Steven D. Pizer

BACKGROUND: Cardiovascular disease (CVD) is the leading cause of morbidity and mortality among US veterans, many receiving cardiovascular care across both the Veterans Health Administration (VA) and non-VA hospitals, driven in part by the growth of Medicare Advantage (MA) enrollment. With both systems federally financed, overlapping payment responsibilities raise concern for duplicative federal spending on inpatient CVD care. METHODS: We conducted a national, cross-sectional study using 2021 linked MA and VA data to identify CVD hospitalizations defined using diagnosis-related groups among MA-enrolled veterans with VA benefits. Each CVD-related hospitalization (eg, heart failure exacerbation) was categorized by source of payment: MA-paid community hospital care, VA-paid community hospital care, or VA-direct care at VA hospitals. High-veteran plans were defined as ≥20% veteran enrollment. Multinomial logistic regressions were used to evaluate associations between high-veteran MA plan enrollment and payment source, adjusting for patient characteristics and state fixed effects. Analyses were stratified by admission complexity and elective status. RESULTS: Sample included 76 722 MA-enrolled veterans (mean [SD] age, 78.0 [8.8] years, 97.9% male), of which 10 366 (13.5%) were in high-veteran MA plans, and 66 356 (86.5%) were in high-veteran MA plans. Veterans in high-veteran MA plans were less likely to have MA-paid community hospitalizations than those in low-veteran MA plans (43.2% versus 69.2%; –26.1 percentage-points [95% CI, –27.1 to –25.1]) but more likely to have VA-paid community care (28.1% versus 16.0%; +12.2 percentage-points [95% CI, 11.3–13.0]) and VA-direct care (28.7% versus 14.8%; +13.9 percentage-points [95% CI, 13.0–14.8]). Results were consistent across admission types and medical complexity. CONCLUSIONS: Veterans enrolled in high-veteran MA plans were more likely to have inpatient CVD care financed by the VA, suggesting that MA plans may be shifting costs despite receiving capitated payments intended for inpatient care. Policy reforms are needed to reduce duplicative federal spending and improve coordination of cardiovascular care.

Objective: To evaluate the impact of Veterans' urgent care use in non‐Veterans Health Administration (VHA) clinics on their later use of specialty and emergency care. Study Setting and Design: A retrospective cohort study analyzing urgent care, specialty care, and emergency department (ED) visits using VHA administrative and claims data. Veterans' first urgent care visit was the index visit, and subsequent utilization was tracked for up to 1 year. We used an instrumental variable (IV) approach to estimate the effect of non‐VHA urgent care visits on the proportion of subsequent visits in non‐VHA clinics. Data Sources and Analytic Sample: We used VHA administrative and community care claims data (2021–2022) obtained from the Corporate Data Warehouse. Our sample comprised VHA enrollees with at least one urgent care visit at a VHA or non‐VHA urgent care clinic in 2021. Principal Findings: Of the 111,898 Veterans in the sample, 90,651 (81%) had their urgent care visit in the VHA, while 21,247 (19%) visited a non‐VHA clinic. IV models showed that Veterans using non‐VHA urgent care had 40 percentage points higher proportions of non‐VHA specialty care within 30 days compared to those using VHA (95% CI: 37, 43). Effects stabilized at 28 percentage points within 365 days post urgent care visit (95% CI: 26, 30). In contrast, the effect of community urgent care on subsequent ED use outside the VHA system after 30 days of urgent care use was negligible and statistically insignificant. Conclusions: Veterans' use of non‐VHA urgent care can lead to increased subsequent use of non‐VHA specialty care but has no effect on non‐VHA emergency care utilization within 60 to 365 days after urgent care visit. While expanded community care improves access in the short run, it also introduces care coordination challenges throughout an episode of care that may negatively affect health outcomes.

