California Emergency Department Visit Rates For Medical Conditions Increased While Visit Rates For Injuries Fell, 2005–11

Abstract

The emergency department (ED) is the source of most hospital admissions, cares for patients with no other point of access to the health care system, receives advanced care referrals from primary care physicians, and provides surveillance data on injuries, infectious disease, violence, and adverse drug events. Understanding the changes in the profile of disease in the ED can inform emergency services administration and planning as well as provide insight into the public’s health. We analyzed the trends in the diagnoses seen in California EDs from 2005–11, finding that while the ED visit rate for injuries decreased by 0.7 percent, the rate of ED visits for non-injury diagnoses rose 13.4 percent. We also found a rise in symptom-related diagnoses, such as abdominal pain, along with nervous system disorders, gastrointestinal disease, and mental illness. These trends point out the increasing importance of EDs in providing care for complex medical cases, as well as the changing nature of illness in the population needing immediate medical attention.

The earliest emergency departments (EDs) were called “accident rooms,” designed to accept patients with traumatic injuries, as well as provide charity care. [1] Although this initial charge to treat injury remains important today, the ED has evolved into a site of comprehensive access to all types of care. Staffed by full-time, often emergency medicine board-certified physicians, and equipped with advanced technology and equipment for acute care, EDs now have the ability to treat a wider array of patients. Furthermore, the passage of the Emergency Medical Treatment and Active Labor Act (EMTALA) in 1986 transformed the ED into a twenty-four-hour source of emergency care for all patients regardless of their ability to pay.

Approximately half of all hospital admissions now come from the ED, and primary care physicians increasingly refer urgent patients to the ED for follow-up care. [2] Visits to the ED rose 47 percent over the past two decades, [3,4] even as the reasons for the visits have become more severe and complex. [5] As an increasingly central part of the US health care system, EDs are used as sentinel surveillance sites for population data on injuries, violence, adverse drug events, and infectious disease. [6–11] Furthermore, as a safety-net provider seeing patients twenty-four hours a day, seven days a week, the ED can serve as a window into the health of vulnerable populations with limited health care access. [3,12]

From this central viewpoint, the distribution of reasons patients visit the ED could serve as a barometer of the health status of and patterns of care in the surrounding community. [9] Additionally, shifts in the breakdown of patient diagnoses in the ED are essential information for emergency medicine physicians, administrators, and policy makers as they staff, design, supply, and fund EDs and hospitals. While previous reports have summarized the cross-sectional distribution of diagnoses within a sample of ED visits, [13] no previous work has examined the recent trends in the distribution of diagnoses within a comprehensive set of ED visits. Therefore, we sought to determine how the diagnoses of patients visiting the ED have changed over recent years in California, both overall and within specific insurance groups.

Study Data And Methods

Design And Data Sources

We conducted a retrospective analysis of California ED visits from 2005–11 using ED visit data that all nonfederal California hospitals report to the Office of Statewide Health Planning and Development (OSHPD).[14] Specifically, we used OSHPD’s emergency discharge data, which captures all outpatient discharges from California EDs, and patient discharge data, which captures all patients seen in the ED and then admitted as an inpatient. Together, these data sources constitute a comprehensive record of ED visits to nonfederal general, acute care hospitals in California. Each visit record captures a range of patient clinical and demographic characteristics, including primary payer (insurance), age, and discharge diagnosis.

This analysis used the nonpublic emergency and patient discharge data files from 2005, the first year the emergency discharge data were available, through 2011, the most recent year available at the time of analysis. We excluded all visits with missing insurance and missing sex information, patient discharge data visits where the source of admission was not the hospital’s ED, and all admissions that were not classified as unscheduled. Our sample selection process is detailed in the online Appendix Exhibit A1. [15]

When calculating overall visit rates, we used annual population data from the Census Bureau’s American Community Survey. For rates of ED use within individual insurance groups, we used California population data stratified by insurance coverage from the Centers for Medicare and Medicaid Services, the Research and Analytic Studies Division of the California Department of Health Care Services, and the Census Bureau’s Health Insurance Historical Tables, derived from the Current Population Survey’s Annual Social and Economic Supplements. [16–18]

Statistical Analysis

To categorize discharge diagnoses into wider and more clinically meaningful categories, we first converted discharge diagnoses from the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes to multilevel diagnosis codes using the Clinical Classifications Software (CCS) [19] developed by the Agency for Healthcare Research and Quality. We further grouped these multilevel diagnosis codes into thirty-eight primary discharge diagnosis categories as specified by Gelareh Z. Gabayan and colleagues. [20] Additionally, we investigated the trends over time in twenty-nine comorbid conditions among admitted patients from the ED using Elixhauser comorbidity measures.

