Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Reference Data from the Automated Neuropsychological Assessment Metrics for Use in Traumatic Brain Injury in an Active Duty Military Sample

Reference Data from the Automated Neuropsychological Assessment Metrics for Use in Traumatic... ABSTRACT The current study examined the performance of active duty soldiers on the Automated Neuropsychological Assessment Metrics (ANAM) traumatic brain injury test battery, to expand the reference data for use in military settings. The effects of age and gender on cognitive performance also were explored. The ANAM traumatic brain injury battery, consisting of six performance tests and two subjective scales, was administered to a sample of healthy active duty soldiers (N = 5,247) as part of a concussion surveillance program. Performance means and SDs, stratified according to age and gender, are reported as reference data. In addition, the impact of age and gender on performance measures was analyzed. Because ANAM is rapidly being adopted for use in many military medical and research applications, the establishment of these reference values is invaluable, particularly for assisting with rapid accurate evaluation and treatment in clinical settings. INTRODUCTION Recent advances in clinical neuroscience have generated rapid expansion in the knowledge of brain functioning and in the methods for treating injury and disease. Many of these advances have been made possible by the introduction of computer-based test systems that provide rapid, objective, and repeatable neuropsychological assessments.1 These computer-based test systems have evolved from research in the disciplines of human factors, cognitive psychology, sensation-perception, and industrial engineering, combined with research and standards of practice in psychiatry, clinical psychology, neuropsychology, and neurology.2 An example of the more accomplished and enduring computer-based test systems is the Automated Neuropsychological Assessment Metrics (ANAM).3 The ANAM test system had its origin in a series of computer test batteries developed by the Department of Defense in the 1980s to assess the influence of pharmaceutical countermeasures for chemical defense. These computer-based test systems were quickly applied to investigate a broad range of risk factors, such as environmental stressors (e.g., heat, cold, hypoxia, and g-force), sleep loss, fatigue, shift work, and drugs.4,5 Currently, ANAM is the most comprehensive and sophisticated test system to evolve from this line of computer-based test development, and it is enhanced by the inclusion of the NATO Advisory Group for Aeronautical Research and Development standards for the development of computer-based test batteries.6 As a result, ANAM has the distinct advantage of providing tests that are comparable to many precursor test systems with one important advancement; ANAM not only embodies some of the best examples of traditional, laboratory-based, human performance assessment metrics but also was designed with enhanced relevance for neuropsychological assessment.7 Construct validity studies demonstrated that ANAM findings correlated well with traditional neuropsychological measures.8,10 This allows laboratory researchers and clinicians to compare and to assess their results not only with respect to ANAM normative values but also with respect to the broad literature bases in these fields. As a result, ANAM subsets of tests have been selected and configured to form stable or preset batteries to address common screening or assessment needs. Alternatively, ANAM provides the flexibility to customdesign batteries for almost any research or clinical application. One area in which ANAM has been especially useful is in the assessment and management of traumatic brain injury (TBI).11,12 The impetus for much of this progress has come from heightened awareness of activities that often result in head injuries, such as explosive war injuries, vehicle accidents, and sports concussions, and the realization that TBIs are more pervasive and pernicious than once thought.13 For example, it is estimated that each year ∼ 1.4 million people in the United States sustain a TBI14 and ∼2% of the U.S. population needs some form of ongoing assistance in activities of daily living as a result of past TBIs.15,16 Although research suggests that TBI is a health threat most often among the young, its incidence is registered across all levels of the population and even more acutely among individuals who are inherently more vulnerable to TBI, such as those serving in the military.17 This study reports the critical results of a multiyear investigation focused on the extended use of a select ANAM battery for the assessment of TBI. Specifically, this study reports reference values for ANAM tests based on data for >5,000 active duty soldiers undergoing paratroop training. Previous studies reported some of the initial results of this ongoing investigation.18,19 However, this study is unique in that it focuses on the largest database currently available for establishing reference values for the application of ANAM for TBI, especially for individuals for whom this reference group would be an appropriate standard of comparison. METHODS Participants and Procedure The ANAM TBI battery was administered to a convenience sample of > 8,000 highly functioning, active duty soldiers at a southern U.S. military base. Participants underwent ANAM baseline testing from 1999 through 2005 as part of an institutional review board-approved study conducted by the Defense and Veterans Brain Injury Center (DVBIC) to examine the cognitive consequences of TBIs. DVBIC personnel briefed unit leaders, who then informed their soldiers of the study. Up to 15 soldiers were tested simultaneously at the DVBIC facility at Fort Bragg, North Carolina. Before testing, DVBIC personnel described the study in person to all potential subjects, emphasizing the voluntary nature of participation. All participants signed informed consent and Health Insurance Portability and Accountability Act documents. Study personnel were present during the entire test administration, to monitor testing and to provide clarification of test instructions if necessary. One goal of this study was to report on the largest possible aggregate sample of participants with regard to this multiyear, computer-based, baseline TBI testing effort. Data from 5,247 participants obtained between June 26, 2002, and July 29, 2005, formed the basis of this reference data report. This constituted the longest time interval during which no changes were made to the ANAM TBI test battery. Data representing an earlier version of the test battery, collected during a different time epoch in this study, were published previously.18,19Table I provides a description of each test in the ANAM TBI battery. TABLE I. ANAM TBI Battery Test Descriptions     View Large TABLE I. ANAM TBI Battery Test Descriptions     View Large Materials The ANAM system is a library of computer-based tests designed for a broad spectrum of clinical and research applications. The ANAM battery constructed for use in this study consisted of six performance tests (code substitution-learning [CDS], code substitution-delayed [CDD], matching to sample [M2S], mathematical processing [MTH], running memory continuous performance test [CPT], and simple reaction time [SRT]) and two subjective scales (sleepiness and mood scales). The ANAM TBI battery was constructed to be maximally sensitive to the effects of concussion and/or mild TBI and took approximately 15 to 20 minutes to complete with a desktop personal computer. Brief descriptions of the tests are provided in Table I, in the sequence of administration. All participants received the same stimuli for each test (i.e., session 1 of the ANAM TBI battery). Individual stimuli were presented with a variable interstimulus interval, to minimize anticipatory responses. Additional information on parameter settings for this administration of the ANAM TBI battery is presented in Table II. TABLE II. ANAM TBI Battery Parameter (Switch) Settings     View Large TABLE II. ANAM TBI Battery Parameter (Switch) Settings     View Large For the purposes of this article, four primary measures were extracted from the available set of variables produced by the ANAM TBI battery, that is, mean correct response time (MCRT), proportion correct (PC), legacy throughput (LTP), and throughput (TP). The reduced subset of variables provides the major performance measures that have traditionally been used to assess performance for each of the tests. The calculation of these measures is described in the Appendix. In general, higher scores on each of these measures reflect better performance. For example, higher PC values reflect better relative performance. The exception is MCRT. MCRT is a reaction time measure, and faster (i.e., lower) reaction times indicate better relative performance. Data Analyses Before construction of the normative database, the following process was used to identify data for possible exclusion. Descriptive statistics were computed for each performance