Emergency Department Visits in the United States for Pediatric Depression: Estimates of Charges and Hospitalization

Transcription

1 ORIGINAL CONTRIBUTION Emergency Department Visits in the United States for Pediatric Depression: Estimates of Charges and Hospitalization Diana Sun, MS, Ivo Abraham, PhD, Marion Slack, PhD, and Grant H. Skrepnek, PhD Abstract Objectives: The objective of this study was to calculate national estimates of depression-related emergency department (ED) visits and associated health care resource use among children and adolescents 17 years or younger. Another goal was to explore the effects of certain sociodemographic and health care system factors and comorbidities on ED charges and subsequent hospitalization in the United States. Methods: The authors analyzed data from the 2006 and 2009 National Emergency Department Sample (NEDS), the largest source of U.S. ED data. ED visits with all listed diagnoses (i.e., principal diagnosis plus secondary conditions) of depression were identified using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes , 296.2X, 296.3X, 300.4, and 311. Population-based estimates of ED visits, hospitalization, resource use, comorbidities, and demographics associated with pediatric depression were calculated. Potentially significant covariate associations were also explored using ED charges and hospital admission from the ED. Results: The 2006 and 2009 NEDS sample contained 365,713 ED visits for pediatric depression; the majority were made by adolescents (87.9%). Of these, 27.2% were admitted to the hospital, 69.5% were treated and released, and <0.1% died in ED. The ED charges in 2012 U.S. dollars summed to a hospital bill of $443.8 million, with the ED plus inpatient charges ($1.2 billion) being more than double that amount. The median inpatient length of stay (LOS) was 4.0 days. Suicide and intentional self-inflicted injury were attempted by 31.4% of the patients. Attention-deficit, conduct, and disruptive disorders; anxiety disorders; substance use disorders; asthma; and infections were the most common comorbidities. In year 2009, a higher number of diagnoses, older age, being female, key comorbidities, and suicide and intentional self-inflicted injury were significantly associated with higher ED charges (all p < 0.05). Increased odds of hospital admission from the ED were significantly associated with a higher number of diagnoses, key comorbidities, and suicide and intentional self-inflicted injury (all p < 0.05). Conclusions: Pediatric depression is common in the ED and is associated with significant burden to the health care system. Certain factors such as a higher number of diagnoses, key comorbidities, and suicide and intentional self-inflicted injury are associated with increased health care costs and resource use. Special attention should be given to these factors, when present. ACADEMIC EMERGENCY MEDICINE 2014;21: by the Society for Academic Emergency Medicine Pediatric depression is a commonly seen medical condition in the emergency department (ED). Children and adolescents often lack optimal access to primary care services and therefore rely heavily on the ED for care. 1 4 According to a statistical brief developed by the Agency for Healthcare Research and Quality (AHRQ), children 17 years or younger had nearly 6.6 million hospital stays in the United States; From the Center for Health Outcomes and PharmacoEconomic Research, College of Pharmacy, University of Arizona (DS, IA, MS), Tucson, AZ; and the Oklahoma Health Sciences Center, College of Pharmacy, University of Oklahoma (GHS), Oklahoma City, OK. Received January 27, 2014; revision received April 4, 2014; accepted May 1, The authors have no relevant financial information or potential conflicts of interest to disclose. Supervising Editor: Jennifer Walthall, MD, MPH. Address for correspondence and reprints: Diana Sun, MS; dianasun@pharmacy.arizona.edu by the Society for Academic Emergency Medicine ISSN doi: /acem PII ISSN

2 1004 Sun et al. ED VISITS IN THE UNITED STATES FOR PEDIATRIC DEPRESSION 17% of those patients were admitted through the ED, with depression and bipolar disorder ranked as the sixth most common reason for hospital admission. 5 However common depression may be in pediatric EDs, there is little research on the estimates of ED care and associated resource use in this population. Factors such as sociodemographic characteristics, insurance status, hospital type, and geographic region are believed to play a significant role in the likelihood of child psychiatric hospitalization. 6,7 Comorbidity is also known to influence service utilization for youths with mental health problems. 8 Common comorbid conditions include anxiety disorders, behavior disorders, mood disorders, and substance use disorders. 9 If left untreated, depression may damage the development of a young person s emotional, cognitive, and social skills and place him or her at higher risk for suicide and self-injurious behaviors. 