hockey (in contrast to ice hockey) are sports which are not characterized by multiple concussions due to collisions between players or players to ground [20], which renders them highly suitable to call upon when making up noncontact sport control groups. Six school boys with a prior history of a learning problem, excluded from the main study, formed two subgroups for further comparative purposes, consisting of 3 rugby players with a learning problem, and 3 field hockey players without a learning problem. All comparative groups in the main study were equivalent for age, education, and estimated IQ (p > 0.05 in all instances) with the exception of the national under-21
216
Shuttleworth-Edwards · Radloff · Whitefield-Alexander Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
In an early research investigation, using paper-and-pencil psychometric testing in a relatively exploratory fashion with a wide-ranging test battery, it was decided to investigate the additive effect of repetitive sports concussion by conducting a comparative analysis of the percentages of players of South African rugby with objective test deficit [14]. The mode of psychometric test score evaluation took the form of a separate personalized analysis of every score within an individual’s test profile, in each case evaluated in comparison with a demographically appropriate norm, thereby tapping the full extent of deficit in a clinically relevant manner that is not possible in a group analysis. On this basis it was hypothesized that due to the aggregation effect of repeated concussion via participation in rugby, in association with longer or more exposed positional play, and/or in association with a prior cognitive vulnerability, rugby players would show more impairment than noncontact sport controls, more at risk forward players would show more impairment than backline players, older and higher level players would show more impairment than younger players, and players who had a recorded history of a learning problem would show more impairment than noncontact sport controls without a recorded history of a learning problem.
Procedure and Outcome Participants were assessed individually by psychology interns using standardized test formats, at the rugby preseason juncture following a 5-month rugby playing break, in order to exclude the acute and subacute effects of concussion. A wide spectrum of commonly employed paper-and-pencil neuropsychological measures, across a comprehensive spectrum of functional modalities known to be sensitive to impairments following mild traumatic brain injury, viz. attention and concentration, visuoperceptual processing speed, memory, verbal fluency, and hand motor dexterity, were used. Deficit was defined as a score greater than 1 SD from the best available normative data in the direction of poorer performance. Taken overall, the outcome of the main group test comparisons, in which percentages of individuals with cognitive deficit were compared via χ2 analyses, was in support of all the hypotheses with rugby players showing more deficit than controls, forward players exhibiting more deficit than backline players, and older players demonstrating more problems than younger school boy players in each of the functional areas of speed of information processing, memory, verbal fluency, and hand motor dexterity. The function that most frequently differentiated these groups significantly in the direction of the hypotheses was in visuospatial processing speed, incorporating outcome from the Wechsler Intelligence Adult Scale Digit Symbol Substitution subtest and the Trail Making Test as described in the source article [14]. The outcome for the subset of school boys with a learning disability was again in the hypothesized direction, with much more impairment for rugby players than controls in the areas of attention, processing speed, and visual memory, despite having a higher estimated IQ than the field hockey subgroup with a learning problem. The percentage of cognitive deficit for these cognitively vulnerable rugby players in both verbal and nonverbal attentional tasks clustered around 67–100% of cases, compared with close on zero indication of deficit across tests for the similarly cognitively vulnerable field hockey players. Although the subgroup sample numbers were small, the descriptive findings are commensurate with the research of Collins et al. [21], who demonstrated significantly increased cognitive deficit in football players with repeated concussions and a learning disability compared with concussed football players without a learning disability. A heuristic strength of this early study using paper-and-pencil testing is its exploratory nature of using a wide battery that allowed for a comparative view on which tests
Visuomotor Vulnerability Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
217
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
rugby players who were significantly lower than field hockey controls with respect to all these variables (p < 0.05 in all instances). Since each level of play was analyzed separately, the outcome of the national open and school cohorts would not be contaminated due to the demographic differences exclusively in the under-21 group. In the subgroup analysis, the rugby group with a learning disability had a lower IQ estimate than the field hockey control group with a learning disability (mean IQ: 113.5 vs. 101.0, respectively), so that any deleterious effects in the rugby group could not be accounted for by lower intellectual ability.
served best to differentiate between the at risk rugby groups and the control groups. Further, the mode of investigation was based on an analysis of the proportion (i.e. number) of rugby players who showed neuropsychological deficits on tests relative to appropriate norms in areas known to be susceptible to damage in association with mild traumatic brain injury. In that this is a mode of analysis typically applied in clinical situations, the outcome can be seen to have clinical as well as statistical relevance. A limitation of the study that might be cause for challenging the validity of the outcome, however, is the very fact of its having such a wide test battery with multiple comparisons calling for Bonferroni’s correction to guard against the chance of a type 1 error (finding erroneous significance by chance). These were not applied thereby opening the study to the criticism of possibly inflating effects [22, 23]. On the other hand, the relatively subtle nature of the findings in respect to the small sample numbers might have served to obscure relevant effects [22, 23]. These criticisms call for more focused hypothesis-testing research in terms of test modality for larger sample numbers, and provided the incentive for further investigation of the study data, to be described below.
