Archives of Clinical Neuropsychology 22 (2007) Accepted 22 June 2006

Transcription

1 Archives of Clinical Neuropsychology 22 (2007) The effects of mild and severe traumatic brain injury on speed of information processing as measured by the computerized tests of information processing (CTIP) Abstract Tom N. Tombaugh a,, Laura Rees a,b, Peter Stormer c, Allyson G. Harrison d, Andra Smith e a Carleton University, Psychology Department, Ottawa, Ont., Canada b The Rehabilitation Centre, Ottawa, Ont., Canada c Queens University, Psychology Department, Kingston, Ont., Canada d Rehabilitation Unit, Hawke s Bay District Health Board, Hastings, NZ e University of Ottawa, Ottawa, Ont., Canada Accepted 22 June 2006 In spite of the fact that reaction time (RT) measures are sensitive to the effects of traumatic brain injury (TBI), few RT procedures have been developed for use in standard clinical evaluations. The computerized test of information processing (CTIP) [Tombaugh, T. N., & Rees, L. (2000). Manual for the computerized tests of information processing (CTIP). Ottawa, Ont.: Carleton University] was designed to measure the degree to which TBI decreases the speed at which information is processed. The CTIP consists of three computerized programs that progressively increase the amount of information that is processed. Results of the current study demonstrated that RT increased as the difficulty of the CTIP tests increased (known as the complexity effect), and as severity of injury increased (from mild to severe TBI). The current study also demonstrated the importance of selecting a non-biased measure of variability. Overall, findings suggest that the CTIP is an easy to administer and sensitive measure of information processing speed National Academy of Neuropsychology. Published by Elsevier Ltd. All rights reserved. Keywords: Computerized tests of information processing; CTIP; TBI The potential value of using reaction time (RT) tests in neuropsychological evaluations was recognized as early as 1971 when Bruhn and Parsons (1971) stated the simple RT test has merit as a diagnostic tool equivalent to that found in more sophisticated and complex psychological tests (p. 614). More recent evidence from a variety of sources supports this contention and indicates that simple and choice reaction times provide a quick, yet easy and valid clinical tool for assessing cognitive status and should be incorporated into the neuropsychological assessment battery for traumatic This research was partially funded by a grant from the National Academy of Neuropsychology. The authors do not receive any current financial benefit from this publication. Corresponding author at: Psychology Department, Carleton University, 1125 Colonel By Drive, Ottawa, Ont., Canada K1S 5B6. Tel.: ; fax: address: tom tombaugh@carleton.ca (T.N. Tombaugh) /$ see front matter 2006 National Academy of Neuropsychology. Published by Elsevier Ltd. All rights reserved. doi: /j.acn

2 26 T.N. Tombaugh et al. / Archives of Clinical Neuropsychology 22 (2007) brain injury (TBI) (Braun, Daigneault, & Champagne, 1989; Elsass & Hartelius, 1985; Ferraro, 1996). Moreover, Bleiberg, Halpern, Reeves, and Daniel (1998) concluded that RT procedures reveal cognitive impairment even when normal performance is shown on traditional neuropsychological measures. Further support for using RT tests clinically is the fact that relatively high test retest reliability coefficients and split half coefficients are reported for RT tests (Godefroy, Lhullier, & Rousseaux, 1994; Hetherington, Stuss, & Finlayson, 1996; Stuss, Pogue, Buckle, & Bondar, 1994; Stuss et al., 1989). However, not all reaction time measures are equivalent or sensitive. Simple RT tests, which reflect the speed at which a stimulus is detected, are less sensitive to the cognitive sequelae of TBI than are choice or discrimination paradigms where responding is contingent on the information contained in two or more stimuli (Collins & Long, 1996; Hugenholtz, Stuss, Stethem, & Richard, 1988; Stuss et al., 1989). This differential sensitivity is referred to as the complexity effect. Since simple reaction time is commonly regarded as a pure measure of speed of information processing, it can be used as a covariate in choice RT analyses to control for decreased speed of processing in basic cognitive processes. Several studies have adopted this procedure and reported that the previously observed RT difference between Control and TBI patients disappeared (Brouwer, Ponds, van Wolffelaar, & van Zomeren, 1989; Felmingham, Baguley, & Green, 2004; Spikman, van Zomeren, & Deelman, 1996; Veltman, Brouwer, van Zomeren, & van Wolffelaar, 1996). This has led to the suggestion that TBI produces a generalized slowing of information processing that has a major impact on various other attentional and cognitive processes such as encoding, verbal comprehension, and adaptive responding to novel situations (Felmingham et al., 2004; Ferraro, 1996). The clinical utility of RT measures rests not only with the initial assessment for level of impairment, but also with tracking recovery. RTs have revealed that recovery of function occurs in cross sectional and longitudinal research, over short (3 6 months) and extended time periods (5 years versus 10 years), and with both mild and severe TBIs (Felmingham et al., 2004; Hetherington et al., 1996; Hugenholtz et al., 1988; MacFlynn, Montgomery, Fenton, & Rutherford, 1984; Spikman, Timmerman, van Zomeren, & Deelman, 1999; van Zomeren & Deelman, 1978; Zwaagstra, Schmidt, & Vanier, 1996). The lack of practice effects with most RT tests make them ideal measures for serial examinations. In spite of the above, reaction time procedures have seldom been used in standard clinical evaluations (Erlanger, Kunter, Barth, & Barnes, 1999). The failure to incorporate RT measures into clinical neuropsychological assessments stems from several sources. One reason is that surprisingly few experiments have directly