ORIGINAL CONTRIBUTION
Emergency Department Utilization Among Frequent Users With Psychiatric Visits Jesse J. Brennan, MA, Theodore C. Chan, MD, Renee Y. Hsia, MD, MSc, Michael P. Wilson, MD, PhD, and Edward M. Castillo, PhD, MPH
Abstract Objectives: The objective was to assess the incidence of psychiatric visits among frequent emergency department (ED) users and utilization among frequent psychiatric users. Methods: This was a multicenter retrospective longitudinal study of 1.76 million adult ED visits from acute care hospitals serving the metropolitan San Diego region (2008 through 2010) using nonpublic data submitted to the Office of Statewide Health Planning and Development. Frequent users were defined as having at least four ED visits within 12 consecutive months and were further classified into three groups based on the number of primary psychiatric visits in the same period, as defined by the primary discharge diagnosis (zero, one to three, and four or more visits). Descriptive, univariate, and logistic regression analyses are reported. Results: Patients with at least one primary psychiatric visit were 4.6 (95% confidence interval [CI] = 4.5 to 4.7) times more likely to be frequent ED users compared to patients with none. However, the majority of frequent ED users (80.1%) did not have any primary psychiatric visits; 16.6% of frequent ED users were classified as occasional psychiatric users, and only 3.3% were classified as frequent psychiatric users. Compared to frequent users without primary psychiatric visits, frequent psychiatric users were more likely to be male, non-Hispanic white, and without private insurance and visited the ED more frequently. Frequent psychiatric users also suffered from psychiatric, medical, and substance abuse comorbidity. Conclusions: While patients with primary psychiatric visits were more likely to be frequent users of the ED, only a small percentage of frequent ED users were seen primarily and repeatedly for psychiatric care. However, this small group of frequent psychiatric users visited the ED at a much higher rate than other frequent users and was burdened with both chronic medical and psychiatric conditions. ACADEMIC EMERGENCY MEDICINE 2014;21:1015–1022 © 2014 by the Society for Academic Emergency Medicine
E
mergency departments (EDs) are becoming increasingly crowded, with the number of ED visits nationwide estimated at 129.8 million in 2010 and rising.1 Between 1997 and 2007, the number of annual ED visits climbed from an estimated 94.9 million to 116.8 million visits, a 23% increase in just 10 years.2 ED crowding can lead to negative outcomes, such as longer waiting times, longer overall length of stay,3,4 and even increased mortality.5 Much research to date has focused on patient factors contributing to
crowding, such as patients who frequently use the ED for care.6–10 Frequent ED users are a relatively small group of patients who account for a large number of ED visits; as few as 4.5% to 8% of an ED patient population may account for up to 21% to 28% of all ED visits.11 In addition, frequent ED users are more likely to have chronic medical conditions, drug and/or alcohol problems, and psychiatric disorders.6,8 Of specific interest in this study was the frequent ED user with a psychiatric disorder. In 2010, there were an
From the Departments of Emergency Medicine, University of California at San Diego (JJB, TCC, MPW, EMC), San Diego, CA; and the University of California at San Francisco (RYH), San Francisco, CA. Received March 7, 2014; revisions received May 5 and May 13, 2014; accepted May 19, 2014. This work was supported by an award from the Office of the National Coordinator for Health Information Technology, US Department of Health and Human Services (cooperative agreement 90BC0015/01). The authors have no relevant financial information or potential conflicts of interest to disclose. Supervising Editor: Mark Mycyk, MD. Address for correspondence and reprints: Jesse J. Brennan, MA; e-mail: [email protected].
© 2014 by the Society for Academic Emergency Medicine doi: 10.1111/acem.12453
ISSN 1069-6563 PII ISSN 1069-6563583
1015 1015
1016
estimated 4.5 million visits nationwide with a primary diagnosis of a mental disorder, accounting for 3.5% of all ED visits.1 In a statewide study between 2008 and 2010, psychiatric-associated visits to the ED increased by 17.7% and accounted for 9.3% of all ED visits in North Carolina in 2010.12 Additionally, patients with psychiatric disorders are more likely to have multiple ED encounters than patients without psychiatric disorders,13 particularly if they suffer from mental illness14 or substance abuse comorbidity.15,16 Understanding the patterns of utilization for patients with psychiatric disorders who frequently seek emergency care in an acute care setting is critical to reshaping perceptions of frequent ED users and informing successful ED-based interventions. The purpose of this study was to assess the incidence of psychiatric visits among frequent ED users and to assess related patterns of utilization among frequent ED users with and without multiple psychiatric visits. METHODS Study Design This was a multicenter retrospective longitudinal cohort study of hospital ED visits from all 18 nonmilitary acute care hospitals serving the metropolitan San Diego region between 2008 and 2010, using data submitted to the California Office of Statewide Health Planning and Development (OSHPD). All licensed hospitals in the state are required to report health care utilization data in a standardized format to OSHPD each year, which are in turn made available to both the public and the nonpublic consumer. Data reported here were available in two separate nonpublic data sources for hospitals and government agencies. Patients included in the Patient Discharge Dataset who were admitted from an ED were extracted and merged with the Emergency Department Dataset to construct a complete ED utilization database. Patients without valid patient identifiers were excluded. Detailed information on these data sources is available elsewhere.17 It is important to note that this dataset allows study of ED visits across any ED in the entire county. Therefore, rather than solely focusing on frequent ED users as defined by a single institution or hospital system, this study captures frequent users of the larger regional health care system. This study was approved by the institutional Human Research Protections Program. Study Setting and Population San Diego regional hospitals serve an estimated population of roughly 3.2 million residents across a 4,206square-mile land mass; approximately 1.4 million reside within the city of San Diego. During this 3-year study period, 788,005 patients with 1,764,559 ED visits were identified who were at least 18 years of age at the first visit in the study period. Measures Measures for this study consisted of standardized utilization data reported to OSHPD, which included limited demographic information, service date, hospital, primary source of payment, discharge disposition, and up