Typically in many of the early neuropsychological studies on contact sports, an expansive battery of tests was used with a view to enhancing sensitivity to subtle mild brain injury effects [7], such as was also applied in the earlier series of South African rugby studies described above [14]. However, as also alluded to above, the use of multiple measures in the research context calls for substantial adjustments towards statistical stringency in order to guard against type I errors. In order to circumvent the problem it has been suggested that data derived from collections of smaller exploratory studies should be combined to gain the increased statistical power that becomes available in a larger sample [22, 23]. Further, the employment of a restricted neurocognitive variable with a view to focused hypothesis testing (rather than the application of a wide-ranging exploratory test battery) is considered the more appropriate methodological route to follow [22, 23]. Accordingly, the aim of this new study [15] was to provide a focused neurocognitive investigation into outcome for the single functional modality of visuomotor processing speed, using data that were available from the earlier wide-ranging investigation described above [14] that explored a spectrum of functional modalities on rugby players at the first team high school, under-21 national, and open national levels. Data selection for the new analysis pertained to all rugby participants and noncontact sport controls in the prior study from each of the three levels of play, and was exclusively in respect of the two tests of visuomotor speed that were identified as being the most discriminatory of possible brain injury effects within the wide battery employed in that study series, viz. Digit Symbol Substitution Test and Trail Making Test parts A
218
Shuttleworth-Edwards · Radloff · Whitefield-Alexander Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
Study 2: Shuttleworth-Edwards and Radloff, 2008 [15]
and B, as described by Lezak et al. [17]. Specifically, the Digit Symbol Substitution Test calls upon visual scanning, motor persistence, sustained attention, response speed, and visuomotor coordination, and is consistently more sensitive to brain damage than the other Wechsler Intelligence Scale subtests [17]. The Trail Making Test similarly taps visual scanning, motor persistence, sustained attention, and response speed, as well as visuomotor tracking, divided attention, and cognitive flexibility [17]. As with Digit Symbol Substitution, the test is highly sensitive to the effects of even mild brain injury. It was hypothesized that there would be clinically relevant signs of diminished performance in visuomotor processing speed for the cerebrally at risk rugby players compared with a noncontact sport group, and amongst cerebrally more at risk forward players compared with backs, due to a history of cumulative concussive and subconcussive events sustained in association with years of participation in the sport. It was expected that synergistically these concussive events would have cumulative effects on rugby players during many years of exposure to the game and predispose them to neuropsychological vulnerability in the prototypically sensitive area of visuomotor processing speed.
Procedure and Outcome As indicated above, from the original wide-ranging battery, only two tests within the single functional modality of visuomotor processing speed were isolated in light of their demonstrated superior sensitivity. These included the Digit Symbol Substitution Test with the score representing the number of pairings of symbol and numbers com-
Visuomotor Vulnerability Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
219
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
Study Participants Participants for the newly combined analysis were top team South African male rugby players (n = 124) versus top team South African noncontact sport controls including a mixture of cricket and field hockey athletes (n = 102), across the three participant stages of play of the earlier study. The total rugby group was divided for comparison between forward positional players who are more exposed to cerebral injury, and backline players who are less exposed to cerebral injury (n = 71 and 53, respectively). As indicated for the original three studies, exclusion criteria were a history of alcohol or substance abuse, neurological or psychiatric disease, a prior failed grade or learning disability, or moderate to severe TBI (loss of consciousness >30 min). All comparative groups were statistically equivalent for age, level of education, and estimated IQ, as well as performance on the Finger Tapping Preferred and Non-Preferred Hand Tests ensure that none of these variables, including physical hand injury, did not account for slowing on visuomotor tracking tasks (p > 0.05 in all instances). Finally, there was a highly significant difference in the percentage of rugby players with a history of 2+ concussions compared with controls (53 vs. 17% respectively, p < 0.0001), but no significant difference in the percentage of forwards with a history of 2+ concussions compared with backs (49 vs. 53% respectively, p > 0.05).