compared the relative sensitivity of RT and standardized neuropsychological measures to detect cognitive deficits. The few comparative studies that are available suggest that RT measures are more sensitive to the long term effects of head injury than are most traditional neuropsychological tests (Bleiberg et al., 1998; Collins & Long, 1996; Maddocks & Saling, 1996). A second reason why RT measures are not commonly used in clinical assessments is the lack of adequately normed RT tests. This is attributed, at least in part, to the fact that until recently computer technology was not sufficiently advanced so that RT programs could be developed with sufficient temporal resolution (1 ms). Additionally, the computer programs that were available experimentally could not be readily adapted to the ordinary clinical environment. Recent advances in computer operating systems now makes it relatively easy to achieve a 1 ms resolution and most programs are compatible with existing operating systems. As a result of these advances, a panel of sports neuropsychologists have recommended the development and validation of computerized tests (Lovell & Collins, 1998). In view of the above, Tombaugh and Rees (2000) developed the computerized test of information processing (CTIP) to provide a clinical tool for evaluating the degree to which various neurological insults, primarily traumatic brain injury, affect the speed at which information is processed. In order to do this, a series of computerized programs were developed that progressively increase the amount of information processed. The most basic test, Simple RT, is often viewed as a pure speed of information processing measure and can serve as a baseline for other tests. Choice procedures were also included. The type of choice paradigm was deemed to be critical and two types of procedures were employed. The first paradigm involved concrete or literal processing where two choice stimuli remained the same over all trials. The second procedure involved conceptual/semantic processing where the items varied between trials and the choice decision required a semantic or lexical search. The selection of this paradigm was based on the semantic search paradigm used in cognitive psychology (Chang, 1986; Loftus, 1973) and guided by neuropsychological research showing that a TBI decreases semantic processing (Haut, Petros, Frank, & Lamberty, 1990; Hinton-Bayre, Geffen, & McFarland, 1997; Levin & Goldstein, 1986; Timmerman & Brouwer, 1999; Wilson et al., 1999).

3 T.N. Tombaugh et al. / Archives of Clinical Neuropsychology 22 (2007) Method 1.1. Participants The study involved three groups: Control (n = 60), Mild TBI (n = 59), and Severe TBI (n = 44). All participants in the Mild TBI group had a Glasgow Coma Scale (GCS) >13 or loss of consciousness (LOC) <5 min and were tested within approximately 1 month of the injury as part of an ongoing study at a local hospital. Some patients received CND$ for their participation. Nine members were involved in assaults, 28 in falls, 13 in motor vehicle accidents (MVAs), 7 in sporting accidents, and 2 unspecified. All Mild TBI participants achieved a score greater than a malingering cut-off score of 18 correct on the 21-Word Test (Iverson, Frazen, & McCracken, 1991). Participants in the Severe TBI group were experiencing some type of significant cognitive problem and were recruited from local neuropsychologists and physiatrists. No GCS scores were available. LOC information was available from 25 patients and ranged from brief to 180 days. A considerable range of time had elapsed from the time of injury to test (mean = 3.3 years, range = 22 days to 25 years). Twenty-seven members of the group were involved in MVAs, 14 in falls, 1 in assaults, 1 in sporting accidents, and 1 unspecified. Ten of the Severe TBI group were administered the TOMM (Tombaugh, 1997) and scored higher than the 45 correct criterion score. No Mild or Severe TBI patient was involved in litigation or disability claim. Members of the Control group were recruited from the community and were selected to provide a sample that closely matched the TBI groups in age and education. English was the first language for all participants Procedure The CTIP (Tombaugh & Rees, 2000) was administered in the context of a flexible battery of neuropsychology tests which contained the following core set of tests: Trails Making Test Making A and B (Reitan & Wolfson, 1985), Digit Span (Schmdit & Tombaugh, 1995) and either Digit Symbol Substitution Test (Wechsler, 1981) or Symbol Digit Modality Test (Smith, 1982). Demographics for all participants were obtained prior to testing, as was a medical history from the both groups. All participants read and signed an informed consent form. Testing took approximately 2 h. 2. Materials 2.1. Computerized tests of information processing Simple RT This program measured the amount of time required to process and to react to a simple stimulus and served as the baseline measure for the other tests. On each of 10 practice and 30 test trials, the space bar was pressed as soon as a single stimulus (X) appeared in the centre of the screen Choice RT The time required to process two bits of information and respond differentially was measured by requiring the person to process one of two words on each of 10 practice and 30 test trials, and then deciding which of two keys should be pressed ( DUCK = right key; KITE = left key) Semantic RT Semantic RT required the person to decide if a word belonged to a specific category (conceptual/semantic processing). On each trial, one of four categories (Weapon, Furniture, Bird, or Fruit) was randomly presented on the screen. The category names remained presented on the screen and then a word appeared below the category that either represented (press right key) or did not represent (press left key) a member of that category. Ten practice and 30 trials were used. 3. Results 3.1. Demographics Demographic information and number of days post-injury are shown in Table 1. One-way analyses of variance (ANOVA) showed the groups were matched on age, number of years of education, and gender, age: F(2,161) = 1.04,

