a b s t r a c t

Background: The Affordable Care Act has impacted the insurance mix of emergency department (ED) visits, yet the degree to which this has influenced provider behavior is not clear.

Methods: This was a difference-in-differences (DID) analysis of ED-visit data from five states in 2013 and 2014. Sample states included 3 expanding Medicaid under the ACA, 1 rejecting ACA funding and delaying an eligibility expansion, and 1 with no eligibility change. We included self-pay and Medicaid patients aged 27 to 64 years. A subsample analysis was done for chest pain visits. DID logistic models were estimated for likelihood of admission for given Medicaid-paid ED visits in expansion states as compared to non-expansion states. Among chest pain visits we assessed likelihood given visits resulted in admission or advanced cardiac imaging, where clinician dis- cretion may be more significant.

Results: A total of 8,157,748 ED visits with primary payer Medicaid and self-pay were included, of which 331,422

were for chest pain. The proportion of visits paid for by Medicaid rose in expansion states by between 15.8% and 38.9%. Medicaid eligibility expansion was associated with increased odds of admission (OR 1.070 [95% CI 1.051-1.089]). Among chest pain visits, expansion was associated with increased odds of admission (OR 1.294 [95% CI 1.144-1.464]), but not advanced cardiac imaging (OR 1.099 [95% CI 0.983-1.229]).

Conclusion: Medicaid expansion was associated with small increases in ED visit admissions across the board and among the subgroup of patients presenting with chest pain.

(C) 2020

1. Introduction

The Patient Protection and Affordable Care Act (ACA) provided nu- merous incentives for state expansion of Medicaid programs in 2014. In particular, the ACA sought to expand eligibility to childless adults up to 133% of the Federal Poverty Line. However, the 2012 supreme court ruling in National Federation of Independent Business versus Sebelius rendered significant expansion of Medicaid under the ACA an invalid exercise of Congress’s spending power [1], prompting many states to reject expanded Medicaid coverage provided by the ACA. Ulti- mately, 21 states expanded Medicaid as a result of the ACA at the time of rollout in 2014 [2]. This state-level variation in Medicaid eligibility by in- come and state provides an opportunity to explore the effects of Medic- aid expansion on health care services utilization [3,4].

? This work was supported by the Emergency Medicine Foundation Medical Student Research grant. This work was presented at American College of Emergency Physicians Scientific Assembly on October 1st, 2018 in San Diego, CA.

* Corresponding author.

E-mail address: [email protected] (A.T. Janke).

Much work has focused on how patients’ health behavior is affected by newly enrolling in health insurance, especially with respect to acute care and emergency department (ED) visits [5-8]. Prior literature on expanding Medicaid eligibility has not always shown benefit for clinical outcomes [7,8]. In general, though, prior literature has established a benefit to insurance status on healthcare access and even mortality [9- 12], and from the ED those with better insurance status are more likely to be admitted to the hospital [13,14].

Controversy remains as to the net effect of statewide Medicaid ex- pansion programs under the ACA on the number of ED visits [15-17], but there is a growing consensus that eligibility expansion did result in a net increase in insurance-payed ED visits [18]. With so much atten- tion payed to patient’s choices, and to quality improvement efforts [19- 21], there has been comparatively less done to decipher how health care provider behavior is affected when the population payer mix changes (towards Public insurance and away from self-pay or no insurance). Specifically, does the provider and system-level management of a given medical condition change with large expansion of a public insur- ance program? Part of the answer lies with changes in the presenting patients – their acuity, comorbidities, and access to other healthcare

https://doi.org/10.1016/j.ajem.2019.12.050

0735-6757/(C) 2020

and social resources. There may also be a change in the general propen- sity of providers to admit or discharge, or to perform advanced diagnos- tic testing.

Provider discretion is often an important driver of variation in utili- zation [22], and chest pain is among the most common conditions asso- ciated with provider variation in testing and admission [23]. Low- and moderate-risk chest pain patients can be discharged without follow up, sent for outpatient testing, or admitted for testing. The effect of ex- pansion on provider or hospital practice patterns is unknown, particu- larly for discretionary conditions such as chest pain.