Brennan et al. • FREQUENT PSYCHIATRIC USERS
to 25 International Classification of Disease 9th Revision Clinical Modification (ICD-9-CM) diagnoses codes. Patient demographic information was based on the first visit in the study period and included age, ethnicity and race, sex, and primary payer. Primary psychiatric visits were defined as visits with a primary discharge diagnosis of a mental disorder, as defined by any of the following ICD-9-CM codes: 290.0–302.x and 306.0–316.x. Primary diagnoses were also used to identify patients with visits for substance abuse (303.0–305.x). The Charlson Comorbidity Index18 was assessed for each patient using up to 25 diagnoses at all visits during the study period. Comorbidity index scores were calculated as a sum of 17 comorbidities weighted for disease severity using the enhanced coding algorithm provided by Quan and colleagues19 and were categorized for analysis as follows: 0, 1, 2, and ≥3. Frequent User Classification. Frequent ED users were defined as patients with four or more visits within any consecutive 12 months of time during the 3-year study period. Frequent ED users were further classified into three smaller groups in the same manner based on the maximum number of primary psychiatric visits within any 12-month period: nonpsychiatric users (frequent ED users with no primary psychiatric visits), occasional psychiatric users (frequent ED users with one to three primary psychiatric visits), and frequent psychiatric users (frequent ED users with four or more primary psychiatric visits). Data Analysis Descriptive analyses of patient demographic and visit characteristics during the study period were conducted on the overall sample and separately for each frequent user group. Relative risk (RR) was assessed to determine the likelihood that patients with a primary psychiatric diagnosis would be classified as frequent ED users compared to patients without a primary psychiatric diagnosis. RR was also assessed to determine the association of substance abuse among the three frequent users groups. A 95% confidence interval (CI) is reported for each RR. The Kruskal-Wallis test was used to compare the total number of ED visits among frequent user groups (using patient as the unit of analysis) and was followed up by nonparametric Mann-Whitney U-tests to specifically compare the numbers of ED visits between each of the three groups. Two separate direct logistic regressions models were developed using patient as the unit of analysis, comparing the nonpsychiatric group to both the occasional and the frequent psychiatric groups, to determine the independent association of each predictor with each outcome. Predictors were based on clinical relevance given the available data from OSHPD and included patient age in years (18 to 24, 25 to 44, 45 to 64, and 65 and older), male sex, ethnicity/race (non-Hispanic white, Hispanic, non-Hispanic African American, non-Hispanic other), primary payer (private, Medicare, Medi-Cal, selfpay/indigent), and the comorbidity index score (0, 1, 2, ≥3) and were entered into each model regardless of statistical association with the outcome. While patients treated within one hospital may be different than
ACADEMIC EMERGENCY MEDICINE • September 2014, Vol. 21, No. 9 • www.aemj.org
patients treated at another hospital, we did not include hospital as a random effects factor to help account for this variation. Given that San Diego–area EDs are accessible to any patient in the region, and that patients commonly go to different EDs or urgent care centers (especially frequent users), controlling for within-hospital variation was not practical. The omnibus chi-square statistic was used to assess goodness of fit for each model. Adjusted odds ratios (OR) and 95% CIs are reported for each predictor. Data were assessed for distribution patterns, cell size, and missing data prior to analyses. Missing data were minimal, affecting less than 0.1% of cases. An alpha level equal to 0.05 was used for interpretation of statistical significance for all statistical analyses with no adjustment for multiple comparisons. Minimum sample size and power were not calculated prior to analyses as no specific hypotheses were being tested; however, sample size for the smallest group (n = 2,394) was considered to be adequate for all data analyses. All statistical analyses were conducted using the IBM SPSS Statistics 19.0 software package. RESULTS Overall, 71,611 (9.1%) patients were classified as frequent users of the ED, identified as having at least four visits in a consecutive 12-month period. These frequent ED users accounted for 646,544 (36.6%) ED visits. While 40,222 (5.1%) patients had at least one primary psychiatric visit and were nearly five times more likely to be classified as frequent users than patients without a primary psychiatric visit (35.5% vs. 7.7%, RR = 4.6, 95% CI = 4.5 to 4.7), the majority of frequent ED users (80.1%) had no primary psychiatric visits in the entire study period, and an additional 16.6% (n = 11,882) had just one to three primary psychiatric visits. Only 3.3% (n = 2,394) of frequent ED users had four or more primary psychiatric visits in a consecutive 12-month period (see Table 1). Demographic Characteristics A detailed description of patient characteristics by type of frequent user is displayed in Table 2. Frequent psychiatric users were different from occasional and nonpsychiatric users on several demographic characteristics. Frequent psychiatric users were predominantly younger than 65 years of age (95.1%), male (56.6%), non-Hispanic white (68.0%), and without private insurance (87.8%). In comparison, 64.5% of nonpsychiatric frequent users were younger than 65 years of age, 41.5% were male, 57.2% were non-Hispanic white, and 76.4% were without private insurance. ED Visit Frequency Approximately 34.2% of frequent psychiatric users visited the ED more than 20 times for all-cause complaints, compared to 10.3% of occasional psychiatric users and 3.1% of non-psychiatric users (see Table 2). Overall, frequent psychiatric users had a significantly higher number of ED visits (median = 15 visits, interquartile range [IQR] = 9 to 26 visits) compared to occasional (median = 8 visits, IQR = 6 to 12 visits) and nonpsychiatric users
1017
Table 1 Number of Patients and Visits by ED Visit Frequency Maximum No. of Visits in 1 Year*
Patients
All ED patients No. of ED visits One to three (occasional 716,344 user) Four or more (frequent 71,661 user) No. of primary psychiatric visits None 747,783 One or more 40,222 Frequent ED users No. of primary psychiatric visits None (nonpsychiatric 57,385 user) One to three (occasional 11,882 psychiatric user) Four or more (frequent 2,394 psychiatric user)
(90.9) (9.1)
Total Visits†
1,118,015 (63.4) 646,544 (36.6)
(94.9) (5.1)
1,516,775 (86.0) 247,784 (14.0)
(80.1)
454,559 (70.3)
(16.6)
136,965 (21.2)
(3.3)
55,020 (8.5)
Data are reported as n (%). *Maximum number of visits in any consecutive 12-month period. †Total number of visits in 3-year study period.
