Ambulatory Care Fragmentation and Hospitalization Among Veterans With Diabetes

ABSTRACT

Objectives: To determine whether having a usual provider of care (UPC) outside the Veterans Health Administration (VHA) and whether having highly fragmented care (regardless of the providers’ health system affiliations) increased the risk of hospitalization among veterans with diabetes.

Study Design: Retrospective dynamic cohort analysis of all veterans with diabetes 65 years and older enrolled nationally in both VHA and Medicare from 2005 to 2010, using VHA-Medicare linked data. We used 5 two-year study periods, assessing ambulatory care in the first year of each 2-year period and any hospitalization in the second year.

Methods: We used longitudinal generalized estimating equation models to test the associations of the affiliation (VHA vs non-VHA) of the UPC and the extent of fragmentation with hospitalization, adjusting for potential confounders. Highly fragmented care was defined as a reversed Bice-Boxerman Index of at least 0.85, which was equivalent to the 75th percentile.

Results: Having a UPC outside the VHA was associated with 11% increased odds of hospitalization (95% CI, 10%-12%). Having highly fragmented care was associated with 7% increased odds of hospitalization (95% CI, 6%-8%). Having both a UPC outside the VHA and highly fragmented care was associated with 19% increased odds of hospitalization (95% CI, 18%-20%).

Conclusions: Among veterans with diabetes enrolled in both VHA and Medicare, having both a UPC outside the VHA and highly fragmented care was associated with higher odds of hospitalization than either of these ambulatory patterns alone.

Am J Manag Care. 2021;27(4):155-160. https://doi.org/10.37765/ajmc.2021.88509

Takeaway Points

In this national sample of veterans with diabetes:

• Having a usual provider of ambulatory care outside the Veterans Health Administration (VHA) was associated with 11% increased odds of hospitalization (95% CI, 10%-12%).
• Having highly fragmented care (spread across many ambulatory providers, regardless of health system affiliation) was associated with 7% increased odds of hospitalization (95% CI, 6%-8%).
• Having both a usual provider of ambulatory care outside the VHA and highly fragmented care was associated with higher (19% increased) odds of hospitalization than for either of these ambulatory patterns alone (95% CI, 18%-20%).

Fragmented care in the ambulatory setting occurs when an individual’s care is distributed across multiple providers, with no single provider accounting for a substantial proportion of visits.¹ Having more fragmented ambulatory care has been found to be associated with costly and potentially avoidable subsequent care, including higher rates of emergency department visits,^1-4 higher rates of hospitalizations,^1,2,5 and more unnecessary tests and procedures,^6,7 compared with having less fragmented ambulatory care.

The Veterans Health Administration (VHA) offers an integrated system of care for eligible veterans, but even among VHA-enrolled veterans, most choose to receive health care outside the VHA. Among VHA patients 65 years and older, about 85% of total health care expenditures are paid for by private insurance, Medicare, or self-pay, with Medicare covering more than half the costs.⁸ Veterans who use both VHA and non-VHA providers are at increased risk of poorer outcomes such as worse glycemic control⁹ and hospitalizations and readmissions among veterans with congestive heart failure¹⁰ and stroke.¹¹ However, such studies are often limited by the use of a dichotomous (VHA only vs both VHA and non-VHA) categorization. This oversimplification does not account for the extent of an individual patient’s fragmentation of care across providers, which may have its own significant effect on a patient’s outcomes.¹

The objective of this study is to determine whether ambulatory care patterns are associated with the risk of subsequent hospitalization, considering 2 different measures of ambulatory care. We will consider the health system affiliation of the usual provider of care (UPC) (VHA or non-VHA), and we will also consider the extent to which care is spread (or fragmented) across all of a patient’s providers, regardless of health system affiliation. Using both VHA medical records and Medicare claims from 2004 to 2010, we will assess UPC and fragmentation separately and also consider interactions between them.

METHODS

Overview

We conducted a retrospective dynamic cohort analysis of all veterans with diabetes 65 years and older enrolled nationally in both VHA and Medicare fee-for-service from 2005 to 2010, using VHA-Medicare linked data. We used 5 two-year study periods, measuring ambulatory care in the first year of each 2-year period (2005, 2006, 2007, 2008, and 2009) and any hospitalization in the second year (2006, 2007, 2008, 2009, and 2010). The study protocol was approved by the Institutional Review Board at the East Orange Veterans Affairs Medical Center.

