Improving Specialty Care Access via Telemedicine

Abstract

Introduction:

Telehealth is a potential solution to persistent disparities in health and health care access by eliminating structural barriers to care. However, its adoption in urban underserved settings has been limited and remains poorly characterized.

Methods:

This is a prospective cohort study of patients receiving telemedicine (TM) consultation for specialty care of diabetes, hypertension, and/or kidney disease with a Federally Qualified Health Center (FQHC) as the originating site and an academic medical center (AMC) multispecialty group practice as the distant site in an urban setting. Primary data were collected onsite at a local FQHC and an urban AMC between March 2017 and March 2020, before the COVID-19 pandemic. Clinical outcomes of study participants were compared with matched controls (CON) from a sister FQHC site who were referred for traditional in-person specialty visits at the AMC. No-show rates for study participants were calculated and compared to their no-show rates for standard (STD) in-person specialty visits at the AMC during the study period. A patient satisfaction questionnaire was administered at the end of each TM visit.

Results:

Visit attendance data were analyzed for 104 patients (834 visits). The no-show rate was 15%. The adjusted odds ratio for no-show for TM versus STD visits was 1.03 [0.66–1.63], p = 0.87. There were no significant differences between TM and CON groups in the change from pre- to intervention periods for mean arterial pressure (p = 0.26), serum creatinine (p = 0.90), or estimated glomerular filtration rate (p = 0.56). The reduction in hemoglobin A1c was significant at a trend level (p = 0.053). Patients indicated high overall satisfaction with TM.

Discussion:

The study demonstrated improved glycemic control and equivalent outcomes in TM management of hypertension and kidney disease with excellent patient satisfaction. This supports ongoing efforts to increase the availability of TM to improve access to care for urban underserved populations.

Introduction

Telehealth is a potential solution to persistent disparities in health and health care access.¹ Its adoption may advance health equity by eliminating structural barriers to care.^2,3 However, the potential to exacerbate disparities through inequitable access, uptake, and effectiveness exists.⁴ Despite the recent increase in adoption of telehealth, its use in urban underserved settings remains poorly characterized.

Federally qualified health centers (FQHC) provide care to almost 30 million people in the United States, including one in three individuals living in poverty.⁵ Although FQHCs serve as a critical and trusted source of primary care in every state, access to specialty care in this setting remains limited.

Telemedicine (TM) and on-site specialty access have been explored as potential solutions. Both options have demonstrated positive outcomes with respect to referral completion and patient satisfaction in limited clinical applications.^6,7

Documented structural and individual barriers to specialty care for FQHC patients include, but are not limited to, office inaccessibility, long travel times, and limited transportation. FQHCs inherently face challenges in implementing telehealth due to regulations that limit long-term strategic integration of TM into their care delivery model.⁷ Out of necessity, the COVID-19 pandemic spurred a significant increase in telehealth visits,⁸ but there is uncertainty that regulatory waivers reducing barriers to telehealth for patients and FQHCs will be made permanent.

We developed a TM model for urban underserved patients through the collaboration between a FQHC and an academic medical center (AMC). Our study occurred before the COVID-19 pandemic and included specialty referrals to endocrinology, cardiology, and nephrology for the management of diabetes, hypertension, and chronic kidney disease. Before the TM program, it was noted that the local FQHC patients had low rates of referral completion for specialty care, while at the AMC, high no-show rates reduced clinician productivity and patient access. We hypothesized that TM could facilitate specialty consultations if the FQHC served as the originating site. We conducted an evaluation of the visit no-show rate, clinical outcomes, and patient satisfaction.

Methods

We conducted an Institutional Review Board-approved prospective cohort study of patients receiving telemedicine (TELE) consultation for specialty care between a FQHC as the originating site and an AMC multispecialty group practice as the distant site in an urban setting (Washington, DC). Participant outcomes were compared with a control (CON) group composed of patients from a sister FQHC site referred for in-person specialty visits at the AMC during the same time period.

The TM system consisted of a secure video platform linking the specialist at the AMC and a dedicated examination room at the FQHC with a Universal Serial Bus (USB) stethoscope, otoscope, and dermatoscope. A nurse or medical assistant served as the tele-presenter after completing 4 h of training on the TM system, including assessment of key physical examination findings.

