Insurance Status and the Care of Adult Patients 19 to 64 Years of Age Visiting the Emergency Department

Transcription

1 ORIGINAL RESEARCH CONTRIBUTION Insurance Status and the Care of Adult Patients 19 to 64 Years of Age Visiting the Emergency Department Rebekah Mannix, MD, MPH, Anne M. Stack, MD*, and Vincent Chiang, MD* Abstract Objectives: The objective was to determine whether insurance status is associated with the care of patients presenting to the emergency department (ED). Methods: This was a retrospective cross-sectional analysis of ED visits using data from the National Hospital Ambulatory Medical Care Survey (NHAMCS; 1999 through 2008). Patients 19 to 64 years of age were categorized as having private insurance, public insurance, or no insurance as their primary insurance. Six components of ED care were assessed: wait time, left prior to discharge, use of diagnostic testing, treatment, instructions for follow-up, and whether the patient had been seen in the past 72 hours. Results: Nonprivate insurance status was associated with all six components of ED care, including higher proportions of leaving before discharge of patients with public insurance (4.1%, 95% confidence interval [CI] = 3.8% to 4.5%) versus patients with no insurance (4.7%, 95% CI = 4.2% to 5.1%) or private insurance (2.2%, 95% CI = 2.0% to 2.4%; p < 0.001). It was also associated with a higher proportion of return visits with 5.1% (95% CI = 4.6% to 5.6%) of patients with public insurance versus 4.7% (95% CI = 4.1% to 4.6%) of patients with no insurance versus 3.8% (95% CI = 3.5% to 4.2%) of patients with private insurance (p < 0.001). Patients with public or no insurance also had decreased odds of ED testing compared to those with private insurance (adjusted odds ratio [AOR] for public = 0.84, 95% CI = 0.80 to 0.88; and AOR for none = 0.82, 95% CI = 0.79 to 0.86). Conclusions: Nonprivate insurance status is associated with different care patterns in adults aged 19 to 64 years visiting the ED. Further studies are needed to evaluate how these disparate care patterns affect health outcomes. ACADEMIC EMERGENCY MEDICINE 2012; 19: ª 2012 by the Society for Academic Emergency Medicine In the past decade, U.S. emergency department (ED) utilization has increased from an estimated million visits in 1999 (about 37.8 visits per 100 persons) to 117 million ED visits in 2007 (approximately 39.4 visits per 100 persons). 1,2 The number of facilities qualifying as safety-net EDs also increased in this time period, from 1,770 in 2000 to 2,489 in While EDs are increasingly serving as safety nets for medically underserved patients, it is unclear if insurance status is associated with differences in patient care within the ED. To From the Division of Emergency Medicine, Department of Medicine, Children s Hospital Boston, Boston, MA. Received January 5, 2012; revisions received January 31 and February 2, 2012; accepted February 3, *These authors contributed equally to authorship. The authors have no relevant financial information or potential conflicts of interest to disclose. Supervising Editor: Lowell Gerson, PhD. Address for correspondence and reprints: Rebekah Mannix, MD, MPH; rebekah.mannix@childrens.harvard.edu. our knowledge, no studies have evaluated national practices in ED care based on insurance status. Analyzing whether insurance status is associated with differences in ED care is important because the majority of patients seen in EDs do not have private insurance, and prior studies have suggested that insurance status significantly influences both hospital-based and outpatient medical care. 4 7 Moreover, it is estimated that emergency physicians provide more acute care to public insurance beneficiaries and the uninsured than all other U.S. physicians combined, 8 and understanding how acute care is delivered to these patients is necessary in an era of health care reform. 9 The objective of this study was to determine whether insurance status is associated with differences in ED care patterns. METHODS Study Design This was a retrospective study of ED visits in the National Hospital Ambulatory Medical Care Survey ISSN ª 2012 by the Society for Academic Emergency Medicine 808 PII ISSN doi: /j x

