1. Introduction
Nearly three million veterans enrolled in the Veterans Health Administration (VA) live in rural areas, where timely access to emergency and urgent care (UC) can be challenging [1, 2]. Rural veterans often face long travel distances to reach VA emergency departments (EDs) or UC sites, with limited alternatives beyond community EDs [2, 3, 4]. The 2018 Maintaining Internal Systems and Strengthening Integrated Outside Networks (MISSION) Act expanded veterans’ access to community care—non‐VA medical services reimbursed by the VA when timely or geographically accessible VA care is unavailable [5]. Although community care offers important key access advantages, especially in acute settings, it also presents challenges related to continuity, oversight, and coordination [6, 7, 8]. Community care provides a critical safety net for veterans who cannot easily reach VA facilities, yet uncertainty surrounding future policy changes to rural provider networks and reimbursement models—including the potential closures of hospitals—raises concerns about the sustainability of this option for rural veterans [9].
To bridge gaps in access, the VA has modernized and expanded services offered through its long‐standing network of national call centers [10]. Historically, veterans with symptom‐based complaints could contact a nurse triage line for advice and care navigation. Following modernization, these call centers now provide real‐time clinical assessment and treatment, serving as a virtual entry point into acute care. Depending on symptom acuity, veterans may be referred to one of two national virtual acute care programs: Tele‐emergency care (TEC), which offers real‐time consultation with emergency physicians or advanced practice providers (APPs), or virtual care visits (VCV), which provide same‐day UC evaluation by video or phone with physicians or APPs from a variety of clinical backgrounds (Figure 1). These programs are particularly relevant for rural veterans, for whom geographic barriers to in‐person VA acute care remain substantial [4].
FIGURE 1.
Veterans who contact the VA nurse advice line are triaged by a registered nurse using a standardized symptom assessment algorithm. On the basis of the recommended follow‐up interval (RFI), callers are assigned to one of five acuity categories. Program guidance recommends that veterans triaged as emergent (0–2 h) be considered for referral to tele‐emergency care (TEC), which provides real‐time consultation with an emergency clinician. Veterans triaged as urgent (2–8 h) may be referred to either TEC or a virtual care visit (VCV), depending on clinical judgment, while those triaged as less urgent (beyond 12 h) are more typically directed to VCV for same‐day, scheduled urgent care evaluation. In practice, referral decisions reflect nurse and clinical discretion, and encounters may occur outside these recommended intervals. Callers assigned a 911 disposition represent the most acute presentations and are directed to emergency services; in some cases, these callers are connected to a TEC provider with the goal of encouraging ED evaluation for patients who are reluctant to call 911 or seek emergency care—a distinct clinical use case from standard virtual acute care delivery. These calls are excluded from the primary study cohort.
Early evaluations demonstrated the feasibility and positive early outcomes of TEC within the VA [11, 12, 13]. However, prior work has not comprehensively examined outcomes across both of VA’s national virtual acute care programs—TEC and VCV—nor has it focused specifically on rural veterans, a population that stands to benefit most given persistent geographic barriers to in‐person acute care. This study examines the reach of TEC and VCV among rural veterans and evaluates whether receiving virtual acute care is associated with subsequent ED visits, hospital admissions, mortality, and short‐term costs.
2. Methods
2.1. Data Sources and Sample
This retrospective study used data from the national call centers covering the period from January 2022 through September 2024. The analytic cohort included all calls made by rural veterans. We used VA Planning Systems Support Group (PSSG) data to identify rural veterans. PSSG data define rurality based on USDA Rural–Urban Commuting Area (RUCA) codes [14]. Patients with RUCA scores higher than 1.1 were considered rural [15]. Nurse triage calls that resulted in a 911 disposition were excluded, as these represent emergent situations outside the scope of virtual care. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for reporting observational research. The Stanford University Institutional Review Board determined that this project does not meet the federal definition of human subject research and is not under the purview of IRB oversight.
Demographic characteristics, comorbidities, prior care utilization, and mortality were obtained from the VA Corporate Data Warehouse (CDW), national repository integrating data from multiple VA clinical and administrative systems. Nurse triage calls were documented through an integrated telephone management system and stored as standardized clinical notes accessible within the CDW. Information on TEC and VCV encounters was extracted from the program’s Customer Relationship Management (CRM), a platform currently used by call centers to track and manage calls, and structured data field in the CDW. We incorporated utilization and cost data on non‐VA encounters reimbursed by VA from national community care data maintained by the Office of Integrated Veteran Care. We obtained inpatient, outpatient, and pharmacy cost data from the VA Managerial Cost Accounting (MCA) System. For each triage call, we calculated total costs (VA and community) incurred within 30 days of the index call date (including the date of the call).
