Frequent Emergency Department Users: A Statewide Comparison Before And After Affordable Care Act Implementation

Frequent users of hospital emergency departments (EDs) have long been a focus of cost-reduction initiatives implemented in both the public and the private sectors.^1,2 The existing literature describing frequent ED users finds that they are typically insured (with overrepresentation among those with public insurance) and are also high users of outpatient services, primary care, and other ambulatory care.³ However, previous work that examined interventions to curtail frequent ED use is either outdated or limited in scope, which suggests that strategies tailored to more specific patient populations and conditions are needed.^4,5

Most studies of frequent ED use were conducted before 2014, when several major coverage expansion provisions of the Affordable Care Act (ACA) were implemented. Twenty million adults gained health insurance coverage under the ACA,⁶ primarily through expansion of eligibility for Medicaid and a combination of new insurance Marketplaces and subsidies. Beyond expanding health coverage, the ACA also included provisions to improve access to health care services by requiring health plans to cover certain so-called essential benefits and by supporting initiatives to improve the coordination of care, particularly for high-need patients. Studies of frequent ED use after these ACA expansions and provisions are necessary to help state and federal efforts address frequent ED use, particularly among people served by public insurance programs.^7,8

It is not clear a priori how ACA coverage expansions might affect frequent ED use. On the one hand, expansion of insurance coverage is associated with increased use of health care services.⁹ As a result, one might expect to find an increase in frequent ED use after the ACA if the newly covered population were similar to the insured population before the ACA, or if there were pent-up demand and limited access to non-ED outpatient care. On the other hand, if the population that was newly covered after the ACA was healthier or better able to access non-ED outpatient care, one might expect to find that frequent ED use declined, mimicking the patterns of overall ED use in prior insurance expansions.¹⁰ In addition, targeted investments directed toward high-need patients under the ACA could cause frequent ED use to decline.

We provide an important first look at changes in frequent ED use before and after ACA coverage expansions in California.

This study used data for the period 2012–15 to examine the likelihood of frequent ED use before and after the largest expansion in insurance coverage in the US in half a century. We describe how the profile of frequent users changed during this time period and provide an important first look at changes in frequent ED use before and after ACA coverage expansions in California—the largest and most diverse state in the country,^11,12 and a state that opted to expand Medicaid eligibility.

Study Data And Methods

Study Design And Data Sources

We performed a retrospective analysis of all visits to the ED (including outpatient visits and those resulting in admission to the hospital) using nonpublic data from the California Office of Statewide Health Planning and Development. All licensed hospitals in California except those that are federally operated are required to report inpatient and ED discharge data to the office.¹³ We used a unique patient identifier to collapse annual visit-level discharge data to the patient level. By linking multiple ED visits across patients in each year, we were able to conduct an analysis of ED users instead of relying solely on visit-level information. Without this linkage information, we would not have been able to identify which patients were frequent users. We pooled four years of information from annual, cross-sectional, patient-level data files and defined the pre-ACA period as the years 2012 and 2013 and the post-ACA period as 2014 and 2015. (Throughout this article, “pre-ACA” and “post-ACA” refer to the periods before and after coverage expansions took effect, not the act’s actual passage in 2010.) Overall, we successfully linked 88 percent of all ED visits using the patient identifier, and this linkage rate did not vary across the four years of data.

Outcome Variable: Definition Of ‘Frequent User’

While there is no agreed-upon standard for what constitutes a frequent user, the most commonly used definition is a person with four or more ED visits in a year.¹⁴ Our main analysis defined a frequent ED user as someone who had four or more ED visits in a single calendar year, and our sensitivity analysis used alternative definitions (someone with seven or more visits and someone with ten or more visits) to measure the robustness of our results.

Predictor Variables: Patient Characteristics

We restricted the study population to nonelderly adults ages 18–64, excluding patients who had any ED visits during the year with Medicare as the expected payer. We focused on nonelderly adults because they were most affected by the ACA coverage expansions. We excluded patients with Medicare visits because Medicare coverage for nonelderly adults is a proxy for being disabled, which could affect the frequency of ED use.