Table 1. t test comparisons of rugby vs. controls and forwards vs. backs for the Digit Symbol Substitution Test, Trail Making Test (TMT) A and B, and Processing Speed (PS) factor
Digit Symbol TMT A TMT B PS factor
Rugby (n = 124) mean ± SD
Control (n = 102) mean ± SD
t value Effect size, δ
95% CI
p value
Forwards (n = 71) mean ± SD
Backs (n = 53) mean ± SD
49.98±9.20 27.86±7.98 58.57±19.68 –0.21±1.06
53.41±8.28 24.87±6.78 53.00±14.53 0.25±0.85
2.92 –3.00 –2.38 3.51
0.13 to 0.65 –0.66 to –0.13 –0.58 to –0.05 0.20 to 0.74
0.004** 0.003** 0.018* 0.0005**
47.99±9.00 27.76±7.42 60.26±20.99 –0.33±1.02
52.63±8.85 2.86 28.00±8.75 0.16 56.33±17.73 –1.10 –0.04±1.11 1.54
0.39 –0.40 –0.32 0.47
t value
Effect size, δ
95% CI
p value
0.52 0.03 –0.20 0.28
0.16 to 0.88 –0.32 to 0.38 –0.55 to 0.16 –0.08 to 0.64
0.005** 0.872 0.273 0.127
pleted in 90 s, and the Trail Making Test parts A and B, which are scored according to the time needed to complete the task, as described in well-known neuropsychological texts [17]. As described earlier in respect of the original series of studies from which the data were taken, the tests were administered individually in the standardized formats by psychology interns at preseason following the 5-month break to identify residual rather than acute effects. The test data were submitted to factor analysis, correlational analyses of test scores with age, and independent samples t test analyses to test for differences in performance on the tests and the mean factor scores between the comparative groups. The factor analysis revealed a single dominant factor with an eigenvalue greater than 1 (λ = 1.88; hereafter termed ‘processing speed factor’), accounting for 62.6% of the variation in the data. This finding confirmed the legitimacy of targeting a single neurocognitive function as the dependent variable via the use of the two designated neurocognitive tests, on the basis of which Bonferroni’s adjustments were not considered indicated. With reference to table 1, it is evident that t test comparisons of the total rugby group versus controls and forwards versus backs revealed highly significant differences in the direction of poorer performance for the rugby group compared with controls on the Digit Symbol Substitution Test, Trail Making Test parts A and B, and the processing speed factor, and for forwards compared with backs on the Digit Symbol Substitution Test (p < 0.001 in most instances), all with clinically relevant effect sizes. Finally, correlations between age and rugby test scores were not significant (p > 0.05 in all instances). Overall, taking significant outcome into account, it was considered that the results implicated depressed performance for the rugby players relative to controls and for the forwards versus backs, and is of clinical relevance due to years of exposure to cumulative concussive and subconcussive events. The absence of significant correlations between age and test scores for rugby participants implicated the presence of deleterious effects on processing speed at all levels of play. In that the comparative groups were well matched for age, education, and IQ, the presence of preexisting differences due to
220
Shuttleworth-Edwards · Radloff · Whitefield-Alexander Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
Reproduced from Shuttleworth-Edwards and Radloff [15], with kind permission from Oxford University Press, publishers of Archives of Clinical Neuropsychology. * p < 0.05; ** p < 0.01.
sports affiliation (rugby vs. noncontact sport) does not appear likely. Further, the presence of pre-existing sports group differences does not provide a satisfactory explanation for the within rugby group findings of the study, which revealed poorer performance on the Digit Symbol Substitution subtest for more at risk forward players compared with backline players. Rather, the substantive differentiating feature of the rugby group from controls was the presence of a significantly higher number of reported concussions for rugby players, without even taking into consideration the known high tendency for underreporting of this injury and an overlay of necessarily unreported subconcussive incidents that are likely to have occurred over a long rugby playing career. It was concluded that this composite study provided compelling support to the growing evidence across a number of studies of lowered performance on speeded visuomotor tasks for participants of rugby at all levels of play starting already at the end of the high school years, being a phenomenon worthy of on-going investigation.