2.2. Measures
All completed TEC or VCV encounters were identified regardless of triage acuity and grouped into two cohorts on the basis of the nurse‐assigned recommended follow‐up interval (RFI): acute (RFI 0–8 h) and non‐acute (RFI > 8 h). Within each cohort, veterans who completed a TEC or VCV encounter were compared with those who did not. Although RFI thresholds reflect program referral guidance, in practice some encounters occurred outside the intended acuity range, reflecting clinical discretion or workflow variation. We examined four short‐term outcomes: (1) any ED visit within 7 days of the index triage call; (2) any hospital admission within 7 days; (3) death from any cause within 30 days; and (4) health care costs within 30 days. Costs were further categorized into total, VA outpatient, VA inpatient, VA pharmacy, and VA‐purchased community care costs. Total costs reflected the sum of all VA and community care payments incurred within 30 days of the triage call, including the day of the call. The cost of the VCV reflected all costs reported for the clinic location and day of the visit.
2.3. Statistical Analysis
We first described the reach of virtual acute care among rural veterans, including the total number of triage calls, the distribution of triage acuity, and the number of completed TEC and VCV encounters within each acuity group. We then compared demographic and clinical characteristics between veterans who received virtual care and those who did not, separately within the acute and non‐acute cohorts, reporting means and proportions as appropriate.
To estimate the effect of virtual acute care encounters (TEC or VCV) on outcomes, we used a propensity‐score matching approach to adjust for observable differences between groups. We estimated propensity scores using covariates that included acuity (RFI), chief complaint, call timing (business hours, weekday, and hour of call), geographic region (the VA’s 18 Veterans Integrated Service Networks [VISNs]), age, sex, race, ethnicity, and VA priority group, which reflects service‐connected disability, income, and co‐pay eligibility [16]. We applied three complementary adjustment strategies:
-
Nearest‐neighbor matching, which paired each veteran who received virtual acute care with a similar veteran who did not (primary analysis).
-
Full matching, which grouped all observations into weighted subclasses to achieve optimal balance.
-
Inverse probability weighting (IPW), which reweighted the sample to create a pseudo‐population balanced on observed characteristics.
Covariate balance was assessed using standardized mean differences and visualized using distributional balance plots. After confirming adequate balance, we estimated treatment effects using linear probability models for each binary outcome. Coefficients are presented as percentage‐point differences in outcome probabilities between recipients and matched controls, alongside control group means for interpretability.
Robustness of findings was evaluated by comparing results across the three matching approaches. Because effect estimates were consistent in direction and magnitude, nearest‐neighbor matching results are presented throughout, with full matching and weighting results reported in Supporting Information Table A1.
As a secondary analysis, we explored whether the associations varied across chief complaint categories within each acuity group. We also examined whether findings generalized to urban veterans by repeating analyses in a parallel urban cohort, applying the same matching procedures and covariate adjustments. Effects were compared qualitatively between rural and urban cohorts to assess consistency in direction and magnitude. All analyses were performed at the call level using Stata version 18 and R version 4.1.1.
3. Results
3.1. Cohort Characteristics and Reach
Between January 2022 and September 2024, there were 1,951,514 nurse triage calls to VA’s national call centers, of which 131,603 (6.7%) were excluded due to 911 dispositions, leaving 1,819,911 calls for analysis. Of these, 631,437 (34.7%) were from rural veterans. Overall, 127,727 (20.2%) rural callers completed a TEC or VCV encounter, whereas the remaining 503,710 (79.8%) did not receive virtual acute care. Among rural callers, 61,408 completed a TEC or VCV encounter with a triage acuity of 0–8 h (acute group) and 66,319 completed an encounter with a triage acuity of >8 h (non‐acute group) (Supporting Information Figure A1). Among rural callers receiving VCV/TEC, the average cost of the virtual acute care encounter was $448 (SD $1040).
Rural veterans who called the nurse triage line were, on average, 63.5 years old (SD 16.0). The majority were male (87.4%) and White (83.1%), with smaller proportions identifying as Black (7.9%) or Native American (1.1%). Over half (52.6%) were assigned to VA Priority Group 1, the highest disability rating category. Most calls (77.0%) occurred during weekday business hours. For calls with a documented (non‐missing) chief complaint, the most common presenting complaints were cough or congestion (20.6%), extremity problems (14.6%), genitourinary issues (8.9%), neck/back pain (8.6%), abdominal pain (7.8%), and chest pain (5.8%) (Table 1).