Adult patients were referred to the program by their FQHC primary care provider if they (1) completed at least one primary care visit at the FQHC, (2) were diagnosed with diabetes, hypertension, and/or kidney disease, (3) had received a referral to a specialist from a FQHC provider for endocrinology, cardiology, or nephrology, and (4) spoke English. A research associate enrolled the patient in the study, obtained consent, completed a specialty-specific intake form, and assisted in scheduling the patient's TM appointment time with the specialist. The FQHC used their usual reminder systems for participants with TM appointments. All visits occurred between March 1, 2017 and March 31, 2020. Specialists used their standard practice. Services were billed through the patient's insurance and if uninsured, were covered by grant funds. Participants did not receive financial incentives.

MEASURES

The primary outcomes were no-show rates and patient satisfaction in the TELE group. A patient satisfaction questionnaire was administered at the end of each TM visit. Some participants also completed a survey about the reason for a missed visit. Quantitative clinical measures for both the TELE and CON groups included hemoglobin A1c (A1c) for glucose control, mean arterial pressure (MAP) for blood pressure control, and serum creatinine (SCr) and estimated glomerular filtration rate (eGFR) for kidney function.

INTERVENTION GROUPS

TELE patients included all those referred for kidney disease, hypertension, or diabetes who agreed to participate during the study period. CON consisted of patients from a sister FQHC site matched for age, gender, race, diagnosis, and disease burden. A greedy matching algorithm attempted to match two CONs to each TELE participant, allowing for a difference of up to 5 years on age, and requiring an exact match on sex, race, and specialty referral.

INTERVENTION PERIOD

The study was conducted between March 2017 and March of 2020. The intervention period for each participant started at the date of the first scheduled TM visit for those in the TELE group. The CON subjects matched to TELE patients used the same date as their matched TELE case to establish the preintervention versus intervention periods.

DATA ANALYSIS

Visit attendance

A within-group comparison was made between the no-show rate of TELE and standard in-person (STAND) specialty visits at the AMC during the intervention period using the Fisher exact test. The primary analysis for this outcome used generalized estimating equations (GEE) with binary outcome, with visits nested within patients, to account for within-patient correlation in the likelihood of having no-shows. Patient satisfaction items were analyzed in the TELE group using the mean score, after confirming that internal consistency reliability was adequate. This was tested using Cronbach's alpha. If Cronbach's alpha >0.80, this would indicate that items formed a single scale, and their mean could be used as a measure of the scale score.

Clinical outcomes

For quantitative clinical outcomes, we compared a baseline time period, including 3 years before the intervention period (PRE), and all visits during the intervention period (INTER) for TELE and CON. For A1c, the baseline value was the last A1c reading of PRE. Quantitative clinical variables were evaluated by examining the main effects of time, and of group, and the group by time interaction in a series of random effects mixed model regressions, with a separate model for each outcome variable (MAP, SCr, eGFR, A1c).

Models were adjusted for age, sex, and race as possible confounds. The time effect measures change over time averaged over groups. Univariate mixed models were used to test time effects within groups, and a random effects mixed model with binary outcome was used to test the group and time effects. SAS (version 9.4, Cary, NC) was used for data analysis, with p < 0.05 considered significant. Results are expressed as mean ± standard deviation (SD), percentage, or odds ratio (OR) with 95% confidence interval (CI).

Results

A total of 114 unique TELE patients participated in the study. After matching, 199 CON and 104 TELE cases were available for analysis. Patient characteristics are shown in Table 1.

Table 1.

Patient Characteristics During Baseline

PATIENT VARIABLE	CON ( n = 199)	TELE ( n = 104)	p
Count female	101 (51%)	51 (49%)	0.78
Mean age (years)	53.5 ± 13	54.2 ± 13	0.67
Race			0.77
Black	194 (97%)	99 (95%)
Hispanic	1 (0.5%)	1 (1%)
White	1 (0.5%)	1 (1%)
Unknown/other	3 (2%)	3 (3%)
Specialty			0.87
Nephrology	70	35
Cardiology	20	9
Endocrinology	109	60
Mean BP (mmHg)
Diastolic	85 ± 14	84 ± 13	0.60
Systolic	142 ± 24	139 ± 24	0.34
MAP	104 ± 16	103 ± 16	0.43
A1c^a	9.4 ± 2.5	9.0 ± 2.9	0.37
Mean kidney function
SCr (mg/dL)	1.19 ± 0.63	1.17 ± 0.55	0.89
eGFR (mL/min)	83 ± 32	78 ± 31	0.51

n (column %) or mean ± SD are shown. For continuous variables, we calculated the mean during the baseline period.