(median = 6 visits, IQR = 5 to 9 visits; p < 0.001 for both comparisons). Frequent psychiatric users also visited more hospitals, with 27.2% visiting five or more hospitals compared to 8.5% of the occasional psychiatric user group and 2.2% of the nonpsychiatric user group (see Table 2). Although not as pronounced, the relationship among the psychiatric user group and the number of different hospitals visited persisted even after accounting for group differences in ED visit frequency (data not shown). Comorbidity Index More than 68% of frequent psychiatric users had comorbidity index scores of one or more, indicating that these patients were diagnosed with at least one comorbid disease during the study period. This finding was similar for both occasional (63%) and nonpsychiatric users (69%; see Table 2). Among frequent psychiatric users, the three most common conditions were chronic pulmonary disease (49%), diabetes with or without chronic complications (22%), and mild liver disease (19%). Similarly, chronic pulmonary disease and diabetes were the most common condition among occasional (36 and 21%, respectively) and nonpsychiatric users (35 and 28%, respectively). However, frequent psychiatric users were disproportionately affected by chronic pulmonary disease, even after accounting for differences in age. Table 3 reports the three most common comorbid diseases among frequent psychiatric users by age group. Substance Abuse The proportion of patients with a primary diagnosis of substance abuse was 25.6% among frequent psychiatric users, compared to 16.1% of occasional psychiatric users and 3.2% of nonpsychiatric users. Even after
1018
Brennan et al. • FREQUENT PSYCHIATRIC USERS
Table 2 Patient Characteristics by Type of Frequent ED User Nonpsychiatric Users (n = 57,385)
Patient Characteristics Age (yr) 18–24 25–44 45–64 ≥65 Sex Male Ethnicity/race Hispanic Non-Hispanic white Non-Hispanic African American Non-Hispanic other Primary payer Private Medicare Medi-Cal Self-pay/indigent Comorbidity index score 0 1 2 ≥3 Total number of ED visits 4–10 11–20 ≥21 Number of different hospitals visited 1–2 3–4 ≥5
6,335 14,712 15,993 20,345
Occasional Psychiatric Users (n = 11,882)
(11.0) (25.6) (27.9) (35.5)
1,379 4,117 4,174 2,212
(11.6) (34.6) (35.1) (18.6)
Frequent Psychiatric Users (n = 2,394) 256 1,077 944 117
(10.7) (45.0) (39.4) (4.9)
23,825 (41.5)
9,254 (35.4)
1,355 (56.6)
13,095 32,812 6,574 4,904
(22.8) (57.2) (11.5) (8.5)
2,229 7,722 1,248 683
(18.8) (65.0) (10.5) (5.7)
363 1,629 256 146
(15.2) (68.0) (10.7) (6.1)
13,537 22,353 10,325 11,170
(23.6) (39.0) (18.0) (19.5)
2,674 3,545 2,557 3,106
(22.5) (29.8) (21.5) (26.1)
291 729 719 655
(12.2) (30.5) (30.0) (27.4)
18,039 10,746 5,604 22,996
(31.4) (18.7) (9.8) (40.1)
4,338 3,013 1,347 3,184
(36.5) (25.4) (11.3) (26.8)
755 729 332 578
(31.5) (30.5) (13.9) (24.1)
48,301 (84.2) 7,303 (12.7) 1,781 (3.1)
7,993 (67.3) 2,670 (22.5) 1,219 (10.3)
828 (34.6) 747 (31.2) 819 (34.2)
45,603 (79.5) 10,526 (18.3) 1,256 (2.2)
6,817 (57.4) 4,053 (34.1) 1,012 (8.5)
804 (33.6) 938 (39.2) 652 (27.2)
Data are reported as n (%).
Table 3 Top Three Comorbid Diseases by Age and Type of Frequent ED User Nonpsychiatric Users (n = 57,385)
Comorbid Diseases* Ages 18–24 yr Chronic pulmonary Diabetes Mild liver disease Ages 25–44 yr Chronic pulmonary Diabetes Mild liver disease Ages 45–64 yr Chronic pulmonary Diabetes Mild liver disease Ages 65 yr and over Chronic pulmonary Diabetes Mild liver disease
Occasional Psychiatric Users (n = 11,882)
Frequent Psychiatric Users (n = 2,394)
disease
1,506 (23.8) 252 (4.0) 128 (2.0)
382 (27.7) 70 (5.1) 38 (2.8)
103 (40.2) 23 (9.0) 4 (1.6)
disease
3,764 (25.6) 1,986 (13.5) 981 (6.7)
1,269 (30.8) 526 (12.8) 511 (12.4)
500 (46.4) 204 (18.9) 181 (16.8)
disease
5,848 (36.6) 5,552 (34.7) 2,629 (16.4)
1,741 (41.7) 1,154 (27.6) 923 (22.1)
505 (53.5) 269 (28.5) 262 (27.8)
disease
8,861 (43.6) 8,117 (39.9) 1,425 (7.0)
872 (39.4) 763 (34.5) 133 (6.0)
60 (51.3) 38 (32.5) 10 (8.5)
Data are reported as n (%). *Derived from Charlson Comorbidity Index.
excluding patients with a primary diagnosis of drug or alcohol psychoses, frequent psychiatric users were seven times as likely (RR = 7.3, 95% CI = 6.7 to 8.0) as nonpsychiatric users and twice as likely (RR = 2.0, 95% CI = 1.8 to 2.2) as occasional psychiatric users to have
had a primary diagnosis of substance abuse. Among frequent psychiatric users, nondependent drug abuse, primarily alcohol abuse, was the most common primary diagnosis of substance abuse (22.1%), followed by alcohol dependence (12.6%) and drug dependence (2.3%).
