Organizational and External Factors Associated with Video Telehealth Use in the Veterans Health Administration Before and During the COVID-19 Pandemic

Abstract

Objectives:

To identify organizational and external factors associated with medical center video telehealth uptake (i.e., the proportion of patients using telemedicine) before and early in the coronavirus disease 2019 (COVID-19) pandemic.

Materials and Methods

: We conducted a retrospective, observational study using cross-sectional data for all 139 U.S. Veterans Affairs Medical Centers (VAMCs). We used logistic regression analyses to identify factors that predicted whether a VAMC was in the top quartile of VA Video Connect (VVC) telehealth uptake for primary care and mental health care.

Results

: All 139 VAMCs increased their VVC uptake at least 2-fold early in the pandemic, with most increasing uptake between 5- and 10-fold. Pre-COVID-19, higher VVC uptake in primary care was weakly and positively associated with having more high-risk patients, negatively associated with having more long-distance patients, and positively associated with the prior fiscal year's VVC uptake. During COVID-19, the positive association with high-risk patients and the negative association with long-distance patients strengthened, while weaker broadband coverage was negatively associated with VVC uptake. For mental health care, having more long-distance patients was positively associated with higher VVC uptake pre-COVID-19, but this relationship reversed during COVID-19.

Discussion:

Despite the marked increase in VVC uptake early in the COVID-19 pandemic, significant VAMC-level variation indicates that VVC adoption was more difficult for some medical centers, particularly those with poorer broadband coverage and less prior VVC experience.

Conclusions and Relevance:

These findings highlight opportunities for medical centers, VA Central Office, and other federal entities to ensure equitable access to video telehealth.

Introduction

In response to the coronavirus disease 2019 (COVID-19) pandemic, Veterans Affairs Medical Centers (VAMCs) tried to minimize patient, staff, and community exposure to COVID-19 by transitioning to virtual appointments whenever possible. In March of 2020, the Department of Veterans Affairs (VA) leadership directed all VAMCs to convert in-person appointments to telephone and VA Video Connect (VVC), a telehealth modality with video visits to the patient's home, whenever clinically appropriate.¹ Although the majority of telehealth care during the pandemic has been conducted via telephone,² leadership indicated that VVC was the preferred modality based on its potential to provide a more comprehensive clinical experience for patients.¹ VA's response to this guidance was swift—by the end of June, the average number of daily VVC visits to patients at home was almost 20,000, up from 1,500 in February 2020.

As an early adopter of telehealth interventions, VA was uniquely positioned to scale up its use of VVC. For instance, since 2016, VA has expanded video telehealth into the home through a national program that disseminates video-enabled tablets to Veterans with access barriers. These tablets have built-in WiFi or mobile data connectivity and prepaid access to a wireless data network.³ However, recent evaluations of the tablet program highlighted significant variation in the types of implementation challenges faced by VAMCs. Low distributing sites, for example, were much more likely to report challenges related to a lack of provider interest, while high distributing sites were more likely to report challenges related to staffing shortages.³ Across VAMCs, these differences are likely to lead to variation in the proportion of patients who use video telehealth and to result in some VAMCs scaling up their use of VVC more quickly than others.

As the adoption of video telehealth remains a priority for VA, it is important to understand how use of VVC telehealth varied across medical centers and the factors associated with VAMCs' uptake of VVC in the early months of the pandemic. Focusing on this period highlights VAMCs that were able to rapidly scale up their use of VVC and, therefore, facilitate better continuity of care for their patients.^4,5 Identifying the types of VAMCs that experienced limited uptake of VVC telehealth early in the pandemic can help the VA Central Office better target its efforts to transition patients to VVC telehealth where appropriate.⁶ Furthermore, the nature of the COVID-19 pandemic may have driven new patterns of VVC telehealth uptake, such as greater demand for telehealth in urban areas.⁷ We aim to identify factors at the medical center level that were associated with VAMCs' uptake of primary care and mental health VVC telehealth before and during the COVID-19 pandemic.

Materials and Methods

ANALYTIC APPROACH

We conducted a retrospective, observational study using cross-sectional VAMC-level data for 139 medical centers from each of the 50 states, the District of Columbia, and Puerto Rico. We determined factors associated with a VAMC being in the top quartile of VVC uptake. We conducted multivariable analyses focusing on the 70 days before and the 70 days after March 11, 2020 (i.e., the date that the World Health Organization declared COVID-19 a pandemic). We used the 70-day window because a prior analysis reported that the largest shifts to VVC telehealth occurred during the first 2 months after March 11th.² We focused on primary care and mental health care separately because VA rates of VVC utilization have historically been much lower in primary care practice relative to mental health care.³ This evaluation was designated as nonresearch quality improvement by VA program office partners in the Office of Rural Health and was therefore exempt from Institutional Review Board approval.