2 ACADEMIC EMERGENCY MEDICINE July 2012, Vol. 19, No (NHAMCS) for the years 1999 through The study was approved by the authors institutional review board. Study Setting and Population The NHAMCS is an annual survey of hospital ED and outpatient department visits, designed by the Centers for Disease Control and Prevention s National Center for Health Statistics, and is administered by the U.S. Census Bureau. The survey measures ambulatory care service use in hospital EDs and outpatient clinics in the United States. Data are obtained from samples of geographically defined areas, hospitals within these areas, clinics and EDs within hospitals, and patient visits within these clinics and EDs as components of the fourstage probability design. A nationally representative sample of noninstitutional general (medical, surgical, and children s) and short-stay hospitals, excluding federal, military, and Veterans Administration hospitals, is selected within geographically defined areas (primary sampling units), after adjustment for size. ED visits and outpatient clinics are sampled separately. Data are collected on approximately 33,000 visits annually to approximately 480 hospital EDs and outpatient departments (sampled separately) and are used to derive national estimates. We limited our analyses to patients 19 to 64 years given that the vast majority of adults over 65 years are covered by Medicare. We excluded the pediatric population because of differences in the availability of insurance, disease prevalence, and access to outpatient care. We categorized patients ED visits by insurance status using the NHAMCS coding for private insurance, public insurance ( Medicaid ), and no insurance (combining self-pay, no charge ). Patients with Medicare were excluded from the analysis, as patients younger than 65 years who qualify for Medicare represent a sicker population. Insurance status was classified based on the primary expected source of payment for the ED visit. Study Protocol NHAMCS Protocol. Visit information is collected during a 4-week reporting period each year by trained staff members at the sampled hospitals with monitoring by NHAMCS field representatives. Patient charts are reviewed and relevant data are abstracted using a standardized patient record form. A field representative from the U.S. Census Bureau reviews the records used for missing visits or missing data. Data consistency is ensured by data processing at a central facility followed by a data management process involving edit determination, computer processing which includes automatic recoding algorithms, or manual editing. The NHAMCS data set is publicly available via the Internet. Data Collection and Processing. We examined the following variables: patient demographics (age, sex, race, and ethnicity), reason for visit, day of visit (weekday vs. weekend), and ED patient disposition (discharge vs. transfer or hospital admission including operating room, catheterization laboratory, or critical care unit). Patient race (white, black or African American, or other) and ethnicity (Hispanic or non-hispanic) were determined based on the observations of hospital personnel, unless it is hospital policy to ask patients directly for this information. This is in accordance with the NHAMCS instructions to record race and ethnicity according to the hospital s usual practice or based on your knowledge of the patient or from information in the medical record. 10 In the categorization of race, there were small sample sizes for Asians, Native Hawaiian or other Pacific Islanders, and American Indian or Alaskan natives, so these were combined into an other category to provide reliable estimates. For overall description of the study population, we divided our population into two age categories (18 through 44 years and 45 through 64 years). In the analysis of reason for visit, we chose complaints of chest pain, abdominal pain, and injury, which represent the top presenting complaints of adults visiting the ED. 11 We looked at reason for visit using the Reason for Visit Classification for Ambulatory Care codes for chest pain ( , , and ), abdominal pain ( ), and injury ( ). The NHAMCS data do not include any direct measure of illness severity, so we used two proxy variables, immediacy with which the patient should be seen and admit or transfer, to adjust for illness severity. Finally, we examined data on other hospital characteristics including region (Northeast, South, Midwest, and West), setting (urban and rural), and hospital ownership (voluntary nonprofit, government, or proprietary). We also included a variable for the proportion of private insurance patients seen at each hospital, to distinguish disparities in care within hospitals from those across hospitals. 