Interventions for Low Health Literate Audiences
The Internet is becoming increasingly ubiquitous, with recent reports indicating that 70% of American adults use the Internet at least occasionally (Pew Internet and American Life Project, 2007). A large majority of those using the Internet (8 in 10) have used it as a source of health information, and on a daily basis six million Americans seek health information on the Internet (Fox, 2006; Schloman, 2003).
Not surprisingly, given Internet users’ existing interest in health information, researchers have become increasingly focused on how the Internet can be used for health education and promotion. Web-based interventions have been used successfully to address an array of health concerns, from hospice care (Willis, Demiris, & Oliver, 2007) to cancer (Friedman, Hoffman-Goetz, & Arocha, 2006) to diabetes management (Bowman, Gregg, Williams, Engelgau, & Jack, 2003). There are many benefits to be gained from receiving health information online. As one example, caregivers of hospice patients, health care providers, and patients use the Internet to meet both support and information needs; they cited anonymity, the ability to research alternative treatments, and ease of finding additional information about their illnesses as advantages of online health information (Willis, Demiris, & Oliver, 2007).
The successful use of digital media to distribute and receive health information is challenging for a number of reasons, however. Chief among these are concerns about quality of information on the Internet and the level of expertise some users might have with the Internet (Suggs, 2006). Disparities in Internet access, variations in computer knowledge, and the speed of Internet connections are all additional factors that must be considered, since at the most basic level it is important that individuals have access to the channel being used to provide health information (Kreuter & McClure, 2004; Leaffer & Mickelberg, 2006).
Beyond access to information and the quality of that information, there is a growing recognition among healthcare researchers and professionals about the role that literacy plays in individuals’ health outcomes (Nielsen-Bohlman, Panzer, & Kindig, 2004; Schwartzberg, Vangeest, & Wang, 2004; Zarcadoolas, Blanco, Boyer, & Pleasant, 2002; Zarcadoolas, Pleasant, & Greer, 2006). This is an important issue, as research has indicated that adults with low literacy are using the Internet to meet information needs (Poftak, 2002; Zarcadoolas, Blanco, Boyer, & Pleasant, 2002). Being able to appropriately process and act on health information, though, is more complex than just the ability to successfully read through health information. Health communication researchers are increasingly focusing on health literacy as an issue, with health literacy defined as the ability of individuals to obtain, process, and understand health information and services to make appropriate health decisions (Ad Hoc Committee on Health Literacy, 1999; United States Department of Health and Human Services, 2000). Issues related to health literacy result in enormous losses to the United States healthcare system, as much as $69 billion per year according to some research (Nielsen-Bohlman, Panzer, & Kindig, 2004). These costs are due to more frequent visits with healthcare professionals, increased rates of medication and treatment errors, more hospitalizations, longer hospital stays, and a lack of the skills necessary to obtain needed services (Cuban, 2006; Schloman, 2003).
The use of interactive health communication and new technology allows for promising strategies that can be used to address some of the issues facing low health literate individuals (Eng, 2002; Friemuth & Quinn, 2004). Indeed, research has already demonstrated the ability of specially-designed websites to provide information to low health literate audiences. Whitten et al. (in press) provided data on the initial evaluation of a website, titled Diabetes and You, designed to provide diabetes information to low health literate audiences. Research participants approved of the site’s design, particularly an interactive risk-assessment tool. Participants increased their knowledge of diabetes after exploring the site’s contents, and the amount they learned was not linked to their level of health literacy.
This is a promising area for further investigation, both to confirm earlier results with a larger sample (pilot work was more qualitative with smaller samples) and gain a deeper understanding of how such websites work and how they can be improved. One area of particular interest relates to culture, as such interventions could potentially be tailored to meet users’ needs, a strategy often considered promising for improving users’ response to interventions and achieving longer lasting results (Kreuter & McClure, 2004; Oenema, Brug, & Lechner, 2001). This research advances research into the proper design of e-health interventions for low health literate audiences and an understanding of how such health communication might be received by more literate audiences.