Population

A veteran was considered to have diabetes if they had at least 1 inpatient or 2 outpatient diabetes International Classification of Diseases, Ninth Revision diagnosis codes (250.xx) in the prior 24-month look-back period in either VHA or Medicare.¹² To be included, veterans had to be enrolled in VHA, Medicare Part A, and Medicare Part B for both years in a given 2-year period. A veteran could contribute up to 5 two-year periods of data. Additional inclusion criteria included being alive on the first day of the outcome year and having at least 4 ambulatory care visits in the baseline year to ensure stability of the measure of fragmentation.5 Non–community-dwelling veterans, as determined by having hospital stays longer than 180 days, any hospice stays, or any skilled nursing stays, were excluded from the cohort.

Ambulatory Visits, UPC, and Fragmentation of Care

Ambulatory visits in VHA and Medicare were defined using a specific set of ambulatory evaluation and management (E&M) codes adapted from the Healthcare Effectiveness Data and Information Set (HEDIS).¹³ Ambulatory visits were restricted to E&M visits for adults that took place in an office setting or hospital-based clinic, excluding home visits and nursing home visits.^1,3,6 Because the HEDIS definition of ambulatory visits does not include emergency department visits, we have not included it in our definition. We counted only 1 ambulatory visit per provider-beneficiary-date combination and excluded ambulatory visits for which the rendering provider was missing. Non-VHA providers are identified in Medicare claims using individual National Provider Index codes. In the VHA, providers are identified using
clinic stop codes.

The provider with the highest proportion of ambulatory care visits was classified as the UPC.¹⁴ UPC was further classified into 2 subcategories, based on whether the provider was inside or outside the VHA. If a veteran had an equal proportion of visits with 2 providers (1 inside the VHA and 1 outside), the UPC was randomly assigned to 1 of them. Approximately 5% (between 4.8% and 5.6% over the 5 years) of patients had an equal number of VHA and non-VHA usual providers and needed to be randomly assigned. The UPC could be either a generalist or a specialist physician.

Fragmentation was defined using the Bice-Boxerman Index (BBI),¹⁵ a validated^1,2,5 measure of dispersion of ambulatory care that uses number of visits, number of unique providers, and the concentration of visits between providers in the following formula:

In this formula, n_i is the number of visits per individual i, and p is the number of unique providers. Although this index was originally labeled as a measure of “continuity,” it actually measures more than consistency of care with a single provider; it captures both “dispersion” (the spread of a patient’s care across multiple providers) and “density” (the relative share of visits by each provider).¹⁶ It is highly correlated with other measures of continuity or fragmentation, such as the Herfindahl-Hirschman Index, the Usual Provider of Care Index, and the Sequential Continuity Index.¹⁴ For ease of interpretation, we inverted the continuous scale so that higher values (ranging from 0 to 1) corresponded to more fragmentation (reversed BBI, or rBBI).

Other Variables

Other measures during each baseline year included age, gender, comorbid condition burden measured using the Charlson-Deyo Comorbidity Index (CCI),^17,18 marital status (married, unmarried, and all other), and VHA priority status. The VHA assigns veterans to priority groups based on the extent of service-connected disability and income.¹⁹ Diabetes severity was captured using the Diabetes Complications Severity Index (DCSI) score, which has been adapted for use with administrative data diagnosis codes.²⁰

Hospitalizations

A single binary dependent variable was used to represent the first acute hospitalization (yes/no) over 5 years (2006-2010), using data from VHA and Medicare inpatient files.

Statistical Analysis

The 2 independent variables (UPC and fragmentation) were measured for each veteran for each baseline year (2005, 2006, 2007, 2008, and 2009). In each baseline year, we dichotomized fragmentation into high (≥ 0.85) vs low (< 0.85), using the 75th percentile of rBBI from the first baseline year as the cut point. We also combined UPC and fragmentation to divide veterans into 4 groups for each baseline year: group 1 (low fragmentation with a VHA UPC), group 2 (high fragmentation with a VHA UPC), group 3 (low fragmentation with a non-VHA UPC), and group 4 (high fragmentation with a non-VHA UPC).

We used descriptive statistics to characterize veterans in each 2-year study period. We also summarized ambulatory utilization. We compared the characteristics of those with high vs low fragmentation using t tests for continuous variables and χ2 tests for dichotomous variables.

