Abstract
Purpose: Electronic health records and patient portals allow patients to communicate with physicians and other members of the health care team. However, few studies have investigated the impact of these tools on chronic disease management. This study examines how patient portal activity can influence the management of type 2 diabetes (T2DM) and hypertension.
Methods: This is a retrospective cohort study from 2018 to 2022 of primary care encounters of large metropolitan health system patients 18 years old and older with a diagnosis of hypertension or T2DM. The primary exposure was total log-ins into the patient portal MyChart (Epic Systems Corporation, Verona, WI). Primary endpoints for hypertension were blood pressure (BP) < 140/90 mmHg, and T2DM was glycohemoglobin (HbA1c) ≤ 8% and tight control (HbA1c ≤ 7%). Multivariate logistic regression models were adjusted for age, ethnicity/race, sex, median household income, Charlson Comorbidity Index, training site status, and baseline disease control.
Results: Regression results showed increased MyChart frequency increased the likelihood of having controlled T2DM (OR 1.77, P < .001). Having baseline BP < 140/90 mmHg or HbA1c control increased the likelihood of outcome control in all models. BP was less likely to be controlled at outcome in African Americans (OR 0.90, P < .001) and more likely to be controlled in males (OR 1.12, P < .001).
Conclusions: Increased patient portal usage is associated with greater likelihood of T2DM and hypertension control. Future studies should examine specific patient portal features, their usage and impact on health outcomes.
- Clinical Medicine
- Cohort Studies
- Continuity of Patient Care
- Disease Management
- Electronic Health Records
- Hypertension
- Logistic Regression
- Patient Portals
- Primary Health Care
- Retrospective Studies
- Type 2 Diabetes Mellitus
Introduction
Hypertension and diabetes are two of the most common chronic diseases seen by primary care physicians, and these two disorders account for most primary care encounters worldwide.1 In the United States, the total health care-associated costs for treating hypertension and type 2 diabetes are approximately $79 billion and $327 billion, respectively.2 The chronic care model is a health care model that outlines the essential elements of medical care that support high-quality treatment of patients with chronic diseases, including the community, the health system, self-management support, delivery system design, decision support, and clinical information systems.3 Importantly, patients with chronic illnesses need consistent care, continuity with health care team members and monitoring of their conditions; however, throughout the COVID-19 global pandemic, many of the chronic-care model elements were either compromised or entirely inaccessible to patients, which ultimately led to numerous difficulties for those with chronic conditions.
During the COVID-19 pandemic, physicians faced significant barriers to diagnosing, treating, and providing follow-up care to patients with chronic diseases. Providing health care via telemedicine technologies became popular during the pandemic because meeting with patients virtually avoided viral exposure and allowed physicians to see patients while following public health physical distancing guidelines. From 2019 to 2021, the use of telemedicine increased for office-based physicians from 15.4% to 85.9% and was higher among primary care physicians and medical specialty physicians than it was among surgical specialty physicians.4 Electronic health records (EHR) and online patient portals, such as MyChart (Epic Systems Corporation, Verona, WI), AthenaOne (AthenaHealth, Boston, MA), and Cerner PowerChart, also provided a way for patients to interact with physicians and other members of the health care team during the pandemic. Studies have shown that patient engagement through the EHR can help mitigate disease burden and increase medication adherence.5 While previous studies have looked at patient portal use and improvement of diabetic outcomes, few studies have looked at hypertension or other chronic conditions.6–11 In addition, those studies were done before the global pandemic and only assessed patients who had an EHR account, not the actual use of the electronic patient portals.7,8,12–14 Mechanisms of EHR use that have been studied include access to patient education material, viewing of laboratory results, and secure messaging systems for communicating with support staff, providers, or physicians. Thus, whether consistent use of patient portals and the EHR by patients indeed helps individuals manage chronic illnesses is not entirely clear. To address this research gap, we performed a retrospective study of patients with chronic illness and explored the association of EHR frequency of use and other patient characteristics with how well the patients managed their own conditions. The study focused on patients with diabetes, hypertension, or both, evaluating various clinical outcomes as indicators of chronic disease management. Understanding how patients use patient portals for self-care of chronic conditions will aid in the development of better, more user friendly and effective EHR-based patient resources.