Baseline value is the last A1c reading before the intervention period.

A1c, hemoglobin A1c; BP, blood pressure; CON, control; eGFR, estimated glomerular filtration rate; MAP, mean arterial pressure; SCr, serum creatinine; SD, standard deviation; TELE, telemedicine.

VISIT ATTENDANCE

A total of 834 visits were scheduled among the TELE patients, of which 126 were no-shows (15%). Among the set of all visits, 380 were TELE visits (46%), and 454 were STAND visits (54%). There were 69 no-shows (18%) among TM visits, and 57 (13%) among STAND visits (p = 0.03). In the primary analysis after adjusting for within-subject correlation of visit status using GEE with visits nested within patients to account for within-patient correlation of visit status, there was no longer a significant effect in all patients (adjusted OR for no show was 1.19 [0.81–1.76], p = 0.38) for TELE versus STAND visits. Of the 22 participants contacted regarding the reason for a missed visit, 9 provided a specific reason, consisting of conflicts with other scheduled appointments, inability to get off work, inability to find transportation, and being hospitalized.

PATIENT SATISFACTION

Mean scores and percent-agreement for the 11 patient satisfaction items are presented in Table 2. There was strong agreement on all items. Cronbach's alpha for the 11-item scale was 0.95, indicating strong internal consistency reliability. The mean of that score across patients was 1.62 ± 0.69 (95% CI [1.51–1.72]), indicating that overall satisfaction with TM was rated midway between “Agree” and “Strongly Agree”.

Table 2.

Patient Satisfaction Items

ITEM	MEAN	SD	%AGREE OR STRONGLY AGREE	%DISAGREE OR STRONGLY DISAGREE
Video visits are a convenient form of health care delivery for me	1.46	0.76	94	2
Video visits save me time	1.47	0.78	95	3
My privacy is protected during video visits	1.53	0.77	93	2
I can always trust the equipment to work	1.92	0.96	78	8
I can explain my medical problems well enough during my video visit	1.49	0.72	96	2
A specialist can get a good understanding of my medical problem during a video visit	1.55	0.81	91	4
Talking to a specialist during a video visit is as satisfying as talking in person	1.58	0.86	90	4
I am more likely to use this technology in the future	1.48	0.87	91	4
In general, I am satisfied with the TM system	1.49	0.76	94	3
Since my first visit, my health is better than it was before I used the TM system	2.03	1.03	65	7
Since my first visit, I follow my doctor's advice better since working with the TM system	1.69	0.92	84	4

Response choices: 1: strongly agree, 2: agree, 3: neutral, 4: disagree, 5: strongly disagree.

TM, telemedicine.

HEMOGLOBIN A1c

A total of 60 TELE and 109 matched CON patients were referred to endocrinology. There was a significant time effect (p = 0.0017), with mean A1c levels dropping from baseline to intervention periods. The group effect was nonsignificant, meaning that CON and TELE patients had similar A1c levels. The group x time interaction was significant at a trend level (p = 0.053), indicating that the decline from pre to intervention periods was different between groups. The decline was larger in the TELE group (Table 3). Using the last A1c reading for the pre and intervention periods for each patient, the mean decline in Tele was 10.6–9.1 = 1.5, while the mean decline in CON patients was 9.9–9.5 = 0.4.

Table 3.

Mean Hemoglobin A1c with 95% Confidence Interval by Time and Group, Adjusted for Age, Sex, and Race

GROUP	TIME	MEAN	95% CI
CON	PRE	9.9	8.8–10.9
CON	INTER	9.5	8.4–10.7
TELE	PRE	10.6	9.4–11.7
TELE	INTER	9.1	8.1–10.1

We used the last A1c reading of the time period for each patient. (CON n = 109, TELE n = 60).

CI, confidence interval; INTER, all visits during the intervention period; PRE, 3 years before the intervention period.

MEAN ARTERIAL PRESSURE

A total of 9 TELE and 20 matched CON patients were referred to the nephrologist or cardiologist for uncontrolled hypertension. The PRE versus INTER MAP change within the CON group was significant, with mean MAP dropping from 112.5 (95% CI [105.8–119.2]) at PRE to 107.1 (95% CI [100.3–114.0]) during INTER (p = 0.016). The TELE group did not change significantly (going from a mean of 106.7 (95% CI [83.6–129.9]) at PRE to 105.4 (95% CI [82.9–127.8]) during INTER, p = 0.56). While the power to detect change was low in this group, the effect size was small and probably not clinically relevant. Comparing the PRE versus INTER change across groups in the mixed model, we found that the difference between groups in the slope of MAP over time was not significant (group x time interaction p = 0.26).

