ÐÇ¿Õ´«Ã½ÊÓƵ

Adult Spinal Cord Injury: (Behrman et al, 2012; n=95,Mean age=43(17) Mean time post SCI 1 year (0.1,25.8) Severity: AIS level  D=64 C= 31),  Phases listed below indicate progression of motor function and are not cutoff scores.

• Phase I ( scores a-c) 50-60% Body weight support for stand retraining, <60%,>20% for treadmill training .27-.54 m/s, >40% to 20% for stand retraining

• Phase II (scores a-c) <10% and >20% body weight support for stand retraining, <20% body weight support for treadmill speed .27-.54m/s, <20% and >10% for stand retraining
• Phase III (scores a-c) 0-9% body weight support for stand retraining, <20% body weight support treadmill speed .58-.89m/s, 0-9% for stand retraining
• Phase IV (only)Independent stepping and <10% body weight support for speeds >1.52 m/s

Adult Spinal Cord injury: (Behrman, Velozo, Suter, Lorenz, & Basso, 2015; n = 69; mean age = 36(15); AIS level: A = 18, B = 17, C = 13, D = 21; time since injury = 3.3 years (7 years); 13 raters: PT = 1, MPT = 2, DPT = 10)

• Excellent Test-retest reliability of NRS items 10/11 items had Spearman correlation coefficients of > .92. Measurement model-derived summary score was (¦Ñ=.99; 95% CI, .96¨C.99).

Interrater Reliability:

Adult Spinal Cord Injury: (Basso, Velozo, Lorenz, Suter, & Behrman, 2015; n = 10; mean age = 43(18); AIS level: A = 1, B = 1, C = 2, D = 6; time since injury = 36 months(3-119); 15 raters: PT = 1; MPT = 2, DPT = 11, PhD = 1)

• Excellent Reliability Kendall Coefficient of concordance (W) Interrater reliability generally strong W=.91-.98 C.I.=.65-.99 
• Excellent Lower reliability for treadmill stand retraining W= .87 C.I. + .06-1
• Excellentseated trunk extension: W=.82 C.I. .28-.94

Construct Validity:

Adult Spinal Cord Injury: (Velozo et al., 2015; n = 188; mean age = 39.3; AIS level: A = 20, B = 19, C = 49, D = 98; time since injury = 1.2 years))

Acceptable overall construct validity for this study.  Areas of opportunity are described below.

• NRS met many Rasch model criteria for construct validity. There were five different components to construct validity for this study:

(1) dimensionality;

(2) item rating structure;

(3) fit of items and patients to the Rasch model;

(4) item difficulty hierarchy; and

(5) person ability-item difficulty match (Velozo et al., 2015).

• Confirmatory Factor Analysis (CFA) was utilized for dimension analysis assessment of the NRS.  Within CFA, principal component analysis was used for dimension reduction.  Instead of traditional psychometric analysis that focuses on test level psychometrics, Rasch analysis was utilized which is more of a probability focused method at the item level (Velozo et al., 2015). Infit and outfit statistics were utilized to validate items and patient fit. (Velozo et al., 2015).  Overall fit was assessed as well to look at observed over expected variance; the ideal for this measurement would be 1.0 (Velozo et al., 2015).
• Results of the CFA chi-squared root mean square error (RMSE) test failed at .217; however results of the CFI were met at .955. (Velozo et al.,2015). The PCA from the Rasch model explained 76.9% of the item and person variance (Velozo et al.,2015). When looking at the finer results of the CFA it appears some of the eigenvalues were over 2.0.  In addition, the variance explained by the first component was 5.3%; the remaining items explained 43.4%, which shows the strength of the measurement construct (Velozo et al., 2015).
• Some model fit statistics include: ten of the eleven items met the criteria for model outfit and infit and 91% of the patients fit the measurement model (Velozo et al., 2015). Precision was assessed using Person separation reliability and results were .92 which is similar to Cronbach¡¯s alpha; and item separation reliability was .90. (Velozo et al, 2015).  It is no surprise based on these results that 9 of the 11 items were highly discriminatory.
• Regarding the person ability-item difficulty match, the distribution of patients according to AIS classifications is adequate ( Velozo et al., 2015).
• Overall these results show a lot of strong characteristics of solid construct validity for the NRS.  However, there is an opportunity for improvement.  CFA criteria was not completely met for unidimensionality and there were a few items in NRS that were not discriminatory (Velozo et al., 2015)

Adult Spinal Cord Injury: (Behrman et al, 2012)

• The NRS was developed between 2000 and 2008 by a team of scientists and physical therapists within the NRN.

Adult Spinal Cord Injury: (Velozo et al., 2015)

• Excellent Rasch analysis identified that the NRS did not show ceiling or floor effects.

Adult Spinal Cord Injury: (Tester et al., 2015; n = 72; mean age = 36 (15); AIS level: A = 17, B = 10, C = 20, D = 25; time since injury = 1.2 yrs (0.1-53.1))

• The NRS was significantly responsive for SCI outpatients (ARM = 1.05; CI = 0.75-1.35).