Management and Outcomes of Patients Undergoing Cardiovascular Evaluation Across Health Care Systems: Comparison of Community Care and Integrated Veterans Affairs Health Care

Abstract

Background

The Veterans Affairs (VA) Healthcare System maintains the largest integrated health system in the United States but also supports fee‐for‐service insurance for veterans receiving care in community facilities outside the VA. We sought to evaluate the management and outcomes of patients referred for consultation in either venue, using cardiovascular evaluation as a model.

Methods

We conducted a retrospective cohort study identifying patients enrolled in the VA Healthcare System referred for cardiovascular evaluation from October 2020 through September 2024 and stratified the population based on the venue in which evaluation was completed. The primary outcome was major adverse cardiovascular events (acute coronary syndromes/stroke/mortality) in a matched population.

Results

Among 235 197 consultations for cardiovascular evaluation, 201 453 were completed in the chosen venue within 6 months. The time between consultation and evaluation was similar across venues (community, 35 days [95% CI, 17–65] versus VA, 33 days [95% CI, 19–53]), with comparable delays to diagnostic testing or therapeutic interventions. Patients receiving care in the community were more likely to undergo stress testing (43.2% versus 36.4%, P=1.5×10⁻⁴⁶) and coronary angiography (23.1% versus 17.4%, P=2.1×10⁻⁵¹) within 2 years compared with those treated in the VA Healthcare System. Despite this, patients treated in the community had a significantly higher rate of major adverse events at 2 years (17.6% versus 15.3%, P=5.9×10⁻¹⁰) compared with those treated in the VA Healthcare System.

Conclusions

Patients undergoing cardiovascular evaluation in community practices were not evaluated more rapidly than those seen in the VA, though they were more likely to receive initial and repeat diagnostic testing. Adverse events were more common among community‐treated patients than those in the VA, suggesting an opportunity to optimize access to care while improving clinical outcomes.

Nonstandard Abbreviation and Acronym

VA

Veterans Affairs

Clinical Perspective.

What Is New?

• There is significant heterogeneity in the evaluation and management of veterans with presumed coronary artery disease across different treatment venues.

• Veterans evaluated in the community had an increased rate of testing and adverse events compared with matched patients who were treated in the Veterans Affairs Healthcare system.

What Are the Clinical Implications?

• There may be an opportunity to optimize the community care program to enhance access for cardiovascular care for veterans while maintaining the highest possible quality.

The Department of Veterans Affairs (VA) maintains the largest integrated health care system in the United States, comprising hundreds of geographically distributed clinics affiliated with dozens of tertiary care medical centers located throughout the country that provide coordinated and integrated health care for its patients. ¹ Funding for this closed system is dictated by congressional legislation and distributed to the facilities in a form of capitation that then allows for the employment of salaried clinicians. ² Previous research has demonstrated that this system provides high‐quality care to a complex patient population, ³ , ⁴ , ⁵ , ⁶ though there have been challenges in providing this care in a timely manner ⁷ , ⁸ near the homes of some veterans. ⁹

The VA Healthcare System also serves as an insurance payer for care performed in non‐VA facilities in the community, termed “community care.” ¹⁰ Patients enrolled in the VA Healthcare System may receive care from nonfederal practitioners with the expenses covered by the government when similar services cannot be provided in a timely manner near the home of a Veteran. ⁹ Expansions in the program that oversees community care via the Veterans Access, Choice and Accountability Act of 2014 and VA MISSION Act of 2018 have led to a significant increase in the proportion of patients enrolled in the health care system who now receive their comprehensive care with fee‐for‐service community providers outside VA. ⁶ , ¹¹ The development of this system provides a unique opportunity to explore care delivery for the same patient population across different health care environments with divergent payment structures.

A comprehensive cardiovascular evaluation for a new patient with stable coronary artery disease is a common reason for referral to specialist care, but it can include significant variation in the diagnostic investigations and therapeutic interventions that are ultimately employed. The varying approach to this patient population can be influenced by different reimbursement patterns, whereby heavily reimbursed testing or therapies may be increased or repeated in certain environments. With this in mind, the present retrospective cohort analysis compared the management and outcomes of patients referred for comprehensive cardiovascular care with VA versus non‐VA community care providers. This analysis provides a contemporary view to the quality of care veterans receive across different venues with divergent levels of integration and reimbursement.