SERUM CREATININE

A total of 35 TELE and 70 matched CON patients were referred for kidney disease. The SCr outcome in the CON group did not change significantly from PRE to INTER periods (going from mean SCr of 1.58 (95% CI [1.33–1.82]) to 1.75 (95% CI [1.46–2.04]), p = 0.09. The TELE group also did not have significant change over time (with PRE mean of 1.57 (95% CI [1.36–1.78]), and INTER mean of 1.77 (95% CI [1.38–2.16]), p = 0.23). The group x time interaction was nonsignificant (p = 0.90), indicating that both groups had similar patterns over time.

ESTIMATED GLOMERULAR FILTRATION RATE

For eGFR, the CON group did not change significantly from PRE to INTER periods (with a mean of 61.6 mL/min/[95% CI (52.2–71.0)] PRE vs. 54.2 mL/min INTER [95% CI (43.6–64.7)], p = 0.09). The TELE group also had a nonsignificant reduction from 50.1 mL/min (95% CI [39.5–60.7]) PRE to 46.3 mL/min (95% CI [32.6–60.1]) INTER, p = 0.45, and the group by time interaction was nonsignificant (p = 0.56), indicating similar degrees of change over time between the groups.

Discussion

We present outcomes from providing specialty TM care at an urban FQHC just before the COVID-19 pandemic. The primary outcome of no-show rate was not significantly different between TELE and STAND visits. A general consensus exists in the literature that the missed appointment, or no-show, occurs due to a combination of patient and provider factors.^9,10 Identification of these factors may be used to target interventions to facilitate visit completion for this population. The evidence base for the effect of TM interventions on no-show rates has expanded during the pandemic, but the results are mixed.

During the pandemic, no-show rates for TM visits were low, however, these were likely impacted by local lock-downs, the need for social distancing and concerns over disease transmission during in person care. A study during this time found a significant reduction in no-show rates for TM appointments at a primary and specialty care clinic, a decrease from a prepandemic in-office rate of 29.8%–7.5%.¹¹ However, other studies, including a large cohort study have found no difference in no-show rate with the introduction of TM.^12,13 No-show rates are of special concern for AMCs and FQHCs since they have been found to be higher among low-income patients and Medicaid beneficiaries.¹⁴

It is anticipated that between 10% and 30% of outpatient visits will be completed via TM in the future. Therefore, the operational impact of TM remains important to understand as AMCs are re-evaluating investments in telehealth. In addition, our study is consistent with existing telehealth literature demonstrating that patient satisfaction with TM is high for convenience, effectiveness, and health outcomes. Despite high patient satisfaction, the no-show rate was not improved. Our follow-up survey on reasons for missed visits revealed numerous barriers, including the requirement of in-person visit for telehealth at the local FQHC.

Secondary outcomes of MAP, SCr, and eGFR did not show a significant difference when compared to a matched control group from a sister FQHC site with similar patient demographics. Encouragingly, TELE patients achieved an absolute reduction of A1c of 1.1% from standard care (TELE improved by 1.5% and CON improved by 0.4%), which was significant on a trend level. Previously published evaluations of TM interventions for diabetes abound, but telehealth definitions are broad.

Often studies include remote monitoring and education-based interventions with or without specialty consultation utilizing real-time audio and video resulting in meta-analyses reaching different conclusions on the efficacy of telehealth interventions.^15,16 In the management of kidney disease, our findings of a stable SCr and eGFR in both TM and in-person care settings are consistent with lack of disease progression during the study period. This finding is also consistent with previously published information from researchers at the Veterans Administration regarding the impact of telenephrology in an underserved population.¹⁷

TM interventions for the management of hypertension in urban underserved populations have demonstrated efficacy in at least two published studies on the topic, which our evaluation was not sufficiently powered to examine.¹⁸ The majority of the existing literature focuses on guided self-management of hypertension. Systematic reviews of TM for the management of hypertension have shown modest improvements based on low-quality evidence.