4 28 T.N. Tombaugh et al. / Archives of Clinical Neuropsychology 22 (2007) Table 1 Age, education, gender (m/f), and days post-accident Group Controls Mild TBI Severe TBI n Age (18.53) (17.23) (14.61) Education (1.74) (2.37) (2.14) Gender (m/f) 31/29 39/20 25/19 Days post-accident Mean (S.D.) (60.26) ( ) Median Range p >.05; education: F(2,161) = 1.05, p >.05; gender: F(2,161) = 1.58, p >.05. The duration of time between the accident and time of testing was significantly shorter for the Mild TBI group than for the Severe TBI group, F(1,102) = 23.66, p < Response latencies The median response latency for each participant was employed to minimize the effects of outliers. Group mean RT scores on each CTIP test are shown in Fig. 1. A repeated measures ANOVA showed all effects were significant, Group: F(2,161) = 18.54, p <.001; Test: F(2,322) = , p <.001; Group Test: F(4,322) = 12.49, p <.001. The significant interaction reflected the increasing differences among the groups as the CTIP tests became progressively more difficult and the severity of the TBI injury became more pronounced. A series of one-way ANOVAs revealed significant group differences for each CTIP test, Simple: F(2,161) = 10.17, p <.001; Choice: F(2,161) = 14.16, p <.001; Semantic: F(2,161) = 17.40, p <.001. Multiple comparisons using the Bonferroni correction procedure (p <.05) revealed that differences between the Severe TBI group and the Control and Mild TBI groups were highly significant for each CTIP test, but the differences between the Control and Mild TBI group was only significant on the Semantic RT test. Since large differences in variance were observed for the TBI groups, two additional non-parametric analyses were performed on each CTIP test. The first analysis used the rank ordering of scores from each CTIP test as the input data for the ANOVAs. The F-values from these analyses were similar to those observed with the RT scores Simple: F(2,161) = 11.88, p <.001; Choice: F(2,161) = 17.94, p <.001; Semantic: F(2,161) = 17.60, p <.001. In addition, the Kruskal Wallis test was employed. This distribution-free test uses a Chi-square rather than an F distribution. Again, results similar to those reported with the one-way ANOVAs were obtained Simple: Chi-square = 20.57, d.f. = 2, p <.001; Choice: Chi-square = 29.60, d.f. = 2, p <.001; Semantic: Chi-square = 29.22, d.f. = 2, p <.001. Since all three types of analyses produced essentially the same results, only the results from ANOVAs will be reported for the remaining data analyses. Fig. 1. Mean reaction time (ms) on the computerized tests of information processing for Controls, Mild TBI, and Severe TBI.

5 Table 2 Paired comparisons for different percentile cut values from norms T.N. Tombaugh et al. / Archives of Clinical Neuropsychology 22 (2007) Tests <50th percentile <25th percentile <10th percentile Percent Mild TBI Severe TBI Percent Mild TBI Severe TBI Percent Mild TBI Severe TBI SRT Normals 49 ns ns Mild TBI ns 41 ns Severe TBI Choice Normals Mild TBI 75 ns 61 ns Severe TBI Semantic Normals Mild TBI 69 ns 59 ns 51 ns Severe TBI To determine the degree to which the speed of responding on the Simple RT could account for the differences on the other two tests, it was used as a covariant. This procedure reduced but did not eliminate the main effects on Choice RT (F(2,161) = 3.65, p <.03) and Semantic RT (F(2,161) = 7.10, p <.001). Multiple comparisons using the Bonferroni correction procedure (p <.05) revealed that the difference between Controls and Severe TBI were significant for both the Choice RT and Semantic RT tests, but the difference between Severe TBI and Mild TBI was only significant on the Semantic RT test Number of correct responses In addition to analyzing latency scores, accuracy of responding was examined in the two tests where the participants were required to select the appropriate stimulus. Highly accurate performance of 97% (i.e., 1 error) occurred for each group on the Choice and Semantic RT tests. One-way ANOVAs failed to yield any significant differences, Choice: F(2,161) =.13, p >.05; Semantic: F(2,161) = 1.89, p > Normative percentiles The previous analyses have focused on comparing groups but have neglected to provide information that the CTIP serves a useful clinical function. One way to provide this information is to compare the scores from the individual groups to percentiles values (50th, 25th, and 10th) derived from normative data. Since the CTIP norms consisted of 320 individuals, aged years (Mean = 29.2, S.D. = 13.70), the scores used for the current comparison were from the same age range (Controls: mean age = 27.14, S.D. = 12.60; Mild TBI: mean age = 32.69, S.D. = 14.08; Severe TBI: mean age = 36.14, S.D. = 14.05). The percentage of each group that fell at each of the three percentiles is shown in Table 2, along with the results from pair-wise comparisons. At each percentile, the proportion of individuals increased as the severity of injury increased. The cognitive load also was a factor, but had a much lower impact than anticipated Different types of injuries Several studies have indicated that mild TBI produced by MVAs result in greater rotation and acceleration/deceleration of the brain than do falls and assaults. Consequently, MVAs are putatively more likely to produce cognitive dysfunctions than are falls and/or assaults (Ommaya & Gennarelli, 1974; Roman, Edwall, Buchanan, & Patton, 1991). One-way ANOVAs for the Mild and Severe TBI subgroups failed to reveal any significance differences among the four subgroups (assaults, falls, MVAs, and sports), Mild TBI Simple: F(3,56) = 2.46, p >.05; Choice: F(3,56) = 1.97, p >.05; Semantic: F(3,56) = 1.98, p >.05; Severe TBI Simple: F(1,40) =.00, p >.05; Choice: F(1,40) = 1.58, p >.05; Semantic: F(1,40) = 2.15, p >.05.

