Abstract
Objective: To determine correspondence between the statistically derived 8-point reliable change index for the Neurobehavioral Symptom Inventory (NSI) against clinically significant item-level change in symptom severity from intake to discharge for mild traumatic brain injury (mTBI).
Setting: Brain Injury Rehabilitation Service at Brooke Army Medical Center, Fort Sam Houston, San Antonio, Texas.
Patients: In total, 655 active-duty service members with a diagnosis of mTBI who received treatment and completed self-report measures between 2007 and 2020.
Design: Observational retrospective analysis of outpatient clinical outcomes data.
Main Measures: NSI total score change was used to divide patients into responder and nonresponders based on whether they met an 8-point decrease between intake and discharge. In addition, patients who had at least one NSI item that changed from a rating of 3 (severe) or 4 (very severe) at intake to a rating of 0 (none) or 1 (mild) at discharge were coded as an individual with significant item-level change.
Results: Forty-five percent of the sample had significant item-level change and were classified as responders according to the reliable change method. Eight percent of the sample had significant item-level change but did not meet the 8-point reliable change threshold. Fifteen percent of the sample did not experience significant item-level change but were classified as responders according to reliable change. Thirty-one percent did not meet either method's criterion for change. Classification agreement between the reliable change and item-level change methods was 76%, which was statistically significant (= 181.32, P < .001).
Conclusion: There is good correspondence between reliable change and item-level change on the NSI in this population. Reliable change is easily calculated and thus much more accessible than the item-level change method. There may be some situations where calculating item-level change may be helpful.