TABLE 1.
Characteristics of rural veterans by virtual acute care group and acuity, January 2022–September 2024.
Acute (RFI 0–8 h)
Non‐acute (RFI >8 h)
Characteristic
All triage calls, n = 631,437
Received TEC/VCV, n = 61,408
Did not receive TEC/VCV, n = 263,763
p value
Received TEC/VCV, n = 66,319
Did not receive TEC/VCV, n = 239,947
p value
Caller characteristics
Age, mean (SD)
63.5 (16.0)
64.4 (15.8)
63.6 (15.8)
<0.001
62.6 (16.4)
63.4 (16.1)
<0.001
Sex, n (%)
0.005
<0.001
Female
79,447 (12.6)
8081 (13.2)
33,465 (12.7)
9282 (14.0)
28,619 (11.9)
Male
551,868 (87.4)
53,313 (86.8)
230,252 (87.3)
57,022 (86.0)
211,281 (88.1)
Race, n (%)
<0.001
<0.001
White
525,029 (83.1)
51,559 (84.0)
217,606 (82.5)
55,259 (83.3)
200,605 (83.6)
Black
50,158 (7.9)
3534 (5.8)
24,107 (9.1)
3742 (5.6)
18,775 (7.8)
Asian
3129 (0.5)
396 (0.6)
1064 (0.4)
562 (0.8)
1107 (0.5)
NHOPI
4863 (0.8)
527 (0.9)
1910 (0.7)
557 (0.8)
1869 (0.8)
AIAN
6801 (1.1)
699 (1.1)
2893 (1.1)
699 (1.1)
2510 (1.0)
Unknown
41,457 (6.6)
4693 (7.6)
16,183 (6.1)
5500 (8.3)
15,081 (6.3)
Ethnicity, n (%)
<0.001
<0.001
Hispanic or Latino
26,298 (4.2)
3739 (6.1)
10,649 (4.0)
4009 (6.0)
7901 (3.3)
Not Hispanic or Latino
585,062 (92.7)
55,506 (90.4)
245,281 (93.0)
59,788 (90.2)
224,487 (93.6)
Unknown
20,077 (3.2)
2163 (3.5)
7833 (3.0)
2522 (3.8)
7559 (3.2)
VA priority group, n (%)
<0.001
<0.001
1
332,025 (52.6)
31,939 (52.0)
143,372 (54.4)
34,236 (51.6)
122,478 (51.0)
2
40,874 (6.5)
3639 (5.9)
16,319 (6.2)
4303 (6.5)
16,613 (6.9)
3
76,684 (12.1)
6716 (10.9)
29,829 (11.3)
7890 (11.9)
32,249 (13.4)
4
10,465 (1.7)
1058 (1.7)
5010 (1.9)
897 (1.4)
3500 (1.5)
5
88,543 (14.0)
9417 (15.3)
38,057 (14.4)
8980 (13.5)
32,089 (13.4)
6
15,895 (2.5)
1567 (2.6)
5637 (2.1)
2017 (3.0)
6674 (2.8)
7
13,625 (2.2)
1462 (2.4)
4969 (1.9)
1767 (2.7)
5427 (2.3)
8
49,154 (7.8)
4971 (8.1)
19,059 (7.2)
5534 (8.3)
19,590 (8.2)
Missing
4172 (0.7)
639 (1.0)
1511 (0.6)
695 (1.0)
1327 (0.6)
Triage call characteristics
Weekday business hours, n (%)
485,982 (77.0)
47,602 (77.5)
189,713 (71.9)
<0.001
51,974 (78.4)
196,693 (82.0)
<0.001
Recommended follow‐up interval (RFI), n (%)
<0.001
<0.001
Now (emergent)
204,776 (32.4)
28,300 (46.1)
176,476 (66.9)
—
—
2–8 h (urgent)
120,395 (19.1)
33,108 (53.9)
87,287 (33.1)
—
—
>8–24h
150,575 (23.8)
—
—
41,199 (62.1)
109,376 (45.6)
2–3 days
84,051 (13.3)
—
—
15,851 (23.9)
68,200 (28.4)
1–2 weeks
40,855 (6.5)
—
—
4140 (6.2)
36,715 (15.3)
Self‐care
20,871 (3.3)
—
—
3654 (5.5)
17,217 (7.2)
Other/Missing
9914 (1.6)
—
—
1475 (2.2)
8439 (3.5)
Top chief complaints, n (%)
a
Abdominal pain
39,056 (7.8)
3739 (7.5)
21,234 (10.3)
2116 (3.8)
11,967 (6.4)
Chest pain
28,772 (5.8)
2740 (5.5)
14,440 (7.0)
1910 (3.4)
9682 (5.2)
Cough/Congestion
102,730 (20.6)
13,145 (26.5)
27,628 (13.4)
22,840 (41.0)
39,117 (20.8)
Extremity problem
72,969 (14.6)
7099 (14.3)
34,367 (16.7)
5267 (9.5)
26,236 (14.0)
Genitourinary issues
44,444 (8.9)
4886 (9.8)
21,229 (10.3)
4261 (7.7)
14,068 (7.5)
Neck/Back pain
42,724 (8.6)
3888 (7.8)
18,040 (8.8)
3579 (6.4)
17,217 (9.2)
Within the acute cohort, veterans who completed a TEC or VCV encounter differed from those who did not in several ways (Table 1). They were less likely to present with emergent acuity (46.1% vs. 66.9%, p < 0.001), more likely to call during business hours (77.5% vs. 71.9%, p < 0.001) and more likely to present with upper respiratory complaints (26.5% vs. 13.4%). They were less likely to present with chest pain and abdominal pain.