We used generalized estimating equations (GEE) models to assess the association between the 2 variables of interest (high vs low fragmentation and whether the UPC was a VHA or a non-VHA provider) in the baseline year and hospitalization in the subsequent year. We generated 3 separate models. Model 1 used fragmentation (high vs low) as the exposure, without including UPC in the model. Model 2 used UPC (VHA vs non-VHA) as the exposure, without including fragmentation in the model. Model 3 used the 4-group variable as the exposure. All models adjusted for age, gender, marital status, VHA priority status, DCSI score, and CCI score and accounted for clustering because veterans could appear across multiple years. We chose an unstructured correlation structure for the models because we have 5 time points and guidelines suggest that with 5 or fewer time points, this option produces the most stable estimates.21

We conducted a sensitivity analysis using Cox proportional hazards models to assess time to first acute hospitalization as the outcome. Patients were censored if they died, or at the end of the outcome year. Proportional hazard assumptions were checked using Schoenfeld residuals. To confirm the robustness of our findings, we also conducted sensitivity analyses stratified by diabetes severity (DCSI score) and, separately, by comorbidity burden (CCI score).

RESULTS

Cohort Characteristics

The cohort of VHA enrollees with diabetes increased over the 5-year period from 492,479 veterans in 2005 to 532,123 in 2009. In 2005, 98% were male, as expected; 87% were White; 8%, Black; and 4%, of Hispanic ethnicity; the mean (SD) age was 75 (5.5) years. The mean (SD) CCI score was 3.4 (2.1), 86% had at least 1 diabetes complication, and 25% were moderately or severely disabled and 30% had very low income according to VHA priority group (Table 1).

As expected, the cohort was aging over time, with 10.0% being 85 years and older in 2009 compared with 4.6% in 2005. Diabetes severity increased slightly, with 32% having 3 or more complications in 2009 vs 30% in 2005. CCI scores decreased over time (31% with a score of 1 in 2009 compared with 22% in 2005), possibly because of the entry into the cohort of veterans with recently diagnosed diabetes and fewer comorbidities (Table 1).

Ambulatory Care Utilization and Provider Characteristics

In unadjusted analysis, 167,286 (34%) of the veterans in 2005 had high fragmentation (rBBI ≥ 0.85). The majority (70%) of veterans had a non-VHA UPC. In 2005, veterans had an average of 15 visits (4 [27%] of which were in VHA) among 6 providers (2 [33%] in VHA), with the UPC accounting for an average of 44% of all visits. Patterns differed significantly between veterans with high and low fragmentation. Veterans with high fragmentation had an average of 17 visits (5 [29%] in VHA) among 8 providers (3 [38%] in VHA), with the UPC accounting for only 27% of total visits, compared with those with low fragmentation, who had an average of 14 visits (4 [29%] in VHA) among 5 providers (2 [40%] in VHA), with the UPC accounting for more than 50% of total visits. eAppendix Table 1, eAppendix Figure 1, and eAppendix Figure 2 show ambulatory patterns for all years from 2005 to 2009 (eAppendix available at ajmc.com).

Of the total veterans in 2005, 19% were in group 1 (low fragmentation with a VHA UPC), 10% were in group 2 (high fragmentation with a VHA UPC), 47% were in group 3 (low fragmentation with a non-VHA UPC), and 24% were in group 4 (high fragmentation with a non-VHA UPC).

Rates of Hospitalization

Table 2 shows that between 27% and 29% of veterans had at least 1 hospitalization in the outcome year between 2005 and 2009. There was a significantly higher proportion of veterans hospitalized in the high vs low fragmentation groups in all years. Separately, having a non-VHA UPC was associated with significantly higher rates of hospitalization compared with having a VHA UPC (Table 2).

Association Between Fragmentation and Hospitalization

Using the GEE models, model 1 revealed that veterans with high fragmentation had 7% higher odds (adjusted odds ratio [AOR], 1.07; 95% CI, 1.06-1.08) of being hospitalized compared with those with low fragmentation, after adjustment for age, gender, marital status, VHA priority status, DCSI score, and CCI score (Table 3). This model did not include UPC as a variable.

Association Between Location of UPC and Hospitalization

Model 2 revealed that veterans with a non-VHA UPC had 11% higher odds of hospitalization (AOR, 1.11; 95% CI, 1.10-1.12) compared with those with a VHA UPC, after adjustment for age, gender, marital status, VHA priority status, DCSI score, and CCI score. This model did not include fragmentation as a variable.

Impact of Interaction of Fragmentation and UPC on Hospitalization

The interaction between the extent of fragmentation and the location of the UPC was significant (P < .0001).