10 Because the burden of pediatric depression is substantial, it is important to determine the epidemiology of depression-related ED visits, hospital admissions, and associated costs to elucidate areas of potential improvement on a nationwide level. Previous studies have focused largely on depression-related outcomes in the inpatient setting, but have ignored the effect of ED visits Continued research can help providers target specific issues surrounding young individuals with possible depression visiting the ED. It can also assist decision makers in identifying driving factors of higher health care resource use and costs that can potentially be avoided if adequate primary care resources are available. The objective of this study was to calculate national estimates of depression-related ED visits and associated health care resource use among children and adolescents 17 years of age or younger in the United States from the perspective of the payer. The effects of sociodemographic characteristics, health care system factors, and relevant comorbidities on ED charges and subsequent hospitalization were also analyzed. We hypothesized that higher ED charges and increased odds of hospitalization would be associated with worse patient outcomes as defined by a greater number of comorbidities. METHODS Study Design This was a retrospective cross-sectional analysis using data from the 2006 and 2009 National Emergency Department Sample (NEDS), a component of the Healthcare Cost and Utilization Project (HCUP) developed by AHRQ. The University of Arizona Institutional Review Board considered this analysis of deidentified, publicly available data as exempt from informed consent requirements. Study Setting and Population The NEDS is the largest source of ED data in the United States and contains discharge data on approximately 26 million (2006) and 29 million (2009) records for ED visits. By design, the NEDS represents a stratified 20% sample of U.S. hospital-based EDs, which the American Hospital Association defines as all non-federal, short-term, general, and other specialty hospitals. 14,15 Public hospitals and academic medical centers are included, as are specialty hospitals. Long-term care, rehabilitation, psychiatric, and alcoholism/chemical dependency hospitals are not included. The NEDS was constructed using records from the HCUP State Emergency Department Databases (SEDD) and the State Inpatient Databases (SID). 14,15 The SEDD contains data on ED visits that do not result in an admission (i.e., treatand-release ED visits or transfers to another hospital), and the SID captures data on patients initially seen in the ED and later admitted to the same hospital. 14,15 Taken together, the resultant NEDS contains information on short-term outcomes for all patients admitted through the ED regardless of disposition. In 2006, the NEDS collected discharge information from 950 hospitals in 24 states; in 2009, data were captured for 964 hospitals in 29 states. For this study, we used 2 years of NEDS data to increase the sample size and to increase the external validity of our findings. Sampling weights were used to produce nationally representative estimates. Additional details of the NEDS can be found elsewhere. 16 We included ED visits made by patients aged 17 years or younger with all listed diagnoses (i.e., principal diagnosis plus secondary conditions) of depression based on International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes: transient organic psychotic condition, depressive type (293.83); major depressive disorder, single episode (296.2X); major depressive disorder, recurrent episode (296.3X); dysthymic disorder (300.4); and depressive disorder, not elsewhere specified (311). 13 Because the NEDS contains up to 15 diagnoses, all listed diagnoses (DX1 DXn) in this study refer to any of the top 15 listed conditions. This criterion was considered because ED visits often focus on the symptom-based evaluation of differential diagnoses, and several conditions may have relevance to the reason for the ED visit; therefore, all listed diagnoses may need to be considered. 17 To assess the overall effect (and to avoid possible misclassifications and resultant exclusions) of depression, we included ED visits with any diagnosis of the condition. Measurements The variables of interest were patient characteristics (age, sex, patient residence, zip code income, and primary payer), hospital characteristics (hospital location, hospital region, and hospital teaching status), year (2006 or 2009), number of diagnoses, and relevant comorbidities. The comorbidities selected for analysis were identified from past literature and included anxiety disorders; attention-deficit, conduct, and disruptive behavior disorders; bipolar disorder; alcohol use disorders; substance use disorders; eating disorders; asthma, infections (e.g., bacterial, viral); fluid and electrolyte disorders; epilepsy; pain (e.g., abdominal, chest); open wounds; fracture; meningitis; noninfectious gastroenteritis; fever of unknown origin; diabetes with complications; cancer; inflammatory bowel disorder; overweight and obesity; and developmental disorders. As the occurrence of self-injurious behaviors tend to be more frequent in young depressed patients, suicide and