We examined patient demographic characteristics including age, sex, primary language, and race/ethnicity. To identify underlying health conditions, we extracted all available discharge abstracts for a patient during each study year, including diagnoses associated with inpatient visits. We examined all available diagnoses—both primary diagnoses and any other diagnoses recorded on the discharge abstract—to create a comprehensive diagnostic history.

We also included a measure of patients’ residence in a rural area based on patients’ ZIP codes and Rural-Urban Commuting Area Codes.¹⁵ For patients who changed ZIP codes during a year, we used the ZIP code or county where visits occurred most frequently (the modal value), and for patients with no ZIP code information available, we used the ZIP code of the hospital. We also constructed indicators of county of residence, again using modal values for people who changed counties and the hospital county for missing values.

Finally, we created mutually exclusive insurance categories to designate the primary coverage source. If patients had multiple coverage sources across ED visits, we assigned coverage based on the largest share of ED visits during the year. In the limited cases where patients had equal shares of ED visits across multiple payers, we used the hierarchy Medicaid, uninsured, and private to assign primary coverage source. For example, if a patient had two ED visits, one covered by Medicaid and the other by private insurance, Medicaid would be assigned as their primary coverage source.

Statistical Analysis

We first examined the share and demographic profiles (for example, race/ethnicity, sex, age, rurality of ZIP code, and principal language) of frequent ED users in our two periods of interest. We then estimated a series of logistic regression models to understand how the likelihood of being a frequent ED user may have changed between these time periods. The first model included only an indicator variable that signified the post-ACA period and patients’ demographic characteristics. Next, we added sets of patient-level controls that includied observed health conditions (model 2), primary coverage source (model 3), and indicator variables for county of residence (model 4) to understand how these added variables might moderate the observed effect of the time period. We included county fixed effects to control for underlying time-invariant differences across counties (such as environmental factors or health care resources) that could affect the likelihood of frequent ED use and the observed relationships between our patient-level controls and ED use. In addition, we clustered standard errors by county to account for correlation of unobservable factors that could affect ED use within counties.

In our final model we also included an interaction term between primary coverage source and the post-ACA indicator, and we calculated the expected odds of being a frequent ED user before and after ACA implementation across each coverage group. These predicted odds took into account differences in baseline odds of being a frequent user across coverage groups and also accounted for differences in the distribution of patient control variables across coverage groups.¹⁶

In addition, we ran separate models for the pre- and post-ACA time periods, as well as models stratified by insurance coverage, to better understand how the ACA coverage expansion and insurance coverage could change the observed effects of patient-level control variables on the likelihood of frequent ED use. We also ran a sensitivity analysis that defined insurance coverage categories based on the first ED visit in the year rather than the largest share of visits.

Statistical analyses were conducted using STATA, version 15. The University of California San Francisco Committee on Human Research approved this study.

Limitations

Our study had several limitations. First, the Office of Statewide Health Planning and Development collects self-reported administrative data from hospitals, which introduces the potential for reporting errors and missing data. An example of this is the determination of expected source of payment. Hospitals report to the office which insurance source paid or is expected to pay the greatest portion of the costs of a patient’s ED visit. Hospitals determine this based on various sources, including approved claims, historical information, and presentation of an insurance card. Concern about reporting error is somewhat mitigated by hospitals’ routine accuracy checks using the office’s Medical Information Reporting for California (MIRCal) online system.¹⁷

Second, our results apply only to California. Despite California’s diverse population, using state-specific data could limit the generalizability of our results—particularly given that California runs the largest Medicaid program in the country.¹⁸

Third, missing patient identifiers kept us from linking all ED visits at the patient level. Although we successfully matched 88 percent of all ED records for nonelderly adults, certain groups—including Latinos, the uninsured, and people with missing information about their residence—were less likely to have patient identifiers. While this is a relatively high linkage rate, other work has found that lack of patient identifiers is concentrated among traditionally vulnerable populations, which could have biased our results if these groups systematically differed in terms of factors that affect frequent ED use.¹⁹

Fourth, our data covered only two years after the coverage expansions took effect. Given the time it can take people to enroll in programs and access care, we did not expect to see the full effects of the ACA in only two years.

Fifth, the lack of information on supply-side changes (that is, the opening of new or the closing of existing EDs) is a potential limitation to our study, given that changes in supply certainly affect the demand for care.

Finally, we were not able to account for how different types of private insurance could differentially affect ED use. For example, high-deductible plans may provide large financial disincentives to ED use compared to Cadillac plans, which may have the opposite effect because of their low copayments.

Study Results

Our analytic data set contained information on about 13.7 million patients who made at least one ED visit to a California hospital in any year of our four-year study period, capturing information on 24.5 million ED visits. In the two years after the ACA coverage expansions, the number of ED users increased by 511,284 (7.7 percent) compared to the two years before the expansions, and the number of ED visits increased by 1,155,772 (9.9 percent) (exhibit 1). The number of frequent ED users increased by 81,800 in the post-ACA period. In the pre-ACA period, 7.9 percent of all ED patients made four or more visits in a single year; in the post-ACA period, 8.5 percent of total ED patientsdid so. This equates to a 7.6 percent increase in the share of frequent users from one period to the next.

Overall, the differences in the demographic characteristics of the total ED patient population and the frequent user population before and after the ACA (exhibit 2) were small, although nearly all were significant ($p<0.001$). After the ACA both patient groups were slightly younger and had higher shares of Hispanic patients. Not surprisingly, the largest change was in primary coverage source. Among the total ED patient population, the share of patients covered by Medicaid increased from 21.5 percent to 37.6 percent, while the share of uninsured patients declined from 26.9 percent to 13.2 percent. The changes in insurance coverage were even more dramatic among frequent ED users. In this group, the share of patients covered by Medicaid increased from 44.6 percent to 67.6 percent, while the share of uninsured patients declined from 29.0 percent to 8.3 percent.

We also found differences in the health profiles of ED patients across the two time periods (exhibit 2). In the post-ACA period greater shares of both patient groups had behavioral health needs, including mental health conditions and substance use disorders (both related and not related to alcohol use), and chronic conditions such as asthma and hypertension.

Online appendix exhibit A1²⁰ shows results from a series of multivariable logistic regression models in which we added covariates iteratively to understand the relationships between frequent use and the covariates before introducing interaction terms. Results from these iterative models indicated that insurance coverage and health conditions changed the direction of the association between the post-ACA period and frequent ED use.

Exhibit 3 presents results from our final model, which included an interaction term between primary coverage source and the post-ACA indicator, to allow the observed relationship between the post-ACA period and frequent ED use to differ across coverage groups. ED patients with Medicaid as their primary insurance source were the reference category and were significantly less likely to have frequent ED use in the post-ACA period (odds ratio: 0.88; $p<0.001$), compared to the pre-ACA period. Frequent ED use also declined for the uninsured post ACA (OR: 0.69; $p<0.001$), while ED patients with private coverage had a slight increase (OR: 1.05; $p<0.001$), although both of these groups were much less likely to be frequent users compared to Medicaid patients, regardless of time period. To account for potential confounding from an increased number of diagnoses for patients admitted to the hospital from the ED, we excluded diagnoses and inpatient admission and found similar results (appendix exhibit A2).²⁰

We calculated the predicted odds of being a frequent ED user based on primary coverage source in the pre- and post-ACA periods. This provided a way to compare changes in frequent ED use separately for each insurance coverage group, accounting for baseline differences in the odds of being a frequent user across coverage groups. The odds of being a frequent ED user declined by 12 percent for Medicaid patients (from 0.35 pre ACA to 0.31 post ACA) and by nearly 50 percent among the uninsured (from 0.13 to 0.07) (appendix exhibit A3).²⁰ Among people with private coverage, we observed a slight increase in the odds of frequent use after implementation of the ACA, although those with private coverage had much lower baseline odds of frequent use (appendix exhibit A4).²⁰

In addition to coverage source, several other patient-level factors were significantly associated with frequent ED use across the four-year study period. Compared to non-Hispanic whites, non-Hispanic blacks had significantly higher odds of being frequent ED users (OR: 1.27; $p<0.001$), while Asians were less likely to be frequent users (OR: 0.62; $p<0.001$) (exhibit 3). Female patients were also more likely to be frequent users (OR: 1.34; $p<0.001$) compared to male patients, as were patients ages 18–25 (OR: 2.25; $p<0.001$) compared to those ages 55–64. Diagnoses of a mental health condition (OR: 3.34; $p<0.001$) and a non-alcohol-related substance use condition (OR: 2.64; $p<0.001$) were also associated with more frequent ED use.

Next, we stratified our sample by the two time periods to examine whether the relationship between frequent ED use and patient-level characteristics other than insurance source might have changed before and after ACA implementation. We saw few differences in the size or direction of the effect of any of the patient-level factors on frequent ED use (exhibit 4). The largest change was the decline in the odds of being a frequent ED user among the uninsured in the post-ACA period, which was consistent with our main results. To further examine the possibility that patient-level factors other than insurance coverage drove our results, we stratified our sample by primary coverage source and ran the models separately for patients with Medicaid, those with private coverage, and the uninsured (appendix exhibit A5).²⁰ Across all insurance types, the same patient-level factors (diagnosis of a mental health condition or substance use disorder) were the largest predictors of frequent ED use.

Results from our sensitivity analysis using alternative definitions of frequent ED user (someone with seven or more or with ten or more ED visits in a single year) were similar to our main results (appendix exhibit A6).²⁰ We also found similar results when we defined insurance coverage categories based on a patient’s first ED visit in the year, not the primary coverage definition used in the main models (appendix exhibit A7).²⁰ Finally, we ran negative binomial count models using total ED visits as the outcome rather than a binary indicator of frequent ED use. Again, these results were consistent with our main results, indicating that the ED visit rate declined by 6 percent in the post-ACA period for Medicaid patients compared to the pre-ACA period (appendix exhibit A6).²⁰

Discussion

Our study contributes to the existing literature by examining changes in frequent ED use before and after the largest expansion of insurance coverage in the US in more than fifty years and examining frequent ED use across multiple types of insurance coverage. We found that the share of ED patients who were frequent users increased after ACA implementation. However, when we controlled for factors such as health status, primary insurance coverage source, and county of residence in our multivariate analyses, we found that ED patients actually had a lower likelihood of being frequent users after implementation. And while Medicaid was the primary payer for a much higher share of frequent ED users after ACA implementation (likely because many uninsured people in the pre-ACA period became Medicaid beneficiaries under the coverage expansion), once we accounted for observed differences in the patient mix—including having a diagnosis of a mental health condition or a non-alcohol-related substance use disorder—the odds of being a frequent user with Medicaid coverage were significantly lower post ACA than pre ACA.

California began expanding Medicaid in 2010 through the state’s temporary Low-Income Health Program, resulting in over 650,000 enrollees by 2013.¹¹ Approximately 2.7 million additional people qualified for Medicaid¹¹ in 2014 as the ACA coverage expansions took effect. Before these expansions, policy makers were unsure about how many people would enroll or, perhaps more importantly, what their health needs would be. California saw larger increases in Medicaid enrollment than expected under the ACA,²¹ and it was largely unknown how new enrollees—many of whom had been uninsured for years and had used the health care system without coverage—would access services after becoming insured. Given historical patterns of ED use among Medicaid beneficiaries and the fact that coverage reduces the out-of-pocket spending required to access care,²² there was concern that frequent ED use would rise.