ACADEMIC EMERGENCY MEDICINE • September 2014, Vol. 21, No. 9 • www.aemj.org
Psychiatric Disorders Frequent psychiatric users were diagnosed with serious and multiple mental disorders during the study period. The most common psychiatric disorders among these patients were schizophrenic disorders (55.6%), followed by episodic mood disorders (53.9%) and neurotic disorders (48.2%). In addition, just over half (51.0%) were diagnosed with at least two of these top three mental disorders. Among occasional psychiatric users, the most common psychiatric disorders included neurotic disorders (32.1%), episodic mood disorders (20.3%), and other depressive disorders (14.4%), although only 8.3% were diagnosed with at least two of these top three mental disorders. A detailed list of the 10 most common primary psychiatric diagnoses for occasional and frequent psychiatric users is presented in Table 4. Logistic Regression Analyses Logistic regression models comparing the nonpsychiatric group to both the occasional and frequent psychiatric groups are reported in Table 5. Compared to patients with no primary psychiatric visits, frequent psychiatric users were less likely to be 65 years of age or older (adjusted OR = 0.04), more likely to be male (adjusted OR = 1.94), less likely to be Hispanic (adjusted OR = 0.45) or non-Hispanic black or African American (adjusted OR = 0.50) than non-Hispanic white, and more likely to have Medicare (adjusted OR = 7.26) or Medi-Cal (adjusted OR = 3.90) or to be self-pay/indigent (adjusted OR = 2.37) than private coverage. Although less dramatic, similar results were found for
Table 4 Ten Most Common Primary Psychiatric Diagnoses by Type of Frequent ED User Primary Psychiatric Diagnoses* (ICD-9-CM code) Occasional psychiatric users (n = 11,882) Neurotic disorders (300.x) Episodic mood disorders (296.x) Depressive disorder, not elsewhere classified (311.x) Alcoholic psychoses (291.x) Schizophrenic disorders (295.x) Drug psychoses (292.x) Other nonorganic psychoses (298.x) Special symptoms or syndromes, not elsewhere classified (307.x) Acute reaction to stress (308.x) Other organic psychoses (294.x) Frequent psychiatric users (n = 2,394) Schizophrenic disorders (295.x) Episodic mood disorders (296.x) Neurotic disorders (300.x) Depressive disorder, not elsewhere classified (311.x) Other nonorganic psychoses (298.x) Alcoholic psychoses (291.x) Drug psychoses (292.x) Special symptoms or syndromes, not elsewhere classified (307.x) Acute reaction to stress (308.x) Paranoid states (297.x) *Documented at any visit in the study period.
n (%) 3,814 (32.1) 2,415 (20.3) 1,711 (14.4) 1,400 1,382 1,077 1,001 724
(11.8) (11.6) (9.1) (8.4) (6.1)
412 (3.5) 380 (3.2) 1,330 1,290 1,153 960
(55.6) (53.9) (48.2) (40.1)
802 354 251 200
(33.5) (14.8) (10.5) (8.4)
160 (6.7) 122 (5.1)
1019
occasional psychiatric users compared to nonpsychiatric users. DISCUSSION In this study, as defined as four or more ED visits in a year, frequent ED users accounted for 9.1% of all ED patients in the region and contributed 36.6% of all visits. This estimate is likely higher than previously reported studies on frequent users using a similar definition for two specific reasons. First, ED utilization was not limited to a single hospital or hospital system, but all 18 nonmilitary acute care hospitals in the region. Second, the identification of frequent users was not limited to a single year, but any consecutive 12-month period over the 3-year study. Emergency department patients with primary psychiatric diagnoses were nearly five times more likely to be seen in the ED on multiple occasions (four or more visits in a year). However, only a small percentage of frequent users (3.3%) actually had four or more primary psychiatric visits in a year. Although frequent psychiatric users comprised a relatively small group of ED patients, they were responsible for a disproportionate number of ED visits compared to other frequent users. Frequent psychiatric users also had a high incidence of substance abuse and were burdened by both comorbid diseases and serious psychiatric conditions. The prevalence of substance use among patients with psychiatric disorders is generally higher than patients without, contributing to poorer health outcomes.20–23 Patients with severe psychiatric and substance use comorbidity also require more intensive interventions for successful treatment.24 The high prevalence of substance abuse among frequent psychiatric users in this study (one in four) is also likely to be a contributing factor to the disproportionate number of ED visits seen in this population. For example, Curran and colleagues15 found that psychiatric patients with substance use comorbidity were more likely to have a higher number of ED visits than psychiatric patients without. This high prevalence of substance abuse in this population of frequent psychiatric users would suggest that treatment plans without consideration of substance use patterns may be at higher risk for failure. In addition, frequent psychiatric users in this study were disproportionately affected by chronic pulmonary disease. This is consistent with other studies indicating a high prevalence of psychiatric disorders among patients with chronic pulmonary disease.25–27 Of specific concern here is that, for example, among patients with chronic pulmonary disease, patients with mood and anxiety disorders are at increased risk for exacerbations of the disease.28,29 Further, while the overall prevalence of chronic pulmonary disease among frequent psychiatric users in this study was approximately one in two, it was also remarkably high for younger patients between the ages of 18 to 24 years (40%) and 25 to 44 years (46%). Considering this young age at which frequent psychiatric users are dealing with chronic pulmonary disease, in combination with the fact that nearly half of frequent psychiatric users were also diagnosed with serious psychiatric disorders, such as schizophrenic and
1020
Brennan et al. • FREQUENT PSYCHIATRIC USERS
Table 5 Adjusted Odds Ratios of Patient Characteristics by Type of Frequent ED User
Patient Characteristics Age (yr) 18–24 25–44 45–64 ≥65 Sex Male Ethnicity/race Non-Hispanic white Hispanic Non-Hispanic African American Non-Hispanic other Payer Private Medicare Medi-Cal Self-pay/indigent Comorbidity index score 0 1 2 ≥3
Occasional vs. Nonpsychiatric User Adjusted OR (95% CI)
p-value
Reference 1.26 (1.18–1.35) 1.19 (1.10–1.28) 0.37 (0.34–0.41)
—
Emergency Department Utilization Among Frequent Users With Psychiatric Visits Jesse J. Brennan, MA, Theodore C. Chan, MD, Renee Y. Hsia, MD, MSc, Michael P. Wilson, MD, PhD, and Edward M. Castillo, PhD, MPH
