Translation and psychometric testing of the Basque version of the SF-36 Health Survey

Introduction: The primary goals of this study were to adapt the Short Form-36 (SF-36) questionnaire to the Basque language and to assess the translated questionnaire's psychometric properties. Methods: Two pilot studies and a field study were conducted in 1999. The pathologies selected were hypertension, eating disorders, back pain, and HIV infection; blood donors were also included. The analyses conducted at item-level were: item internal consistency, item discriminant validity, equal item variance, and equal item-scale correlation. Internal consistency was measured at the scale level. Reproducibility, convergent validity and discriminant validity were also examined. Results: 285 individuals took part in the study; sufficient data were obtained for 265 (93%). All items correlated higher than 0.4 with their hypothesized scales. Within each scale, item standard deviations were similar. Item-scale correlations were also similar. Cronbach's alpha coefficients varied from 0.76 to 0.92. In the test-retest reliability study, which included 54 additional individuals, all scales were above 0.50. Regarding the inter-scale correlation, the majority exceeded the 0.40 coefficient. In the correlation of similar scales of the SF-36 and the Nottingham Health Profile, intraclass correlation coefficient results ranged from 0.29 to 0.62. Conclusions: These results confirm that this new version of the SF-36 has been translated and adapted correctly for the Basque language and that it fulfils, at least partially, the psychometric properties required for this instrument. Further studies, however, are needed to completely validate the Basque version of the SF-36.