measure across all participants for each test. These statistics, along with histograms of the distributions, were reviewed to determine the integrity of the dataset. As a result of this examination, data were excluded for a participant on a test if any of the following four criteria were met: (1) no responses (number of response lapses equal to the number of trials), (2) one response (response lapses for all except 1 item), (3) no correct responses, or (4) MCRT >6 SDs above or below the MCRT of the sample. These criteria resulted in the exclusion of data for 14 participants on code substitution-learning, 15 participants on CDD, 17 participants on CPT, 19 participants on M2S, 13 participants on MTH, and 33 participants on SRT. Although these criteria seem rather liberal, we examined the impact of various outlier-removal methods and found minimal differences in the summary statistics when more-conservative criteria were used. Additional data were excluded from the present analysis for the CPT test (n = 681; 13%) after an examination revealed that many participants performed the task incorrectly (responding to every second item rather than every item). In the first phase of the analysis, histograms were constructed for each performance measure to examine the data distributions for each test in the ANAM TBI battery. Goodness of fit was examined by using the Lilliefors modification of the Kolmogorov-Smirnov test. The Kolmogorov-Smirnov test belongs to the family of empirical distribution function statistics commonly used to test the goodness of fit of continuous distributions.20 In the second phase of the analysis, the effects of age and gender on ANAM test performance were examined by using a series of two-way analysis of variance models including gender and age as between-subjects factors. Age was divided into five categories (18–25, 26–30, 31–35, 36–40, and 41–51 years). Separate analyses of variance were conducted for each ANAM test on each of the four performance measures (MCRT, PC, TP, and LTP). When statistical significance was indicated, subsequent pairwise comparisons were made with the Games-Howell post hoc multiple-comparison procedure (α = 0.01). Finally, reference tables were developed to describe the distributional properties of each measure for specific age groups in men and women; here we present the mean, SD, minimum (0%) and maximum (100%) percentile scores, first (25%), second (50%), and third (75%) quartile scores, and 2nd, 9th, 91st, and 98th percentile scores. The 2nd, 9th, 91st, and 98th percentile scores were chosen to serve as cutoff points for “below/above average” (9th/91st percentiles) and “clearly below/above average” performance (2nd/98th percentiles), as described by Hannay and Lezak.21 All analyses were performed with SAS for Windows 8.2 (SAS Institute, Cary, North Carolina) or SPSS for Windows 12.0.0 (SPSS, Chicago, Illinois). A two-sided α level of 0.01 was used to determine statistical significance. RESULTS The number of available participants for analysis varied slightly across ANAM TBI tests depending on the frequency of missing data or the number of individuals excluded for anomalous performance (as described above). Demographic data from the ANAM TBI test with the largest number of participants (N = 5,247, sleepiness scale) are presented in Table III. Participants ranged in age from 18 to 51 years (mean, 26 years; SD, 5.8 years) and were predominantly male (91%). Table IV to Table XII contain the number of participants, means, SDs, and selected percentiles for each ANAM test included in the ANAM TBI battery, stratified according to age group and gender. TABLE III. Participant Group Characteristics     View Large TABLE III. Participant Group Characteristics     View Large TABLE IV. Mean, SD, and Percentile Scores for the ANAM CDS Test     View Large TABLE IV. Mean, SD, and Percentile Scores for the ANAM CDS Test     View Large TABLE V. Mean, SD, and Percentile Scores for the ANAM CDD Test     View Large TABLE V. Mean, SD, and Percentile Scores for the ANAM CDD Test     View Large TABLE VI. Mean, SD, and Percentile Scores for the ANAM CPT Test     View Large TABLE VI. Mean, SD, and Percentile Scores for the ANAM CPT Test     View Large TABLE VII. Mean, SD, and Percentile Scores for the ANAM CDD Test     View Large TABLE VII. Mean, SD, and Percentile Scores for the ANAM CDD Test     View Large TABLE VIII. Mean, SD, and Percentile Scores for the ANAM MTH Test     View Large TABLE VIII. Mean, SD, and Percentile Scores for the ANAM MTH Test     View Large TABLE IX. Mean, SD, and Percentile Scores for the ANAM SRT Test     View Large TABLE IX. Mean, SD, and Percentile Scores for the ANAM SRT Test     View Large TABLE X. Mean, SD, and Percentile Scores for the ANAM SRT Test, Run 2     View Large TABLE X. Mean, SD, and Percentile Scores for the ANAM SRT Test, Run 2     View Large TABLE XI. Sleepiness Scale Ratings     View Large TABLE XI. Sleepiness Scale Ratings     View Large TABLE XII. Sleepiness Scale Ratings     View Large TABLE XII. Sleepiness Scale Ratings     View Large Distributional Tests Assumptions for performing traditional parametric tests include normality and homogeneity of variances. The shape of the data distributions of each of the four performance measures was examined by using the Kolmogorov-Smirnov test. Both normal and logarithmic-normal distributions were fit to the data. All test results were statistically significant, indicating that the data were not well described by either a normal or logarithmic-normal distribution. However, because of the large sample size, it is likely that these tests are detecting statistically significant but practically unimportant deviations from these distributions (see Figs. 1–7 for examples with MCRT). With large sample sizes, most statistical tests with underlying assumptions of normality are robust to deviations from normality, particularly with the apparent magnitude of departure in these data. Nonetheless, all analyses were performed on logarithmically transformed (for MCRT, LTP, and TP) or arcsine-transformed (for PC) data. The distributions after transformation more closely approximated a normal distribution; however, there were still statistically significant departures from normality according to goodness-of-fit testing. FIGURE 1. View largeDownload slide Distribution of MCRT values for the CDS test. FIGURE 1. View largeDownload slide Distribution of MCRT values for the CDS test. FIGURE 2. View largeDownload slide Distribution of MCRT values for the CDD test. FIGURE 2. View largeDownload slide Distribution of MCRT values for the CDD test. FIGURE 3. View largeDownload slide Distribution of MCRT values for CPT test. FIGURE 3. View largeDownload slide Distribution of MCRT values for CPT test. FIGURE 4. View largeDownload slide Distribution of MCRT values for the M2S test. FIGURE 4. View largeDownload slide Distribution of MCRT values for the M2S test. FIGURE 5. View largeDownload slide Distribution of MCRT values for the MTH test. FIGURE 5. View largeDownload slide Distribution of MCRT values for the MTH test. FIGURE 6. View largeDownload slide Distribution of MCRT values for the SRT test. FIGURE 6. View largeDownload slide Distribution of MCRT values for the SRT test. FIGURE 7. View largeDownload slide Distribution of MCRT values for the SRT test, run 2. FIGURE 7. View largeDownload slide Distribution of MCRT values for the SRT test, run 2. The preponderance of male subjects (91%) in the sample presented additional challenges for statistical testing. Age and gender were found to be significantly related on all tests, with the relative proportions of male and female subjects differing according to age group (p<0.01). Bartlett's test of homogeneity of variances indicated that the variability of the performance measures between genders differed according to age for all tests (p<0.01). This is most likely because of the large differences in cell sizes, with male subjects outnumbering female subjects by ratios of 8:1 to 9:1 (varying according to test). The transformations used to achieve normality in the sample distribution also resulted in equalizing the variances across age groups (p > 0.05). Therefore, we proceeded with standard parametric procedures for analyzing the transformed (logarithmic or arcsine) data for all four performance measures. Effects of Age and Gender on Performance MCRT MCRT values from each test in the ANAM TBI battery were subjected to individual two-way analyses of variance. No significant gender-age interactions were indicated for any of the tests. Significant main effects of both gender and age were observed for CDS, CPT, M2S, and SRT (p<0.007) (Table XIII). Follow-up pairwise analyses indicated that MCRTs were significantly lower (faster) for male subjects than for female subjects on all tests (mean differences of 20 milliseconds for SRT, 61 milliseconds for CPT, 168 milliseconds for CDS, and 164 milliseconds for M2S). Furthermore, the comparisons indicated an increase (slowing) in MCRT with increasing age, although not all pairwise comparisons were significant. A significant gender effect was observed