We used graphical and tabular analysis to describe the trends in diagnosis groups during the study period. We presented both absolute number of visits and visit rates per 1,000 residents to account for population growth. We then performed nonparametric testing to assess the statistical significance of the trends in diagnoses across years. We used SAS 9.2 software for all analyses. The University of California San Francisco Committee on Human Research and the Stanford Institutional Review Board approved this study.

Limitations

Our findings must be interpreted in light of several limitations. First, hospitals self-report their administrative data to OSHPD, introducing potential for reporting errors and missing data. However, OSHPD performs routine accuracy checks on all submitted data, reducing the likelihood of significant reporting errors. In addition, given that there had not been significant changes to the reporting surveys or systems during our study period, we do not expect differential misreporting or misclassification to occur over time to bias these results.

Second, our rates were calculated using population denominators from survey data capturing all applicable insurance statuses of each respondent over a year-long window. This methodology is significantly different from that used to collect the OSHPD data, which records a patient’s insurance status at the time of treatment. The different methodologies do introduce the possibility of measurement bias that could affect our insurance-specific rates of ED visits. However, the survey data provides the most accurate data on health insurance enrollment available in the time periods and subgroups necessary for our analysis. Furthermore, neither the survey nor the OSHPD collection methods changed significantly during the study period, so we again do not expect the methods to differentially bias our results over time.

Third, there is the possibility that our findings regarding trends in individual diagnosis categories, especially the decline in “other residual codes,” could reflect increasing financial pressures on the part of hospitals and physicians to capture and code more specific or acute diagnoses that are reimbursed at higher rates. [21]

Finally, our results only apply to California, a large, diverse state representing 12 percent of the US population with unique comprehensive, statewide ED data. Although a useful case study of changes in ED use, these results cannot be generalized to the rest of the nation.

Study Results

The descriptive characteristics of all visits in our study sample are shown in Appendix Exhibit A2. [15] Briefly, there has been a slight increase in the age of the overall population of visitors to the ED, a notable increase in the proportion of minorities, and a rise in the proportion of ED visitors covered by public insurance. Visitors to the ED also have an increasing number of comorbidities (of twenty-nine Elixhauser comorbidities, all but two have seen a statistically significant increase over time), and the overall case-mix index has risen as well (p < 0.0001), indicating that the patients seen in the ED are more medically complex and seriously ill. The total number of ED visits in California grew from 10.1 million in 2005 to 11.9 million in 2011, representing a rise from 287 to 317 ED visits per 1,000 residents (10.3 percent). This growth, however, was not equally distributed across diagnosis groups.

We noted a particular discrepancy between ED visits for injuries and visits for all other diagnoses (Exhibit 1). The rate of visits for injuries did not change significantly over the time period of our study, decreasing slightly from 61.5 visits per 1,000 Californians in 2005 to 61.0 visits per 1,000 in 2011 (0.7 percent). The rate of visits with a non-injury diagnosis, on the other hand, rose statistically significantly from 225.3 to 255.5 visits per 1,000 residents (13.4 percent). As a result, injuries actually declined as a proportion of visits, from 21.4 percent in 2005 to 19.3 percent in 2011.

Exhibit 1: Figure.

ED Visits For Injury Versus Non-injury Diagnoses Per 1,000 California Residents, 2005–11

Sources: SOURCES Authors’ analysis of emergency department discharge data and patient discharge data from the California Office of Statewide Health Planning and Development as well as population data from the Census Bureau’s American Community Survey.

Within the non-injury diagnoses, we observed the largest absolute growth in the number of visits per 1,000 Californians among gastrointestinal (GI) system diseases (20.1 to 23.6 visits per 1,000), symptoms of abdominal pain (11.2 to 14.5 visits per 1,000), and nervous system disorders (14.7 to 18.1 visits per 1,000). The fastest-growing non-injury diagnoses are presented in Exhibit 2. The largest percent increases in visit rates were noted in renal disease (49 percent), diseases of the blood and blood forming organs (43.8 percent), abdominal pain (30 percent), other undefined symptoms (27 percent), and hypertension (27 percent). Full data on changes in rates of ED use by diagnosis are presented in Appendix Exhibit A3. [15]

Exhibit 2: Figure.

Absolute Change In Number Of Emergency Department (ED) Visits Per 1,000 California Residents, 2005–11, By Non-injury Diagnosis Group

Sources/Notes: SOURCES Authors’ analysis of emergency department discharge data and patient discharge data from the California Office of Statewide Health Planning and Development as well as population data from the Census Bureau’s American Community Survey.

When we further stratified the visits by payer, the rate of ED visits for injuries decreased among the uninsured and rose more slowly than non-injury diagnoses among Medicaid beneficiaries and the privately insured. Medicare beneficiaries were the only group that exhibited a similar percent growth for injury and non-injury diagnoses (Exhibit 3).

Exhibit 3: Figure.

