An Observational Study of Disparities in Telemedicine Utilization in Primary Care Patients Before and During the COVID-19 Pandemic

Introduction

The COVID-19 pandemic rapidly changed the delivery of health care in the United States. In the first weeks of the pandemic, many health care activities were prohibited from continuing in-person. In response, hospitals and payers adjusted delivery models and reimbursement structures to accommodate telemedicine visits, defined as the remote delivery of care through video and telephone calls, the number of these visits grew exponentially and to date remains much higher than prepandemic levels.^1,2

The digital divide in medicine and the social determinants that contribute to it have been well described,^3,4 and inevitable concerns have been raised that the rapid telemedicine scale-up during the pandemic would further disadvantage vulnerable patient groups.^5–7 Some studies suggest that this has occurred, showing increased utilization of telemedicine by younger patients (55 years or less) compared with older, and greater utilization by patients with commercial insurance.⁸ However, other reports suggest that the picture is not clear cut, with variable impact in different patient groups.

A study of commercially insured patients showed that utilization of telemedicine during the pandemic increased more in urban patients with higher incomes, compared with insured lower income patients in rural areas.⁵ Another study from New York at the peak of the pandemic showed that black patients and patients with less than average income, were less likely to seek help through telemedicine than other groups. However, the same study showed that black patients and patients with less than average income increased their overall telemedicine use compared with prepandemic levels.⁹ In contrast, a study reviewing the U.S. National Disease and Therapeutic Index audit showed regional variation in adoption of telemedicine in response to the pandemic, but with black and white patients adopting at equal rates.⁸ Therefore, there remains a need for more work to understand how best to ensure health care access using all potential routes for patients, particularly for groups affected by disparities highlighted by differences in outcome during the pandemic.¹⁰

Before the pandemic, patient utilization of telemedicine was limited not only by patient issues such as access to the Internet and digital devices, but importantly also to provider issues such as a lack of providers offering telemedicine visits. Some of the substantial provider hurdles to the widespread delivery of telemedicine included staff reticence and reimbursement issues.^6,7 Health systems lacked the incentives to adopt the required infrastructure for providing telemedicine visits, including videoconferencing platforms, additional digital devices, and employee and patient training.⁷ Telemedicine provision was uncommon in urban academic medical centers.²

The rapid scale-up of telemedicine during the COVID-19 pandemic has provided an opportunity to challenge some of the assumptions about patient adoption and to examine health care disparities in telemedicine utilization in settings where telemedicine had not previously been practiced. As the U.S. Department of Health and Human Services embraced telemedicine, including video and phone visits across many platforms, unforeseen disparities between modes of delivery may have been realized. Data from the Pew Research Center shows differences among phone and computer ownership in the United States: 96% of Americans own a cellphone and 81% own a smartphone, but only 74% own a desktop or laptop computer.¹¹ These differences could potentially exacerbate health disparities through telemedicine delivery during COVID-19.

As the COVID-19 pandemic has continued and health systems grapple with resulting changes in how to deliver effective medical care differently in the long term, it is important to understand how best to reach all patients in need of primary care. Given the disproportionate toll that COVID-19 has had on underserved populations, it is imperative that primary care and preventive medicine delivery systems evolve delivery modes to effectively reach all patients equitably.^10,12

The purpose of this study was to compare patient demographics and type of primary care visit before the onset of the pandemic with demographics and type of primary care visit after the rapid scale-up of telemedicine. We sought to test the hypothesis that older patients and those with non-English preference, from racial minority backgrounds and lower income areas would have fewer telemedicine visits due to anticipated structural barriers in accessing and completing telemedicine visits.

Methods

The study was based at Keck Medical Center of the University of Southern California (USC), an academic medical center situated in East Los Angeles with a large local Hispanic population comprising nearly one third of our patients and a large percentage of Medicare and Medicaid/MediCal patients (data captured by our local system data and census data).^13,14

Patients scheduled for a primary care visit with a clinician in either Internal Medicine or Family Medicine departments between April 1 through May 31, 2019 and April 1 through May 31, 2020 were included in the study. The period of April 1–May 31 was selected for each year to account for the unstable use of telemedicine at the start of the pandemic in March 2020. Most scheduled visits in 2019 were in-person appointments, but visits in 2020 included both in-person and telemedicine visits. Patients who lived outside of California were excluded.

The electronic health record (EHR) was interrogated to extract data for consecutive patients who were scheduled for, and received, primary care at Keck Medicine of USC during the dates of interest. All identifiable patient information was masked or redacted before analysis. Variables of interest included age (child, <18 years; adult, 18–64 years; older adult, ≥65 years), gender, race, ethnicity, preferred language, visit type (televideo, telephone, in-person), visit status (completed, rescheduled, no show, canceled), insurance type, and zip code. Televideo visits were defined as those which utilized audio and video technology (including Zoom, Microsoft Teams, In-Touch, etc.). Zip codes were collected to approximate average income, and socioeconomic status by proxy.