6 30 T.N. Tombaugh et al. / Archives of Clinical Neuropsychology 22 (2007) Practice effects In order to determine if there were any practice effects, the scores were grouped in three, 10-trial blocks. Repeated measures ANOVAs showed that performance was constant over the three blocks of trials. These results clearly illustrate that performance on the CTIP tests did not show any within session practice effects, Simple RT Group: F(2,161) = 10.25, p <.00l; Blocks: F(2,322) =.35, p >.05; Group Block: F(2,322) = 1.46, p >.05; Choice RT Group: F(2,161) = 13.18, p <.00l; Blocks: F(2,322) = 1.76, p >.05; Group Block: F(2,322) = 1.56, p >.05; Semantic RT Group: F(2,161) = 12.83, p <.00l; Blocks: F(2,322) = 2.20, p >.05; Group Block: F(2,322) = 1.21, p > Variability 4.1. Semi inter-quartile range/median score (Q/Med) CTIP scores were also examined to determine the relative amount of variability that occurred among the three groups. Traditionally standard deviation (S.D.) scores have been used to measure variability. Using S.D. scores in a repeated measure ANOVA yielded the following, Group: F(2,161) = 13.02, p <.001; Test: F(2,161) = , p <.001; Group Test: F(4,322) = 5.69, p <.001. One-way ANOVAs on each test revealed significant variability effects on each CTIP test, Simple RT: F(2,161) = 4.46, p <.05; Choice RT: F(2,161) = 8.45, p <.001; Semantic RT: F(2,161) = 12.05, p <.001. Paired comparisons using the Bonferroni correction procedure (p <.05) yielded significant differences between the Severe TBI group and Control group on each test and between the Severe TBI group and Mild TBI group on the Semantic RT test. However, van Zomeren and Brouwer (1987) reported that S.D. scores are often proportional to increases in reaction times and suggested using the ratio of semi-interquartile values to median values (Q/Med) to control for the corre- Fig. 2. Mean reaction time (ms) for nine different percentiles on the computerized tests of information processing for Controls, Mild TBI, and Severe TBI.

7 T.N. Tombaugh et al. / Archives of Clinical Neuropsychology 22 (2007) lation between RTs and S.D. When the Q/Med measure was submitted to a repeated measure ANOVA many of the previously reported differences were reduced disappeared Group: F(2,161) = 4.40, p <.05; Test: F(2,161) = 91.88, p <.001; Group Test: F(4,322) =.42, p >.05. Subsequent one-way ANOVAs and pair-wise comparisons using the Bonferroni correction procedure (p <.05) revealed that the increased variability only occurred on the Simple RT and was limited to a difference between the Severe TBI group and Control group, Simple RT: F(2,161) = 4.47, p <.05; Choice RT: F(2,161) = 2.15, p >.05; Semantic RT: F(2,161) = 1.54, p > Percentile analysis The Q/Med measure provides a relatively clear picture of the variability with the extreme scores eliminated. However, examining only the outlying scores may also provide useful information. A percentile ranking system was used which allowed variability to be examined at the extreme ends of the distribution and at the intermediate levels. Fig. 2 shows the reaction times associated with the different percentile ranks for each CTIP test. Repeated measures ANOVAs on each CTIP test revealed that RT scores increased over the percentiles with a progressively greater increase occurring for the Mild and Severe TBI groups, Simple RT Group: F(2,161) = 9.95, p <.001; Percentile: F(8,1288) = , p <.001; Group Percentile: F(16,1288) = 8.25, p <.001; Choice RT Group: F(2,161) = 14.36, p <.001; Percentile: F(8,1288) = , p <.001; Group Percentile: F(16,1288) = 9.37, p <.001; Semantic RT Group: F(2,161) = 15.72, p <.001; Percentile: F(8,1288) = , p <.001; Group Percentile: F(16,1288) = 10.13, p <.001. To determine if there was a difference in only the extreme scores, one-way ANOVAs were performed over the 10th percentile scores (i.e., those outlying scores having the longest latencies). All results were significant, Simple: F(2,161) = 11.84, p <.001; Choice: F(2,161) = 12.71, p <.001; Semantic: F(2,161) = 11.70, p <.001: pair-wise comparisons using the Bonferroni correction procedure (p <.05) revealed significant differences between all groups except between the Mild TBI and Control groups on Simple RT and between the Mild TBI and Severe TBI groups on Semantic RT. Table 3 Correlations between CTIP and neuropsychological tests Groups Tests Simple RT Choice RT Semantic RT Trails A Trails B Digit Symbol Tests Normals Choice RT.57 ** Semantic RT.52 **.73 ** Trails A Trials B ** Digit Symbol **.34 ** Digit Span Mild TBI Choice RT.65 ** Semantic RT.62 **.73 ** Trails A **.16 Trials B.30 *.45 ** ** Digit Symbol **.31 *.48 **.39 ** Digit Span **.33 * Severe TBI Choice RT.59 ** Semantic RT ** Trails A Trials B ** Digit Symbol **.70 ** Digit Span * p <.05. ** p <.01.