Methods
Study Population
This was a retrospective cohort study of patients with chronic health conditions who were treated between January 2018 through December 2020 (baseline), and who had received follow-up care at Henry Ford Health primary care sites within Detroit and the surrounding metropolitan area between January 2021 and December 2022. Inclusion criteria were age ≥ 18 years old; previous diagnosis of primary hypertension or type 2 diabetes, at the time of the initial encounter; as well as having received follow-up care between January 2021 and December 2022. Patients were also required to have an activated MyChart patient portal. Conditions were defined as the presence of International Statistical Classification of Diseases and Related Health Problems, 10th revision, Clinical Modification codes for type 2 diabetes (E11.xx) or hypertension (I10.xx).
Exposure Measure
Frequency of MyChart patient portal use was measured electronically by account activation status and number of log-ins to the MyChart patient portal during the study period. Log-ins were grouped based on the quartile averages over the entire 5-year study period as 0, 1 to 34, 35 to 150, 151 to 350, and > 350.
Outcome Measures
The main outcomes were disease-specific and were based on current American Academy of Family Physicians guidelines.15 Outcomes for hypertension was maintenance of blood pressure < 140/90 mmHg. Outcomes for type 2 diabetes were glycohemoglobin (HbA1c) percentages: general control at 7% to 8% and at tight control < 7%. Hospital use was also assessed by the mean number of emergency department visits, hospital admissions, and subsequent duration of hospital stay.
Dataset Source and Study Variables
Data were obtained from the Henry Ford Health medical system administrative EHR databases and filtered based on inclusion/exclusion criteria. The following baseline patient characteristics were recorded: age, sex, race/ethnicity as reported in the medical record, median household income, and Charlson Comorbidity Index.
Age was stratified as follows: 18 to 29, 30 to 39, 40 to 49, 50 to 64, 65 to 79, and ≥ 80 years. Sex was categorized as male and female as self-reported in the EHR. Race/ethnicity was categorized as Asian, Black/African American, Hispanic, White, and other/unknown. Median household income was obtained by mapping patients’ addresses to census block group level median household income.
Statistical Analysis
Baseline characteristics were used as the main predictors in all models. Categorical data were described as frequencies and percentages and continuous data were described as the mean with standard deviation. A logistic regression model was built using predictors to quantify the likelihood of disease control and calculate odds ratios (OR) and 95% confidence intervals (95% CI).
This study was approved by the Institutional Review Board at Henry Ford Health.
Results
Baseline Characteristics
There were 69,308 unique patients who received care at Henry Ford Health primary care sites between January 1, 2018, and December 31, 2022. After excluding those without a diagnosis of hypertension or diabetes, we had an initial cohort of 23,340. Our final analytic sample only included 22,465 patients that had an activated patient portal. Excluding patients without primary care visit or disease outcome recorded during 2021 to 2022 (blood pressure or HbA1c), we were left with our final cohorts of hypertension with 20,722, diabetes mellitus with 8,814, and for 7,071 patients with comorbid disease hypertension and diabetes (Figure 1). The mean age of the overall cohort was 63.6 ± 13.4 years, 60.5% were female and Black/African American comprised 49.2% (Table 1).
Criteria to determine final study cohort. Abbreviations: A1c, glycohemoglobin; BP, blood pressure; MRN, medical record number.
Patient Demographics and Clinical Characteristics by Chronic Disease Diagnosis
We first characterized MyChart users, defined as anyone with a patient portal account that had been activated at any point. Of the final cohort population, only a relatively small number, 3.5%, had not activated their account and were excluded from further analysis, whereas 63% activated their MyChart patient portal account before 2018 and 33% activated their account after 2018. Patient portal log-ins ranged from 0 to 2,651 (Table 2). Over the entirety of the study, only 9.6% patient portal users never accessed their account despite being activated. Access to the EHR rose over the course of the study with the largest increase in patient portal log-ins taking place during 2020 and 2021 (Figure 2). Mean number of patient portal log-ins also increased each year from 2018 to 2022 from 33.5 ± 59.4 to 70.1 ± 105.6 (Table 2).