While this research on the design and usability of e-health interventions for low health literate audiences was not intended to explicitly test a theoretical model, the Health Belief Model (HBM) remains an extremely relevant framework for considering the work and guiding this investigation (Becker, 1974; Janz & Becker, 1984; Janz, Champion, & Strecher, 2002). Research utilizing the HBM has been used to study a range of health communication problems, including how the content of health-related advertising might contribute to health disparities among different ethnic groups (Duerksen et al., 2005), racial and ethnic barriers to flu vaccinations (Chen, Fox, Cantrell, Stockdale, & Kagawa-Singer, 2007), and knowledge of diabetes and management of the disease (Powell, Hill, & Clancy, 2007). This study’s interventions include content related to core HBM constructs, as they were designed primarily to increase knowledge of specific health issues (perceived severity, perceived susceptibility, etc.) to promote behavior change. As such, this work lays the foundation for research into how interventions designed for low health literate audiences might educate and promote behavior change, given the primary role of knowledge as the foundation for individuals’ perceptions of health issues and preventive health behaviors (Becker, 1974).
This paper continues with an overview of the research questions and hypothesis that guided this study, the websites studied, and the methods used to conduct the investigation. Study results are reported, followed by a discussion of the implications for researchers and health communication professionals. Future directions for further investigation are also considered.
This study was meant to further investigate the potential of two specially designed websites to provide health information to low health literate audiences. The research was guided by two exploratory research questions and one hypothesis:
• RQ1: How does a broader, more health literate audience perceive the websites in regard to content, design, and usability?
• RQ2: Can websites specifically designed for low literate and low health literate populations be effective at delivering health information to a wider audience?
• H1: Respondents will prefer information providers that match their ethnicity.
This research explored two websites that were designed to provide health information to low literate and low health literate audiences. Both websites were created by researchers and designers from the Michigan State University Communication Technology Laboratory. The websites were Flash-based and designed so that any current Web browser could view the material. The content of the websites was developed and approved by a team of healthcare providers and educators. These interventions were designed for audiences with potentially low reading levels, so users were verbally presented with information by animated health providers. Text is used sparingly to highlight key points, while information is presented via audio and supported by relevant images.
The first website, Diabetes and You1 (Figure 1), was created to provide information about diabetes to non-diabetics. Diabetes and You featured basic information about diabetes, signs and symptoms of diabetes, diet and exercise tips that could help prevent diabetes, and a tool that let users determine their own risk of developing diabetes. All of these sections served to increase perceived benefits of behavior change, while removing concerns about potential barriers; the risk assessment tool was a feature explicitly designed to impact users’ perceived susceptibility of disease. Diabetes and You featured two characters, an African American male doctor and a white female doctor. The website was controlled by simple, VCR-style controls that let users rewind, pause, and advance through the content sections. While users can explore the website’s content areas in any order they might prefer, users who follow the doctors’ instructions progress through the website in a simple, linear fashion.
The second intervention, the Child Care Center2 (Figure 2), provided information about promoting healthy eating habits in children, dealing with newborns, and what to do when children become sick. This website featured four characters: (1) Nancy, a female nurse that welcomes users to the website; (2) Dr. Garcia, a Hispanic male doctor; (3) Dr. Smith, a white female doctor; and (4) Dr. Lee, an Asian male doctor. The Child Care Center website featured a more sophisticated design that let users choose their own path through the site, as well as more advanced production values (e.g., the impression of a moving camera as the user enters a doctor’s office).
Study Procedures and Sample
Two online surveys were conducted to complete this research, one for the Diabetes and You website and one for the Child Care Center. The same procedure was used for each survey, with participants recruited from a panel of survey participants by e-mail invitations. The panel is an opt-in pool of over 18,000 international participants who are entered into monthly drawings for membership in the panel. Subjects are randomly selected and invited to participate in specific research studies, typically with a cash incentive provided for participation in the project.