12 This variable was calculated by dividing the number of private patients seen in each hospital by the total number of patients seen at that hospital (e.g., a hospital whose payer mix is normally 60% private would have a single hospital-level variable of 0.6). We included this variable because prior studies suggest that some disparities in care may be the result of hospital-based disparities; e.g., African American patients are more likely to be treated by a physician or hospital where all patients receive lower-quality care. 13 For ease of comparison, we transformed the variable into a categorical variable in which the lowest quartile was given a value of 1 and the highest was given a value of 4. Outcomes The primary outcomes were six components of ED care: wait time, left prior to discharge, use of diagnostic testing (laboratory or radiologic), treatment (medication or procedure), instructions for follow-up, and whether the patient had been seen in the past 72 hours. These outcomes were chosen based on prior studies demonstrating both disease-specific and general population-based associations of race, sex, and insurance with these outcomes in the ED For the categorization of left prior to discharge, we combined the variables left, left before medical screening

3 810 Mannix et al. INSURANCE STATUS IN THE ED exam, left after medical screening exam, and left against medical advice. For the follow-up variable, we used the variable no follow-up to ascertain whether the patient had been given any specific instructions for follow-up or not, excluding patients who left prior to discharge. The secondary outcome was the proportion of ED visits in which a significant illness was diagnosed. We defined significant illness using the Agency for Healthcare Research and Quality (AHRQ) Most Common Specific Reasons for Hospitalization 11 (Table 1). We included this outcome to understand whether insurance-associated differences in testing or treatment are also associated with differences in diagnosis of significant illness. Further, using AHRQ definitions of ambulatory care sensitive conditions, we divided significant illness into diagnoses that were ambulatory care sensitive conditions and those that were not, to assess whether insurance-associated differences in testing or treatment are potentially due to differences in outpatient care. 11 Data Analysis Weights, strata, and primary sampling unit design variables provided by the NHAMCS were used for all analyses. We used descriptive statistics, with appropriate weighting, to account for the survey sampling methodology, using the svy commands available in Stata 10.1 (StataCorp, College Station, TX). The relationship between insurance status and demographic variables was examined in a univariate fashion using the svy chi-square command. For the primary outcome of wait time, we compared median ED wait times in private versus public versus no insurance groups using the Kruskal-Wallis test and the unweighted sample, because medians and IQRs were virtually identical between weighted and unweighted samples. To assess the statistical significance of the relationship between insurance status and the proportions of ED visits during which patients left prior to discharge, were discharged Table 1 ICD-9 Codes for Most Common Specific Reasons for Hospitalizations Diagnoses ICD-9 Code Adults Pneumonia* 480xx-486xx COPD and allied conditions* 490xx-496xx Heart failure* 428xx Coronary atherosclerosis Osteoarthritis 715xx Mood disorder 296xx Cardiac dysrhythmia 427xx Nonspecific chest pain 786.5x Septicemia 038xx Complications of a device or graft 996xx Myocardial infarction 410xx Acute cerebrovascular disease 430xx-436xx COPD = chronic obstructive pulmonary disease; ICD-9 = International Classification of Diseases, Ninth Revision. *Ambulatory care sensitive condition. without follow-up, or were seen in the prior 72 hours, we used the weighted chi-square test. To evaluate the relationship between insurance status and diagnostic testing (laboratory or radiologic) and treatment (medication or procedure) we used both univariate and multivariate models, since the use of diagnostic testing could be associated with patient, provider, and hospital factors. For the multivariate model, we decided a priori to include age, sex, race, ethnicity, region, urban status, time of treatment (daytime defined as 8AM-5PM or not daytime), day of treatment (weekday or weekend), proportion of private patients seen in the hospital, triage status, admission transfer status, ambulatory care sensitive condition status, and year in the model. We used the same multivariate model to evaluate the relationship of insurance status to significant illness. RESULTS Visit Rates and Characteristics Of the 370,018 observed ED visits over the 10-year study period, 178,276 visits were for adults aged 19 to 64 years with private, public, or no insurance. Fifty percent (95% confidence interval [CI] = 49% to 52%) of patients were classified as having primary private insurance, while 22% (95% CI = 21% to 23%),and 28% (95% CI = 26% to 29%) were classified as having public or no insurance, respectively. Age, race, sex, and ethnicity were all associated with type of insurance. Demographics for the study population are shown in Table 2. Patients