only for SRT run 2 (p<0.0001). As in previous tests, male subjects were significantly faster than female subjects. Although no gender effects were identified for MCRTs for CDD, a significant effect of age was indicated (p<0.0001), with values increasing with age. No significant effects were observed in MCRTs for MTH. TABLE XIII. Analysis of Variance Source Table     View Large TABLE XIII. Analysis of Variance Source Table     View Large PC PC for M2S varied significantly between genders (p<0.0001), in favor of male subjects, and age (p = 0.01), declining with age. For MTH, PC varied with age (p = 0.0003) and was greater in the age groups spanning 26 to 35 years, relative to the 18- to 25-year age group (p<0.001). A significant gender-age interaction was indicated for CDS (p = 0.003). Follow-up contrasts showed differences between male and female subjects at ages 18 to 25. In addition, female subjects 18 to 25 years of age differed from male subjects 31 to 35 years of age. A significant age effect was indicated for CDD (p = 0.0004). Follow-up tests indicated that PC was significantly smaller in the 41- to 51-year age group, relative to both the 18- to 25-year and 26- to 30-year age groups (p<0.009). No significant gender or age effects were observed in PCs for CPT, SRT, or SRT run 2. TP Cell mean plots of TP values are presented in Figure 8. A significant gender-age interaction (F = 4.2, p = 0.002) was observed for TP for CDS. TP values for female subjects 18 to 25 years and 41 to 51 years of age were significantly lower than values for male subjects of the same ages (p<0.05). TP values for male subjects were significantly higher for M2S, SRT, and SRT run 2 (p<0.004). This follows from the results for MCRT and PC, with male subjects demonstrating both faster response times and greater accuracy on those tests. TP was associated with age for CDD, CPT, M2S, and MTH (p<0.002). For CDD, a general decline across age groups was indicated, with significant declines observed for all age groups relative to the 18- to 25-year group and the 26- to 30-year group (p<0.004). TP remained constant for M2S from ages 18 through 30, with significant declines observed after age 30. TP was higher for CPT for ages 18 to 25 and 26 to 30, relative to all other age groups, with no significant differences observed between groups after age 30. In contrast, TP significantly increased for MTH from the 18- to 25-year group to the 26- to 30-year group, with a decline after age 40. FIGURE 8. View largeDownload slide TP means according to gender and age. FIGURE 8. View largeDownload slide TP means according to gender and age. LTP The pattern of results for LTP mirrors the pattern for TP, with the exception of CPT and SRT. As described in the Appendix, LTP deviates from TP in proportion to the number of lapses or response omissions. If a participant responds to each item, then TP and LTP are identical. CPT is a test that often results in more lapses than other tests, leading to greater differences in the values of TP and LTP. The significant age main effect observed for TP remained intact. In addition, a significant gender effect was observed for LTP, with male subjects showing higher LTP scores than female subjects (p = 0.007). This is most likely attributable to the greater (although not significant) number of lapses recorded by female subjects. For SRT, a significant age main effect (F = 3.4, p = 0.009) was observed in addition to the previously mentioned gender difference, with LTP declining as age increased. The actual mean difference between genders remained virtually identical. However, an increase in power because of decreased variability in the groups resulted in a statistically significant outcome. Subjective Scales The average rating on the sleepiness scale was 2.68 (SD, 1.4), falling between able to concentrate but not quite at peak (rating = 2) and relaxed and awake but not fully alert (rating = 3). As shown in Table XI, the majority of participants (89%) reported sleepiness levels at or below a rating of 4 (a little tired and having mild difficulty concentrating). Percentile values for the ANAM mood scale are presented in Table XII. Percentage of adjective category was the variable analyzed for the ANAM mood scale. This variable is calculated as the sum of the ratings on the six adjectives for each subscale divided by the maximum possible rating (36 for i = 6 adjectives) for a given subscale and represents a relative percentage of a given mood state reported by each participant. A significant main effect of age was seen for 5 of the 6 mood subscales, with a general decline with increasing age on the anger [F(4,5,000) = 2.92, p = 0.02], depression [F(4,5,000) = 5.30, p = 0.0003], fatigue [F(4,5,000) = 5.88, p = 0.0001], and restlessness [F(4,2,278) = 2.82, p = 0.024] subscales and an increase on the vigor subscale [F(4,5,000) = 4.29, p = 0.0018]. Follow-up comparisons were conducted with Tukey's honestly significant difference test. The 18- to 25-year age group reported significantly more anger, in comparison with the 31- to 35-year age group (p = 0.017). Similarly, participants 18 to 25 years of age reported significantly more depression than did those 26 to 30 years of age (p = 0.0003) and 31 to 35 years of age (p = 0.0002). Participants 18 to 25 years of age reported significantly more fatigue and significantly less vigor, in comparison with all other age groups (all p<0.025) except the 41- to 51-year age group. Reports of restlessness were also higher in the 18- to 25-year age group, compared with the 36- to 40-year age group (p = 0.023). Furthermore, significant gender main effects were observed on the anxiety [F(1,5,000) = 7.38, p = 0.007], fatigue [F(1,5,000) = 9.91, p = 0.002], and vigor [F(1,5,000) = 25.81, p<0.0001] subscales. Male subjects reported significantly less anxiety and fatigue and significantly more vigor than did female subjects. No significant age-gender interactions were noted. DISCUSSION This article presents reference data for a military population tested with the ANAM TBI battery. Furthermore, the statistical impact of gender and age stratification is presented. Overall, the data suggest that a general decline in performance with age should be expected on most tests in the ANAM TBI battery. This was reflected in both increasing (slower) MCRT and decreasing accuracy (lower PC) in older participants. In addition, recorded response times for male subjects were faster than those for female subjects, with only minimal differences in PC. Therefore, the observed TP (and LTP) differences appear to be driven primarily by differences in MCRT between genders. The relatively small numbers of female subjects and older participants could limit the generalizability of these reference values. Although it would be ideal to have proportional representation across all demographic categories, the composition of the military tends to include primarily younger male subjects. The effects of education were not examined in this study, because age and education were found to be highly correlated. In this sample, older individuals generally had completed higher levels of education. This could be attributed to increased opportunities to complete educational endeavors. Also, military career longevity (e.g., promotions) is more likely for those with higher levels of education. The ANAM battery of tests is highly flexible and can be customized by changing many parameter settings within each of the tests. Therefore, it is important to realize that the reference data presented are useful when the ANAM TBI battery is used in its typical/default configuration (Table II). The tremendous flexibility offered for configuring an ANAM battery can influence the resulting data, which should be considered when comparisons with these or other reference values are made. There are some notable differences between the values reported in this article and those reported by Reeves et al.18 and Roebuck-Spencer et al.19 Many of these changes could be explained by differences in the configuration of the test parameters or differences in outlier exclusion methods. The largest differences were observed for CDS, CDD, and MTH tests. In the learning phase (CDS), the current sample was 26% faster (285 milliseconds), with no change in accuracy. The number of trials in the current ANAM TBI battery is twice that in the earlier version of the battery (72 vs. 36 trials), which could easily explain this difference. The data suggest that “learning” occurred, with participants responding more quickly as the test proceeded. Similar differences were observed in the delayed portion of the code substitution test (CDD), with respondents in the current sample showing a 33% decrease in reaction time with approximately equal PC values. Because the number of learning trials was increased, it is likely that the participants had better recall ability, leading to the observed differences in MCRT. The current sample also showed improved response times (32%; 800 milliseconds) for MTH, with a slight decrease in accuracy (10%). Little information is available regarding the parameter settings from