Model 3, which included the 4-level exposure variable of high and low fragmentation and VHA vs non-VHA UPC, demonstrated a statistically significant incremental increase in odds of hospitalization compared with the reference group (veterans with a VHA UPC and low fragmentation). Compared with the reference group (group 1), the odds of hospitalization increased to an AOR of 1.11 (95% CI, 1.10-1.13)for those with high fragmentation index and VHA UPC (group 2), AOR of 1.13 (95% CI, 1.12-1.14) for those with low fragmentation and a non-VHA UPC (group 3), and AOR of 1.19 (95% CI, 1.18-1.20) for those with both high fragmentation index and non-VHA UPC (group 4) (Table 3).

Sensitivity Analyses

Applying Cox proportional hazard models (eAppendix Table 2) corroborated the findings described above, with high fragmentation associated with an increased adjusted hazard of hospitalization (HR, 1.06; 95% CI, 1.05-1.07). Having a non-VHA provider as the UPC resulted in an increased hazard of hospitalization (HR, 1.10; 95% CI, 1.09-1.10). Using the combined 4-category variable, the results were similar. However, because the sensitivity analyses using Cox models did not satisfy the proportional hazards assumption, these results can only be interpreted as an “average hazard ratio” over a 1-year follow up period.

When we stratified results by diabetes severity, the main findings of the study were observed within each level of diabetes severity (eAppendix Table 3). Separately, when we stratified results by CCI scores, the main findings were observed within each level of CCI scores (eAppendix Table 4).

DISCUSSION

In this cohort of veterans 65 years and older with diabetes who use VHA and Medicare, high fragmentation of ambulatory care and having a non-VHA UPC in the baseline year were each associated with higher odds of hospitalization. Veterans with both high fragmentation and a non-VHA UPC had the highest odds of hospitalization compared with all other combinations of fragmentation level and UPC affiliation. These results suggest that the extent of ambulatory care fragmentation and the setting of the UPC each make independent contributions to the odds of hospitalization.

Some literature has focused on the differences between users who use both VHA and non-VHA providers (mainly Medicare) compared with those who use the VHA only; however, the results are mixed for the different types of outcomes studied.^9,22,23 Another study looked at VA/private dual use among veterans with diabetes, stratified by whether care by private providers was for specialty care or for primary care; that study showed a beneficial effect on preventable hospitalizations when care by private providers was for specialty services and no effect when they were for primary care services.²⁴ To our knowledge, our study is the first to tease apart the affiliation of the UPC from the extent of fragmentation overall.

The 2 groups with the lowest odds of hospitalization both had a UPC at the VHA. This finding could be because the VHA is an integrated delivery system²⁵ with a single electronic health record that enables information continuity across providers and sites.^26,27 However, the finding that, among those with a VHA UPC, having more fragmented care was associated with a higher risk of hospitalization compared with having less fragmented care is novel.

Strengths and Limitations

The study has numerous strengths. It uses the entire population of older Medicare-enrolled VHA users with diabetes over a 6-year period, representing one of the largest samples of patients with diabetes. Our use of the VHA and Medicare data ensures the most complete administrative data set possible and captures all hospitalizations. By utilizing 5 cohorts over the 6-year period, we were able to show consistency of our results over time. We also used previously validated measures of fragmentation of care and applied robust statistical analysis.

This study has several limitations. First, because it is observational, we cannot infer causality and cannot rule out the presence of unmeasured confounders. It uses administrative data records, which could result in incomplete recording or miscoding of comorbidities. Second, we cannot determine the appropriateness of the hospital admissions. Third, although we excluded individuals with prolonged hospitalizations (> 180 days), we did not account for prior hospitalization, in part because of the dynamic cohort design. Fourth, for VHA services, fragmentation was potentially partially obscured by the use of clinic stop codes to capture providers, which may have the effect of counting multiple providers in the same clinic as 1 “provider” for this analysis. This limitation would cause us to underestimate the extent of fragmentation. Finally, this study uses data through 2010, which preceded the passage of the Choice Act of 2014^28,29 and the MISSION Act of 2018,^30,31 which are legislative initiatives designed to make it easier for veterans to seek care outside the VHA at the VHA’s expense. From fiscal year (FY) 2014 to FY 2018, the number of individual veterans using community care increased from 1.3 million to 1.8 million, with the latter number equivalent to 30% of all veterans accessing VHA health care services that year.³² From FY 2014 to FY 2018, VHA spending on community care increased from $8.2 billion to $14.9 billion (an 82% increase), and that figure is expected to reach $17.8 billion in FY 2021.32 Although we cannot be sure that this represents an overall increase in the number of providers engaged in the care of these veterans, the magnitude of the increases suggests this and, from the perspective of the VHA, highlights the issue of fragmentation of care. Thus, it will be important for the VHA to continue to monitor the impact of fragmentation of care on hospitalization and cost in the era of the Choice and MISSION Acts, and our results can serve as a baseline for those future analyses.