METHODS

The data that support the findings of this study are available from the corresponding author upon reasonable request, though they will be subject to the stringent data privacy rules of the VA Healthcare System and the US government.

Population

The present retrospective cohort study identified patients who have previously received their care within the VA Healthcare System and subsequently had a completed cardiovascular evaluation for a presumptive diagnosis of stable coronary artery disease from October 2020 through September 2024 (Table S1). For the purposes of this analysis, an active user of the health care system was defined as a patient with at least 2 encounters and 1 medication fill in the prior 2 years. A completed evaluation was defined as a referral for cardiovascular care followed by an evaluation with a cardiologist in the chosen venue within 6 months. This definition ensures that incomplete consults that were deemed unnecessary and subsequently discontinued were removed. Further, those who had a clinical status change requiring an urgent visit in the non‐chosen venue would also be excluded from analysis. For example, we excluded patients who suffered a major adverse cardiovascular event (death/acute coronary syndromes/stroke) that occurred between the consult request and the cardiovascular evaluation as this likely would change subsequent management. To ensure patients included in the cohort were from a geographic location that had an opportunity to receive care in either venue, the analytic cohort was limited to patients enrolled at sites where no less than 10% or no more than 90% of the cardiology consultations were seen at community facilities. For those receiving an index evaluation in the community, we excluded patients who were referred outside of the VA Healthcare System because it was deemed in their best medical interest and included only patients who were evaluated because of the distance to a VA facility. The analysis was performed in an operational capacity for the Department of Veterans Affairs, and neither institutional review board approval nor informed consent was deemed necessary by the local authorities.

Measures

All data were ascertained from the VA Corporate Data Warehouse and the Consolidated Data Set. Medical comorbidities were based on administrative diagnosis and procedure codes from the VA Healthcare System, requiring at least 1 inpatient or 2 outpatient encounters with the codes of interest within 2 years before the clinic evaluation. The cohort was divided into 2 groups based on the venue in which comprehensive cardiovascular evaluation was requested and initial clinic evaluation completed, either community care or VA direct care. Postclinic evaluation testing included noninvasive diagnostic studies such as transthoracic echocardiography, stress testing, or computed tomography, and invasive diagnostic studies such as coronary angiography. Therapeutic interventions included percutaneous coronary intervention (Table S1). Fees for professional services provided in the community setting associated with the Current Procedural Terminology codes of interest were collected from paid claims. All Current Procedural Terminology‐related professional fees were summed within a given claim for the study or intervention of interest to provide a billed cost for that professional service. To evaluate process metrics that could reflect the quality of care provided, the proportion of patients who filled a guideline‐indicated prescription for coronary artery disease was assessed using VA outpatient pharmacy data. For this analysis, the index clinic evaluation as the baseline when measuring time to fill for all medications except P2Y12, which were measured from the date of a procedure (percutaneous coronary intervention) that would mandate that therapy.

Covariates and Outcomes

The primary outcome was a major adverse cardiovascular event, defined as a composite of death, hospitalization, or procedures for acute coronary syndromes and stroke at 1 and 2 years following the initial evaluation (Table S1). Secondary outcomes included the rates and timing of diagnostic testing or therapeutic interventions at 1 and 2 years following the initial clinical evaluation. Adjustment covariates included patient demographics, medical history, and prior medication usage. Given that distance to VA facilities is a major determinant of care venue, adjustment for patient straight‐line distances was also performed to ensure that included patients would be similarly likely to use VA direct care or community care.

