Impact of Telemedicine Modalities on Equitable Access to Ambulatory Gastroenterology Care

The coronavirus disease 2019 (COVID-19) pandemic led to widespread adoption of telemedicine for ambulatory care, including videoconferencing (VV) or telephone (TV) visits. In light of this public health emergency (PHE), insurance carriers expanded their telemedicine coverage and the Centers for Medicare & Medicaid Services (CMS) established Waiver-1135 to extend telemedicine in Medicare to routine healthcare.¹ A more recent Blanket Waiver also broadened reimbursement of telehealth services across additional healthcare fields.¹

While telemedicine may increase access to care among certain patient populations, it can simultaneously create barriers to access among others.^{2, 3} Consequently, telehealth modalities such as VV may create a digital divide and exacerbate existing healthcare disparities among vulnerable populations including Black, Latino/a/x, low socioeconomic status (SES), and older individuals.^{2, 3} However, there is a paucity of literature evaluating the uptake of telemedicine through TV versus VV by race/ethnicity, insurance status and age.² In addition, telehealth may affect subspeciality care differently compared to primary care. The expanded telemedicine services during this PHE provide a unique opportunity to systemically assess differential telehealth uptake across diverse populations.

Considering the role that telehealth will likely play in the future of healthcare delivery, it is critical that we understand the potential impact that this paradigm shift may have among vulnerable populations. Therefore, we aimed to examine the patient characteristics associated with completion of in-person (IPV) and telemedicine visits in a high-volume gastroenterology (GI) clinic.

METHODS

We analyzed all ambulatory GI clinic visits at a large, tertiary care center in Massachusetts between 4/1/2020 and 5/15/2020, with visits from the same period in 2019 serving as controls. An electronic healthcare record (EHR) database query was conducted for all completed GI clinic visits, which were classified into IPV and telemedicine visits. All telemedicine visits were manually reviewed to further sub-classify into those completed as VV or TV visits. This was performed by reviewing the clinic and billing reports for each visit individually. Our institution required inclusion of a standardized statement in each clinic report and a supplementary billing code to identify the visit modality. Variables assessed included race/ethnicity, age, sex, median income by home address zip code, insurance (private, public, public with supplement), and type of patient appointment (new vs return visits). (Supplementary Methods)

RESULTS

There were 6120 completed GI clinic visits during the study periods (3589 from 2019 and 2531 from 2020). All visits from 2019 were IPV; while 9 from 2020 were IPV and were excluded. In total, 2522 telemedicine visits were completed (958 VV and 1564 TV). The mean age of all included patients was 52.0±17.8 years and 4060 (66.4%) were women. Overall, 4626 (75.7%) patients identified as White, 433 (7.09%) as Black, 573 (9.38%) as Latino/a/x, and 200 (3.28%) as Asian. The average median income by zip code was $76372.2±26577. All patients had health insurance coverage, with 56.8% private, 20.0% public, and 23.2% public with private supplement. Comparing IPV from 2019 and all telemedicine visits (VV and TV) in 2020, there were no significant differences in mean age, racial/ethnic distribution, median income by zip code, insurance, or appointment type (new vs return).

Telemedicine versus Pre-Telemedicine: VV versus IPV

On univariate analyses, the VV cohort was significantly younger than the IPV group (46.0±17.0 vs 53.1±17.8 years, p<0.0001), with less patients >60 years (25.7% vs 39.8%, p<0.0001). It also had a lower proportion of Black (3.48% vs 7.02%, p<0.0001) and Latino/a/x (3.48% vs 9.92%, p<0.0001) patients, more private commercial insurance coverage (74.1% vs 54.6%, p=0.0001), and higher income by zip code ($75,850 vs $72,292, p<0.0001) compared to IPV. The percentage of new patient visits was also lower among VV versus IPV (22.6% vs 29.1%, p<0.0001). (Table 1)

Table 1:

Characteristics of all outpatient gastroenterology clinic visits between April 1^st and May 15^th in 2019 and 2020.