Our results suggest that the opposite occurred. After we controlled for compositional changes in the ED population, we found that the likelihood of being a frequent ED user significantly decreased post ACA for people covered by Medicaid and the uninsured. While our findings do not provide evidence that the ACA caused these changes, they suggest that expanded Medicaid coverage might have allowed patients to access needed medical services outside of the ED. This might have been especially true among people with chronic conditions who used the ED frequently pre ACA but who became connected to a primary care provider as a result of the ACA Medicaid expansion via Medicaid managed care plans. The steeper post-ACA decline in frequent ED use among the uninsured signals that people who remained uninsured after ACA implementation might have been healthier and thus had less incentive to obtain insurance. We did not find declines in frequent ED use among ED patients with private coverage as their primary source (we actually found slight increases), which suggests that the ACA may have had a different impact on this group’s ED use. It is again important to note that we were not able to account for the various types of private coverage or how those may have changed post ACA.

Our analysis focused on the population of people who visit the ED in any given year, rather than the broader population of people who might have gained coverage under ACA expansions. While this limited the scope of our results, it also allowed us to focus on a population of particular importance: people with frequent ED use who have high disease burdens²³ and also appear to have more difficulty accessing care in other ambulatory care settings.²⁴

Consistent with the results of past research, our findings indicate that frequent ED users have serious health conditions,^23,25 which require care that may be unavailable or inaccessible in other parts of the health care delivery system, especially for Medicaid beneficiaries.²⁶ In our analysis, ED patients in the post-ACA period had higher shares of comorbidities than those in the pre-ACA period. This is unsurprising if people who were uninsured before ACA implementation had pent-up demands that could not be fully absorbed by the non-ED ambulatory care system. It could also be the case that patients with certain health conditions are more likely to be diagnosed when they have insurance coverage.

The implementation of the ACA—in particular, the expansion of Medicaid to low-income single adults—underscores the success of coverage expansions in providing health insurance to many people who had been previously uninsured and had unmet health care needs. We found that nearly 70 percent of frequent ED users in California in the post-ACA period were covered by the state’s Medicaid program, compared to 45 percent in the pre-ACA period. Because nearly all nonelderly adults covered by the state’s Medicaid program are enrolled in managed care plans,²⁷ frequent ED users with Medicaid as their primary coverage source should be assigned to health plans that can manage a diverse set of health care needs and that have the financial incentive to do so. Given the large role that behavioral health conditions play in explaining frequent ED use, Medicaid managed care plans and county behavioral health agencies need to better coordinate physical and behavioral health services. Furthermore, to the extent that hospitals have seen a reduction in their bad debt and charity care expenditures in Medicaid expansion states,^28,29 hospitals might consider using some of these additional resources for these coordination efforts, which could ultimately reduce the number of frequent ED users.

Conclusion

We found that after changes in the health status and coverage sources of the ED patient population were accounted for, the likelihood of being a frequent ED user decreased in the two years following implementation of the major coverage provisions of the ACA. Still, with the sizable increase in ED patients covered by state Medicaid programs and higher baseline odds of frequent ED use among Medicaid patients relative to those with other coverage, there has been an overall increase in both the share and the absolute number of ED patients who are frequent users. Medicaid coverage and enrollment in managed care offer a vehicle to better serve high-need patients. However, the large growth in enrollment in such a short period of time also poses challenges for Medicaid managed care plans, which have absorbed large numbers of new beneficiaries, as well as for state budgets.

ACKNOWLEDGMENTS

The authors thank the California Office of Statewide Health Planning and Development for assistance in preparing the data sets used in this project.