3 ACADEMIC EMERGENCY MEDICINE September 2014, Vol. 21, No self-inflicted injury was also included as a risk factor. 35,36 For all analyses, the Clinical Classification Software for Mental Health and Substance Abuse (CCS-MHSA) was applied, when applicable. 37 The CCS-MHSA categories and ICD-9-CM codes for defining comorbidities in children and adolescents with depression are presented in Data Supplement S1 (available as supporting information in the online version of this paper). The primary outcome measures were ED charges and hospital admission from the ED. Identifying risk factors of higher resource use can alert physicians to detect high-risk patients and trigger prompt initiation of the treatment process. The secondary outcome measures were inpatient length of stay (LOS) and ED plus inpatient charges. Charges were adjusted for inflation by using the Medical Care component of the Consumer Price Index, with all amounts reported as 2012 U.S. dollars. Data on ED LOS and actual ED costs are not available in the NEDS. Data Analysis Descriptive statistics were used to summarize patient and hospital characteristics as well as relevant comorbidities. Frequencies, proportions, means, standard deviations (SDs), medians, interquartile range (IQR), and sums are reported, as appropriate. Multivariate statistics were used to analyze ED charges and hospital admission from the ED. Two generalized linear models were fitted independently while controlling for covariates (patient s residence, zip code income, primary payer, hospital location, hospital region, and hospital teaching status). ED charges were log transformed and assessed using a generalized linear model with Gaussian distribution and identity link, and hospital admission from the ED was assessed using a generalized linear model with binomial distribution and log link. Results are reported as exponentiated beta (exp ) coefficients, which correspond to relative risk measures (e.g., odds ratios). The optimal model for each outcome was selected based on the evaluation of residual analyses (e.g., deviance, Anscombe) and goodness-of-fit statistics (e.g., Akaike information criterion, Bayesian information criterion). 38 Heteroskedasticity, multicollinearity, and other diagnostics associated with regression models were also examined. To adjust for age-related differences, results were stratified into three age groups: children (0 5 years), youth (6 12 years), and adolescents (13 17 years). Data analyses were conducted using SPSS version 20 and STATA IC version 12 software. Sampling weights were incorporated to consider the complex survey design and sampling procedures of the NEDS, allowing findings to be nationally representative. Missing data were handled using the weighting approach. All statistical tests were two-sided, and a p-value of less than 0.05 was considered significant. Bonferroni corrections were used whenever multiple pairwise comparisons were made. Sensitivity Analyses To test the robustness of our findings and to address the confounding effect of listing depression as a secondary condition, we performed a series of sensitivity analyses: a conservative case definition with only a principal diagnosis of depression and a semiconservative case definition with tertiary diagnoses of depression. The principal diagnosis (DX1) describes the diagnosis, condition, problem or other reason for encounter/visit shown in the medical records to be chiefly responsible for the services provided ; tertiary diagnoses (DX1, DX2, and DX3) refer to the principal diagnosis plus secondary conditions. However, the conservative case definition had small sample size issues and did not permit the regression model to perform properly; therefore, we only repeated the analysis using the semiconservative case definition. RESULTS After the weighting procedures, there were an estimated 365,713 ED visits for children and adolescents with all listed diagnoses of depression from 2006 and Among these, 27.2% were admitted to the hospital, 69.5% were treated and released, and <0.1% died in ED. The majority of ED visits were made by adolescents (87.9%), followed by youths (11.8%) and then children (0.3%). The ED charges summed to $443.8 million (2012 U.S. dollars), with the ED plus inpatient charges ($1.2 billion) being more than double that amount. The mean (SD) ED charge was $1,508 ($1,568), and the median was $1,132 (IQR = $718.7 to $2,173.4). When we include the hospital services, the mean (SD) ED plus inpatient charge was $14,941 ($33,204), and the median was $8,579 (IQR = $6,564.5 to $16,630.9). Adolescents had higher total ED charges and total ED plus inpatient charges ($400.1 million and $1.0 billion, respectively) compared with youths ($42.8 million and $180.9 million) and children ($873,502 and $2.7 million). The median inpatient LOS was 4.0 days (IQR = 2.0 to 6.0 days), and the mean (SD) was 5.2 days (8.2 days). Patient and Hospital Characteristics Table 1 presents the patient and hospital characteristics of ED visits associated with pediatric depression. Across all ED visits, the mean (SD) age was 3.1 (1.9) years for children, 10.6 (1.6) years for youths, and 15.4 (1.3) years for adolescents. Females (62.9%) had a greater proportion of ED visits than males (37.0%). With regard to patient disposition following ED visits, 52.8% were discharged routinely, while 5.3% were transferred to other short-term hospitals, 16.2% were transferred to other facilities (including skilled nursing and intermediate care facilities), <0.1% were discharged to home health care, 0.5% were discharged against medical advice, 21.9% were admitted as inpatients to the same hospital, <0.1% died in ED, and 3.3% were not admitted with destination unknown. Among ED visits that that resulted in hospital admission, 18.5% were routinely discharged, 0.7% were transferred to short-term hospitals, 2.3% were transferred to other facilities, 0.2% were discharged to home health care, 0.2% were discharged against medical advice, <0.1% died in hospital, and <0.1% were discharged alive with destination unknown. Most of the patients lived in large metropolitan areas (48.9%), with

