Community Urgent Care Use Following Implementation of the Veterans Affairs Maintaining Internal Systems and Strengthening Integrated Outside Networks Act.

BACKGROUND: Effective June 6, 2019, Veterans Affairs (VA) began offering a new urgent care (UC) benefit that provides eligible Veterans with greater choice and access to care for the treatment of minor injuries and illnesses in their local communities.
OBJECTIVES: The aim was to describe trends in UC use, identify predictors of UC benefit use, and understand the factors associated with community UC use versus VA emergency department (ED) or urgent care center (UCC) use. STUDY
DESIGN: Using VA administrative data, this was a retrospective cross-sectional study of Veterans that were enrolled in VA in FY19. Veterans were classified into 3 groups: UC benefit users, benefit non-users, and VA ED/UCC users.
METHODS: We used summary statistics to compare population characteristics across user groups. To determine whether predisposing, enabling, and need factors predicted UC benefit use and setting choice (community UCC vs. VA ED/UCC), 2 logistic regression models were fitted to assess odds of UC use.
RESULTS: From June 6, 2019 through February 29, 2020, 138,305 Veterans made 175,821 community UC visits. The majority of visits were made by White males who were not subject to co-pays. The average cost to VA for UC visits was $132 (SD=$135). Upper respiratory infections were the most common reason for UC use. Being younger, female, and living farther from a VA ED/UCC was associated with greater UC benefit use compared with both benefit non-users and VA ED/UCC users.
CONCLUSIONS: The new benefit expands Veteran access to UC services for low-acuity conditions.

Approximately 1 in 3 emergency department (ED) visits in the United States are nonemergent,1 potentially leading to unnecessary testing, treatment, and cost. Payers have long struggled to discourage nonemergent ED visits through patient education and higher copayments for ED visits.1 Delivery systems have built alternatives like same day or after-hours primary care, urgent care centers (UCCs), retail clinics, and telemedicine.1–3

In its approach to this problem, the Veterans Affairs (VA) health care system is unusual in that it is both a large payer and delivery system. Historically, the VA has provided acute, unscheduled care at 111 EDs and 34 distinct UCCs across the country.4 The VA also sometimes pays for Veterans to receive emergency medical care at non-VA EDs,5 subject to some eligibility requirements, and this has grown to become one of the costliest components for VA purchased community care.6 Aside from same day access to primary care, VA had not until recently provided a robust, easy to access network of alternative care options for nonemergent care in VA or the community.

On June 6, 2019, as part of the VA Maintaining Internal Systems and Strengthening Integrated Outside Networks (MISSION) Act,7 VA began offering a new urgent care (UC) benefit that allows eligible Veterans to receive UC from providers within VA’s community network, without prior authorization from VA. As Veterans take advantage of the new benefit, understanding the extent to which access to acute care improves, and for which Veterans, will help guide the future of the program.

In this study, we specifically focused on UC use during early implementation of the MISSION Act UC benefit. Our objectives were threefold: (1) to describe the characteristics of UC users and their visits; (2) to assess individual characteristics that predict UC benefit use; and (3) to assess factors that are predictive of setting choice (community UCC use vs. VA ED/UCC use). Beyond informing VA policymakers, these results may help other health systems and payers as they consider the benefits of integrating financing and delivery of UC services.

METHODS

Setting

UCCs have no formal definition, and have limited regulatory and state licensing requirements. They are typically defined as ambulatory medical facilities outside of a traditional hospital-based or freestanding ED that focus on the treatment of minor illnesses and injuries.8 An UCC may have the following characteristics: multiple exam rooms and on-site diagnostic equipment; a licensed physician acting as medical director; accepts walk-in patients during business hours; treats a broad spectrum of illnesses and injuries; performs minor medical procedures; and is open 7 days a week. The difference between UC medicine and emergency medicine is that UC primarily focuses on nonemergent problems at the lower end of the severity spectrum. Generally, patients at UCCs are evaluated and treated with simple, office-based laboratory tests (eg, urinalysis, pregnancy test, rapid strep assay) and x-rays. A typical UCC does not have immediate access to extensive laboratory testing or advanced imaging.9