Despite large and growing urban populations, increasing racial and ethnic diversity in the United States, and persistent disparities in health outcomes, dedicated research on telehealth interventions for urban and predominantly black populations remains lacking. With the rapid expansion of telehealth during the COVID-19 pandemic, medical practices pivoted to near-exclusive use of TM. However, several cohort studies of outpatient and specialty clinic visit rates during the initial surge have observed a significantly decreased visit rate by patients with older age, non-English language preference, and Medicaid status.^2,12,19,20 Furthermore, in patients who complete TM visits, a decreased proportion of synchronous video use (vs. telephone use) is seen in those with older age, black race, Latinx ethnicity, and lower household income.¹⁹

TM program implementation may exacerbate existing disparities in access and outcomes. Therefore, it is crucial to carefully design and evaluate emerging telehealth interventions, especially in underserved and hard to reach populations. For example, while our study supports the efficacy of specialty consultation via TM for poorly controlled diabetes, hypertension, and kidney disease, it does not add insights to how shifting specialty consultation to the home via TM may affect its impact.

Inconsistent specialist scheduling at clinics, high patient no-show rates, and inadequate or sporadic specialty referral volumes prevent on-site specialty care from being a viable long-term solution for FQHCs and their patients.⁶ As a result, patients requiring specialty medical services are often referred to an outside specialist. An opportunity for well-designed telehealth intervention to expand access to care for underserved patients lies in the integration of telehealth with existing FQHCs practices.

Our model, with the FQHC as originating site, may address some of these challenges and expand access to care for underserved patients by using a designated space at a qualified facility where TM services can be received. Advantages of this model include colocating TM with existing facilities in medically underserved neighborhoods to reduce transportation barriers, increasing patient familiarity with the environment, engagement with culturally competent staff, as well as access to on-site equipment and laboratory testing.^21
–23

In addition, the model preserves the ability to perform a physical examination via the use of telepresenters. Staff, and even laypersons, may be trained to become telepresenters with a brief educational intervention.¹⁴ We believe that this model may be applied to other urban settings that are qualified originating sites for TM. Of course, there may be further improvement in outcomes and access when patients with health disparities have access to care via TM from home, as we found that access issues exist even to reach the community clinic. This emphasizes the need to continue to advocate for expanded qualified originating sites and telehealth payment parity.

Limitations of this study include a lack of sufficient STAND visit data for the subspecialties of cardiology, endocrinology, and nephrology in the TELE cohort during the study period (<10% of the STAND specialties), given most of their disease management was via TM. Furthermore, given the study design, clinical staff and study participants were unblinded and no randomization was performed. The occurrence of this study before the COVID-19 pandemic may limit the generalizability of this study, given limited knowledge of TM before the pandemic, which may have decreased engagement in tele-specialty consultation. It should also be noted that in this study, the FQHC was the originating site, while during the pandemic, home was the most common originating site for most TM services.

Overall, our data regarding no-show rates and biochemical parameters indicate that TM was equivalent with in-person care in this context; however, a randomized controlled trial may control for unknown confounding variables and provide more support for causality.

Conclusion

Improving access to care and clinical outcomes for urban underserved patients remains challenging. Telehealth remains a potential solution that requires rigorous evaluation as it may address some existing barriers to care while creating new ones such as internet and device access. Establishing TM from a FQHC in this study allowed patients to be seen by a specialist for care of diabetes, hypertension, and kidney disease, with similar no-show rates and high patient satisfaction. We demonstrated equivalent clinical outcomes for nephrology and cardiology and improved glycemic control when compared to standard endocrinology care. These outcomes are an important finding that supports efforts to make care for underserved populations via TM more widely available and accessible.

Footnotes

Acknowledgment

Keisha Herbert, James Betz, Nick Reed, Davin Combs, Kai Neander, Jen Puryear, Waddaa Redha, MD, and Jeff Jacobs, MD.

Authorship Confirmation Statement

All contributors agreed to submit this article for publication.

Disclosure Statement

G.V.B., S.Q.L., and N.S. received funding support from Care First Foundation to conduct the study. N.E. has received investigator-initiated grants from Dexcom and educational grants from Novo Nordisk and Merck. She is a consultant for Novo Nordisk. S.Q.L. consults for TrioMed.

Funding Information

This study was funded through a grant from CareFirst BlueCross BlueShield.

References

Westby

, Nissly

, Gieseker

, Timmins

, Justesen

. Achieving equity in telehealth: “Centering at the margins” in access, provision, and reimbursement. J Am Board Fam Med, 2021; 34(Suppl.):S29–S32.

Chunara

, Zhao

, Chen

, et al. Telemedicine and healthcare disparities: A cohort study in a large healthcare system in New York City during COVID-19. J Am Med Inform Assoc, 2021; 28:33–41.

