Objective: To develop a short-form version of the Quadriplegia Index of Function (QIF) that would be more practical to use than the original version.
Design: Cross-sectional data collected at 6 months post spinal cord injury.
Setting: Regional Spinal Cord Injury Center.
Patients: Consecutive sample of 95 patients with tetraplegia, non-ambulatory at 6 months, admitted to a regional SCI center between December 1987 and August 1992.
Methods: A short-form QIF was developed by using regression analysis to determine the best six items that would predict the sum of the 37 items selected from seven mobility and self-care categories of the original scale. This short-form QIF was evaluated for internal consistency and discriminant validity. Validity of the short-form QIF was assessed by correlation with motor scores and using analysis of variance by motor levels and motor score groupings.
Results: Regression analysis identified the following items as best predictors of the 37-item QIF score: (1) wash/dry hair, (2) turn supine to side in bed, (3) put on lower body clothing, (4) open carton/jar (feeding), (5) transfer from bed to chair, and (6) lock wheelchair. These items explained 99% of the variance in total scores. Short-form QIF scores (simple sum of the six best items) ranged from 0 to 24, with a median score of seven, interquartile range 0-16. Item response patterns were largely dichotomous. Item-total correlations ranged from 0.60 to 0.80; Cronbach's alpha was 0.89. Spearman correlation coefficient between upper extremity motor score and short-form QIF was 0.82. Analysis of variance indicated that the motor score groupings and motor levels accounted for 63 and 54% of the variance in short-form QIF scores, respectively. Post hoc analyses indicated that motor levels from C5 to T1 had different mean QIF scores, except for C7 versus C8. There may be ceiling effects for individuals with low level injuries.
Conclusion: There is significant redundancy in the QIF. Six items, selected from five categories, yield results comparable to the 37-item QIF. The short-form QIF must next be assessed for sensitivity to change. A brief disability measure would improve data quality and completeness, and may permit ongoing collection of observational rather than self-report data.