	2019	2020
	In-Person n=3589	Video Visit n=958	p-value*	Telephone Visit n=1564	p-value*	p-value**
Age
Mean±SD (years)	53.1±17.8	46.0±17.0	<0.0001	53.5±17.6	0.414	<0.0001
Groups, n (%)
<40	1069 (29.8)	409 (42.7)	<0.0001	416 (26.6)	0.046	<0.0001
40–60	1092 (30.4)	303 (31.6)		480 (30.7)
>60	1428 (39.8)	246 (25.7)		668 (42.7)
Male, n (%)	1226 (34.2)	323 (33.7)	0.797	502 (32.1)	0.149	0.400
Race/Ethnicity, n (%)
White	2699 (76.1)	830 (87.5)	<0.0001	1097 (70.9)	0.0002	<0.0001
Black	249 (7.02)	33 (3.48)		151 (9.76)
Latino/a/x	352 (9.92)	33 (3.48)		188 (12.2)
Other	247 (6.96)	53 (5.58)		111 (7.18)
Median Income by Zip Code
Median	72292	75850	<0.0001	70446	0.016	<0.0001
Quartiles, n (%)
4th (>75th percentile)	879 (24.5)	290 (30.3)	<0.0001	341 (21.8)	0.126
3rd (50th–75th percentile)	863 (24.1)	251 (26.2)		398 (25.5)
2nd (25th–50th percentile)	888 (24.7)	236 (24.6)		378 (24.2)
1st (<25th percentile)	959 (26.7)	181 (18.9)		447 (28.6)
Insurance, n (%)
Private Insurance	1958 (54.6)	710 (74.1)	<0.0001	796 (50.9)	0.040	<0.0001
Public Insurance	760 (21.2)	98 (10.2)		366 (23.4)
Public with Private Supplement	869 (24.2)	150 (15.7)		402 (25.7)
Appointment Type, n (%)
New	1044 (29.1)	216 (22.6)	<0.0001	142 (9.08)	<0.0001	<0.0001
Return	2545 (70.9)	742 (77.5)		1422 (90.9)

^*Compared to in-person visits in 2019

^**Telephone Visit versus Video Visit

On multivariable analysis, Black [adjusted odds ratio (aOR)=0.56; 95%CI:0.38–0.82, p=0.039] and Latino/a/x (aOR=0.43; 95%CI:0.29–0.63, p=0.0009) patients, living in zip codes in the lowest quartile of income (aOR=0.72; 95%CI:0.58–0.90, p=0.017), age >60 years (aOR=0.53; 95%CI:0.43–0.65, p<0.0001), use of public insurance only (aOR=0.51; 95%CI:0.40–0.65, p<0.0001) or public insurance with private supplement (aOR=0.64, 95%CI:0.51–0.80, p=0.035), and new patient appointment (aOR=0.72; 95%CI:0.60–0.85, p=0.0002) were independently associated with lower odds of completing VV compared to IPV. (Supplementary Table 1a)

Telemedicine versus Pre-Telemedicine: TV versus IPV

Compared to IPV, TV was associated with more patients age >60 years (42.7% vs 39.8%, p<0.046), higher proportions of Black (9.76% vs 7.02%, p=0.0002) and Latino/a/x (12.2% vs 9.92%, p=0.0002) patients, and lower income by zip code ($70,466 vs $72,292, p=0.016). The rates of private commercial insurance coverage (50.9% vs 54.6%, p=0.04) and new patient visits (9.08% vs 29.1%, p<0.0001) were also lower in the TV cohort compared to IPV. (Table 1)

On multivariable logistic regression, Black (aOR=1.53; 95%CI:1.21–1.94, p=0.016) and Latino/a/x (aOR=1.32; 95%CI:1.07–1.66, p=0.034) patients independently correlated with higher odds of engaging in TV versus IPV, while new patient appointment (aOR=0.24; 95%CI:0.20–0.29, p<0.0001) was a negative predictor. (Supplementary Table 1b)