8 32 T.N. Tombaugh et al. / Archives of Clinical Neuropsychology 22 (2007) Neuropsychological tests and the CTIP Scores from the neuropsychological tests were converted to percentile scores using the appropriate norms. ANOVAs performed on the percentile scores for the neuropsychological tests yielded significant group differences for each test, Trails A: F(2,150) = 4.49, p <.01; Trails B: F(2,150) = 5.78, p <.01; Digit Symbol: F(2,135) = 8.41, p <.001; Digit Span (F + B): F(2,150) = 10.49, p <.001. The significant effects for Trails B disappeared when the scores from Trails A were used as a covariant, F(2,150) = 2.47, p >.05. Pair-wise comparisons using the Bonferroni correction procedure (p <.05) revealed that the Severe TBI group was significantly different on all neuropsychological tests (except Trails A) from the Mild TBI and Control groups which were not significantly different from each other. On Trails A, the only significant difference was between the Severe TBI group and the Mild TBI group. Correlational analyses (Pearson s r) were performed to determine the relationships between scores on the CTIP and those on the neuropsychological tests (see Table 3). Inspection of Table 3 clearly shows that scores on the CTIP tests were highly intercorrelated but they were not correlated with any of the neuropsychological tests for the Control and Severe TBI groups. The scores on the CTIP were moderately correlated with neuopsychological measures for the Mild TBI group. 6. Discussion The present results clearly demonstrated that RT increased as the difficulty of the CTIP tests increased. This effect, commonly referred to as the complexity effect, has been reported repeatedly when the simple RT procedure is changed to require an element of choice. The most popular way to experimentally increase complexity has been to vary the number of stimuli by using a four-choice, reaction time procedure similar to that introduced by van Zomeren (1981). Research using this procedure has universally demonstrated that (1) decision time and movement time can be dissociated and (2) choice decision time is more sensitive to TBI than simple reaction time (Miller, 1970; Ponsford & Kinsella, 1992; van Zomeren & Brouwer, 1994; Veltman et al., 1996; Zwaagstra et al., 1996). Although this procedure has generated numerous research findings showing that speed of information processing decreases with severity of TBI injury, it has the practical disadvantage of requiring a specific type of apparatus that may be difficult to transport from one location to another. Moreover, the choice among different response buttons necessitates measuring two types of response latencies, one putatively measuring decision time and the other movement. Finally, decision time only reflects the type of concrete or literal processing that is associated with varying the physical or numerical properties of the stimulus. Manipulations of this type that require the individual to process the form of the stimulus generally involve a minimal amount of internal search and processing and, as such, are regarded as representing a relatively light cognitive load. In contrast, the CTIP employs two types of choice procedures involving different cognitive loads while maintaining the same two-item response requirement (left key versus right key). This was achieved by using choice paradigms that require (1) concrete/literal processing where the decision was based on the form of the two stimuli which were repeatedly administered (i.e., Choice RT) and (2) conceptual/semantic processing where processing of the meaning of a changing set of stimuli was required (i.e., Semantic RT). The cognitive processes involved in searching through a semantic lexicon to determine if the exemplar is from the same or different category are more extensive and require more extensive internal processing than did the Choice RT procedure which only required recognition of the stimulus form ( Kite versus Duck ) before selecting the appropriate response key. Thus, the complexity effect demonstrated with the CTIP represents performance of TBI patients over a continuum of cognitive difficulty ranging from Simple RT through Choice RT to Semantic RT and provides evidence that speed of information processing decreased with increased severity of injury. Examination of each of the CTIP tests was undertaken to further determine the degree to which each test was sensitive to the effects of TBI. Simple RT and Choice RT tests did not reveal any significant differences between the Mild TBI and Control groups. Only the Semantic RT yielded significant differences between these two groups. On the other hand, all three tests provided evidence that speed of information processing was slower for the Severe TBI group than for the Mild TBI and Control groups. Thus, it appears that all three tests were not equally capable of detecting performance differences between the different groups. The inclusion of the Semantic RT test was particularly critical in detecting cognitive deficits in the Mild TBI group. Timmerman and Brouwer (1999), using a similar categorization task, reported that the difference between severe TBI patients and controls increased as the category exemplar became