CONCLUSIONS

Among veterans with diabetes, having highly fragmented health care in one year was associated with 11% increased odds of hospitalization in the following year. Separately, having a non-VHA UPC one year was associated with 7% increased odds of hospitalization in the following year. Having both highly fragmented care and a non-VHA UPC one year was associated with 19% increased odds of hospitalization in the following year. The magnitude of this effect and the presence of this interaction indicate that where veterans seek care and how diffusely their care is delivered both matter. Future studies should apply the methods used in this study to understand whether this problem is exacerbated in the context of legislative initiatives that are likely to increase the fragmentation of care for veterans.

Author Affiliations: Department of Veterans Affairs–New Jersey Health Care System (MRa, DH, MRo, DF), East Orange, NJ; Department of Medicine, Weill Cornell Medicine (MRa, LMK), New York, NY; Center for Innovations in Quality, Effectiveness, and Safety (IQuESt), Michael E. DeBakey VA Medical Center (DH), Houston, TX; Rutgers University School of Public Health (DF), Piscataway, NJ.

Source of Funding: This material is based upon work supported by the Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Health Services Research and Development Service (VA HSRD IIR 12-401) with resources and the use of facilities at the VA–New Jersey Health Care System, East Orange, NJ. 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 United States government.

Author Disclosures: Dr Fried has applied for grant funding related to care fragmentation. Dr Kern is a consultant to Mathematica and Partners Healthcare and has received a grant from the National Heart, Lung, and Blood Institute. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (MRa, DH, MRo, DF, LMK); acquisition of data (MRa, DH, MRo, DF); analysis and interpretation of data (MRa, DH, MRo, DF, LMK); drafting of the manuscript (MRa, DH, DF); critical revision of the manuscript for important intellectual content (DH, MRo, DF, LMK); statistical analysis (MRa, DH, MRo, DF); obtaining funding (DH); administrative, technical, or logistic support (DH, MRo); and supervision (DH).

Address Correspondence to: Mangala Rajan, MBA, Weill Cornell Medicine, 420 E 70th St, Room LH-348, New York, NY 10021. Email: mar2834@med.cornell.edu.

REFERENCES

1. Kern LM, Seirup JK, Rajan M, Jawahar R, Stuard SS. Fragmented ambulatory care and subsequent healthcare utilization among Medicare beneficiaries. Am J Manag Care. 2018;24(9):e278-e284.

2. Hussey PS, Schneider EC, Rudin RS, Fox DS, Lai J, Pollack CE. Continuity and the costs of care for chronic disease. JAMA Intern Med. 2014;174(5):742-748. doi:10.1001/jamainternmed.2014.245

3. Kern LM, Seirup JK, Rajan M, Jawahar R, Stuard SS. Fragmented ambulatory care and subsequent emergency department visits and hospital admissions among Medicaid beneficiaries. Am J Manag Care. 2019;25(3):107-112.

4. Liu CW, Einstadter D, Cebul RD. Care fragmentation and emergency department use among complex patients with diabetes. Am J Manag Care. 2010;16(6):413-420.

5. Nyweide DJ, Anthony DL, Bynum JPW, et al. Continuity of care and the risk of preventable hospitalization in older adults. JAMA Intern Med. 2013;173(20):1879-1885. doi:10.1001/jamainternmed.2013.10059

6. Kern LM, Seirup JK, Casalino LP, Safford MM. Healthcare fragmentation and the frequency of
radiology and other diagnostic tests: a cross-sectional study. J Gen Intern Med. 2017;32(2):175-181. doi:10.1007/s11606-016-3883-z

7. Romano MJ, Segal JB, Pollack CE. The association between continuity of care and the overuse of medical procedures. JAMA Intern Med. 2015;175(7):1148-1154. doi:10.1001/jamainternmed.2015.1340

8. West AN, Weeks WB. Health care expenditures for urban and rural veterans in Veterans Health Administration care. Health Serv Res. 2009;44(5, pt 1):1718-1734. doi:10.1111/j.1475-6773.2009.00988.x

9. Helmer D, Sambamoorthi U, Shen Y, et al. Opting out of an integrated healthcare system: dual-system use is associated with poorer glycemic control in veterans with diabetes. Prim Care Diabetes. 2008;2(2):73-80. doi:10.1016/j.pcd.2008.02.004

