Data on the activation and utilization of an electronic health record patient portal in an adult inpatient population at an academic medical center

Electronic health record patient portals allow patients to access their own health data online and interact with the healthcare team. Many studies have focused on use of patient portals in the outpatient setting. Relatively less is known about use of patient portals for hospitalized patients. The data in this article include analysis of patient portal activation and utilization for adults hospitalized in 2018 at an academic medical center in a Midwestern state in the United States. Activation was assessed by percentage of patients who had activated their patient portal by the time of data review. Utilization of the patient portal was determined by whether patients or their legal proxies accessed one or more reports from diagnostic testing ordered during inpatient encounter(s) in 2018. The data include 826,843 diagnostic tests on 40,640 unique patients. Patient characteristics include sex, age, whether outpatient diagnostic tests were also performed in 2018, preferred language (English or non-English), health insurance status (private, public, or uninsured), self-declared race (White or non-White), and whether there was a legal proxy for the patient. Association of these covariates with patient portal activation and utilization were analyzed.


a b s t r a c t
Electronic health record patient portals allow patients to access their own health data online and interact with the healthcare team. Many studies have focused on use of patient portals in the outpatient setting. Relatively less is known about use of patient portals for hospitalized patients. The data in this article include analysis of patient portal activation and utilization for adults hospitalized in 2018 at an academic medical center in a Midwestern state in the United States. Activation was assessed by percentage of patients who had activated their patient portal by the time of data review. Utilization of the patient portal was determined by whether patients or their legal proxies accessed one or more reports from diagnostic testing ordered during inpatient encounter(s) in 2018. The data include 826,843 diagnostic tests on 40,640 unique patients. Patient characteristics include sex, age, whether outpatient diagnostic tests were also performed in 2018, preferred language (English or non-English), health insurance status (private, public, or uninsured), self-declared race (White or non-White), and whether there was a legal proxy for the patient. Association of these covariates with patient portal activation and utilization were analyzed.