DATA SOURCES

Encounter data were drawn from the VA's electronic health records. VAMCs are funded through a capitated payment model, with workload for primary care and mental health encounters captured using the VA's Managerial Cost Accounting stop codes.² VAMC rurality and complexity levels were obtained from the Veterans Health Administration Support (VHA) Site Tracking System. Patients were assigned to a VAMC using the Patient Centered Management Module database. Veterans' distance to the nearest facility was determined using VA's Planning System Support Group. Baseline telehealth performance was obtained from the VHA Center Capital Assets database. The number of tablets distributed by each VAMC was obtained from VA's Denver Acquisitions and Logistics Center.³ Fixed broadband and mobile coverage data were obtained from the Federal Communications Commission.⁸

MEASURES

Outcome measures

VVC uptake was adapted from a metric used to assess VAMC telehealth uptake (Tele9) in the VA's Telehealth Scorecard, a quarterly telehealth outcome report provided to VA leadership for quality management purposes.⁹ The metric divides the number of patients with a VVC encounter at each VAMC in a given fiscal year by the number of patients assigned to that VAMC in the previous fiscal year (i.e., 2019). We modified this measure by using alternate time periods (i.e., 70 days pre- and post- March 11, 2020). We used the same denominator for all measures to ensure we captured changes in VVC use and not changes in patient volume. The two primary care measures included patients who had at least one primary care VVC encounter from January 1, 2020, to March 10, 2020, and from March 11, 2020, to May 20, 2020. The two mental health measures included patients with at least one mental health VVC encounter over the same time periods. Outcome variables were dichotomized to compare VAMCs in the top quartile with all other VAMCs.

VAMC characteristics

VAMC characteristics were selected with guidance from the Non-adoption, Abandonment, and challenges to the Scale-up, Spread, and Sustainability of health and care technologies (NASSS) framework.¹⁰ This framework suggests that a medical center's adoption of new health care technologies is influenced by the following domains: patient conditions, key features of the technology, value proposition of the technology, adopter system and organizational characteristics, wider context, and organizational resilience. Table 1 summarizes the variables chosen to represent key NASSS concepts, variable definitions, the hypothesized association of each variable with VVC use, and whether the characteristic is modifiable at the VAMC-level.

Table 1.

VAMC Characteristics Selected to Capture NASSS Framework Domains and Their Expected Association with VVC Use

NASSS DOMAIN	VARIABLE	DEFINITION	HYPOTHESIZED ASSOCIATION WITH VAMC VVC USE	VAMC MODIFIABLE CHARACTERISTIC
Condition or illness	Proportion of patient population that is high risk	Patients with a CAN score ≥90 as a proportion of assigned patients at the VAMC at year end 2019, by quartile^a	Negative—patients with higher medical needs and more intense utilization patterns may be deemed less suitable candidates for telehealth technology¹⁰	No
Technology	Tablet distribution	Indicator for whether the VAMC was in the top quartile of centers distributing VA-issued tablets to Veterans with access issues up to March 11, 2020	Positive—greater experience with technology and interaction with suppliers leads to ease of rapid scale-up¹⁰	Yes—through support of the tablet distribution program
Value proposition	Proportion of long-distance patients	Patients who live ≥40 miles from assigned primary care facility as a proportion of assigned patients at the VAMC at year end 2019, by quartile	Positive (patient perspective)—patients who live further from VAMCs may have greater cost savings from use of video telehealth¹¹ Positive (health system perspective)—addressing the needs of long-distant patients via video telehealth could reduce logistical hassles and missed/late appointments¹²	No
Value proposition	Rurality of VAMC	Indicator for whether the primary facility of the VAMC is rural or highly rural	Uncertain Positive—VA has a long-standing commitment to improve access to care for rural patients¹³ Negative—patients in rural facilities may be less likely to have adequate broadband coverage for video telehealth¹⁴	No
Wider context	Broadband coverage in VAMC's primary facility county	Proportion of population in county where primary facility is located with 25/3 fixed broadband and 5/1 mobile broadband coverage, by quartile	Positive—facilities with better broadband coverage are more likely to use video telehealth technology¹⁴	No
Wider context (sensitivity analyses)	Broadband coverage in patients' county of residence	Proportion of Veteran population living in a county where <80% of residents have access to 25/3 fixed broadband and 5/1 mobile broadband coverage,³ by quartile^b	Positive—patients with better broadband coverage are more likely to use video telehealth technology¹⁵	No
Adopter system	VAMC encounter volume	Encounter volume 12 months preceding March 11, 2020	Positive—higher encounter volume can support investment in telehealth technology¹⁶	No
Adopter system	Patient count per provider	Patient count per primary care or mental health provider across all teams a provider may be assigned in the health care system	Positive—higher patient volume can support investment in telehealth technology¹⁶	Yes—through investments in staff
Adopter system	Facility complexity level	Highest facility complexity level of the health care system^c	Uncertain Positive—facility complexity is partially based on patient volume, which can support investment in telehealth technology¹⁵ Negative—facility complexity is partially based on intensity of medical need, which may be correlated with lower use of telehealth¹⁰	No
Organization	Baseline VVC use	Indicator for whether the VAMC tele9 score was in the top quartile of centers in fiscal year 2020, where tele9 indicates the proportion of assigned patients with ≥1 VVC encounter as a proportion of all assigned patients at the VAMC in 2019	Positive—organization's readiness for technology is strongly associated with its previous use¹⁰	Yes—through support of VVC programs at facilities

