Predictors of eHealth UsageInsights on The Digital Divide From the Health Information National Trends Survey 2012

For years, the dominant story about health technology and race was one of deep digital division, with minority groups locked out and the gap widening with every new app and patient portal. Then, the 2012 Health Information National Trends Survey data arrived, and Kontos and colleagues quietly overturned that assumption. Race wasn't the divide; something else was, and it had been hiding in plain sight. The setup matters. EHealth, which includes online doctor communication, health tracking tools, patient portals, and mobile health downloads, arrived with a genuine promise: that better information, more conveniently delivered, would help the people who need health care most. Kontos and colleagues frame this within what's called the Chronic Care Model, which imagines an "informed and activated patient" working alongside a "prepared and proactive practice team." The digital toolset was supposed to make that vision real. However, the chronic conditions it was meant to help manage — cancer, obesity, diabetes — remain stubbornly unequal across social groups. This raises the blunt question: does eHealth actually reach the people who need it most? To find out, Kontos and colleagues turned to the National Cancer Institute's Health Information National Trends Survey, known as HINTS. The 2012 dataset covered three thousand nine hundred fifty-nine U.S. adults, nationally representative, drawn from a mailed survey collected between October 2011 and February 2012. The key analytic move was to restrict the sample to the two thousand three hundred fifty-eight respondents who reported ever going online. The question wasn't who could access the internet; it was what people did once they got there. Eleven eHealth behaviors were grouped into three domains: health care tasks, such as emailing a provider or looking up a doctor; health information-seeking, like searching for medical information or using the internet to manage diet and weight; and user-generated content, like participating in online health support groups or posting on health social networks. Kontos and colleagues used weighted multivariable logistic regression to estimate how much socioeconomic status, race and ethnicity, age, and sex each predicted use within these domains, while controlling for insurance status, health status, regular provider access, and several other factors. The prevalence numbers set the scene. Searching for health information online was nearly universal among internet users — about seventy-nine percent had done it for themselves, and fifty-seven percent for someone else. However, the more interactive, care-oriented tasks were far less common. Only about nineteen percent had emailed a provider, nineteen percent had tracked personal health information online, and roughly thirty-eight percent had searched for a health care provider. User-generated content was the rarest: somewhere between three and five percent participated in health blogs or support groups. The question was who, within those numbers, was doing what. Here’s the finding that reshapes the whole conversation. Among online adults, race and ethnicity did not predict eHealth use in any consistent way. Kontos and colleagues are explicit: "among online adults, we saw no evidence of a digital use divide by race and ethnicity." That qualifier is everything. Prior research, and much of the conventional narrative around digital health equity, had focused on race. Earlier studies found Black and Hispanic adults significantly less likely to seek health information online. However, those gaps, the authors note, had begun to narrow. In this dataset, once you're looking only at people who are already online, race drops out of the picture as a reliable predictor across domains. What takes its place is socioeconomic status, and particularly education. The education findings are striking in their consistency. Adults with a high school degree or less had about half the odds of several core eHealth behaviors compared to college graduates. Their odds of using email or the internet to communicate with a doctor were 0.46, which is roughly fifty-four percent lower. Their odds of looking up a health care provider online were 0.50. Their odds of tracking personal health information online were 0.53. For using a website to help manage diet, weight, or physical activity, the odds ratio was 0.64 for those with a high school education or less, and 0.67 for those with some college, both significantly below college graduates. Even having some college, not just stopping at high school, was enough to meaningfully reduce the odds of downloading health information to a mobile device, where the odds ratio was 0.54. These aren't marginal gaps; they represent substantial, consistent disadvantage across exactly the tasks that connect patients to ongoing care. Income mattered too, but less reliably. Lower-income households showed reduced odds of buying medicine or vitamins online — households earning under twenty thousand dollars had an odds ratio of 0.34, and those earning between twenty and thirty-five thousand had 0.38. However, income effects didn't show up uniformly across the other eHealth behaviors. Education was the more stable signal. Kontos and colleagues point to eHealth literacy as the likely mechanism. EHealth literacy is defined as "the ability to seek, find, understand, and appraise health information from electronic sources and apply knowledge gained to addressing or solving a health problem." It's not the same as general literacy, but it closely tracks with education. The authors treat it as a probable mediator: lower educational attainment leads to lower eHealth literacy, which leads to less engagement with the online tools that could actually improve health management. They call explicitly for future research to test that pathway directly. Age adds another layer. Younger adults used nearly every eHealth domain more than older ones, and the differences are large. Adults aged eighteen to thirty-four were three point five one times as likely as those sixty-five and older to search the internet for health information for themselves. They were three point three seven times as likely to use a website to help with diet or exercise. Even just being in the thirty-five to forty-nine bracket conferred more than twice the odds of searching for health information compared to the oldest group. For finding a health care provider online, the eighteen to thirty-four group had odds two point two four times higher than adults sixty-five and older. The age gradient runs across virtually every outcome. Sex was also a consistent predictor, and consistently in one direction. Women had higher odds of looking for a health care provider online, with an odds ratio of 1.53, higher odds of searching for health information for themselves, at around 1.46, and higher odds of tracking personal health information online, at 1.52. Across health care and information-seeking tasks, being female was a reliable positive predictor of eHealth engagement. The social media domain showed a different texture. Overall engagement was low, with only a few percent of online adults participating. However, Kontos and colleagues note that prior Pew data cited in the paper showed higher mobile health access among Black adults at thirty-five percent, and Hispanic adults at thirty-eight percent, compared to twenty-seven percent for white adults. This suggests that for mobile and social platforms specifically, minority groups may be more engaged than the aggregate numbers imply. One subgroup in the HINTS analysis, labeled "other race," showed increased odds of downloading health information to mobile devices and searching for a health care provider online. The picture in user-generated content is more complicated than in the other two domains, and the authors treat it accordingly. What does this add up to? The digital divide, for people who are already online, is primarily an education divide. It is reinforced by age and shaped by sex. This divide runs through the exact tasks — talking to doctors electronically, tracking health data, managing chronic conditions through digital tools — where the stakes are highest for people who already carry a disproportionate burden of disease. Kontos and colleagues are careful about what the data can and can't say. The HINTS survey is cross-sectional, so causality can't be established. The thirty-six point seven percent response rate introduces potential sampling error. Furthermore, the technology landscape of 2011 and 2012 has shifted considerably. But the structural finding is durable: designing eHealth tools without explicit attention to education-linked literacy skills means building for the people who need them least. The paper recommends matching technology design to users' eHealth literacy, supplementing digital content with non-digital materials where needed, and using user-generated and mobile platforms strategically for populations more likely to engage there. The promise of eHealth was democratization. What Kontos and colleagues found is that achieving it requires more than just putting tools online. It requires designing those tools for the people who arrive at them with the fewest advantages, and recognizing that education, not race, is the fault line that most reliably predicts who gets left behind. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.