with no insurance were less likely than those with private or public insurance to present with chest pain (no insurance 6.7%, 95% CI = 6.4% to 7.0%; private 8.8%, 95% CI = 8.5% to 9.1%; public 7.7%, 95% CI = 7.3% to 8.1%; p < 0.001) or abdominal pain (no insurance 10%, 95% CI = 10% to 11%; private 11%, 95% CI = 10% to 11%; public 12%, 95% CI = 11% to 12%; p < 0.001). Patients with public insurance (12%, 95% CI = 12% to 13%) were less likely than those with private (19%, 95% CI = 18% to 20%) or no insurance (19%, 95% CI = 19% to 21%) to present with an injury complaint (p < 0.001). Association of Wait Times With Insurance Status Median ED wait times to see a physician were similar in patients with private (40 minutes, interquartile range [IQR] = 15 to 97 minutes) and public insurance (39 minutes, IQR = 14 to 97 minutes), but higher in patients without insurance (44 minutes, IQR = 15 to 108 minutes; p < 0.001). Association of Left Prior to Discharge, Follow-up, and Seen in the Prior 72 Hours With Insurance Status The percentage of total ED visits in which the patient left before discharge was highest in patients without insurance at 4.7% (95% CI = 4.2% to 5.1%), compared to 4.1% in patients with public insurance (95% CI = 3.8% to 4.5%), and 2.2% in those with private insurance (95% CI = 2.0% to 2.4%; p < 0.001). The proportion of ED visits without follow-up plans was

4 ACADEMIC EMERGENCY MEDICINE July 2012, Vol. 19, No Table 2 Demographics of the Study Population for Adults Aged Years Presenting to the ED From 1999 to 2008 Private Insurance Public Insurance No Insurance Age (yr) 18 to <45 56,600 (63) 31,471 (71) 36,894 (79) 45 to <65 32,270 (37) 10,990 (29) 9,244 (21) Sex Male 38,108 (44) 14,855 (32) 24,733 (51) Female 48,367 (56) 29,601 (68) 22,572 (49) Race White 67,315 (79) 27,134 (65) 32,140 (70) Black or African American 15,379 (18) 15,107 (32) 13,694 (27) Other 3,781 (3) 2,215 (3) 1,511 (3) Ethnicity: Hispanic 7,820 (9) 7,642 (14) 7,697 (15) Region Northeast 20,674 (20) 14,585 (22) 8,941 (15) Midwest 19,369 (24) 8,488 (24) 7,983 (19) South 27,555 (36) 13,189 (36) 23,132 (52) West 18,877 (20) 8,194 (18) 7,289 (14) Urban: MSA 75,136 (84) 38,711 (82) 42,037 (84) Triage status* <15 minutes 19,860 (25) (24) 9,253 (21) 15 minutes <1 hour 31,827 (41) 16,287 (41) 16,092 (39) 1 hr < 2 hours 17,169 (23) 8,619 (23) 10,187 (24) 2 hours 8,439 (11) 4,662 (12) 6,593 (16) Admission status: admitted transferred 10,898 (12) 6,814 (13) 4,825 (9) Time of day: daytime (8a-5p) 29,993 (35) 15,346 (34) 16,2256 (35) Day of week: weekday 60,320 (70) 32,335 (72) 34,455 (73) Hospital ownership Nonprofit 57,863 (77) 25,7307 (71) 26,787 (65) Government 8,961 (12) 10,965 (19) 10,289 (23) Proprietary 8,088 (11) 3,731 (10) 4,138 (12) Data are reported as n (survey weighted %). *Only patients with a recorded triage status. MSA = Metropolitan Statistical Area. highest in patients without insurance (7.5%, 95% CI = 6.8% to 8.4%), compared to those with public insurance (5.9%, 95% CI = 5.2% to 6.7%) and those with private insurance (6.8%, 95% CI = 6.1% to 7.7%) (p < 0.001). More than 5% of patients with public insurance (5.1%, 95% CI = 4.6% to 5.6%) and 4.7% of patients without insurance (95% CI = 4.1% to 4.6%) had been seen in the ED in the prior 72 hours, compared to 3.8% of patients with private insurance (95% CI = 3.5% to 4.2%; p < 0.001). Association of Testing and Treatment With Insurance Status Because patients without insurance were overall less likely to be triaged to be seen within the hour (60% vs. 66% for patients with private insurance and 65% for patients with pubic insurance, p < 0.001), we first looked to see if there were insurance-based differences in ED testing within triage statuses. Insurance-associated differences in testing were seen across all triage categories, with increased differences at the least acute triage categories. For patients triaged to be seen within 15 minutes, 72% (95% CI = 70% to 74%) of those with private insurance underwent testing, compared to 72% (95% CI = 70% to 73%) of those with public insurance and 69% (95% CI = 67% to 71%) of those with no insurance (p = 0.006). For patients triaged to be seen after 2 hours, 50% (95% CI = 48% to 51%) of those with Figure 1. Percentage of patients who receive any testing (laboratory or radiologic) by triage category and insurance status. Patients with nonprivate insurance were less likely to undergo testing (laboratory or radiologic) across all triage statuses, an effect that increased with decreasing urgency of triage status (p < for each triage status). private insurance underwent testing compared to 45% (95% CI = 43% to 47%) of those with public insurance and 43% (95% CI = 41% to 45%) of those with no insurance (p < 0.001; Figure 1).