The goal of this investigation is to examine the effect of statewide Medicaid expansion under the ACA on Hospital admission rates from the ED specifically for the payer groups Medicaid and self-pay. In a sub- set analysis of chest pain visits among this payer group, where provider discretion in practice may be more significant, we examine the effect of expansion on admission rates and advanced cardiac imaging.

1. Methods
   1. Study design and setting

This was a secondary analysis of existing publicly available data from the Healthcare Cost and Utilization Project (HCUP). We used the State Inpatient Databases (SID), which includes data from inpatient stays in- cluding those that were the result of admission from a hospital affiliated ED, and State Emergency Department Databases (SEDD), which include visits to hospital affiliated EDs which did not result in admission to the same institution [24,25]. When combined, these datasets represent all ED visits in a year in the given state. Our study dataset included five states where both inpatient and ED data in the year immediately before (2013) and after (2014) the ACA Medicaid expansion were available: Florida, Wisconsin, Arizona, Iowa, and Kentucky. Florida did not expand Medicaid eligibility in the study period. Wisconsin reduced eligibility for adults with children from 200% of the federal poverty level (FPL) to 100% in April 2014 and simultaneously expanded enrollment to include childless adults, but did not expand eligibility under the terms of the ACA. Arizona, Iowa, and Kentucky all expanded eligibility in January 2014 to include childless adults up to 133% of the FPL under the terms of the ACA.

Selection of participants

The study was limited to ED visits for patients’ whose primary payer was either listed as Medicaid or self-pay. Here the self-pay designation means uninsured. Only those patients aged 19 through 64 years were included, as the ACA Medicaid expansion did not change eligibility for children (covered by Medicaid already) and older adults (who qualify for Medicare). For our secondary analysis limited to ED visits we only in- cluded visits with a Clinical Classification Software (CCS) code for undif- ferentiated chest pain (CCS code 102). This definition has been used in prior work to identify and profile hospital admission and imaging rates for ED patients for whom coronary artery disease remains a diag- nostic consideration, but where a firm diagnosis, such as non-ST seg- ment elevation myocardial infarction (NSTEMI), unstable angina, or pneumonia, is not given. [26,27].

Outcomes

The primary outcome was admission to the hospital (as opposed to discharge from the ED or transfer to another facility). The secondary outcome, addressed only in our secondary analysis, was performance of advanced cardiac imaging, either during the ED visit or during the re- sultant hospital admission. This was identified using the following Cur- rent Procedural Technology (CPT) codes: exercise electrocardiography (93015-93,018), exercise echocardiography (93350), coronary com- puted tomography angiography (75574), and myocardial perfusion

imaging (78452). These were chosen as they are the most common non-Invasive approaches to testing for coronary artery disease, and have been the focus of prior research on chest pain evaluations initiated in the ED [22].

Analysis

Prior to limiting the analysis to those with primary payer Medicaid or self-pay, we tabulated number of ED visits across all primary payers to summarize overall changes in the payer mix. We utilized a difference-in-differences (DID) analysis to assess if ED visits in states expanding Medicaid eligibility were more likely to result in admission after implementation of those changes, and in secondary analysis lim- ited to chest pain visits if these were more likely to result in admission or in advanced diagnostic testing for coronary artery disease. We con- structed a logistic regression model [28] of ED visit-level data with each of the following binary outcomes: admission, and, for the second- ary analysis, any of the four advanced diagnostic tests as noted above for coronary artery disease. Covariates included age, sex, race, income quar- tile of patient’s zip code, and primary payer. We also included state- level fixed effects. The analysis was structured so that those Medicaid visits which took place in a state that expanded Medicaid after the eligi- bility change in January 2014 (Arizona, Iowa, and Kentucky) and April 2014 (Wisconsin) were the intervention group and all other visits were the control group (that is, all other Medicaid visits, and all self- pay visits in expansion and non-expansion periods are part of the con- trol group). Because Wisconsin chose to opt out of the Medicaid expan- sion but did ultimately expand eligibility among childless adults starting in April 2014; we include Wisconsin among expanding states in our analysis. This study was deemed exempt by the Institutional Review Board. All analyses were performed Stata version 13 (Stata Corp, College Station, TX).

1. Results
   1. Characteristics of study subjects

Overall, there were 17,981,410 ED visits for patients aged 27-64 across all five states in 2013 and 2014. 8,157,879 of these were self- pay and Medicaid ED visits, and 331,427 of these were discharged from the ED or hospital with a CCS code for chest pain. Table 1 combines data from our sample with publicly available data from Kaiser Family

Table 1

Population proportion of ED visit payer proportion of insurance status.

	Medicaid			Self-pay
	ED visit payer	Population		ED visit payer	Population
Iowa 2013	23%	17%		9%	17%
2014	31%	18%		6%	10%
Kentucky 2013	22%	18%		14%	24%
2014	41%	23%		9%	10%
Arizona 2013	29%	19%		17%	21%
2014	40%	21%		14%	13%
Wisconsin 2013	27%	17%		9%	17%
2014	33%	17%		7%	12%
Florida 2013	27%	17%		20%	26%
2014	27%	18%		17%	24%

This table depicts the proportion of ED visits payed for by Medicaid and under self-pay sta- tus, drawn from the study sample, against the proportion of the population insured by Medicaid and under self-pay status. Population-level insurance status was obtained from the Kaiser Family Foundation [2].

Foundation [2] on population-level insurance status to depict the per- centage of ED visits paid by Medicaid or self-pay alongside the percent- age of the population with these insurance statuses. Among visits that were either Medicaid or self-pay, the proportion of overall Medicaid and self-pay visits that were paid for by Medicaid rose in all expansion states, from +19.5% in Arizona, +21% in Iowa, and + 34.9% in Kentucky. In Florida, which did not expand, the proportion rose by 2.1% and in Wisconsin, which delayed an eligibility expansion and rejected ACA funding, the proportion rose by 15.8%. Table 2 gives a summary of total visits by state and year, as well as proportions for each of age cat- egories, sex, race, primary payer. Table 3 provides similar results for the sample as limited to chest pain visits.

Main results

The adjusted odds ratio (aOR) for Medicaid eligibility expansion on likelihood of ED visit resulting in admission to the hospital was 1.070 (95% CI 1.051-1.089). Table 4 shows the difference-in-differences anal- ysis for Medicaid expansion on our primary outcome, the likelihood of hospital admission, in the primary analysis sample of all Medicaid and self-pay ED visits.

Secondary analysis

Among self-pay and Medicaid ED visits, we found that Medicaid ex- pansion was associated with a given visit’s increased odds of admission (aOR 1.294 [95% CI 1.144 to 1.464]), but not a statistically significant in- creased odds of advanced cardiac imaging (aOR 1.099 [95% CI 0.983-1229]). Table 5 depicts results from the secondary analysis re- stricted to chest pain visits.

1. Discussion

Beyond the increased propensity of individuals to visit the ED after insurance enrollment, our study suggests that Medicaid expansion is as- sociated with an increase in hospital admissions. Our study addresses both the broad outcome, likelihood of admission among all ED visits, as well as the narrower case of the common, discretionary condition of chest pain. Notably, although the overall trend in the states studied was towards decreased admission rates and advanced cardiac testing

(as seen in Tables 1 and 2), our study aims to isolate only the effect of Medicaid expansion policy, independent of other broad trends in re- source utilization. In absolute terms, our findings correspond to approx- imately one additional admission for every 100 ED visits, and in the secondary analysis approximately one additional admission for every 40 chest pain visits. To what extent this is Uninsured patients closing the gap in access to care or health care providers and systems responding to the financial incentives independent of patient presenta- tion is not elucidated in this study.

The study design has a number of unique advantages. First, while in- dividual insurance status changes, by accounting for it as a covariate, our study focuses on the effect of population wide changes in coverage. Sec- ond, our analysis focuses on the ED, where physicians cannot choose which patients to see, thus established physician-patient relationships do not act as a confounder. Third, the DID analysis allows us to isolate changes attributable exclusively to the insurance expansion by subtracting underlying trends common to both expanding and non- expanding states alike. Consequently, while introduction of health ex- changes, modification to Medicaid programs, technology, practice, and financial changes occurring concurrent to the ACA may affect the ob- served change in outcomes, these are common to all states, and are therefore accounted for by the DID design.

Our data provide evidence that health insurance expansion results in an overall increase in resource utilization among Medicaid beneficiaries [29-31]. This corroborates with the existing evidence that insurance sta- tus affects admission and resource utilization [32]. As pressure mounts to control Healthcare costs, for example by stemming admissions from the ED or tackling overutilization of Advanced imaging modalities, health policy makers and researchers must address how the patient payer mix changes the effectiveness of such initiatives. For example, as is the test case for the present study, the benefit of admission and advanced testing for coronary artery disease, as an alternative to a close follow-up visit with a cardiologist, depends on patients’ access to outpatient follow up [23,33]. A patient with barriers to obtaining follow up care or prescrip- tions may be better off with a one-stop shopping model of acute care for chest pain (those same individuals who might have been previously uninsured). Another patient, more comfortable with their ability to access urgent outpatient follow up after evaluation in the ED, may be discharged. The effect of improved follow up access after new insurance enrollment, might facilitate a lower overall admission rate.

Table 2

Summary statistics all Medicaid and self-pay ED visits by state.

	All states		Arizona		Iowa		Kentucky		Wisconsin		Florida
	2013 2014		2013 2014		2013 2014		2013 2014		’13-Apr ’14 Apr-Dec ’14		2013 2014
N	4,046,980 4,110,768		588,160 664,318		198,539 218,367		522,159 618,458		521,620 340,045		2,216,568 2,216,568
Female	59.0% 58.8%		57.6% 49.4%		60.4% 59.7%		58.6% 49.3%		58.6% 0.5745912		59.3% 59.2%
Age
27-35	38.1% 37.9%		36.9% 36.2%		41.7% 40.4%		37.2% 35.7%		40.8% 40.9%		37.7% 38.4%
36-45	27.5% 27.6%		28.2% 27.9%		27.1% 27.2%		28.2% 28.8%		27.7% 27.3%		27.2% 27.3%
46-55	22.7% 22.5%		23.0% 23.3%		21.2% 21.7%		22.9% 23.7%		21.4% 21.7%		22.9% 22.2%
56-64	11.7% 11.9%		11.9% 12.7%		10.0% 10.6%		11.6% 11.9%		10.1% 10.2%		12.2% 12.1%
Race
White	57.4% 57.1%	53.1%	52.0%	76.7%	76.9%	84.5%	83.8%	61.9%	58.9%	49.3%	49.2%
Black	21.2% 21.1%	8.9%	8.8%	12.7%	13.2%	10.9%	11.0%	24.6%	27.4%	26.9%	27.3%
Hispanic	17.7% 17.5%	31.0%	30.7%	4.6%	4.5%	3.7%	3.1%	8.5%	8.4%	20.8%	20.3%
Asian	0.6% 0.8%	1.1%	2.0%	0.7%	0.8%	0.2%	0.3%	1.2%	1.1%	0.5%	0.5%
Other	1.1% 1.4%	4.7%	5.5%	0.8%	0.8%	0.0%	1.2%	2.1%	2.4%	0.1%	0.1%
Income of patient’s ZIP code
Quartile 1	47.8% 49.4%	49.8%	50.4%	21.9%	15.6%	62.0%	59.9%	34.5%	38.2%	49.4%	51.2%
Quartile 2	31.3% 31.3%	27.5%	28.6%	48.8%	50.0%	25.3%	26.9%	37.9%	36.4%	30.5%	30.7%
Quartile 3	14.3% 12.7%	13.7%	12.3%	23.1%	26.6%	8.9%	8.8%	20.0%	17.8%	13.5%	11.8%
Quartile 4	3.6% 3.5%	4.6%	4.3%	4.7%	6.5%	1.3%	2.2%	6.4%	6.5%	3.0%	2.8%
Payer
Medicaid	51.8% 63.3%	57.1%	76.6%	55.0%	76.0%	46.8%	81.7%	61.7%	77.4%	48.9%	51.0%
Self-pay	48.2% 36.7%	42.9%	23.4%	45.0%	24.0%	53.2%	18.3%	38.3%	22.6%	51.1%	49.0%
Admission	17.0% 16.5%	18.2%	17.7%	13.4%	13.3%	14.9%	15.0%	17.3%	16.2%	17.5%	17.0%

Table 3

Summary statistics Medicaid and self-pay ED visits for chest pain by state.

	All states			Arizona			Iowa			Kentucky		Wisconsin			Florida
	2013	2014		2013	2014		2013	2014		2013 2014		’13-Apr ’14	Apr-Dec ’14		2013	2014
N	164,775	166,647		26,281	29,270		7872	8692		23,860 27,977		21,054	21,054		85,711	88,539
Female	54.3%	54.0%		52.8%	49.9%		53.4%	53.4%		54.9% 49.8%		54.7%	0.53956125		54.5%	54.4%
Age
27-35	23.6%	23.2%		21.8%	21.4%		25.2%	23.1%		21.1% 20.9%		26.1%	25.9%		24.1%	24.3%
36-45	30.1%	30.0%		29.8%	28.7%		31.7%	29.9%		31.4% 32.0%		30.5%	30.5%		29.6%	29.7%
46-55	31.1%	31.1%		31.9%	32.3%		30.1%	32.9%		32.3% 31.7%		30.0%	29.8%		30.8%	30.5%
56-64	15.2%	15.7%		16.6%	17.6%		13.0%	14.1%		15.2% 15.4%		13.4%	13.8%		15.5%	15.6%
Race
White	54.8%	54.9%	50.9%		51.0%	75.5%		74.6%	85.7%	84.8%	55.6%		53.9%	45.3%		45.0%
Black	22.8%	22.6%	10.8%		10.7%	14.3%		15.4%	10.7%	10.7%	31.2%		32.4%	28.7%		29.6%
Hispanic	18.9%	18.5%	32.9%		31.7%	4.2%		4.7%	2.9%	2.8%	9.3%		9.0%	22.7%		21.8%
Asian	0.6%	0.7%	1.1%		1.9%	0.6%		0.6%	0.2%	0.2%	1.0%		0.9%	0.5%		0.4%
Other	0.8%	1.1%	3.1%		3.7%	1.1%		0.9%	0.0%	1.0%	1.4%		1.9%	0.1%		0.1%
Income of patient’s ZIP code
Quartile 1	49.3%	50.6%	49.8%		49.8%	21.2%		16.1%	63.5%	61.9%	38.1%		40.4%	50.6%		52.0%
Quartile 2	29.6%	30.3%	27.8%		29.5%	47.2%		48.3%	24.1%	25.6%	35.1%		35.2%	28.8%		29.7%
Quartile 3	14.2%	12.4%	13.7%		12.7%	25.4%		27.2%	8.2%	7.8%	19.2%		17.1%	13.8%		11.6%
Quartile 4	3.5%	3.4%	4.5%		4.0%	4.5%		6.8%	1.2%	2.1%	6.2%		6.3%	3.0%		2.8%
Payer
Medicaid	51.2%	50.0%	56.9%		77.8%	55.1%		77.5%	48.8%	83.9%	60.6%		78.3%	47.5%		49.7%
Self-Pay	48.8%	50.0%	43.1%		22.2%	44.9%		22.5%	51.2%	16.1%	39.4%		21.7%	52.5%		50.3%
Admission	10.1%	7.9%	5.1%		3.5%	3.5%		3.1%	5.6%	4.5%	13.1%		9.9%	12.6%		10.6%
Advanced cardiac imaging	7.4%	7.6%	11.4%		11.0%	8.3%		7.1%	11.0%	11.1%	5.9%		3.5%	6.9%		6.4%

While our approach has some methodological strengths, there are inherent limitations as well. Our analysis relies on discharge diagnoses rather than patient presenting complaint. That is, our subsample analy- sis of chest pain patients is those discharged with a diagnosis of nonspe- cific chest pain (a lower risk group than the full sample of patients presenting for chest pain). In addition, our DID modeling can be interpreted to yield unbiased estimators of the effect of Medicaid expan- sion on ED resource utilization only if certain assumptions are met. Most importantly, our model requires the assumption that changes in insur- ance status in the period from 2013 to 2014 were driven primarily by

changes in Medicaid eligibility and not by changes in the underlying de- mographic changes in the population of patients and visits. The benefit of including a ‘control’ state, in this case Florida, is that this risk is miti- gated. We are also unable to provide a ‘washout period’ as is sometimes

Table 5

Admission rate and advanced cardiac imaging among chest pain visits.

Admission rate Advanced cardiac imaging

% aOR 95% CI % aOR 95% CI

		Medicaid expansion	5.1%	1.294	1.144-1.464	10.0%	1.099	0.983-1.229
Table 4		estimator
Admission rate among all ED visits.		Primary payer	8.9%	1.220	1.171-1.271	7.9%	0.815	0.772-0.219
		Medicaid
	All ED visits	After expansion	7.9%	0.826	0.791-0.862	8.0%	0.956	0.909-1.005
	Admission rate (%) aOR 95% CI	indicator

N 8,157,748

				Expansion state	6.0%	1.166	1.102-1.235	9.9%	1.076	0.998-1.159
Medicaid expansion estimator	17.7%	1.070	1.051-1.089	indicator
Primary payer Medicaid	20.6%	2.360	2.343-2.378	Female	8.2%	0.815	0.794-0.835	8.0%	1.017	0.990-1.043
After expansion indicator	16.5%	0.971	0.963-0.979	Age
Expansion state indicator	16.2%	1.050	1.039-1.061	27-35	3.3%	Omitted	–	3.3%	Omitted	–
Female	16.6%	0.894	0.890-0.897	36-45	7.5%	2.457	2.347-2.573	8.1%	2.515	2.398-2.636
Age				46-55	12.0%	4.143	3.965-4.328	10.2%	3.267	3.120-3.421
27-35	13.2%	Omitted	–	56-64	14.3%	5.053	4.822-5.295	10.8%	3.489	3.312-3.669
36-45	13.8%	1.076	1.070-1.081	Race
46-55	20.0%	1.727	1.719-1.739	White	8.4%	1.020	0.948-1.097	8.2%	0.919	0.851-0.992
56-64	29.0%	2.664	2.649-2.679	Black	10.2%	1.054	0.978-1.137	6.8%	0.986	0.909-1.069
Race				Hispanic	9.2%	1.044	0.967-1.127	9.0%	1.157	1.068-1.253
White	16.5%	0.820	0.811-0.828	Asian	9.1%	1.156	0.978-1.367	13.0%	1.486	1.273-1.735
Black	15.3%	0.721	0.713-0.729	Other	7.9%	Omitted	–	8.2%	1.486	–
Hispanic	18.4%	0.905	0.895-0.914	Income of
Asian	26.0%	1.411	1.380-1.441	patient’s ZIP
Other	19.5%	Omitted	–	code
Income of patient’s ZIP code				Quartile 1	9.5%	Omitted	–	8.4%	Omitted	–
Quartile 1	16.7%	Omitted	–	Quartile 2	8.5%	0.907	0.881-0.933	7.4%	0.923	0.895-0.953
Quartile 2	16.3%	1.011	1.007-1.016	Quartile 3	8.2%	0.860	0.827-0.894	8.2%	1.060	1.017-1.104
Quartile 3	17.2%	1.108	1.102-1.115	Quartile 4	8.0%	0.865	0.805-0.929	8.7%	1.124	1.044-1.209
Quartile 4	18.9%	1.283	1.271-1.300	N	331,422			308,890

This table depicts adjusted odds ratios from a difference-in-differences model and logistic regression for likelihood of ED visit resulting in admission, using data before and after Medicaid expansion in 2013 and 2014 and from each of five states (Arizona, Iowa, Ken- tucky, Wisconsin [expanders], and Florida [non-expander]). The sample is limited to ED visits with primary payer Medicaid and self-pay.

This is the main statistical estimate of interest in the analysis.

This table depicts adjusted odds ratios from a difference-in-differences model and logistic regression for likelihood of ED visit resulting in admission or in advanced cardiac testing, using data before and after Medicaid expansion in 2013 and 2014 and from each of five states (Arizona, Iowa, Kentucky, Wisconsin [expanders], and Florida [non-expander]). The sample is limited to ED visits with primary payer Medicaid and self-pay with a CCS code for chest pain.

The is the main statistical outcome of interest in the analysis.

included in difference-in-differences analyses, where data points occur- ring near the time of Policy change are excluded to allow time for the new policy to take effect. This is because the HCUP data do not reliably include granular data on month that a given ED visit took place. Another limitation is that our analysis does not in any way address the private in- surance market or health insurance exchange market. We limit our study to Medicaid and self-pay, but this allows the possibility that some individuals who appear in the 2013 data as self-pay may have be- come commercially insured in 2014. Our analysis does not speak to the impact of any policy change on this group. The more important under- lying limitation in this work is that we cannot comment on how out- comes after these ED visits change. For the test case of chest pain, it may be that the previously uninsured cohort now enrolled in Medicaid benefits greatly from more risk stratification for ACS and work up and management for CAD, and future work may pursue this.

1. Conclusion

In this analysis of data from five states, Medicaid eligibility expan- sion after passage of the ACA was associated with increases in hospital-based acute care service utilization, namely admission rates. This finding was consistent with respect to chest pain, a common condi- tion known to have substantial discretionary utilization. This pattern may represent improvements in access to care that result from insur- ance expansion, financially motivated utilization by providers or a tran- sient increase in the utilization of hospital services among patients who are newly insured and accessing care. Policymakers should consider these effects with respect to quality improvement initiatives aimed at outcomes such as hospital admission that may be sensitive to insurance driven effects.

Author contributions

PDL, SD, ATJ conceptualized the study and developed the methodol- ogy. ATJ performed the analysis and drafted the manuscript. PDL, SD, ATJ, AKV contributed to the revision and editing of the manuscript.

Acknowledgements

This work was supported by the Emergency Medicine Foundation Medical Student Research grant.