Within the non‐acute cohort, those who received virtual care were more likely to be female (14.0% vs. 11.9%, p < 0.001) and more likely to identify as Hispanic or Latino (6.0% vs. 3.3%, p < 0.001) compared with those who did not. Non‐acute recipients were also more likely to present with cough or congestion (41.0% vs. 20.8%) and less likely to present with extremity problems (9.5% vs. 14.0%), neck or back pain (6.4% vs. 9.2%), abdominal pain (3.8% vs. 6.4%), and chest pain (3.4% vs. 5.2%). Unlike the acute cohort, non‐acute recipients were slightly less likely to call during business hours than non‐recipients (78.4% vs. 82.0%, p < 0.001). Overall, 10.7% of all VA and community ED visits (n = 313,143) made by rural veterans were preceded by a nurse triage call, compared with 9.4% of ED visits (n = 611,182) among urban veterans.
3.2. Unadjusted Outcomes
In unadjusted analyses, rural veterans who received a virtual acute care encounter (TEC or VCV) had substantially lower rates of subsequent acute care utilization compared with those who did not (Table 2). Among acute callers, 29.5% of virtual care recipients visited an ED within 7 days compared with 48.3% of non‐recipients. Hospitalization within 7 days was also less frequent among recipients (3.8% vs. 5.2%), and UC visits were lower (6.0% vs. 11.6%). Among non‐acute callers, patterns were consistent but attenuated, reflecting the lower baseline acuity of this group: ED visits occurred in 11.8% of recipients versus 13.4% of non‐recipients, and hospitalizations were 1.2% versus 1.3%. Primary care visits within 21 days and 30‐day mortality remained low and comparable between groups in both cohorts.
TABLE 2.
Unadjusted outcomes and costs following triage calls, by acuity group.
Acute RFI (0–8 h)
Received TEC/VCV, n = 61,408
Did not receive TEC/VCV, n = 263,763
p value
Outcomes, n (%)
ED visit within 7 days
18,108 (29.5)
127,285 (48.3)
<0.001
UC visit within 7 days
3674 (6.0)
30,662 (11.6)
<0.001
PC visit within 21 days
4955 (8.1)
28,377 (10.8)
<0.001
Hospital admission within 7 days
2326 (3.8)
13,587 (5.2)
<0.001
30‐day mortality
352 (0.6)
1593 (0.6)
0.40
Costs ($), mean (SD)
a
Total VA
6300 (14,922)
6307 (14,199)
<0.001
VA inpatient
1365 (10,522)
1488 (9977)
<0.001
VA outpatient
3228 (4046)
2810 (3845)
<0.001
VA pharmacy
359 (2007)
358 (2101)
<0.001
Total community care
1348 (6621)
1652 (6504)
<0.001
Non‐acute RFI (>8 h)
Received TEC/VCV n = 66,319
Did not receive TEC/VCV n = 239,947
p value
Outcomes, n (%)
ED visit within 7 days
7800 (11.8)
32,065 (13.4)
<0.001
UC visit within 7 days
3526 (5.3)
25,651 (10.7)
<0.001
PC visit within 21 days
6811 (10.3)
34,312 (14.3)
<0.001
Hospital admission within 7 days
809 (1.2)
3027 (1.3)
0.40
30‐day mortality
180 (0.3)
659 (0.3)
0.90
Costs ($), mean (SD)
a
Total VA
4298 (10,045)
3802 (9612)
<0.001
VA inpatient
587 (6755)
554 (6313)
0.50
VA outpatient
2736 (3369)
2249 (3159)
<0.001
VA pharmacy
255 (1501)
248 (1836)
<0.001
Total community care
720 (4535)
751 (4203)
<0.001
Within 30 days of the triage call, mean total costs were slightly lower among virtual care recipients in the acute group ($6300 vs. $6307) and higher among recipients in the non‐acute group ($4298 vs. $3802) (Table 2). Across both cohorts, outpatient costs were higher among recipients while community care costs were lower.