The CAN score calculates hospitalization risk based on sociodemographics, medical conditions, vital signs, prior use of health services, medications, and laboratory tests.

The 80% threshold reflects the mean level of broadband coverage for rural counties in the United States.

Facility complexity levels are assigned based on the characteristics of the patient population, clinical services, educational and research missions, and administrative complexity.

CAN, Care Assessment Need; NASSS, non-adoption, abandonment, scale-up, spread, sustainability; VA, Veterans Affairs; VAMC, Veterans Affairs Medical Center; VVC, VA Video Connect.

Patient conditions were captured by calculating the proportion of patients with a 1-year risk of hospitalization at or above the 90th percentile, using the VA's Care Assessment Need (CAN) score.¹⁷ Familiarity with video-based technology was captured using (1) the proportion of patients with ≥1 VVC encounter in the previous fiscal year and (2) a variable indicating that the VAMC was in the top quartile of VAMCs distributing tablets to Veterans. Value proposition was captured through parent facility rurality and the proportion of assigned patients living ≥40 miles from their primary care facility because VVC time and money savings are associated with Veterans living further from their primary VAMC.¹¹ Adopter system and organization characteristics included encounter volumes and patients per provider in the year preceding the pandemic. A VAMC's wider context was captured with broadband access in the county where each VAMC was located. We used the proportion of residents in the primary facility's county that had access to broadband at the Federal Communications Commission benchmark for high-speed broadband (i.e., 25 Mbps download/3 Mbps upload fixed terrestrial broadband service access and 5 Mbps download/1 Mbps upload mobile broadband access).⁸ This variable captured a potential constraint on VAMCs' ability to consistently provide VVC.¹⁴ Organizational resilience was captured using facility complexity levels.¹⁸

To capture the wider context related to patient demand, sensitivity analyses included the proportion of the VAMC-assigned Veterans living in a county where <80% of residents had 25/3 and 5/1 broadband coverage. These models excluded the proportion of long-distance patients, as it was highly correlated with our broadband access measure.

STATISTICAL ANALYSES

We used t-test and chi-square statistics to identify significant differences between VAMCs in the top quartile of VVC uptake and all other facilities during the COVID-19 pandemic. Descriptive analyses also assessed the correlation between VVC uptake in the 70 days before March 11th and the uptake in the 70 days after March 11th. We used multivariable logistic regressions to examine factors associated with being in the top quartiles of VVC uptake for primary care and mental health encounters. We used separate models for encounters between January 1, 2020, and March 10, 2020 (i.e., the pre-COVID-19 period) and between March 11, 2020, and May 20, 2020 (i.e., during COVID-19). We report marginal effects for each regression. All analyses were conducted using Stata 14 (StataCorp LLC).

Results

Table 2 presents VVC uptake pre- and during COVID-19, as well as VVC uptake for primary care and mental health care across 139 VAMCs. The proportion of assigned patients who had a VVC encounter increased from 0.87% pre-COVID-19 to 5.65% during the COVID-19 pandemic. VVC uptake was higher for Veterans using mental health services than for primary care services, increasing from 0.12% to 1.60% for primary care and from 0.59% to 2.81% for mental health care. Figure 1 illustrates the substantial increases in VVC uptake early in the pandemic. All 139 VAMCs increased their VVC uptake by at least 2 times what it was before COVID-19, with most VAMCs increasing their VVC uptake between 5- and 10-fold.

Fig. 1.

Scatterplot of VVC uptake pre-COVID-19 versus during COVID-19. COVID-19, coronavirus disease 2019; VVC, VA Video Connect.

Table 2.

Proportion of Assigned VAMC Patients with a VVC Encounter Pre-COVID-19 and During COVID-19 for 139 VA Medical Centers

	PRE-COVID^a PROPORTION	PRE-COVID-19 TOP QUARTILE RANGE	DURING COVID^b PROPORTION	DURING COVID-19 TOP QUARTILE RANGE
VVC uptake—all encounter types	0.87%	1.10–3.48%	5.65%	6.57–15.08%
VVC uptake—primary care encounters	0.12%	0.15–0.56%	1.60%	2.16–5.02%
VVC uptake—mental health encounters	0.59%	0.76–3.07%	2.81%	3.42–11.09%

January 1, 2020, to March 10, 2020.

