Independent freestanding emergency departments and implications for the rural emergency physician workforce in Texas

Abstract

Objective

Independent freestanding emergency departments (IFEDs) have proliferated over the last decade, largely in Texas. We examined the IFED physician workforce composition and changes in emergency physician workforce supply across states and in rural Texas over the period of IFED proliferation following a 2009 legislation allowing the licensing of these sites.

Data Sources

IFED websites, Texas Medical Board lookup tool, National Plan & Provider Enumeration System (NPPES), Provider Enrollment and Chain/Ownership System (PECOS), Medicare Physician Shared Patient Patterns, CareSet DocGraph Hop Teaming, Healthcare Provider Database.

Study Design

Descriptive analysis of the IFED physician workforce; quasi‐experimental difference‐in‐difference analysis of Texas emergency physician movement into and out of the state; and difference‐in‐difference‐in‐difference analysis of the change in emergency physician supply between rural and urban areas in Texas compared with other states.

Data Extraction Methods

Using the NPIs obtained through Texas IFED websites and Texas Medical Board data, we examined NPPES/PECOS files, Medicare Physician Shared Patient Patterns, and CareSet DocGraph Hop Teaming for IFED physician practice locations from 2009 to 2017. We extracted all active emergency physicians from a Healthcare Provider Database, derived from a 5% Medicare claims (1999‐2017).

Principal Findings

In 2019, 545 physicians practiced in Texas IFEDs, of which 515 (94.5%) were emergency physicians. We located 533 in previous practice, of whom 522 (97.9%) previously practiced in Disproportionate Share Hospitals and 100 (18.8%) in rural areas. Following legislation to begin licensing IFEDs in 2009, there were on average 42.1 (P < .01) moving into Texas and 17.0 (P < .01) fewer moving out compared with all other states. Our results also indicated that the difference in emergency physician supply between rural and urban Texas was 1,002 (P < .01) fewer than for other states.

Conclusions

New models of health care organizations such as IFEDs have workforce implications that may further exacerbate rural and underserved workforce and access challenges.

What is Known on this Topic

• The number of independent freestanding emergency departments (IFEDs) has grown over the last two decades, with the majority located in the state of Texas.

• IFEDs predominantly locate in areas of higher income, private‐pay patients, and early evidence suggests freestanding emergency departments are associated with increased health care costs and may negatively impact nearby hospital‐based EDs.

What This Study Adds

• Texas‐independent freestanding emergency departments (IFEDs) have recruited emergency medicine physicians largely from rural and Disproportionate Share Hospitals.

• Over the time period of Texas IFED proliferation, the inflow of emergency medicine physicians into Texas has increased more than for other states, and rural Texas saw a decrease in emergency medicine physician supply with the difference between rural and urban Texas worsening compared with other states.

1. INTRODUCTION

Freestanding emergency departments (EDs) have proliferated in the past two decades with supporters of the model promoting the benefits of increased access to care and reduced ED wait times. While most EDs are located within hospitals, freestanding EDs are located away from hospital campuses and come in two major types: hospital‐affiliated and independent. Independent freestanding emergency departments (IFEDs) are distinctly unique entities in that Medicare does recognize them and will not reimburse for services provided in an IFED. Therefore, the policies, incentives, behaviors, and implications of IFEDs are the focus of this study.

There have been a number of concerns raised recently in regard to freestanding EDs and what their true implications are for access, cost, and health care systems. Freestanding EDs, particularly IFEDs, have been more likely to locate in areas of higher income and better payer mixes, suggesting this model of care delivery is not improving access for high‐need communities. ¹ Freestanding EDs have been associated with increased health care costs, ² perhaps not only due to increased access to needed emergency care but also concerning for inappropriate utilization, particularly for services that might be delivered in less costly settings, such as urgent care centers. ³ There has also been concern around freestanding EDs affecting the financial viability of nearby hospital‐based EDs by attracting less complicated, private‐pay patients away from local competitors. ⁴

Related to concerns over location, utilization, and local health care systems, IFEDs may have additional health workforce implications that go beyond their local health care systems. These entrepreneurial models of health care delivery represent new entrants into the workforce market, creating increased competition for the limited resource of emergency medicine physicians. IFEDs’ preferential location in areas with better payer mixes may be creating additional workforce competition for hospitals serving lower income populations, such as Disproportionate Share Hospitals. Access to emergency health care is a particular concern for rural communities, which have struggled with health workforce recruitment. While an estimated 19% of the US population live in rural areas, ⁵ only 14.3% of all emergency medicine clinicians and 10.5% of emergency medicine physicians are practicing in rural counties. ⁶ Even as rural communities have received little benefit from freestanding EDs, they may also be particularly susceptible to workforce loss due to more challenging work environments and less competitive pay. In addition, as the physician workforce can move between states, the policies and practices of one state can have indirect workforce implications for other states.

