1 / 18

Psychometric Assessment of the Neurobehavioral Symptom Inventory (NSI-22)

Psychometric Assessment of the Neurobehavioral Symptom Inventory (NSI-22). Mark Meterko, PhD 1 Errol Baker, PhD 1 Kelly L. Stolzmann, MS 1 Ann Hendricks, PhD 1 Keith D. Cicerone, PhD, ABPP-Cn 2 Henry L. Lew, MD, PhD 3. 1 VA Boston Healthcare System

mab
Download Presentation

Psychometric Assessment of the Neurobehavioral Symptom Inventory (NSI-22)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Psychometric Assessment of the Neurobehavioral Symptom Inventory (NSI-22) Mark Meterko, PhD1 Errol Baker, PhD1 Kelly L. Stolzmann, MS1 Ann Hendricks, PhD1 Keith D. Cicerone, PhD, ABPP-Cn2 Henry L. Lew, MD, PhD3 1VA Boston Healthcare System 2JFK-Johnson Rehabilitation Institute3Defense and Veterans Brain Injury Center This work supported by VA HSR&D Grant: SDR 08-405

  2. Background 1: Postconcussive Syndrome? • Studies of postconcussive symptoms have raised several issues: • Is there a postconcussive syndrome (PCS)? • If yes, is there a single cluster of symptoms, or several? • What symptom patterns distinguish among them? • Previous studies suggest different answers, depending on: • Etiology of injury • Evaluation instrument • Target population • Statistical procedures

  3. Background 2: The VA Context • VA Policy & Process Regarding TBI • Screening • Comprehensive TBI Evaluation (CTE) • Includes Neurobehavioral Symptom Inventory (NSI-22) • Clinical evaluator’s overall judgment regarding history and course “consistent with a diagnosis of TBI”

  4. Background 3: Prior Work Cicerone KD & Kalmar K, 1995 (JHTR) • 22-item self-report inventory of symptoms • 50 mostly vehicular accident patients • Using cluster analysis, 17 items grouped into 4 factors: • Cognitive • Affective • Somatic • Sensory • 5 orphan items

  5. Background 4: NSI22 Recent Work • Benge JF, Pastorek NJ & Thornton GM, 2009. • Postconcussive symptoms in OEF-OIF Veterans: Factor structure and impact of posttraumatic stress. Rehab Psych, 54(3), 270-278. • Exploratory factor analysis yielded 6-factor model: 1. Cognitive 4. Sensory 2. Vestibular 5. Headaches, Sensitivity to light 3. Affective 6. Hearing, Sensitivity to noise • Caplan LJ, Ivins B, Poole JH, Vanderploeg RD, Jaffee MS, Schwab K, 2010. • The structure of postconcussive symptoms in 3 US military samples. JHTR, 25(6), 447-458. • Exploratory & confirmatory factor analysis • Three models supported: 2,3 and 9 factors • Endorsed 3-factor solution 1. Somatic/sensory 2. Affective 3. Cognitive

  6. Purpose: Study Aims • Examine the factor structure of NSI-22 • In large sample of deployed veterans judged to have mTBI • Examine whether & how pain related to other symptoms in the NSI-22 • Assess utility of the factor-based NSI-22 scales • Compare subgroups defined by: • Etiology of concussion • Presence/absence of PTSD

  7. Methods 1: Sample • VA National CTE database for FY08 & most of FY09 • N=36,919 • Random split into 2 samples • Derivation (n=18,459) • Confirmation (n=18,460) • Applied inclusion/exclusion criteria to both samples • Keep only those with “symptoms consistent with TBI” (n=18,649) • Drop those with either pre- or post-deployment concussion history (n=5945) • Drop cases missing on pain (n=663) • Drop duplicate and invalid (test case) entries (n=53) • Final samples • Derivation (n=6001) • Confirmation (n=5987)

  8. Methods 2: Analyses • Check success of randomization • Compare derivation & confirmation samples on • Demographics & etiology (chi-square) • NSI-22 items and pain item (MANOVA) • Examine factor structure of NSI-22 • Derivation sample • Exploratory factor analysis (EFA) – four runs • Empirical criteria for n of factors retained, NSI-22 only • Empirical criteria, NSI-22 plus pain • Specify 4 factors, NSI-22 only • Specify 4 factors, NIS-22 plus pain

  9. Methods 3: Analyses • Confirm factor structure • Confirmation sample • Confirmatory factor analysis (CFA) • Utility of proposed factor-based scales • Confirmation sample • Stratified respondents by: • Etiology • Blast, Non-Blast, Both (“Blast Plus”) • PTSD co-morbidity • Dichotomous based on clinical evaluator judgment during CTE • Two-way MANOVA • Grouping factors (IV): Etiology, PTSD, Etiology x PTSD • Dependent variables: NSI-22 factor scores, with and without pain

  10. Results 1: Randomization Success • No significant differences, derivation vs. confirmation samples on: • NSI-22 symptoms • Pain • Blast injury exposure • TBI diagnosis • Marital status • Education • Borderline exception (p=.07): Employment status • Derivation sample: 7.2% working part time • Confirmation sample: 7.9% working part time • Very small effect size (Cramer’s V = .02)

  11. Results 2: EFA in Derivation Sample Using empirical criteria for N of factors to retain • Three criteria • Percent variance accounted for • Horn’s parallel analysis • Velicer’s Minimum Average Partial (MAP) test • 2- and 3-factor models emerged • Same results with and without pain • Preponderance of evidence favored 3-factor model • Somatosensory (11 or 12 items) • Pain loaded cleanly here when included • Affective (6 items) • Cognitive (4 items) • Orphan items (2 items) • Hearing difficulties (no loading >=.40) • Change in appetite (equal loadings < .40 on two factors)

  12. Results 3: EFA in Derivation Sample Specify 4 factors a-priori • Three items from Somato-sensory form a separate, Vestibular factor • Loss of balance • Dizziness • Poor coordination/clumsiness • Pain remained affiliated with Somato-sensory • Orphans – same as before

  13. Results 5: CFA in Confirmation Sample

  14. Results 6: Utility Analyses in Confirmation Sample • Significant main effects for both Etiology and PTSD co-morbidity • Regardless of whether 3 or 4 factors were compared • Regardless of whether pain was/was not included • No significant Etiology x PTSD interactions • Results for PTSD and no-PTSD respondents the same across Etiology groups

  15. Results 7: Utility Analyses in Confirmation Sample

  16. Results 8: Utility Analyses in Confirmation Sample

  17. Conclusions

  18. Conclusions • PCS for Veterans injured during deployment as measured by NSI-22 are multi-dimensional • Pain associated with Somto-sensory factor in all solutions • By technical criteria, no substantial difference between 3- and 4-factor models • Prefer 4-factor model • In EFA: No dual-loading items in EFA • In CFA: Fit statistics marginally but consistently better • Interpretability and utility of 4-factor model • Increased potential for differentiation among clinical groups

More Related