Abstract Objectives: The objective was to assess the incidence of psychiatric visits among frequent emergency department (ED) users and utilization among frequent psychiatric users. Methods: This was a multicenter retrospective longitudinal study of 1.76 million adult ED visits from acute care hospitals serving the metropolitan San Diego region (2008 through 2010) using nonpublic data submitted to the Office of Statewide Health Planning and Development. Frequent users were defined as having at least four ED visits within 12 consecutive months and were further classified into three groups based on the number of primary psychiatric visits in the same period, as defined by the primary discharge diagnosis (zero, one to three, and four or more visits). Descriptive, univariate, and logistic regression analyses are reported. Results: Patients with at least one primary psychiatric visit were 4.6 (95% confidence interval [CI] = 4.5 to 4.7) times more likely to be frequent ED users compared to patients with none. However, the majority of frequent ED users (80.1%) did not have any primary psychiatric visits; 16.6% of frequent ED users were classified as occasional psychiatric users, and only 3.3% were classified as frequent psychiatric users. Compared to frequent users without primary psychiatric visits, frequent psychiatric users were more likely to be male, non-Hispanic white, and without private insurance and visited the ED more frequently. Frequent psychiatric users also suffered from psychiatric, medical, and substance abuse comorbidity. Conclusions: While patients with primary psychiatric visits were more likely to be frequent users of the ED, only a small percentage of frequent ED users were seen primarily and repeatedly for psychiatric care. However, this small group of frequent psychiatric users visited the ED at a much higher rate than other frequent users and was burdened with both chronic medical and psychiatric conditions. ACADEMIC EMERGENCY MEDICINE 2014;21:1015–1022 © 2014 by the Society for Academic Emergency Medicine
E
mergency departments (EDs) are becoming increasingly crowded, with the number of ED visits nationwide estimated at 129.8 million in 2010 and rising.1 Between 1997 and 2007, the number of annual ED visits climbed from an estimated 94.9 million to 116.8 million visits, a 23% increase in just 10 years.2 ED crowding can lead to negative outcomes, such as longer waiting times, longer overall length of stay,3,4 and even increased mortality.5 Much research to date has focused on patient factors contributing to
crowding, such as patients who frequently use the ED for care.6–10 Frequent ED users are a relatively small group of patients who account for a large number of ED visits; as few as 4.5% to 8% of an ED patient population may account for up to 21% to 28% of all ED visits.11 In addition, frequent ED users are more likely to have chronic medical conditions, drug and/or alcohol problems, and psychiatric disorders.6,8 Of specific interest in this study was the frequent ED user with a psychiatric disorder. In 2010, there were an
From the Departments of Emergency Medicine, University of California at San Diego (JJB, TCC, MPW, EMC), San Diego, CA; and the University of California at San Francisco (RYH), San Francisco, CA. Received March 7, 2014; revisions received May 5 and May 13, 2014; accepted May 19, 2014. This work was supported by an award from the Office of the National Coordinator for Health Information Technology, US Department of Health and Human Services (cooperative agreement 90BC0015/01). The authors have no relevant financial information or potential conflicts of interest to disclose. Supervising Editor: Mark Mycyk, MD. Address for correspondence and reprints: Jesse J. Brennan, MA; e-mail: [email protected].
© 2014 by the Society for Academic Emergency Medicine doi: 10.1111/acem.12453
ISSN 1069-6563 PII ISSN 1069-6563583
1015 1015
1016
estimated 4.5 million visits nationwide with a primary diagnosis of a mental disorder, accounting for 3.5% of all ED visits.1 In a statewide study between 2008 and 2010, psychiatric-associated visits to the ED increased by 17.7% and accounted for 9.3% of all ED visits in North Carolina in 2010.12 Additionally, patients with psychiatric disorders are more likely to have multiple ED encounters than patients without psychiatric disorders,13 particularly if they suffer from mental illness14 or substance abuse comorbidity.15,16 Understanding the patterns of utilization for patients with psychiatric disorders who frequently seek emergency care in an acute care setting is critical to reshaping perceptions of frequent ED users and informing successful ED-based interventions. The purpose of this study was to assess the incidence of psychiatric visits among frequent ED users and to assess related patterns of utilization among frequent ED users with and without multiple psychiatric visits. METHODS Study Design This was a multicenter retrospective longitudinal cohort study of hospital ED visits from all 18 nonmilitary acute care hospitals serving the metropolitan San Diego region between 2008 and 2010, using data submitted to the California Office of Statewide Health Planning and Development (OSHPD). All licensed hospitals in the state are required to report health care utilization data in a standardized format to OSHPD each year, which are in turn made available to both the public and the nonpublic consumer. Data reported here were available in two separate nonpublic data sources for hospitals and government agencies. Patients included in the Patient Discharge Dataset who were admitted from an ED were extracted and merged with the Emergency Department Dataset to construct a complete ED utilization database. Patients without valid patient identifiers were excluded. Detailed information on these data sources is available elsewhere.17 It is important to note that this dataset allows study of ED visits across any ED in the entire county. Therefore, rather than solely focusing on frequent ED users as defined by a single institution or hospital system, this study captures frequent users of the larger regional health care system. This study was approved by the institutional Human Research Protections Program. Study Setting and Population San Diego regional hospitals serve an estimated population of roughly 3.2 million residents across a 4,206square-mile land mass; approximately 1.4 million reside within the city of San Diego. During this 3-year study period, 788,005 patients with 1,764,559 ED visits were identified who were at least 18 years of age at the first visit in the study period. Measures Measures for this study consisted of standardized utilization data reported to OSHPD, which included limited demographic information, service date, hospital, primary source of payment, discharge disposition, and up
Brennan et al. • FREQUENT PSYCHIATRIC USERS