March 11, 2020, to May 20, 2020.

COVID-19, coronavirus disease 2019.

Bivariate analyses in Table 3 indicate that VAMCs with the highest proportions of patients using VVC during the pandemic had more high-risk patients and fewer long-distance patients. Higher performing VAMCs for primary care VVC were more likely to be in areas with higher mobile broadband coverage and more likely to be in the top quartile for VVC uptake in the fiscal year 2019.

Table 3.

VA Medical Center Characteristics Among VAMCs in the Top Quartile Versus Remaining Quartiles of VVC Uptake During COVID-19 (March 11, 2020, to May 20, 2020)

	PRIMARY CARE			MENTAL HEALTH
	TOP PERFORMING VVC FACILITIES—PROPORTION/MEAN (SD)	REMAINING FACILITIES—PROPORTION/MEAN (SD)	SIG	TOP PERFORMING VVC FACILITIES—PROPORTION/MEAN (SD)	REMAINING FACILITIES—PROPORTION/MEAN (SD)	SIG
Proportion of high-risk patients
Lowest quartile (3.7–9.2%)	8.82%	30.48%	^*	26.47%	24.76%
Second quartile (9.3–11.0%)	32.35%	22.86%		23.53%	25.71%
Third quartile (11.1–12.8%)	26.47%	24.76%		14.71%	28.57%
Fourth quartile (12.9–20%)	32.35%	21.90%		35.29%	20.95%	^†
Proportion of patients ≥40 miles from assigned PC facility
Lowest quartile (0–0.14%)	50.00%	17.14%	^**	50.00%	17.14%	^**
Second quartile (0.15–5.2%)	23.53%	25.71%		26.47%	24.76%
Third quartile (5.5–10.9%)	20.59%	26.67%		11.76%	29.52%	^*
Fourth quartile (10.9–38.1%)	5.88%	30.48%	^**	11.76%	28.57%	^*
Primary care encounter volume in the 12 months pre-COVID
Less than 10,000	29.41%	20.00%		26.47%	20.95%
10,000 to 19,999	35.29%	57.14%	^*	47.06%	53.33%
20,000 or more	35.29%	22.86%		26.47%	25.71%
VAMC primary facilities classified as rural or highly rural	11.76%	12.38%		5.88%	14.29%
Facility complexity^a
Most complex	67.65%	65.71%		70.59%	64.76%
Intermediate complex	11.76%	15.24%		11.76%	15.24%
Least complex	20.59%	19.05%		17.65%	20.00%
Top quartile in number of VA-issued tablet recipients (>105 recipients)	21.88%	25.96%		23.53%	25.71%
Patients per PCP	861.16 (119.575)	865.593 (90.925)		853.48 (104.68)	868.08 (96.32)
Proportion of residents in primary facility county with 25/3 fixed broadband and 5/1 mobile
60–94.4%	14.71%	28.57%		17.65%	27.62%
94.5–97.7%	17.65%	27.62%		20.59%	26.67%
97.8–99.3%	26.47%	25.71%		35.29%	22.86%
99.4–100%	41.18%	18.10%	^**	26.47%	22.86%
In top quartile of VVC uptake in FY19	41.18%	19.05%	^**	58.82%	13.33%	^**
N	35	104		35	104

Significance for chi-squared and t-tests: ^*** p < 0.001; ^** p < 0.01; ^* p < 0.05; ^† p < 0.10.

Composite measure based on patient, clinical service, education, and research characteristics.

PCP, primary care provider; SD, standard deviation.

PRIMARY CARE

Multivariable analyses summarized in Table 4 indicate that pre- and during COVID-19, VAMCs having more high-risk patients and fewer long-distance patients were positively associated with higher VVC uptake. Relative to facilities with the lowest proportion of high-risk patients, VAMCs with the highest proportions of high-risk patients were 35% points (95% confidence interval [CI]: 0.11 to 0.62) more likely to have high VVC uptake. Facilities in the top quartile for long-distance patients were 37% points (95% CI: −0.597 to −0.149) less likely than those in the bottom quartile to have higher VVC uptake. Before COVID-19, broadband access was not associated with primary care VVC uptake, but facility broadband was a strong predictor of higher uptake during COVID-19. VAMCs in the bottom quartile of broadband access were 23% points less likely to have a higher VVC uptake than VAMCs in the top quartile of broadband access. In the pre-COVID-19 period, having higher VVC uptake in the previous year increased a VAMC's likelihood of having a higher VVC uptake by 15% points (95% CI: 0.01 to 0.30) but was not a significant predictor of VVC uptake during COVID-19.

Table 4.