Scaled frequency of total EHR logins by year. Abbreviation: EHR, electronic health records.
Patient Portal Log-Ins by Year
A multivariate logistic regression was completed to assess the impact of patient portal usage on blood pressure or HbA1c control at encounters in 2021/2022. With MyChart patient portal use between 1 and 34, the data demonstrated a 22% decreased likelihood of blood pressure control, blood pressure < 140/90 mmHg, a 2% to 77% increased likelihood of HbA1c < 8%, and 2% to 35% increased likelihood of HbA1c < 7%. The same regression demonstrated that the strongest predictors of outcome blood pressure and HbA1c control at outcome were baseline blood pressure and HbA1c control. Baseline blood pressure control had OR 1.79 (95% CI, 1.68-1.90), indicating an almost 2 times likelihood of blood pressure control at outcome in 2021/2022 with baseline blood pressure control. Similarly, HbA1c < 8% at baseline had OR 6.55 (95% CI, 5.77-7.44, P < .001) and HbA1c < 7% at baseline had OR 8.10 (95% CI, 7.28-9.01, P < .01), indicating nearly 8 times likelihood of HbA1c control at outcome (Table 3).
Adjusted Logistic Regression: Hypertension Control or Diabetes Control in Patients during 2021/2022
Given these results, we then sought to examine only those patients with uncontrolled blood pressure at baseline (>140/90 mmHg). Logistic regression demonstrated a non significant increased likelihood of blood pressure at goal at outcome encounter with MyChart patient portal use between 51 and 350 log-ins (Table 4). A similar multivariate logistic regression that looked only at patients with uncontrolled HbA1c at baseline demonstrated a 35%–58% increased likelihood of HbA1c < 8% at outcome encounters in 2021/2022 with MyChart use strata between 151–350 log-ins and over 350 log-ins, respectively (Table 4). Charlson Comorbidity Index and median household income were not identified to have statistically significant effect on either blood pressure or HbA1c outcomes (Table 4). Finally, we assessed the impact that patient portal usage had on health care system utilization; however, when comparing across log-in strata, there was no statistically significant difference in emergency department visits, likelihood of hospital admission, or length of stay (Table 5).
Adjusted Logistic Regression: Hypertension Control or Diabetes Control in Patients Not in Control at Baseline
Patient Portal Usage by Emergency Department Visits, Hospitalizations and Length of Stay During Hospitalizations
Discussion
In this retrospective study, we examined the role of patient portal usage, among the myriads of drivers for achieving hypertension and diabetes managements goals in an urban ambulatory setting. We saw a wide variance in usage patterns. Surprisingly despite having an active patient portal, most users never accessed their account, a trend that was redemonstrated annually throughout the study period. In contrast a single user recorded 2,651 log-in events over 12 months. We did observe a small, but appreciable decline in the proportion of users without log-ins over the course of the 5 years examined, with a concomitant increase in both the mean and median of log-in events.
When first analyzing the data, we identified that the largest determinant of disease control at the end of our study was unsurprisingly disease control at baseline. However, subgroup analysis of patients who were uncontrolled at baseline showed that blood pressure was improved with each increase in patient portal use strata for those who were assigned male at birth. Conversely, worse outcomes were observed among patients identifying as Black/African American. A similar subgroup analysis for both liberal and tight glycemic cutoffs, demonstrated that achieving goal HbA1c levels was more likely associated with increase patient portal usage and increasing age.
Interestingly for patients with both hypertension and diabetes, blood pressure control at baseline was the only factor associated with improved blood pressure outcomes, while blood sugar control was associated with patient portal usage, as well as baseline HbA1c, and age, and negatively associated with Charlson Comorbidity Index strata. We hypothesize that this may be related to the complexity of patient care involved in blood sugar management, including the need for more frequent care team encounters, invasive laboratory monitoring, and nonoral medications. Though we investigated trends regarding health care system utilization by examining emergency department visits, hospitalizations, and increased length of stay, and saw that higher and lower extremes of patient portal usage were associated with higher rates of health care system use, these trends are not statistically significant.