The series of cross-sectional South African rugby studies described above, using paper-and-pencil instruments for preseason testing alone, were followed by three new prospective rugby studies employing computerized testing for pre- and postseason testing. These three new studies, the details of which are summarized elsewhere [16], employed the Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT) computerized test [24, 25], and were specifically targeted at high school and university levels of play. A common feature of note when perusing the outcome of these new studies was that the noncontact sports groups improved on the computerized ImPACT Visual Motor Speed (VMS) composite between the preand postseason test intervals, whereas the rugby players did not. It is well documented that brain damage populations typically fail to learn with practice on cognitive tests where there is an expected practice effect [17, 26], and therefore it was considered that this feature had important investigative potential as a possible diagnostic indicator of residual cerebral impairment in association cumulative concussive events sustained in rugby. Accordingly, for the purposes of a new analysis, it was decided to combine the prospective pre- versus postseason outcome data available from these three new South African rugby studies on the single functional modality of visuoperceptual speed as tapped by the ImPACT VMS composite. It was hypothesized that cognitive vulnerability would be apparent in the contact sport groups compared to equivalent noncontact sport controls due to failure to benefit from practice on this task, as a consequence of a long-term history of many years of participation in a sport with a high risk of concussive insults, in interaction with the effects of any unrecognized concussive and subconcussive event sustained during the most recent season of participation in the sport (i.e. hidden within season effects), none of which would be characteristic of the
Visuomotor Vulnerability Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
221
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
Study 3: Shuttleworth-Edwards et al., 2014 [16]
Table 2. Combined sample: dependent t test comparisons of ImPACT Visual Motor Speed (VMS) scores at the pre- vs. postseason intervals for the rugby and noncontact sports groups Source
Sample size, n
Preseason mean ± SD
Postseason mean ± SD
t value
Effect size δ-value
95% CI
p value
Rugby VMS Noncontact VMS
145 106
36.57±6.80 36.65±7.24
36.77±6.62 38.39±6.70
–0.393 –3.967
–0.03 –0.39
–0.20 to 0.13 –0.58 to –0.19
0.695
Vulnerability in Speed of Visuomotor Ability Ann B. Shuttleworth-Edwards a · Sarah E. Radloff b · Victoria J. Whitefield-Alexander a
Departments of a Psychology and b Statistics, Rhodes University, Grahamstown, South Africa
Abstract
Concussive brain injury is a common feature characterizing the field contact sports [1, 2]. Alongside the growing popularity of these sports at both child and adult levels, optimal management of the injury has evolved as a major area of concern among the medical and neuropsychological community [3, 4]. It has long been acknowledged that boxing is associated with repeated head insults resulting in structural neurological damage, permanent neurocognitive deficits, and the possible long-term outcome of chronic traumatic encephalopathy [5, 6]. However, it is only more recently that permanent effects of the injury have been raised as an issue of concern in field contact sports. The disquiet has arisen in the wake of multiple neuropsychological studies of American football, soccer, and Rugby Union (hereafter
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
The aim of this report is to illustrate the utility of neurocognitive testing as an investigative method to establish the presence of persistent effects of concussive brain injury amongst players of a contact sport with high risk of such injury. The report reviews the outcomes of three previously published neuropsychological studies on South African Rugby Union (hereafter ‘rugby’) from school through to the national adult level. The diagnostic utility of differential effects per se, as well as differential practice effects on visuomotor function, as a means of distinguishing poorer neurocognitive outcome for rugby versus demographically equivalent noncontact sports players, is described. From various methodological angles, at each level of play, the reviewed studies attest to the presence of long-term vulnerability in visuomotor speed in association with participation in rugby, in turn implicating diffuse frontotemporal dysfunction due to repetitive concussive and subconcussive injury amongst rugby players. As visuomotor speed is a prime function called upon for optimal scholastic and occupational performance, the robustly demonstrated decrement is of immediate clinical importance. Of critical heuristic relevance, however, is evidence of residual brain dysfunction in association with rugby as early as school level, which may ultimately be the precursor of chronic trau© 2014 S. Karger AG, Basel matic encephalopathy.
214
Shuttleworth-Edwards · Radloff · Whitefield-Alexander Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
‘rugby’) that reveal lingering deleterious psychometric test outcome for players of these sports [7]. In light of the interpretive nature of such investigations, which involves measurement of brain output as distinct from direct examination of the brain, as well as the cross-sectional nature of many of the studies, there has been a tenacious reserve, if not skepticism expressed amongst some about the validity of these findings [8]. In very recent times, however, potent cross-validation of the repeated psychometric indications of residual brain dysfunction is being achieved from more direct evidence of structural brain damage from neuroimaging and autopsy studies for older, and even in some instances for younger, participants of these sports [9–12]. In other words, the new relatively ‘hard’ evidence makes it more difficult to ignore or dismiss as alarmist the long history of psychometric test outcome presented over the past two decades similarly in support of residual effects on brain function of cumulative concussive injury [7]. Added impetus in this regard is being derived from the media frenzy around a multimillion dollar legal settlement by the National Football League (NFL) for suppressing the dangers of the injury [13]. Such cross-validation in turn ratifies the use of psychometric testing, when contextualized appropriately within the formal conceptual parameters of clinical neuropsychology, as a critically important investigative method in this area. Importantly, while the imaging and autopsy studies in question have used samples predominantly of players of American football or ice hockey, on logical grounds the disturbing implications can clearly be extrapolated beyond these convenience populations within the USA and Canada to other contact sports including rugby [13]. In other words, there is the compelling indication from these imaging and autopsy studies, supported indirectly by psychometric evidence, for developing neurodegenerative disease in association with long-term participation in any sport where player-to-player and player-to-ground collisions are intrinsic to the sport, thereby resulting in subconcussive (microtrauma) and concussive events being incurred on a repetitive basis. Of the at risk field contact sports, rugby (played extensively in the UK, New Z ealand, Australia, and South Africa) is considered to be the most dangerous, and reveals a higher incidence of reported concussion than American football [2]. Participants of rugby, therefore, are a highly suited population on which to evaluate concussion sequelae per se, and especially the hotly debated issue of whether or not there are damaging long-term effects of repetitive concussive and subconcussive brain injury. To this end, specifically using psychometric testing as an investigative method, a number of studies and/or collation of various studies have been completed and published by the present first author and colleagues [14–16], the key features of which will be reviewed below. Prior to embarking on the review, however, it is important to provide a synthesis of key conceptual parameters relevant to using neuropsychological evaluation in particular as an investigative tool in the concussion arena, as distinct from other investigative techniques, and to define other key concepts as applied in the three research studies.