5 812 Mannix et al. INSURANCE STATUS IN THE ED The Association of Significant Illnesses and Insurance Status Patients with public insurance (9.9%, 95% CI = 9.4% to 10%) were more likely to be diagnosed with a significant illness when compared to those with private insurance (7.8%, 95% CI = 7.5% to 8.1%) or no insurance (6.6%, 95% CI = 6.3% to 7.0%; p < 0.001). Among patients diagnosed with significant illness, 48% (95% CI = 46% to 50%) of those with private insurance were diagnosed with an ambulatory care sensitive condition versus 52% (95% CI = 49% to 54%) of those with public insurance and 56% (95% CI = 53% to 58%) of those with no insurance (p < 0.001). Multivariate Modeling of the Association of Insurance Status With Testing, Treatment, and Significant Illness To explore whether the relationship between testing and treatment in the ED was confounded by factors in addition to triage status, we conducted a multivariate model that included age, sex, race, ethnicity, region, urban status, time of treatment (daytime or not Table 3 Multivariate Model of the Association of Insurance Status With the Use of Diagnostic Testing and Intervention in Adults Aged Years Visiting the ED From 1999 to 2008 Receive Any Test Receive Any Medication Undergo Any Procedure Insurance Private 1 (reference) 1 (reference) 1 (reference) Public 0.84 ( )* 0.91 ( )* 0.83 ( )* None 0.82 ( )* 0Æ88 ( )* 0.79 ( )* Age (yr) 1.6 ( )* 0.98 ( ) 1.2 ( )* Sex Female 1 (reference) 1 (reference) 1 (reference) Male 0.85 ( )* 0.96 ( )* 1.0 ( )* Age (yr) 1.6 (1.5, 1.7)* 0.98 (0.90, 1.1) 1.2 (1.1, 1.3)* Race White 1 (reference) 1 (reference) 1 (reference) Black or African American 0.97 ( ) 0.90 ( )* 1.0 ( ) Other 0.99 ( ) 0.92 ( ) 0.98 ( ) Ethnicity Not Hispanic 1(reference) 1 (reference) 1 (reference) Hispanic 1.0 ( ) 0.92 ( )* 1.1 ( ) Region Northeast 1 (reference) 1 (reference) 1 (reference) Midwest 1.1 ( )* 1.1 ( ) 1.0 ( ) South 1.2 ( )* 1.2 ( )* 0.97 ( ) West 0.96 ( ) 1.2 ( )* 1.2 ( )* Urban Non-MSA 1 (reference) 1 (reference) 1 (reference) MSA 1.3 ( )* 0.90 ( ) 1.2 ( ) Triage statusà 0.79 ( )* 1.0 ( ) 0.84 ( )* Admitted transferred No 1 (reference) 1 (reference) 1 (reference) Yes 5.4 ( )* 0.85 ( )* 3.6 ( )* Triage statusà 0.79 (0.76, 0.81)* 1.0 (0.98, 1.1) 0.84 (0.82, 0.88)* Time of day Nighttime 1 (reference) 1 (reference) 1 (reference) Daytime (8AM-5PM) 1.0 ( )* 0.92 ( )* 0.97 ( ) Day of week Weekend 1 (reference) 1 (reference) 1 (reference) Weekday 1.1 ( )* 0.90 ( )* 0.96 ( )* Hospital ownership Nonprofit 1 (reference) 1 (reference) 1 (reference) Government 0.88 ( )* 0.90 ( ) 0.89 ( ) Proprietary 0.95 ( ) 1.0 ( ) 1.0 ( ) Year 1.0 ( )* 1.1 ( )* 1.1 ( )* Proportion of private patients 1.0 ( ) 1.0 ( ) 1.0 ( ) Ambulatory care sensitive condition No 1 (reference) 1 (reference) 1 (reference) Yes 1.3 ( ) 4.3 ( ) 1.2 ( ) Data are reported as AOR (95% CI). AOR = adjusted odds ratio. *Values have p < Categorical age (19 44 years, years). àcategorical triage status, of those with recorded triage status (<15 minutes, 15 minutes 1 hour, 1 2 hour, >2 hours, OR for each increase in less acute triage status). For each increase in quartile.

