Day 1 to 3: 10 mg bid placebo. Day 3 to 21: 20 mg bid placebo. Day 21 to 168: 10 mg bid placebo.

Brief Visuospatial Memory Test-Revised (BVMT-R) Delayed Recall measures the correctly recalled designs (i.e., standard scoring of accuracy and location as described in the manual) after a 25 minute delay. The delayed recall raw score ranges from 0 to 12 with 12 being the highest and best possible score. The raw score is converted to a T-score (Mean=50; SD=10) which was used for statistical analysis.

Brief Visuospatial Memory Test-Revised (BVMT-R) Learning measures the correctly recalled designs (i.e., standard scoring of accuracy and location as described in the manual) over 3 learning trials. The Learning raw score is the sum of the higher number of correctly recalled designs on either Trial 2 or Trial 3 minus the number of correctly recalled designs on Trial 1. A higher score is a better score. The raw score was converted to a T-score (Mean=50; SD=10) for analysis.

Neuropsychological test of visual attention and executive functioning. The trail making tests are thought to reflect a variety of cognitive processes including attention, visual search and scanning, sequencing and shifting, psychomotor speed, abstraction, flexibility, ability to execute and modify a plan of action, and ability to maintain two trains of thought simultaneously. In Trails B, the participant is instructed to draw lines to connect numbers and letters in an alternating numeric and alphabetic sequence as rapidly as possible. Lower scores are better scores and the range of scores can be from 0 to no limit for Trail Making Test part B. This is a timed test and the number of seconds to complete the task is recorded. The unit of measure in seconds is converted to a scaled score (mean =10, SD = 3) using the Heaton et al. norms with lower scores indicating better performance.

Stroop Interference Test is a neuropsychological test to assess a person's executive function. Specifically, the test is thought to reflect selective attention, cognitive flexibility and processing speed. The raw score for this measure is the number of items correctly identified within 45 seconds. The raw score was converted to a T-score (mean=50; SD=10) for analysis. Higher scores reflect better performance and less interference on reading ability.

The Behavior Rating Inventory of Executive Function (BRIEF) Inhibit subscale is a rating scale completed by the participant and independently by an observer that assesses the ability to control impulses (inhibitory control) and to stop engaging in a behavior. The frequency of behaviors indicated by items is rated on a 3-point scale (never, sometimes, often). The raw score for the Inhibit subscale is the sum of ratings for the 8 items included in this measure. This sum was converted to a T-score (mean=50; SD=10) for analysis. Higher scores suggest a higher level of dysfunction in a specific domain of executive functions.

The TBI-QOL Anger item bank includes 38 hierarchically ordered items designed to measure the full continuum of anger in a way which is both sensitive and appropriate for TBI. The TBI-QOL Anger item bank can be administered as a computer-adaptive test (CAT), allowing precise measurement of self-reported anger using only 4-8 adaptively selected items. Using CAT technology, an individual participant's responses to the TBIQoL Anger scale generated a T-score (Mean=50; SD=10) with a range of 0 (lowest anger) to 100 (greatest anger).

Total of all reporting groups

Neuropsychological test used to assess an individual's verbal memory abilities. California Verbal Learning Test-Second Edition (CVLT-II) Trials 1-5 Free Recall Total measures the sum of all word list items correctly recalled on learning trials 1 through 5. This raw score is converted to a T-score (Mean=50; SD=10) which was used for statistical analysis. The total A1-5T score reflects accurate recall over the five learning trials of the first list, and is most often used as a summary index of learning on the CVLT-II, with higher scores reflecting better performance.

Brief Visuospatial Memory Test-Revised (BVMT-R) Delayed Recall measures the correctly recalled designs (i.e., standard scoring of accuracy and location as described in the manual) after a 25 minute delay. The delayed recall raw score ranges from 0 to 12 with 12 being the highest and best possible score. The raw score is converted to a T-score (Mean=50; SD=10) which was used for statistical analysis.

Brief Visuospatial Memory Test-Revised (BVMT-R) Learning measures the correctly recalled designs (i.e., standard scoring of accuracy and location as described in the manual) over 3 learning trials. The Learning raw score is the sum of the higher number of correctly recalled designs on either Trial 2 or Trial 3 minus the number of correctly recalled designs on Trial 1. A higher score is a better score. The raw score was converted to a T-score (Mean=50; SD=10) for analysis.

Neuropsychological test of visual attention and executive functioning. The trail making tests are thought to reflect a variety of cognitive processes including attention, visual search and scanning, sequencing and shifting, psychomotor speed, abstraction, flexibility, ability to execute and modify a plan of action, and ability to maintain two trains of thought simultaneously. In Trails B, the participant is instructed to draw lines to connect numbers and letters in an alternating numeric and alphabetic sequence as rapidly as possible. Lower scores are better scores and the range of scores can be from 0 to no limit for Trail Making Test part B. This is a timed test and the number of seconds to complete the task is recorded. The unit of measure in seconds is converted to a scaled score (mean =10, SD = 3) using the Heaton et al. norms with lower scores indicating better performance.

Stroop Interference Test is a neuropsychological test to assess a person's executive function. Specifically, the test is thought to reflect selective attention, cognitive flexibility and processing speed. The raw score for this measure is the number of items correctly identified within 45 seconds. The raw score was converted to a T-score (mean=50; SD=10) for analysis. Higher scores reflect better performance and less interference on reading ability.

The Behavior Rating Inventory of Executive Function (BRIEF) Inhibit subscale is a rating scale completed by the participant and independently by an observer that assesses the ability to control impulses (inhibitory control) and to stop engaging in a behavior. The frequency of behaviors indicated by items is rated on a 3-point scale (never, sometimes, often). The raw score for the Inhibit subscale is the sum of ratings for the 8 items included in this measure. This sum was converted to a T-score (mean=50; SD=10) for analysis. Higher scores suggest a higher level of dysfunction in a specific domain of executive functions.

The TBI-QOL Anger item bank includes 38 hierarchically ordered items designed to measure the full continuum of anger in a way which is both sensitive and appropriate for TBI. The TBI-QOL Anger item bank can be administered as a computer-adaptive test (CAT), allowing precise measurement of self-reported anger using only 4-8 adaptively selected items. Using CAT technology, an individual participant's responses to the TBIQoL Anger scale generated a T-score (Mean=50; SD=10) with a range of 0 (lowest anger) to 100 (greatest anger).

24 weeks of placebo matched to memantine formulation, beginning within 48 hours of injury.

Placebo: Day 1 to 3: 10 mg bid placebo. Day 3 to 21: 20 mg bid placebo. Day 21 to 168: 10 mg bid placebo.