4 1006 Sun et al. ED VISITS IN THE UNITED STATES FOR PEDIATRIC DEPRESSION Table 1 Patient and Hospital Characteristics of ED Visits Associated With All Listed Diagnoses of Pediatric Depression Stratified by Age Group, 2006 and 2009* Characteristics Children (n = 945, 0 5 yr) Youths (n = 43,103, 6 12 yr) Adolescents (n = 321,665, yr) Overall (n = 365,713, 0 17 yr) Patient characteristics Age (yr), mean (SD) 3.1 (1.9) 10.6 (1.6) 15.4 (1.3) 14.8 (2.1) Sex Male 479 (50.7) 23,219 (53.9) 111,729 (34.7) 135,426 (37.0) Female 460 (48.7) 19,859 (46.1) 209,759 (65.2) 230,078 (62.9) Unspecified 6 (0.6) 25 (0.1) 178 (0.1) 209 (0.1) Patient s residence Large metropolitan 380 (40.2) 21,686 (50.3) 156,849 (48.8) 178,915 (48.9) Medium metropolitan 339 (35.9) 10,209 (23.7) 75,536 (23.5) 86,084 (23.5) Small metropolitan 99 (10.5) 4,239 (9.8) 31,346 (9.7) 35,684 (9.8) Micropolitan 96 (10.2) 4,524 (10.5) 36,929 (11.5) 41,549 (11.4) Rural 24 (2.6) 2,271 (5.3) 19,176 (6.0) 21,471 (5.9) Unspecified 6 (0.6) 174 (0.4) 1,830 (0.6) 2,010 (0.5) Zip code income $1-$39, (24.1) 11,140 (25.8) 70,170 (21.8) 81,537 (22.3) $40,000-$49, (44.9) 12,230 (28.4) 83,873 (26.1) 96,527 (26.4) $50,000-$65, (17.1) 10,771 (25.0) 84,084 (26.1) 95,016 (26.0) $66,000 or more 107 (11.4) 8,118 (18.8) 77,039 (24.0) 85,264 (23.3) Unspecified 24 (2.5) 845 (2.0) 6,499 (2.0) 7,368 (2.0) Primary payer Medicare 15 (1.6) 74 (0.2) 673 (0.2) 762 (0.2) Medicaid 457 (48.4) 22,361 (51.9) 125,168 (38.9) 147,986 (40.5) Private 304 (32.1) 16,419 (38.1) 158,732 (49.3) 175,455 (48.0) Uninsured 88 (9.3) 2,300 (5.3) 21,576 (6.7) 23,964 (6.6) Other 76 (8.0) 1,835 (4.3) 14,195 (4.4) 16,106 (4.4) Unspecified 5 (0.6) 113 (0.3) 1,321 (0.4) 1,439 (0.4) Disposition from ED Routine 602 (63.7) 23,140 (53.7) 169,269 (52.6) 193,011 (52.8) Transfer to short-term hospital 37 (4.0) 1,955 (4.5) 17,218 (5.4) 19,211 (5.3) Other transfers 65 (6.9) 5,906 (13.7) 53,434 (16.6) 59,405 (16.2) Home health care 0 8 (<0.1) 103 (<0.1) 111 (<0.1) Against medical advice 4 (0.5) 160 (0.4) 1,512 (0.5) 1,677 (0.5) Admitted as an inpatient to 159 (16.9) 10,208 (23.7) 69,803 (21.7) 80,170 (21.9) this hospital Died in ED 0 4 (<0.1) 54 (<0.1) 58 (<0.1) Not admitted, destination unknown 76 (8.0) 1,720 (4.0) 10,273 (3.2) 12,069 (3.3) Disposition from hospital admission Routine 133 (14.1) 9,363 (21.7) 58,203 (18.1) 67,699 (18.5) Transfer to short-term hospital 3 (0.4) 205 (0.5) 2,206 (0.7) 2,424 (0.7) Other transfers 9 (0.9) 470 (1.1) 7,974 (2.5) 8,453 (2.3) Home health care 9 (0.9) 97 (0.2) 579 (0.2) 685 (0.2) Against medical advice 0 51 (0.1) 590 (0.2) 641 (0.2) Died in hospital 5 (0.6) 5 (<0.1) 80 (<0.1) 90 (<0.1) Discharged alive, 0 10 (<0.1) 71 (<0.1) 81 (<0.1) destination unknown Unspecified 785 (83.1) 32,904 (76.3) 251,961 (78.3) 285,650 (78.1) Hospital characteristics Hospital location Rural 82 (8.6) 4,351 (10.1) 37,042 (11.5) 41,475 (11.3) Urban 848 (89.7) 37,277 (87.5) 271,667 (84.5) 309,792 (84.7) Other 16 (1.6) 1,475 (3.4) 12,955 (4.0) 14,446 (4.0) Hospital region Northeast 300 (31.7) 11,747 (27.3) 79,636 (24.8) 91,683 (25.1) Midwest 187 (19.8) 12,781 (29.7) 95,161 (29.6) 108,129 (29.6) South 345 (36.6) 13,955 (32.4) 96,652 (30.0) 110,952 (30.3) West 112 (11.9) 4,619 (10.7) 50,217 (15.6) 54,948 (15.0) Hospital teaching status Nonteaching 398 (42.2) 15,757 (36.6) 128,558 (40.0) 144,713 (39.6) Teaching 449 (47.5) 21,519 (49.9) 143,110 (44.5) 165,078 (45.1) Other 97 (10.3) 5,826 (13.5) 49,998 (15.5) 55,921 (15.3) All data are reported as n (%) unless otherwise noted. *Results are based on Healthcare Cost and Utilization Project Nationwide Emergency Department Sample weighted sampling of discharged children and adolescents aged younger than 18 years with all listed diagnoses of codes , 296.2X, 296.3X, 300.4, and 311 from the International Classification of Diseases, Ninth Revision, Clinical Modification. Percentages may not add to 100% due to rounding.

5 ACADEMIC EMERGENCY MEDICINE September 2014, Vol. 21, No % in medium metropolitan areas, 9.8% in small metropolitan areas, 11.4% in micropolitan areas, and 5.9% in rural areas. About one-fourth (22.3%) were identified as having zip code income of less than $39,000. Most patients were covered by either private insurance (48.0%) or Medicaid (40.5%), although some (6.6%) were uninsured. EDs attached to hospitals in urban locations (84.7%), the South region (30.3%), and in teaching hospitals (45.1%) treated most patients. Comorbidities There were 114,774 (31.4%) patients with attempted suicide and intentional self-inflicted injury (Table 2). Among these, 13.7% were children, 22.4% were youths, and 32.6% were adolescents. Of the 20 relevant comorbidities, attention-deficit, conduct, and disruptive behavior disorders (17.6%); anxiety disorders (17.2%); substance use disorders (8.6%); asthma (7.7%); and infections (6.7%) were the most prevalent conditions identified in this population. Multivariate Analyses After the adjustment of several covariates (patient residence, zip code income, primary payer, hospital location, hospital region, and hospital teaching status), higher ED charges were associated with year 2009, a higher number of diagnoses, older age, and being female (Table 3). Patients with alcohol use disorders, substance use disorders, fluid and electrolyte disorders, epilepsy, pain, fracture, meningitis, noninfectious gastroenteritis, and diabetes with complications attributed