IMPROVED DISCRIMINATIVE AND EVALUATIVE CAPABILITY OF A REFINED VERSION OF SKINDEX, A QUALITY-OF-LIFE INSTRUMENT FOR PATIENTS WITH SKIN DISEASES

Objective: To improve Skindex, a dermatologic quality-of-life instrume nt. Design: Cross-sectional and longitudinal questionnaire study. Sett ing: Dermatology clinic of a Veterans Affairs hospital and private der matology practices. Patients: Patients waiting for dermatology appoint ments; 201 patients responded to the original version of Skindex and 6 92 additional patients to the revised version. Main Outcome Measures: Reproducibility, internal consistency reliability, and validity of the revised version of Skindex. The revised version was compared with the original in 3 ways: the amount of time patients need to complete it; discriminative capability, determined as the number of items to which patients chose the same response; and evaluative capability, determine d as the number of scales that were responsive to patients' reports of clinical change. Results: With the revised 29-item version of Skindex , scale scores were reproducible after 72 hours (v=0.88-0.92) and were internally reliable (Cronbach alpha=0.87-0.96). The instrument demons trated both construct and content validity: patients with psoriasis an d eczema responded with higher scores than those with isolated lesions ; in an exploratory principal axes factor analysis with an oblique rot ation, 97% of the common variance was explained by 3 factors that corr elated with the a priori scales; and most patients' responses to an op en-ended question about their skin disease were addressed by items in the instrument. The average time to complete the revised instrument wa s 5 minutes (compared with 15 minutes for the original version). For o nly 3 items (10%) did 70% or more of patients choose the same response (vs 17 [28%] of items in the original version). All scales changed si gnificantly in the expected direction in patients who reported that th eir skin had changed after 3 months (vs only 3 of 8 scales originally) . Conclusion: The 29-item version of Skindex remains reliable and vali d, but has decreased respondent burden and improved discriminative and evaluative capability.