Select Page

Table 17.35 Characteristics of the Neurobehavioral Functioning Inventory

 
Criterion Evidence
Reliability Internal Consistency: Cronbach’s α values for each scale were reported to be 0.93 (depression), 0.86 (somatic), 0.95 (memory/attention), 0.88 (communication), 0.89 (aggression), and 0.87 (motor impairment). For the entire scale, α-0.97 (Kreutzer et al. 1996). Awad (2002) reported α coefficients for each NFI subscale: depression=0.93, somatic=0.83, memory/attention=0.95, communication=0.88, aggression=0.87 and motor=0.88.
Validity

Construct Validity: Factorial analysis of the original 105 scale items revealed a 70-item, 6 factor model with a comparative fit index of 0.89 that was superior to other models tested. Intercorrelations between total subscale scores ranged from 0.44 to 0.67. Awad (2002) reported a Goodness-of-Fit index of 0.71 and Comparative Fit Index of 0.71 for the six factor, 70-item NFI. In general, fit indices for each subscale were higher than for the total scale. 20 items had squared multiple correlations <0.40 (1-depression, 9-somatic, 4-memory, 3-communication, 2-aggression and 1-motor item). Intercorrelations between subscales ranged from 0.56-0.58 and were all significant (p<0.001). This suggests that the NFI may be measuring a single, large underlying construct.

Construct Validity (Known Groups): Scores on depression (p<0.002), memory/attention (p<0.002), communication (p<0.001), aggression (p<0.002) and motor (p<0.002) subscales could distinguish between groups based on employed versus unemployed persons who had sustained traumatic brain injury (Sander et al. 1997). Comparison of subscale scores for individuals with TBI versus non-clinical controls via ANCOVA revealed no significant differences between groups on the depression, memory/attention, communication and motors subscales when controlling for the effects of age and sex. The only significant differences appeared on the somatic subscale (p<0.01) on which non-clinical controls achieved higher scores than the TBI group (Awad, 2002).

Concurrent Validity: Scores on NFI subscales were correlated with the following scales from the MMPI-hypochondriasis, depression, hysteria, psychasthenia, schizophrenia. Correlations between NFI subscales and MMPI scales were all significant (p<0.001). Correlations between MMPI hypochondriasis and NFI subscales ranged from 0.24 (aggression) to 0.65 (somatic), for MMPI depression correlations ranged from 0.21 (aggression) and 0.47 (depression, motor and somatic), for MMPI hysteria from 0.25 (communication) to 0.50 (somatic), for MMPI psychasthenia from 0.26 communication) to 0.43 (depression) and for MMP schizophrenia from 0.25 (aggression) to 0.40 (depression) (Kreutzer et al. 1996) NFI Communication subscale correlated significantly with scores on neuropsychological measures of attention, memory and learning, communication and visual and motor functioning (p<0.001). No other subscale correlated significantly with any of the neuropsychological tests administered with the exception of memory/attention which correlated with scores on the Symbol Digits Modalities Test Oral (Kreutzer et al. 1996). NFI memory/attention correlated significantly with WMS-Logical Memory raw scores (r=-0.26, p<0.001) and with the WMS-R Logical Memory recall scores (r=-0.26, p<0.001), NFI motor scores correlated with Trailmaking tests A (r=0.27, p<0.001) and B (r=0.25, p<0.001) and Grooved Pegboard scores (r=-0.28, p<0.001). NFI communication correlated with Controlled Word Association Test adjusted scores (r=-0.18, p<0.001) (Awad, 2002).

Responsiveness N/A
Tested for ABI/TBI patients? Head injury specific.
Other Formats

NFI-66: Developed by Kreutzer & Devany (unpublished). Weinfurt et al. (1999) performed factor analysis revealing 4 components with eigenvalues>2.0; cognitive deficits, depression, aggression and somatization. Internal reliability of the 4 scale NFI-66 ranged from 0.79 (aggression) to 0.92 (cognitive deficits and depression). Significant correlations were reported between NFI-66 scale scores and the GOS ranging from 0.21 (depression)-0.26 (somatization). Aggression subscale scores did not correlate with GOS scores. Scores on the Euroqol VAS were significantly and inversely correlated with NFI subscale scores ranging from 0.17 (aggression) to 0.50 (depression).

NFI-D: A 13-item Depression subscale of the NFI. Seel & Kreutzer (2003) reported high internal consistency (α=0.93). Convergent and discriminant validity was supported as scores on the NFI-D correlated with both Beck Depression Inventory scores (r=0.765) and MMPI-2 Depression scale T-scores (r=0.752) but not significantly with MMPI-2 hypomania scale scores (r=0.159). Normal and clinically depressed BDI scores were accurately predicted by NFI-D scores 81% & 87% of the time, respectively. Patients who were classified with mild or borderline depression on the BDI were less likely to be correctly classified as such by the NFI-D. Using the MMPI-2 Depression score classifications-normal versus depressed classifications could be accurately predicted by NFI-D scores 75% & 83% of the time, respectively. Via mapping to the BDI, the following score ranges were proposed for the identification and classification of depression: ≤ 28 (minimal depression), ≥43 (clinical depression-moderate to severe), 29-42 (mood disturbance). However, classification in the last range is considered to be a borderline region and contains many false positives and false negatives.

Use by proxy?

Test contains forms for ratings by self and by significant other (proxy). Correlations between self and SO ratings were moderate for communication and memory/attention and weaker/not significant for motor, depression, somatic and aggression scales. Self-ratings were significantly higher than SO ratings from somatic, memory/attention and communication scales (Rush et al. 2004).

Concordance coefficients (between patient and significant other) ranged from 0.63 (aggression) to 0.76 (somatic). Significant others rated symptoms in the aggression scale as being significantly more frequent than the patients. A similar trend was observed for ratings of symptoms on the depression subscale. Such discrepancies were noted more for cognitive or behavioural symptoms, not for physical or somatic ones (Hart et al. 2003).

Seel et al. (1997) (TBI) reported that agreement between family and patient ratings ranged from 48% to 84% and, for the most part, family members and patients tend to rate problems as occurring at the same frequency. On an item by item analysis, there were no statistical differences for ratings on 57 of 70 items. On the 13 statistically different items, patients rated problems as more frequent than family members. The only scale score that demonstrated statistically different ratings (family versus patient) was the communication scale (p<0.01).