Marginal Effects from Logistic Regressions Predicting Being in the Top Quartile of VVC Uptake for Primary Care Encounters Pre-COVID-19 and During COVID-19

PRIMARY CARE TELE9 COVID	PRE-COVID-19				DURING COVID-19
PRIMARY CARE TELE9 COVID	MARGINAL EFFECT	95% CI		SIG	MARGINAL EFFECT	95% CI		SIG
Proportion of high-risk patients
Lowest quartile (3.7–9.2%) (Ref.)
Second quartile (9.3–11.0%)	0.249	0.029	0.469	^*	0.386	0.141	0.630	^**
Third quartile (11.1–12.8%)	0.134	−0.082	0.349		0.275	0.027	0.523	^*
Fourth quartile (12.9–20%)	0.155	−0.082	0.391		0.361	0.107	0.615	^**
Proportion of patients ≥40 miles from assigned PC facility
Lowest quartile (0–0.14%) (Ref.)
Second quartile (0.15–5.2%)	0.002	−0.180	0.183		−0.125	−0.288	0.038
Third quartile (5.5–10.9%)	−0.041	−0.235	0.154		−0.093	−0.264	0.077
Fourth quartile (10.9–38.1%)	−0.329	−0.575	−0.083	^**	−0.373	−0.597	−0.149	^***
Primary care encounter volume in the 12 months pre-COVID
Less than 10,000 (Ref.)
10,000 to 19,999	0.087	−0.135	0.310		−0.076	−0.278	0.126
20,000 or more	0.132	−0.148	0.413		0.092	−0.171	0.356
VAMC primary facilities classified as rural or highly rural	0.228	−0.041	0.497	^†	0.152	−0.086	0.389
Facility complexity^a
Most complex (Ref.)
Intermediate complex	−0.099	−0.350	0.152		0.093	−0.144	0.331
Least complex	0.085	−0.169	0.338		0.211	−0.029	0.450	^†
Top quartile in number of VA-issued tablet recipients (>105 recipients)	−0.008	−0.176	0.160		−0.019	−0.184	0.146
Patients per PCP	−0.001	−0.001	0.000		0.000	−0.001	0.000
Proportion of residents in primary facility county with 25/3 fixed broadband and 5/1 mobile
60–94.4%	−0.122	−0.377	0.132		−0.226	−0.444	−0.007	^*
94.5–97.7%	0.108	−0.079	0.295		−0.106	−0.280	0.067
97.8–99.3%	−0.072	−0.256	0.112		−0.172	−0.333	−0.010	^*
99.4–100% (Ref.)
In top quartile of VVC uptake in FY19	0.155	0.013	0.297	^*	0.093	−0.047	0.233
N	139				139

Significance: ^*** p < 0.001; ^** p < 0.01; ^* p < 0.05; ^† p < 0.10.

Composite measure based on patient, clinical service, education, and research characteristics.

CI, confidence interval.

MENTAL HEALTH CARE

Multivariable analyses summarized in Table 5 indicate that VAMCs with higher proportions of high-risk patients were less likely to have higher VVC penetration rates. Pre-COVID-19, VAMCs with higher uptake were those with a higher proportion of long-distance patients. During COVID-19, the direction of this association reversed. Facilities in the top quartile of long-distance patients were 22% points (95% CI: −0.41 to −0.03) less likely than those in the bottom quartile to have a higher VVC uptake. Pre-COVID-19, VAMCs with somewhat lower broadband coverage (i.e., those in the third quartile of coverage) were 28% points (95% CI: 0.099 to 0.46) more likely to have higher VVC uptake. This association was no longer significant during COVID-19. Finally, having higher VVC uptake in the prior fiscal year was positively associated with having higher VVC uptake both before and during COVID-19.

Table 5.

Marginal Effects from Logistic Regressions Predicting Being in the Top Quartile of VVC Uptake for Mental Health Encounters Pre-COVID-19 and During COVID-19