3.3. Adjusted Outcomes
After propensity‐score adjustment and nearest‐neighbor matching, receipt of a TEC or VCV encounter was associated with statistically significant reductions in short‐term acute care use in both cohorts (Table 3). Among acute callers, veterans who received virtual care had an 11.87 percentage point (pp) lower probability of an ED visit within 7 days compared with matched controls (95% CI, −12.47 to −11.26; p < 0.001), and a 0.55 pp lower probability of hospitalization (95% CI, −0.81 to −0.28; p < 0.001). Among non‐acute callers, ED visits were 3.62 pp lower (95% CI, −4.02 to −3.22; p < 0.001), and hospitalizations were 0.15 pp lower (95% CI, −0.28 to −0.02; p = 0.026). Thirty‐day mortality did not differ significantly in both acute and non‐acute groups.
TABLE 3.
Association between virtual acute care (tele‐emergency care [TEC] or virtual care visits [VCV]) and outcomes, by acuity group.
Outcome
Adjusted estimate
a
95% CI
p value
Control mean
b
Acute RFI (0–8 h)
Utilization and mortality outcomes (percentage‐point difference)
ED visit within 7 days
−11.87
(−12.47, −11.26)
<0.001
48.1%
Hospital admission within 7 days
−0.55
(−0.81, −0.28)
<0.001
5.0%
30‐day mortality
0.09
(−0.01, 0.20)
0.064
0.59%
Costs (mean dollar difference, 30‐day)
Total VA
$549
($359, $740)
<0.001
$6168
VA outpatient
$517
($464, $570)
<0.001
$2772
VA inpatient
$223
($92, $354)
<0.001
$1425
VA pharmacy
−$3
(−$31, $25)
0.834
$350
Total community care
−$188
(−$282, −$94)
<0.001
$1622
Non‐acute RFI (>8 h)
Utilization and mortality outcomes (percentage‐point difference)
ED visit within 7 days
−3.62
(−4.02, −3.22)
<0.001
13.5%
Hospital admission within 7 days
−0.15
(−0.28, −0.02)
0.026
1.2%
30‐day mortality
0.00
(−0.06, 0.06)
0.915
0.26%
Costs (mean dollar difference, 30‐day)
Total VA
$386
($274, $499)
<0.001
$3751
VA outpatient
$469
($430, $509)
<0.001
$2226
VA inpatient
$20
(−$54, $94)
0.602
$543
VA pharmacy
−$5
(−$23, $14)
0.632
$243
Total community care
−$98
(−$150, −$45)
<0.001
$739
Adjusted 30‐day costs were modestly higher among virtual care recipients in both cohorts. Among acute callers, total costs were $549 higher (95% CI, $359–$740; p < 0.001), driven by increased VA outpatient spending (+$517; 95% CI, $464–$570; p < 0.001) and higher inpatient costs (+$223; 95% CI, $92–$354; p < 0.001). Pharmacy costs did not differ significantly. Community care spending was significantly lower (−$188; 95% CI, −$282 to −$94; p < 0.001). Among non‐acute callers, total costs were $386 higher (95% CI, $274–$499; p < 0.001), again driven by outpatient spending (+$469; 95% CI, $430–$509; p < 0.001), with inpatient and pharmacy costs nonsignificant. Community care costs were also lower in the non‐acute group (−$98; 95% CI, −$150 to −$45; p < 0.001).