6 ACADEMIC EMERGENCY MEDICINE July 2012, Vol. 19, No daytime), day of treatment (weekday or weekend), proportion of private patients seen in the hospital (categorical), triage status, admission transfer status, ambulatory care sensitive condition, and year. On multivariate analyses (Table 3), when compared to patients with private insurance, patients with public insurance were overall less likely to receive any test, receive any medication, or undergo any procedure. The effect was even larger in those without insurance, who were less likely to receive any test, receive any medication, or undergo any procedure. On multivariate modeling of the association of insurance status with the diagnosis of significant illness, patients with public insurance were more likely to be diagnosed with a significant illness when compared to those with private insurance (adjusted odds ratio [AOR] = 1.2, 95% CI = 1.1 to 1.2), but there were no differences in significant illnesses in those without insurance when compared to those with private insurance (AOR 0.96, 95% CI = 0.87 to 1.1). DISCUSSION In this national probability-based sample, we report significantly different ED care patterns associated with insurance status for adult patients presenting to the ED between 1999 and To our knowledge, this is the first nationally based study that demonstrates an association with insurance status and emergency care in the general adult emergency patient population. It is unclear whether our findings reflect appropriate patterns of care delivery or disparate care. However, there is decreased testing and treatment in patients without private insurance, despite similar or increased odds of significant illness and process-level items including wait times, return visits, and follow-up instructions suggest that the care patterns may be less effective for these patients. In addition, increased testing in the privately insured, especially in the least acute triage statuses, may offer opportunities for cost containment. While the effect sizes described in our study are small, these small differences in testing or treatment represent very large numbers when extrapolated to the entire U.S. population. For example, while the differences in testing in the least acute triage status do not seem large (50% of patients with private insurance tested compared to 45% with public insurance and 43% of those with no insurance), they represent a difference of 7,500,000 tests in patients with private versus public insurance, and 4,400,000 tests in patients with private versus no insurance. Understanding these different practice patterns could therefore have significant economic effect. Our findings are similar to other regional or institution-specific studies that have reported patterns of decreased testing in uninsured ED patients. 18,19 It is of pivotal importance to understand the quality of care given to all patients in the emergency setting, but this is of particular import given debate around health care reform. 9 Prior studies have suggested that patients with public insurance may be less likely to receive evidencebased therapies and have worse outcomes. 20 Insurancebased differences in processes of care have also been previously described including wait times, 15 leaving prior to discharge, 14 and access to follow-up after ED visits. 21 It is uncertain how or why insurance status is associated with ED evaluation. However, several explanations are possible. First, it is plausible that insurance status influences clinical decision-making due to differences in physician perception of the patient. Insurance-based differences in medical evaluation may in part be due to differences in patient-specific qualities, such as assertiveness, language barriers, 22 or cultural expectation (e.g., advanced imaging if imaging is requested). Prior studies have also demonstrated that patients socioeconomic characteristics affect physician behaviors during medical encounters, as well as the treatments patients receive Studies in emergency medicine have shown significant disparities in complaint-based triage status 26 and wait times 15 associated with race, ethnicity, and insurance status. It is certainly possible that these initial care decisions could establish a cascade of downstream effects including decreased use of medical testing and treatment and increased risk of leaving prior to discharge. Second, financial considerations may play a role. Physicians may be credited with honoring patient requests for less testing or treating if such testing results in a major financial blow to the patient. Conversely, it is possible that financial incentives lead to overuse in patients with private insurance. Silverman et al. 27 reported increased Medicare spending in for-profit hospitals. However, most emergency physicians are salaried or paid by the hour, so it is not clear how financial incentives would affect care decisions in the ED. Nonetheless, overuse of testing in the ED could offer an opportunity for cost containment if not associated with improved outcomes. Third, it is possible that our findings represent referral bias and that patients with private insurance are referred to the EDs for specific workups, thus leading to improved triage processing, shorter wait times, and increased testing and treatment. Patients with private insurance are significantly more likely to have a usual primary care physician, 11 and prior studies show that referral by a primary care physician predicts resource utilization. 