to 25 International Classification of Disease 9th Revision Clinical Modification (ICD-9-CM) diagnoses codes. Patient demographic information was based on the first visit in the study period and included age, ethnicity and race, sex, and primary payer. Primary psychiatric visits were defined as visits with a primary discharge diagnosis of a mental disorder, as defined by any of the following ICD-9-CM codes: 290.0–302.x and 306.0–316.x. Primary diagnoses were also used to identify patients with visits for substance abuse (303.0–305.x). The Charlson Comorbidity Index18 was assessed for each patient using up to 25 diagnoses at all visits during the study period. Comorbidity index scores were calculated as a sum of 17 comorbidities weighted for disease severity using the enhanced coding algorithm provided by Quan and colleagues19 and were categorized for analysis as follows: 0, 1, 2, and ≥3. Frequent User Classification. Frequent ED users were defined as patients with four or more visits within any consecutive 12 months of time during the 3-year study period. Frequent ED users were further classified into three smaller groups in the same manner based on the maximum number of primary psychiatric visits within any 12-month period: nonpsychiatric users (frequent ED users with no primary psychiatric visits), occasional psychiatric users (frequent ED users with one to three primary psychiatric visits), and frequent psychiatric users (frequent ED users with four or more primary psychiatric visits). Data Analysis Descriptive analyses of patient demographic and visit characteristics during the study period were conducted on the overall sample and separately for each frequent user group. Relative risk (RR) was assessed to determine the likelihood that patients with a primary psychiatric diagnosis would be classified as frequent ED users compared to patients without a primary psychiatric diagnosis. RR was also assessed to determine the association of substance abuse among the three frequent users groups. A 95% confidence interval (CI) is reported for each RR. The Kruskal-Wallis test was used to compare the total number of ED visits among frequent user groups (using patient as the unit of analysis) and was followed up by nonparametric Mann-Whitney U-tests to specifically compare the numbers of ED visits between each of the three groups. Two separate direct logistic regressions models were developed using patient as the unit of analysis, comparing the nonpsychiatric group to both the occasional and the frequent psychiatric groups, to determine the independent association of each predictor with each outcome. Predictors were based on clinical relevance given the available data from OSHPD and included patient age in years (18 to 24, 25 to 44, 45 to 64, and 65 and older), male sex, ethnicity/race (non-Hispanic white, Hispanic, non-Hispanic African American, non-Hispanic other), primary payer (private, Medicare, Medi-Cal, selfpay/indigent), and the comorbidity index score (0, 1, 2, ≥3) and were entered into each model regardless of statistical association with the outcome. While patients treated within one hospital may be different than
ACADEMIC EMERGENCY MEDICINE • September 2014, Vol. 21, No. 9 • www.aemj.org
patients treated at another hospital, we did not include hospital as a random effects factor to help account for this variation. Given that San Diego–area EDs are accessible to any patient in the region, and that patients commonly go to different EDs or urgent care centers (especially frequent users), controlling for within-hospital variation was not practical. The omnibus chi-square statistic was used to assess goodness of fit for each model. Adjusted odds ratios (OR) and 95% CIs are reported for each predictor. Data were assessed for distribution patterns, cell size, and missing data prior to analyses. Missing data were minimal, affecting less than 0.1% of cases. An alpha level equal to 0.05 was used for interpretation of statistical significance for all statistical analyses with no adjustment for multiple comparisons. Minimum sample size and power were not calculated prior to analyses as no specific hypotheses were being tested; however, sample size for the smallest group (n = 2,394) was considered to be adequate for all data analyses. All statistical analyses were conducted using the IBM SPSS Statistics 19.0 software package. RESULTS Overall, 71,611 (9.1%) patients were classified as frequent users of the ED, identified as having at least four visits in a consecutive 12-month period. These frequent ED users accounted for 646,544 (36.6%) ED visits. While 40,222 (5.1%) patients had at least one primary psychiatric visit and were nearly five times more likely to be classified as frequent users than patients without a primary psychiatric visit (35.5% vs. 7.7%, RR = 4.6, 95% CI = 4.5 to 4.7), the majority of frequent ED users (80.1%) had no primary psychiatric visits in the entire study period, and an additional 16.6% (n = 11,882) had just one to three primary psychiatric visits. Only 3.3% (n = 2,394) of frequent ED users had four or more primary psychiatric visits in a consecutive 12-month period (see Table 1). Demographic Characteristics A detailed description of patient characteristics by type of frequent user is displayed in Table 2. Frequent psychiatric users were different from occasional and nonpsychiatric users on several demographic characteristics. Frequent psychiatric users were predominantly younger than 65 years of age (95.1%), male (56.6%), non-Hispanic white (68.0%), and without private insurance (87.8%). In comparison, 64.5% of nonpsychiatric frequent users were younger than 65 years of age, 41.5% were male, 57.2% were non-Hispanic white, and 76.4% were without private insurance. ED Visit Frequency Approximately 34.2% of frequent psychiatric users visited the ED more than 20 times for all-cause complaints, compared to 10.3% of occasional psychiatric users and 3.1% of non-psychiatric users (see Table 2). Overall, frequent psychiatric users had a significantly higher number of ED visits (median = 15 visits, interquartile range [IQR] = 9 to 26 visits) compared to occasional (median = 8 visits, IQR = 6 to 12 visits) and nonpsychiatric users
1017
Table 1 Number of Patients and Visits by ED Visit Frequency Maximum No. of Visits in 1 Year*
Patients
All ED patients No. of ED visits One to three (occasional 716,344 user) Four or more (frequent 71,661 user) No. of primary psychiatric visits None 747,783 One or more 40,222 Frequent ED users No. of primary psychiatric visits None (nonpsychiatric 57,385 user) One to three (occasional 11,882 psychiatric user) Four or more (frequent 2,394 psychiatric user)
(90.9) (9.1)
Total Visits†
1,118,015 (63.4) 646,544 (36.6)
(94.9) (5.1)
1,516,775 (86.0) 247,784 (14.0)
(80.1)
454,559 (70.3)
(16.6)
136,965 (21.2)
(3.3)
55,020 (8.5)
Data are reported as n (%). *Maximum number of visits in any consecutive 12-month period. †Total number of visits in 3-year study period.
