The 14 102 participants were ethnically diverse with varying educational attainment (). During the 12-month study period, 5671 (40%) requested a password via the internet for the patient portal. Among these, whom we assume had sufficient access to the internet by virtue of request for a password for the patient portal, we observed a decrement in each step of the use process, 1360 (25%) did not activate their accounts using the mailed, default password, and an additional 389 (7%) never logged on to the patient portal using their own passwords (). Three thousand nine hundred and twenty-two logged on. Although each user could use multiple functions, among those who requested a password, 2990 participants (53%) viewed their laboratory results, 2132 (38%) requested prescription medication refills, 2093 (37%) sent email messages to their physicians/care team, and 835 (15%) made medical appointments using the internet-based patient portal.
Patient characteristics (N=14 102)
Proportion of users who requested a password who used each function one or more times during the study period (N=5671).
There were marked race/ethnic differences in use (), with African–American (31%), Latino (34%), and Filipino (32%) participants least likely, and Asian (53%) and White (51%) participants most likely to both request a password for the internet-based patient portal (a marker for internet access and intent to use) and log on to the portal after requesting a password was completed (an early marker for navigability once access is obtained).
Figure 2 (A) Proportion who requested a password and logged on to the internet-based patient portal, by race/ethnicity (for difference across ethnicities, p<0.01) and immigration (for difference by immigration status p=0.51) status (N=14 102). (more ...)
For example, 2006 (51%) of Caucasians compared to 887 (31%) of African–Americans requested a password for the patient portal. Of those who requested a password, 395 (21%) of Latinos logged on to the patient portal compared to 1496 (38%) of Caucasians. Moreover, we observed a consistent gradient with respect to educational attainment, such that those with higher educational attainment were more likely to both request a password and log on to the internet-based patient portal (). We did not see any differences in patient portal use by immigration status ().
In analyses adjusted for age, sex, educational attainment, employment, immigration status, and race/ethnicity, we found that race/ethnicity and educational attainment were independently associated with logging on to the patient portal (). Compared to non-Hispanic Caucasians, African–American and Latino participants were more likely not to have ever logged on to the patient portal (AOR 2.6 and 2.3, respectively). This disparity persisted when we restricted the analysis to users that requested a password.
ORs for not logging on to patient portal. GED, General Educational Development Test.
Compared to those with a college degree, those with a lower educational attainment were more likely to never have logged on. However, when we examined users who requested a password only, the strength of this pattern was attenuated and, with the exception of the comparison between high school and college graduates, became non-significant. Importantly, age over 70 years was associated with lack of use among the entire cohort; however, among users who requested a password, older subjects were more likely to log on to the patient portal compared to those under 40 (). When we examined use of the individual functions (lab results view, emails, appointments, medication refills), we found the relationships between race/ethnicity and educational attainment to be remarkably consistent with logging on and across all user functions (see appendix 1 and table 2 available online at www.jamia.org
Finally, compared to those who used the patient portal, non-users were more likely to have suboptimal control of their diabetes and related risk factors, as demonstrated by a higher proportion with HbA1c>9% (17% vs 14%, p<0.01), SBP>130 mm
Hg (50% vs 46%, p<0.01), and LDL>130 mg/dl (20% vs 18%, p<0.01).