The survey was created by an internal team with experts in the areas of clinical informatics, healthcare quality, and survey methodology. Many survey questions were adapted from previously validated questionnaires.8,9,13–15
The survey was pilot tested for clarity and content with 25 adult outpatients, and was approved by the Institutional Review Board of Weill Cornell Medical College. Survey questions are included in the online appendix
Study Population and Survey Administration
The Hudson Valley is an eight county region in southern New York State (NYS) bordering the Hudson River that contains more than 10% of the total NYS population.16
Many of the healthcare centers in the Hudson Valley are affiliated with the Taconic Health Information Network and Community (THINC) including several hospitals, a clinical laboratory, and the Taconic Independent Practice Association (IPA), comprised of over 4,000 physicians at 1,145 practices. THINC has implemented electronic health records (EHRs) in ambulatory physicians’ offices and established HIE among many practices, laboratories and hospitals.The telephone survey was administered by the Cornell Survey Research Institute (SRI) in Ithaca, NY from January–April 2008 using the Computer-Assisted Survey Execution System.17
Survey respondents were identified using a random digit dial sample of fixed-line telephone numbers within the residential zip codes of the eight Hudson Valley counties. Trained interviewers contacted potential respondents, obtained verbal consent using a standard IRB-approved script, confirmed eligibility, and offered a $10 incentive to eligible respondents. Eligible participants had to be at least 18 years old, English speaking and residing in the Hudson Valley. If there was no answer at a phone number, the number was called a maximum of five times unsuccessfully before retiring the number.
Dependent Variables Before assessing participants’ attitudes towards physician use of HIE, EHRs and HIE were described, and examples of HIE use in an emergency room and an outpatient setting were given. Participants were asked how they feel about their medical records being shared electronically between their medical providers using a five-point likert scale (strongly support, somewhat support, no opinion, somewhat against, strongly against). Those who responded they strongly or somewhat support the idea were categorized as supporting physician use of HIE.Personal HIE was described as “us[ing] the computer to find information about your health or view[ing] your own electronic medical record.” Participants were then asked to respond to the following statement: “I am interested in using the computer and the Internet to manage my health care” on a five-point likert scale (strongly agree, agree, neutral, disagree and strongly disagree). The question was designed to be broad in order not to restrict to one type of personal HIE architecture (i.e. personal health record or patient portal). Those who responded they strongly agree or agree with the statement were categorized as interested in personal use of HIE. Participants were later asked about whether they currently do or would like to do specific healthcare-related tasks (i.e. viewing their medical record, e-mailing doctors, and requesting appointments) online.
We adapted the Technology Adoption Model18
to identify factors potentially associated with participants’ attitudes towards HIE (Fig. ): demographic factors, experience with computers, perceived benefits and risks of HIE use, and the perceived need for HIE. The demographic characteristics assessed in the survey included age, gender, race, ethnicity and income. To ascertain participants’ experience with computers, we asked if any of their doctors used computerized medical records and the frequency of their Internet use (several times daily, daily, weekly, monthly, rarely/never). For analysis, we dichotomized the categorical variables to compare the oldest (≥ 65 years), wealthiest (income >$100,000/year), least frequent Internet users (rarely or never) and non-minority (white, non-Latino) participants to their counterparts.
To determine participants’ perceptions about the risks and benefits of HIE use, we asked participants to indicate on a 5-point scale how they thought physician and personal use of HIE would affect the following items: the security and privacy, completeness and accuracy of their medical record, and communication among their physicians and between their physicians and themselves. For analysis, those who reported HIE would improve (greatly improve, improve) an item were compared to those who reported HIE would not improve (no change, worsen, greatly worsen) an item. To determine concerns about Internet security, we asked participants to rate the security (“safe from unwanted viewing”) of personal information on regular and secure Internet connections. Participants were grouped into those who believed 1) both types of connections were secure, 2) only secure connections were secure, or 3) both types of connections were insecure.Perceived need for HIE was measured by assessing participants’ health, healthcare usage, role as a caregiver, and perceived efficacy of current healthcare communication. Questions to assess health included participants’ self-rated overall health status15
and if they were current being treated for a chronic medical condition. We grouped respondents into three categories based upon their response to how often they had visited different healthcare providers and facilities during the past year: those who had no healthcare visits, those who had visit(s) only
with their primary care physician, and those who had visit(s) with other providers or at other facilities (i.e specialists, ER visit or hospitalization). We also asked participants if they were making healthcare decisions for someone with a debilitating or chronic illness. Participants were asked to rate the efficacy of physician communication across different practice locations based on how often their physicians knew about what happened or had records from their visits at other locations. For analysis, we compared those who thought communication was good (very well, well), those who thought the communication was suboptimal (it varies, poor) and those who reported they had only been to one practice or location.
Modified technology acceptance model.
We used descriptive statistics to summarize respondents’ characteristics. In order to compare the study population with the Hudson Valley population, we calculated weighted averages of age, race, ethnicity and median income from the census data of eight Hudson Valley counties.19
We estimated bivariate associations between each of the two dependent variables, support of physician HIE use and interest in personal HIE use, and the independent variables using odds ratios. Multivariate logistic regression models were used to determine independent associations with each dependent variable. All data were analyzed using SAS® software, version 9.2.1