Rasch Measurement Properties of the KIDSCREEN Quality of Life Instrument in Children with Cerebral Palsy and Differential Item Functioning between Children with and without Cerebral Palsy.
ABSTRACT Objective: To assess if the Rasch-scaled KIDSCREEN-52 generic health-related quality of life measure was valid in children with cerebral palsy (CP). Methods: The Rasch measurement properties and differential item functioning (DIF) of the KIDSCREEN-52 were examined in children with CP. Data were available from the KIDSCREEN project from 3219 children aged 8 to 12 years and 2126 parents in the general population; and from the SPARCLE project from 501 children aged 8 to 12 years with CP and 823 parents. Analysis used Zumbo's logistic regression DIF approach. Partial credit model analyses were conducted. Results: All items of the KIDSCREEN self-report version fitted the partial credit model (smallest P-value: 0.256). Only one item of the parent version did not fit the data well (smallest P-value 0.001). Statistically significant DIF was observed in some items, but was of substantial magnitude (DeltaR(2) = 0.046, 0.049) for only two items in two dimensions of the parent version. The practical impact of DIF was small. DIF-adjusted standardized mean differences between children with and without CP being 1.07 and 0.34 for the physical and school dimensions, respectively (unadjusted: 1.09 and 0.16). Conclusion: The KIDSCREEN-52 functions in a similar way in children with CP and in the general population. Comparisons of quality of life between such children are therefore likely to be valid.
ABSTRACT Objective: To assess if the Rasch-scaled KIDSCREEN-52 generic health-related quality of life measure was valid in children with cerebral palsy (CP). Methods: The Rasch measurement properties and differential item functioning (DIF) of the KIDSCREEN-52 were examined in children with CP. Data were available from the KIDSCREEN project from 3219 children aged 8 to 12 years and 2126 parents in the general population; and from the SPARCLE project from 501 children aged 8 to 12 years with CP and 823 parents. Analysis used Zumbo's logistic regression DIF approach. Partial credit model analyses were conducted. Results: All items of the KIDSCREEN self-report version fitted the partial credit model (smallest P-value: 0.256). Only one item of the parent version did not fit the data well (smallest P-value 0.001). Statistically significant DIF was observed in some items, but was of substantial magnitude (DeltaR(2) = 0.046, 0.049) for only two items in two dimensions of the parent version. The practical impact of DIF was small. DIF-adjusted standardized mean differences between children with and without CP being 1.07 and 0.34 for the physical and school dimensions, respectively (unadjusted: 1.09 and 0.16). Conclusion: The KIDSCREEN-52 functions in a similar way in children with CP and in the general population. Comparisons of quality of life between such children are therefore likely to be valid.