PRIMARY CARE TELE9 COVID	PRE-COVID-19				DURING COVID-19
PRIMARY CARE TELE9 COVID	MARGINAL EFFECT	95% CI		SIG	MARGINAL EFFECT	95% CI		SIG
Proportion of high-risk patients
Lowest quartile (3.7–9.2%) (Ref.)
Second quartile (9.3–11.0%)	0.006	−0.176	0.188		−0.070	−0.254	0.114
Third quartile (11.1–12.8%)	−0.134	−0.339	0.071		−0.278	−0.498	−0.058	^*
Fourth quartile (12.9–20%)	0.142	−0.071	0.356		−0.064	−0.266	0.138
Proportion of patients ≥40 miles from assigned PC facility
Lowest quartile (0–0.14%) (Ref.)
Second quartile (0.15–5.2%)	0.324	0.136	0.512	^***	−0.151	−0.302	0.000	^*
Third quartile (5.5–10.9%)	0.194	0.008	0.380	^*	−0.226	−0.406	−0.045	^*
Fourth quartile (10.9–38.1%)	0.035	−0.176	0.246		−0.222	−0.413	−0.032	^*
MH encounter volume in the 12 months pre-COVID
Less than 10,000 (Ref.)
10,000 to 19,999	−0.034	−0.237	0.168		0.084	−0.103	0.272
20,000 or more	0.053	−0.218	0.323		0.031	−0.214	0.277
VAMC primary facilities classified as rural or highly rural	0.129	−0.074	0.332		−0.205	−0.471	0.062
Facility complexity^a
Most complex (Ref.)
Intermediate complex	0.001	−0.263	0.265		−0.010	−0.243	0.223
Least complex	0.188	−0.071	0.448		−0.032	−0.281	0.218
Top quartile in number of VA-issued tablet recipients (>105 recipients)	0.034	−0.111	0.179		−0.024	−0.181	0.132
Patients per MHP	−0.0001	−0.001	0.001		−0.001	−0.001	0.000	^†
Proportion of residents in primary facility county with 25/3 fixed broadband and 5/1 mobile
60–94.4%	0.034	−0.180	0.247		0.017	−0.195	0.229
94.5–97.7%	−0.031	−0.235	0.174		0.058	−0.119	0.235
97.8–99.3%	0.281	0.099	0.462	^**	−0.032	−0.193	0.130
99.4–100% (Ref.)
In top quartile of VVC uptake in FY19	0.270	0.154	0.386	^***	0.296	0.192	0.401	^***
N	139				139

Significance: ^*** p < 0.001; ^** p < 0.01; ^* p < 0.05; ^† p < 0.10.

Composite measure based on patient, clinical service, education, and research characteristics.

Results were similar in models that included Veterans' broadband access ( Appendix Tables A1 and A2 ). During COVID-19, VAMCs whose patient population had the lowest levels of broadband access were 33% points (95% CI: −0.57 to −0.08) and 32% points (95% CI: −0.54 to −0.99) less likely to be in the top quartile of primary care and mental health VVC uptake, respectively.

Discussion

In this study, we describe VVC uptake across 139 VAMCs in the 70 days immediately preceding and following the WHO's declaration of COVID-19 as a pandemic. We found a 6.5-fold increase in the proportion of patients with one or more VVC visit during the pandemic. This is consistent with prior work that found that weekly video visits increased from 10,000 to 60,000 over the same time period, while in-person encounters decreased from 2 million to 600,000 and telephone encounters increased from 327,275 to 869,251.² Before COVID-19, a higher VVC uptake for primary care was weakly and positively associated with having more high-risk patients and negatively associated with having more long-distance patients. During COVID-19, the positive association with high-risk patients strengthened, as did the negative association with the proportion of long-distance patients. Weaker broadband coverage was also negatively associated with VVC uptake during COVID-19. For mental health care, the positive association between having more long-distance patients and higher VVC uptake that was observed pre-COVID-19 reversed during the pandemic, while the association between lower broadband coverage and VVC uptake disappeared during COVID-19.

Our findings that both primary care and mental health VVC uptake during COVID-19 were negatively associated with having more long-distance patients is likely a reflection of the greater use of video care among VA patients residing in urban areas affected by COVID-19. This is supported by prior work, where we found that new users of video care during the pandemic were 17% more likely to be urban-dwelling Veterans.² These findings are also reflective of urban populations being more likely to have the required access to a device and high-quality connectivity necessary to accomplish VVC into the home, as supported by sensitivity analyses that included controls for individual-level broadband access. The fact that the direction of this association reversed for mental health VVC uptake suggests a comparatively larger growth in VVC use for urban patients during the pandemic. While mental health providers were more likely to use VVC for rural patients with access issues prepandemic, during the pandemic, VVC helped to expand access in urban areas that were significantly impacted by COVID-19 and were likely to be early adopters of shelter-in-place recommendations.⁷

The strengthening of the association between higher VVC uptake and the proportion of high-risk patients suggests that during the pandemic, VAMCs may have prioritized VVC appointments for higher risk patients. This is in line with guidance from the VA Central Office that suggested that VA primary care teams reach out to their high-risk Veterans to inquire about their needs.¹⁹ The negative association observed between the proportion of high-risk patients and higher VVC penetration for mental health might be due to the greater use of telephone outreach among high-risk patients who also had mental health needs.