Statistical Analysis

Absolute standardized differences were used to assess differences in demographics and comorbidities based on treatment venue. To balance these characteristics, propensity score matching was used for our population. Propensity scores were estimated with a generalized boosted regression model using the ‘gbm’ package and controlled for the observed variables listed in Table 1. ¹² Patients were first matched within the same VA facility where the index consultation originated from using 1:1 matching. We then performed 1:1 matching among all remaining unmatched patients without consideration of VA facility. To assess the time from index clinical evaluation to outcomes and testing, we estimated survival curves and standard errors based on treatment venue from Cox proportional hazards models using the matched population and tested for absolute group differences in survival rates at select time points. ¹³ We estimated the median time from original referral for cardiovascular care to first test along with the interquartile range, as well as the time to a second test of the same type, from a rematched subpopulation of patients with at least 1 test and 1 year of follow‐up. With nonparametric bootstrapping, resampling observations with replacement, the difference in median time to test based on treatment venue was calculated using the mean along with the 2.5th and 97.5th percentiles across all 1000 samples to summarize group differences with 95% CIs. For costs analyses, we summarized costs from the first event after the index clinic visit with medians (interquartile range). The effect of a hypothetical confounder was assessed using a simulation (Data S1).

Table 1.

Demographic and Clinical Characteristics of Patients Referred for an Index Comprehensive Cardiovascular Evaluation for Presumed Coronary Artery Disease

	Unweighted population			Matched population
	VA	Community	SD	VA	Community	SD
	N=72 687	N=50 510		Weighted N=30 099	Weighted N=30 099
Demographics
Age, y	68.3 (11.3)	70.1 (10.6)	0.17	69.3 (10.8)	69.2 (10.8)	0.01
Male sex	0.94 (68 620)	0.96 (48 413)	0.07	0.95 (28 673)	0.95 (28 700)	<0.01
Race
White	0.74 (53 875)	0.81 (40 830)	0.16	0.81 (24 358)	0.81 (24 431)	0.01
Black	0.16 (11 652)	0.09 (4529)	0.21	0.1 (2887)	0.1 (2868)	<0.01
Something else*	0.03 (2188)	0.02 (1213)	0.04	0.03 (846)	0.02 (686)	0.03
Missing	0.07 (4972)	0.08 (3938)	0.04	0.07 (2008)	0.07 (2114)	0.01
Ethnicity
Hispanic	0.05 (3745)	0.04 (1807)	0.08	0.03 (923)	0.03 (798)	0.02
Non‐Hispanic	0.9 (65 392)	0.9 (45 672)	0.02	0.92 (27 610)	0.92 (27 651)	<0.01
Missing	0.05 (3550)	0.06 (3031)	0.05	0.05 (1566)	0.05 (1650)	0.01
Medical history
Congestive heart failure	0.11 (7969)	0.13 (6614)	0.07	0.11 (3458)	0.11 (3430)	<0.01
Chronic kidney disease	0.13 (9371)	0.12 (6176)	0.02	0.12 (3588)	0.12 (3620)	<0.01
Chronic obstructive pulmonary disease	0.21 (15 166)	0.22 (11 333)	0.04	0.22 (6643)	0.22 (6742)	0.01
Cardiovascular disease	0.11 (8267)	0.1 (5284)	0.03	0.11 (3282)	0.11 (3269)	<0.01
Depression	0.27 (19 647)	0.23 (11 738)	0.09	0.25 (7396)	0.24 (7364)	<0.01
Dialysis	0.01 (1066)	0.01 (538)	0.04	0.01 (351)	0.01 (346)	<0.01
Diabetes	0.43 (31 169)	0.44 (21 991)	0.01	0.43 (12 967)	0.43 (12 889)	0.01
Hyperlipidemia	0.7 (50 756)	0.7 (35 529)	0.01	0.71 (21 347)	0.71 (21 518)	0.01
Hypertension	0.76 (55 120)	0.77 (38 734)	0.02	0.76 (22 842)	0.76 (22 890)	<0.01
Prior coronary artery bypass graft	0.2 (14 675)	0.2 (10 313)	0.01	0.21 (6280)	0.21 (6270)	<0.01
Prior myocardial infarction	0.31 (22 369)	0.3 (15 122)	0.02	0.3 (9074)	0.3 (9017)	<0.01
Prior percutaneous coronary intervention	0.2 (14 628)	0.18 (9097)	0.05	0.19 (5806)	0.19 (5812)	<0.01
Prior stroke	0.08 (5568)	0.07 (3340)	0.04	0.07 (2068)	0.07 (2094)	<0.01
Peripheral artery disease	0.12 (8468)	0.12 (5884)	<0.001	0.12 (3566)	0.12 (3496)	0.01
Sleep apnea	0.3 (22 108)	0.27 (13 791)	0.07	0.28 (8565)	0.28 (8471)	0.01
Tobacco use	0.29 (21 189)	0.26 (13 214)	0.07	0.28 (8352)	0.28 (8511)	0.01
Patient location
Distance to noninvasive diagnostics	39.8 (37.8)	79.9 (40.2)	>0.99	69 (40.3)	70.6 (38.1)	0.04
Distance to invasive diagnostics	56.9 (55.6)	109.9 (70)	0.84	88.6 (54.4)	93.8 (59)	0.09
Distance to invasive therapeutics	61.5 (64)	113.9 (72.5)	0.77	93.9 (69.9)	98.9 (62.9)	0.07
Urban/rural
Urban	0.63 (45 599)	0.31 (15 888)	0.66	0.4 (11 890)	0.38 (11 451)	0.03
Rural	0.37 (27 087)	0.69 (34 622)	0.66	0.6 (18 208)	0.62 (18 648)	0.03
Medication usage
Before visit
Statin	0.66 (48 215)	0.66 (33 100)	0.02	0.66 (19 937)	0.66 (19 911)	<0.01
Beta blocker	0.49 (35 914)	0.51 (25 512)	0.02	0.5 (15 104)	0.5 (15 160)	<0.01
Angiotensin‐converting enzyme inhibitor	0.28 (20 047)	0.27 (13 589)	0.02	0.28 (8372)	0.28 (8382)	<0.01
P2Y12	0.22 (15 731)	0.24 (12 305)	0.06	0.22 (6757)	0.23 (6888)	0.01

