Operating Characteristics of Executive Functioning Tests Following Traumatic Brain Injury

Abstract

The primary purposes of this study were to determine if controls, and mild and moderate/severe traumatic brain injury (TBI) patients performed differently on a battery of executive functioning (EF) tests, and to identify the operating characteristics of EF tests in this population. Participants consisted of 46 brain-injured individuals and 24 healthy controls. All participants completed an extensive battery of EF tests. Results showed that mild TBI participants performed worse than controls on the Trail Making Test Part B, and that moderate/severe TBI participants consistently performed worse than either group on a variety of EF measures. Tests of EF exhibited a wide range of operating characteristics, suggesting that some EF tests are better than others in identifying TBI-related neurocognitive impairment. Predictive values were better for individuals with moderate/severe TBI than mild TBI. Overall, the Digit Span Backward Test showed the best positive predictive power in differentiating TBI. Our results provide useful data that may guide test selection in evaluating EF in patients with traumatic brain injury.

Keywords:

• Operating characteristics
• Executive functioning
• Traumatic brain injury

Acknowledgments

Supported by grants from the Evelyn F. McKnight Brain Research Grant Program, the Florida Brain and Spinal Cord Injury Trust Fund, and the National Institute of Mental Health (K01 MH01857) to W.M.P. We thank Kay Waid-Ebbs and Paul J. Seignourel for their assistance with patient recruitment and data collection. This research was conducted in partial fulfillment of the first author's Ph.D. dissertation.

Notes

¹We evaluated the accuracy of our retrospective interview method by comparing interview-only brain injury severity classification to brain injury severity classification in seven participants (one mild, six M/S) for whom medical records were reviewed. In each case the medical record review confirmed our initial interview-only method of brain injury severity classification as not a single participant was misclassified.

²Due to difficulties with data collection, the sample size for the WCST consists of 15 individuals with mild TBI, 11 individuals with M/S TBI, and 10 controls.

³We conducted a frequency count on the Digit Span Age-Corrected Scaled Scores (ACSS) and found that there was one participant with mild TBI and one control with ACSS less than 7 (both had ACSS of 6). When these participants were excluded, the pattern of significance in the between-groups differences did not change (i.e., the same values that were statistically significant remained as such). Thus we chose to include all participants in the analyses in order to most fully represent our data.

⁴Levene's test of homogeneity of variance for the Trail Making Test Parts A and B revealed significantly different variances between-groups, Levene's Statistic >6.32, p < .003. Thus we conducted a Welch's ANCOVA that accounts for the group differences in variance. Results were consistent with the original ANCOVA, F(2, 42) = 9.82, p < .001.