VA UCCs are defined as a center that “provides acute medical care for patients without a scheduled appointment who are in need of immediate attention for an acute medical or mental health illness and/or minor injuries.”4 Unlike VA EDs, VA UCCs generally do not operate on a 24 hours, 7 days a week schedule, and they do not accept patients through ambulance. Because VA UCCs do not operate in facilities that have an ED, both VA EDs and UCCs treat nonemergent or low-acuity conditions.

Overview of Urgent Care and the Veterans Affairs Community Care Urgent Care Benefit

The UC benefit is considered open access: Veterans may go to any in-network UCC without a referral from VA. The program initially utilized VA’s existing nation-wide Patient-Centered Community Care network of community providers. In March 2020, VA transitioned to the Community Care Network, a network that has 6 regional networks that VA will use to purchase care in the community.10

The UC benefit covers injuries and illnesses that are not life-threatening but require immediate attention, such as respiratory infections, urinary complaints, and minor injuries. To be eligible to use the benefit, the Veteran must be enrolled in the VA health care system and have received care through VA, from either VA or a VA-contracted community provider, within the past 24 months. UC providers must be in VA’s contracted network. An online directory of VA network providers is available for Veterans to access.11

Before providing care, in-network UC providers are required to call a centralized number to verify the Veteran’s eligibility. During the visit, the UCC does not collect a copayment from the Veteran; however, VA may bill the Veteran for an applicable copayment based on the Veteran’s priority group and the number of UC visits made in a calendar year. UC providers are expected to advise Veterans to follow up with their VA primary care provider following the UC encounter, and medical documentation must be submitted directly to the Veteran’s VA facility within 30 days of the visit. Finally, UCCs must submit claims to third party administrators within 6 months of the service. UC providers can write a prescription for up to a 14-day supply to be filled at a VA or in-network pharmacy.11

Data and Measures

Our initial cohort included Veterans enrolled in VA health care in FY19. Three Veteran user types were defined: Veterans that utilized the community UC benefit (UC benefit users), Veterans that did not utilize the community UC benefit (benefit non-users), and Veterans that utilized a VA ED/UCC for low-acuity conditions (VA ED/UCC users). These groups were not mutually exclusive. VA ED/UCC users are a subset of benefit non-users. Thus, if a Veteran utilized both community and VA UC during the study period, they were classified as a community UC benefit user.

We examined UC patients and visits from June 6, 2019 (start of the benefit) to February 29, 2020, using data provided by the VA Office of Community Care, the office that tracks all UC encounters provided through the VA Community Care program.12 We included all processed and paid claims as of October 7, 2020, 7 months after our study end date, to minimize the impact of lags in claims submissions and processing and changes in acute care utilization because of coronavirus disease 2019 (COVID-19).

To create a comparable sample of Veterans that used VA care with similar health care needs and acuity, we defined low-acuity visits as any VA ED/UCC visit with a triage assignment of an emergency severity index (ESI) level 4 or 5, which corresponds to the need for 0 or 1 resource based on the patient’s chief complaint.13

Our investigation was guided by the Andersen health care utilization model, a conceptual framework that suggests health care utilization is determined by 3 dynamics: predisposing factors, enabling factors, and need.14 We measured all 3 dynamics at the individual characteristic level. Predisposing variables included age, sex, and race. Enabling variables included housing instability (yes/no), rurality (urban, rural, highly rural), and ease of accessing the treatment site [driving time (in minutes) from a Veteran’s residence to closest VA primary care center and VA ED or VA UCC]. Driving times were derived from VA Planning Systems Support Group data. Need variables (clinical need and perceived need) included: VA priority group15 (which determines VA enrollment priority based on military service, service-connected disabilities, and income, and also determines co-pay), total number of Elixhauser comorbidities,16 primary care utilization in the 1 year before the study period (total number of VA outpatient primary care visits including telephone, virtual, home-based, and face-to-face encounters), and acute care utilization in the 1 year before the study period (total number of VA ED/UCC visits, total number of community ED visits paid for by the VA). Prior use variables were top-coded using their respective 99.5th percentile values (36 for VA primary care visits, 13 for VA ED/UCC visits, and 19 for community ED visits).