In October 2009, the 81st Texas State legislature passed the Texas Freestanding Emergency Medical Care Facility Licensing Act and began licensing its first IFEDs in 2010. This legislation was important as it opened an explicit avenue for freestanding EDs not affiliated with a hospital to be licensed. The number of Texas IFEDs went from none to 191 between 2010 and 2016. In 2016, there were only 12 additional IFEDs located in Colorado, Minnesota, and Rhode Island. The majority of IFEDs were for‐profit entities, located in urban areas. ⁷ Regulation of overall freestanding EDs is largely at the state level and varies widely between states. In a 2015 analysis of state regulations, 24 states required a certificate of need, 21 states required state licensure, and only one state, California, did not allow freestanding EDs. ⁸ States with certificate of need requirements, in which a state body must approve certain new health care facilities, had significantly fewer EDs per capita. ⁸ By the legislation, IFEDs in Texas do not require a certificate of need. ⁸ However, freestanding EDs in Texas are required to have at least one physician physically present at all times, ⁹ and they often market themselves based on the availability of ER‐trained physicians. ¹⁰

This study examines the workforce implications of IFEDs in the state of Texas: who are IFED providers, where did they come from, and what happened to the emergency medicine physician workforce across states and in rural communities over the time of IFED proliferation.

2. METHODS

2.1. IFED physician practice patterns

Physicians working at IFEDs are difficult to identify in any one dataset due to the billing behavior in the private setting. In order to identify the supply of the IFED physicians in Texas, we compiled a list of 211 IFEDs released by the Texas Department of State Health Services, as of July 2019. Using information on the location and name of each center, we recorded a list of physicians’ names through individual IFED webpages and retrieved license numbers from the Texas Medical Board. Of the 211 centers, we identified 545 physicians and their license numbers, which we matched to their National Provider Identifier (NPI) numbers through the National Plan & Provider Enumeration System (NPPES) registry (September 2019 Version). From the NPPES registry, we also extracted information on the IFED physician self‐reported specialty, cross‐referencing this with the information available in the Provider Enrollment and Chain/Ownership System (PECOS) files. The NPPES provided information on all 545 IFED physicians, while the PECOS files contained information for 520 physicians on the compiled list. The NPPES registry reported 481 of the 545 IFED physicians as emergency medicine specialists, while the PECOS files reported 490 of the 520 IFED physicians as ED practitioners. When identifying all ED physicians in the country, we adopted a widely inclusive criterion of cataloging physicians as emergency medicine specialists if they reported ED in any one of the sources between the NPPES registry, Unique Physician Identification Number (UPIN) directory, or PECOS files.

Using the constructed list of IFED physicians and their corresponding NPI numbers, we then tracked the physicians’ historical location of practice using Medicare Physician Shared Patient Patterns data for the period of 2009‐2013, and the CareSet DocGraph Hop Teaming data for the period of 2014‐2017. ¹¹ These two data sources are two iterations of a data that attempts to capture patient flows between two NPIs. Both datasets utilize Medicare Part A and Part B claims data to track individual patients visiting different health care providers, using individual NPI, and health care institutions and group practices, using organizational NPI. The data provide a count of unique patients and encounters flowing between two NPIs. In the case with emergency physicians, we are able to capture individual emergency physicians’ NPI from the professional claims and the emergency departments’ NPI from the organizational NPI using the corresponding institutional claims. We exploit the fact that ED physicians working in hospitals will have high volumes of patient flow between their individual NPIs and hospital NPIs at which they work. We are able to use the two patient flow datasets, Physician Shared Patient Patterns and DocGraph, to track these transactions thereby identifying ED physicians working in hospital settings. Using these data sources, we found at least one previous practice location for 533 of the 545 IFED physicians from the compiled list and tracked these physicians between 2009 and 2017 to determine whether they were engaged in hospital ED services. Since we assume that IFED centers provide up‐to‐date information on currently practicing physicians at their respective locations, we associated any attrition of NPIs from the referral data as indication that those physicians transitioned to working from hospital EDs to IFED practices. Given that attrition may also be a result of larger changes in the supply of emergency medicine physicians at the national or state level, we also examined movement and supply of all ED physicians, not just IFED practitioners.

2.2. Movement and supply of ED physicians

To empirically measure the changes in movement of emergency medicine physicians, we used a Healthcare Provider Database from 1999 to 2017 that allows us to examine the preexisting trend before the 2009 Texas IFED legislation. ¹² These data combine several publicly available data sources, including NPPES, PECOS, and ResDAC UPIN directory allowing the tracking of individual health care providers over time. This data source also included the top 5 carrier line performing provider zip codes for each year, which represents the service location, derived from 5% Medicare claims data. Using reported specialties from PECOS files, NPPES registry, and UPIN directory, we identified 49 212 emergency medicine physicians. Using data from 1999 to 2001, we established the baseline of existing emergency medicine physicians before 2002 and where they were practicing using their most frequently used zip codes for each year. We further tracked new emergency medicine physicians joining the workforce and the movements of all emergency medicine physicians between 2002 and 2017 using the top zip codes in every year. We define the inflow of a physician to a state as the physician started to provide services in a state in two consecutive years, and the outflow of a physician from a state as the physician stopped providing services in a state in two consecutive years but still provided services from other states, and new physicians are defined as physicians with no previous service record in any state who started to provide service in a state.

We employed a difference‐in‐difference estimation to examine the effect of the Texas Freestanding Emergency Medical Care Facility Licensing Act in October 2009 on the movement of ED physicians. We model the general specification below:

$${\mathit{Movement}}_{\mathit{st}}={\gamma }_0+{\gamma }_1{P}_s\times T_t+\beta X_{\mathit{st}}+{\eta }_t+{\theta }_s+{\epsilon }_{\mathit{st}}$$

For each of these variables, states are indexed by s in the year t. The outcome variable, ${\mathit{Movement}}_{\mathit{st}}$, includes four different measures of inflow, outflow, net change, and the entry of new ED physicians in Texas compared with other all other states. P_s represents the policy as a pre‐ and post‐2009 indicator variable for state s (1 for 2010 and later, 0 otherwise), and T_t represents a binary variable for Texas and other states in year t. ${\eta }_t$and ${\theta }_s$ are year and state fixed effects. $X_{\mathit{st}}$ is a vector of state‐level covariates obtained from the Census Bureau data including population size, age, race/ethnicity, sex, income, unemployment, and Medicare managed care penetration rate. The policy effect of interest on the movement of ED physicians in Texas relative to other states is represented by ${\gamma }_1$.

Movement of ED physicians in Texas relative to other states may be driven by population trends related to urban‐ and rural‐specific growth. To explore this further, we developed a difference‐in‐difference‐in‐difference estimation to parse out the impact of the policy on ED physician supply in rural Texas relative to urban Texas compared with the same difference in other states. Building from the initial specification, the second model is as follows:

$${\mathit{Supply}}_{\mathit{rst}}={\gamma }_0+{\gamma }_1{P}_{\mathit{rs}}\times T_{\mathit{rt}}+{\gamma }_2{R}_{\mathit{rs}}\times T_{\mathit{rt}}+{\gamma }_3{P}_{\mathit{rs}}\times T_{\mathit{rt}}\times R_{\mathit{st}}++\beta X_{\mathit{rst}}+{\eta }_t+{\omega }_{\mathit{rs}}+{\epsilon }_{\mathit{rst}}$$

The difference‐in‐difference‐in‐difference model includes the addition of rural status, R_st, of state s in year t. The state fixed effects also changed to the state‐rurality level with ${\omega }_{\mathit{rs}}$. For this model, we constructed the rurality indicator based on physician zip codes from the Healthcare Provider Database. We first identified the corresponding counties of the physician zip codes, and we then assigned the rural‐urban status to counties using the Rural‐Urban Continuum Codes (RUCCs). We defined urban counties using metropolitan counties (ie, RUCC 1, 2, and 3) and adjacent nonmetro counties (ie, RUCC 4 and 6), and rural counties using all remaining RUCCs. ¹³ For each state, we aggregated ED physician supply at the rural‐urban level. The control variables, $X_{\mathit{rst}}$, are aggregated to the state rural‐urban level from US Census Bureau county‐level data. Therefore, each state s in year t is also indexed by r for urban‐rural groups.

In the difference‐in‐difference‐in‐difference model, movement no longer represented an appropriate outcome variable as we were not interested in modeling the movement of an ED physician from a rural Texas county to another rural Texas county. Instead, we used supply as the outcome variable using the total number of ED physicians and ED physicians per 10 000 population. This allowed us to identify the differential effect of the 2009 policy on the supply of ED physicians in rural Texas relative to urban Texas, compared with the same rural‐urban difference of all other states. The policy effect of interest in this difference‐in‐difference‐in‐difference estimation is represented by ${\gamma }_3$.

3. RESULTS

3.1. IFED physician practice patterns

We found that 378 of the current IFED physicians as of July 2019 had a practice hospital in 2009, of which 34 were in rural hospitals and 361 practiced at locations participating in Disproportionate Share Hospital (DSH) programs (Table 1). By 2017, we identified practice hospitals for 267 IFED physicians, of which 20 were in rural hospitals and 216 practiced in locations under DSH programs. Since we assume that IFED physicians in the recently compiled list are currently still in practice at the IFEDs, the difference of 111 from the data indicates a decreased productivity at hospitals from these physicians from 2009 to 2017. In addition, we observed that rural hospitals and DSH programs were proportionally more affected by this decline in IFED physicians’ participation. Between 2009 and 2013, the number of IFED physicians practicing in hospitals remains constant as there is an initial growth period of IFED sites during this time. We observed noticeable changes in the number of IFED physicians between 2013 and 2017, during the period that IFED sites increased almost sixfold from 27 to 153. We further examined the geographic distribution of IFEDs opened in Texas since 2010. We found that most IFEDs are in urban areas, consistent with our findings on IFED physicians.

Table 1.

Texas IFED ^a physician practice patterns at hospitals (2009‐2017)

Year	Cumulative Number of IFEDs	IFED Physicians Practicing in hospitals	IFED Physicians in rural counties	IFED Physicians in DSH b programs
2009	0	378	34 (8.9%)	361 (95.5%)
2010	4	404	36 (8.9%)	383 (94.8%)
2011	7	419	39 (9.3%)	397 (94.8%)
2012	15	442	36 (8.1%)	420 (95.0%)
2013	27	438	37 (8.5%)	417 (95.2%)
2014	47	365	27 (7.4%)	328 (89.9%)
2015	87	335	22 (6.6%)	294 (87.8%)
2016	125	313	23 (7.4%)	258 (82.4%)
2017	153	267	20 (7.5%)	216 (80.9%)
Cumulative total c		533	100 (18.8%)	522 (97.9%)

^aIFED stands for independent freestanding emergency departments.

^bDSH stands for Disproportionate Share Hospital (DSH) programs.

^cCumulative total refers to total unique number of physicians identified from 2009 to 2017. Data from author's calculation. Number in parentheses is the percent of IFED physicians identified in that category for a given year.

3.2. ED physician movement

Figure 1 below illustrates the changes in inflow, outflow, and new ED physicians in Texas between 2002 and 2017. ED physician movement trends toward a decline in outflow and increase in inflow and new physicians without adjusting for any covariates.

Figure 1.

Numbers of inflow, outflow, and new Texas emergency medicine physicians (2002‐2016). This Figure exhibits the number of inflow, outflow, and new emergency medicine physicians from 2002 to 2016. We define the inflow of physicians into Texas by physicians that started to bill in Texas for two consecutive years; the outflow of physicians from Texas as physicians that stopped billing in Texas for two consecutive years (but still billed from other states), and new physicians are defined as physicians with no previous billing records that started to bill in Texas. Blue line indicates inflow of physicians, red line indicates outflow of physicians, and green line indicate new physicians. Data were from the Healthcare Provider Database (see ¹² )

Table 2 shows results for the difference‐in‐difference estimation of the 2009 policy on the movement of ED physicians in Texas relative to non‐Texas ED physicians. Texas experienced an inflow increase of 42.1 ED physician, an outflow decrease of 17.0 ED physicians, and a 25.1 increase in new ED physicians in the state compared with all other states after the 2009 policy period. These estimations are all adjusted for a set of state‐level characteristics, in addition to state and year fixed effects. Overall, these trends reflect that Texas experienced greater growth in the supply of ED physicians relative to other states following the passing of the 2009 policy. We examined the plausibility of the parallel trend assumption using leads and lags models (see Figure S1).

Table 2.

Effect of Texas freestanding emergency medical care facility licensing act on emergency physician movements

	(1)	(2)	(3)	(4)
	Inflow (SE a )	Outflow (SE)	Net flow (SE)	New (SE)
Texas X Postenactment	42.05 (3.95)***	‒16.99 (1.62)***	59.04 (4.42)***	25.10 (1.69)***
ln(Population)	‒44.21 (35.30)	‒18.91 (23.54)	‒25.30 (39.42)	36.91 (28.76)
Percent of elder population (Age 65+)	1.69 (1.60)	1.89 (1.24)	‒0.21 (2.06)	2.28 (1.83)
Percent of black population	‒0.33 (1.05)	0.14 (0.43)	‒0.48 (1.27)	0.22 (0.66)
Percent of Hispanic population	4.71 (1.65)***	2.49 (0.85)***	2.22 (1.77)	0.26 (1.30)
Percent of male population	‒15.40 (11.52)	‒2.10 (4.33)	‒13.30 (13.52)	2.01 (6.17)
ln(area median income)	47.28* (26.94)	37.38 (22.91)	9.90 (30.43)	16.43 (31.77)
Unemployment rate	‒0.59 (0.52)	‒0.41 (0.62)	‒0.18 (0.98)	‒0.70 (0.56)
Medicare managed care penetration rate	0.22 (0.19)	0.07 (0.12)	0.15 (0.23)	0.04 (0.16)
Constant	888.38 (775.39)	‒28.58 (357.93)	916.97 (959.62)	‒823.34 (562.08)
State fixed effects	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes
Observations	765	765	765	765
R 2	0.91	0.92	0.56	0.96
Number of states	51	51	51	51

^aStandard errors clustered at the state level in parentheses.

^*** P < .01

^** P < .05

^* P < .1.

3.3. ED physician supply

We found that on average, the rate of decline in the supply of ED physicians in rural regions compared with urban areas was greater in Texas than in other states (Figure 2). Table 3 summarizes the findings of the difference‐in‐difference‐in‐difference estimation, which parses out the effect of the policy in rural and urban Texas, relative to the same difference in all other states. The difference in ED physician supply between rural and urban Texas after the 2009 legislation was 1002 fewer than the difference in ED physician supply between rural and urban regions of all other states. When adjusting for the population, we find a similar relationship with a decrease of 0.55 ED physicians per 10 000 population in rural Texas relative to urban Texas, compared with the same difference in all other states.

Figure 2.

Emergency medicine physician supply in Texas rural and urban areas compared with other states’ rural and urban regions (2002‐2017). A, shows absolute numbers of emergency medicine physicians, and (B) shows emergency medicine physician‐to‐population ratio (per 10,000). Data were from the Healthcare Provider Database (see ¹² )

Table 3.

Effect of texas freestanding emergency medical care facility licensing act on emergency medicine physician supply between rural and urban areas

	(1)	(2)
	Emergency medicine physicians (SE a )	Emergency medicine physicians per 10,000 (SE)
Texas X Rural X Postenactment	‒1,001.75 (27.61)***	‒0.55 (0.13)***
Rural X Post	‒127.82 (34.01)***	0.12 (0.12)
Texas X Post	835.38 (51.22)***	‒0.05 (0.09)
ln(Population)	‒179.59 (301.27)	1.00 (1.01)
Percent of elder population (Age 65+)	‒9.10 (10.16)	‒0.21 (0.13)
Percent of Black Population	‒3.03 (12.36)	0.08 (0.04)*
Percent of Hispanic population	50.06 (21.80)**	‒0.09 (0.06)
Percent of male population	‒83.92 (46.00)*	‒0.59 (0.29)**
ln(area median income)	54.54 (232.25)	0.57 (1.03)
Unemployment rate	‒1.24 (3.23)	0.01 (0.02)
Medicare managed care penetration rate	3.26 (1.54)**	‒0.01 (0.01)*
Constant	6,250.56 (4,093.04)	14.51 (22.19)
State rural/urban area fixed effects	Yes	Yes
Year fixed effects	Yes	Yes
Observations	1,536	1,536
R 2	0.99	0.70
Number of states	51	51

^aStandard errors clustered at the state level in parentheses.

^*** P < .01

^** P < .05

^* P < .1.

4. DISCUSSION

Our study demonstrates Texas IFED providers are predominantly emergency medicine physicians who previously worked in Disproportionate Share Hospitals, with nearly 19% of IFED physicians previously working in rural areas. These findings might reflect normal stock and flow movement of the workforce over time. However, we further found that over the study period, in which the number of Texas IFEDs grew rapidly, the inflow of emergency medicine physicians into Texas increased more than for other states and rural Texas saw both an overall decrease in emergency medicine physicians per population and a greater change relative to urban Texas areas than other states.

Physician movement between states and rural‐urban settings is complex and likely driven by many factors, including health care market changes, organization characteristics, and individual factors. Rural hospital closures, which have been particularly concerning over the past two decades, ¹⁴ are an important push factor that should be considered in interpreting our findings. The implications of our findings may also be varied for different states and communities. Recent growth of the emergency medicine physician and advanced practice clinician workforce suggests a potential oversupply of this workforce. ¹⁵ , ¹⁶ In this case, increased flow of emergency medicine physicians to Texas may not be an issue. However, the emergency medicine workforce is not evenly distributed between states or rural‐urban areas. In 2019, Texas had one emergency medicine physician per 8609 people, while its neighbor, Oklahoma, had one per 11 529 and Massachusetts had one per 5535. ¹⁷ For states and communities that are struggling to recruit this workforce, the growth of freestanding EDs may present a further challenge. Policies to address access to emergency health care should consider IFEDs and the potential implications for the emergency health care workforce.

As IFEDs cannot bill Medicare, it is, perhaps, unsurprising that they do not locate in rural areas where patient populations are often older. Freestanding EDs have been proposed as a policy solution for rural emergency access. MedPAC’s 2018 Report recommended allowing rural isolated IFEDs to bill Medicare and providing additional payments to assist with operational costs. In addition, they recommended cutting payment rates for hospital‐affiliated freestanding EDs within six miles of a hospital‐based ED to control the costs associated with these service sites. ¹⁸ Payments to rural IFEDs would be expected to increase the financial viability of rural IFEDs; however, the sustainability of these sites will also depend on their ability to recruit providers. ¹⁹ Of note, with COVID‐19, IFEDs were temporarily allowed to bill Medicare and Medicaid. ²⁰ Whether this policy is continued remains to be seen. A global recognition of IFEDs by Medicare may ultimately lead to further proliferation of this model. Within existing incentives, it seems likely IFED expansion would occur first in more urban and higher socioeconomic communities, leaving rural and underserved communities further challenged.

In further considering the potential of freestanding EDs to address emergency health care access, the role of the emergency medicine physicians in relation to other clinicians should be further considered. Nurse practitioners, physician assistants, and nonemergency physicians make up nearly 39% of clinicians providing care in emergency departments, ⁶ and the likelihood of being seen by a family medicine physician, physician assistant, or other clinician increases as rurality increases. ²¹ As telehealth expands, the opportunity arises to strategically redesign emergency health care teams, particularly for rural communities. Regional consortia could connect family medicine physicians, physician assistants, and other clinicians to distant emergency medicine physicians to support high intensity care and training. However, financial incentives and payment policies are needed to support new models of care and existing regulatory barriers, such as state scope of practice limitations and freestanding ED requirements for physicians (in some cases, specifically emergency medicine physicians) to be physically present, ²¹ would need to be re‐evaluated.

Our study had a number of limitations. First, the list of Texas IFED providers was obtained through an online review of IFED websites and may reflect organizational self‐interests to promote emergency medicine physicians over other types of physicians or providers. Second, medical specialty in NPPES and PECOS files is self‐reported. We noted a small discrepancy in reported specialty between the two datasets—481 compared to 490 with an emergency medicine specialty, respectively. Third, we were unable to determine the relative amount of service and time physicians spent in rural or Disproportionate Share Hospitals. Physicians can work in multiple sites. As we did not have access to all payer data, we could not determine what proportion of a physician's time and services were provided in different sites and therefore could not determine the total loss of physician effort in rural and Disproportionate Share Hospitals related to the movement to IFEDs. As the most recent Medicare data available were from 2017, we were also unable to determine whether IFED physicians continued to provide some care in rural or Disproportionate Share Hospitals.

Our study highlights important workforce implications to consider with new models of health care delivery. Texas IFEDs have largely recruited emergency medicine physicians from rural and Disproportionate Share Hospitals. While the predominance of these physicians in the IFED workforce is undoubtedly also driven by pay, work environments, and financially challenged health care systems, competition from freestanding EDs will increase the challenge of recruitment and retention for rural and underserved communities. These workforce implications may also hold for other developing models of health care, such as urgent care centers, retail clinics, and direct primary care. These models of care deserve further study. Finally, it is important to note that the growth of Texas IFEDs was supported by state policies and that one state's policies can affect another state's workforce.

Supporting information

Supplementary Material

Appendix S1

Luo Q, Chong N, Chen C. Independent freestanding emergency departments and implications for the rural emergency physician workforce in Texas. Health Serv Res. 2020;55:1013–1020. 10.1111/1475-6773.13587