the earlier portion of the study; therefore, it is unclear whether the observed differences were attributable to changes in the test parameters or another cause. Minimal differences (<10% change) were noted for the remaining ANAM TBI tests (SRT, CPT, and M2S). These differences may be attributed to changes in the parameter settings of the battery. In the SRT, the number of trials was decreased from 25 to 20. In the CPT, the timeout parameter was changed from 1,500 milliseconds to 1,000 milliseconds. The shorter timeout in the current battery results in a truncated data distribution, with higher reaction time values being excluded from calculation of the performance measures. In addition, the decreased timeout pushes the participants to respond more quickly, thereby perhaps reducing accuracy. Finally, for M2S, two changes were made to the test parameters; the timeout value was lengthened, resulting in a data distribution that was less truncated, and the number of trials was increased (from 15 to 20). Although these reference data were derived from a military sample, their generalizability may be equal to or greater than that of normative data derived from many college samples because of the diversity of the sample. With interest in and knowledge of TBI expanding so rapidly, and with a lack of computer-based tests as broadly distributed in both military and civilian populations as the ANAM TBI battery, these data can be of crucial value to neuropsychologists and other trained health care professionals in interpreting ANAM data in research or clinical contexts. APPENDIX: CALCULATION OF MAJOR PERFORMANCE MEASURES MCRT is the arithmetic average of the response times corresponding to correct responses. It is calculated as MCRT = ΣRTC/C, where RTC is the response time for a correct response and C is the number of correct responses. MCRT is measured in milliseconds. PC represents the accuracy of responding. It is calculated as PC = 100 × [C/(C + I + T)], where C is the number of correct responses, I is the number of incorrect responses, and T is the number of timeouts, or response omissions (also known as, lapses). A “bad” is an early anticipatory response (sometimes called an “impulsive” response) that occurs before any reasonably short response interval (<130 milliseconds). “Bad” responses are excluded from all calculations. TP scores provide an adjustment for speed-accuracy tradeoffs known to result from interindividual variability of performance on choice reaction time tasks.22 These scores provide an estimate of the rate of correct responses per minute. The term LTP is used to describe the TP measure as calculated in ANAM versions before ANAM4. LTP is calculated as LTP = (60,000/mean RT)/[C/(C + I + T)] = speed × PC, where mean RT is the mean response time for all responses (correct and incorrect), C is the number of correct responses, I is the number of incorrect responses, and T is the number of timeouts or response omissions (i.e., lapses). The new calculation of TP includes lapse time in the calculation of available response time. Previously, the number of timeouts, or lapses, was included in the calculation, but the allowable response times associated with lapses were not counted toward the available response time (i.e., mean overall response time). The new calculation of TP is TP = 60,000 × [C/[(C + I )mean RT + (T × timeout)]], where C is the number of correct responses, I is the number of incorrect responses, mean RT is the mean overall response time (correct and incorrect responses), and timeout is the time allowed for a response before a lapse is recorded. When a user does not lapse on any trial, LTP equals TP. As the number of lapses increases, TP deviates from LTP in proportion to the length of the specified response timeout. ACKNOWLEDGMENTS This study was funded primarily by the DVBIC, with additional support from Cooperative Agreement DAMD17-00-1-0056 between the U.S. Army Medical Research and Materiel Command and the National Rehabilitation Hospital. REFERENCES 1. Kane RL, Kay GG Computerized assessment in neuropsychology: a review of tests and test batteries. Neuropsychol Rev  1992; 3: 1– 117. Google Scholar CrossRef Search ADS PubMed  2. Schatz P, Browndyke J Applications of computer-based neuropsychological assessment. J Head Trauma Rehabil  2002; 17: 395– 410. Google Scholar CrossRef Search ADS PubMed  3. Reeves D, Kane R, Winter K, Raynsford K, Pancella T Automated Neuropsychological Assessment Metrics (ANAM): Test Administrator's Guide, Version 1.0 . St. Louis, MO, Mssouri Institute of Mental Health, 1993. 4. Englund CE, Reeves DL, Shingledecker CA, Thorne DR, Wilson KP, Hegge FW Unified Tri-Services Cognitive Performance Assessment Battery (UTC-PAB), Part I: Design and Specification of the Battery. Report 87-10 . Bethesda, MD, Naval Medical Research and Development Command, Joint Working Group on Drug Dependent Degradation in Mlitary Performance, 1987. 5. Perez WA, Masline PJ, Ramsey EG, Urban KE Unified Tri-Services Cognitive Performance Assessment Battery: Review and Methodology. Report AAMRL-TR-87-007 . Wright-Patterson Air Force Base, OH, Armstrong Aerospace Medical Research Laboratory, 1987. 6. Santucci G, Farmer E, Grissett J, et al.   Human Performance Assessment Methods. AGARDograph 308 . Neuilly Sur Seine, France, NATO Advisory Group for Aerospace Research and Development, Working Group 12, 1989. 7. Reeves D, Winter K, Bleiberg J, Kane RL ANAM Genogram: historical perspectives, description, and current endeavors. Arch Clin Neuropsychol  2007; 22(Suppl 1): 15– 37. Google Scholar CrossRef Search ADS PubMed  8. Bleiberg J, Kane RL, Reeves DL, Garmoe WS, Halpern E Factor analysis of computerized and traditional tests used in mild brain injury research. Clin Neuropsychol  2000; 14: 287– 94. Google Scholar CrossRef Search ADS PubMed  9. Kabat MH, Kane RL, Jefferson AL, DiPino RK Construct validity of selected Automated Neuropsychological Assessment Metrics (ANAM) Battery Measures. Clin Neuropsychol  2001; 15: 498– 507. Google Scholar CrossRef Search ADS PubMed  10. Short P, Cernich A, Kabat M, Wilken JA, Kane RL Initial construct validation of frequently employed ANAM measures through structural equation modeling. Arch Clin Neuropsychol  2007; 22(Suppl 1): 63– 77. Google Scholar CrossRef Search ADS PubMed  11. Bleiberg J, Garmoe WS, Halpern EL, Reeves DL, Nadler JD Consistency of within-day and across-day performance after mild brain injury. Neuropsychiatry Neuropsychol Behav Neurol  1997; 10: 247– 53. Google Scholar PubMed  12. Warden DL, Bleiberg J, Cameron KL, et al.   Persistent prolongation of simple reaction time in sports concussion. Neurology  2001; 57: 524– 6. Google Scholar CrossRef Search ADS PubMed  13. Williamson DJG, Scott JG, Adams RL Traumatic brain injury. In: Neuropsychology for Clinical Practice , pp 9– 64. Edited by Adams RL, Parsons OA, Culbertson JL, Nixon SJ Washington, DC, American Psychology Association, 1996. Available at http://oai.dtic.mil/oai/oai?verb=getRecord&metadataPrefix=html&identifier=ADA18240. Google Scholar CrossRef Search ADS   14. Langlois JA, Rutland-Brown W, Thomas KE Traumatic Brain Injury in the United States: Emergency Department Visits, Hospitalizations, and Deaths . Atlanta, GA, Centers for Disease Control and Prevention, 2004. 15. Thurman D, Alverson C, Dunn K, Guerrero J, Sniezek J Traumatic brain injury in the United States: a public health perspective. J Head Trauma Rehabil  1999; 14: 602– 15. Google Scholar CrossRef Search ADS PubMed  16. National Institute of Neurological Disorders and Stroke Traumatic brain injury: hope through research . Available at http://www.ninds.nih.gov/disorders/tbi/pubs_tbi/detail_tbi.htm124893218; accessed April 2, 2007. 17. Ivins BJ, Schwab K, Warden D, et al.   Traumatic brain injury in U.S. Army paratroopers: prevalence and character. J Trauma  2003; 55: 617– 21. Google Scholar CrossRef Search ADS PubMed  18. Reeves DL, Bleiberg J, Roebuck-Spencer T, et al.   Reference values for performance on the Automated Neuropsychological Assessment Metrics V3.0 in an active-duty military sample. Mlit Med  2006; 171: 982– 4. 19. Roebuck-Spencer TM, Reeves DL, Bleiberg J, et al.   Influence of demographics on computerized cognitive testing in a military sample. Mil Psychol  2008; 20: 187– 203. Google Scholar CrossRef Search ADS   20. D'Agostino RB, Stephens MA Goodness-of-Fit Techniques . New York, NY, Marcel Dekker, 1986. 21. Hannay HJ, Lezak MD The neuropsychological examination: interpretation. In: Neuropsychological Assessment , pp 133– 56. Edited by Lezak MD, Howieson DB, Loring DW New York, NY, Oxford University Press, 2004. 22. Salthouse TA, Hedden T Interpreting reaction time measures in between-group comparisons. J Clin Exp Neuropsychol  2002; 24: 858– 72. Google Scholar CrossRef Search ADS PubMed  Reprint & Copyright © Association of Military Surgeons of the U.S. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Military Medicine Oxford University Press

Reference Data from the Automated Neuropsychological Assessment Metrics for Use in Traumatic Brain Injury in an Active Duty Military Sample

Loading next page...
