Use of the Internet to Communicate with Health Care Providers in the United States: Estimates from the 2003 and 2005 Health Information National Trends Surveys (HINTS)

Background: Despite substantial evidence that the public wants access to Internet-based communication with health care providers, online patient-provider communication remains relatively uncommon, and few studies have examined sociodemographic and health-related factors associated with the use of online communication with health care providers at a population level. Objective: The aim of the study was to use nationally representative data to report on the prevalence of and changes in use of online patient-provider communication in 2003 and 2005 and to describe sociodemographic and health-related factors associated with its use. Methods: Data for this study are from two iterations of the Health Information National Trends Survey (HINTS 2003, HINTS 2005). In both years, respondents were asked whether they had ever used email or the Internet to communicate with a doctor or a doctor’s office. Adult Internet users in 2003 (n = 3982) and 2005 (n = 3244) were included in the present study. Multivariate logistic regression analysis was conducted to identify predictors for electronic communication with health care providers. Results: In 2003, 7% of Internet users had communicated online with an health care provider; this prevalence significantly increased to 10% in 2005. In multivariate analyses, Internet users with more years of education, who lived in a metro area, who reported poorer health status or who had a personal history of cancer were more likely to have used online patient-provider communication. Conclusions: Despite wide diffusion of the Internet, online patient-provider communication remains uncommon but is slowly increasing. Policy-level changes are needed to maximize the availability and effectiveness of online patient-provider communication for health care consumers and health care providers. Internet access remains a significant barrier to online patient-provider communication. (J Med Internet Res 2007;9(3):e20) doi: 10.2196/jmir.9.3.e20 J Med Internet Res 2007 | vol. 9 | iss. 3 | e20 | p. 1 http://www.jmir.org/2007/3/e20/ (page number not for citation purposes) Beckjord et al JOURNAL OF MEDICAL INTERNET RESEARCH


Introduction
For more than a decade, studies have consistently shown that some members of the public want access to Internet-based communication with health care providers, with preference estimates for online patient-provider communication ranging from 40% to 83% [1][2][3][4][5][6][7][8][9]. The Institute of Medicine has characterized patient-centered care as health care that addresses patient needs and preferences, and it has identified information technology as crucial to advancing health care quality [10]. There is evidence that online communication with health care providers promotes health-related quality of life (eg, [11]) and, further, that health care consumers would benefit from increased partnerships between health information technology and health care providers: health care providers are a more trusted source of health information than the Internet, but the Internet is the source most often used by the public to retrieve health information [12]. Thus, as the presence of eHealth applications In 2003, the National Cancer Institute launched the first biennial Health Information National Trends Survey (HINTS) [22,23]. HINTS was designed to capture nationally representative data on the impact of the cancer information environment and to specifically assess the public's use of information technology for health. Though previous estimates of the online public's use of the Internet to communicate with health care providers have been reported from nationally representative data [15,17,24], the purpose and design of HINTS allow for a more in-depth examination of sociodemographic and health-related factors in relation to online communication with health care providers than has been possible in previous investigations. Further, using cross-sectional iterations of HINTS from 2003 and 2005, we can examine whether the prevalence of online communication with health care providers among American Internet users has changed over time.
In 2001, an estimated 6% of Internet users had emailed health care providers [15], up to 7% in 2003 [17]. In this paper, we provide estimates of Internet-based patient-provider communication derived from national probability samples of online American adults in 2003 and 2005. In addition, we report on changes in use of the Internet to communicate with health care providers over this 2-year period and identify sociodemographic and health-related factors associated with online patient-provider communication. Overall, we expect that the prevalence of online patient-provider communication will be low, but that use of the Internet to communicate with health care providers will significantly increase between 2003 and 2005. Further, based on results of previous studies [2,25], we expect that Internet users with more years of education, who are of higher socioeconomic status, and who are more engaged with the health care system will be more likely to report having communicated online with an health care provider.

Data Source
Data for this study are from HINTS 2003 and HINTS 2005, two iterations of the nationally representative survey designed to assess the impact of the cancer information environment and the public's knowledge of, attitudes toward, and behaviors related to cancer and cancer prevention [26]. Comprehensive reports on the conceptual framework of HINTS and sample designs are published elsewhere [22,23]. The cross-sectional surveys were administered by trained interviewers to representative samples of American households drawn from all telephone exchanges in the United States. Exchanges with high numbers of African Americans and Hispanics were oversampled in 2003. One adult (age 18 or older) was selected from each household to participate in the full survey during a household screening. In 2003, response rates were 55% at the household screening level (ie, the initial contact with the household used for sampling purposes) and 63% at the sampled person interview level (ie, completion of the interview by the sampled household member); in 2005, the respective response rates were 34% and 61%. Every sampled adult who completed a questionnaire in HINTS 2003 and HINTS 2005 was assigned a final sampling weight and a set of 50 replicate sampling weights. These sampling weights were used for the purpose of computing nationally representative estimates, to adjust for nonresponse and to reduce the sampling variance of estimators through utilization of information with less sampling and nonsampling error than the corresponding HINTS estimates (eg, estimates obtained through the Current Population Survey, which has much larger sample sizes than HINTS).