(median = 6 visits, IQR = 5 to 9 visits; p < 0.001 for both comparisons). Frequent psychiatric users also visited more hospitals, with 27.2% visiting five or more hospitals compared to 8.5% of the occasional psychiatric user group and 2.2% of the nonpsychiatric user group (see Table 2). Although not as pronounced, the relationship among the psychiatric user group and the number of different hospitals visited persisted even after accounting for group differences in ED visit frequency (data not shown). Comorbidity Index More than 68% of frequent psychiatric users had comorbidity index scores of one or more, indicating that these patients were diagnosed with at least one comorbid disease during the study period. This finding was similar for both occasional (63%) and nonpsychiatric users (69%; see Table 2). Among frequent psychiatric users, the three most common conditions were chronic pulmonary disease (49%), diabetes with or without chronic complications (22%), and mild liver disease (19%). Similarly, chronic pulmonary disease and diabetes were the most common condition among occasional (36 and 21%, respectively) and nonpsychiatric users (35 and 28%, respectively). However, frequent psychiatric users were disproportionately affected by chronic pulmonary disease, even after accounting for differences in age. Table 3 reports the three most common comorbid diseases among frequent psychiatric users by age group. Substance Abuse The proportion of patients with a primary diagnosis of substance abuse was 25.6% among frequent psychiatric users, compared to 16.1% of occasional psychiatric users and 3.2% of nonpsychiatric users. Even after
1018
Brennan et al. • FREQUENT PSYCHIATRIC USERS
Table 2 Patient Characteristics by Type of Frequent ED User Nonpsychiatric Users (n = 57,385)
Patient Characteristics Age (yr) 18–24 25–44 45–64 ≥65 Sex Male Ethnicity/race Hispanic Non-Hispanic white Non-Hispanic African American Non-Hispanic other Primary payer Private Medicare Medi-Cal Self-pay/indigent Comorbidity index score 0 1 2 ≥3 Total number of ED visits 4–10 11–20 ≥21 Number of different hospitals visited 1–2 3–4 ≥5
6,335 14,712 15,993 20,345
Occasional Psychiatric Users (n = 11,882)
(11.0) (25.6) (27.9) (35.5)
1,379 4,117 4,174 2,212
(11.6) (34.6) (35.1) (18.6)
Frequent Psychiatric Users (n = 2,394) 256 1,077 944 117
(10.7) (45.0) (39.4) (4.9)
23,825 (41.5)
9,254 (35.4)
1,355 (56.6)
13,095 32,812 6,574 4,904
(22.8) (57.2) (11.5) (8.5)
2,229 7,722 1,248 683
(18.8) (65.0) (10.5) (5.7)
363 1,629 256 146
(15.2) (68.0) (10.7) (6.1)
13,537 22,353 10,325 11,170
(23.6) (39.0) (18.0) (19.5)
2,674 3,545 2,557 3,106
(22.5) (29.8) (21.5) (26.1)
291 729 719 655
(12.2) (30.5) (30.0) (27.4)
18,039 10,746 5,604 22,996
(31.4) (18.7) (9.8) (40.1)
4,338 3,013 1,347 3,184
(36.5) (25.4) (11.3) (26.8)
755 729 332 578
(31.5) (30.5) (13.9) (24.1)
48,301 (84.2) 7,303 (12.7) 1,781 (3.1)
7,993 (67.3) 2,670 (22.5) 1,219 (10.3)
828 (34.6) 747 (31.2) 819 (34.2)
45,603 (79.5) 10,526 (18.3) 1,256 (2.2)
6,817 (57.4) 4,053 (34.1) 1,012 (8.5)
804 (33.6) 938 (39.2) 652 (27.2)
Data are reported as n (%).
Table 3 Top Three Comorbid Diseases by Age and Type of Frequent ED User Nonpsychiatric Users (n = 57,385)
Comorbid Diseases* Ages 18–24 yr Chronic pulmonary Diabetes Mild liver disease Ages 25–44 yr Chronic pulmonary Diabetes Mild liver disease Ages 45–64 yr Chronic pulmonary Diabetes Mild liver disease Ages 65 yr and over Chronic pulmonary Diabetes Mild liver disease
Occasional Psychiatric Users (n = 11,882)
Frequent Psychiatric Users (n = 2,394)
disease
1,506 (23.8) 252 (4.0) 128 (2.0)
382 (27.7) 70 (5.1) 38 (2.8)
103 (40.2) 23 (9.0) 4 (1.6)
disease
3,764 (25.6) 1,986 (13.5) 981 (6.7)
1,269 (30.8) 526 (12.8) 511 (12.4)
500 (46.4) 204 (18.9) 181 (16.8)
disease
5,848 (36.6) 5,552 (34.7) 2,629 (16.4)
1,741 (41.7) 1,154 (27.6) 923 (22.1)
505 (53.5) 269 (28.5) 262 (27.8)
disease
8,861 (43.6) 8,117 (39.9) 1,425 (7.0)
872 (39.4) 763 (34.5) 133 (6.0)
60 (51.3) 38 (32.5) 10 (8.5)
Data are reported as n (%). *Derived from Charlson Comorbidity Index.
excluding patients with a primary diagnosis of drug or alcohol psychoses, frequent psychiatric users were seven times as likely (RR = 7.3, 95% CI = 6.7 to 8.0) as nonpsychiatric users and twice as likely (RR = 2.0, 95% CI = 1.8 to 2.2) as occasional psychiatric users to have
had a primary diagnosis of substance abuse. Among frequent psychiatric users, nondependent drug abuse, primarily alcohol abuse, was the most common primary diagnosis of substance abuse (22.1%), followed by alcohol dependence (12.6%) and drug dependence (2.3%).