References

1. Oyez. National Federation of Independent Business v. Sebelius. https://www.oyez. org/cases/2011/11-393. (Accessed on September 20, 2018).
2. Henry J Kaiser Family Foundation. Status of state Medicaid expansion decisions: In- teractive map. https://www.kff.org/medicaid/issue-brief/status-of-state-medicaid- expansion-decisions-interactive-map.
3. Price CC, Eibner C. For states that opt out of Medicaid expansion: 3.6 million fewer insured and $8.4 billion less in federal payments. Health Aff 2013;32(6):1030-6.
4. Nikpay S, Buchmueller T, Levy HG. Affordable care act Medicaid expansion reduced uninsured hospital stays in 2014. Health Aff 2016;35:106-10.
5. Brook RH, Ware JE, Rogers WH, Keeler EB, Davies AR, Sherbourne CA, et al. The effect of coinsurance on the health of adults. Results from the RAND health insurance ex- periment. Report R-3055-HHSSanta Monica, CA: RAND Corporation; 1984 [ISBN 0- 8330-0614-2].
6. Taubman S, Allen H, Wright B, Baicker K, Finkelstein A, the Oregon Health Study Group. Medicaid increases emergency department use: evidence from Oregon’s health insurance experiment. Science 2014;343(6168):263-8.
7. Baicker K, Taubman S, Allen H, Bernstein M, Gruber J, the Oregon Health Study Group. The Oregon experiment-effects of Medicaid on clinical outcomes. NEJM 2013;368(18):1713-22.
8. Sommers BD, Gunja MZ, Finegold K, Musco T. Changes in self-reported insurance coverage, access to care, and health under the affordable care act. JAMA 2015;314 (4):366-74.
9. Bittoni MA, Wexler R, Spees CK, Clinton SK, Taylor CA. Lack of private health insur- ance is associated with higher mortality from cancer and other chronic diseases, poor diet quality, and inflammatory biomarkers in the United States. Prev Med 2015;81:420-6.
10. LaPar DJ, Bhamidipati CM, Mery CM, Stukenborg GJ, Jones DR, Schirmer BD, et al. Pri- mary payer status affects mortality for major surgical operations. Ann Surg 2010;252 (3):544-50.
11. Chikani V, Brophy M, Vossbrink A, Hussaini K, Salvino C, Skubic J, et al. Association of insurance status with health outcomes following traumatic injury: statewide multi- center analysis. West J Emerg Med 2015;16(3):408-13.
12. Janke AT, Brody AM, Overbeek DL, Bedford JD, Welch RD, Levy PD. Access to care is- sues and the role of EDs in the wake of the Affordable Care Act. Am J Emerg Med 2015;33(2):181-5.
13. Ruger JP, Richter CJ, Lewis LM. Association between insurance status and admission rate for patients evaluated in the emergency department. Acad Emerg Med 2003;10 (11):1285-8.
14. Wilson S, Sharmistha D, Mahan M, Malhotra M, Miller J. Identifying disparity in emergency department length of stay and admission likelihood. World J Emerg Med 2016;7(2):111-6.
15. Pines JM, Socchi M, Moghtaderi A, Black D, Farmer SA, Hufstetler G, et al. Medicaid expansion in 2014 did not increase emergency department use but did change in- surance payer mix. Health Aff 2016;35(8).
16. Nikpay S, Freedman S, Levy H, Buchmueller T. Effect of the affordable care act Med- icaid expansion on emergency department visits: evidence from state-level emer- gency department databases. Ann Emerg Med 2017;70(2):2015-225e6.
17. Friedman AB. The uncertain economics of insurance enabling more emergency de- partment visits. Ann Emerg Med 2017;70(2):226-8.
18. Singer AJ, Thode HC, Pines JM. US emergency department visits and hospital dis- charges among uninsured patients before and after implementation of the afford- able care act. JAMA Netw Open 2019;2(4):e192662.
19. Venktatesh AK, Goodrich K. Emergency care and the national quality strategy: high- lights from the centers for Medicare & Medicaid services. Ann Emerg Med 2015;65 (4):396-9.
20. Venkatesh AK, Goodrich K, Conway PH. Opportunities for quality measurement to improve the value of care for patients with multiple chronic conditions. Ann Intern Med 2014;161:S76-80.
21. Venkatesh AK, Slesinger T, Whittle J, Osborn T, Aaronson E, Rothenberg C, et al. Pre- liminary performance on the new CMS sepsis-1 national quality measure: early in- sights from the emergency quality network (E-QUAL). Ann Emerg Med 2018;17 (1):10-15e1.
22. Sabbatini AK, Brahmajee K, Kocher KE. Reducing variation in hospital admissions from the emergency department for low-mortality conditions may produce savings. Health Aff 2014;33(9):1655-63.
23. Sandhu AT, Heidenreich PA, Bhattacharya J, Bundorf MK. Cardiovascular testing and clinical outcomes in emergency department patients with chest pain. JAMA Intern Med 2017;177(8):1175-82.
24. AHRQ. AHRQ healthcare cost and utilization project. Overview of the statewide emergency department database (SEDD). https://www.hcup-us.ahrq.gov/ seddoverview.jsp.
25. AHRQ. AHRQ healthcare cost and utilization project. Overview of the statewide inpa- tient database (SID). https://www.hcup-us.ahrq.gov/sidoverview.jsp.
26. Venkatesh AK, Dai Y, Ross JS, Schuur JD, Capp R, Krumholz HM. Variation in US hos- pital emergency department admission rates by clinical condition. Med Care 2014; 53(3):237-44.
27. Schuur JD, Venkatesh AK, Bohan JS. Variation in the use of Observation status evalu- ation in Massachusetts acute care hospitals, 2003-2006. Int J Emerg Med 2010;3(4): 367-72.
28. Dimick JB, Ryan AM. Methods for evaluating changes in health care policy: the difference-in-differences approach. JAMA 2014;312(22):2401-2.
29. Taubman SL, Allen HL, Wright BJ, Baicker K, Finkelstein AN. Medicaid increases emergency department use: evidence from Oregon’s health insurance experiment. Science 2014;343:263-8.
30. Finkelstein AN, Taubman SL, Allen HL, Wright BJ, Baicker K. Effect of Medicaid cover- age on ED use-further evidence from Oregon’s experiment. NEJM 2016;375:1505-7.
31. AHRQ. AHRQ healthcare cost and utilization project. Effect of Medicaid expansion on hospital use. http://www.hcup-us.ahrq.gov/faststats/statepayer/states.jsp.
32. Hunter AE, Spatz ES, Bernstein SL, Rosenthal MS. Factors influencing hospital admis- sion of non-critically ill patients presenting to the emergency department: a cross- sectional study. J Gen Intern Med 2016;31(1):37-44.
33. Foy AJ, Liu G, Davidson WR, Sciamanna C, Leslie DL. Comparative effectiveness of di- agnostic testing strategies in emergency department patients with chest pain: an analysis of downstream testing, interventions, and outcomes. JAMA Intern Med 2015;175(3):428-36.