Clinical neuropsychology is defined as ‘an applied science concerned with the behavioural expression of brain dysfunction’ [17]. Intrinsic to the science is a vast body of research starting in the first half of the 20th century with screening of brain-injured servicemen during the First World War that has linked various alterations and limitations of cognitive function with disturbances in specific parts of the brain and/or more complex neuropathological presentations. The critical point about this form of investigation is that it is an indirect measure of the brain, and therefore is necessarily interpretive and hypothetical to a much greater extent than the more direct examinations of brain structure and function enabled by autopsy and/or neuroimaging investigative methods. Nevertheless, the crucial contribution of neuropsychological research to neurological knowledge in general resides in the need to establish the behavioral effect of brain changes in addition to the concrete indication of brain alteration in and of itself. Specifically, neuropsychological investigation can be defined as a specialist level of inquiry about brain output rather than straight brain observation, which can only be established on the basis of cognitive, emotional, and behavioral testing, clinical observation, and inquiry, as this is a relatively noninvasive and cost-effective mode of investigation. Importantly, expense aside, it is not always possible to identify brain changes by means of other more direct investigative techniques despite massive technological advances in those fields, especially when anomalous brain function is in the early stages, and of a diffuse nature such as with concussive brain injury [4, 17]. In contrast, evidence of behavioral change, especially on probing neurocognitive examination, has the power to identify early signs of critical brain alterations in advance of their subsequent confirmation via other more direct modes of examination. In turn this knowledge may provide crucial clues as to etiological links and stage of onset of a particular pathology. Finally, for the conceptual clarity in this report, it is noted that the term ‘concussive brain injury’ will be alternatively referred to in this report as ‘mild traumatic brain injury’ such as is conventionally applied in the sports literature, although it is acknowledged that some use the terms separately to refer to different types of injury. In their research protocols, all the studies to be reported on below [14–16] use the term concussion to refer to a relatively low-velocity injury that causes brain ‘shaking’, being a subset of traumatic brain injury [4]. Operationally, the research studies in question define a concussion as any event following a head or body collision where there was any loss of consciousness or period of amnesia, or any changes in mental and physical status implicating postconcussion symptoms as typically described in the literature [18]. The focus of the three studies to be reviewed is on the long-term, i.e. chronic, outcome of concussion, as distinct from acute and subacute effects. In these studies, the term ‘chronic’ has been operationally differentiated from more immediate acute sequelae that typically resolve within 3 months after the injury, with effects which persist for longer than this being viewed as chronic, i.e. relatively intractable [19].
Visuomotor Vulnerability Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
215
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
Conceptual Issues
A review of the three designated studies on rugby follows below [14–16], the full methodological details are available in the original reports and will not be replicated here.
Study 1: Shuttleworth-Edwards et al., 2004 [14]
Study Participants Exclusion criteria for the study were a history of substance abuse, neurological or psychiatric condition, prior moderate or severe head injury for any reason (i.e. loss of unconsciousness >30 min), any prior grade failure, and/or a recorded learning problem. The final sample of predominantly white English- and Afrikaans-speaking adults consisted of 26 national open rugby players (15 forwards and 11 backs), 19 national under-21 rugby players (11 forwards and 8 backs), and 47 school rugby players (28 forwards and 19 backs), with control groups being made up, respectively, of 21 national open cricketers, a younger group of 21 national open field hockey players, and 34 schoolboy field hockey players. Cricket (as with baseball) and field hockey (in contrast to ice hockey) are sports which are not characterized by multiple concussions due to collisions between players or players to ground [20], which renders them highly suitable to call upon when making up noncontact sport control groups. Six school boys with a prior history of a learning problem, excluded from the main study, formed two subgroups for further comparative purposes, consisting of 3 rugby players with a learning problem, and 3 field hockey players without a learning problem. All comparative groups in the main study were equivalent for age, education, and estimated IQ (p > 0.05 in all instances) with the exception of the national under-21
216
Shuttleworth-Edwards · Radloff · Whitefield-Alexander Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
In an early research investigation, using paper-and-pencil psychometric testing in a relatively exploratory fashion with a wide-ranging test battery, it was decided to investigate the additive effect of repetitive sports concussion by conducting a comparative analysis of the percentages of players of South African rugby with objective test deficit [14]. The mode of psychometric test score evaluation took the form of a separate personalized analysis of every score within an individual’s test profile, in each case evaluated in comparison with a demographically appropriate norm, thereby tapping the full extent of deficit in a clinically relevant manner that is not possible in a group analysis. On this basis it was hypothesized that due to the aggregation effect of repeated concussion via participation in rugby, in association with longer or more exposed positional play, and/or in association with a prior cognitive vulnerability, rugby players would show more impairment than noncontact sport controls, more at risk forward players would show more impairment than backline players, older and higher level players would show more impairment than younger players, and players who had a recorded history of a learning problem would show more impairment than noncontact sport controls without a recorded history of a learning problem.
