US Emergency Department Visits and Hospital Discharges Among Uninsured Patients Before and After Implementation of the Affordable Care Act

IMPORTANCE The US Patient Protection and Affordable Care Act of 2010 (ACA) was enacted in 2010 with several provisions that targeted reducing numbers of uninsured Americans. OBJECTIVE To assess the numbers and proportion of emergency department (ED) visits (2006-2016) and hospital discharges (2006-2016) by uninsured patients, focusing on the 2014 ACA insurance reforms (Medicaid expansion, individual mandate, and private insurance exchanges). DESIGN, SETTING, AND PARTICIPANTS Cross-sectional study of visitors to US EDs and patients discharged from US hospitals using National Hospital Ambulatory Care Survey data and Healthcare Cost and Utilization Project data, respectively, from 2006 to 2016. Data analysis took place in February 2019. MAIN OUTCOMES AND MEASURES Numbers and proportions of total and uninsured ED visits and hospital discharges. Simple descriptive statistics and interrupted time-series analysis were used to assess changes in uninsured visits over time and after the implementation of insurance provisions in 2014.

95% CI, −4.3 to −1.8 percentage points; P = .003). The proportion of hospital discharges by uninsured patients remained steady at approximately 6% from 2006 to 2013, then declined to 5% in 2014 and 4% in 2016. Similar changes were seen for patients aged 18 to 64 years, with a decrease in hospital discharges from 10% to 7% over the study period.

Introduction
One of the main goals of the Patient Protection and Affordable Care Act of 2010 (ACA) was to reduce the number of people without health insurance in the United States while also improving health care quality and reducing cost. 1 Whether these changes have achieved their goals remains to be seen.
Several ACA provisions targeted health insurance: (1) expanding Medicaid eligibility, which later became optional for states (2014); (2) the creation of health insurance exchanges (2014); (3) regulations on health plans that allowed young adults to remain on their parents' insurance until age 26 years and disallowed excluding or charging higher rates for patients with preexisting conditions (2010); (4) requirements that most individuals have health insurance (implemented 2014 but repealed in 2018); and (5) penalties to employers for not offering coverage for employees (2016). In addition, there were other ways the ACA might affect coverage, such as limits on experience rating and subsidies. To our knowledge, no reports have presented nationally representative data describing the longitudinal associations of the ACA with both ED visits and hospital discharges together.
We examined trends in ED visits and hospitalizations over the decade from 2006 to 2016, including visit rates and changes in insurance coverage, with a focus on visits after 2013, when the ACA insurance provisions of Medicaid eligibility expansion, health insurance exchanges, and the individual mandate 2 went into effect. Specifically, we aimed to answer the following question: was ACA expansion in 2014 associated with significant changes in the number of ED visits and hospitalizations in the US overall and by uninsured individuals in particular?

Study Design and Setting
We conducted a retrospective secondary analysis of data from the National Hospital Ambulatory

National Hospital Ambulatory Care Survey
Data from the NHAMCS include a nationally representative sample of visits to hospital-based EDs. sample design, data collection procedures, field quality control, data processing, and estimation procedures, is available on the National Center for Health Statistics website. 3

Healthcare Cost and Utilization Project
The data from the HCUP National Inpatient Sample were extracted from HCUPnet, the online tool available on the website of the Agency for Healthcare Research and Quality. The National Inpatient Sample has information on all hospital discharges, including data on demographic characteristics (age, sex, payer) as well as diagnosis and other variables from US community hospitals as a 20% stratified sample from the Agency for Healthcare Research and Quality's State Inpatient Databases.

Outcomes
Primary outcomes of interest were the number of ED visits, hospital admissions from the ED, and hospital discharges over the study period. Each outcome was examined by insurance type (uninsured, Medicaid, private insurance, and Medicare) both in terms of total numbers and relative payer mix expressed as a percentage of all patients. A planned subgroup analysis included a subset of patients aged 18 to 64 years, as this group has less access to government-sponsored insurance.