The average income was calculated for California zip codes using the most recent available IRS income and tax data (from 2018).¹⁵ Total reported income was divided by the total number of returns for each zip code, providing average income per zip code. Income categories were divided into quartiles for further analysis.¹⁶ Insurance types were classified into three categories: commercial, including nongovernment, employer-provided health insurance plans; government, including Medicare, Medicaid, dual eligibility, and veterans’ insurance; and uninsured, including self-pay, uninsured patients, hospital donor funds, and charity funds.

Statistical tests with a p-value of <0.05 were considered significant. SAS version 9.4 (SAS Institute, Inc., Cary, NC) was used for data analysis. To determine whether there was a difference in populations by year (2019 vs. 2020), continuous variables (age in years and average income) were compared using Wilcoxon–Mann–Whitney tests and chi-squared tests were used to compare categorical variables.

Differences in patient characteristics between televideo, telephone, and in-person visits were also compared using Kruskal–Wallis tests (continuous variables) and chi-squared or Fisher’s exact tests (categorical variables). We focused further analysis on patients who completed visits, both by year and visit type. Logistic regression determined odds ratios (ORs) for patients who completed each visit type by age category, gender, income category, ethnicity, race, preferred language, and insurance type.

This was a retrospective analysis of patient demographics and visit type, and was reviewed and declared exempt by the Institutional Review Board at USC (UP-20-00661). The study was performed and the article reported in line with STROBE guidelines for observational studies.¹⁷

Results

The initial dataset included 58,202 scheduled visits by 19,376 unique patients. There were 7,893 patients scheduled for a visit in 2019, which rose to 11,483 in 2020, an increase of 45.5% in booked visits. These bookings did not all translate into completed visits.

There was a much smaller increase of 5.2% in the number of patients who completed one or more primary care visits between April 1 and May 31, 2020, compared with April 1 and May 31, 2019 (n = 4,986, n = 4,742 respectively). This increase in patient counts occurred despite the ongoing COVID-19 pandemic, which caused a substantial shift in the type of primary care visits, from in-person in 2019 to virtual (telephone or video) in 2020 (100% in-person 2019, 83% virtual 2020). Of the total primary care visits completed during the early stages of the COVID-19 pandemic, 17% occurred in person, 60% occurred by video, and 23% occurred by telephone.

When comparing characteristics for patients with a scheduled visit during April 1–May 31, 2019 to the same timeframe in 2020, there was a statistically significant difference in continuous age, continuous average income, categorical age, gender, income category, ethnicity, race, preferred language, appointment type, and appointment status (Table 1). There was also a statistically significant difference in these characteristics between the different visit types in 2020 (Table 2).

Further analysis of completed visits in 2020 (Table 3) revealed that the OR of an older adult (65 years of age or more) having a telephone visit was more than two and a half times greater than a younger adult (OR = 2.8, confidence interval [95% CI] = 2.5–3.2), whereas the OR of an older adult completing a televideo visit was half that of a younger adult (OR = 0.5, 95% CI = 0.7–0.8). In addition, as the average income category increased, the OR of completing a televideo visit increased, whereas the OR of completing a telephone visit decreased.

Patients of Hispanic/Latino ethnicity (OR = 1.7, 95% CI = 1.4–2.1) and racial minority (Black, OR = 1.5, 95% CI = 1.0–2.2; and Asian patients, OR = 1.4, 95% CI = 1.1–1.9) had higher OR of completing a telephone visit when compared with non-Hispanic and white patients, but lower OR of completing a televideo visit (OR = 0.7, 95% CI = 0.6–0.9; OR = 0.6, 95% CI = 04.–0.9; and OR = 0.7, 95% CI = 0.5–0.9, respectively).

This same theme was evident when examining preferred language. The OR of a patient with non-English language preference having a telephone visit was more than three times that of patients whose preferred language was English (Spanish preference: OR = 3.7, 95% CI = 2.9–4.7; other language preference: OR = 3.6, 95% CI = 2.3–5.4), whereas the OR of completing a televideo visit was less than half that of patients with English preference (OR = 0.4, 95% CI = 0.3–0.5; OR = 0.4 95% CI = 0.3–0.7, respectively).

Discussion

The COVID-19 pandemic served as an impetus for health systems to change the way they delivered primary care. Protecting both patients and providers against virus transmission required a swift pivot to deliver an increasing percentage of visits virtually. The sudden shift to virtual care, which was heavily dependent on patient and provider comfort with technology and the availability of reliable broadband Internet may have left, and may continue to leave, some patients out.

Our initial hypothesis proposed that older patients with non-English-speaking preference from a racial minority background and lower income areas would have fewer telemedicine visits due to expected structural barriers in accessing and completing telemedicine visits. However, the data from our sample indicate that these patients were able to manage telemedicine visits. As noted similarly in another recent study, the differences in patient characteristics surfaced most when the modality of the telemedicine visit, video versus telephone, was examined.¹⁸

The finding that patient counts increased during 2020 compared with 2019, with ∼83% of patients completing telemedicine visits, demonstrates the ability of telemedicine to scale up and reach similar or greater patient volumes than in-person visits before the pandemic. Although the overall visit volume did increase, there was also a high cancellation rate, which may be influenced by many factors at the start of the pandemic but could also reflect less patient commitment to a telemedicine appointment rather than to in-person contact.