28 Fourth, it is possible that hospital-specific factors are at least in part responsible for our findings. Prior studies have suggested that African American patients are more likely to be treated by a physician or hospital where all patients receive lower-quality care, 13 and this may also be true of patients without private insurance. It is therefore possible that patients without private insurance have longer wait times, higher rates of leaving prior to discharge, and higher return visit rates because they are treated at hospitals where these variables are overall worse. However, in our multivariate modeling, we included a variable for the proportion of patients with private insurance seen at each hospital, to attempt to account for this clustering effect, and still observed different patterns of ED evaluation and treatment associated with insurance status. Fifth, it is possible that the relationship between insurance status and ED care is confounded by other clinical factors for which we were not able to control

7 814 Mannix et al. INSURANCE STATUS IN THE ED and that the care patterns are appropriate. Although we attempted to control for acuity in our multivariate model using triage and admission status, it is possible that there is residual confounding regarding decision for testing or treatment. However, significant illness was actually diagnosed more frequently in patients with public insurance compared to those with private insurance. This is supported by prior studies that suggest that patients without private insurance tend to present for care later in the course of disease 29,30 often with worse outcome. 31,32 However, given the insuranceassociated differences in the frequency of both presenting chief complaint and the diagnosis of an ambulatory care sensitive condition, it is possible that case mix explains some of the differences found in our study. LIMITATIONS This study has several important limitations. First, it represents a cross-sectional analysis of previously collected data, and the national estimates may not be completely accurate. However, the consistency and rigor of the multistage sampling framework makes this unlikely. We further safeguarded the accuracy of our estimates by complying with the statistical directives of the CDC s National Center for Health Statistics and ensuring that the outcomes we measured were based on more than 30 records of patient visits (the cutoff for reliability) and that relative standard error of the estimates was less than 30%. Second, the NHAMCS database provides limited clinical information. We used International Classification of Diseases, Ninth Revision (ICD-9), codes to ascertain diagnostic outcomes, similar to prior published studies, but these codes do not reflect subtleties faced by clinicians. Information as to the severity of illness is also lacking. Although we attempted to adjust for illness severity with the variables immediacy with which the patient should be seen and admission or transfer, we had no means of adjusting for severity of illness with validated instruments such as the Acute Physiology and Chronic Health Evaluation score (APACHE). However, we assume that the combined admission transfer variable would encompass all severe illness and, if anything, bias our results toward the null, as less severely ill patients could be included. Last, and perhaps most importantly, the NHAMCS database also does not offer data on outcomes, so we are unable to demonstrate the effect of the disparate care patterns on outcomes. However, we believe observing insuranceassociated differences in ED care is the first important step in assessing whether there are also associated differences in outcomes. CONCLUSIONS From 1999 to 2008, nonprivate insurance status was associated with different patterns of ED care in adults aged 19 to 64 years, despite the same or increased likelihood of significant illness. Work is needed to understand the patient, hospital, and physician factors responsible for these care patterns. Further research into outcomes after ED visits in private insurance versus not private insurance patients is clearly warranted. References 1. McCaig LF, Burt CW. National Hospital Ambulatory Medical Care Survey: 1999 emergency department summary. Adv Data. 2001; (320): Niska R, Bhuiya F, Xu J. National Hospital Ambulatory Medical Care Survey: 2007 emergency department summary. Natl Health Stat Report. 