Reasons for UC visits were classified using primary diagnosis into meaningful categories.17 Medical procedures and services were classified using Current Procedure Terminology (CPT) codes.18 CPT codes related to evaluation and management of new (99201-99205) and established (99211-99215) patients were excluded from analysis.

Statistical Analysis

We used summary statistics to compare population characteristics across user groups. To determine whether predisposing, enabling, and need (individual) factors predicted UC benefit use and setting preference (community UCC vs. VA ED/UCC), 2 logistic regression models were fitted to assess odds of UC benefit use, the first in the full population and the latter in the subset with any community UCC or VA ED/UCC use.

To assess collinearity, measures of association (Pearson correlation for numeric variables and the Cramér V for categorical variables) were evaluated for pairs of variables likely to have a high degree of association. Analysis for collinearity found the presence of moderate to strong pairwise association between driving time to VA primary care and driving time to VA ED/UCC and between each driving time variable and urban/rural/highly rural residence. Though these associations were not strong enough to require that we choose between the variables, we ran sensitivity analyses dropping driving time to VA primary care in both models (this variable was less tied to our outcome of interest than driving time to VA ED/UCC and because it was one of the variables in the strongest pairwise association, with rurality).

Data were also assessed for distribution patterns, cell size, and missingness before analyses. To address missing data, we performed multiple imputation for both models. In each population of interest, we used all variables in our full model to create 5 imputed datasets, and fewer than 10% of patients had any missing values. We used a fully conditional specification approach and all variables with missing values were categorical and had the discriminant function applied to them. We used 50 burn-in iterations for all imputations; imputations and the combining of results across imputed datasets were performed with PROC MI and PROC MIANALYZE.

All analyses were conducted using SAS version 9.04 (SAS Institute Inc., Cary, NC), Microsoft SQL Server Enterprise version 14.0 (Microsoft, Redmond, WA), and R version 3.6.1 (R Core Team, 2019). This study was approved by the Stanford University Institutional Review Board.

RESULTS

Characteristics of the Veteran population by user type are presented in Table 1. The UC benefit users in our cohort (n=138,305) had a mean age of 53.1 years (SD=16.8), were predominately male (82.2%), White (75.3%), and resided in urban areas (64.1%). Less than half of Veterans utilizing the benefit (43.4%) lived more than 60 minutes from a VA ED/UCC and the vast majority (81.1%) lived within 30 minutes of a VA primary care site.