Subgroup Analysis: Telemedicine - VV versus TV

On subgroup analyses of only telemedicine patients in 2020, TV patients had higher mean age (53.5±17.6 vs 46.0±17.0 years, p<0.0001) and were more likely to be Black (9.8% vs 3.5%, p<0.0001) and Latino/a/x (12.2% vs 3.5%, p<0.0001) compared to VV patients (Figure 1a). Notably, over 80% of Black and Latino/a/x patients completed their telemedicine appointments using TV, compared to about half of White patients (Figure 1b). The average median income by zip code ($70446 vs $75850, p<0.0001) and proportion of patients with exclusively private commercial insurance (50.9% vs 74.1%, p<0.0001) or presenting for new consultation (9.1% vs 22.6%, p<0.0001) were also significantly lower in the TV group. (Table 1)

Figure 1:

(A) Race/ethnicity composition of gastroenterology ambulatory clinic patients in the 2019 cohort (in-person visits), 2020 videoconferencing visit group, and 2020 telephone visit group. The proportion of White patients completing video virtual visits was significantly higher than those of in-person visits and telephone visits. (B) The proportion of telephone versus video visits among all gastroenterology telehealth visits by race/ethnicity. Significantly higher proportions of Black and Latino/a/x patients completed telephone visits compared to White patients.

On adjusted multivariable model, Black (aOR=2.95; 95%CI:1.94–4.48, p=0.015) and Latino/a/x (aOR=3.12; 95%CI:2.07–4.71, p=0.005) patients, lowest quartile of income by zip code (aOR=1.44; 95%CI:1.11–1.88, p=0.054), age >60 years (aOR=2.05; 95%CI:1.61–2.63, p<0.0001), public insurance only (aOR=1.96; 95%CI:1.49–2.56, p=0.002), public insurance with supplement (aOR=1.62, 95%CI:1.24–2.11, p=0.028), and new patient appointments (aOR=0.38; 95%CI:0.30–0.49, p<0.001) were independent predictors for engaging in TV compared to VV. (Supplementary Table 2)

DISCUSSION

The COVID-19 pandemic has led to a rapid paradigm shift towards telehealth. Prior to this PHE, there was limited coverage for telemedicine practices, with only 0.1% of Medicare primary care visits being conducted through telehealth services before March 2020 compared to nearly half of them in April 2020.⁴ Similar widespread adoption of telemedicine has been reported in subspecialty practices. While telehealth may increase access to ambulatory care, it may also create barriers for vulnerable populations and worsen existing healthcare disparities. In this study of ambulatory visits at a high-volume GI clinic during the initial phase of the COVID-19 pandemic, we found that Black and Latino/a/x patients, lower SES, non-commercial insurance coverage, and older age were independent risk factors for lower engagement in VV compared to TV or to IPV from 2019.

Several potential barriers to telemedicine may contribute to the disparate uptake among various patient groups, including infrastructural or implementation factors and patient-related considerations. Access to electronic devices, availability of reliable internet connection, and usability of digital tools comprise the former category, while the latter may include patient health and digital technology literacy, language preferences, educational level, cultural considerations, and personal attitudes.^{5, 6} These barriers may disproportionality affect underserved racial/ethnic groups including Black and Latino/a/x populations, individuals with decreased educational attainment or limited English proficiency (LEP), lower SES and older adults.^{5, 6} It is therefore crucial to understand that adoption of technology-based solutions such as telemedicine can widen these existing healthcare inequities.⁶

Telehealth in Black and Latino/a/x Populations

Prior studies have shown that Black and Latino/a/x households have lower rates of home broadband internet access compared to White populations, while smartphone ownership is comparable.⁷ Black and Latino/a/x patients also utilize smartphones to access healthcare information more often.⁷ One study conducted during the COVID-19 pandemic found that Black patients reported higher odds of engaging in telehealth services such as secure messaging compared to White populations.⁸ However, factors beyond smartphone accessibility may affect engagement with digital health tools among Black and Latino/a/x communities. For instance, ownership and usage of smartphones is nuanced by access to affordable WiFi, speed of internet service or adequate data plans. It is imperative that these barriers be considered when offering telehealth options to ensure equitable access to care.