Our finding that broadband coverage became more important for primary care during the pandemic may suggest that when there was lower demand for primary care VVC encounters pre-COVID-19, VAMCs were able to address issues related to broadband access through existing programs. These include the VA's tablet program, partnerships offering broadband access to Veterans, or through programs that promote VVC access points outside of Veterans' homes.^3,20,21 Early in the pandemic, when demand for these types of services likely outstripped supply, insufficient broadband coverage could have become a greater deterrent to primary care VVC use. This indicates that the smaller scale programs that were addressing broadband coverage issues likely need to be expanded if VA wants to continue to promote the growth of VVC primary care encounters. In particular, additional resources should be geared toward strategies affordably boosting connectivity for the 25% of rural-dwelling Veterans more often challenged with accessing quality broadband.²²

In addition, the more frequent need for a physical examination and peripheral devices in primary care encounters, relative to mental health encounters, relies on better broadband coverage for high-definition video streaming. Pre-COVID, VAMCs with somewhat lower broadband coverage were more likely to have higher VVC uptake. This may be another indication that existing programs were helping close the gaps in broadband access for Veterans when there was less demand for these services. The advent of 5G cellular connectivity, coupled with substantial federal investment in augmenting connectivity in geographically remote areas, implies that those living in areas with limited broadband will have the opportunity to more readily engage in all types of virtual care.²³

Our results should be interpreted while considering certain limitations. For instance, in our analyses focused on the association between VVC uptake and broadband, there were relatively small differences in coverage between the top and bottom quartiles. We do note, however, that the cutoff for the bottom quartile (i.e., less than 94.4% of residents in the primary facility's county) is the same as the national average for 25/3 and 5/1 broadband coverage across all U.S. counties.⁸ As such, counties in the first quartile all fall below the national average of broadband coverage. We also note a relatively short follow-up period (i.e., 70 days) for our analysis. The intent of this article was to identify characteristics associated with more rapid ramp-up of VVC telehealth, and so, this analysis does not capture VAMCs that achieved higher uptake over a longer period of time in our outcome measure. Enhancements to VA video infrastructure occurred during May 2020, and so, our analysis does not capture improvements in access due to these changes. We also note that although our patient complexity measure is correlated with age,²⁴ we did not explicitly account for the mean age of assigned patients at the VAMC, which may be associated with telehealth use and internet access. Finally, we note that our definition of “high-risk” patients focused on high-cost, high-need patients which, while correlated with mental health diagnoses,²⁴ does not explicitly account for mental health complexity.

Conclusion

Despite an overall increase in VVC uptake early in the COVID-19 pandemic, VVC adoption rates were highly varied. Some factors that predicted VVC uptake can inform implementation strategies at the VAMC level. For instance, the finding that higher VVC uptake in the prior year predicted rapid VVC expansion during the pandemic suggests that a VAMC's prior infrastructure investment and familiarity with the technology facilitated its rapid expansion. Other factors, while not modifiable by VAMCs, may suggest ways in which the VA Central Office and other federal entities might better target future investments to ensure equitable access to this modality. For instance, the fact that VAMCs with poorer broadband coverage had difficulties rapidly expanding VVC use highlights the importance of investments in programs that help close gaps in Veterans' broadband access and of federal investment in expanding connectivity in geographically remote areas.

Footnotes

Authors' Contributions

Conceptualization: J.J., J.M.F., M.Y., J.V.C., L.G., L.H., and D.M.Z. Data curation: J.V.C. Analysis: J.J. Methodology: J.J., J.M.F., M.Y., and D.M.Z. Visualization: J.M.F. Writing and editing: J.J., J.M.F., M.Y., L.G., L.H., and D.M.Z.

Acknowledgments

The authors thank Cindie Slightam, Camila Chaudhary, the VA Office of Connected Care, and the Virtual Access QUERI team for their help in supporting this evaluation.

Disclosure Statement

No competing financial interests exist.

Funding Information

This work was supported, in part, by a Career Development Award 19-120 (J.J.) from the U.S. Department of Veterans Affairs Health Services Research and Development Service and by a Quality Enhancement Research Imitative (QUERI) (PEI 18-205).

Appendix

References

Heyworth

, Kirsh

, Zulman

, Ferguson

, Kizer

. Expanding access through virtual care: The VA's early experience with Covid-19. NEJM Catal Innov Care Deliv, 2020; 1:1–11.

Ferguson

, Jacobs

, Yefimova

, Greene

, Heyworth

, Zulman

. Virtual care expansion in the Veterans Health Administration during the COVID-19 Pandemic: Clinical services and patient characteristics associated with utilization. J Am Med Inform Assoc, 2020; 28:453–462.

Zulman

, Wong

, Slightam

, et al. Making connections: Nationwide implementation of video telehealth tablets to address access barriers in Veterans. JAMIA Open, 2019; 2:323–329.

Wosik

, Fudim

, Cameron

, et al. Telehealth transformation: COVID-19 and the rise of virtual care. J Am Med Inform Assoc, 2020; 27:957–962.

Wosik

, Clowse

MEB

, Overton

, et al. Impact of the COVID-19 pandemic on patterns of outpatient cardiovascular care. Am Heart J, 2021; 231:1–5.

