Abstract
Introduction: Does telehealth decrease health disparities by improving connections to care or simply result in new barriers for vulnerable populations who often lack access to technology? This study aims to better understand the role of telehealth and social determinants of health in improving care connections and outcomes for Community Health Center patients with diabetes.
Methods: This retrospective analysis of Electronic Health Record (EHR) data examined the relationship between telehealth utilization and glycemic control and consistency of connection to the health care team (“connectivity”). EHR data were collected from 20 Community Health Centers from July 1, 2019 through December 31, 2021. Descriptive statistics were calculated, and multivariable linear regression was used to assess the associations between telehealth use and engagement in care and glycemic control.
Results: The adjusted analysis found positive, statistically significant associations between telehealth use and each of the 2 primary outcomes. Telehealth use was associated with 0.89 additional months of hemoglobin A1c (HbA1c) control (95% confidence interval [CI], 0.73 to 1.04) and 4.49 additional months of connection to care (95% CI, 4.27 to 4.70).
Discussion: The demonstrated increased engagement in primary care for telehealth users is significant and encouraging as Community Health Center populations are at greater risk of lapses in care and loss to follow up.
Conclusions: Telehealth can be a highly effective, patient-centered form of care for people with diabetes. Telehealth can play a critical role in keeping vulnerable patients with diabetes connected to their care team and involved in care and may be an important tool for reducing health disparities.
- Chronic Disease
- Community Health Centers
- Continuity of Patient Care
- Diabetes Mellitus
- Health Inequities
- Health Care Disparities
- Linear Regression
- Patient Care Team
- Patient-Centered Care
- Quality Improvement
- Retrospective Studies
- Telemedicine
- Social Determinants of Health
Introduction
When the COVID-19 pandemic hit the US in 2020, access to primary care services suffered. Health centers across the country quickly mobilized to meet the needs of their patients virtually through telehealth1⇓⇓–4. Telehealth remains a fixture in the delivery of care although in-person visits have nearly returned to prepandemic numbers.3,5⇓–7 Although many health centers continue to rely on telehealth, little research exists regarding its impact on chronic conditions such as diabetes in diverse and vulnerable populations.
Community Health Centers are often recognized as the nation’s safety net system, reducing health disparities and providing high quality care. They serve patients who are disproportionately low-income, members of racial/ethnic minority groups, uninsured or publicly insured, and more likely to experience significant food insecurity and housing instability in comparison to the general population.8 They also suffer from chronic conditions at higher rates than the general population. According to the National Association of Community Health Centers (NACHC), the prevalence of diabetes in Community Health Center populations is almost double that of the US general population and continues to grow rapidly.8
Like most delivery systems, Community Health Centers experienced rapid adoption of telehealth during the pandemic.6 Forty-three percent of federally funded health centers reported offering telehealth services in 2018 compared with 98% in 2020.9 Yet numerous studies have also identified barriers to accessing technology in vulnerable populations.10⇓⇓⇓⇓–15 For example, almost one-fourth of individuals with household incomes less than $30,000 per year do not have access to a smart phone and 41% do not have access to a computer.16
As virtual care becomes a more permanent fixture in health care delivery, questions remain about the impact of this utilization especially in safety net populations. Does telehealth decrease disparities by improving connection to care or simply result in new barriers because vulnerable population often lack access to technology?16,17 A survey by NACHC found that 92% of community health centers reported use of audio-only telehealth and 85% reported audio-only options increased their ability to reach vulnerable populations.18 These findings underscore the need for flexible, patient-centered approaches to ensure timely access and avoid further widening of disparities.
If these challenges can be overcome, telehealth has the potential to be a valuable tool to increase connection to care. This continuity is critical for diverse populations experiencing social determinant of health factors at increased risk of lapses in care.19,20 Social determinants of health are defined as the environmental conditions in which people are born, live, and age that affect health and quality-of-life.21 These factors impact health outcomes for patients with diabetes22 and other chronic disease23 and are significant contributors to health inequities.24 Little research exists about how telehealth is being used by populations whose social determinants of health place them at higher risk of care gaps and whether telehealth plays a role in increasing continuity of care for these patients.