3.4. Findings by Chief Complaint
Across all chief complaints in the acute cohort, receipt of a virtual acute care encounter was associated with substantially lower 7‐day ED visit rates (all p < 0.001), with reductions ranging from −9.76 to −18.39 pp (Table 4). The largest reductions were seen for stool issues, neck/back pain, and shortness of breath. In the non‐acute cohort, ED visit reductions were smaller but significant for most complaints; abdominal pain, chest pain, and stool issues showed no significant reductions in this group (Supporting Information Table A2).
TABLE 4.
Association between virtual acute care (tele‐emergency care [TEC] or virtual care visits [VCV]) and short‐term outcomes by chief complaint among rural veterans—acute group (recommended follow‐up interval [RFI] 0–8 h).
ED visit within 7 days
Hospital admission within 7 days
30‐day mortality
Chief complaint
Adjusted estimate (95% CI)
a
p value
Adjusted estimate (95% CI)
a
p value
Adjusted estimate (95% CI)
a
p value
Abdominal pain
−9.76 (−12.16, −7.37)
<0.001
−1.21 (−2.54, 0.12)
0.075
0.81 (0.33, 1.30)
0.001
Chest pain
−11.13 (−13.92, −8.35)
<0.001
−0.76 (−2.02, 0.50)
0.239
−0.07 (−0.42, 0.29)
0.708
Cough/Congestion
−11.33 (−12.38, −10.29)
<0.001
−0.12 (−0.49, 0.24)
0.507
−0.03 (−0.17, 0.11)
0.688
Extremity problem
−11.52 (−13.14, −9.90)
<0.001
−0.14 (−0.72, 0.45)
0.644
0.01 (−0.23, 0.26)
0.915
Genitourinary issues
−12.14 (−14.14, −10.13)
<0.001
−0.85 (−1.71, 0.01)
0.052
−0.05 (−0.38, 0.28)
0.756
Neck/Back pain
−14.77 (−17.03, −12.51)
<0.001
−0.51 (−1.23, 0.20)
0.159
0.19 (−0.10, 0.48)
0.199
Shortness of breath
−13.90 (−17.41, −10.40)
<0.001
−2.65 (−4.90, −0.39)
0.022
−0.71 (−1.79, 0.36)
0.194
Skin problem
−12.17 (−14.77, −9.57)
<0.001
−1.38 (−2.36, −0.41)
0.005
−0.29 (−0.60, 0.03)
0.076
Stool issues
−18.39 (−21.43, −15.36)
<0.001
−0.67 (−2.36, 1.01)
0.433
−0.48 (−1.11, 0.15)
0.135
Vision/Eye problems
−11.81 (−14.97, −8.65)
<0.001
−0.09 (−1.08, 0.89)
0.857
−0.04 (−0.23, 0.15)
0.715
Hospital admission reductions were more selective. In the acute cohort, the largest significant reductions were observed for shortness of breath and skin problems. Most other acute complaints showed nonsignificant admission differences. In the non‐acute cohort, significant reductions were limited to extremity problems and neck/back pain. Thirty‐day mortality did not differ significantly for most chief complaints in either cohort. One exception was abdominal pain in the acute group, where mortality was slightly but significantly higher among virtual care recipients.
Across both cohorts and all presenting complaints, outpatient costs were consistently higher among virtual care recipients, whereas community care costs were lower, suggesting substitution of VA‐based outpatient care for non‐VA acute services (Supporting Information Table A3). The largest community care reductions in the acute group were observed for shortness of breath and genitourinary issues. Total costs were significantly higher for a small number of conditions: cough/congestion, extremity problems, neck/back pain, and abdominal pain in the acute group, and cough/congestion, abdominal pain, and stool issues in the non‐acute group.
3.5. Comparison With Urban Veterans
Uptake of virtual acute care was somewhat higher among urban than rural veterans, with 22.8% of acute urban callers and 26.6% of non‐acute urban callers completing a TEC or VCV encounter, compared with 18.9% and 21.7%, respectively, among rural callers (Supporting Information Figures A1 and A2). Despite this difference in uptake, the direction and pattern of findings were consistent across settings (Supporting Information Table A4). Among urban veterans in the acute cohort, virtual care was associated with a 10.01 pp reduction in ED visits within 7 days (p < 0.001) and a 1.08 pp reduction in hospitalizations (p < 0.001), whereas 30‐day mortality did not differ significantly. In the non‐acute cohort, ED visit reductions were similar in magnitude to the rural non‐acute group (−3.97 pp, p < 0.001), and hospital admissions were also lower (−0.17 pp, p < 0.001).