Introduction: Obesity is an increasing chronic medical condition facing the veteran population. Glucagon-like peptide 1 receptor agonists (GLP-1s) have emerged as an optimal treatment for obesity- and related-medical disorders. As the Veterans Health Administration (VHA) continues to expand community-based care purchased under the MISSION Act, it is unknown whether veterans receiving community-based primary care experience comparable access to GLP-1 prescriptions. Methods: In this national analysis of VHA data, we assessed differences in GLP-1 prescribing among veterans with overweight or obesity in VHA direct care compared with community care. Results: Veterans with 3 or more primary care visits receiving VHA-direct primary care had 1.2 percentage points (pp) higher adjusted probability of GLP-1 prescriptions (20% higher odds) than veterans receiving mostly community primary care. Black and Asian veterans had 1.8-pp and 2.2-pp lower probabilities of GLP-1 prescriptions (26% and 34% lower odds, respectively) compared with White counterparts across both settings. Conclusion: Researchers and policymakers should continue to monitor whether differences in care between the direct and community settings contribute to disparities in GLP-1 access.

Objective: To estimate the effect of two provider supply components—(1) clinic time from providers and (2) the number of new patient appointments provided per day—on the wait times for new patient appointments in specialty care. Study Setting and Design: In this observational analysis, we examine wait times in the largest public health care delivery system in the US—the Veterans Health Administration (VHA). We introduce new data on wait times for specialty appointments at the VHA. We estimate the impact of the two supply components on new patient wait times for specialty care appointments, while controlling for patient characteristics, local area characteristics, medical center fixed effects, and national changes over time and across seasons. We use an instrumental variables approach to handle endogeneity concerns. Data Sources and Analytic Sample: Secondary administrative data from the VHA were analyzed, focusing on four specialties: cardiology, orthopedics, urology, and gastroenterology. The sample includes 126 medical centers with the specialty departments observed at 2‐week intervals spanning July 2014 to December 2019. Principal Findings: Our estimates indicate that a 10% increase in the sample average clinician time from specialists (equivalent to 2.5 full‐time clinicians per 100 enrollees) would reduce the average wait time for a new patient appointment by 1.1 days [95% CI: 1.08–1.13] or 3.9% of the sample average wait time. A 10% increase in the average number of new patients seen per clinic day by providers (0.37 visits) would reduce the average wait time by 4.0 days [95% CI: 3.97–4.03] or 12.9% of the sample average. Conclusions: Increasing clinic time and the new patient visit rate among providers can improve access to care for new patients. Scheduling policies that prioritize new patients can affect this visit rate without substantially reducing access for other patients.

Introduction: Telehealth delivers more than half of mental health care in the United States amid persistent mental health workforce shortages. However, little is known about how telehealth relates to clinicians’ work behavior, specifically their visits per clinic day and probability of turnover. Methods: We examined mental health clinicians working at the Veterans Health Administration who collectively delivered 26 million visits. We regressed clinicians’ monthly visit volume (supply), number of clinic days worked (input), visits per clinic day (throughput), and probability of turnover (retention) on the proportion of their mental health visits delivered via video. We used variation in broadband availability as an instrumental variable for video visit use. Results: A 10 percentage-point higher proportion of video visits was associated with no significant difference in total visits, 0.9 fewer clinic days per month (8.5% of the sample mean), 0.5 more visits per clinic day (6.2% of the sample mean), and a 0.4 percentage-point lower probability of monthly turnover (46% of the sample mean). Among psychologists and social workers, higher video share was associated with more visits per clinic day and fewer clinic days, whereas psychiatrists showed fewer visits per clinic day. Conclusion: Telehealth may shape clinicians’ work behavior, including visits per clinic day, allocation of clinic days, and retention.

