RELIABILITY OF SPANISH TRANSLATIONS OF SELECT UROLOGICAL QUALITY-OF-LIFE INSTRUMENTS

Purpose: Many patients with urological disease do not speak English. I n medical studies restricting patients to those who speak only English undermines efforts to understand disease because restrictions decreas e efficiency of patient recruitment, and because language and culture are associated with variable outcomes. In Spanish speaking locations, such as South Florida, studies would suffer severe selection bias if p atients were required to speak English. To allow grouping in future st udies of English and Spanish speaking patients we examined the English -Spanish reliability of select instruments that measure health related quality of life in patients with urological disease. Materials and Me thods: We assembled available Spanish versions and translated English versions of questions regarding satisfaction, the American Urological Association symptom index, the University of California, Los Angeles P rostate Cancer Index and a pain inventory. We then examined English-Sp anish reliability bg asking bilingual men 50 years old or older to com plete English and Spanish versions at the same sitting. A convenience sample was recruited from outpatients and volunteers at the Miami Vete rans Affairs Medical Center and population based subjects living in la rgely Hispanic Hialeah, Florida. Reliability estimates were calculated with kappa coefficients for categorical data and intraclass correlati on coefficients for quantitative data. Results: A total of 100 subject s a median of 59 years old completed the questionnaire, including 55 b orn in Puerto Rico or Cuba, while the remainder were born at various s ites throughout the Americas and Spain. Reliability estimates showed t hat kappa = >0.81 for almost all items. For 2 items relating to health and social interactions reliability was poor, and stratification show ed that poor reliability was primarily a feature of subjects in good h ealth who are theoretically socially active. Conclusions: Almost all i tems tested have excellent English-Spanish reliability in a mixed samp le of bilingual men. Nonreliability of 2 items relating to health and social interactions probably originates from the effect of language on perception, and invalidates English and Spanish grouping of these ite ms. Because the sample represents many dialects of Spanish, the transl ations tested may be transported to other cities. In studies that use these instruments investigators can reasonably group answers from Engl ish and Spanish speaking study subjects or study the effects of accult uration on quality of life.