Adler-Milstein

, Kvedar

, Bates

. Telehealth among US hospitals: Several factors, including state reimbursement and licensure policies, influence adoption. Health Aff (Millwood), 2014; 33:207–215.

Dave

, Friedson

, Matsuzawa

, Sabia

, Safford

. JUE Insight: Were urban cowboys enough to control COVID-19? Local shelter-in-place orders and coronavirus case growth. J Urban Econ, 2020; [Epub ahead of print]; DOI: https://doi.org/10.1016/j.jue.2020.103294. (last accessed April 2, 2021 ).

Federal Communications Commission. 2020. Broadband Deployment Report. Published June 8, 2020. Available at https://www.fcc.gov/reports-research/reports/broadband-progress-reports/2020-broadband-deployment-report (last accessed September 15, 2020 ).

Darkins

, Foster

, Anderson

, Goldschmidt

, Selvin

. The design, implementation, and operational management of a comprehensive quality management program to support National Telehealth Networks. Telemed J E Health, 2013; 19:557–564.

10.

Greenhalgh

, Wherton

, Papoutsi

, et al. Beyond adoption: A new framework for theorizing and evaluating nonadoption, abandonment, and challenges to the scale-up, spread, and sustainability of health and care technologies. J Med Internet Res, 2017; 19:e367.

11.

Jacobs

, Hu

, Slightam

, Gregory

, Zulman

. Virtual savings: Patient-reported time and money savings from a VA National Telehealth Tablet Initiative. Telemed J E Health, 2019; 26:1178–1183.

12.

Jacobs

, Blonigen

, Kimerling

, et al. Increasing mental health care access, continuity, and efficiency for Veterans through telehealth with video tablets. Psychiatr Serv, 2019; 70:976–982.

13.

Lum

, Nearing

, Pimentel

, Levy

, Hung

. Anywhere to anywhere: Use of telehealth to increase health care access for older, rural Veterans. Public Policy Aging Rep, 2020; 30:12–18.

14.

Whitacre

, Wheeler

, Landgraf

. What can the National Broadband Map Tell us about the health care connectivity gap?. J Rural Health, 2017; 33:284–289.

15.

Slightam

, Gregory

, Hu

, et al. Patient perceptions of video visits using Veterans Affairs telehealth tablets: Survey study. J Med Internet Res, 2020; 22:e15682.

16.

Park

, Erikson

, Han

, Iyer

. Are state telehealth policies associated with the use of telehealth services among underserved populations?. Health Aff (Millwood), 2018; 37:2060–2068.

17.

Wang

, Porter

, Maynard

, et al. Predicting risk of hospitalization or death among patients receiving primary care in the Veterans Health Administration. Med Care, 2013; 51:368–373.

18.

Department of Veterans Affairs. VHA Facility Quality and Safety Report—Fiscal Year. 2012. Data.; 2013:64. Available at https://www.va.gov/HEALTH/docs/VHA_Quality_and_Safety_Report_2013.pdf (last accessed September 1, 2020 ).

19.

Veterans Health

Administration

. COVID-10 response plan: Incident-specific annex to the VHA high consequence infection (HCI) base plan

Version

. Published

August

, 2020. Available at https://www.publichealth.va.gov/docs/n-coronavirus/COVID-19_Response_Plan_07Aug2020_signed_final.pdf (last accessed April 4, 2021 ).

20.

Galpin

. VA partners with Microsoft to bring internet to more

Veterans

. VAntage

Point

. Published

June

, 2019. Available at https://www.blogs.va.gov/VAntage/61913/multiple-vha-partnerships-work-bring-veterans-online/ (last accessed September 15, 2020 ).

21.

Kaupanger

. VA healthcare: At home, in a library and soon, to a

VSO near you

. Connecting

Vets

. Published

July

, 2018. Available at https://connectingvets.radio.com/articles/telehealth-va-expanding-access-rural-veterans-receive-healthcare (last accessed September 15, 2020 ).

22.

U.S. Department of Veterans Affairs. RURAL VETERANS—Office of Rural Health. Available at https://www.ruralhealth.va.gov/aboutus/ruralvets.asp (last accessed October 21, 2020 ).

23.

Sadowski

. Rural Broadband Will Need Federal

Help

. Here's What Trump Has Done and What Biden

Proposes

. UpNorthNews. Published

August

, 2020. Available at https://upnorthnewswi.com/2020/08/24/rural-broadband-will-need-federal-help-heres-what-trump-has-done-and-what-biden-proposes/ (last accessed October 21, 2020 ).

24.

Fihn

, Box

. Care Assessment Need (CAN) Score and Patient Care Assessment System (PCAS): Tools for care management. Presented at the VIReC Clinical Informatics Semicar: June 27, 2013. Available at https://www.hsrd.research.va.gov/for_researchers/cyber_seminars/archives/713-notes.pdf (last accessed February 25, 2021 ).