VA indicates Veterans Affairs.

^*Something else: Includes any self‐described race that is not White or Black.

All analyses were performed with R 4.2.1. ¹⁴ A 2‐sided P value <0.05 was considered statistically significant.

RESULTS

Population

There were 235 197 new consultations for comprehensive cardiovascular evaluation and 201 453 that were completed in the chosen venue within 6 months. After removing patients who met the exclusion criteria, including those who were not active users of the internal VA Healthcare System (8847), were connected to sites that did not have an equal opportunity for internal VA or non‐VA community care (27546), or were eligible for community care for reasons other than distance (20880), we identified an analytic cohort of 123 197 patients with 50 510 (41%) being seen by community care providers and 72 687 (59%) being evaluated in the VA Healthcare System (Figure S1). After performing propensity score matching, our population was reduced to 30 099 patients in both the VA and community care setting, of whom 71% were matched within the same VA facility.

Clinical Characteristics

The demographics and clinical comorbidities of the patients treated in either venue were similar (Table 1). As expected, unmatched patients treated at non‐VA facilities in the community were more likely to reside in rural locations (69%) compared with those treated in the VA Healthcare System (37%) with a much greater straight‐line distance to receive noninvasive testing (79.9 miles versus 39.8 miles) or invasive testing (113.9 miles versus 61.5 miles). After matching the patient population, the 2 groups were similar without any absolute standardized differences >0.10.

Diagnostic Testing

A higher proportion of patients referred for cardiovascular evaluation received stress testing in the community at 1 year (difference, 4.8% [95% CI, 4.0%–5.5%], P=1.8×10⁻³²) and 2 years (difference, 6.8% [95% CI, 5.9%–7.8%], P=1.5×10⁻⁴⁶) after initial evaluation (Table 2). Similarly, patients evaluated in the community had a significantly higher rate of invasive diagnostic testing with coronary angiography (1‐year difference, 4.3% [95% CI, 3.7%–4.9%], P=3.5×10⁻⁴²; 2‐year difference, 5.7% [95% CI, 5.0%–6.5%], P=2.1×10⁻⁵¹) and therapeutic intervention with percutaneous coronary intervention (1‐year difference, 1.9% [95% CI, 1.6%–2.3%], P=2.2×10⁻²⁷; 2‐year difference, 2.5% [95% CI, 2.1%–2.9%], P=8.2×10⁻³⁰) after initial evaluation. Among those who received a first diagnostic test, a significantly larger proportion of patients were likely to have a repeat echocardiogram (1‐year difference, 3.3% [95% CI, 2.3%–4.3%], P=1.8x10⁻¹¹; 2‐year difference, 7.5% [95% CI, 6.1%–8.8%], P=1.6×10⁻²⁷) or stress test (2‐year difference, 4.4% [95% CI, 3.0%–5.7%], P=6.9×10⁻¹⁰) after initial evaluation in the community.