Study Variables
As in prior investigations of Internet users' use of online patient-provider communication (eg, [15]), sociodemographic variables included in the present study were age, gender, education, annual income, race/ethnicity, and metropolitan statistical area (metro or non-metro county). Previous studies have also suggested that health-related variables, such as poorer self-reported health status [2] and having health insurance [25], are associated with use of online patient-provider communication. Health-related variables included in the present study were self-reported health status, possession of health insurance, and personal cancer history.

Data Analysis
Analyses were conducted using SAS-callable SUDAAN, version 9.0 [27] to account for the complex survey design of HINTS and to obtain appropriate standard errors and 95% confidence intervals (CIs) for point estimates. Responses of "refused" or "don't know" were counted as missing. Unweighted and weighted descriptive statistics are presented, and weighted data were used in all inferential statistical analyses. Bivariate analyses  Bivariate associations between study variables and emailing health care providers are displayed in Table 2. In 2003, respondents who had communicated online with an health care provider had significantly more years of education and were more likely to reside in a metro county. In 2005, they were more likely to be female, had significantly more years of education, and were more likely to have a personal history of cancer.   Internet users who were college graduates had over three times the odds of communicating online with an health care provider compared to those with less than a high school education (OR = 3.73; 95% CI = 1.10-12.59; P = .03). Those who lived a non-metro area were less likely to have used online patient-provider communication compared to Internet users who resided in metro area counties (OR = 0.62; 95% CI = 0.41-0.95; P = .03). Finally, Internet users who reported "fair" or "poor" health status had higher odds of communicating online with an health care provider (OR = 1.46; 95% CI = 1.00-2.04; P = .05).

Sample Characteristics and Bivariate Analyses
For Internet users in 2005, women were more likely to have communicated online with an health care provider compared to men (OR = 1.47; 95% CI = 1.00-2.15; P = .05), and cancer survivors were more likely to have used online patient-provider communication compared to those without a history of cancer (OR = 1.99; 95% CI = 1.27-3.12; P = .002). These results are consistent with bivariate analyses; however, education was not associated with online patient-provider communication in the multivariate model.

Discussion
Despite over a decade of research and the availability of guidelines for use of Internet-based communication by health care providers [28], the number of health care consumers using online patient-provider communication is still far below estimates of the number who would prefer to do so. Though data from HINTS suggest that use is slowing increasing, diffusion of online patient-provider communication is occurring at a pace far slower than diffusion of Internet use in general [29].
users with more medical problems or who are more engaged with the health care system due to a significant medical history may be more "hooked in" to Internet-based health communication resources or may have more a frequent need to use them. Finally, in 2005, women were more likely to use online patient provider communication compared to men. This result is consistent with findings that online women are more likely to search specifically for health information compared to men [41] and that higher percentages of women use the Internet for interpersonal communication related to health (eg, use of online support groups or health-based chat rooms [42]).
We did not observe associations between online communication with health care providers and characteristics such as race/ethnicity or annual income that have been documented in other studies (eg, [2,25]) as evidence of a "digital divide" [43,44]. Nonetheless, research and policy should continue to address groups potentially affected by the digital divide to ensure that advances in health information technology benefit all health care consumers [45]. Finally, our results were not consistent with previous studies that observed younger Internet users to be more likely to engage in online communication with health care providers (eg, [2,6]), suggesting a potential growth in comfort with online communication among Internet users of all ages.

Limitations
Though HINTS data are nationally representative, the generalizability of our results may be limited by survey response rates and the drop in response rates between 2003 and 2005. However, HINTS response rates are comparable to those of other national random digit dial surveys [46], and the agreement between our findings regarding prevalence of online patient-provider communication with other reports [17] supports the reliability of the HINTS estimates. Further, estimates of Internet penetration vary widely in the published literature; HINTS penetrance estimates may be more conservative than data reported through market analysis firms due to the degree of sampling precision mandated for federal surveys that provide publicly available data. Due to item wording, we can only discuss our results at a generalized level of "online patient-provider communication" or "Internet-based communication with health care providers" and cannot characterize this behavior in more specific ways (eg, use of personal email, use of a Web portal) that could potentially affect our findings and resulting conclusions [38]. Finally, though HINTS provides a valuable population-level perspective on the prevalence of Internet-based health care provider communication and information on the characteristics of those who use it, all data are based on self-report, and HINTS does not allow for more in-depth examinations of barriers to communicating online with health care providers or the perceived benefits for those who do. To best meet the needs of patients and health care providers, research should continue to assess health care consumers' and health care providers' perspectives on barriers and benefits related to use of Internet-based communication as health information technology increasingly becomes part of standard medical care.