TABLE 1

Patient Characteristics by Urgent Care User Type

	UC Benefit Users	UC Benefit Non-users	VA ED/UCC Users
Characteristics	n=138,305*	n=5,755,942	n=431,843
Sex, male, %	82.2	91.5	88.5
Age, mean (SD)	53.1 (16.8)	62.6 (16.6)	58.2 (15.7)
Age, y, %
18–44	34.6	17.1	21.7
45–64	34.8	29.3	38.3
≥65	30.7	53.6	40.0
Missing	<0.05	<0.05	<0.05
Race, %
White	75.3	72.3	63.2
Non-White	18.2	19.9	31.3
Missing	6.5	7.9	5.5
Unhoused, %	3.6	3.9	10.0
Residential location, %
Urban	64.1	64.9	74.0
Rural	30.5	30.9	23.3
Highly rural	2.8	4.1	2.6
Missing	2.7	0.1	<0.05
US census region, %
Northeast	9.9	12.7	12.1
South	55.1	44.1	44.1
Midwest	14.8	21.9	22.5
West	20.2	21.3	21.4
Drive time to closest VA ED/UCC, median (IQR)	55 (31–91)	43 (23–81)	23 (14–40)
Drive time to closest VA ED/UCC, min, %
≤30	23.5	35.8	63.4
>30≤60	29.3	27.2	23.8
>60	43.4	36.1	12.7
Missing	3.7	0.9	0.1
Drive time to closest VA PC services, median (IQR)	17 (11–27)	17 (11–27)	15 (10–23)
Drive time to closest VA PC services, min, %
≤30	81.1	79.3	85.4
>30≤60	17.0	17.7	13.1
>60	1.9	2.9	1.5
Missing	0.1	0.1	<0.05
VA priority group, %
1 and 4: high disability (≥50% SC)	51.1	40.1	48.6
2, 3, 5: low/moderate disability (10%–40% SC) or low income	35.5	38.4	39.2
6–8: nondisabled (0% SC), UC copayment required	13.4	21.4	12.2
Missing	<0.05	<0.05	<0.05
Elixhauser comorbidity score, mean (SD)	2.1 (2.2)	2.3 (2.4)	3.0 (2.7)
Elixhauser conditions, %
0	25.0	24.4	14.8
1–2	41.7	39.9	36.8
≥3	33.3	35.7	48.4
VA primary care visits in 1 y prior, mean (SD)	4.5 (4.7)	3.7 (4.7)	5.7 (5.7)
VA ED or UCC visits in 1 y prior, mean (SD) †	0.5 (1.2)	0.4 (1.1)	1.6 (2.2)
Community care ED visits in 1 y prior, mean (SD)	1.1 (2.6)	0.6 (1.9)	0.9 (2.3)

Includes 15,159 (11%) Veterans who also used VA ED or UC services for low-acuity conditions during the study period.

Defined as VA ED or VA UCC visits regardless of visit acuity level.

ED indicates emergency department; PC, primary care; SC, service-connected disability; UC, urgent care; UCC, urgent care center.

Trends in Urgent Care Benefit Utilization

The number of UC benefit users and visits generally increased over time (Fig. 1). Across all months in our study period, there were 175,821 UC visits. After adjusting by state for the number of Veterans who used VA health care in FY19, we found that Arizona and South Carolina had the highest proportion of UC benefit users (Fig. 2). The mean charge of UC visits was $132 (SD=$135), and the charges for all UC visits in the first 9 months of the program totaled $23,273,792. During the study period, the vast majority of Veterans (80.4%) utilized the UC benefit only once, with an additional 14.6% making 2 visits. UC visits largely (83.7%) occurred during weekdays, with the highest volume of visits occurring on Mondays (18.3%). The most common reasons for UC visits were upper respiratory infections, back pain, and bronchitis (Table 2). The most common medical procedures and treatments associated with UC visits were chest radiograph, subcutaneous or intramuscular medication injection, influenza assay, Streptococcus group A assay, and urinalysis.

FIGURE 1

Number of Veterans and community urgent care visits by month, June 2019 to February 2020. UC indicates urgent care.

FIGURE 2

Veteran utilization of community urgent care by state, June 2019 to February 2020. To account for population differences, we calculated state-specific proportions by dividing the number of Veteran urgent care users residing in the state by the total number of Veterans in the state who used VA healthcare in FY19. Map created using Excel Office 365 Maps (Microsoft Corp).

TABLE 2

Most Common Conditions Associated With Community Urgent Care Visits, June 2019 to February 2020

Community UC Visits (n=175,821)
Rank	Condition Group*	%
1	Upper respiratory infections	17.9
2	Spondylosis; intervertebral disc disorders; other back problems	4.7
3	Acute bronchitis	4.5
4	Skin and subcutaneous tissue infections	4.4
5	Lower respiratory disease	4.3
6	Connective tissue disease	3.4
7	Sprains and strains	3.3
8	Nontraumatic joint disorders	3.3
9	Influenza	2.8
10	Otitis media and related conditions	2.5

Condition groups from the AHRQ Clinical Classifications Software (CCS) for ICD-10-CM diagnoses.