Cost patterns were broadly consistent with rural findings. Total 30‐day costs were higher among urban virtual care recipients in both the acute (+$585; p < 0.001) and non‐acute (+$564; p < 0.001) groups, driven by increased outpatient spending. Community care costs were lower across both acuity groups, mirroring the rural pattern of redistribution toward VA‐provided services. The primary difference from rural findings was in VA inpatient costs, which were nonsignificant in the urban acute group but significantly higher in the urban non‐acute group (+$118; p < 0.001), a pattern not observed among rural non‐acute callers.
4. Discussion
In this national evaluation of VA’s virtual acute care programs, completion of a TEC or VCV encounter was associated with substantially lower short‐term emergency and hospital utilization among rural veterans across both acuity groups. Among acute callers, veterans who received virtual care were 11.87 pp less likely to visit an ED and 0.55 pp less likely to be hospitalized compared with matched controls; among non‐acute callers, reductions were smaller but consistent (−3.62 and −0.15 pp, respectively). These reductions spanned most chief complaints, including traditionally high‐acuity conditions such as chest pain and shortness of breath, and were achieved without significant increases in 30‐day mortality in either group. Importantly, similar patterns were observed among urban veterans, indicating that the benefits of virtual acute care extend beyond geographic context and reflect a scalable model for improving timely access and care coordination across the VA system. Together, these findings suggest that VA’s national virtual acute care programs are associated with reductions in near‐term reliance on in‐person, unplanned acute care services, without evidence of increased mortality risk.
The consistency of findings across rural and urban settings raises important questions about the mechanisms driving these associations. Although reduced travel burden is often cited as a primary motivation for virtual care among rural populations [4, 17], the similar magnitude of ED reductions across settings suggests this explanation may be incomplete. Veterans classified as urban may still live far from a VA ED, and future work examining drive time would help clarify whether distance is a meaningful pathway. Beyond geography, two other mechanisms may help explain our findings. First, timely clinician‐level assessment may reduce clinical uncertainty at the point of triage and reduce ED visits regardless of location. For example, a veteran calling about chest pain who is reassured by an emergency physician that their symptoms are musculoskeletal and given a clear follow‐up plan may be safely managed without an in‐person visit. Second, the virtual encounter may function as a care navigation tool that facilitates appropriate follow‐up (e.g., primary care follow‐up) and reduces unnecessary acute care‐seeking. Disentangling these pathways will be important for understanding where virtual acute care has the greatest impact.
Despite significant reductions in ED and inpatient utilization, these improvements did not translate into lower overall costs in either acuity group. Instead, total costs were modestly higher among virtual care recipients in both the acute (+$549) and non‐acute (+$386) groups. The rise in costs most likely reflects two components: the cost of the virtual encounter itself (on average, $448) and subsequent VA‐based follow‐up care prompted by the virtual visit, such as referrals for laboratory testing, imaging, or specialist evaluation. This pattern parallels findings from prior evaluations of direct‐to‐consumer and urgent telemedicine models outside VA, which have shown that greater convenience can increase downstream utilization and shift, rather than reduce spending [18, 19]. Conversely, studies of integrated delivery systems demonstrate that virtual UC can achieve cost neutrality or modest savings when deployed within coordinated systems that can leverage downstream efficiencies [20, 21]. That this redistribution was consistent across rural and urban settings and both acuity groups suggests it reflects a systemic feature of how virtual care integrates into VA’s delivery model. From a policy standpoint, this reduction in community care spending is meaningful given ongoing concerns about the rising costs of community care under the MISSION Act and the growing fragility of rural provider networks amid hospital closures and mergers [1, 9, 22]. A full cost‐effectiveness analysis, accounting for the direct cost of delivering TEC and VCV and longer term utilization patterns, will be needed to assess the value proposition of these programs, particularly in resource‐constrained rural settings.
Historically, nurse triage lines have served primarily as navigation tools—providing advice and directing patients to appropriate care settings—but few have integrated clinicians capable of diagnosing and managing acute conditions in real time [23, 24]. Research on nurse‐only telephone triage, including a recent scoping review of nurse triage in primary care, indicates that these programs demonstrate acceptable safety and quality but have shown limited and inconsistent effects on downstream ED utilization, in part, because nurses typically lack the authority to diagnose, prescribe, or initiate treatment during the call itself [25]. A recent large‐scale VA evaluation illustrated this gap directly: Among callers triaged as urgent or emergent, veterans followed ED referral recommendations only 59% of the time, and receipt of TEC was associated with a 16.5 pp reduction in 7‐day ED visits compared with nurse triage alone (28.5% vs. 45.0%), suggesting that embedding a clinician within the triage encounter meaningfully changes the care trajectory [13].
