Pii: S0895-4356(00)00345-0

The purpose of this study was to determine whether individual items in a disability questionnaire were answered differently depending on whether or not the questions were attributed to the upper limb (i.e., “do you have difficulty eating due to your arm or hand problem?” or “do you have difficulty eating?”, respectively). The a priori hypothesis was that the same or more disability would be detected by nonattributed items. Four hundred sixty-seven patients with upper extremity disorders completed the SF-36 general health survey, which does not attribute health problems to affected areas. Patients also completed six additional questions, modified from the SF-36, regarding work (four questions) and social function with friends and family (two questions), which attributed their disability to their affected upper extremity. Of 467 patients, 419–431 (89–92%) responded to both versions of the questions. Although we demonstrated a significant order effect (Generalized Estimating Equation; P 5 .003), comparison of the responses to the six questions showed that for five of the six questions (Generalized Estimating Equation; P < .001) patients reported more disability when the questions were worded with attribution to the upper extremity. Even considering the order effect, patients demonstrated a counterintuitive result by reporting more disability when questions were attributed to their affected area. Thus, both the wording of questions and order of questions can significantly affect patients’ responses about their disability and raises questions about the validity of patients’ reports of their disability. © 2001 Elsevier Science Inc. All rights reserved.