In our study, Latino/a/x patients comprised the second largest racial/ethnic population seen by our providers. In Massachusetts, approximately 9% of the population has LEP, with the majority Spanish-speaking, while 41% of all Spanish speakers report LEP.⁹ Individuals with LEP have been shown to be less likely to access their patient portals.¹⁰ The use of trained medical interpreters in LEP populations improves the quality of care and patient satisfaction,¹¹ and video/telephone medical interpretation is non-inferior to in-person interpretation.¹² While several three-way telephone interpretation services are available that allow patients, providers and medical interpreters to be on the same call, the use of medical interpreters in VV is often more challenging and costly, and many healthcare organizations have not integrated this technology into their telemedicine workflow. Our results may be explained, in part, by the higher likelihood of necessitating a medical interpreter in Latino/a/x populations and providers preferencing TV for easier integration of interpreter services.

Among Black and Latino/a/x communities, a history of structural racism has led to founded medical mistrust and fears concerning the utilization of personal medical information, which may also affect the uptake of digital health tools.¹³ Moreover, healthcare tools are generally not designed for underserved racial/ethnic populations including Black and Latino/a/x groups.¹⁴ It is imperative that investments be made in infrastructures that allow for the development of telehealth options specifically designed to serve Black, Latino/a/x, LEP and low literacy populations.¹⁴ This may include healthcare tools created in languages most predominantly spoken by local patient populations. Patient navigation has also been shown to help decrease healthcare disparities across multiple cancer screening initiatives and can potentially be leveraged to increase telehealth literacy.¹⁵ Investment in patient navigation services to guide patients through the telehealth continuum may help instill trust in the technology, improve usability, and reduce healthcare disparities.

Telehealth and Socioeconomic Status

Our results also found that patients from lower income neighborhoods are more likely to engage in telephone-based visit modalities, independent of race/ethnicity, age, and visit types. Individuals with lower income level may have less technology adoption, including smartphones.¹⁶ Even among low income families with access to these technologies, they may worry about out-of-pocket expenses such as home broadband or smart phone data use charges.¹⁷ While our study was unable to specifically assess the impact of education level on telemedicine use on a patient level, as this information was not consistently recorded, the average educational attainment on a population level by zip codes generally parallels that of income. Moreover, close to half of individuals with a high school degree or less reported internet non-adoption¹⁸ and individuals with less than vocational/some college education are less likely to access their EHR.¹⁰ With the increasing adoption of telehealth, institutions and providers should recognize the potential economic burden on and limited resources of patients with lower socioeconomic backgrounds.

Telehealth in Older Adult Populations

Some studies have shown that telehealth may help overcome barriers to healthcare access for older adults, such as limited transportation and geographic isolation.¹⁹ However, our study, similar to some recent reports, found that older patients are more likely to complete TV than VV despite availability of both.^{20, 21} This preference may represent patient-perceived challenges to VV use, including lack of technological confidence.²² Furthermore, adults >65 years-old represent the largest group of internet non-users.¹⁸ Despite these challenges, telehealth use in older adults may be increased through targeted interventions. In a pilot study of older patients who received telehealth patient navigation and a technological needs assessment, all 32 participants successfully completed VV.²² Given the increasing proportion of elderly patients, institutions and providers should ensure availability of various telehealth options and provide resources to navigate technological challenges.