 
/lp/oxford-university-press/reference-data-from-the-automated-neuropsychological-assessment-czrAjUClcm

References (16)

Publisher
Oxford University Press
Copyright
Reprint & Copyright © Association of Military Surgeons of the U.S.
ISSN
0026-4075
eISSN
1930-613X
DOI
10.7205/MILMED.173.9.836
Publisher site
See Article on Publisher Site

Abstract

ABSTRACT The current study examined the performance of active duty soldiers on the Automated Neuropsychological Assessment Metrics (ANAM) traumatic brain injury test battery, to expand the reference data for use in military settings. The effects of age and gender on cognitive performance also were explored. The ANAM traumatic brain injury battery, consisting of six performance tests and two subjective scales, was administered to a sample of healthy active duty soldiers (N = 5,247) as part of a concussion surveillance program. Performance means and SDs, stratified according to age and gender, are reported as reference data. In addition, the impact of age and gender on performance measures was analyzed. Because ANAM is rapidly being adopted for use in many military medical and research applications, the establishment of these reference values is invaluable, particularly for assisting with rapid accurate evaluation and treatment in clinical settings. INTRODUCTION Recent advances in clinical neuroscience have generated rapid expansion in the knowledge of brain functioning and in the methods for treating injury and disease. Many of these advances have been made possible by the introduction of computer-based test systems that provide rapid, objective, and repeatable neuropsychological assessments.1 These computer-based test systems have evolved from research in the disciplines of human factors, cognitive psychology, sensation-perception, and industrial engineering, combined with research and standards of practice in psychiatry, clinical psychology, neuropsychology, and neurology.2 An example of the more accomplished and enduring computer-based test systems is the Automated Neuropsychological Assessment Metrics (ANAM).3 The ANAM test system had its origin in a series of computer test batteries developed by the Department of Defense in the 1980s to assess the influence of pharmaceutical countermeasures for chemical defense. These computer-based test systems were quickly applied to investigate a broad range of risk factors, such as environmental stressors (e.g., heat, cold, hypoxia, and g-force), sleep loss, fatigue, shift work, and drugs.4,5 Currently, ANAM is the most comprehensive and sophisticated test system to evolve from this line of computer-based test development, and it is enhanced by the inclusion of the NATO Advisory Group for Aeronautical Research and Development standards for the development of computer-based test batteries.6 As a result, ANAM has the distinct advantage of providing tests that are comparable to many precursor test systems with one important advancement; ANAM not only embodies some of the best examples of traditional, laboratory-based, human performance assessment metrics but also was designed with enhanced relevance for neuropsychological assessment.7 Construct validity studies demonstrated that ANAM findings correlated well with traditional neuropsychological measures.8,10 This allows laboratory researchers and clinicians to compare and to assess their results not only with respect to ANAM normative values but also with respect to the broad literature bases in these fields. As a result, ANAM subsets of tests have been selected and configured to form stable or preset batteries to address common screening or assessment needs. Alternatively, ANAM provides the flexibility to customdesign batteries for almost any research or clinical application. One area in which ANAM has been especially useful is in the assessment and management of traumatic brain injury (TBI).11,12 The impetus for much of this progress has come from heightened awareness of activities that often result in head injuries, such as explosive war injuries, vehicle accidents, and sports concussions, and the realization that TBIs are more pervasive and pernicious than once thought.13 For example, it is estimated that each year ∼ 1.4 million people in the United States sustain a TBI14 and ∼2% of the U.S. population needs some form of ongoing assistance in activities of daily living as a result of past TBIs.15,16 Although research suggests that TBI is a health threat most often among the young, its incidence is registered across all levels of the population and even more acutely among individuals who are inherently more vulnerable to TBI, such as those serving in the military.17 This study reports the critical results of a multiyear investigation focused on the extended use of a select ANAM battery for the assessment of TBI. Specifically, this study reports reference values for ANAM tests based on data for >5,000 active duty soldiers undergoing paratroop training. Previous studies reported some of the initial results of this ongoing investigation.18,19 However, this study is unique in that it focuses on the largest database currently available for establishing reference values for the application of ANAM for TBI, especially for individuals for whom this reference group would be an appropriate standard of comparison. METHODS Participants and Procedure The ANAM TBI battery was administered to a convenience sample of > 8,000 highly functioning, active duty soldiers at a southern U.S. military base. Participants underwent ANAM baseline testing from 1999 through 2005 as part of an institutional review board-approved study conducted by the Defense and Veterans Brain Injury Center (DVBIC) to examine the cognitive consequences of TBIs. DVBIC personnel briefed unit leaders, who then informed their soldiers of the study. Up to 15 soldiers were tested simultaneously at the DVBIC facility at Fort Bragg, North Carolina. Before testing, DVBIC personnel described the study in person to all potential subjects, emphasizing the voluntary nature of participation. All participants signed informed consent and Health Insurance Portability and Accountability Act documents. Study personnel were present during the entire test administration, to monitor testing and to provide clarification of test instructions if necessary. One goal of this study was to report on the largest possible aggregate sample of participants with regard to this multiyear, computer-based, baseline TBI testing effort. Data from 5,247 participants obtained between June 26, 2002, and July 29, 2005, formed the basis of this reference data report. This constituted the longest time interval during which no changes were made to the ANAM TBI test battery. Data representing an earlier version of the test battery, collected during a different time epoch in this study, were published previously.18,19Table I provides a description of each test in the ANAM TBI battery. TABLE I. ANAM TBI Battery Test Descriptions     View Large TABLE I. ANAM TBI Battery Test Descriptions     View Large Materials The ANAM system is a library of computer-based tests designed for a broad spectrum of clinical and research applications. The ANAM battery constructed for use in this study consisted of six performance tests (code substitution-learning [CDS], code substitution-delayed [CDD], matching to sample [M2S], mathematical processing [MTH], running memory continuous performance test [CPT], and simple reaction time [SRT]) and two subjective scales (sleepiness and mood scales). The ANAM TBI battery was constructed to be maximally sensitive to the effects of concussion and/or mild TBI and took approximately 15 to 20 minutes to complete with a desktop personal computer. Brief descriptions of the tests are provided in Table I, in the sequence of administration. All participants received the same stimuli for each test (i.e., session 1 of the ANAM TBI battery). Individual stimuli were presented with a variable interstimulus interval, to minimize anticipatory responses. Additional information on parameter settings for this administration of the ANAM TBI battery is presented in Table II. TABLE II. ANAM TBI Battery Parameter (Switch) Settings     View Large TABLE II. ANAM TBI Battery Parameter (Switch) Settings     View Large For the purposes of this article, four primary measures were extracted from the available set of variables produced by the ANAM TBI battery, that is, mean correct response time (MCRT), proportion correct (PC), legacy throughput (LTP), and throughput (TP). The reduced subset of variables provides the major performance measures that have traditionally been used to assess performance for each of the tests. The calculation of these measures is described in the Appendix. In general, higher scores on each of these measures reflect better performance. For example, higher PC values reflect better relative performance. The exception is MCRT. MCRT is a reaction time measure, and faster (i.e., lower) reaction times indicate better relative performance. Data Analyses Before construction of the normative database, the following process was used to identify data