Procedure and Outcome Participants were assessed individually by psychology interns using standardized test formats, at the rugby preseason juncture following a 5-month rugby playing break, in order to exclude the acute and subacute effects of concussion. A wide spectrum of commonly employed paper-and-pencil neuropsychological measures, across a comprehensive spectrum of functional modalities known to be sensitive to impairments following mild traumatic brain injury, viz. attention and concentration, visuoperceptual processing speed, memory, verbal fluency, and hand motor dexterity, were used. Deficit was defined as a score greater than 1 SD from the best available normative data in the direction of poorer performance. Taken overall, the outcome of the main group test comparisons, in which percentages of individuals with cognitive deficit were compared via χ2 analyses, was in support of all the hypotheses with rugby players showing more deficit than controls, forward players exhibiting more deficit than backline players, and older players demonstrating more problems than younger school boy players in each of the functional areas of speed of information processing, memory, verbal fluency, and hand motor dexterity. The function that most frequently differentiated these groups significantly in the direction of the hypotheses was in visuospatial processing speed, incorporating outcome from the Wechsler Intelligence Adult Scale Digit Symbol Substitution subtest and the Trail Making Test as described in the source article [14]. The outcome for the subset of school boys with a learning disability was again in the hypothesized direction, with much more impairment for rugby players than controls in the areas of attention, processing speed, and visual memory, despite having a higher estimated IQ than the field hockey subgroup with a learning problem. The percentage of cognitive deficit for these cognitively vulnerable rugby players in both verbal and nonverbal attentional tasks clustered around 67–100% of cases, compared with close on zero indication of deficit across tests for the similarly cognitively vulnerable field hockey players. Although the subgroup sample numbers were small, the descriptive findings are commensurate with the research of Collins et al. [21], who demonstrated significantly increased cognitive deficit in football players with repeated concussions and a learning disability compared with concussed football players without a learning disability. A heuristic strength of this early study using paper-and-pencil testing is its exploratory nature of using a wide battery that allowed for a comparative view on which tests
Visuomotor Vulnerability Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
217
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
rugby players who were significantly lower than field hockey controls with respect to all these variables (p < 0.05 in all instances). Since each level of play was analyzed separately, the outcome of the national open and school cohorts would not be contaminated due to the demographic differences exclusively in the under-21 group. In the subgroup analysis, the rugby group with a learning disability had a lower IQ estimate than the field hockey control group with a learning disability (mean IQ: 113.5 vs. 101.0, respectively), so that any deleterious effects in the rugby group could not be accounted for by lower intellectual ability.
served best to differentiate between the at risk rugby groups and the control groups. Further, the mode of investigation was based on an analysis of the proportion (i.e. number) of rugby players who showed neuropsychological deficits on tests relative to appropriate norms in areas known to be susceptible to damage in association with mild traumatic brain injury. In that this is a mode of analysis typically applied in clinical situations, the outcome can be seen to have clinical as well as statistical relevance. A limitation of the study that might be cause for challenging the validity of the outcome, however, is the very fact of its having such a wide test battery with multiple comparisons calling for Bonferroni’s correction to guard against the chance of a type 1 error (finding erroneous significance by chance). These were not applied thereby opening the study to the criticism of possibly inflating effects [22, 23]. On the other hand, the relatively subtle nature of the findings in respect to the small sample numbers might have served to obscure relevant effects [22, 23]. These criticisms call for more focused hypothesis-testing research in terms of test modality for larger sample numbers, and provided the incentive for further investigation of the study data, to be described below.