Telehealth in GI and Hepatology Care

Prior to the COVID-19 pandemic, several studies have demonstrated telemedicine to be an effective mode of care delivery in GI, including for services such as nutrition counseling.²³ However, telemedicine uptake within GI continued to lag behind other medical subspecialties before 2020.²³ Among inflammatory bowel disease patients, data suggest that both TV and VV may be appropriate for routine follow-up care during remission, while patients prefer IPV during flares.²¹ Studies of patients with chronic liver disease (CLD) showed a predilection for use of VV over TV.²⁴ This may be partially due to the benefit of visual assessments in identifying clinical signs of decompensation. However, our study shows that VV use is lower among Black and Latino/a/x patients, groups that are disproportionately affected by CLD. More research is needed to assess the impact of telehealth on long-term clinical outcomes in various chronic GI/hepatology conditions, and to identify strategies to increase adoption among vulnerable populations.

In our study, there were significantly lower proportions of new patient encounters with the use of telemedicine during the PHE. This may have resulted from patients deferring elective new appointments for chronic symptoms. Alternatively, telemedicine may be more acceptable by patients for established follow-up care, rather than initial consultations. As such, telemedicine should be viewed as an adjunct to, rather than replacement of, IPV.

Telehealth Options and Reimbursement

The CMS traditionally defined telemedicine as a face-to-face interaction that utilized combined audio-video technology.²⁵ While this definition of telemedicine did not specifically include audio-only services, the CMS expanded their coverage guidelines during the PHE to include TV. However, it remains unclear whether all telemedicine visit modalities will continue to be reimbursed after the current PHE. If coverage for telemedicine reverts to excluding TV, institutions and providers may be disincentivized to offer audio-only options for telehealth. Our results suggest that if only VV is offered as a telemedicine option and barriers to VV use cannot be mitigated, underserved and vulnerable populations would be disproportionately affected. However, if equally reimbursed, telemedicine (both TV and VV) has the potential to improve equitable healthcare access beyond this PHE.

Our analysis has several strengths. First, our hospital system has a standardized and validated methodology for collecting patients’ self-reported race/ethnicity at the time of registration. Secondly, all patient data for race/ethnicity, insurance type and visit modality completed were manually reviewed to minimize misclassification. Lastly, our study included data for nearly all patients with ambulatory encounters at our GI clinic during the study periods, thereby reducing potential selection bias. There are also several limitations. First, home address zip code was utilized as a surrogate marker for patient income, as individual income data was not recorded in the EHR. Secondly, our study was limited to a single-site design at an academic tertiary medical care center in Massachusetts, a state with near universal health insurance coverage after its healthcare reform in 2006. This may limit our study’s generalizability to safety-net hospitals or community health centers primarily serving underserved racial/ethnic populations including Black and Latino/a/x groups and uninsured/underinsured patients. However, any potential bias resulting from our cohort would be toward less disparate uptake of telemedicine modalities, given the available resources and health coverage. The significant disparity in VV versus TV utilization found in our population despite available resources and high insurance rates further supports the impact of other social factors on the uptake of different telehealth options beyond health coverage alone. Importantly, our results highlighted the need for equal coverage and offerings across telehealth options to avoid disenfranchising already vulnerable populations. Such a need would be even more critical for safety-net and community centers serving a high proportion of underserved patients.

In summary, we found that Black and Latino/a/x patients, low SES, non-private commercial insurance, and older age independently predicted lower odds of engaging in VV compared to TV or IPV. Factors such as digital health literacy, resource availability, reliable internet access, and systemic mistrust by vulnerable populations should be considered when employing telemedicine technology to decrease healthcare inequities. Healthcare stakeholders should ensure equal availability, coverage, and reimbursement across telehealth options to prevent disincentivizing certain modalities and potentially worsening healthcare disparities. Patients from vulnerable groups may need additional support, including needs/barriers assessments, culturally congruent patient navigators, and development of culturally/linguistically sensitive technology with high usability for older, Black, Latino/a/x, low SES, and LEP populations. Blanket adoption of technology in healthcare would not address population specific barriers and could further widen existing healthcare disparities.