10. Axon RN, Gebregziabher M, Everett CJ, Heidenreich P, Hunt KJ. Dual health care system use is associated with higher rates of hospitalization and hospital readmission among veterans with heart failure. Am Heart J. 2016;174:157-163. doi:10.1016/j.ahj.2015.09.023

11. Jia H, Zheng Y, Reker DM, et al. Multiple system utilization and mortality for veterans with stroke. Stroke. 2007;38(2):355-360. doi:10.1161/01.STR.0000254457.38901.fb

12. Miller DR, Safford MM, Pogach LM. Who has diabetes? best estimates of diabetes prevalence in the Department of Veterans Affairs based on computerized patient data. Diabetes Care. 2004;27(suppl 2):B10-B21. doi:10.2337/diacare.27.suppl_2.b10

13. HEDIS volume 2: technical specifications. Healthcare Effectiveness Data and Information Set. 2015. Accessed September 17, 2020. https://www.ncqa.org/hedis/measures

14. Pollack CE, Hussey PS, Rudin RS, Fox DS, Lai J, Schneider EC. Measuring care continuity: a comparison of claims-based methods. Med Care. 2016;54(5):e30-e34. doi:10.1097/MLR.0000000000000018

15. Bice TW, Boxerman SB. A quantitative measure of continuity of care. Med Care. 1977;15(4):347-349. doi:10.1097/00005650-197704000-00010

16. Jee SH, Cabana MD. Indices for continuity of care: a systematic review of the literature. Med Care Res Rev. 2006;63(2):158-188. doi:10.1177/1077558705285294

17. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383. doi:10.1016/0021-9681(87)90171-8

18. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45(6):613-619. doi:10.1016/0895-4356(92)90133-8

19. Health care benefits overview online edition. US Department of Veterans Affairs. Accessed April 17, 2020. https://www.va.gov/healthbenefits/resources/publications/hbco/index.asp#en

20. Chang HY, Weiner JP, Richards TM, Bleich SN, Segal JB. Validating the adapted Diabetes Complications Severity Index in claims data. Am J Manag Care. 2012;18(11):721-726.

21. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika.
1986;73(1):13-22. doi:10.1093/biomet/73.1.13

22. Wolinsky FD, Miller TR, An H, Brezinski PR, Vaughn TE, Rosenthal GE. Dual use of Medicare and the Veterans Health Administration: are there adverse health outcomes? BMC Health Serv Res. 2006;6:131. doi:10.1186/1472-6963-6-131

23. Asch SM, McGlynn EA, Hogan MM, et al. Comparison of quality of care for patients in the Veterans Health Administration and patients in a national sample. Ann Intern Med. 2004;141(12):938-945. doi:10.7326/0003-4819-141-12-200412210-00010

24. Rose DE, Rowneki M, Sambamoorthi U, et al. Variations in VA and Medicare use among veterans with diabetes: impacts on ambulatory care sensitive conditions hospitalizations for 2008, 2009, and 2010. Med Care. 2019;57(6):425-436. doi:10.1097/MLR.0000000000001119

25. Veterans Health Administration. Accessed October 3, 2019. https://www.va.gov/health/

26. Veterans Health Information Systems and Technology Architecture (VistA). 2019. Accessed October 3, 2019. https://catalog.data.gov/dataset/veterans-health-information-systems-and-technology-architecture-vista

27. Saultz JW, Lochner J. Interpersonal continuity of care and care outcomes: a critical review. Ann Fam Med. 2005;3(2):159-166. doi:10.1370/afm.285

28. Veterans Access, Choice, and Accountability Act of 2014, Pub L No. 113-146 (2014).

29. Veterans Access, Choice and Accountability Act of 2014 (“Choice Act”). US Department of Veterans Affairs. Accessed October 3, 2019. https://www.va.gov/opa/choiceact/documents/Choice-Act-Summary.pdf

30. Maintaining Internal Systems and Strengthening Integrated Outside Networks (MISSION) Act of 2018, Pub L No. 115-182 (2018).

31. VA launches new health care options under MISSION Act. US Department of Veterans Affairs. June 6, 2019. Accessed October 3, 2019. https://www.blogs.va.gov/VAntage/61286/va-launches-new-health-care-options-mission-act/

32. VA health care: estimating resources needed to provide community care. June 12, 2019. Accessed October 3, 2019. https://www.gao.gov/products/GAO-19-478