Table 2.

Proportion of Patients Undergoing Testing or Therapeutic Intervention With Differences for Those Treated in the Community Relative to Those in the VA

	1‐year results (est 95% CI; P value)			2‐year results (est 95% CI; P value)
	VA	Community	Difference	VA	Community	Difference
Proportion with 1 test
Computed tomography	3.5	3.4	−0.1; (−0.4 to 0.2); 0.59	4.3	4.5	0.2; (−0.2 to 0.6); 0.28
Transthoracic echocardiogram	53.3	53.1	−0.2; (−1 to 0.7); 0.73	63.7	66.1	2.5; (1.6 to 3.4); 1.2×10−7
Stress test	28.7	33.5	4.8; (4 to 5.5); <0.001	36.4	43.2	6.8; (5.9 to 7.8); <0.001
Stress test	28.7	33.5	4.8; (4.0 to 5.5); 1.8×10−32	36.4	43.2	6.8; (5.9 to 7.8); 1.5×10−46
Coronary angiography	13.6	17.9	4.3; (3.7 to 4.9); 3.5×10−42	17.4	23.1	5.7; (5.0 to 6.5); 2.1×10−51
Percutaneous intervention	3.5	5.4	1.9; (1.6 to 2.3); 2.2×10−27	4.5	7.0	2.5; (2.1 to 2.9); 8.2×10−30
Proportion with second test*
Computed tomography	4.9	4.7	−0.2; (−2.8 to 2.4); 0.88	6.8	10.2	3.4; (−0.4 to 7.1); 0.08
Transthoracic echocardiogram	16.4	19.7	3.3; (2.3 to 4.3); 1.8×10−11	35.4	42.9	7.5; (6.1 to 8.8); 1.6×10−27
Stress test	7.4	8.0	0.6; (−0.3 to 1.5); 0.17	15.8	20.1	4.4; (3.0 to 5.7); 6.9×10−10

VA indicates Veterans Affairs.

^*For analyses looking at the time to second tests, the study population was subset to patients with at least 1 test of interest.

Time to Testing

The median time between initial cardiovascular consultation and the first visit was similar in the VA (33 days [95% CI, 19–53]) and community (35 days [95% CI, 17–65], Figure 1). The time interval between the original referral for cardiovascular care and initial testing for those who received testing is depicted in Table 3, with transthoracic echocardiography (113 versus 102 days, difference, 11.4 days [95% CI, 7–16 days], P=0) delayed for those seen in the community. In contrast, the time to coronary angiography (111 versus 110 days, difference, 0.4 days [95% CI, −7 to 9 days], P=0.99) and percutaneous coronary intervention (108 versus 117 days, difference, −7.5 days [95% CI, −23 to 9 days], P=0.41) was similar across venues.

Figure 1. Time between consult and visit.

VA indicates Veterans Affairs.

Table 3.

Time to First Diagnostic Test or Therapeutic Intervention for Those Treated in the Community Relative to Those in the VA

	Time to first test in days median (interquartile range)		Difference in medians in days
	VA	Community	Difference (95% CI), P value
Coronary computed tomography	112 (61 to 288)	127 (63 to 357)	14.1; (−7 to 35); 0.21
Transthoracic echocardiogram	102 (50 to 242)	113 (55 to 290)	11.4; (7 to 16); 0
Stress test	97 (52 to 273)	100 (49 to 293)	2.7; (−2 to 8); 0.37
Coronary angiography	110 (50 to 295)	111 (50 to 300)	0.4; (−7 to 9); 0.99
Percutaneous coronary intervention	117 (55 to 288)	108 (47 to 303)	−7.5; (−23 to 9); 0.41

VA indicates Veterans Affairs.