This study aims to use Electronic Health Record (EHR) data to examine the role telehealth and social determinants of health play in decreasing gaps in care and improving outcomes for Community Health Center patients with diabetes. We hypothesized that telehealth would be associated with more consistent connections to care and improved glycemic control for Community Health Center patients with diabetes.
Methods
This retrospective analysis of EHR data examined the relationship between telehealth utilization and consistency of connection to the health care team (“connectivity”) and glycemic control. All patient data were deidentified to protect confidentiality. This study was granted an exemption from review by the Chicago Department of Public Health Institutional Review Board.
EHR data were collected from 20 Community Health Center or Community Health Center look alike organizations in 12 states that are part of a health center-controlled network between July 1, 2019, and December 31, 2021. The 30-month study period included a 6-month Pre period (July 2019-December 2019) and a 24-month Post period (January 2020 to December 2021). The Pre period and Post period definitions were based on the World Health Organization’s declaration of a Public Health Emergency of International Concern in January 2020.25 The study population included patients aged 18 years and older with an active diagnosis of diabetes and at least one encounter and one hemoglobin A1c (HbA1c) lab result during the study period. Lab results were included in this analysis only if the result was available in the EHR. Virtual encounters included both video and telephone visits. All Community Health Centers in the sample offered virtual encounters (the percent of included patients at each Community Health Center who used telehealth ranged from 11.5% to 77.4%.). Visit types included medical, behavioral health, and enabling services. Visits for services performed by nonbillable staff were excluded as were pregnant patients and those in long-term care or hospice.
Variables collected from the EHR included: patient demographics, type of encounter, date of encounter, diagnosis ICD-10 code, provider type, service location, HbA1c dates and results. Various social determinant of health variables including income, housing status, and insurance status were also collected. Self-reported patient characteristics data were coded as missing or unknown when relevant. We collected data on patients’ comorbid chronic illnesses via conditions defined by the Charlson Comorbidity Index.26,27
The independent variable was a binary measure of telehealth use (ie, whether the patient had any telehealth encounters) during the Post period. There were 2 primary outcomes, one for HbA1c control and one for connectivity, both of which were continuous variables. To create each primary outcome, we first calculated a binary (0/1) metric each month. The monthly measure of HbA1c control assessed whether the patient’s most recent HbA1c result was less than or equal to 7%. This threshold aligned with the American Diabetes Association recommendation of <7% HbA1c for many nonpregnant adults without significant hypoglycemia.28 Over each subsequent month, any new HbA1c result was used in respective monthly calculations (if a patient had ≥2 new HbA1c results in a month, the last result from that month was used). During months when a patient had no new HbA1c results, their most recent result was carried over for relevant monthly calculations. Then, the number of months of HbA1c control was summed to create a continuous measure (range, 0 to 24).
Using similar calculations, we also created a primary outcome for the number of months that a patient was connected to care, but with varying definitions according to patients’ level of glycemic control. Each month, patients whose most recent HbA1c result was less than or equal to 7% were categorized as connected to care if they had an encounter or HbA1c lab result within the previous 6 months. Patients whose most recent HbA1c result was greater than 7% were considered connected if they had an encounter or HbA1c lab result within the previous 3 months. This definition was based on clinical care guidelines that recommend glycemic status assessment at least 2 times a year in patients who are meeting treatment goals and have stable glycemic control and at least quarterly or as needed in patients whose therapy has recently changed and/or who are not meeting glycemic goals.28
In unadjusted analysis, descriptive statistics, including frequency counts, percentages, means, and standard deviations were calculated for all quantitative variables. For patient characteristics variables, differences between telehealth users and nonusers were calculated via t test (mean age) and Chi-squared tests (all categorical characteristics).