9 T.N. Tombaugh et al. / Archives of Clinical Neuropsychology 22 (2007) more difficult. They concluded that this was because these words took proportionally longer to retrieve from semantic memory. Further evidence showing the sensitivity of semantically based paradigms to TBI has been reported using depth of processing procedures (Goldstein, Levin, Boake, & Lohrey, 1990), speed of sentence comprehension (Comerford, Geffen, May, Medland, & Geffen, 2002; Hinton-Bayre et al., 1997; Wilson et al., 1999), Sternberg paradigm (Haut et al., 1990), and category naming tasks (Maring, Deelman, & Bouwer, 1984). Overall, the results are consistent with the position that slowness of information processing occurs with TBI and increases with the severity of the injury. Several authors have speculated that this deficit in information process is related to diffuse axonal injury (DAI) that typically occurs with TBI (Felmingham et al., 2004; van Zomeren & Brouwer, 1994). The diffuse injury to white matter tracts presumably reduces the interconnections between neural networks, thereby reducing the speed at which information is transmitted. Moreover, the slowing in the speed of information processing is often assumed to represent a generalized slowing which is observed at the most basic level and underlies the increased latencies of more complex tasks (Sarno et al., 2003; Spikman et al., 1996; van Zomeren & Brouwer, 1994, Zahn & Mirsky, 1999). However, when the effects of the general slowing are statistically controlled by using the simple RT scores as a covariate, the increased latencies on the more complex task were removed for the Severe TBI participants (Felmingham et al., 2004; Spikman et al., 1996). The finding in the present experiment that performance differences occurred on the Simple RT raises the question as to whether these initial differences in speed of information processing could also explain the differences on the two more complex tests. Covariate analyses, using Simple RT scores as the covariate, eliminated the differences for the Mild TBI group indicating that the injuries resulted in a minor reduction in basic speed of information processing which contributed significantly to performance on the two choice tasks. However, the covariant analyses greatly reduced, but did not eliminate the effects for the Severe TBI group. The reduction was greater for Choice RT than Semantic RT. The failure of these results to replicate the previous findings of Felmingham et al. (2004) and Spikman et al. (1996) probably reflects the use of different choice procedures. Both previous studies used a procedure similar to the current Choice RT. As previously indicated, this paradigm requires concrete/literal processing with few cognitive demands which probably reflected the decreased speed of information processing observed on the Simple RT. The conceptual processing required in the Semantic RT task employed a greater number of interrelated cognitive components, each of which was affected by the generalized decrease in processing speed and which summed together to produce an overall decline in performance. This position is also consistent with Timmerman and Brouwer s (1999) speculation that diffuse axonal injury associated with TBI disrupts the spread of activation among the nodes of semantic memory thereby increasing RT. In a more general sense, the above results argue strongly for employing a series of tests that have a processing continuum that reflects the degree of cognitive effort, where a baseline measure can be used as a measure of pure speed and which can be used as a statistical control to better estimate effects that increasing cognitive load has on performance. The effects of TBI on variability of responding was also addressed in the current study. A critical issue in this determination was selecting a non-biased measure of variability. When standard deviations of reaction times were used, variability increased on each test with most of this effect due to the performance of the Severe TBI group. Similar findings have been reported by other investigators (Collins & Long, 1996; Stuss et al., 1989, 1994; Zahn & Mirsky, 1999). However, van Zomeren and Brouwer (1987) noted that RT and S.D. scores are often correlated. Consequently, any increased variability attributed to a TBI that is based on S.D. scores may be an artifact of the increased RT scores rather than reflecting a genuine inconsistency effect. In view of this, the present study followed the suggestion of van Zomeren and Brouwer (1987) and used the ratio between the semi-interquartile range and median score as a statistic to control for differences in overall speed of responding. Since this measure uses the difference between the 25th and 75th percentile, it is not affected by the extreme scores at the tails of the distribution and represents a relatively stable measure of inconsistency of behavior rather than merely measuring momentary lapses of attention. Analyses employing this measure eliminated most of the variability effects reported with S.D. except an increase in variability between the Severe TBI and the Control group on the Simple RT test. Other investigators using the same ratio between interquartile deviation and median scores have also reported that TBI did not increase variability in a four-choice visual RT test (van Zomeren and Brouwer, 1987) or on any of the five computerized tests on the automated neuropsychological assessment metrics (ANAM; Bleiberg, Garmoe, Halpern, Reeves, & Nadler, 1997; Reeves, Thorne, Winter, & Hegge, 1989). Thus, it appears that the increased variability associated with absolute measures such as S.D. can be largely attributable to a proportional increase in RTs associated primarily with more severe TBIs. The effects of TBI on variability were further explored by using the scores which represented the extreme end of the distribution of response scores (i.e., 10th percentile). This measure showed that variability was associated with the