Participants first completed a pre-test of knowledge on the topic covered by the website. They were then asked to spend 10–15 minutes exploring the website, followed by a post-test of knowledge to assess any gain in knowledge that resulted from viewing the intervention. Subjects completed a survey to gauge their reaction to both the content and design of the website. Finally, participants completed the Short Test of Functional Health Literacy in Adults (STOFHLA) to assess their health literacy (Baker, Williams, Parker, Gazmarian, & Nurss, 1999; Parker, Baker, Williams, & Nurss, 1995).
There were 783 respondents (N = 783) to the Diabetes and You survey for a response rate of 26.8%. The sample was 57.2% female and 42.8% male, with an average age of 45.1 (SD = 13.1). Subjects were asked to report their ethnicity by checking all appropriate ethnicities. The most common ethnicity was white (53.3%), followed by Hispanic (5.1%), Asian (4.1%), African American (1.8%), and Native American (.6%).
The Child Care Center survey was completed by 547 subjects (N = 547) for a response rate of 23.0%. That sample included a larger proportion of females (71.0%) to males (29.0%), with an average age of 43.0 (SD = 11.9). The Child Care Center was viewed by more whites (46.3%) than other ethnicities. The sample also included Asians (2.4%), Hispanics (2.2%), African Americans (1.6%), and Native Americans (1.1%).
Results showed that respondents were generally positive about both websites, approving of both the content and design of these interventions. The results are provided below, organized by the research questions and hypothesis that guided this research.
RQ1: How does a broader, more health literate audience perceive the websites in regard to content, design, and usability?
Given these websites’ initial intended audience, it was important to investigate how more health literate audiences would respond to these interventions. Respondents were asked to assess the quality of the content in terms of accuracy, subjects’ ability to understand the information, and the amount of information provided on a scale of 1–7 (with 7 a positive evaluation of the content). Both websites received favorable reviews, with Diabetes and You (M = 5.7; SD = 1.2) rated more positively than the Child Care Center (M = 5.4; SD = 1.1).
Not surprisingly, the Child Care Center (M = 5.5; SD = 1.3) was more popular than Diabetes and You (M = 4.6; SD = 1.5) when respondents were asked to comment on website design. The Diabetes and You website was the designers’ first effort at creating this sort of intervention, so the doctors’ mouths are not matched as well with their words and the graphics in general are not as sophisticated as the Child Care Center. The Child Care Center (M = 5.7; SD = 1.3) also achieved higher ratings of usability than Diabetes and You (M = 5.4; SD = 1.6).
Respondents were also asked to assess the websites’ characters in terms of helpfulness and appearance. Again, the more advanced production values of the Child Care Center were evident. Nancy (M = 5.5; SD = 1.4), Dr. Garcia (M = 5.3; SD = 1.5), Dr. Smith (M = 5.5; SD = 1.5), and Dr. Lee (M = 5.5; SD = 1.4) were all rated more highly than the African American male doctor (M = 5.0; SD = 1.3) and white female doctor (M = 4.8; SD = 1.3) from the Diabetes and You website.
A more sophisticated design could also have contributed to subjects’ self-reported interest in the websites, as that also favored the Child Care Center (M = 4.8; SD = 1.3) over Diabetes and You (M = 4.4; SD = 1.2).
RQ2: Can websites specifically designed for low literate and low health literate populations be effective at delivering health information to a wider audience?
As the last step in data collection, subjects completed an online version of the STOFHLA. Scores on the STOFHLA range from 1–36; scores from 0–16 represent inadequate health literacy, 17–22 is a marginal level of health literacy, and scores in the 23–36 demonstrate adequate health literacy. Not surprisingly, given the nature of these online samples, both groups demonstrated relatively strong health literacy. The Diabetes and You sample (M = 33.6; SD = 6.6) had slightly higher health literacy than the Child Care Center sample (M = 33.1; SD = 5.9). Small percentages of respondents (4.7% for Diabetes and You, 6.0% for the Child Care Center) had inadequate or marginal health literacy.