2010; (26): Tang N, Stein J, Hsie RY, Maselli JH, Gonzales R. Trends and characteristics of US emergency department visits, JAMA. 2010; 304: Fowler RA, Noyahr LA, Thronton JD, et al. An official American Thoracic Society systematic review: the association between health insurance status and access, care delivery, and outcomes for patients who are critically ill. Am J Respir Crit Care Med. 2010; 181: Hadley J. Insurance coverage, medical care use, and short-term health changes following an unintentional injury or the onset of a chronic condition. JAMA. 2007; 297: Abdullah F, Zhang Y, Lardaro T, et al. Analysis of 23 million US hospitalizations: uninsured children have higher all-cause in-hospital mortality. J Public Health (Oxf). 2010; 32: Sox CM, Burstin HR, Edwards RA, O Neil AC, Brennan TA. Hospital admissions through the emergency department: does insurance status matter? Am J Med. 1998; 105: Pitts SR, Carrier ER, Rich EC, Kellermann AL. Where Americans get acute care: increasingly, it s not at their doctor s office. Health Aff (Millwood). 2010; 29: Rosenthal MB. Hard choices alternatives for reining in Medicare and Medicaid spending. N Engl J Med. 2011; 364: National Center for Health Statistics. Ambulatory Health Care Data. April 8, Agency for Healthcare & Human Services. New and Numbers: The Most Common Reasons People Visit the Emergency Room. Available at: care411.ahrq.gov/transcript.aspx?id=759&type=seg. Accessed May 2, Pines JM, Localio AR, Hollander JE. Racial disparities in emergency department length of stay for admitted patients in the United States. Acad Emerg Med. 2009; 16: Skinner J, Chandra A, Staiger D, Lee J, McClellan M. Mortality after acute myocardial infarction in hospitals that disproportionately treat black patients. Circulation. 2005; 112: Bourgeois FT, Shannon MW, Stack AM. Left without being seen : a national profile of children who leave the emergency department before evaluation. Ann Emerg Med. 2008; 52: James CA, Bourgeois FT, Shannon MW. Association of race ethnicity with emergency department wait times. Pediatrics. 2005; 115:e Pezzin LE, Keyl PM, Green GB. Disparities in the emergency department evaluation of chest pain patients. Acad Emerg Med. 2007; 14: Wu BU, Banks PA, Conwell DL. Disparities in emergency department wait times for acute

8 ACADEMIC EMERGENCY MEDICINE July 2012, Vol. 19, No gastrointestinal illnesses: results from the National Hospital Ambulatory Medical Care Survey, Am J Gastroenterol. 2009; 104: Jackson P. The impact of health insurance status on emergency room services. J Health Soc Policy. 2001; 14: Svenson JE, Spurlock CW. Insurance status and admission to hospital for head injuries: are we part of a two-tiered medical system? Am J Emerg Med. 2001; 19: Calvin JE, Roe MT, Chen AY, et al. Insurance coverage and care of patients with non-st-segment elevation acute coronary syndromes. Ann Intern Med. 2006; 145: Asplin BR, Rhodes KV, Levy H, et al. Insurance status and access to urgent ambulatory care follow-up appointments. JAMA. 2005; 294: Hostetler MA, Auinger P, Szilagyi PG. Parenteral analgesic and sedative use among ED patients in the United States: combined results from the National Hospital Ambulatory Medical Care Survey (NHAMCS) Am J Emerg Med. 2002; 20: van Ryn M, Burke J. The effect of patient race and socio-economic status on physicians perceptions of patients. Soc Sci Med. 2000; 50: Epstein AM, McNeil BJ. The effects of patient characteristics on ambulatory test ordering. Soc Sci Med. 1985; 21: Epstein AM, Taylor WC, Seage GR 3rd. Effects of patients socioeconomic status and physicians training and practice on patient-doctor communication. Am J Med. 1985; 78: Lopez L, Wilper AP, Cervantes MC, Betancourt JR, Green AR. Racial and sex differences in emergency department triage assessment and test ordering for chest pain, Acad Emerg Med. 2010; 17: Silverman EM, Skinner JS, Fisher ES. The association between for-profit hospital ownership and increased Medicare spending. N Engl J Med. 1999; 341: Rinderknecht AS, Ho M, Matykiewicz P, Grupp- Phelan JM. Referral to the emergency department by a primary care provider predicts severity of illness. Pediatrics. 2010; 126: Brown DL, Schneider DL, Colbert R, Guss D. Influence of insurance coverage on delays in seeking emergency care in patients with acute chest pain. Am J Cardiol. 1998; 82: Canto JG, Rogers WJ, French WJ, Gore JM, Chandra NC, Barron HV. Payer status and the utilization of hospital resources in acute myocardial infarction: a report from the National Registry of Myocardial Infarction 2. Arch Intern Med. 2000; 160: Roetzheim RG, Pal N, Gonzalez EC, Ferrante JM, Van Durme DJ, Krischer JP. Effects of health insurance and race on colorectal cancer treatments and outcomes. Am J Public Health. 2000; 90: Ayanian JZ, Kohler BA, Abe T, Epstein AM. The relation between health insurance coverage and clinical outcomes among women with breast cancer. N Engl J Med. 1993; 329: