Consumer Perspectives on Personal Health Records: A 4-Community Study
Published Online: April 17, 2014
Erika L. Abramson, MD, MS; Vaishali Patel, PhD, MPH; Alison Edwards, MStat; and Rainu Kaushal, MD, MPH
Unprecedented federal initiatives are under way to transform healthcare delivery by promoting health information technology (HIT).1 One key focus of these initiatives is to deliver more patient-centered care. Personal health records (PHRs), which are HIT applications that allow consumers to electronically “access, manage, and share their health information… in a private, secure, and confidential environment” have potential to play a key role in these efforts.2
There are several PHR models. Tethered PHRs are integrated with an electronic health record (EHR) and receive automatically prepopulated information. Untethered PHRs are not linked and generally require users to enter and maintain data. Importantly, there are no specified standards for the content, functionalities, and architecture that should comprise a PHR, resulting in widespread variability.3
Multiple potential benefits exist with PHR use, although literature demonstrating actual benefits is limited.3 Theoretically, PHRs can allow consumers to become more actively engaged in managing their health and can improve patient and provider communication. Certain PHR features such as online appointment scheduling also should improve efficiency by allowing administrative issues to be handled electronically.4
Despite these benefits, PHR usage remains low, even though many nonusers endorse interest in using a PHR.5 Importantly, Stage 2 requirements for the EHR Incentive Program will require that providers give more than 50% of their patients online access to their health information.6 Furthermore, they require providers to demonstrate that 5% of their patients have accessed their health information online and that 5% have engaged in secure messaging. This will likely spur PHR adoption.
Addressing the gap between reported interest and actual usage requires more detailed information on consumer preferences for PHRs and barriers. In this way, PHRs can be better designed and marketed toward consumers broadly. Evidence already suggests that those with less access to and experience using the Internet, as well as people from certain minority groups and of lower socioeconomic status, may be less likely to use PHRs.7-10 Finally, a detailed understanding of consumer preferences can help lead to more uniform standards for PHR products.
We therefore conducted this study to better understand consumer preferences for PHRs and to identify factors associated with potential usage. We used pooled data previously collected and published from individual communities within New York state (NYS) with HIT initiatives already under way to allow us to generate a more broadly representative assessment of consumer preferences and predictors of potential usage.7,9,11,12
We conducted cross-sectional surveys of consumers in 4 NYS communities to assess attitudes toward PHRs. This study was led by researchers from the Health Information Technology Evaluation Collaborative (HITEC), which comprises 4 research institutions across NYS (Weill Cornell Medical College, Columbia University, State University of New York at Albany, and University of Rochester) and is designated to evaluate NYS HIT and health information exchange (HIE) initiatives. The 4 communities, located in Buffalo, Rochester, the Hudson Valley, and Brooklyn, were all recipients of NYS grant funding to promote community-wide HIT adoption and HIE. They are geographically spread across NYS; include rural, urban and suburban areas; and are home to consumers who are socioeconomically, ethnically, and racially diverse.
The novel survey instrument was developed based upon literature review and consultation with experts in clinical informatics, healthcare quality, and survey methodology from Weill Cornell Medical College, Columbia University Medical Center, and the Center for Medical Consumers, a not-for-profit consumer advocacy organization. Wherever possible, survey items were drawn from previously validated questions from other national surveys.13-16 The survey was piloted with 25 outpatients at an adult internal medicine clinic affiliated with an academic, tertiary hospital.
Survey Domains and Measurement
The survey domains included demographic characteristics, health characteristics, healthcare experience, Internet access and use, attitudes toward HIE and EHRs, and attitudes toward and preferences related to PHRs. In the survey, a PHR was described as a tool “primarily used by you to view your health information and manage your healthcare (for example, to make appointments) on the Internet.” We asked consumers about PHR preferences for content and functionalities, and questions regarding health characteristics such as health status, use of medications, treatment for chronic conditions, and caring for someone with a chronic condition. “Healthcare experience” included questions regarding satisfaction with care, interest in collaborative decision making, and problems understanding medical information. We also asked consumers about their access to and use of the Internet, concerns related to privacy and security, and attitudes toward HIE and EHRs.
Most questions were either yes/no, on a 5-point Likert-like scale, or 3-point questions (ie, improve/no effect/ worsen). The initial survey was administered in the Hudson Valley, and tailored slightly for each subsequent community. For example, in Brooklyn, “Russian” was explicitly listed for the question “What is the main language you speak at home?” whereas in the other communities, only “English, Spanish, Chinese, or Other” were answer choices. However, most questions were consistent across communities to allow for cross-community comparisons.
Survey Population and Administration
The mode of survey administration was not uniform. In the Hudson Valley, Cornell Survey Research Institute (CSRI) administered the survey via telephone from January through April 2008, using the Computer-Assisted Survey Execution System, which is a comprehensive computer-assisted interviewing system that can be used in all stages of data collection.11 Survey respondents were identified using a random digit dial sample of fixed-line telephone numbers within the residential zip codes of the 8 Hudson Valley counties. Potential respondents were contacted by trained interviewers, consented using a standardized script, and offered a $10 incentive. A single phone number was called a maximum of 5 times. CSRI similarly administered the survey in Buffalo in February 2009, although no incentives were offered. Eligible adults were English-speaking residents of the 8 counties comprising the greater Buffalo region of NYS, which is served by the HEALTHeLINK regional health information organization (RHIO).
In Brooklyn, from October to November 2008, trained volunteers who were bilingual in Russian, Spanish, or Mandarin distributed the survey to consecutive adults who entered 3 ambulatory care sites and an emergency department participating in the community RHIO’s HIE pilot program. In Rochester, the survey was administered during August 2008 by research staff to consecutive adults presenting for a visit to 1 of 5 primary care practices involved in the RHIO’s HIE pilot program, which at the time included 4700 patients. For all sites, institutional review board approval was obtained.
In order to describe consumers’ potential use of PHRs, we examined responses to a survey question about the frequency of potential PHR use to view health information or manage healthcare. Except for 1 community, responses to this question were on a 5-point scale ranging from “rarely or not at all” to “about once daily.” The last community used a 5-point scale that ranged from “rarely or not at all” to “once a week.” To keep responses consistent across communities, we dichotomized respondents who indicated they would use the PHR “rarely or not at all” as nonusers while the remaining respondents were categorized as potential users. We also recategorized many of the independent variables, by either dichotomizing or reducing the number of categories, to be consistent across communities: language spoken at home, education, employment, children in the household, satisfaction with quality of care, doctor uses computer, frequency of Internet use, and each perceived benefit. Responses to age, income, health status, and understanding of health information were collapsed into 3 categories (or 4 for income) before merging community data. We also combined race and ethnicity into 1 binary variable (white and non-Hispanic vs nonwhite or Hispanic).
We used descriptive statistics to characterize respondents’ demographic and health characteristics, healthrelated experiences, Internet access and frequency of use, and attitudes toward PHRs. We examined bivariate associations between these variables and potential use of PHRs using χ² tests or Fisher’s exact tests for categorical variables. We selected those significantly associated with our outcome (P <.05) for inclusion into the multivariable model. To reduce the number of related factors included as independent variables, we considered pairwise percent agreement among the 6 questions inquiring about perceived PHR benefits.
PDF is available on the last page.