[Consequences of errors in the translation of questionnaires: Spanish version of Downton index]

The application of screening tools to detect the risk of falls in hospitalized patients is in general use. During the development of a systematic review a serious disparity in three items of the Spanish version of the Downton index was detected, compared to the original version. The aim of this study was to determine the impact of this error and to compare the estimated risk of falls with each of these versions, its validity and internal consistency.

A descriptive cross-sectional study in acute hospitalised patients was performed during 2011 in Hospital Costa del Sol, Marbella. The patients' risk of falling was assessed by the Spanish version of the Downton index, and then it was re-calculated according to the items in the original version. Sensitivity, specificity and Cronbach's alpha were calculated.

Application of the original version of the index reduced the number of patients classified as "high risk" of falling by 24.2%. With the Spanish version of the tool, the possibility of being classed as "high risk" of falling was considerably 3.3 times higher (OR: 3.3). Both versions of the Downton index showed low accuracy and diagnostic validity. The sensitivity of the original scale was 28% and specificity of 82%. Its internal consistency was low (Cronbach's alpha: .51).

The Downton index, given its poor accuracy and diagnostic validity, low internal consistency, and the significant error observed in its Spanish translation, is not the most appropriate tool to assess the risk of falls in hospitalised acute patients.