higher ED charges than those without key comorbidities. Suicide and intentional self-inflicted injury attempters had higher ED charges than nonattempters. While controlling for the same covariates, hospital admission from the ED increased by 1.47 (95% CI = 1.38 to 1.57) per diagnosis (Table 4). Of the relevant comorbidities, attention-deficit, conduct, and disruptive behavior disorders; alcohol use disorders; substance use disorders; eating disorders; fluid and electrolyte disorders; meningitis; diabetes with complications; cancer; inflammatory bowel disorder; overweight and obesity; and developmental disorders were associated with increased odds of hospital admission from the ED. Patients who attempted suicide and intentional selfinflicted injury were more likely to be admitted than patients who did not attempt. Covariate associations differed across age groups (all p < 0.01 after Bonferroni correction), with most factors identified among adolescents. Estimates for each age group are presented in their corresponding Tables 3 and 4. The multivariate results did not materially change in the sensitivity analyses (data not shown). In the overall sample, higher ED charges remained significantly associated with year 2009, older age, being female, key comorbidities, and suicide and intentional self-inflicted injury (all p < 0.05; Table 3). Similarly, increased odds of hospital admission from the ED remained significantly associated with a higher number of diagnoses, being female, key comorbidities, and suicide and intentional Table 2 Comorbidities Associated With All Listed Diagnoses of Pediatric Depression Stratified by Age Group, 2006 and 2009* Comorbidities Children (n = 945, 0 5 yr) Youths (n = 43,103, 6 12 yr) Adolescents (n = 321,665, yr) Overall (n = 365,713, 0 17 yr) Anxiety disorders 131 (13.9) 7,990 (18.5) 54,676 (17.0) 62,796 (17.2) Attention-deficit, conduct, 161 (17.0) 14,482 (33.6) 49,762 (15.5) 64,404 (17.6) and disruptive behavior disorders Bipolar disorder 34 (3.6) 2,713 (6.3) 16,279 (5.1) 19,026 (5.2) Alcohol use disorders 36 (3.8) 140 (0.3) 14,279 (4.4) 14,455 (4.0) Substance use disorders 44 (4.6) 363 (0.8) 31,221 (9.7) 31,628 (8.6) Eating disorders 6 (0.7) 139 (0.3) 2,405 (0.7) 2,550 (0.7) Asthma 61 (6.5) 4,132 (9.6) 24,051 (7.5) 28,244 (7.7) Infections 130 (13.7) 2,882 (6.7) 21,362 (6.6) 24,374 (6.7) Fluid and electrolyte disorders 66 (7.0) 591 (1.4) 7,238 (2.3) 7,895 (2.2) Epilepsy 65 (6.9) 758 (1.8) 5,720 (1.8) 6,544 (1.8) Pain 39 (4.1) 1,238 (2.9) 12,663 (3.9) 13,939 (3.8) Open wounds 28 (3.0) 1,162 (2.7) 16,302 (5.1) 17,492 (4.8) Fracture 14 (1.5) 540 (1.3) 3,162 (1.0) 3,716 (1.0) Meningitis 0 28 (0.1) 111 (0.0) 140 (0.0) Noninfectious gastroenteritis 30 (3.2) 191 (0.4) 1,157 (0.4) 1,379 (0.4) Fever of unknown origin 61 (6.5) 367 (0.9) 1,693 (0.5) 2,122 (0.6) Diabetes with complications (0.4) 1,719 (0.5) 1,884 (0.5) Cancer 14 (1.5) 265 (0.6) 1,108 (0.3) 1,387 (0.4) Inflammatory bowel disorder 0 5 (0.0) 132 (0.0) 137 (0.0) Overweight and obesity 15 (1.6) 837 (1.9) 6,013 (1.9) 6,864 (1.9) Developmental disorders 13 (1.4) 1,083 (2.5) 4,960 (1.5) 6,056 (1.7) Attempted suicide and intentional self-inflicted injury 129 (13.7) 9,654 (22.4) 104,991 (32.6) 114,774 (31.4) All data are reported as n (%). *Results are based on Healthcare Cost and Utilization Project Nationwide Emergency Department Sample weighted sampling of discharged children and adolescents aged younger than 18 years with all listed diagnoses of codes , 296.2X, 296.3X, 300.4, and 311 from the International Classification of Diseases, Ninth Revision, Clinical Modification. Percentages may not add to 100% due to rounding.

6 1008 Sun et al. ED VISITS IN THE UNITED STATES FOR PEDIATRIC DEPRESSION Table 3 Multivariate Analyses of ED Charges Associated With All Listed Diagnoses of Pediatric Depression Stratified by Age Group, 2006 and 2009* Children (n = 945, 0 5 yr) Youths (n = 43,103, 6 12 yr) Adolescents (n = 321,665, yr) Overall (n = 365,713, 0 17 yr) Characteristics exp ^b p-value 95% CI exp ^b p-value 95% CI exp ^b p-value 95% CI exp ^b p-value 95% CI Year (ref = 2006) < < < No. of diagnoses < < < Age (yr) < < < Sex (ref = male) female < < Comorbidities Anxiety disorders Attention-deficit, conduct, and disruptive behavior disorders < < Bipolar disorder Alcohol use disorders Substance use disorders Eating disorders Asthma < < Infections < < < Fluid and electrolyte disorders < < Epilepsy < < Pain < < < Open wounds < < Fracture < < < Meningitis Omitted Noninfectious gastroenteritis Fever of unknown origin Diabetes with complications Omitted Cancer Inflammatory bowel disorder Omitted Omitted Overweight and obesity < < Developmental disorders Attempted suicide and intentional < < < self-inflicted injury exp ^b = exponentiated beta coefficient. *Results are based on Healthcare Cost and Utilization Project Nationwide Emergency Department Sample weighted sampling of discharged children and adolescents aged younger than 18 years with all-listed diagnoses of codes , 296.2X, 296.3X, 300.4, and 311 from the International Classification of Diseases, Ninth Revision, Clinical Modification. All generalized linear models are controlled for patient s residence, zip code income, primary payer, hospital location, hospital region, and hospital teaching status. Omitted because strata do not contain subpopulation members. Statistically significant at p < Statistically significant at p < 0.01, after Bonferroni correction.

7 ACADEMIC EMERGENCY MEDICINE September 2014, Vol. 21, No Table 4 Multivariate Analyses of Hospital Admission From the ED Associated With All Listed Diagnoses of Pediatric Depression Stratified by Age Group, 2006 and 2009* Children (n = 945, 0 5 yr) Youths (n = 43,103, 6 12 yr) Adolescents (n = 321,665, yr) Overall (n = 365,713, 0 17 yr) Characteristics exp ^b p-value 95% CI exp ^b p-value 95% CI exp^b p-value 95% CI exp^b p-value 95% CI Year (ref = 2006) No. of diagnoses < < < < Age (yr) < < Sex (ref = male) female Comorbidities Anxiety disorders Attention-deficit, conduct, and disruptive behavior disorders Bipolar disorder Alcohol use disorders < < Substance use disorders < < Eating disorders < < < Asthma Infections < < < Fluid and electrolyte < < disorders Epilepsy Pain < < < < Open wounds < < < Fracture Meningitis Omitted < Noninfectious gastroenteritis Fever of unknown origin < < < Diabetes with Omitted < < complications Cancer < < Omitted < < Inflammatory bowel disorder 10 6 < Overweight and obesity < < < Developmental disorders < < Attempted suicide and < < < intentional self-inflicted injury exp ^b = exponentiated beta coefficient;. *Results are based on Healthcare Cost and Utilization Project Nationwide Emergency Department Sample weighted sampling of discharged children and adolescents aged younger than 18 years with all-listed diagnoses of codes , 296.2X, 296.3X, 300.4, and 311 from the International Classification of Diseases, Ninth Revision, Clinical Modification. All generalized linear models are controlled for patient s residence, zip code income, primary payer, hospital location, hospital region, and hospital teaching status. Omitted because strata do not contain subpopulation members. Statistically significant at p < Statistically significant at p < 0.01, after Bonferroni correction.

8 1010 Sun et al. ED VISITS IN THE UNITED STATES FOR PEDIATRIC DEPRESSION self-inflicted injury (all p < 0.05; Table 4). Slight differences were seen in the key comorbidities, but the associations with mental health problems remained significant. DISCUSSION This study provides nationally representative data on pediatric ED visits with all listed diagnoses of depression in the United States as obtained from the 2006 and 2009 NEDS database. We found that, although less than 0.1% of the visits died in ED, almost one-third (27.2%) were hospitalized, with more than half being treated and released, and the majority of the visits were made by adolescents (87.9%). The ED charges summed to a substantial hospital bill of $443.8 million nationwide, with the ED plus inpatient charges ($1.2 billion) being more than double that amount. The median inpatient LOS was 4.0 days, and the mean (5.2 days) was slightly higher. Suicide and intentional self-inflicted injury were a major issue in this population, attempted by 31.4% of the patients. Not surprisingly, the most common comorbidities in this population were mental health problems, specifically, attention-deficit, conduct, and disruptive disorders; anxiety disorders; and substance use disorders. Although certain variables were significantly associated with higher resource use, we urge caution in interpreting these results because they could be due to large sample sizes easily yielding statistically significant results that may be clinically difficult to reconcile. Results of the current study extend past research in several important ways. First, a previous study has shown that hospitalization of California youths with psychiatric diagnoses is largely predicted by clinical need. 39 Therefore, it is of great interest to examine hospitalization after ED visits and the predictors of those hospitalization decisions on a population level. Second, our analysis defines depression using the broader case definition of all listed diagnoses where previous studies considered only a predominant principal diagnosis, thereby capturing a more accurate representation of the condition in the ED. Third, research on pediatric depression-related visits to the ED is limited, particularly those studies that incorporate both economic and clinical outcomes. Finally, through a multivariate approach, potentially significant factor associations were explored using two key indicators of resource use: ED charges (Table 3) and hospital admission from the ED (Table 4). Over the past decade, with the significant increase in pediatric ED visits, concerns have been raised about EDs becoming sources of substitute care for mental health problems. 40,41 According to the National Ambulatory Medical Care Surveys, 5% of all U.S. ED visits (n = million) were made by children with mental health problems, and nearly 478,000 patients were admitted to the hospital. 42 In our study, we found a substantial number of ED patients with depression-related diagnoses to enter the ED (n = 365,713), and 27.2% were then hospitalized. Although our findings are not directly comparable to those of Grupp-Phelan et al., 42 it is in keeping with the notion that the national emergency care system is overloaded due to youths with mental health problems. With an ongoing shortage of pediatric mental health professionals, EDs will face new challenges in providing quality and accessible care under constrained resources. 43 To date, there are very little data on how many children with depression are missed in the emergency setting, probably because depression is particularly difficult to diagnose in this age group. 22,35,36 For example, children do not have the ability to express feelings of sadness in language, so they tend to have more temper tantrums, somatic complaints, hallucinations, and social withdrawal. 44,45 Adolescents, on the other hand, tend to have higher rates of social dysfunction (e.g., school absenteeism) and self-injurious behaviors (e.g., substance abuse, suicide attempts). 45,46 Because of the unique characteristics at each developmental stage, it can be difficult to distinguish these normal but intense emotional behaviors from depressive symptoms. Nonetheless, several works in the general population and other clinical settings have shed some light on this subject. In a cohort of 167 patients, Taggart et al. 47 compared diagnoses obtained from routine psychiatric assessment in emergency settings with those from a standardized diagnostic interview. They found only moderate agreement (Cohen s kappa = 0.47) between clinical assessment and standardized diagnoses, which suggests that certain disorders may be underdiagnosed at emergency evaluation. In primary care settings, Cassidy and Jellinek 48 noted that psychiatric disorders occurred in 14% to 20% of children and adolescents, yet fewer than one in five of these children were recognized. The most common psychiatric disorders presenting to pediatricians include attention-deficit hyperactive disorder, anxiety disorders, depression, substance use disorders, and conduct disorders. Experts from a consensus conference also argued that individuals with depression are being seriously undertreated. 49 Some reasons include failure to recognize the symptoms, poor professional school education about depression, and inadequate time to evaluate and treat depression. Given the complexity of the disorder and the potential for misdiagnosis, brief screening tools for mental health problems should be routinely used in the ED After adjusting for the multiple covariates, we found that a higher number of diagnoses, key comorbidities, and suicide and intentional self-inflicted injury are associated with an increased risk of hospital admission. These findings are not unexpected and are consistent with those of Huffman et al. 39 that hospitalization after ED visits among child psychiatric patients was largely a function of key clinical factors. Previous studies have also suggested that, in children and adolescents, behavioral problems, alcoholism, substance abuse, and eating disorders frequently co-occur in the presence of other psychiatric and personality disorders. 20,53 55 In our study, we were able to investigate these phenomena in a nationally representative sample. We also found that the predominant drivers of higher ED charges were year 2009, a higher number of diagnoses, older age, being female, key comorbidities, and suicide and intentional self-inflicted injury. Although earlier work has provided cost estimates of inpatient, 11 13,56 58 outpatient, 11,56,57 physician, 56,57 and ED services in