In adjusted analysis, we estimated 2 multivariable linear regression models examining the adjusted association between telehealth use (binary 0/1 independent variable) and each continuous primary outcome. In both regression models, we adjusted for measured sociodemographic characteristics, comorbidities, and the number of months during the Post period when no prior HbA1c data were available. In the model for glycemic control, the outcome variable was the continuous number of months of HbA1c control (range, 0 to 24), and we additionally adjusted for Pre period HbA1c control and the number of HbA1c tests the patient completed during the Post period. In the model for connectivity, the outcome variable was the continuous number of months of connectivity (range, 0 to 24), and we additionally adjusted for Pre period connectivity and the number of in-person encounters during both the Pre and Post periods.
All analyses were conducted using Stata, version 17.0 (StataCorp; College Station, TX), with a 2-sided significance threshold of P < .05.
Results
There were 35,328 patients with diabetes who met all inclusion criteria. Twenty-three patients were excluded due to missing age or sex data, resulting in a final study sample of 35,305 patients.
As shown in Table 1, the study sample had a mean age of 54.1 year (standard deviation [S.D.], 12.8) and was majority female (56.8%). Approximately one-fifth of included patients were Black race (22.2%), and nearly one-half were Latino/Hispanic ethnicity (48.7%). Patients’ insurance was most likely to be defined as unknown or uninsured (42.3%), followed by Medicaid (28.0%). Among patients with available household income data, approximately two-thirds had income at or below the federal poverty line (66.5% of those with nonmissing income data; 41.2% of full sample). Besides diabetes, the most commonly identified comorbidities were chronic pulmonary disease (16.3%) and mild/moderate renal disease (8.6%). Among all 35,305 included patients, only 105 (0.3%) had one or more telehealth encounters during the last 6 months of the year 2019 before the onset of the COVID-19 pandemic.
Patient Characteristics During the Post Period
Telehealth use increased dramatically during the years 2020 and 2021. During this 2-year Post period, 18,111 patients (51.3%) had one or more telehealth encounters (ie, telehealth users) and 17,194 (48.7%) had zero telehealth encounters (ie, telehealth nonusers). There were statistically significant differences between the 2 groups in all measured sociodemographic characteristics (P < .001 for all), though some absolute differences were small in magnitude. Telehealth users had a slightly lower mean age (mean, 53.4 years, vs 54.8 years among nonusers), and were more likely to be female (59.9%, vs 53.6%). Nearly one-third of telehealth users were insured by Medicaid (32.3%), versus only 23.6% of nonusers. We were unable to draw a conclusion about differences between uninsured and insured patients because uninsured patients were grouped with those with unknown insurance status. A slightly lower proportion of telehealth users were homeless (2.9%, vs 4.5%). Several comorbidities were more common among telehealth users, such as chronic pulmonary disease (17.9%, vs 14.6% among nonusers), mild liver disease (6.8%, vs 5.0%), and HIV (2.4%, vs 1.2%). Telehealth users were also more likely to have identified diabetes complications such as nephropathy, cataract, or neuropathy (18.2%, vs 15.9%).
In unadjusted analysis, many telehealth users were found to have few telehealth encounters (Figure 1). Telehealth users had a mean of 4.0 telehealth encounters (S.D., 5.4) and a median of 2 telehealth encounters (interquartile range, 1 to 4), during the Post period. When outcome data were compared between groups, telehealth users had slightly improved glycemic control, and greater connectivity, than nonusers (P < .001 for both) (Figure 2). During the 24-month Post period, telehealth users had controlled HbA1c for a mean of 9.3 months (S.D., 9.9), versus a mean of 8.1 month (S.D., 10.1) among nonusers (P < .001). Telehealth users met the definition of connectivity for a mean of 17.7 months (S.D., 5.7), versus 10.9 months (S.D., 8.1) among nonusers.
Distribution of telehealth encounters among telehealth users during the Post period.
Months of glycemic control and connectivity among telehealth users and nonusers during the Post period. Abbreviation: HbA1c, hemoglobin A1c.