UC indicates urgent care.

Predictors of Urgent Care Benefit Use

First, we examined the predisposing, enabling, and need characteristics associated with UC benefit use. Veterans who had a driving time >60 minutes to a VA ED/UCC had twice the odds of seeking care at community UCCs compared with Veterans whose driving time was 30 minutes or less (odds ratio=2.08, 95% confidence interval: 2.05–2.11) (Fig. 3). In addition, being young, female, and White was associated with increased odds of UC benefit use. For every community ED visit paid for by VA in the prior year, there was 5% increased odds of utilizing the UC benefit.

FIGURE 3

Predictors of community urgent care utilization, odds ratios (OR) and 95% confidence intervals (95% CI). Drive time is in minutes and the prior use variables are the number of visits in the 1 year prior to the study period. ED indicates emergency department; PC, primary care; UC, urgent care; UCC, urgent care center; VA, Veterans Affairs. P-values for the odds ratios estimates were P<0.0001 except for prior VA PC use in Model 2 was P=0.0381.

Predictors of Setting Choice

A second logistic model compared UC benefit users to patients who only utilized VA EDs/UCCs for low-acuity conditions. Compared with patients who used VA EDs/UCCs, community UC users were more likely to live farther from a VA ED/UCC and be younger, female, housed, and urban dwelling (Fig. 3). Interestingly, living farther from VA primary care was associated with lower odds of using community UC. In addition, despite having no copay, Veterans with more disabilities (priority groups 1–5) were more likely to use VA care and had lower odds of utilizing the UC benefit. For every VA ED/UCC visit in the prior year, there was a 33% reduction in odds of using the UC benefit.

Sensitivity Analysis

Removing the term for driving time to VA primary care in both models had little effect on the odds ratios for the remaining terms in either model.

DISCUSSION

From June 2019 to February 2020, 138,305 Veterans made 175,821 UC visits under the new MISSION Act UC benefit, costing the VA $23,273,792. We found that the program’s reach increased over time but still was lower volume than VA delivered UC. Overall, 2.4% of the 5.9 million potentially eligible Veterans in our cohort utilized the UC benefit, while 7.3% visited a VA ED/UCC for a low-acuity condition during the same period.

Convenient operating hours of UC services is one likely driving factor behind the increasing use of the UC benefit. Prior studies have shown that nearly half of retail clinic visits occur after-hours.19,20 While the vast majority of UC visits in this study occurred during the weekdays, we do not know if these visits occurred after-hours. Consistent with other studies, we found younger patients had higher odds of utilizing the UC benefit compared with benefit non-users.21 Younger Veterans may prefer walk-in or online booking options for making appointments rather than through a call center, where wait times may be lengthy.22,23 Working hours may also increase difficulty in accessing usual care at convenient times.

UCCs can also expand geographic access points. We found that driving time to a VA ED/UCC was the greatest predictor of UC benefit use when compared with both benefit non-users and VA ED/UCC users. Rurality, paradoxically, was associated with lower odds of UC benefit use. In part, this may be driven by the location of UCCs, which tend to be in higher-income urban and suburban settings.24,25 Another potential explanation is patient preference. Campbell et al26 found that rural residents were more likely to delay care until their own doctor was available.