Objective: To investigate the relationship between community care (CC) treatment, surgical complexity, and postoperative surgical outcomes in spine surgeries among Veterans. Data Sources and Study Setting: Veterans Health Administration (VHA) sample with data sourced from the Corporate Data Warehouse and CC claims. Study Design: To evaluate differences in VHA and CC spine surgical complexity and outcomes, we first characterized VHA patients with lumbar spinal stenosis (LSS) who received spine surgery in the VHA or CC. Then, we estimated adjusted naïve logistic regression models to calculate the effect of CC on the probability of having a complex spine surgery, 30‐day readmission, 30‐day complication, and 1‐year reoperation. Finally, we estimated adjusted 2‐stage models using an instrument for primary care provider's historical CC referral rates and imaging rates as a semi‐parametric Newey correction for sample selection. Analytic Sample: LSS‐diagnosed patients living ≤ 80 miles from a VHA facility that performed at least one spine surgery between January 1, 2019 and December 31, 2022. Principal Findings: Of the 41,726 LSS‐diagnosed patients, 7496 (18.0%) had spine surgery within 1 year of diagnosis. 2920 (39.0%) were VHA surgeries and 4576 (61.0%) were in CC. In the naïve model, CC surgery was associated with a 26.61 percentage point (pp) increase in the probability of having a complex surgery (95% CI 24.17, 29.05), a 4.31 pp increase in readmission (95% CI 2.76, 5.85), and a 6.80 pp increase in reoperation (95% CI 5.21, 8.40). After accounting for characteristics associated with the likelihood of surgery, CC, and outcomes, only the effect of CC use on the probability of a complex surgery was significant (36.48; 95% CI 22.69, 50.27). Conclusions: We found no difference in surgical outcomes between VHA and CC patients. Since CC patients were more likely to receive complex spine surgeries, the VHA paid for more costly, resource‐intensive procedures with no improvements in quality.

Objective: To examine the relationship between mental health provider productivity, staffing levels, and suicide‐related events (SREs) among U.S. Veterans receiving care within the Veterans Health Administration (VHA), focusing on therapy and medication management providers. Data Sources/Setting: We analyzed administrative data from the Department of Defense and VHA (2014–2018), encompassing 109,376 Veterans who separated from active duty between 2010 and 2017. Design: A longitudinal design estimated the effects of facility‐level provider work rate and staffing on SREs, adjusting for patient and facility characteristics. An instrumental variables (IV) approach addressed potential endogeneity. Data Collection/Extraction Methods: Data were obtained from the VHA Corporate Data Warehouse and the VHA Survey of Enrollees. Principal Findings: A 1% increase in therapy provider work rate led to a 12.1% increase in SRE probability, regardless of staffing levels. Conversely, a 1% increase in staffing levels led to a 1.6% reduction in SREs, with the largest effect in low‐staffed facilities. For medication management providers, work rate had no overall impact on SREs, except in medium‐staffed facilities. A 1% increase in staffing levels for medication management providers led to a 1.7% reduction in SREs. Conclusions: Increased work rates, particularly in low‐staffed VHA facilities, may elevate suicide‐related risks. In contrast, staffing increases simultaneously improve access and reduce adverse outcomes. Where possible, policymakers should prioritize staffing growth over productivity gains to improve access to mental health clinics and ensure Veteran safety and care quality.

Objective: The objective of this analysis was to evaluate the effect of resident program training size on clinician productivity and turnover in the Veterans Health Administration (VHA), the largest education and training platform for medical professionals in the United States.Data Sources: We retrieved administrative data on training programs and training facilities from the VA Office of Academic Affiliations and the VHA Corporate Data Warehouse. Data on primary care physician shortage areas were retrieved from the Health Resources and Services Administration.Study Design: We used a quasi‐experimental instrumental variables 2SLS design and constructed an exogenous predicted training allocation treatment variable as a function of the total national training program allocation. The outcome was clinical staff productivity and turnover. Secondary analyses stratified results using Health Professional Shortage Areas data (HPSA).Data Collection/Extraction Methods: Data were obtained for a national dataset of 141 VHA medical facilities and 26 specialties that hosted training programs across 11 years from 2011 to 2021 (N = 132,177).Principal Findings: Instrumental variables results showed that on average, an increase of one training slot in a specialty leads to a decrease of 0.039 visits per standardized clinic day (p < 0.001) and a 0.02 percentage point increase in turnover (p < 0.001). The direction of this association varied by specialty: while psychiatry and psychology specialties saw a decline in productivity, fields such as primary care and cardiology experienced an increase in productivity. HPSA stratified results indicate that negative effects on productivity and turnover are driven by areas with little to no primary care physician shortage, whereas shortage areas experienced a small increase in productivity and no effect on turnover.Conclusions: This quasi‐experimental evaluation indicates that resident training program size is associated with reduced productivity and increased turnover in specialties such as psychiatry and in facilities with high baseline productivity. However, in specialties like primary care and cardiology, as well as areas with shortages of primary care, larger training programs are associated with increased productivity.