In multivariable linear regression analysis, there were positive, statistically significant associations between telehealth use and each of the 2 primary outcomes (Table 2). In the multivariable model for our glycemic control outcome, telehealth use was associated with 0.89 additional months of HbA1c control (95% confidence interval [CI], 0.73 to 1.04). In the multivariable model for our connectivity outcome, telehealth use was associated with 4.49 additional months of HbA1c control (95% CI, 4.27 to 4.70). In both regression models, age of 35 or greater (compared with age 18 to 34), female sex (compared with male), and non-Hispanic ethnicity (compared with Latino/Hispanic) were positively associated with each respective primary outcome. Homelessness was not significantly associated with either primary outcome variable. For other defined demographic characteristics—including race, insurance type, and household income—statistical significance of defined categories varied across the 2 regression models.
Results of Multivariable Linear Regressions Investigating Associations Between Telehealth Use and Each Primary Outcome Variable
In sensitivity analyses using an alternative definition of HbA1c control at the ≤8% threshold, differences between telehealth users and nonusers were largely similar to those observed in main analyses that used the ≤7% threshold for HbA1c control (detailed results available from authors on request).
Discussion
This study found telehealth use in the Community Health Center setting was associated with increased connection to care, controlling for age, sex, race, ethnicity, various comorbidities and other social determinant of health factors for patients with diabetes. We also found a modest positive association between telehealth use and glycemic control. These findings demonstrate expansion of access due to telehealth coupled with limited positive influence on an important diabetes health outcome.
In our model, telehealth users were connected to care for nearly 4 and a half months longer than nonusers. Although our definition of connectivity did not concentrate on diabetes care specifically, this increased connection to primary care is significant and encouraging as Community Health Center populations are at greater risk of lapses in care and loss to follow up.19,20 Reducing this risk is critical for patients with diabetes given studies demonstrating that the delays in care resulting from the COVID-19 pandemic resulted in poor health outcomes for patients with chronic disease.29 Further, patients with diabetes who connect with their primary care provider are more likely to receive more current and recommended diabetes care than those without these visits,30 a finding supported by the modest increase in glycemic control we saw for telehealth users. Because the postperiod frame of this analysis was limited, the magnitude of glycemic control difference was predictably small but meaningful nonetheless.
Also of interest, our findings demonstrate the dramatic uptake of telehealth utilization during the COVID-19 pandemic described by other researchers.3,31,32 Previous studies have documented this rapid increase but not specifically among patients with diabetes in a safety net environment. The COVID-19 pandemic emphasized disparities across vulnerable populations,33⇓⇓–36 thus we are encouraged by the swift uptake of telehealth by chronic disease patients in these underserved communities.
Our study has several limitations. First, because this study included only billable encounters, it does not account for the often vital role that ancillary services such as care team education, nutritionist consults, and others play in diabetes education and management. Second, the gaps in data for social determinants of health limited our ability to fully analyze some of the characteristics described in this review. Almost 18% of homeless status data were missing from the sample, and 38% were missing household income data. In addition of note, the data were collected from a single health center-controlled network in which all entities share a common technology infrastructure. This commonality may limit the generalizability to other types of organizations. A final limitation is that the cross-sectional nature of our data does not allow us to address the question of causation; thus further exploration is necessary to explore direct clinical impacts of telehealth usage and diabetes technologies on clinical outcomes related to diabetes management.
Telehealth can be a highly effective, patient-centered form of care for people with diabetes.30,32,37 Our study demonstrates the key role telehealth can play in keeping at risk patients with diabetes connected to their care team and involved in care and may be an important tool for reducing health disparities. Further study is recommended exploring how telehealth can be optimized in conjunction with home monitoring devices and other technologies to improve glycemic control and other health outcomes. In addition, gaps in understanding of the role of social determinant of health factors in telehealth utilization for vulnerable chronic disease populations persist and warrant further attention. Telehealth is an important tool for increasing patient connection to care with the potential to help improve outcomes and reduce health disparities.
Notes
This article was externally peer reviewed.
Funding: This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health under Award Number P30DK092949. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Conflict of interest: The authors report no conflicts of interest.
To see this article online, please go to: http://jabfm.org/content/37/2/206.full.
- Received for publication September 15, 2023.
- Revision received November 27, 2023.
- Accepted for publication December 4, 2023.