Typically in many of the early neuropsychological studies on contact sports, an expansive battery of tests was used with a view to enhancing sensitivity to subtle mild brain injury effects [7], such as was also applied in the earlier series of South African rugby studies described above [14]. However, as also alluded to above, the use of multiple measures in the research context calls for substantial adjustments towards statistical stringency in order to guard against type I errors. In order to circumvent the problem it has been suggested that data derived from collections of smaller exploratory studies should be combined to gain the increased statistical power that becomes available in a larger sample [22, 23]. Further, the employment of a restricted neurocognitive variable with a view to focused hypothesis testing (rather than the application of a wide-ranging exploratory test battery) is considered the more appropriate methodological route to follow [22, 23]. Accordingly, the aim of this new study [15] was to provide a focused neurocognitive investigation into outcome for the single functional modality of visuomotor processing speed, using data that were available from the earlier wide-ranging investigation described above [14] that explored a spectrum of functional modalities on rugby players at the first team high school, under-21 national, and open national levels. Data selection for the new analysis pertained to all rugby participants and noncontact sport controls in the prior study from each of the three levels of play, and was exclusively in respect of the two tests of visuomotor speed that were identified as being the most discriminatory of possible brain injury effects within the wide battery employed in that study series, viz. Digit Symbol Substitution Test and Trail Making Test parts A
218
Shuttleworth-Edwards · Radloff · Whitefield-Alexander Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
Study 2: Shuttleworth-Edwards and Radloff, 2008 [15]
and B, as described by Lezak et al. [17]. Specifically, the Digit Symbol Substitution Test calls upon visual scanning, motor persistence, sustained attention, response speed, and visuomotor coordination, and is consistently more sensitive to brain damage than the other Wechsler Intelligence Scale subtests [17]. The Trail Making Test similarly taps visual scanning, motor persistence, sustained attention, and response speed, as well as visuomotor tracking, divided attention, and cognitive flexibility [17]. As with Digit Symbol Substitution, the test is highly sensitive to the effects of even mild brain injury. It was hypothesized that there would be clinically relevant signs of diminished performance in visuomotor processing speed for the cerebrally at risk rugby players compared with a noncontact sport group, and amongst cerebrally more at risk forward players compared with backs, due to a history of cumulative concussive and subconcussive events sustained in association with years of participation in the sport. It was expected that synergistically these concussive events would have cumulative effects on rugby players during many years of exposure to the game and predispose them to neuropsychological vulnerability in the prototypically sensitive area of visuomotor processing speed.
Procedure and Outcome As indicated above, from the original wide-ranging battery, only two tests within the single functional modality of visuomotor processing speed were isolated in light of their demonstrated superior sensitivity. These included the Digit Symbol Substitution Test with the score representing the number of pairings of symbol and numbers com-
Visuomotor Vulnerability Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
219
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
Study Participants Participants for the newly combined analysis were top team South African male rugby players (n = 124) versus top team South African noncontact sport controls including a mixture of cricket and field hockey athletes (n = 102), across the three participant stages of play of the earlier study. The total rugby group was divided for comparison between forward positional players who are more exposed to cerebral injury, and backline players who are less exposed to cerebral injury (n = 71 and 53, respectively). As indicated for the original three studies, exclusion criteria were a history of alcohol or substance abuse, neurological or psychiatric disease, a prior failed grade or learning disability, or moderate to severe TBI (loss of consciousness >30 min). All comparative groups were statistically equivalent for age, level of education, and estimated IQ, as well as performance on the Finger Tapping Preferred and Non-Preferred Hand Tests ensure that none of these variables, including physical hand injury, did not account for slowing on visuomotor tracking tasks (p > 0.05 in all instances). Finally, there was a highly significant difference in the percentage of rugby players with a history of 2+ concussions compared with controls (53 vs. 17% respectively, p < 0.0001), but no significant difference in the percentage of forwards with a history of 2+ concussions compared with backs (49 vs. 53% respectively, p > 0.05).
Table 1. t test comparisons of rugby vs. controls and forwards vs. backs for the Digit Symbol Substitution Test, Trail Making Test (TMT) A and B, and Processing Speed (PS) factor
Digit Symbol TMT A TMT B PS factor
Rugby (n = 124) mean ± SD
Control (n = 102) mean ± SD
t value Effect size, δ
95% CI
p value
Forwards (n = 71) mean ± SD
Backs (n = 53) mean ± SD
49.98±9.20 27.86±7.98 58.57±19.68 –0.21±1.06
53.41±8.28 24.87±6.78 53.00±14.53 0.25±0.85
2.92 –3.00 –2.38 3.51
0.13 to 0.65 –0.66 to –0.13 –0.58 to –0.05 0.20 to 0.74
0.004** 0.003** 0.018* 0.0005**
47.99±9.00 27.76±7.42 60.26±20.99 –0.33±1.02
52.63±8.85 2.86 28.00±8.75 0.16 56.33±17.73 –1.10 –0.04±1.11 1.54
0.39 –0.40 –0.32 0.47
t value
Effect size, δ
95% CI
p value
0.52 0.03 –0.20 0.28
0.16 to 0.88 –0.32 to 0.38 –0.55 to 0.16 –0.08 to 0.64
0.005** 0.872 0.273 0.127
