Reliability of Spanish translations of select urological quality of life instruments

Many patients with urological disease do not speak English. In medical studies restricting patients to those who speak only English undermines efforts to understand disease because restrictions decrease 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 patients were required to speak English. To allow grouping in future studies 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.

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 Prostate Cancer Index and a pain inventory. We then examined English-Spanish reliability by asking bilingual men 50 years old or older to complete English and Spanish versions at the same sitting. A convenience sample was recruited from outpatients and volunteers at the Miami Veterans Affairs Medical Center and population based subjects living in largely Hispanic Hialeah, Florida. Reliability estimates were calculated with kappa coefficients for categorical data and intraclass correlation coefficients for quantitative data.

A total of 100 subjects a median of 59 years old completed the questionnaire, including 55 born in Puerto Rico or Cuba, while the remainder were born at various sites throughout the Americas and Spain. Reliability estimates showed that kappa = > 0.81 for almost all items. For 2 items relating to health and social interactions reliability was poor, and stratification showed that poor reliability was primarily a feature of subjects in good health who are theoretically socially active.

Almost all items tested have excellent English-Spanish reliability in a mixed sample 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 items. Because the sample represents many dialects of Spanish, the translations tested may be transported to other cities. In studies that use these instruments investigators can reasonably group answers from English and Spanish speaking study subjects or study the effects of acculturation on quality of life.