Cost of Testing

Over three‐quarters of claims submitted for reimbursement in the community included professional bills where cost estimates were captured. The median professional fees billed for community care providers were inversely proportional to the time elapsed to complete this care in the community (Pearson correlation: −0.924). Procedures that had less generous professional fees billed occurred more rapidly in the VA Healthcare System and procedures with more generous reimbursement occurring at a similar pace in the fee‐for‐service non‐VA community (Table 4).

Table 4.

Median Billed Professional Fees by Diagnostic Test or Therapeutic Intervention

	Community procedures with cost estimates (%)	Professional costs, median (interquartile range)
CT angiography	71.6	$387 ($255, $684)
Transthoracic echocardiography	82.6	$390 ($200, $666)
Stress test	80.8	$620 ($294, $1400)
Coronary angiography	76.7	$983 ($756, $1526)
Percutaneous coronary intervention	77.2	$1897 ($1462, $2988)

CT indicates computed tomography.

Medical Therapies

Figure 2 summarizes the proportion of patients who received prescription fills for guideline recommended therapies in the propensity matched cohort, including statins, beta blockers, angiotensin receptor inhibitors, and antiplatelet agents (P2Y12). For all medications, the proportion of patients with referrals for non‐VA community care who received medication fills after their comprehensive cardiovascular evaluation was significantly lower at 1 year (P<0.001) and lower at 2 years (P<0.001, P=0.01 for antiplatelet agents).

Figure 2. Medication prescription fills among patients with presumed coronary artery disease, based on treatment venue.

The proportion of patients receiving a prescription fill for a statin (A), beta blocker (B), angiotensin receptor inhibitor (C), or P2Y12 inhibitor (D) was significantly lower for patients evaluated in the community at 1 year (P<0.01) and 2 years (P<0.01), compared with those treated in the VA Healthcare System. Statin: atorvastatin, fluvastatin, lovastatin, pitavastatin, pravastatin, rosuvastatin, simvastatin. Beta blocker: acebutolol, atenolol, betaxolol, bisoprolol, carvedilol, esmolol, labetalol, metoprolol, nadolol, pindolol, propranolol, sotalol. ACEi: benazepril, captopril, enalapril, fosinopril, lisinopril, moexipril, perindopril, quinapril, ramipril, trandolapril. P2Y12: cangrelor, clopidogrel, prasugrel, ticagrelor, ticlopidine. ACEi indicates angiotensin‐converting enzyme inhibitor; CABG, coronary artery bypass graft; PCI, percutaneous coronary intervention; and VA, Veterans Affairs.

Clinical Outcomes

The proportion of matched patients suffering major adverse cardiovascular events at 1 (9.7% versus 8.7%, difference, 1.0% [95% CI, 0.5%–1.5%], P=7.3×10⁻⁵) and 2 (17.6% versus 15.3%, difference, 2.3% [95% CI, 1.6%–3.0%], P=5.9×10⁻¹⁰) years after initial consultation was significantly higher for those referred to the community (Figure 3A), compared with those treated in the VA Healthcare System after a median follow‐up of 543 days (interquartile range, 244–730) in the matched cohort. The proportion of matched patients who died was not significantly different at 1 year (P=0.13) but was higher for patients treated in the community at 2 years (9.0% versus 8.3%, difference, 0.7% [95% CI, 0.1%–1.3%], P=0.01) (Figure 3B). Acute coronary syndromes (1‐year: 5.3% versus 4.7%, difference 0.6%, [95% CI, 0.3%–1.0%], P=1.1×10⁻³; 2‐year: 9.1% versus 7.5%, difference, 1.5% [95% CI, 1.0%–2.1%], P=1.6×10^−0.8, Figure 3C) as well as stroke (1‐year: 1.8% versus 1.5%, difference, 0.3% [95% CI, 0.1%–0.5%], P=0.01; 2‐year: 3.2% versus 2.8%, difference, 0.4% [95% CI, 0.01%–0.7%], P=0.02, Figure 3D) were significantly greater at 1 and 2 years for those matched patients treated in the community.