9 ACADEMIC EMERGENCY MEDICINE September 2014, Vol. 21, No pediatric patients with depression, none has identified the cost drivers of their ED visits. Merikangas et al. 8 provided a potential explanation for our findings in a national survey that examined the rates of sociodemographic correlates of lifetime mental health service use by severity type and number of DSM-IV disorders. They found that disorder severity was significantly associated with an increased likelihood of receiving treatment. 8 Comorbidity and severe impairment were also strongly associated with service utilization, particularly in youth with behavior disorders. Therefore, reducing depression-related comorbidities may decrease the number of ED visits and hospitalizations, which ultimately reduces the economic burden of depression in pediatric EDs. Patients with childhood depression represent a particularly high-risk group in the ED because they engage in life-threatening attempts, and this makes appropriate care and disposition decision challenging. According to the Centers for Disease Control and Prevention, suicide in the United States currently ranks as the third leading cause of death for 15- to 25-year-olds, accounting for 20% of all deaths in this age group annually. 59 Furthermore, about 90% of all youths who commit suicide had at least one prior psychiatric diagnosis, with substance use and mood disorders being the most common. 60 Of great concern is that the risk of suicidal behavior is often undetected due to a lack of routine screening. In one survey study, 328 physicians reported that one or more adolescent patients attempted suicide in the previous year, but only 23% frequently or always screened adolescent patients for suicide risk factors. 61 Furthermore, fewer than 50% of adolescents seen for suicidal behavior in the ED were ever referred for treatment and, even when they were referred, adherence to treatment was low. 62 Repeat attempters are particularly problematic because they are more likely to be diagnosed with psychiatric disorders, have a higher chance of being hospitalized, and are less compliant with outpatient follow-up. 63,64 Nonetheless, there are interventions that work to improve suicide-related outcomes. For example, ED visits followed up with interim psychiatric care were found to be associated with significant risk reduction (59%) in suicide-related hospitalizations (95% CI = 0.28 to 0.60). 65 Given that effective strategies are available, identifying risk factors of suicide behavior in people with depression should be a top priority in pediatric emergency care. Once patients in our study were stratified into three age groups, different associations emerged between the variables and outcomes. This finding is not unexpected given the developmental differences between age groups. 22,45,66 At the consensus conference on Mental Health Aspects of Emergency Medical Services for Children, Horowitz et al. 67 emphasized the importance of addressing children s mental health needs in the ED based on age and developmental stage. They recommended that ED staff members be educated in the various issues associated with each stage. Given the complexity of child and adolescent development, it is important that providers consider each age group separately and independently, as it may provide unique insights into patient outcomes. Also of note is the sensitivity analysis, in which we used tertiary diagnoses as a criterion for case definition. Although the overall conclusions from the analyses were consistent, it is evident that there are some minor differences in outcomes. For example, ED charges for visits with tertiary diagnoses summed to a hospital bill of $375 million (data not shown), which is not substantially different from those with all listed diagnoses ($443.8 million). However, ED plus inpatient charges with all listed diagnoses ($1.2 billion) nearly doubled compared to those with tertiary diagnoses ($716.8 million, data not shown). These findings indicate that results may differ depending on the case definition, and perhaps researchers should expand their criteria to more accurately assess the overall effect of pediatric depression-related visits in the ED. LIMITATIONS This study has several limitations that warrant comment. First, coding errors, omissions, and unmeasured confounders may exist, and the lack of validating clinical information, such as degree of impairment, may limit definitive conclusions regarding the appropriateness of clinical care. Second, given the large sample size, some multivariate estimates may have attained statistical significance without reaching clinical relevance. Third, because records were deidentified, we could not examine long-term outcomes, such as ED revisits or hospital readmissions. Fourth, direct admits to mental facilities were not captured in the database; therefore, we advise against generalizing beyond the study population. Fifth, charges may differ from the actual amount paid because of discounts, deductibles, copayments, and coinsurance. 68 Also not reported are expenditures associated with outpatient care and postdischarge care, indirect costs associated with seeking care or time lost from work, and societal costs. Sixth, while we adjusted for available demographic and other covariates in our models, unobserved factors (e.g., psychiatric service providers, psychiatric beds) might have contributed to the effect on outcomes. Finally, accuracy of the case definition used in this study for identifying depression-related ED visits is unknown; therefore, our results may underestimate or overestimate the true burden of pediatric depression in the ED. Despite these limitations, the present study provides valuable insights for evaluating depression-related ED visits in the pediatric population. CONCLUSIONS Pediatric depression presenting to the ED is common, and presents a considerable burden to the U.S. health care system. Certain factors, such as a higher number of diagnoses, key comorbidities, and suicide and intentional self-inflicted injury indicate higher health care resource use and costs. With the recent changes in and growing financial pressures on the health care system, physicians are in need of education regarding the risk factors of pediatric depression presenting to the ED. Special attention should be given to these measures, when present.