A patient’s personal preference for a medical care setting may also influence where they choose to receive acute care. Our results indicate that Veterans with prior history of VA ED/UCC visits continued to choose VA for low-acuity issues, regardless of the availability of the UC benefit. Along the same lines, Veterans with prior experiences at non-VA EDs were more likely to utilize the UC benefit. These results suggest that Veterans may have a prior preference for VA or non-VA ED care settings, and this preference could predict future use. Preference for VA may also explain why, despite having no co-pay for community UC use, Veterans assigned to priority groups associated with more disability and lower income were more likely to use VA. These patients may also be influenced by the convenience of using VA to streamline services in a single location and receive same day test results.27 Alternatively, we found female Veterans were much more likely to use the UC benefit compared with both nonbenefit users and VA users. Because of variations in availability of comprehensive women’s health services, female Veterans commonly rely on both VA and non-VA care.28 Our findings may reflect female Veterans’ familiarity with non-VA settings or may signal continued sex disparities in acute care services for female Veterans.29,30

Proper use of UCCs require patients to self-triage their symptoms and select an appropriate care venue. Similar to Wilson et al,31 our analysis suggests that patients are selecting UCCs for appropriate low-acuity conditions that can be resolved in one visit. Reassuringly, older Veterans and those with increasing comorbidity burdens are more likely to use a VA ED/UCC than a community UCC. While this decision may be driven by preference for and familiarity with VA facilities, it also likely also reflects medical necessity.

The impact that the UC benefit will have on care continuity is not yet known. One study has found that patients with retail clinic visits had fewer future visits to their primary care physicians.32 The new UC benefit may exacerbate care fragmentation for Veterans since community UCCs are not electronically linked to VA primary care practices and hospitals. In addition, guidelines state that UC providers should advise Veterans to follow up with their VA primary care providers after an UC encounter, and medical documentation is expected to be shared with the Veteran’s VA facility. However, these measures are not currently tracked and could potentially affect care continuity.

As usage of the UC benefit continues to grow, it is important to measure the benefit’s impact on ED utilization (in and outside of the VA) and the overall cost to VA. Evidence from prior studies have been mixed; providing access to UC has been shown to reduce health care costs, as UCCs often rely on less expensive staffing models and have lower fixed costs compared with EDs.33 For low-acuity conditions, evidence suggests that care delivered at UCCs cost less per episode than similar care delivered at primary care offices and EDs.3,34,35 However, despite these potential savings, total costs of care for VA could potentially increase if the UC benefit attracts patients that would not have otherwise sought medical attention and results in increased total utilization of care, as has been true in other settings.2

VA is currently undergoing a modernization of their Clinical Contact Centers (CCC),36 which will ultimately be available to Veterans 24 hours a day, 7 days a week.37 Intended as an alternative to ED/UCCs or primary care clinics for many low-acuity conditions, the CCCs will include services like nurse advice, triage, and virtual visits with providers. VA-wide implementation of CCCs is planned for late 2021 and could have significant implications for both UC benefit and VA ED/UCC use, particularly for Veterans facing temporal and geographic barriers to acute care.

Our findings should be interpreted in the context of our limitations. First, we elected to only include data from the first 9 months of the UC benefit program to allow for lag time in claims processing and reporting, and to avoid disruptions in care-seeking behavior because of COVID-19. As program uptake increases, future evaluations should examine if usage of UC has continued to grow over time. Second, because VA does not have its own robust network of UCCs, we created a comparison group of patients that use VA acute care by selecting those that made VA ED/UCC visits and had an ESI level 4 or 5—a commonly used method to identify low-acuity patients.13,38,39 The 3 most common ESI level 4 or 5 VA ED/UCC visits were for back pain, nontraumatic joint disorders, and upper respiratory infections, and we also found these to be common reasons for visits in the UC user group. Finally, although other access factors such as patient preferences, wait times, primary care access, having non-VA insurance, and other social determinants of health are important to consider, we focused our study on the predisposing, enabling, and need factors readily available with VA data.

In conclusion, the recent UC benefit included in the VA MISSION Act provides a new way to deliver unscheduled, low-acuity acute care to Veterans. Our results highlight characteristics of early adopters of the benefit. As the program continues to expand and evolve, impacts on care coordination, medication safety, outcomes, and shifts in utilization of VA primary and emergency services must be evaluated.