Data Description
We compiled data on 826,843 diagnostic test (pathology/laboratory and radiology) reports from 40,460 unique patients that had testing performed during one or inpatient admission(s) at an academic medical center in 2018. There is growing literature on factors influencing patient engagement and use of electronic health record (EHR) patient portals, especially in the ambulatory setting [1][2][3][4][5][6] . More recent research has focused on patient portal use in the acute care [7] , emergency care [8] , and inpatient settings [9][10][11][12][13][14] . In the United States, a federal mandate known as the Cures Act will dictate that health systems release all medical notes and diagnostic test reports to patients when they are available [15] . Patient portals provide a major route to achieve these goals, and increased use will have significant impacts on patient engagement with their own healthcare [4 , 11 , 16 , 17] . The Cures Act was scheduled to take effect November 2, 2020 but has been delayed likely until spring of 2021. The data presented provide a baseline for inpatient portal use prior to any effects of the Cures Act.
Our institution (University of Iowa Hospitals and Clinics) adopted the Epic EHR for both inpatient and outpatient care in May 2009 and implemented MyChart in 2010 [8] . MyChart is a patient portal 'tethered' to the Epic EHR, meaning that MyChart allows patients (or legal proxies such as parents/guardians of young children) to access their own patient data from the Epic database once a password-protected account is set up [2 , 4 , 5 , 14] . Patients who are seen in either inpatient or outpatient settings within our institution are provided information to activate MyChart accounts. Once set up, the accounts can be viewed from either computers or mobile devices (including a smartphone app) that are connected to the internet. Patients can use My-Chart to schedule appointments, view diagnostic test reports, renew prescriptions, fill out medical questionnaires or forms, access inpatient discharge summaries, and send messages to the healthcare team. The hardware hosting Epic and MyChart are maintained by Health Care Information Services of the University of Iowa. Hospital staff receive training in the functions of MyChart and often utilize MyChart in their own healthcare or that of their families. Specialized database reporting tools within Epic (described in more detail below) allow hospital staff to analyze patient utilization of MyChart, provided that ethical approval has been granted.
For the 40,460 unique patients that had diagnostic testing performed during hospitalization in 2018, account activation was assessed by percentage of patients who had activated their patient portal by the time of data review (January 2020). Utilization of the patient portal was determined by whether patients used the patient portal to access one or more reports from diagnostic tests ordered during hospitalization. During the retrospective time period, there were limited features of the patient portal for inpatient encounters. This included viewing of diagnostic test reports, discharge summaries, and post-visit summaries. Of these, only viewing of diagnostic test reports was captured as a metric in Epic that could be later analyzed. Hard copies of the post-visit summary were given to patients/families on discharge; this also contained information on activating a MyChart account for those who had not done so already.
Original data for this study are available at Mendeley Data ( http://dx.doi.org/10.17632/ 3wc6t92rrg.1 ) which include data for 826,843 diagnostic test results on 40,640 unique patients. Specific data fields include: unique patient identification number (deidentified), number of inpatient admissions in 2018, whether test result was during the initial admission in 2018 or a subsequent admission, specific diagnostic test ordered, diagnostic test category, sex (as officially noted in the electronic health record), patient portal status (active or inactive), whether diagnostic test result was viewed in the patient portal, the most common self-declared race categories, simplified race category (White or non-White), age at time of testing, age category, primary/preferred language (as recorded in electronic health record; English on non-English), whether there was a legal proxy that had access to the patient's portal account, and insurance category (private/commercial or public/uninsured). Table 1 shows statistics for viewing of diagnostic tests (pathology/laboratory and radiology), with utilization defined as viewing at least one test report by time of data review. The data is broken down by sex (male/female), whether outpatient diagnostics were also ordered in 2018, whether patient had only one or more than one inpatient admissions in 2018, preferred/primary language (English or non-English), health insurance status in 2018 (public insurance/uninsured or private insurance), self-declared race (White or non-White), and whether patient had legal proxy who could access the patient's EHR portal (establishing a proxy required committee approval and was typically due to patient mental incapacitation, intellectual disability, or other factors that impacted ability to make medical decisions). We employed univariate comparisons using X 2 tests for categorical variables. We also performed multivariate logistic regression analysis to determine the association between patient medical and demographic characteristics and the endpoints of patient portal activation and utilization. On multiple logistic regression analysis, female sex, outpatient studies, performance of both laboratory and imaging studies during hospitalization, English as preferred language, commercial insurance, self-declared race as White, age, and presence of legal proxy for the patient showed significant association with MyChart account activation ( Table 2 ) and viewing of diagnostic test reports ( Table 3 ). Table 4 uses the same dataset as Table 1 but with data expressed as percent viewing of total diagnostic tests. Table 5 uses the same format as Table 1 but with the dataset restricted to pathology/laboratory tests only. This table also includes statistics for whether patients who had pathology/laboratory tests also had radiology tests performed in 2018 during hospitalization. Table 6 uses the same dataset as Table 5 but with data expressed as percent viewing of total tests. Table 7 uses the same format as Table 1 but with the dataset restricted to imaging studies only. This table also includes statistics for whether patients who had radiology tests also had   pathology/laboratory tests performed in 2018 during hospitalization. Table 8 shows the same dataset as Table 7 but with data expressed as percent viewing of total tests. Fig. 1 displays the percent of patients within various subcategories who have active patient portal account (panel A) and who viewed at least one diagnostic test reports (pathology/laboratory and radiology) performed during hospitalization in 2018 (panel B). This figure uses the entire dataset of pathology/laboratory and radiology tests summarized in Table 1 . Fig. 2 displays the percent of patients within various age brackets who have active patient portal account (panel A) and who viewed at least one diagnostic test report (pathology/laboratory and radiology) performed during hospitalization in 2018 (panel B). Fig. 3 shows the percent of all diagnostic test reports viewed in various test categories. The test categories are Anatomic Pathology (which includes surgical pathology, cytopathology, and dermatopathology), Chemistry (includes clinical chemistry, therapeutic drug monitoring, and toxicology), computed tomography (CT) scan, Hematology (includes bone marrow and hemostasis/thrombosis), Send-out Tests (sent to reference laboratory), Microbiology, magnetic resonance imaging (MRI)/nuclear scans, and X-rays. Data is also aggregated as All Labs (pathology/laboratory) and All Imaging.

Experimental Design Materials and Methods
Epic Reporting Workbench (RWB), a software tool for retrieving specific data from the EHR, was used to retrieve all diagnostic test orders (pathology/laboratory and radiology) ordered during hospitalization between January 1, 2018 and December 31, 2018 on patients 18 years and older [18] . With each diagnostic test order, the RWB search also retrieved age, sex (male or female), self-identified race, self-identified preferred language, insurance status, prior hospitalization, presence of outpatient lab reports, and whether patient had a proxy that could legally access their patient portal. Race was categorized as White or race other than White (included African-American, American Indian / Alaska Native, Asian, Hispanic/Latino, Mixed Race, Native Hawaiian / Pacific Islander, Other, Unknown and Declined), language as English or Non-English (included a total of 75 other languages with the six most common other than English being Spanish, Arabic, French, Chinese, Swahili and Vietnamese) and insurance as commercial payor or public insurance (Medicare or Medicaid)/uninsured. Data was also analyzed by testing category. Pathology/laboratory tests were categorized as Anatomic Pathology (including surgical biopsies and resections, cytopathology, and dermatopathology), Chemistry (includes toxicology, therapeutic drug monitoring, and clinical chemistry), Hematology, Microbiology, and Send-out (referred to outside reference laboratories). Imaging tests were categorized as computerized tomography (CT) scan, magnetic resonance imaging (MRI), nuclear scans, and X-rays.

Ethics Statement
The analyses had approval by the University of Iowa Institutional Review Board (protocol # 201907709) as a retrospective project.

Declaration of Competing Interest
The authors declare they have no known competing financial interests or personal relationships which have, or be perceived to have, influenced the work in this article.