Development of a Diabetes-Focused Print Health Literacy Scale Using the Rapid Estimate of Adult Literacy in Medicine Model

BACKGROUND: A diabetes mellitus (DM)-specific health literacy (HL) measure that focuses on both oral and print HL is needed in clinical and research settings. OBJECTIVE: The present study developed a psychometrically sound DM-specific HL instrument that measures oral and print HL. METHODS: We developed the measure in three steps. First, we reviewed clinical guidelines and conducted focus groups with experts to generate items. Next, we conducted a psychometric evaluation of the scale in three language versions (English, Spanish, and Korean). Lastly, we identified and removed items with potential cultural bias and duplicate functions to produce shorter versions of the scale, using item response theory (IRT). KEY RESULTS: We initially developed an 82-item DM-specific oral HL scale using the Rapid Estimate of Adult Literacy in Medicine (DM-REALM) model. To improve the clinical utility of the DM-REALM, we created shorter forms, a 40-item and 20-item version, and evaluated them by using IRT. All DM-REALM versions had high Cronbach alphas (.985, .974, and .945, respectively) and yielded sufficient convergent validity by positive correlations with existing functional HL scale (r = .49, p < .001), education (r = .14, p = .14 to r = .54, p < .001), and DM knowledge (r = .04, p = .70 to r = .36, p < .001). DM-REALM also demonstrated adequate sensitivity as an intervention evaluation tool that captures the changes induced by an intervention. CONCLUSIONS: All forms of the DM-REALM tool were reliable, valid, and clinically useful measures of HL in the context of DM care. Both researchers and clinicians can use this tool to assess DM-specific HL across multiple racial and ethnic populations. [HLRP: Health Literacy Research and Practice. 2020;4(4):e237-e249.] PLAIN LANGUAGE SUMMARY: This article reported the process and findings of a newly developed health literacy scale for people with diabetes mellitus using three different language versions. Both long and short versions of the scale demonstrated adequate validity and reliability.