10 34 T.N. Tombaugh et al. / Archives of Clinical Neuropsychology 22 (2007) severity of injury and suggests that much of the variability attributed to TBI is the result of extreme scores that probably reflected momentary lapses of attention. Within this context, it should be noted that Rabbitt and Banerji (1989) reported that prolonged practice on a four-choice RT task produced improvements in speed of responding that were mainly due to a substantial reduction in the skews of the distribution of RT scores. Several authors have recommended that clinical neuropsychological evaluations should contain reaction time measures (Braun et al., 1989; Bruhn & Parsons, 1971; Collins & Long, 1996). The present results support this suggestion and provide additional evidence that the CTIP holds considerable promise as a clinical neuropsychological test that measures speed of information processing. The CTIP utilizes a continuum of RT tests that is progressively more cognitively demanding, with Simple RT serving as a baseline measures for the two choice procedures therefore controlling for generalized cognitive slowing, movement difficulties, etc. The CTIP does not require any additional equipment as it is a computer-based program with Windows-compatible software. Moreover, there is no need to separate decision time (i.e., choice) from movement time as is necessary when complexity is manipulated by varying number of response alternatives. Perhaps most clinically relevant is the ability of different percentiles derived from normative data to classify participants in the expected manner. Even though the TBI produced significant effects on the traditional neuropsychological tests, correlation analyses revealed little overlap between the CTIP and the more traditional measure of attention, particularly for the Controls and Severe TBI groups. Thus, the CTIP appears to be measuring a different aspect of attention, presumably speed of information processing. Consequently, comparing the scores from the CTIP with those obtained from Trails, Digit Span, and Digit Symbols offers the clinician a potential avenue for determining how various neurological illnesses affect different types of attentional processes. Previous research has reported that the CTIP is a reliable test possessing high test retest coefficients, and it is ideally suited for tracking recovery since it is not sensitive to the effects of repeated administrations occurring within or between sessions (Baird, Tombaugh, & Francis, in press; Willison & Tombaugh, 2006). The present results provide additional evidence that performance on the CTIP is not affected by prior experience and that it is very stable over a single testing session. That is, practice effects measured over three 10-trial blocks did not occur for either the control or TBI participants. Acknowledgements We wish to thank Drs. Marshall and Zaharia for their assistance in collecting the data for the TBI patients and Denise Wozney, BA, for her assistance in collecting data for the control subjects. References Baird, B., Tombaugh, T. N., & Francis, M. (in press). The effects of practice on speed of information processing using the adjusting-psat (A-PSAT) and the computerized test of information processing (CTIP). Journal of Applied Neuropsychology. Bleiberg, J., Garmoe, W. S., Halpern, E. L., Reeves, D. L., & Nadler, J. D. (1997). Consistency of within-day and across-day performance after mild brain injury. Neuropsychiatry, Neuropsychological Behavioral Neurology, 10, Bleiberg, J., Halpern, E. L., Reeves, D., & Daniel, J. C. (1998). Future directions for the neuropsychological assessment of sports concussion. Journal of Head Trauma Rehabilitation, 13, Braun, D. M. J., Daigneault, S., & Champagne, D. (1989). Information processing deficits as indexed by reaction time parameters in severe closed head injury. International Journal of Clinical Neuropsychology, 11, Brouwer, W. H., Ponds, R. W., van Wolffelaar, P. C., & van Zomeren, A. H. (1989). Divided attention 5 to 10 years after severe closed head injury. Cortex, 25, Bruhn, P., & Parsons, O. A. (1971). Continuous reaction time in brain damage. Cortex, 7, Chang, T. M. (1986). Semantic memory: Facts and models. Psychological Bulletin, 99, Collins, L. F., & Long, C. J. (1996). Visual reaction time and its relation to neuropsychological test performance. Archives of Clinical Neuropsychology, 11, Comerford, V. E., Geffen, G. M., May, C., Medland, S. E., & Geffen, L. B. (2002). A rapid screen of the severity of mild traumatic brain injury. Journal of Clinical and Experimental Neuropsychology, 24, Elsass, P., & Hartelius, H. (1985). Reaction time and brain disease: Relation to location, etiology and progression of cerebral dysfunction. Acta Neurologica Scandinavica, 71, Erlanger, D. M., Kutner, K. C., Barth, J. T., & Barnes, R. (1999). Neuropsychology of sports-related head injury: Dementia pugilistica to post concussion syndrome. The Clinical Neuropsychologist, 13,

11 T.N. Tombaugh et al. / Archives of Clinical Neuropsychology 22 (2007) Felmingham, K. L., Baguley, I. J., & Green, A. S. (2004). Effects of diffuse axonal injury on speed of information processing following severe traumatic brain injury. Neuropsychology, 18, Ferraro, F. R. (1996). Cognitive slowing in closed-head injury. Brain Cognition, 32, Godefroy, O., Lhullier, C., & Rousseaux, M. (1994). Reliability of reaction time measurement in brain-damaged patients. Journal of the Neurological Sciences, 126, Goldstein, F. C., Levin, H. S., Boake, C., & Lohrey, J. H. (1990). Facilitation of memory performance through induced semantic processing in survivors of severe closed-head injury. Journal of Clinical and Experimental Neuropsychology, 12, Haut, M. W., Petros, T. V., Frank, R. G., & Lamberty, G. (1990). Short-term memory processes following closed head injury. Archives of Clinical Neuropsychology, 5, Hetherington, C. R., Stuss, D. T., & Finlayson, M. A. (1996). Reaction time and variability 5 and 10 years after traumatic brain injury. Brain Injury, 10, Hinton-Bayre, A. D., Geffen, G., & McFarland, K. (1997). Mild head injury and speed of information processing: A prospective study of professional rugby league players. Journal of Clinical and Experimental