In an effort to test the educational value of these websites, respondents were tested for knowledge before and after the interventions using 18-item pre-tests and post-tests. For Diabetes and You, the average score on the diabetes pre-test was 14.7 (SD = 1.7), while the post-test was 15.9 (SD = 2.0). A paired samples t-test (t(488) = 13.99, p < .001) demonstrated the statistical significance of that difference. For the Child Care Center, the average score on the child care pre-test was 10.9 (SD = 2.4), which improved to an average of 11.6 (SD = 3.1). A paired samples t-test (t(282) = 7.06, p < .001) indicated that this difference was statistically significant.
Finally, the relationship between health literacy (as measured by the STOFHLA) and gain in knowledge (as measured by subjects’ improvement from pre-test to post-test) was explored to assess the role of health literacy in knowledge improvement. For the Diabetes and You sample, health literacy and knowledge gain were indeed related (r = .18, p < .01). This was also the case for the Child Care Center sample (r = .09, p < .05). While both of these relationships were statistically significant, the actual variance explained in knowledge gain by health literacy level is actually relatively small (3% and 1%, respectively).
H1: Respondents will prefer information providers that match their ethnicity.
One of the potential benefits of digital media is the ability to customize content to match users’ cultural backgrounds and preferences. While neither of these websites were designed to enable explicit customization, it is possible to evaluate the potential preference of users for an information provider matching their own ethnicity.
For the Diabetes and You website, African Americans’ ratings of the African American male doctor (M = 4.9; SD = 1.7) and the white female doctor (M = 4.9; SD = 1.7) did not demonstrate any significant preference based on ethnicity (t(13) = .22, p = .83). In contrast, whites had higher ratings of the African American doctor (M = 5.0; SD = 1.3) than the white doctor (M = 4.9; SD = 1.3). While this difference is indeed statistically significant (t(410) = 5.62, p < .001), there is little practical difference as it might influence intervention design.
In the Child Care Center website, Hispanics’ preference for Dr. Garcia (M = 6.1; SD = 1.1) over Dr. Smith (M = 5.8; SD = 1.2) and Dr. Lee (M = 6.1; SD = 1.1) was assessed. This indicates that Hispanics did not prefer Dr. Garcia over Dr. Smith (t(11) = 1.6, p = .13) or Dr. Lee (t(11) = .76, p = .46). A similar comparison was used to explore Asians’ preference for Dr. Lee (M = 5.4; SD = 1.2) over Dr. Garcia (M = 5.3; SD = 1.0) and Dr. Smith (M = 5.5; SD = 1.0). Again, paired sample t-tests indicated that Asians did not favor Dr. Lee over Dr. Garcia (t(12) = .16, p = .87) or Dr. Smith (t(12) = .74, p = .48). Finally, whites’ preference for Dr. Smith (M = 5.4; SD = 1.5) over Dr. Garcia (M = 5.3; SD = 1.5) and Dr. Lee (M = 5.5; SD = 1.5). The difference between Dr. Smith and Dr. Lee was not significant (t(247) = .51, p = .61), but Dr. Smith was preferred to Dr. Garcia (t(247) = 3.02, p < .01). Again, that statistical significance might overstate any difference that could be leveraged to improve intervention design.
Overall, these results from tests of both websites indicate that there is little support for H1. Indeed, the two statistically significant differences were more indicative of a large sample size than any practically useful difference.
This project built on earlier work (Whitten, Love, Buis, & Mackert, in press) to explore the potential for websites designed for low health literate groups to provide health information to a more general audience. The findings are useful for several reasons, not the least of which is that it is the first large-scale evaluation of the efficacy of websites designed specifically to provide health information to low health literate audiences. While further work must continue to explore effective methods for improving health education for low health literate audiences, it is important to recognize that even individuals with adequate health literacy scores (as measured by the established scales of health literacy) have favorable impressions of this kind of intervention and appreciate a simpler and clearer model of acquiring health information. This is an important finding, as even more literate audiences can appreciate health information being presented in a simpler fashion. This is certainly the case as patients are expected to understand increasingly complex medical information as science and medicine continue to advance.