Objective: To estimate a causal relationship between mental health staffing and time to initiation of mental health care for new patients. Data Sources and Study Setting: As the largest integrated health care delivery system in the United States, the Veterans Health Administration (VHA) provides a unique setting for isolating the effects of staffing on initiation of mental health care where demand is high and out‐of‐pocket costs are not a relevant confounder. We use data from the Department of Defense and VHA to obtain patient and facility characteristics and health care use. Study Design: To isolate exogenous variation in mental health staffing, we used an instrumental variables approach—two‐stage residual inclusion with a discrete time hazard model. Our outcome is time to initiation of mental health care after separation from active duty (first appointment) and our exposure is mental health staffing (standardized clinic time per 1000 VHA enrollees per pay period). Data Collection/Extraction Methods: Our cohort consists of all Veterans separating from active duty between July 2014 and September 2017, who were enrolled in the VHA, and had at least one diagnosis of post‐traumatic stress disorder, major depressive disorder, and/or substance use disorder in the year prior to separation from active duty ( N = 54,209). Principal Findings: An increase of 1 standard deviation in mental health staffing results in a higher likelihood of initiating mental health care (adjusted hazard ratio: 3.17, 95% confidence interval: 2.62, 3.84, p < 0.001). Models stratified by tertile of mental health staffing exhibit decreasing returns to scale. Conclusions: Increases in mental health staffing led to faster initiation of care and are especially beneficial in facilities where staffing is lower, although initiation of care appears capacity‐limited everywhere.

Objective: To investigate whether the Veterans Health Administration's (VA) 2019 Referral Coordination Initiative (RCI) was associated with changes in the proportion of VA specialty referrals completed by community‐based care (CC) providers and mean appointment waiting times for VA and CC providers.Data Sources/Study Settings: Monthly facility level VA data for 3,097,366 specialty care referrals for eight high‐volume specialties (cardiology, dermatology, gastroenterology, neurology, ophthalmology, orthopedics, physical therapy, and podiatry) from October 1, 2019 to May 30, 2022.Study Design: We employed a staggered difference‐in‐differences approach to evaluate RCI's effects on referral patterns and wait times. Our unit of analysis was facility‐month. We dichotomized facilities into high and low RCI use based on the proportion of total referrals for a specialty. We stratified our analysis by specialty and the staffing model that high RCI users adopted: centralized, decentralized, and hybrid.Data Collection/Extraction Methods: Administrative data on referrals and waiting times were extracted from the VA's corporate data warehouse. Data on staffing models were provided by the VA's Office of Integrated Veteran Care.Principal Findings: We did not reject the null hypotheses that high RCI use do not change CC referral rates or waiting times in any of the care settings for most specialties. For example, high RCI use for physical therapy—the highest volume specialty studied—was associated with −0.054 (95% confidence interval [CI]: −0.114 to 0.006) and 2.0 days (95% CI: −4.8 to 8.8) change in CC referral rate and waiting time at CC providers, respectively, among centralized staffing model adopters.Conclusions: In the initial years of the RCI program, RCI does not have a measurable effect on waiting times or CC referral rates. Our findings do not support concerns that RCI might be impeding Veterans' access to CC providers. Future evaluations should examine whether RCI facilitates Veterans' ability to receive care in their preferred setting.