10 1012 Sun et al. ED VISITS IN THE UNITED STATES FOR PEDIATRIC DEPRESSION References 1. Grove DD, Lazebnik R, Petrack EM. Urban emergency department utilization by adolescents. Clin Pediatr 2000;39: Ma J, Wang Y, Stafford RS. U.S. adolescents receive suboptimal preventive counseling during ambulatory care. J Adolesc Health 2005;36: Merrill CT, Owens PL, Stocks C. Pediatric Emergency Department Visits in Community Hospitals From Selected States, Rockville, MD: Agency for Healthcare Research and Quality, Ziv A, Boulet JR, Slap GB. Emergency department utilization by adolescents in the United States. Pediatrics 1998;101: Elixhauser A. Hospital Stays for Children. Rockville, MD: Agency for Healthcare Research and Quality, Levine LJ, Schwarz DF, Argon J, Mandell DS, Feudtner C. Discharge disposition of adolescents admitted to medical hospitals after attempting suicide. Arch Pediatr Adolesc Med 2005;159: Mahajan P, Alpern ER, Grupp-Phelan J, et al. Epidemiology of psychiatric-related visits to emergency departments in a multicenter collaborative research pediatric network. Pediatr Emerg Care 2009;25: Merikangas KR, He JP, Burstein M, et al. Service utilization for lifetime mental disorders in U.S. adolescents: results of the National Comorbidity Survey-Adolescent Supplement (NCS-A). J Am Acad Child Adolesc Psychiatry 2011;50: Merikangas KR, He JP, Burstein M, et al. Lifetime prevalence of mental disorders in U.S. adolescents: results from the National Comorbidity Survey Replication-Adolescent Supplement (NCS-A). J Am Acad Child Adolesc Psychiatry 2010;49: Nock MK, Green JG, Hwang I, et al. Prevalence, correlates, and treatment of lifetime suicidal behavior among adolescents: results from the National Comorbidity Survey Replication Adolescent Supplement. JAMA Psychiatry 2013;70: Martin A, Leslie D. Psychiatric inpatient, outpatient, and medication utilization and costs among privately insured youths, Am J Psychiatry 2003;160: Sclar DA, Robison LM, Gavrun C, Skaer TL. Hospital length of stay for children and adolescents diagnosed with depression: is primary payer an influencing factor? Gen Hosp Psychiatry 2008;30: Lasky T, Krieger A, Elixhauser A, Vitiello B. Children s hospitalizations with a mood disorder diagnosis in general hospitals in the united states Child Adolesc Psychiatry Ment Health 2011;5: Agency for Healthcare Research and Quality. Introduction to the HCUP Nationwide Emergency Department Sample (NEDS): Rockville, MD: Agency for Healthcare Research and Quality, Agency for Healthcare Research and Quality. Introduction to the HCUP Nationwide Emergency Department Sample (NEDS): Rockville, MD: Agency for Healthcare Research and Quality, Healthcare Cost and Utilization Project. Overview of the Nationwide Emergency Department Sample (NEDS). Available at: nedsoverview.jsp. Accessed Jun 26, Healthcare Cost and Utilization Project. Special Study on the Meaning of the First-Listed Diagnosis on Emergency Department and Ambulatory Surgery Records. Rockville, MD: Agency for Healthcare Research and Quality, Birmaher B, Ryan ND, Williamson DE, Brent DA, Kaufman J. Childhood and adolescent depression: a review of the past 10 years. Part II. J Am Acad Child Adolesc Psychiatry 1996;35: Birmaher B, Ryan ND, Williamson DE, et al. Childhood and adolescent depression: a review of the past 10 years. Part I. J Am Acad Child Adolesc Psychiatry 1996;35: Baren JM, Mace SE, Hendry PL, Dietrich AM, Goldman RD, Warden CR. Children s mental health emergencies part 2: emergency department evaluation and treatment of children with mental health disorders. Pediatr Emerg Care 2008;24: McQuaid EL, Kopel SJ, Nassau JH. Behavioral adjustment in children with asthma: a meta-analysis. J Dev Behav Pediatr 2001;22: Son SE, Kirchner JT. Depression in children and adolescents. Am Fam Physician 2000;62: , Webb WL Jr, Gehi M. Electrolyte and fluid imbalance: neuropsychiatric manifestations. Psychosomatics 1981;22: Dunn DW, Austin JK, Huster GA. Symptoms of depression in adolescents with epilepsy. J Am Acad Child Adolesc Psychiatry 1999;38: Raymer D, Weininger O, Hamilton JR. Psychological problems in children with abdominal pain. Lancet 1984;1: Kellogg ND. Evaluation of suspected child physical abuse. Pediatrics 2007;119: McCracken GH Jr. Current management of bacterial meningitis in infants and children. Pediatr Infect Dis J 1992;11: Fitzgerald M, McGee HM. Psychological health status of mothers and the admission of children to hospital for gastroenteritis. Fam Pract 1990;7: Grey M, Whittemore R, Tamborlane W. Depression in type 1 diabetes in children: natural history and correlates. J Psychosom Res 2002;53: Mulhern RK, Fairclough DL, Smith B, Douglas SM. Maternal depression, assessment methods, and physical symptoms affect estimates of depressive symptomatology among children with cancer. J Pediatr Psychol 1992;17: Engstrom I. Mental health and psychological functioning in children and adolescents with inflammatory bowel disease: a comparison with children having other chronic illnesses and with healthy children. J Child Psychol Psychiatry 1992;33: Goodman E, Whitaker RC. A prospective study of the role of depression in the development and