When back translation is not enough: Questionnaire development for a diabetes intervention targeting low income Latinos

Diabetes self-care can significantly decrease diabetes-related complications, including death. Latinos diagnosed with diabetes tend to have higher levels of A1C over time, suggesting poor diabetes control. We developed a pilot intervention (Latino Wellness, Prevention and Advocacy (LWPA) Project of New Jersey) based on core elements proposed by the American Diabetes Association to increase diabetes knowledge, daily monitoring of blood glucose, and change in dietary and physical activity behaviors among low literacy, low income Latinos. The use of previously tested, reliable instruments proved challenging to apply to our target population. Specifically, participants had difficulties interpreting 5 or 6-item Likert scales, although these items had been shown to be reliable measures in other studies and had used back translation techniques. We present two waves of pilot data on the use of these original scales (n=40), and a modified version of the scales using a 3-item response options (n=30) across various health belief domains. We discuss the relative trade-off between the use of multiple response items versus low literacy ‘friendly' measures, content areas where most challenges arose, and the reliability of both sets of items. We suggest the need to incorporate literacy and socioeconomic factors, in addition to back translation techniques, to ensure that evaluation instruments adequately measure expected domains.