The evolution of these two websites from the initial work with Diabetes and You to the Child Care Center also demonstrates the value of pilot testing and learning design lessons from earlier work. More positive ratings of the doctors, graphics, and level of engagement with the Child Care Center website are the direct result of addressing issues and concerns that arose in the evaluation of Diabetes and You. One of the chief complaints of respondents in pilot testing of the Diabetes and You website was that the doctors’ mouths did not match their speech particularly well; improved ratings of the doctors in the Child Care Center would seem to suggest that efforts at improving this design concern were successful.
The test of the hypothesis that subjects would prefer to receive information from providers of matching ethnicity is also informative. One of the most frequently discussed benefits of new digital media is the ability to customize content to match users’ preferences, including ethnic and cultural backgrounds (Kreuter & McClure, 2004; Oenema, Brug, & Lechner, 2001). The failure to confirm that hypothesis should not be construed as contradicting such claims. Instead, it points to the importance of truly customizing content to match users’ ethnic and cultural preferences. Merely changing the ethnicity of the information provider – leaving all other content the same – will not be enough “customization” to successfully engage users with information designed specifically to meet their needs. These findings must be tempered by the small number of minority participants; continued work in this area should explicitly focus on recruiting a more culturally and ethnically diverse sample.
This research points to a variety of useful directions for further investigation. First, the effect of true cultural customization must be explored. Superficial customization (e.g., simply presenting users with information providers of varied ethnicities) seems to not have a significant impact on users’ evaluations of the intervention. Effective tailoring of content, based on user-provided information, could help realize the potential advantage of new digital media to customize content to meet user needs. Alternative models of delivery, such as mobile devices or kiosks in medical facilities, also merit attention. As stated earlier, users must have access to the delivery channel for health information to be effective in promoting positive behavior change. While low literate and low health literate audiences represent a hard-to-reach audience, a delivery model based in the waiting rooms of medical facilities could provide a useful strategy for (1) reaching these audiences with important health information and (2) making it possible to test such interventions on a large scale with a group that is traditionally hard to access.
This work also demonstrates the need to better understand the connection between health literacy and the HBM. Low health literacy could, for example, make it less likely that individuals will receive health information from an educational intervention and adopt appropriate preventive health behaviors. Continued work with interventions for low health literate audiences should follow participants over time, to track changes in actual behavior and investigate how health literacy fits within the HBM. While one recent study related to diabetes did not demonstrate a link between health literacy and the HBM (Powell, Hill, & Clancy, 2007), this remains an issue that merits further investigation. A better understanding of the role of health literacy and its influence on the HBM will help health communication researchers and professionals design improved interventions that can more successfully promote behavior change in low health literate populations. The Diabetes and You and Child Care Center websites feature content aligned with core HBM constructs, and continued work with these interventions will allow for explicit tests of how these websites can promote behavior change in low health literate audiences.
As science and medicine become increasingly complicated, effectively communicating health information to low health literate audiences will become more important – the costs associated with low health literacy make this clear. While research into the most effective ways to design interventions for these groups will continue to move forward, it is important to recognize that more health literate audiences also appreciate the simplicity that this type of health communication represents. Straightforward, effective health communication to all audiences should be the goal for health communication researchers and professionals, and this research demonstrates the potential for interventions designed for low health literate individuals to meet the needs of a wider audience.
This project was supported by funding from The University of Texas at Austin Office of the Vice President for Research and the Michigan Department of Community Health.
About the Authors
Michael Mackert is an Assistant Professor in the Department of Advertising at The University of Texas at Austin. His research focuses on health literacy, with a particular interest in the best ways to design health messages to reach low health literate populations using both new digital and traditional media.Address: 1 University Station A1200, Austin, TX 78712, USA
Pamela Whitten is an Assistant Dean in the College of Communication Arts and Sciences at Michigan State University. Her research focuses on the use of technology in health care with a specific interest in telehealth and its impact on the delivery of health care services and education.Address: 409 Communication Arts and Sciences Building, East Lansing, MI 48824, USA
Adriana Garcia is a doctoral student in the Department of Advertising at The University of Texas at Austin. She is primarily interested in the study of non-profit organizations and the way that organizations communicate with potential Latino donors.Address: 1 University Station A1200, Austin, TX 78712, USA