pleted in 90 s, and the Trail Making Test parts A and B, which are scored according to the time needed to complete the task, as described in well-known neuropsychological texts [17]. As described earlier in respect of the original series of studies from which the data were taken, the tests were administered individually in the standardized formats by psychology interns at preseason following the 5-month break to identify residual rather than acute effects. The test data were submitted to factor analysis, correlational analyses of test scores with age, and independent samples t test analyses to test for differences in performance on the tests and the mean factor scores between the comparative groups. The factor analysis revealed a single dominant factor with an eigenvalue greater than 1 (λ = 1.88; hereafter termed ‘processing speed factor’), accounting for 62.6% of the variation in the data. This finding confirmed the legitimacy of targeting a single neurocognitive function as the dependent variable via the use of the two designated neurocognitive tests, on the basis of which Bonferroni’s adjustments were not considered indicated. With reference to table 1, it is evident that t test comparisons of the total rugby group versus controls and forwards versus backs revealed highly significant differences in the direction of poorer performance for the rugby group compared with controls on the Digit Symbol Substitution Test, Trail Making Test parts A and B, and the processing speed factor, and for forwards compared with backs on the Digit Symbol Substitution Test (p < 0.001 in most instances), all with clinically relevant effect sizes. Finally, correlations between age and rugby test scores were not significant (p > 0.05 in all instances). Overall, taking significant outcome into account, it was considered that the results implicated depressed performance for the rugby players relative to controls and for the forwards versus backs, and is of clinical relevance due to years of exposure to cumulative concussive and subconcussive events. The absence of significant correlations between age and test scores for rugby participants implicated the presence of deleterious effects on processing speed at all levels of play. In that the comparative groups were well matched for age, education, and IQ, the presence of preexisting differences due to
220
Shuttleworth-Edwards · Radloff · Whitefield-Alexander Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
Reproduced from Shuttleworth-Edwards and Radloff [15], with kind permission from Oxford University Press, publishers of Archives of Clinical Neuropsychology. * p < 0.05; ** p < 0.01.
sports affiliation (rugby vs. noncontact sport) does not appear likely. Further, the presence of pre-existing sports group differences does not provide a satisfactory explanation for the within rugby group findings of the study, which revealed poorer performance on the Digit Symbol Substitution subtest for more at risk forward players compared with backline players. Rather, the substantive differentiating feature of the rugby group from controls was the presence of a significantly higher number of reported concussions for rugby players, without even taking into consideration the known high tendency for underreporting of this injury and an overlay of necessarily unreported subconcussive incidents that are likely to have occurred over a long rugby playing career. It was concluded that this composite study provided compelling support to the growing evidence across a number of studies of lowered performance on speeded visuomotor tasks for participants of rugby at all levels of play starting already at the end of the high school years, being a phenomenon worthy of on-going investigation.
The series of cross-sectional South African rugby studies described above, using paper-and-pencil instruments for preseason testing alone, were followed by three new prospective rugby studies employing computerized testing for pre- and postseason testing. These three new studies, the details of which are summarized elsewhere [16], employed the Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT) computerized test [24, 25], and were specifically targeted at high school and university levels of play. A common feature of note when perusing the outcome of these new studies was that the noncontact sports groups improved on the computerized ImPACT Visual Motor Speed (VMS) composite between the preand postseason test intervals, whereas the rugby players did not. It is well documented that brain damage populations typically fail to learn with practice on cognitive tests where there is an expected practice effect [17, 26], and therefore it was considered that this feature had important investigative potential as a possible diagnostic indicator of residual cerebral impairment in association cumulative concussive events sustained in rugby. Accordingly, for the purposes of a new analysis, it was decided to combine the prospective pre- versus postseason outcome data available from these three new South African rugby studies on the single functional modality of visuoperceptual speed as tapped by the ImPACT VMS composite. It was hypothesized that cognitive vulnerability would be apparent in the contact sport groups compared to equivalent noncontact sport controls due to failure to benefit from practice on this task, as a consequence of a long-term history of many years of participation in a sport with a high risk of concussive insults, in interaction with the effects of any unrecognized concussive and subconcussive event sustained during the most recent season of participation in the sport (i.e. hidden within season effects), none of which would be characteristic of the
Visuomotor Vulnerability Niranjan A, Lunsford LD (eds): Concussion. Prog Neurol Surg. Basel, Karger, 2014, vol 28, pp 213–225 DOI: 10.1159/000358785
221
Downloaded by: Kellogg Health Sciences Libr. 129.173.72.87 - 5/19/2015 10:11:44 AM
Study 3: Shuttleworth-Edwards et al., 2014 [16]
Table 2. Combined sample: dependent t test comparisons of ImPACT Visual Motor Speed (VMS) scores at the pre- vs. postseason intervals for the rugby and noncontact sports groups Source
Sample size, n
Preseason mean ± SD
Postseason mean ± SD
t value
Effect size δ-value
95% CI
p value
Rugby VMS Noncontact VMS
145 106
36.57±6.80 36.65±7.24
36.77±6.62 38.39±6.70
–0.393 –3.967
–0.03 –0.39
–0.20 to 0.13 –0.58 to –0.19
0.695