NOTES

• 1 California Health Care Foundation. Frequent Users of Health Services Initiative [Internet]. Oakland (CA): CHCF; 2008 Oct 20 [cited 2018 Mar 2]. Available from: https://www.chcf.org/project/frequent-users-of-health-services-initiative/ Google Scholar
• 2 United Way of San Diego County. Project 25: “frequent user initiative” for chronically homeless [Internet]. San Diego (CA): United Way of San Diego County; 2010 Aug 18 [cited 2018 Mar 2]. Available from: https://uwsd.org/Frequent-User-Initiative Google Scholar
• 3 Zuckerman S, Shen YC. Characteristics of occasional and frequent emergency department users: do insurance coverage and access to care matter? Med Care. 2004;42(2):176–82. Crossref, Medline, Google Scholar
• 4 Soril LJJ, Leggett LE, Lorenzetti DL, Noseworthy TW, Clement FM. Reducing frequent visits to the emergency department: a systematic review of interventions. PLoS One. 2015;10(4):e0123660. Crossref, Medline, Google Scholar
• 5 Pines JM, Asplin BR, Kaji AH, Lowe RA, Magid DJ, Raven Met al. Frequent users of emergency department services: gaps in knowledge and a proposed research agenda. Acad Emerg Med. 2011;18(6):e64–9. Crossref, Medline, Google Scholar
• 6 Uberoi N, Finegold K, Gee E. Health insurance coverage and the Affordable Care Act, 2010–2016 [Internet]. Washington (DC): Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation; 2016 Mar 3 [cited 2018 Mar 3]. (ASPE Issue Brief). Available from: https://aspe.hhs.gov/system/files/pdf/187551/ACA2010-2016.pdf Google Scholar
• 7 California Department of Health Care Services. Health homes Program: health home for patients with complex needs (HHPCN) [Internet]. Sacramento (CA): The Department; [last modified 2018 Jan 22; cited 2018 Mar 2]. Available from: http://www.dhcs.ca.gov/services/Pages/HealthHomesProgram.aspx Google Scholar
• 8 California Department of Health Care Services. Whole Person Care Pilots [Internet]. Sacramento (CA): The Department; [last modified 2018 Feb 1; cited 2018 Mar 2]. Available from: http://www.dhcs.ca.gov/services/Pages/WholePersonCarePilots.aspx Google Scholar
• 9 Collins SR, Gunja M, Doty MM, Beutel S. Americans’ experiences with ACA Marketplace and Medicaid coverage: access to care and satisfaction: findings from the Commonwealth Fund Affordable Care Act Tracking Survey, February–April 2016 [Internet]. New York (NY): Commonwealth Fund; 2016 May [cited 2018 Mar 2]. Available from: http://www.commonwealthfund.org/~/media/files/publications/issue-brief/2016/may/1879_collins_americans_experience_aca_marketplace_feb_april_2016_tb.pdf Google Scholar
• 10 Miller S. The effect of insurance on emergency room visits: an analysis of the 2006 Massachusetts health reform. J Public Econ. 2012;96(11–12):893–908. Crossref, Google Scholar
• 11 Henry J. Kaiser Family Foundation. The California health care landscape [Internet]. Menlo Park (CA): KFF; 2015 Aug 26 [cited 2018 Mar 2]. Available from: https://www.kff.org/health-reform/fact-sheet/the-california-health-care-landscape/ Google Scholar
• 12 Weinberg M, Kallerman P (Bay Area Council Economic Institute, San Francisco, CA). A study of Affordable Care Act competitiveness in California [Internet]. Washington (DC): Center for Health Policy at Brookings; 2017 Feb [cited 2018 Mar 2]. (Field Research Report). Available from: https://www.brookings.edu/wp-content/uploads/2017/02/ca-aca-competitiveness.pdf Google Scholar
• 13 Office of Statewide Health Planning and Development. Hospital annual utilization data [Internet]. Sacramento (CA): OSHPD; [last updated 2018 Feb 28; cited 2018 Apr 9]. Available from: https://www.oshpd.ca.gov/HID/Hospital-Utilization.html Google Scholar