Percent Change From 2005 To 2011 In Number Of Visits For Injury Versus Non-injury Diagnoses Per 1,000 Californians, By Payer

Sources: SOURCES Authors’ analysis of emergency department discharge data and patient discharge data from the California Office of Statewide Health Planning and Development as well as population data from the Centers for Medicare and Medicaid Services, the Research and Analytic Studies Division of the California Department of Health Care Services, and the Census Bureau’s Health Insurance Historical Tables.

We found slightly differing granular trends in ED visit rates by diagnosis group when stratified by payer (see Appendix Exhibit A4). [15] For instance, nervous system disorders, GI system diseases, and mental illness were in the top-five fastest-growing diagnoses of multiple insurance groups. More specifically, the rate of ED use for mental illness rose faster than any other diagnosis group specifically among patients who were uninsured. “Other injuries” was one of the fastest-growing diagnoses for people with Medicare but not for those with other insurance types, and the rate of ED visits for upper respiratory infections was only in the top-five fastest-growing diagnoses among Medicaid patients.

Finally, when stratified by age, injury diagnoses were in the top-five diagnoses among the older population (45–64 and sixty-five and older), as opposed to patients in the younger age groups (5–19 and 20–44), whose fastest-growing diagnoses were non-injury related. Exhibit A5 in the Appendix provides a detailed table of the top-five diagnoses with the largest absolute changes in rate of ED visits stratified by the following age categories: < 5, 5–19, 20–44, 45–64, and sixty-five and older. [15]

Discussion

We found that from 2005–11, the rate of ED visits in California for injuries decreased by 0.7 percent, while visit rates for all other non-injury diagnoses rose by 13.4 percent. These results indicate that the growth of ED visits in California was not driven by injuries but rather by nontrauma-related, mainly medical diagnoses. In addition, we found that EDs are caring for a more complex patient cohort with more chronic comorbidities and a higher case-mix index.

The overall trend away from ED care for injuries mirrors recent changes in the US disease burden. The mortality rate from injuries has remained relatively stable from 2005–11, [22] and the number of adults with multiple chronic diseases has grown. [23] In fact, one of every two American adults has heart disease, COPD, hypertension, arthritis, diabetes, cancer, or asthma. [24] Although the fastest-growing diagnoses we documented did not reflect these chronic illnesses specifically, the increasing complexity of health management for Americans with multiple medical conditions is consistent with the pattern we observed in the ED.

When stratified by payer, we found that the rate of growth in ED visits for injuries kept up with non-injury diagnoses only among Medicare beneficiaries. This could reflect an increasing incidence of injury among the elderly as a result of a variety of factors, including increasingly active lifestyles[25] or an increasing severity of injury among the elderly requiring growing amounts of emergency care.

Finally, when looking at the diagnoses with the largest growth in rates of ED visits between 2005 and 2011, we found a few notable insurance-specific trends. For instance, mental illness was in the top-five diagnoses with the largest absolute changes in ED visit rates among the uninsured, Medicare, and the privately insured. Although this could reflect changes in the prevalence of mental health conditions, trends in the prevalence of psychological distress appear stable. [26,27] This trend could also reflect worse access to preventive nonemergent care for mental illness, especially because of issues of affordability and cost sharing, the primary barrier cited by patients and physicians to receiving needed mental health services. [28,29] Furthermore, especially among chronically uninsured patients with severe mental illness, worsening access to social services could perpetuate the need for ED care. [30]

As the rising number of ED visits and the stagnant supply of ED and critical care beds in California hospitals contribute to ED crowding, [31] it is worth noting that rising visit rates for trauma are not major contributors to this effect. Rather, medical conditions, including nervous system disorders, GI system diseases, nonspecific symptoms, and mental illness, are driving the rising rate of ED use. Whether this trend reflects changing disease incidence in the California population, changes in access to alternative or preventive treatment for these concerns, or numerous other potential explanations for the rise in demand, it certainly reflects the growing role the ED plays in the modern health care system in treating complex medical cases in addition to traumatic injuries. These findings suggest that human and capital resources may need to shift in ways that more accurately reflect the demands placed on EDs.

For example, the large growth in mental health conditions presenting to the ED could require an increased presence of mental health professionals and services in the emergency department. In addition, the rise in injuries in older adults, particularly among adults between forty-five and sixty-four years old, may point to a need for trauma services that are more finely attuned to the needs of an aging population. Current efforts, such as creating different triage criteria based on age in the prehospital setting and embedding geriatricians within the trauma service in the postacute care setting, are examples of initiatives that may improve outcomes and require further study.

Conclusion

We found that the rate of ED use in California for non-injury, mainly medical diagnoses rose 13.4 percent from 2005–11, while the rate of ED use for injuries decreased by 0.7 percent. Thus, California EDs are providing increasing amounts of care for complex emergencies related to chronic conditions, infections, and even nonspecific symptoms. This trend reflects both changes in the population disease burden and the ED’s more central role in health care compared to its original charge to treat injured patients and provide charity care.