Reevaluating the Digital Divide: Current Lack of Internet Use Is Not a Barrier to Adoption of Novel Health Information Technology

We conducted a mail survey of 4,024 patients with type 2 diabetes in eastern Massachusetts identified via our primary care network's administrative database. After excluding 102 ineligible subjects, our response rate was 29% (1,146 of 3,922). Excluding subjects who left blank the question about current Internet use, 952 responses remained for analysis (24% of eligible cohort). Compared with nonrespondents, survey respondents were somewhat older (65.8 vs. 64.2 years, P = 0.002), more often male (58 vs. 49%, P < 0.001), more often white (86 vs. 78%, P < 0.001), and lived in neighborhoods with slightly higher median family incomes ($49,746 vs. $44,101 based on annual tax returns by zip code, P < 0.001). However, the proportion of individuals insured by Medicare (50 vs. 53%) and the last measured A1C level (7.3 vs. 7.4%) were similar between groups.

The survey was developed in collaboration with diabetes and information technology specialists. New questions developed to assess current technology use, receptivity to using wireless technologies to share health information with a provider remotely, and specific barriers to technology adoption were tested in patient and provider focus groups but not formally validated. The survey also included one question from the 12-item short-form health survey (SF-12) (8) and self-reported medication adherence (Morisky scale) (9).

In this hypothesis-generating exploratory analysis, we examined the relationship between current Internet use (frequent and infrequent users vs. nonusers) and attitudes toward future diabetes-related technology use. We constructed logistic regression models to test the hypothesis that the association between current Internet use and willingness to adopt new technologies would be attenuated when adjusting for potentially ameliorable self-reported barriers (SAS version 9.1; SAS Institute, Cary, NC). The number of subjects included in each model varied from 855 to 936 due to missing covariate data. Analyses limited to patients with complete data yielded similar results.

The majority of respondents were white (84.2%), male (58.1%), and had some college education (56.8%). The median duration of diabetes was 8 years, 56.6% individuals reported no problems with medication adherence, and most respondents (84.2%) self-reported their health status as average or above. Compared with nonusers, Internet users were younger (60.9 vs. 70.2 years, P < 0.001), more had attended college (78.7 vs. 39.7%, P < 0.001), and more knew their most recent A1C result (66.0 vs. 39.7%, P < 0.001).

Internet users and nonusers differed greatly in their level of confidence with technology (85.5 vs. 48.2%, P < 0.001), privacy concerns (41.1 vs. 60.9%, P < 0.001), and cost concerns (49.3 vs. 58.9%, P = 0.01). A wide gap exists between Internet users and nonusers in the use of non–health care technology such as cell phones (89.1 vs. 37.8%, P < 0.001) and digital cameras (55.8 vs. 5.9%, P < 0.001). This gap, however, was significantly reduced for health care technologies, with similar proportions using glucometers (85.5 vs. 81.2%, P = 0.08) and home blood pressure monitors (42.2 vs. 37.1%, P = 0.12).

Table 1 illustrates subjects' receptivity to using new devices or services if offered in the future. Although differences between Internet users and nonusers were statistically significant, absolute levels of interest were high in both groups. Among current Internet nonusers, over 80% would use a wireless glucometer, over 70% a wireless blood pressure cuff, and over two-thirds wireless activity monitors, scales, or pillboxes to automatically transfer health data to their primary care physician. Similarly, technology to permit personalized goals and feedback was of interest to nearly all respondents. In contrast, the idea of conducting an online visit in place of an office visit was unpopular in both groups.

In multivariate models, the association between Internet use and willingness to adopt new HITs was eliminated or greatly attenuated after adjusting for factors that are potentially ameliorable with patient training (e.g., concerns about privacy and cost or confidence using new technology). This change emphasizes that the digital divide seen in Internet use is not itself a barrier to the adoption of novel diabetes-related HITs.

This study examined views toward clinical information-sharing using new devices and services and barriers to increasing technology implementation in over 900 primary care patients with type 2 diabetes. Among respondents, Internet users were nearly 10 years younger, were twice as likely to have a college education, were more likely to know their A1C level, and had fewer reservations about new technology than nonusers, confirming previous reports in the literature that Internet use represents a key dividing line among patients with diabetes (10).

However, results of our survey indicate that the digital divide may not apply to advanced HIT tools being developed for diabetes care. Both users and nonusers were eager to benefit from advances in technology to facilitate information-sharing with their physicians. This reinforces the point that if new technology can be made both relevant and valuable, patients will use it. Moreover, the widespread acceptance of wireless communication technologies in our survey population indicates that future care models incorporating extensive home monitoring and tailored feedback have potential for broad adoption by patients with type 2 diabetes. The challenge remains to design these new systems to fit with current clinical workflow and reimbursement mechanisms.

Our study was limited by a relatively low response rate that, while on par with similar mail-based surveys, raises the possibility of response bias. Although demographic differences between responders and nonresponders were small, the possibility remains that responders may have been more enthusiastic about technology use. Another limitation was that the survey asked about potential interest in adopting new technologies. This is only a proxy measure of adoption and may overestimate actual uptake if these services were offered. Several recent studies were able to enroll only 6–23% of potential subjects into Internet-based diabetes interventions (11–13).

In conclusion, our results support a more complex view of the role technology can play in improving diabetes care. Internet use is at best a partial indicator of a patient's willingness to adopt technology to improve their health. The finding that patients who currently never use the Internet are enthusiastic about adopting future HIT tools for diabetes care underscores the insight that we should not underestimate the potential audience for diabetes technology innovations.

This project was funded by the Center for Connected Health, Boston, Massachusetts. Dr. Grant is supported by an National Institute of Diabetes and Digestive and Kidney Diseases Career Development Award (K23 DK067452).

Preliminary findings were presented in abstract form at the American Diabetes Association 67th Scientific Session, Chicago, 22–26 June 2007.