for possible exclusion. Descriptive statistics were computed for each performance measure across all participants for each test. These statistics, along with histograms of the distributions, were reviewed to determine the integrity of the dataset. As a result of this examination, data were excluded for a participant on a test if any of the following four criteria were met: (1) no responses (number of response lapses equal to the number of trials), (2) one response (response lapses for all except 1 item), (3) no correct responses, or (4) MCRT >6 SDs above or below the MCRT of the sample. These criteria resulted in the exclusion of data for 14 participants on code substitution-learning, 15 participants on CDD, 17 participants on CPT, 19 participants on M2S, 13 participants on MTH, and 33 participants on SRT. Although these criteria seem rather liberal, we examined the impact of various outlier-removal methods and found minimal differences in the summary statistics when more-conservative criteria were used. Additional data were excluded from the present analysis for the CPT test (n = 681; 13%) after an examination revealed that many participants performed the task incorrectly (responding to every second item rather than every item). In the first phase of the analysis, histograms were constructed for each performance measure to examine the data distributions for each test in the ANAM TBI battery. Goodness of fit was examined by using the Lilliefors modification of the Kolmogorov-Smirnov test. The Kolmogorov-Smirnov test belongs to the family of empirical distribution function statistics commonly used to test the goodness of fit of continuous distributions.20 In the second phase of the analysis, the effects of age and gender on ANAM test performance were examined by using a series of two-way analysis of variance models including gender and age as between-subjects factors. Age was divided into five categories (18–25, 26–30, 31–35, 36–40, and 41–51 years). Separate analyses of variance were conducted for each ANAM test on each of the four performance measures (MCRT, PC, TP, and LTP). When statistical significance was indicated, subsequent pairwise comparisons were made with the Games-Howell post hoc multiple-comparison procedure (α = 0.01). Finally, reference tables were developed to describe the distributional properties of each measure for specific age groups in men and women; here we present the mean, SD, minimum (0%) and maximum (100%) percentile scores, first (25%), second (50%), and third (75%) quartile scores, and 2nd, 9th, 91st, and 98th percentile scores. The 2nd, 9th, 91st, and 98th percentile scores were chosen to serve as cutoff points for “below/above average” (9th/91st percentiles) and “clearly below/above average” performance (2nd/98th percentiles), as described by Hannay and Lezak.21 All analyses were performed with SAS for Windows 8.2 (SAS Institute, Cary, North Carolina) or SPSS for Windows 12.0.0 (SPSS, Chicago, Illinois). A two-sided α level of 0.01 was used to determine statistical significance. RESULTS The number of available participants for analysis varied slightly across ANAM TBI tests depending on the frequency of missing data or the number of individuals excluded for anomalous performance (as described above). Demographic data from the ANAM TBI test with the largest number of participants (N = 5,247, sleepiness scale) are presented in Table III. Participants ranged in age from 18 to 51 years (mean, 26 years; SD, 5.8 years) and were predominantly male (91%). Table IV to Table XII contain the number of participants, means, SDs, and selected percentiles for each ANAM test included in the ANAM TBI battery, stratified according to age group and gender. TABLE III. Participant Group Characteristics     View Large TABLE III. Participant Group Characteristics     View Large TABLE IV. Mean, SD, and Percentile Scores for the ANAM CDS Test     View Large TABLE IV. Mean, SD, and Percentile Scores for the ANAM CDS Test     View Large TABLE V. Mean, SD, and Percentile Scores for the ANAM CDD Test     View Large TABLE V. Mean, SD, and Percentile Scores for the ANAM CDD Test     View Large TABLE VI. Mean, SD, and Percentile Scores for the ANAM CPT Test     View Large TABLE VI. Mean, SD, and Percentile Scores for the ANAM CPT Test     View Large TABLE VII. Mean, SD, and Percentile Scores for the ANAM CDD Test     View Large TABLE VII. Mean, SD, and Percentile Scores for the ANAM CDD Test     View Large TABLE VIII. Mean, SD, and Percentile Scores for the ANAM MTH Test     View Large TABLE VIII. Mean, SD, and Percentile Scores for the ANAM MTH Test     View Large TABLE IX. Mean, SD, and Percentile Scores for the ANAM SRT Test     View Large TABLE IX. Mean, SD, and Percentile Scores for the ANAM SRT Test     View Large TABLE X. Mean, SD, and Percentile Scores for the ANAM SRT Test, Run 2     View Large TABLE X. Mean, SD, and Percentile Scores for the ANAM SRT Test, Run 2     View Large TABLE XI. Sleepiness Scale Ratings     View Large TABLE XI. Sleepiness Scale Ratings     View Large TABLE XII. Sleepiness Scale Ratings     View Large TABLE XII. Sleepiness Scale Ratings     View Large Distributional Tests Assumptions for performing traditional parametric tests include normality and homogeneity of variances. The shape of the data distributions of each of the four performance measures was examined by using the Kolmogorov-Smirnov test. Both normal and logarithmic-normal distributions were fit to the data. All test results were statistically significant, indicating that the data were not well described by either a normal or logarithmic-normal distribution. However, because of the large sample size, it is likely that these tests are detecting statistically significant but practically unimportant deviations from these distributions (see Figs. 1–7 for examples with MCRT). With large sample sizes, most statistical tests with underlying assumptions of normality are robust to deviations from normality, particularly with the apparent magnitude of departure in these data. Nonetheless, all analyses were performed on logarithmically transformed (for MCRT, LTP, and TP) or arcsine-transformed (for PC) data. The distributions after transformation more closely approximated a normal distribution; however, there were still statistically significant departures from normality according to goodness-of-fit testing. FIGURE 1. View largeDownload slide Distribution of MCRT values for the CDS test. FIGURE 1. View largeDownload slide Distribution of MCRT values for the CDS test. FIGURE 2. View largeDownload slide Distribution of MCRT values for the CDD test. FIGURE 2. View largeDownload slide Distribution of MCRT values for the CDD test. FIGURE 3. View largeDownload slide Distribution of MCRT values for CPT test. FIGURE 3. View largeDownload slide Distribution of MCRT values for CPT test. FIGURE 4. View largeDownload slide Distribution of MCRT values for the M2S test. FIGURE 4. View largeDownload slide Distribution of MCRT values for the M2S test. FIGURE 5. View largeDownload slide Distribution of MCRT values for the MTH test. FIGURE 5. View largeDownload slide Distribution of MCRT values for the MTH test. FIGURE 6. View largeDownload slide Distribution of MCRT values for the SRT test. FIGURE 6. View largeDownload slide Distribution of MCRT values for the SRT test. FIGURE 7. View largeDownload slide Distribution of MCRT values for the SRT test, run 2. FIGURE 7. View largeDownload slide Distribution of MCRT values for the SRT test, run 2. The preponderance of male subjects (91%) in the sample presented additional challenges for statistical testing. Age and gender were found to be significantly related on all tests, with the relative proportions of male and female subjects differing according to age group (p<0.01). Bartlett's test of homogeneity of variances indicated that the variability of the performance measures between genders differed according to age for all tests (p<0.01). This is most likely because of the large differences in cell sizes, with male subjects outnumbering female subjects by ratios of 8:1 to 9:1 (varying according to test). The transformations used to achieve normality in the sample distribution also resulted in equalizing the variances across age groups (p > 0.05). Therefore, we proceeded with standard parametric procedures for analyzing the transformed (logarithmic or arcsine) data for all four performance measures. Effects of Age and Gender on Performance MCRT MCRT values from each test in the ANAM TBI battery were subjected to individual two-way analyses of variance. No significant gender-age interactions were indicated for any of the tests. Significant main effects of both gender and age were observed for CDS, CPT, M2S, and SRT (p<0.007) (Table XIII). Follow-up pairwise analyses indicated that MCRTs were significantly lower (faster) for male subjects than for female subjects on all tests (mean differences of 20 milliseconds for SRT, 61 milliseconds for CPT, 168 milliseconds for CDS, and 164 milliseconds for M2S). Furthermore, the comparisons indicated an increase (slowing) in MCRT with increasing age, although not all pairwise comparisons were significant. A significant gender effect was observed only for SRT run 2 (p<0.0001). As in previous tests, male subjects were significantly faster than female subjects. Although no gender effects were identified for MCRTs for CDD, a significant effect of age was indicated (p<0.0001), with values increasing with age. No significant effects were observed in MCRTs for MTH. TABLE XIII. Analysis of Variance Source Table     View Large TABLE XIII. Analysis of Variance Source Table     View Large PC PC for M2S varied significantly between genders (p<0.0001), in favor of male subjects, and age (p = 0.01), declining with age. For MTH, PC varied with age (p = 0.0003) and was greater in the age groups spanning 26 to 35 years, relative to the 18- to 25-year age group (p<0.001). A significant gender-age interaction was indicated for CDS (p = 0.003). Follow-up contrasts showed differences between male and female subjects at ages 18 to 25. In addition, female subjects 18 to 25 years of age differed from male subjects 31 to 35 years of age. A significant age effect was indicated for CDD (p = 0.0004). Follow-up tests indicated that PC was significantly smaller in the 41- to 51-year age group, relative to both the 18- to 25-year and 26- to 30-year age groups (p<0.009). No significant gender or age effects were observed in PCs for CPT, SRT, or SRT run 2. TP Cell mean plots of TP values are presented in Figure 8. A significant gender-age interaction (F = 4.2, p = 0.002) was observed for TP for CDS. TP values for female subjects 18 to 25 years and 41 to 51 years of age were significantly lower than values for male subjects of the same ages (p<0.05). TP values for male subjects were significantly higher for M2S, SRT, and SRT run 2 (p<0.004). This follows from the results for MCRT and PC, with male subjects demonstrating both faster response times and greater accuracy on those tests. TP was associated with age for CDD, CPT, M2S, and MTH (p<0.002). For CDD, a general decline across age groups was indicated, with significant declines observed for all age groups relative to the 18- to 25-year group and the 26- to 30-year group (p<0.004). TP remained constant for M2S from ages 18 through 30, with significant declines observed after age 30. TP was higher for CPT for ages 18 to 25 and 26 to 30, relative to all other age groups, with no significant differences observed between groups after age 30. In contrast, TP significantly increased for MTH from the 18- to 25-year group to the 26- to 30-year group, with a decline after age 40. FIGURE 8. View largeDownload slide TP means according to gender and age. FIGURE 8. View largeDownload slide TP means according to gender and age. LTP The pattern of results for LTP mirrors the pattern for TP, with the exception of CPT and SRT. As described in the Appendix, LTP deviates from TP in proportion to the number of lapses or response omissions. If a participant responds to each item, then TP and LTP are identical. CPT is a test that often results in more lapses than other tests, leading to greater differences in the values of TP and LTP. The significant age main effect observed for TP remained intact. In addition, a significant gender effect was observed for LTP, with male subjects showing higher LTP scores than female subjects (p = 0.007). This is most likely attributable to the greater (although not significant) number of lapses recorded by female subjects. For SRT, a significant age main effect (F = 3.4, p = 0.009) was observed in addition to the previously mentioned gender difference, with LTP declining as age increased. The actual mean difference between genders remained virtually identical. However, an increase in power because of decreased variability in the groups resulted in a statistically significant outcome. Subjective Scales The average rating on the sleepiness scale was 2.68 (SD, 1.4), falling between able to concentrate but not quite at peak (rating = 2) and relaxed and awake but not fully alert (rating = 3). As shown in Table XI, the majority of participants (89%) reported sleepiness levels at or below a rating of 4 (a little tired and having mild difficulty concentrating). Percentile values for the ANAM mood scale are presented in Table XII. Percentage of adjective category was the variable analyzed for the ANAM mood scale. This variable is calculated as the sum of the ratings on the six adjectives for each subscale divided by the maximum possible rating (36 for i = 6 adjectives) for a given subscale and represents a relative percentage of a given mood state reported by each participant. A significant main effect of age was seen for 5 of the 6 mood subscales, with a general decline with increasing age on the anger [F(4,5,000) = 2.92, p = 0.02], depression [F(4,5,000) = 5.30, p = 0.0003], fatigue [F(4,5,000) = 5.88, p = 0.0001], and restlessness [F(4,2,278) = 2.82, p = 0.024] subscales and an increase on the vigor subscale [F(4,5,000) = 4.29, p = 0.0018]. Follow-up comparisons were conducted with Tukey's honestly significant difference test. The 18- to 25-year age group reported significantly more anger, in comparison with the 31- to 35-year age group (p = 0.017). Similarly, participants 18 to 25 years of age reported significantly more depression than did those 26 to 30 years of age (p = 0.0003) and 31 to 35 years of age (p = 0.0002). Participants 18 to 25 years of age reported significantly more fatigue and significantly less vigor, in comparison with all other age groups (all p<0.025) except the 41- to 51-year age group. Reports of restlessness were also higher in the 18- to 25-year age group, compared with the 36- to 40-year age group (p = 0.023). Furthermore, significant gender main effects were observed on the anxiety [F(1,5,000) = 7.38, p = 0.007], fatigue [F(1,5,000) = 9.91, p = 0.002], and vigor [F(1,5,000) = 25.81, p<0.0001] subscales. Male subjects reported significantly less anxiety and fatigue and significantly more vigor than did female subjects. No significant age-gender interactions were noted. DISCUSSION This article presents reference data for a military population tested with the ANAM TBI battery. Furthermore, the statistical impact of gender and age stratification is presented. Overall, the data suggest that a general decline in performance with age should be expected on most tests in the ANAM TBI battery. This was reflected in both increasing (slower) MCRT and decreasing accuracy (lower PC) in older participants. In addition, recorded response times for male subjects were faster than those for female subjects, with only minimal differences in PC. Therefore, the observed TP (and LTP) differences appear to be driven primarily by differences in MCRT between genders. The relatively small numbers of female subjects and older participants could limit the generalizability of these reference values. Although it would be ideal to have proportional representation across all demographic categories, the composition of the military tends to include primarily younger male subjects. The effects of education were not examined in this study, because age and education were found to be highly correlated. In this sample, older individuals generally had completed higher levels of education. This could be attributed to increased opportunities to complete educational endeavors. Also, military career longevity (e.g., promotions) is more likely for those with higher levels of education. The ANAM battery of tests is highly flexible and can be customized by changing many parameter settings within each of the tests. Therefore, it is important to realize that the reference data presented are useful when the ANAM TBI battery is used in its typical/default configuration (Table II). The tremendous flexibility offered for configuring an ANAM battery can influence the resulting data, which should be considered when comparisons with these or other reference values are made. There are some notable differences between the values reported in this article and those reported by Reeves et al.18 and Roebuck-Spencer et al.19 Many of these changes could be explained by differences in the configuration of the test parameters or differences in outlier exclusion methods. The largest differences were observed for CDS, CDD, and MTH tests. In the learning phase (CDS), the current sample was 26% faster (285 milliseconds), with no change in accuracy. The number of trials in the current ANAM TBI battery is twice that in the earlier version of the battery (72 vs. 36 trials), which could easily explain this difference. The data suggest that “learning” occurred, with participants responding more quickly as the test proceeded. Similar differences were observed in the delayed portion of the code substitution test (CDD), with respondents in the current sample showing a 33% decrease in reaction time with approximately equal PC values. Because the number of learning trials was increased, it is likely that the participants had better recall ability, leading to the observed differences in MCRT. The current sample also showed improved response times (32%; 800 milliseconds) for