Our findings contrast with a very large study reviewing 125.8 million primary care visits nationally across 10 calendar quarters between January 2018 and June 2020, which found a decrease in quarter 2 of 2020 of 21% in total primary care visits compared with the average of 2018 and 2019. During the first and second quarters of 2020 there was regional variation in telemedicine utilization across the United States with the highest utilization in the Pacific region (Washington, Oregon, California), where there were 40 million primary care but only 26.8% of those visits delivered through telemedicine.⁸

Adoption of telemedicine within primary care at our institution overwhelmingly exceeded the findings of that study; for a comparable period in our organization 83% of completed primary care visits were delivered through telemedicine. This may be due in part to our status as an academic medical center, which had a small telemedicine team already in place before the pandemic and as we have reported previously, our scale-up in response to the pandemic was extremely rapid.²

The difference in our primary care patient demographics between 2019 and 2020 demonstrates the ability of telemedicine to reach older patients of lower income, higher Spanish-speaking preference, and with government-provided health insurance plans. This suggests the potential for continued use of telemedicine beyond the COVID-19 pandemic to help reduce health disparities among patients who have been previously disadvantaged in accessing medical care due to their income, language, or lack of private health insurance coverage.¹⁹ Because patients who were older adults of lower income and non-English language preference were more likely to complete a telephone over a video visit, these populations will be disadvantaged by any potential limitations in care provided by telephone compared with video.²⁰

A limitation of this study is that it was performed in one academic medical center over two short windows of time and restricted to primary care. Although there have been similar studies performed at other centers in the United States, including primary care and specialty care with broadly similar findings, the population base in our study (over one third Hispanic/Latino) provides further data to understand how best to deliver telemedicine to diverse populations with various needs.^21,22

It must also be considered that this study relied solely on what was entered into our EHR. Our data show that there were zero telephone visits in 2019. It is likely that phone calls did take place but were not recorded. Changes in the reimbursement structure in 2020 made it beneficial to record telephone communication with patients. Using the EHR, there is no way to confirm whether the number of calls in 2019 is correct.

While our results on utilization according to race, ethnicity, and language reflect statistical changes for patients where these variables were recorded, in 2020 there was an increase in the number of patients where race and ethnicity was not captured in the EHR in marked contrast to insurance status, which was captured for 99% of patients in both years.

In 2019, 26.6% of race and 32% of ethnicity were not captured, these percentages are in line with those reported from large U.S. databases and are better than recorded in other U.S. academic medical centers.²³ However, with the switch to telemedicine care, the percentage of race and ethnicity not captured in our organization rose to 52.5%, and 54.6%, respectively. Studies show that patients are more willing to record race and ethnicity using patient facing tools than when asked directly,²³ this provides an area for further monitoring and study to ensure these important data are collected effectively during virtual visits.

Further research is recommended to determine the extent to which there is a difference in quality of health care provided with telephone versus video. Recent studies indicate that video visits are longer in duration, allow for better observation of patient responses and living environment, and facilitate more provider diagnoses compared with telephone visits.^20,24 Additional analysis in the primary care setting will be valuable, especially considering the likely permanence of telemedicine beyond the COVID-19 pandemic. If televideo does indeed allow for higher quality primary care than telephone visits, efforts should be made to improve the ability of all patients to access and utilize televideo.

However, the potential benefit of increased access for all patients, including through telephone calls should not be dismissed, and again research is needed to establish what type of connection is as effective through a telephone call. Our study of an urban multiethnic population with a large percentage of patients without commercial insurance exceeded regional and national averages for delivery of total primary care patient visits and telemedicine utilization during the COVID-19 pandemic, suggesting that future studies should focus on telemedicine as a means to minimize health care disparities in the delivery of primary care.

Conclusions

This study demonstrates an increase in primary care visits from early 2019 to 2020. Most visits moved from in-person to telemedicine and there were increased numbers of older patients, patients of lower income, and non-English language preference who accessed primary care. While it is reassuring that the rates of telemedicine versus in-person care were similar for minority groups, the difference in utilization of the telemedicine modality, video versus phone in older patients of lower income, and non-English preference groups suggests there is risk of telemedicine furthering health care disparities for care that cannot be addressed using telephone visits alone. Further research is needed to understand the benefits of televideo compared with telephone, and to better understand how best to improve video access to all patients.

Authors’ Contributions

All authors contributed to the writing and review of this article.

Disclosure Statement

V.M.P. and K.S.M. have nothing to disclose. C.J.P. receives consultancy fees from the Institute for Healthcare Improvement and the American College of Surgeons.

Funding Information

No funding was received for this article.