Figure 3. Clinical outcomes among patients with presumed coronary artery disease.

Clinical outcomes among patients with presumed coronary artery disease based on treatment venue for MACE (A), death (B), ACS (C), and stroke hospitalization (D). ACS indicates acute coronary syndrome; CABG, coronary artery bypass graft; MACE, major adverse cardiovascular event; PCI, percutaneous coronary intervention; and VA, Veterans Affairs.

Unmeasured Confounder

Using a hypothetical confounder with a difference in prevalence similar to rurality (VA: 46%; community: 56.5%) and associated with a 2‐fold increase in adverse events (11.7 versus 20.3), the absolute incidence of major adverse events in the community would still be 1.4 (95% CI, 0.7–2.1) higher and remain statistically significant. This would suggest that an unmeasured confounder would need to be very prevalent (≈50%), have a higher prevalence among community patients (approximately 10%), and have a significant differential effect (approximately 173%) on adverse events between those treated at community care or VA facilities to only partially negate the observed discrepancy in clinical outcomes (Table S2).

DISCUSSION

The present analysis evaluated the diagnostic testing and clinical outcomes among matched patients undergoing a cardiovascular evaluation in 2 distinct venues with divergent reimbursement systems. Though the community care system was designed to accelerate evaluation of patients, the present study demonstrates that the median time between initial consultation and evaluation was similar across venues with comparable delays in diagnostic testing or therapeutic interventions regardless of where the patient was seen. Patients evaluated in the community fee‐for‐service environment, however, were more likely to undergo noninvasive and invasive testing, with a significantly higher proportion of patients also receiving repeat testing than those treated in the integrated VA Healthcare System. Further, process metrics indicative of quality, such as prescription of guideline‐recommended medical therapies, were higher in the integrated health care system compared with the community. With these differences in management in mind, clinical outcomes were also divergent with a significantly increased rate of adverse events among matched patients who were treated in the community. Collectively, these data suggest an opportunity to optimize the community care program for veterans.

The time between a consultation and completion of a cardiovascular evaluation was similar across treatment venues, leading to comparable delays in diagnostic testing or therapeutic interventions. Over the past decade, there have been concerns that the VA Healthcare System could not meet the clinical needs of its patients within a timely manner. ⁷ , ⁸ Because of this, the Veterans Access, Choice and Accountability Act of 2014 and VA MISSION Act of 2018 have been passed to improve access to care for these patients through an expansion of partnerships with community care practitioners. ⁸ These efforts have increased the use of health care in the community for those with cardiovascular disease, though there have been reports that community care providers have also struggled to meet the need for increased capacity. ¹⁵ The present study suggests that the median time to initial evaluation after consultation was similar in the VA Healthcare System and the community. Further, the time to first diagnostic evaluation or therapeutic procedure was either shorter in the integrated health care system (transthoracic echocardiography) or similar to those treated in the community. These findings are consistent with studies evaluating other patient populations that suggest referrals to community care practitioners do not result in a rapid evaluation for patients enrolled in the VA Healthcare System, perhaps due to increased administrative burden in executing insurance agreements that do not exist in the integrated system. ¹⁶ , ¹⁷ This indicates that prior legislation designed to improve rapid access to care may not be functioning as intended.

The proportion of patients who underwent initial and repeat testing for cardiovascular disease was significantly higher among those treated in fee‐for‐service community practices outside VA. Previous research has suggested that physicians may alter their practice patterns based upon reimbursement policies across a range of specialties ¹⁸ with higher rates of stress testing after coronary revascularization in fee‐for‐service environments by physicians who billed for technical or professional fees. ¹⁹ The present study adds to these data, suggesting that a homogenous cohort of matched patients undergoing a cardiovascular evaluation are significantly more likely to undergo noninvasive or invasive diagnostic testing in a fee‐for‐service community practice environment than in the integrated VA Healthcare System. Further, patients undergoing community evaluation are more likely to have repeat computed tomography, transthoracic echocardiography or stress testing in the 2 years following the initial consultation. The increased testing rate was most prominent among highly reimbursed invasive procedures, including coronary angiography and percutaneous coronary revascularization. This is particularly surprising given recent data highlighting the importance of trialing medical therapy for patients with stable coronary artery disease before coronary revascularization. ²⁰ , ²¹ The financial incentives that may encourage this practice do not, however, ensure improved outcomes.