MTH, with a slight decrease in accuracy (10%). Little information is available regarding the parameter settings from the earlier portion of the study; therefore, it is unclear whether the observed differences were attributable to changes in the test parameters or another cause. Minimal differences (<10% change) were noted for the remaining ANAM TBI tests (SRT, CPT, and M2S). These differences may be attributed to changes in the parameter settings of the battery. In the SRT, the number of trials was decreased from 25 to 20. In the CPT, the timeout parameter was changed from 1,500 milliseconds to 1,000 milliseconds. The shorter timeout in the current battery results in a truncated data distribution, with higher reaction time values being excluded from calculation of the performance measures. In addition, the decreased timeout pushes the participants to respond more quickly, thereby perhaps reducing accuracy. Finally, for M2S, two changes were made to the test parameters; the timeout value was lengthened, resulting in a data distribution that was less truncated, and the number of trials was increased (from 15 to 20). Although these reference data were derived from a military sample, their generalizability may be equal to or greater than that of normative data derived from many college samples because of the diversity of the sample. With interest in and knowledge of TBI expanding so rapidly, and with a lack of computer-based tests as broadly distributed in both military and civilian populations as the ANAM TBI battery, these data can be of crucial value to neuropsychologists and other trained health care professionals in interpreting ANAM data in research or clinical contexts. APPENDIX: CALCULATION OF MAJOR PERFORMANCE MEASURES MCRT is the arithmetic average of the response times corresponding to correct responses. It is calculated as MCRT = ΣRTC/C, where RTC is the response time for a correct response and C is the number of correct responses. MCRT is measured in milliseconds. PC represents the accuracy of responding. It is calculated as PC = 100 × [C/(C + I + T)], where C is the number of correct responses, I is the number of incorrect responses, and T is the number of timeouts, or response omissions (also known as, lapses). A “bad” is an early anticipatory response (sometimes called an “impulsive” response) that occurs before any reasonably short response interval (<130 milliseconds). “Bad” responses are excluded from all calculations. TP scores provide an adjustment for speed-accuracy tradeoffs known to result from interindividual variability of performance on choice reaction time tasks.22 These scores provide an estimate of the rate of correct responses per minute. The term LTP is used to describe the TP measure as calculated in ANAM versions before ANAM4. LTP is calculated as LTP = (60,000/mean RT)/[C/(C + I + T)] = speed × PC, where mean RT is the mean response time for all responses (correct and incorrect), C is the number of correct responses, I is the number of incorrect responses, and T is the number of timeouts or response omissions (i.e., lapses). The new calculation of TP includes lapse time in the calculation of available response time. Previously, the number of timeouts, or lapses, was included in the calculation, but the allowable response times associated with lapses were not counted toward the available response time (i.e., mean overall response time). The new calculation of TP is TP = 60,000 × [C/[(C + I )mean RT + (T × timeout)]], where C is the number of correct responses, I is the number of incorrect responses, mean RT is the mean overall response time (correct and incorrect responses), and timeout is the time allowed for a response before a lapse is recorded. When a user does not lapse on any trial, LTP equals TP. As the number of lapses increases, TP deviates from LTP in proportion to the length of the specified response timeout. ACKNOWLEDGMENTS This study was funded primarily by the DVBIC, with additional support from Cooperative Agreement DAMD17-00-1-0056 between the U.S. Army Medical Research and Materiel Command and the National Rehabilitation Hospital. REFERENCES 1. Kane RL, Kay GG Computerized assessment in neuropsychology: a review of tests and test batteries. Neuropsychol Rev  1992; 3: 1– 117. Google Scholar CrossRef Search ADS PubMed  2. Schatz P, Browndyke J Applications of computer-based neuropsychological assessment. J Head Trauma Rehabil  2002; 17: 395– 410. Google Scholar CrossRef Search ADS PubMed  3. Reeves D, Kane R, Winter K, Raynsford K, Pancella T Automated Neuropsychological Assessment Metrics (ANAM): Test Administrator's Guide, Version 1.0 . St. Louis, MO, Mssouri Institute of Mental Health, 1993. 4. Englund CE, Reeves DL, Shingledecker CA, Thorne DR, Wilson KP, Hegge FW Unified Tri-Services Cognitive Performance Assessment Battery (UTC-PAB), Part I: Design and Specification of the Battery. Report 87-10 . Bethesda, MD, Naval Medical Research and Development Command, Joint Working Group on Drug Dependent Degradation in Mlitary Performance, 1987. 5. Perez WA, Masline PJ, Ramsey EG, Urban KE Unified Tri-Services Cognitive Performance Assessment Battery: Review and Methodology. Report AAMRL-TR-87-007 . Wright-Patterson Air Force Base, OH, Armstrong Aerospace Medical Research Laboratory, 1987. 6. Santucci G, Farmer E, Grissett J, et al.   Human Performance Assessment Methods. AGARDograph 308 . Neuilly Sur Seine, France, NATO Advisory Group for Aerospace Research and Development, Working Group 12, 1989. 7. Reeves D, Winter K, Bleiberg J, Kane RL ANAM Genogram: historical perspectives, description, and current endeavors. Arch Clin Neuropsychol  2007; 22(Suppl 1): 15– 37. Google Scholar CrossRef Search ADS PubMed  8. Bleiberg J, Kane RL, Reeves DL, Garmoe WS, Halpern E Factor analysis of computerized and traditional tests used in mild brain injury research. Clin Neuropsychol  2000; 14: 287– 94. Google Scholar CrossRef Search ADS PubMed  9. Kabat MH, Kane RL, Jefferson AL, DiPino RK Construct validity of selected Automated Neuropsychological Assessment Metrics (ANAM) Battery Measures. Clin Neuropsychol  2001; 15: 498– 507. Google Scholar CrossRef Search ADS PubMed  10. Short P, Cernich A, Kabat M, Wilken JA, Kane RL Initial construct validation of frequently employed ANAM measures through structural equation modeling. Arch Clin Neuropsychol  2007; 22(Suppl 1): 63– 77. Google Scholar CrossRef Search ADS PubMed  11. Bleiberg J, Garmoe WS, Halpern EL, Reeves DL, Nadler JD Consistency of within-day and across-day performance after mild brain injury. Neuropsychiatry Neuropsychol Behav Neurol  1997; 10: 247– 53. Google Scholar PubMed  12. Warden DL, Bleiberg J, Cameron KL, et al.   Persistent prolongation of simple reaction time in sports concussion. Neurology  2001; 57: 524– 6. Google Scholar CrossRef Search ADS PubMed  13. Williamson DJG, Scott JG, Adams RL Traumatic brain injury. In: Neuropsychology for Clinical Practice , pp 9– 64. Edited by Adams RL, Parsons OA, Culbertson JL, Nixon SJ Washington, DC, American Psychology Association, 1996. Available at http://oai.dtic.mil/oai/oai?verb=getRecord&metadataPrefix=html&identifier=ADA18240. Google Scholar CrossRef Search ADS   14. Langlois JA, Rutland-Brown W, Thomas KE Traumatic Brain Injury in the United States: Emergency Department Visits, Hospitalizations, and Deaths . Atlanta, GA, Centers for Disease Control and Prevention, 2004. 15. Thurman D, Alverson C, Dunn K, Guerrero J, Sniezek J Traumatic brain injury in the United States: a public health perspective. J Head Trauma Rehabil  1999; 14: 602– 15. Google Scholar CrossRef Search ADS PubMed  16. National Institute of Neurological Disorders and Stroke Traumatic brain injury: hope through research . Available at http://www.ninds.nih.gov/disorders/tbi/pubs_tbi/detail_tbi.htm124893218; accessed April 2, 2007. 17. Ivins BJ, Schwab K, Warden D, et al.   Traumatic brain injury in U.S. Army paratroopers: prevalence and character. J Trauma  2003; 55: 617– 21. Google Scholar CrossRef Search ADS PubMed  18. Reeves DL, Bleiberg J, Roebuck-Spencer T, et al.   Reference values for performance on the Automated Neuropsychological Assessment Metrics V3.0 in an active-duty military sample. Mlit Med  2006; 171: 982– 4. 19. Roebuck-Spencer TM, Reeves DL, Bleiberg J, et al.   Influence of demographics on computerized cognitive testing in a military sample. Mil Psychol  2008; 20: 187– 203. Google Scholar CrossRef Search ADS   20. D'Agostino RB, Stephens MA Goodness-of-Fit Techniques . New York, NY, Marcel Dekker, 1986. 21. Hannay HJ, Lezak MD The neuropsychological examination: interpretation. In: Neuropsychological Assessment , pp 133– 56. Edited by Lezak MD, Howieson DB, Loring DW New York, NY, Oxford University Press, 2004. 22. Salthouse TA, Hedden T Interpreting reaction time measures in between-group comparisons. J Clin Exp Neuropsychol  2002; 24: 858– 72. Google Scholar CrossRef Search ADS PubMed  Reprint & Copyright © Association of Military Surgeons of the U.S.

Journal

Military MedicineOxford University Press

Published: Sep 1, 2008

There are no references for this article.