The American Recovery and Reinvestment Act mandated meaningful use of the EHR by physicians, nonphysician providers, and allied health professionals by the start of 2014. Goals of this project by the US Department of Health and Human Services include improvements to patient care, patient participation, care coordination, diagnostics and patient outcomes, as well as practice efficiencies and cost savings. Since that time, EHR and patient portal utilization has broadly enabled patients to have episodic, asynchronous, nearly real-time access to health care teams, which has improved patient outcomes.16 Accessing an online patient portal has previously been associated with younger age, higher baseline weight with no gender differences, or impact on follow-up6; however, patient access to EHRs was linked to improvements in 2 of the 6 Institute of Medicine’s domains of health care quality: patient safety and effectiveness.10 This contrasts with our study, which identified some differences based on race/ethnicity as well as gender.
This research represents the first study to demonstrate the link in changing behaviors in patient portal utilization and chronic disease outcomes spanning the COVID-19 pandemic. Interestingly, while patient portal log-ins increased during that time, there remains a wide variability in the frequency of usage. This can at least be partially attributed to the skewed distribution in patient age in our study population and technology literacy but may also be impacted by regular access to internet capable devices such as a telephone, tablet, or computer. Similarly, a significant enrollment-disuse gap was observed, with the largest proportion of users never logging in throughout the study. Although this group decreased following the onset of the COVID-19 pandemic, it underscores a critical opportunity for the health care system to enhance patient engagement. While these individuals activated their MyChart profiles, they did not use them to log in. Further studies should explore the reasons behind this behavior. For example, were patients uncomfortable with or lacking confidence in using technology, or did they have concerns about its trustworthiness? At this stage, we can only hypothesize.
Future directions of this research include prospective studies examining specific patient portal features, their usage, and impact on health outcomes. Furthermore, at a systems level promoting widespread efforts to drive EHR patient portal enrollment, proxy assignment, and utilization may increase patients access to care and drive down associated morbidity and health care system costs. A similar study with post-COVID-19 pandemic data would also be interesting to further look at how patient portal usage may have changed emergency department visits and hospitalizations as well.
A potentially unexpected consequence of patient portal utilization was an increase in the administrative work associated with EHRs. Before the COVID-19 pandemic, primary care physicians were estimated to spend 12.4% of an average workday and 23.7% of total EHR time managing their electronic in-basket, defined as refills and results management, letter generation, responding to messages in the patient portal, or telephone calls.17 Though multiple strategies have been implemented to decrease administrative fatigue associated with EHR in-basket management, there have been varying effects on mitigating burnout and promoting resilience.18 In response to the COVID-19 pandemic, telehealth and remote access to health care services resulted in increased rates of EHR use,5,19 further exacerbating physician burnout through asynchronous tasking.20,21 Future studies should aim to look at how increased patient portal usage affects physical well-being and overall job satisfaction.
Limitations
This study had several limitations. First, this was a retrospective cohort characterization of patients within a single health system, and therefore cause and effect cannot be deduced, and results may not be generalizable. In addition, account log-ins were used as a proxy for patient portal usage. Patient portals have a variety of functions including scheduling appointments, reviewing results, requesting refills, as well as communication with care teams. While in practice and we were not able to ascribe a specific function or patient portal utility accessed during each digital encounter, we believe that any log-in reflects patient engagement in navigating their own care and should be considered. Relatedly, the broad distribution of usage, ranging from 0 to 2,651 log-ins for a single user, led to skewed log-in strata based on quartile information over the entire study range, losing nuance with less frequent and nonusers. Regarding population sampling, the patients examined were predominantly Black/African American and this is not representative of the broader US population.
Conclusions
Increased patient portal usage is associated with greater likelihood of diabetes and hypertension control. Future studies should examine specific patient portal features, their usage and impact on health outcomes.
Acknowledgments
The authors thank Dr. Passalacqua and Ms. Stebens for their assistance with the study.
Notes
This article was externally peer reviewed.
Funding: No financial disclosures or financial support were reported by the authors of this paper.
Conflict of interest: The authors declare that they have no conflicts-of-interest to disclose related to the conduct or writing of this study.
- Received for publication February 19, 2025.
- Revision received April 8, 2025.
- Accepted for publication May 12, 2025.
References
Jump to section
Related Articles
Cited By...
- No citing articles found.