The VA’s TEC and VCV programs represent an important evolution of this model, moving beyond navigation to deliver same‐day clinical evaluation and treatment within the call itself. The present study is the first to examine both virtual acute care programs together specifically among rural veterans, who face unique barriers to in‐person access. Our findings extend prior VA work [11, 12, 13] by demonstrating clinically meaningful reductions in downstream utilization across both TEC and VCV and across two distinct acuity groups, including, in the acute cohort, traditionally high‐acuity presentations such as chest pain, abdominal pain, and shortness of breath that are typically directed to the ED under standard triage protocols. Outside of integrated systems like Kaiser Permanente, which has shown that physician‐directed tele‐triage for chest pain can safely reduce ED referrals without increasing mortality, few health systems have extended virtual care to this level of acuity [26]. Our findings suggest that virtual acute care may safely extend to conditions traditionally managed in the ED, underscoring its potential to expand the scope of telehealth in acute care.
Even with encouraging results, the reach of virtual acute care remains limited, and its distribution raises important questions about program targeting. In our cohort, completed TEC and VCV encounters were roughly equally split between the acute and non‐acute group yet the clinical benefits were substantially more pronounced among acute callers. This pattern suggests that current referral practices may not be optimally directing virtual care to the veterans who stand to benefit most. Nurses may be more comfortable referring less urgent presentations to virtual care, while acute callers may be more likely to be directed immediately to the ED rather than offered a virtual evaluation first [27]. On the patient side, veterans with more severe symptoms may be less willing to accept virtual care or may bypass the call center entirely. Strengthening referral guidance and training to encourage virtual evaluation for higher acuity presentations, particularly conditions like shortness of breath, chest pain, and stool issues, where reductions were largest, could meaningfully expand the impact of these programs. At the same time, evaluating patient experiences among those who declined or did not complete a virtual encounter will be important to ensure these programs are meeting veterans’ needs across the acuity spectrum.
Low overall engagement with the nurse triage system represents a broader and more fundamental gap. Only about one in ten VA or community ED visits among rural veterans was preceded by a triage call, suggesting that most veterans seek acute care directly rather than contacting the call center first. This finding is consistent with our prior work showing that veterans who do use VA call centers are overwhelmingly satisfied with the service, yet many remain unaware that it exists [28]. Broader research similarly highlights challenges in virtual care engagement [29]. The VA’s Virtual Care State of the Art conference also identified “access” and “engagement” as top priorities for future research and improvement, emphasizing that veterans must first be aware of, able to access, and willing to use virtual care before its benefits can be realized [30]. Increasing awareness and promoting a “call‐first” culture could expand the reach of virtual acute care and ensure that, when appropriate, veterans benefit from timely triage and treatment before turning to the ED.
This study should be interpreted in the context of its limitations. First, as an observational analysis, our findings may be subject to residual confounding despite the use of propensity‐score matching and multiple adjustment methods. Specifically, unmeasured factors such as patient preference for in‐person versus virtual care, symptom evolution after the triage call, individual technology access, and nurse or clinician discretion in referral decisions may influence both the likelihood of completing a virtual encounter and subsequent utilization and cannot be fully accounted for through propensity‐score methods alone. Accordingly, our findings should be interpreted as associations rather than causal effects. Second, reliance on administrative data may introduce misclassification or incomplete capture of variables such as symptom severity or patient preference, and variation in nurse documentation and triage decisions could affect eligibility classification and referral patterns. Third, our outcome measures reflect short‐term utilization, mortality, and costs, which do not capture longer term effects on health status. Future work should evaluate whether virtual acute care influences ongoing care engagement and downstream utilization patterns over time. Fourth, our cost analyses compare cumulative cost differences between veterans who received acute virtual care and those who did not rather than directly measuring the causal pathway from virtual encounter to downstream spending/utilization. A more granular claims‐based analysis linking individual encounters to specific costs would be needed to fully understand how virtual acute care encounters impact care costs. Finally, while our analysis included both VA and community encounters reimbursed by VA, we could not identify care obtained entirely outside the VA system and not billed to VA. Despite these limitations, this study provides the first comprehensive national evaluation to date of VA’s virtual acute care programs and their association with short‐term outcomes among rural veterans.