Neuropsychology, 19, Hugenholtz, H., Stuss, D. T., Stethem, L. L., & Richard, M. T. (1988). How long does it take to recover from a mild concussion? Neurosurgery, 22, Iverson, G. L., Franzen, M. D., & McCracken, L. M. (1991). Evaluation of an objective assessment technique for the detection of malingering memory deficit. Law and Human Behavior, 15, Levin, H. S., & Goldstein, F. C. (1986). Organization of verbal memory after severe closed-head injury. Journal of Clinical and Experimental Neuropsychology, 8, Loftus, E. F. (1973). Category dominance, instance dominance, and categorization time. Journal of Experimental Psychology, 97, Lovell, M. R., & Collins, M. W. (1998). Neuropsychological assessment of the college football player. Journal of Head Trauma Rehabilitation, 13, MacFlynn, G., Montgomery, E. A., Fenton, G. W., & Rutherford, W. (1984). Measurement of reaction time following minor head injury. Journal of Neurology, Neurosurgery, and Psychiatry, 47, Maddocks, D., & Saling, M. (1996). Neuropsychological deficits following concussion. Brain Injury, 10, Maring, W., Deelman, B. G., & Bouwer, W. H. (1984). Improving deficient memory in closed head injury patients. The INS Bulletin, 24. Miller, E. (1970). Simple and choice reaction time following severe head injury. Cortex, 6, Ommaya, A. K., & Gennarelli, T. A. (1974). Cerebral concussion and traumatic unconsciousness: correlation and clinical observations on blunt head injuries. Brain, 97, Ponsford, J., & Kinsella, G. (1992). Attentional deficits following closed-head injury. Journal of Clinical and Experimental Neuropsychology, 14, Rabbitt, P., & Banerji, H. (1989). How does very prolonged practice improve speed? Journal of Experimental Psychology: General, 118, Reeves, D., Thorne, R., Winter, A., & Hegge, F. (1989). The United Tri-Service Cognitive Performance Assessment Battery (UTC-PAB). Report San Diego, California: US Naval Aerospace Medical Research Laboratory and Walter Reed Army Institute of Research. Reitan, R. M., & Wolfson, D. (1985). The Halstead Reitan neuropsychological test battery. Tucson, Arizona: Neuropsychology Press. Roman, D. D., Edwall, G. E., Buchanan, R. J., & Patton, J. H. (1991). Extended norms for the Paced Auditory Serial Addition Task. The Clincial Neuropsychologist, 5, Sarno, S., Erasmus, L., Lipp, B., & Schlaegel, W. (2003). Multisensory integration after traumatic brain injury: A reaction time study between pairings of vision, touch and audition. Brain Injury, 17, Schmdit, J. P., & Tombaugh, T. N. (1995). The learning and memory battery (LAMB). Toronto: Multi-Health Systems. Smith, A. (1982). Symbol Digit Modalities Test Manual, Revised Los Angeles: Western Psychological Services. Spikman, J. M., Timmerman, M. E., van Zomeren, A. H., & Deelman, B. G. (1999). Recovery versus retest effects in attention after closed head injury. Journal of Clinical and Experimental Neuropsychology, 21, Spikman, J. M., van Zomeren, A. H., & Deelman, B. G. (1996). Deficits of attention after closed-head injury: Slowness only? Journal of Clinical and Experimental Neuropsychology, 18, Stuss, D. T., Pogue, J., Buckle, L., & Bondar, J. (1994). Characterization of stability of performance in patients with traumatic brain injury: Variability and consistency on reaction time tests. Neuropsychology, 8, Stuss, D. T., Stethem, L. L., Hugenholtz, H., Picton, T., Pivik, J., & Richard, M. T. (1989). Reaction time after head injury: Fatigue, divided and focused attention and consistency of performance. Journal of Neurology, Neurosurgery, and Psychiatry, 52, Timmerman, M. E., & Brouwer, W. H. (1999). Slow information processing after very severe closed head injury: Impaired access to declarative knowledge and intact application and acquisition of procedural knowledge. Neuropsychologia, 37, Tombaugh, T. N. (1997). The Test of Memory Malingering (TOMM): Normative data from cognitively intact and cognitively impaired individuals. Psychological Assessment, 9, Tombaugh, T. N., & Rees, L. (2000). Manual for the computerized tests of information processing (CTIP). Ottawa, Ont.: Carleton University. van Zomeren, A. H. (1981). Reaction time and attention after closed head injury. Lisse, The Netherlands: Swets and Zeitlinger. van Zomeren, A. H., & Brouwer, W. H. (1987). Head injury and concepts of attention. In H. S. Levin, J. Grafman, & H. M. Eisenberg (Eds.), Neurobehavioral recovery from head injury (pp ). New York: Oxford University Press. van Zomeren, A. H., & Brouwer, W. H. (1994). Clinical neuropsychology of attention. New York: Oxford University Press. van Zomeren, A. H., & Deelman, B. G. (1978). Long-term recovery of visual reaction time after closed head injury. Journal of Neurology, Neurosurgery, and Psychiatry, 4, Veltman, J. C., Brouwer, W. H., van Zomeren, A. H., & van Wolffelaar, P. C. (1996). Central executive aspects of attention in subacute severe and very severe closed head injury patients: Planning, inhibition, flexibility and divided attention. Neuropsychology, 19,

12 36 T.N. Tombaugh et al. / Archives of Clinical Neuropsychology 22 (2007) Wechsler, D. (1981). Wechsler Adult Intelligence Scales Revised. New York: Psychological Corporation. Willison, J., & Tombaugh, T. N. (2006). Detecting malingering of attention deficits using reaction time tests. Archives of Clinical Neuropsychology, 21, Wilson, B. A., Evans, J. J., Emslie, H., Balleny, H., Watson, P. C., & Baddeley, A. D. (1999). Measuring recover from post traumatic amnesia. Brain Injury, 13, Zahn, T. P., & Mirsky, A. F. (1999). Reaction time indicators of attention deficits in closed head injury. Journal of Clinical and Experimental Neuropsychology, 21, Zwaagstra, R., Schmidt, I., & Vanier, M. (1996). Recovery of speed of information processing in closed-head-injured patients. Journal of Clinical and Experimental Neuropsychology, 18,