ACADEMIC EMERGENCY MEDICINE • September 2014, Vol. 21, No. 9 • www.aemj.org
Psychiatric Disorders Frequent psychiatric users were diagnosed with serious and multiple mental disorders during the study period. The most common psychiatric disorders among these patients were schizophrenic disorders (55.6%), followed by episodic mood disorders (53.9%) and neurotic disorders (48.2%). In addition, just over half (51.0%) were diagnosed with at least two of these top three mental disorders. Among occasional psychiatric users, the most common psychiatric disorders included neurotic disorders (32.1%), episodic mood disorders (20.3%), and other depressive disorders (14.4%), although only 8.3% were diagnosed with at least two of these top three mental disorders. A detailed list of the 10 most common primary psychiatric diagnoses for occasional and frequent psychiatric users is presented in Table 4. Logistic Regression Analyses Logistic regression models comparing the nonpsychiatric group to both the occasional and frequent psychiatric groups are reported in Table 5. Compared to patients with no primary psychiatric visits, frequent psychiatric users were less likely to be 65 years of age or older (adjusted OR = 0.04), more likely to be male (adjusted OR = 1.94), less likely to be Hispanic (adjusted OR = 0.45) or non-Hispanic black or African American (adjusted OR = 0.50) than non-Hispanic white, and more likely to have Medicare (adjusted OR = 7.26) or Medi-Cal (adjusted OR = 3.90) or to be self-pay/indigent (adjusted OR = 2.37) than private coverage. Although less dramatic, similar results were found for
Table 4 Ten Most Common Primary Psychiatric Diagnoses by Type of Frequent ED User Primary Psychiatric Diagnoses* (ICD-9-CM code) Occasional psychiatric users (n = 11,882) Neurotic disorders (300.x) Episodic mood disorders (296.x) Depressive disorder, not elsewhere classified (311.x) Alcoholic psychoses (291.x) Schizophrenic disorders (295.x) Drug psychoses (292.x) Other nonorganic psychoses (298.x) Special symptoms or syndromes, not elsewhere classified (307.x) Acute reaction to stress (308.x) Other organic psychoses (294.x) Frequent psychiatric users (n = 2,394) Schizophrenic disorders (295.x) Episodic mood disorders (296.x) Neurotic disorders (300.x) Depressive disorder, not elsewhere classified (311.x) Other nonorganic psychoses (298.x) Alcoholic psychoses (291.x) Drug psychoses (292.x) Special symptoms or syndromes, not elsewhere classified (307.x) Acute reaction to stress (308.x) Paranoid states (297.x) *Documented at any visit in the study period.
n (%) 3,814 (32.1) 2,415 (20.3) 1,711 (14.4) 1,400 1,382 1,077 1,001 724
(11.8) (11.6) (9.1) (8.4) (6.1)
412 (3.5) 380 (3.2) 1,330 1,290 1,153 960
(55.6) (53.9) (48.2) (40.1)
802 354 251 200
(33.5) (14.8) (10.5) (8.4)
160 (6.7) 122 (5.1)
1019
occasional psychiatric users compared to nonpsychiatric users. DISCUSSION In this study, as defined as four or more ED visits in a year, frequent ED users accounted for 9.1% of all ED patients in the region and contributed 36.6% of all visits. This estimate is likely higher than previously reported studies on frequent users using a similar definition for two specific reasons. First, ED utilization was not limited to a single hospital or hospital system, but all 18 nonmilitary acute care hospitals in the region. Second, the identification of frequent users was not limited to a single year, but any consecutive 12-month period over the 3-year study. Emergency department patients with primary psychiatric diagnoses were nearly five times more likely to be seen in the ED on multiple occasions (four or more visits in a year). However, only a small percentage of frequent users (3.3%) actually had four or more primary psychiatric visits in a year. Although frequent psychiatric users comprised a relatively small group of ED patients, they were responsible for a disproportionate number of ED visits compared to other frequent users. Frequent psychiatric users also had a high incidence of substance abuse and were burdened by both comorbid diseases and serious psychiatric conditions. The prevalence of substance use among patients with psychiatric disorders is generally higher than patients without, contributing to poorer health outcomes.20–23 Patients with severe psychiatric and substance use comorbidity also require more intensive interventions for successful treatment.24 The high prevalence of substance abuse among frequent psychiatric users in this study (one in four) is also likely to be a contributing factor to the disproportionate number of ED visits seen in this population. For example, Curran and colleagues15 found that psychiatric patients with substance use comorbidity were more likely to have a higher number of ED visits than psychiatric patients without. This high prevalence of substance abuse in this population of frequent psychiatric users would suggest that treatment plans without consideration of substance use patterns may be at higher risk for failure. In addition, frequent psychiatric users in this study were disproportionately affected by chronic pulmonary disease. This is consistent with other studies indicating a high prevalence of psychiatric disorders among patients with chronic pulmonary disease.25–27 Of specific concern here is that, for example, among patients with chronic pulmonary disease, patients with mood and anxiety disorders are at increased risk for exacerbations of the disease.28,29 Further, while the overall prevalence of chronic pulmonary disease among frequent psychiatric users in this study was approximately one in two, it was also remarkably high for younger patients between the ages of 18 to 24 years (40%) and 25 to 44 years (46%). Considering this young age at which frequent psychiatric users are dealing with chronic pulmonary disease, in combination with the fact that nearly half of frequent psychiatric users were also diagnosed with serious psychiatric disorders, such as schizophrenic and
1020
Brennan et al. • FREQUENT PSYCHIATRIC USERS
Table 5 Adjusted Odds Ratios of Patient Characteristics by Type of Frequent ED User
Patient Characteristics Age (yr) 18–24 25–44 45–64 ≥65 Sex Male Ethnicity/race Non-Hispanic white Hispanic Non-Hispanic African American Non-Hispanic other Payer Private Medicare Medi-Cal Self-pay/indigent Comorbidity index score 0 1 2 ≥3
Occasional vs. Nonpsychiatric User Adjusted OR (95% CI)
p-value
Reference 1.26 (1.18–1.35) 1.19 (1.10–1.28) 0.37 (0.34–0.41)
—