• 14 LaCalle E, Rabin E. Frequent users of emergency departments: the myths, the data, and the policy implications. Ann Emerg Med. 2010;56(1):42–8. Crossref, Medline, Google Scholar
• 15 Morrill R, Cromartie J, Hart G. Metropolitan, urban, and rural commuting areas: toward a better depiction of the United States settlement system. Urban Geogr. 1999;20(8):727–48. Crossref, Google Scholar
• 16 Buis ML. Stata tip 87: interpretation of interactions in nonlinear models. Stata J. 2010;10(2):305–8. Crossref, Google Scholar
• 17 Office of Statewide Health Planning and Development. MIRCal—California Inpatient Data Reporting Manual: 7th edition [Internet]. Sacramento (CA): OSHPD; [last updated 2018 Feb 21; cited 2018 Mar 3]. Available from: https://www.oshpd.ca.gov/HID/MIRCal/IPManual.html Google Scholar
• 18 Norris L. California and the ACA’s Medicaid expansion. Healthinsurance.org [serial on the Internet]. 2017 Jan 21 [cited 2018 Mar 3]. Available from: https://www.healthinsurance.org/california-medicaid/ Google Scholar
• 19 Zingmond DS, Ye Z, Ettner SL, Liu H. Linking hospital discharge and death records—accuracy and sources of bias. J Clin Epidemiol. 2004;57(1):21–9. Crossref, Medline, Google Scholar
• 20 To access the appendix, click on the Details tab of the article online.
• 21 Lucia L, Jacobs K, Watson G, Dietz M, Roby DH. Medi-Cal expansion under the Affordable Care Act: significant increase in coverage with minimal cost to the state [Internet]. Berkeley (CA): University of California Berkeley Center for Labor Research and Education; 2013 Jan [cited 2018 Mar 2]. Available from: http://laborcenter.berkeley.edu/pdf/2013/medi-cal_expansion13.pdf Google Scholar
• 22 Golberstein E, Gonzales G, Sommers BD. California’s early ACA expansion increased coverage and reduced out-of-pocket spending for the state’s low-income population. Health Aff (Millwood). 2015;34(10):1688–94. Go to the article, Google Scholar
• 23 Billings J, Raven MC. Dispelling an urban legend: frequent emergency department users have substantial burden of disease. Health Aff (Millwood). 2013;32(12):2099–108. Go to the article, Google Scholar
• 24 Kangovi S, Barg FK, Carter T, Long JA, Shannon R, Grande D. Understanding why patients of low socioeconomic status prefer hospitals over ambulatory care. Health Aff (Millwood). 2013;32(7):1196–203. Go to the article, Google Scholar
• 25 Doran KM, Raven MC, Rosenheck RA. What drives frequent emergency department use in an integrated health system? National data from the Veterans Health Administration. Ann Emerg Med. 2013;62(2):151–9. Crossref, Medline, Google Scholar
• 26 Coffman JM, Hulett D, Fix M, Bindman AB. Physician participation in Medi-Cal: ready for the enrollment boom? [Internet]. Oakland (CA): California HealthCare Foundation; 2014 Aug [cited 2018 Mar 2]. Available from: https://www.chcf.org/wp-content/uploads/2017/12/PDF-PhysicianParticipationMediCalEnrollmentBoom.pdf Google Scholar
• 27 McConville S. Just the facts: the Medi-Cal Program [Internet]. San Francisco (CA): Public Policy Institute of California; 2017 Apr [cited 2018 Mar 2]. Available from: http://www.ppic.org/publication/the-medi-cal-program/ Google Scholar
• 28 Dranove D, Garthwaite C, Ody C. Uncompensated care decreased at hospitals in Medicaid expansion states but not at hospitals in nonexpansion states. Health Aff (Millwood). 2016;35(8):1471–9. Go to the article, Google Scholar
• 29 Blavin F. Association between the 2014 Medicaid expansion and US hospital finances. JAMA. 2016;316(14):1475–83. Crossref, Medline, Google Scholar