Statistical Analysis
Binary data are presented as numbers and frequencies. Interrupted time series using Joinpoint 4 software was used to determine whether the proportions of ED visits and discharges changed, especially in the uninsured category, from 2014 to 2016 compared with those in earlier years.
Joinpoint models are linear regression models where lines with different slopes are connected together at join points, which are points where changes can be detected. 5 Joinpoint software takes trend data and fits the simplest model that the data allow. The program was started with the minimum number of specified join points equal to 0 (ie, a straight line) and tested whether more join points must be added to the model to identify changes in trend that were statistically significant. The tests of significance are determined using a Monte Carlo permutation method, with 5000 replications with unautocorrelated errors, to identify models that are statistically different from a single linear regression line. 5 The level of significance was set at a 2-sided P value of .05.  (Table).    Figure 1B). There was no clear decrease in the number of hospital discharges after 2014. However, when examining changes in payer mix, we observed similar changes as were seen with ED visits after the ACA insurance expansions, specifically by insurance type (Figure 4).

Discussion
The ACA has several provisions focused on improving health insurance coverage and reducing the number of uninsured Americans. Some of these provisions rolled out early in the ACA (ie, allowing people to stay on their parents' insurance until age 26 years). However, more dramatic changes in insurance coverage were implemented in January 2014: the expansion of Medicaid in selected states, establishment of health insurance exchanges intended to create private (non-employer-based) markets for coverage, and the individual mandate. These changes resulted in extended coverage to  20 million previously uninsured individuals and a 21% increase in Medicaid enrollment. 6 To date, 32 states and the District of Columbia have elected to expand coverage.
Our study is one of the few that have simultaneously examined ED visits and all hospital discharges. We demonstrate clear associations between ACA legislation and changes in ED visits and hospital discharges, as well as other broad trends in US health care policy. Visits to the ED increased rapidly until 2010, but then grew more slowly after that increasing rates of insurance coverage suggests that ED visits may be less preventable than inpatient hospitalizations, which may be more discretionary.
After 2014, there was a significant change in the payer mix for ED visits and hospital discharges, with a proportional shift from uninsured individuals to those receiving Medicaid, as well as declines in private insurance. These changes were accentuated among adult patients aged 18 to 64 years, who are at the highest risk for being uninsured because many are not eligible for state-level coverage or Medicare. Findings were similar for hospitalizations, which showed similar reductions in uninsured visits. Prior studies [7][8][9][10][11][12][13][14] have also demonstrated that Medicaid expansion in particular is associated with changes in insurance mix in EDs and hospitals; however, association between expansion and the numbers of people using the ED has not been demonstrated. Future work is needed in this area to further dissect the effects of these policies at the national level, on individual states, in specific populations of patients, and with respect to health outcomes. Additional work is needed to examine how these trends have affected out-of-pocket costs for patients and overall costs of care. Longerterm studies will also be helpful to determine the continued or delayed impact of insurance changes.  At the end of the study period (2016), we found that, despite decreasing rates of uninsured ED visits and hospitalizations, nearly 1 in 10 ED visits and 1 in 20 hospitalizations were still uninsured.
This represents an important gap that policy makers should continue to address, as lack of insurance coverage is associated with worse health outcomes, and the United States is one of the only developed countries that does not guarantee health insurance coverage to all citizens. 15

Limitations
There are several limitations to this study. The first limitation is the descriptive nature of our work, which does not explicitly isolate the effects of the insurance expansions (ie, separating states by Medicaid expansion), but rather examines trends as a whole. It is possible that factors other than insurance expansions or the ACA may account for some of the trends reported. Another major limitation is the retrospective nature of the study and our reliance on publicly available hospitalbased databases. These are limited because of the inability to account for repeat visits, or how visits evolved in other parts of the health care system, such as in outpatient clinics. In addition, because the databases contain a limited set of variables, we could not adjust for many confounding variables that may have affected ED visits and hospital discharges, such as changes in treatment patterns, reasons for visit, or other health policies that may have occurred concurrently. Also, our study only included 3 years after Medicaid expansions. Therefore, we cannot estimate or comment on more current and future trends.

Conclusions
Based on nationally representative administrative databases, the percentage of ED visits and hospital discharges by uninsured patients decreased considerably after the ACA insurance expansions of 2014. This was balanced by increases in ED visits and hospital discharges by patients with Medicaid.
Furthermore, these shifts in payer mix were especially prominent among individuals aged 18 to 64 years. Overall, ED visits did not appear to be associated with ACA expansions.