Background: The majority of Americans now access the Internet, thereby expanding prospects for Web-based health-related education and intervention. However, there remains a digital divide among those with lower income and education, and among Spanish-speaking populations in the United States. Additional concerns are the low eHealth literacy rate among these populations and their interest in Internet-delivered interventions with these components. Given these factors, combined with the prevalence of type 2 diabetes among low socioeconomic status and Spanish-speaking Americans, strides need to be taken to reach these populations with online tools for diabetes prevention and management that are at once accessible and efficacious.

Objective: Using a formative evaluation of an eHealth diabetes prevention and control website, we tested the extent to which African Americans, Caucasians, and Mexican Americans at risk for type 2 diabetes gained knowledge and intended to modify their dietary intake and physical activity subsequent to viewing the website. We also examined their general Internet use patterns related to type 2 diabetes.

Methods: A mixed methods approach was undertaken. The diabetes prevention and control website provided educational and behavioral change information in English and Spanish. For this study, eligible participants (1) completed a prequantitative survey, (2) interacted with the website, (3) completed a qualitative interview, and (4) completed a postquantitative survey.

Results: After finding a significant differences in posttest diabetes knowledge scores (P<.001), a regression analysis controlling for pretest score, health literacy, ethnicity, Transtheoretical Model Stage for exercise and fruit and vegetable consumption, and Internet literacy was conducted. Internet literacy score (P=.04) and fruit and vegetable consumption stage (P<.001) were significantly associated with posttest scores indicating that those in precontemplation stage and with low Internet literacy scores were less likely to show improved diabetes knowledge scores. We found significant difference in posttest intention to eat a healthy diet each day in the next 2 months after controlling for pretest score, health literacy, ethnicity, Transtheoretical Model Stage for fruit and vegetable consumption and Internet literacy. Those in the Action stage of the Transtheoretical model for exercise were significantly less likely (P=.023) to improve the posttest score for intention to eat a healthy diet compared to those in the Preparation stage for exercise. We also found that health information is sought commonly across ethnic groups, but that diabetes-related information is less commonly sought even among those at risk. Other specific ethnic usage patterns were identified in the qualitative data including content sought on Web searches and technology used to access the Internet.

Conclusions: This study provides in-depth qualitative insight into the seeking, access, and use of Web-based health information across three ethnic groups in two languages. Additionally, it provides evidence from pre-post measures of exposure to Web-based health content and related changes in diabetes knowledge and intention to eat a healthy diet.