The clinical outcomes among patients undergoing comprehensive cardiovascular evaluation in the community were not superior to those treated in the VA Healthcare System. Previous research has demonstrated that patients receiving care in the community had similar or inferior clinical outcomes to those treated in the VA Healthcare System across a range of medical specialties. ²² This was particularly true for patients undergoing coronary revascularization for stable coronary artery disease, with a significant increase in mortality among patients treated via non‐VA care in the community. ⁶ This discrepancy was in part related to difficulties coordinating postprocedure follow‐up visits and guideline‐recommended medications across fractured community health care systems, highlighting the benefits of an integrated health care system such as the VA. ¹¹ The present study reinforces these findings among a broader matched population of stable patients undergoing cardiovascular evaluation, with an increased rate of major adverse events for those treated in the community. Differences in adverse event rates could be a manifestation of better adherence to guideline‐recommended therapy or improved coordination in the integrated health system. Alternatively, this may be related to significant heterogeneity in the community care settings that evaluated patients, including differences in accessibility and resources. Regardless of the cause, though, the data suggest that increased access to care in fee‐for‐service community care did not accelerate an evaluation for these patients, nor did increases in diagnostic testing obtained in the community improve outcomes. Further discussion is warranted to determine the best mechanism to optimize access to care while maintaining quality and improving outcomes among patients enrolled in the VA Healthcare System.

Limitations

The present analysis should be interpreted in the context of several limitations. The data are retrieved from administrative sources and thus residual confounding is possible. Specifically, there may be patient characteristics that lead to preferential treatment in the community compared with the VA Healthcare System. The present study attempted to mitigate this by limiting the analysis to only those patients who were seen in the community due to extensive drive times to a VA facility. Patients who were preferentially seen in the community for other reasons, including for increased medical acuity or due to the best medical interest of the veteran, were excluded from this analysis. Additionally, comparisons were then made among patients who were matched for demographics, comorbidities, and distance to a VA facility. Given divergent financial incentives, it is possible that administrative coding is different across treatment venues for clinical outcomes. An unmeasured confounding analysis was undertaken to address this, suggesting that the difference in coding or other confounding variables would have to be large to completely negate the differential outcomes observed. Further, the observed difference in mortality was substantial between venues and not subject to biases in coding. Prescription data were not consistently available for community‐based pharmacies, and thus differences in prescription rates were derived from fills at VA outpatient pharmacies. The impact of this is unclear, as the entire cohort analyzed had previously used the VA outpatient pharmacy within the previous 2‐years and there are significant financial incentives for the patient to continue doing so in the form of reduced copayments. Regardless, it is possible that this could undermeasure community‐based pharmacy prescriptions. The population analyzed was predominantly male and thus the results may not be generalizable to other groups with greater diversity. Finally, the analysis is focused on the unique dual health care environment facilitated by the VA Healthcare System with both fee‐for‐service non‐VA care and integrated VA care.

CONCLUSIONS

In conclusion, predominantly White male veterans undergoing cardiovascular evaluation in fee‐for‐service community practices did not receive a more rapid evaluation but did undergo increased rates of initial and repeat diagnostic testing compared with those treated an integrated health care system. Despite this, patients treated in the community had lower rates of adherence to guideline‐recommended medications and higher rates of major adverse events suggesting an opportunity to optimize the community care program to enhance access while maintaining quality for patients.

Sources of Funding

None.

Disclosures

The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the US Government. Dr Waldo has received grants from the National Institutes of Health and VA Health Services Research. The remaining authors have nothing to disclose.

Supporting information

Data S1

Tables S1–S2

Figure S1