Mitchell

, Chebli

, Ruggiero

, Muramatsu

. The digital divide in health-related technology use: The significance of race/ethnicity. Gerontologist, 2019; 59:6–14.

Nouri

, Khoong

, Lyles

, Karliner

. Addressing equity in telemedicine for chronic disease management during the Covid-19 pandemic. NEJM Catal Innov Care Deliv, 2020; 1, DOI: 10.1056/CAT.20.0123.

Health Center Program: Impact and Growth. Bureau of Primary Health Care. Available at https://bphc.hrsa.gov/about/healthcenterprogram/index.html (last accessed August 15, 2021 ).

Nakamura

, Laberge

, Davis

, Formoso

. Barriers and strategies for specialty care access through federally qualified health centers: A scoping review. J Health Care Poor Underserved, 2019; 39:910–933.

Experiences of Medicaid Programs and Health Centers in Implementing Telehealth—PubMed. Available at https://pubmed.ncbi.nlm.nih.gov/32582468/ (last accessed August 15, 2021 ).

Koonin

, Hoots

, Tsang

, et al. Trends in the use of telehealth during the emergence of the COVID-19 pandemic—United States, January–March 2020. Morb Mortal Wkly Rep, 2020; 69:1595–1599.

Rosenbaum

, Mieloszyk

, Hall

, Hippe

, Gunn

, Bhargava

. Understanding why patients no-show: Observations of 2.9 million outpatient imaging visits over 16 years. J Am Coll Radiol, 2018; 15:944–950.

10.

Miller

, Chae

, Peterson

, Ko

. Predictors of repeated “no-showing” to clinic appointments. Am J Otolaryngol, 2015; 36:411–414.

11.

Drerup

, Espenschied

, Wiedemer

, Hamilton

. Reduced no-show rates and sustained patient satisfaction of telehealth during the COVID-19 pandemic. Telemed J E Health, 2021; 27:1409–1415.

12.

Franciosi

, Tan

, Kassamali

, et al. The impact of telehealth implementation on underserved populations and no-show rates by medical specialty during the COVID-19 pandemic. Telemed J E-Health, 2021; 27:874–880.

13.

Siow

, Walker

, Britt

, et al. What was the change in telehealth usage and proportion of no-show visits for an orthopaedic trauma clinic during the COVID-19 pandemic?. Clin Orthop, 2020; 478:2257–2263.

14.

Dantas

, Fleck

, Cyrino Oliveira

, Hamacher

. No-shows in appointment scheduling—A systematic literature review. Health Policy, 2018; 122:412–421.

15.

Flodgren

, Rachas

, Farmer

, Inzitari

, Shepperd

. Interactive telemedicine: Effects on professional practice and health care outcomes. Cochrane Database Syst Rev, 2015; 2015:CD002098.

16.

Zhai

, Zhu

, Cai

, Sun

, Zhao

. Clinical- and cost-effectiveness of telemedicine in type 2 diabetes mellitus: A systematic review and meta-analysis. Medicine (Baltimore), 2014; 93:e312.

17.

Ladino

, Wiley

, Schulman

, et al. Tele-nephrology: A feasible way to improve access to care for patients with kidney disease who reside in underserved areas. Telemed J E Health, 2016; 22:650–654.

18.

Bove

, Homko

, Santamore

, et al. Managing hypertension in urban underserved subjects using telemedicine—A clinical trial. Am Heart J, 2013; 154:615–621.

19.

Eberly

, Kallan

, Julien

, et al. Patient characteristics associated with telemedicine access for primary and specialty ambulatory care during the COVID-19 pandemic. JAMA Netw Open, 2020; 3:e2031640.

20.

Eberly

, Khatana

SAM

, Nathan

, et al. Telemedicine outpatient cardiovascular care during the COVID-19 pandemic: Bridging or opening the digital divide?. Circulation, 2020; 142:510–512.

21.

Shin

, Sharac

, Jacobs

. Provision of telemedicine services by community health centers. Online J Public Health Inform, 2014; 6:e185.

22.

Arora

, Kalishman

, Dion

, et al. Partnering urban academic medical centers and rural primary care clinicians to provide complex chronic disease care. Health Aff (Millwood), 2011; 30:1176–1184.

23.

Marcin

, Shaikh

, Steinhorn

. Addressing health disparities in rural communities using telehealth. Pediatr Res, 2016; 79:169–176.