Social workers in emergency departments (EDs) are part of an interdisciplinary medical team that provides critical services to patients with multiple general medical, psychiatric, social, and economic needs who do not access services elsewhere (
1–
5). It has been suggested that social workers in EDs can save valuable medical provider time (
2) and help patients enroll in insurance programs (
6). It has been demonstrated that the majority of patients seen by social workers in the ED are discharged home rather than admitted (
7).
EDs are increasingly serving patients with multifaceted needs. Patients with psychiatric disorders (
8) and substance use disorders (
9,
10) have higher rates of ED use than the general population. Uninsured patients with psychiatric disorders are also at risk of high ED utilization (
11). In 2007, visits related to mental health and substance use problems represented 12.5%—or one out of eight—ED visits in the United States (
12). In 2010, psychosis was one of the top 20 diagnostic groups for ED visits for men ages 15–65 (
13). Social workers are often referred the most complicated patients who require mental health, social, and referral services in order to be discharged (
7,
14,
15).
Patients treated for traumatic injuries also have many social work needs. Specifically, more than one-quarter of the 2.8 million intentional injuries treated in EDs were self-inflicted, and nearly one-third of the 2.7 million patients transferred from EDs to other hospitals were transferred to a psychiatric hospital (
13). Over 80% of the more than 1.3 million patients seen in EDs nationally for traumatic brain injury (TBI) were discharged home (
16). Patients with TBI have high rates of comorbid mental illness (
17–
19), and many would benefit from early intervention and identification by social workers. The U.S. National Hospital Ambulatory Medical Care Survey (NHAMCS) consistently documents high prevalence rates of ED patients with needs potentially amenable to social work services, including patients with mental or substance use disorders and patients who are homeless, poor, or unable to access outpatient services (
20). With training in mental health, substance abuse, and injury-related services, social workers can serve as frontline providers for many of these patients.
Despite these contributions, ED social work remains an understudied and underfunded area of emergency medical services (
4). There are currently no national practice standards or specific guidelines for ED social workers, and there are no national data repositories. More research is needed to understand social work’s impact on patient health outcomes, ED recidivism, and cost-effectiveness. This is increasingly important because the Affordable Care Act (ACA) calls for improved care coordination and integration of previously siloed service sectors, such as emergency medicine, outpatient psychiatry, and general medical care.
Using comprehensive data from a population of patients treated in the ED of a large, urban, level 1 trauma center, this investigation characterized patients served by ED social workers, categorized ED social work services, and determined predictors of length of ED stay. In addition, predictors of ED social work services were identified with the goal of developing social work referral and treatment algorithms. We hypothesized that social workers would provide services to patients with multiple comorbidities and recurrent visits and that social work contacts would add only marginally to overall length of stay.
Methods
Data Source and Study Sample
This was a cross-sectional study of all patients treated in the ED of a large, urban, academic, level 1 trauma center from January 1, 2012, to March 31, 2013 (N=49,354). Level 1 trauma centers are comprehensive care facilities able to provide leadership in injury care, conduct teaching and research, and provide 24-hour injury care (
21). Data were extracted from the electronic medical record (EMR) by using the Microsoft Amalga platform (
22). The University of Washington Institutional Review Board approved the study.
Demographic Characteristics, Discharge Disposition, and Visit Counts
Demographic characteristics available in the EMR included age, sex, primary language, and race-ethnicity. Discharge disposition was entered into the EMR by the treating ED provider. For the predictor models, discharge disposition was dichotomized as discharged to home or home with care versus nonhome discharge (including death, admitted, transfer, court or jail, against medical advice, skilled nursing facility, or other). A count variable was created documenting ED visits (including index) for each patient during the study period.
ICD-9-CM Diagnoses
General medical, mental, and substance use disorder diagnoses associated with the first ED visit in the study period were identified. We used the NHAMCS to identify the seven most common primary diagnoses at ED visits by major disease category (
13). Approximately 75% of all primary ED diagnoses are represented in these seven disease categories. In this investigation, we excluded the category “symptoms, signs and ill-defined conditions” because of the broad, ill-defined nature of this category. Medical diagnoses were identified by using the remaining six categories. The following
ICD-9-CM categories were included for medical diagnoses: all injury (codes 800–959), diseases of the respiratory system (460–519), diseases of the musculoskeletal system and connective tissue (710–739), diseases of the digestive system (520–579), diseases of the genitourinary system (580–629), and diseases of the nervous system and sense organs (320–389). These diagnostic categories were combined and included as a dichotomous variable for the analysis involving overall diagnostic category (general medical versus mental illness versus substance use disorder diagnoses). Injury (codes 800.0–801.9, 803.0–804.9, 850.0–854.1, and 959.01) was also analyzed separately because of its relevance to social work services. Injury etiology was further classified by using the Centers for Disease Control and Prevention–recommended framework of E-code groupings in the Web-based Injury Statistics Query and Reporting System (
23): unintentional injury (E800–E869 and E880–E929), self-inflicted injury (E950–E959), assault (E960–E969, E979, and E999.1), and undetermined (E980–E989).
A separate variable for mental disorders was created by combining anxiety and acute stress disorders, depressive disorders, and psychotic disorders. The prevalence of each major mental disorder is not reported in NHAMCS; therefore, we included major mental disorder diagnostic categories prevalent in the United States (
24) as well as those disorders that were listed as common reasons for ED visits among adults (
13). On the basis of our clinical experience, we excluded conditions rarely diagnosed in the ED (for example, hyperkinetic syndrome of childhood) and psychiatric diagnoses not typically treated by mental health professionals (for example, organic psychotic conditions). Specifically, the following categories were included: anxiety and acute stress disorders (including posttraumatic stress disorder, 309.81; acute stress disorders, 308.0–308.9; adjustment disorders, 309.0–309.9; panic disorder, 300.01, 300.21, and 300.22; phobia, 300.29; social anxiety, 300.23; obsessive-compulsive disorder, 300.3; generalized anxiety disorder, 300.02; and other anxiety, 293.84, 300.00, 300.09, 300.2, and 313), depressive disorders (including major depressive disorder, 296.2–296.99; dysthymia, 300.4; and other depressive disorders, 309.1 and 311), psychotic disorders (including schizophrenia, 295.0–295.99; bipolar disorders, 296.0; and psychosis not otherwise specified, 300.9).
To capture substance use disorders, we combined the following exhaustive diagnoses list: alcohol dependence (303.00–303.99), alcohol abuse (305.00–305.09), opioid dependence (304.00–304.01), opioid abuse (305.50–305.59), cocaine dependence (304.20–304.29), cocaine abuse (305.60–305.69), amphetamine dependence (304.40–304.49), amphetamine abuse (305.70–305.79), marijuana dependence (304.30–304.39), marijuana abuse (305.20–305.23), sedative abuse (305.4), polydrug dependence (304.70–304.99), polydrug abuse (305.90–305.99), and tobacco use (305.1).
Social Work Services
Social workers documented each patient encounter in the EMR. At the study site, there are at least two master’s-level social workers on duty providing social work services 24 hours a day, seven days a week. During the study period, 15 of 20 social workers were licensed clinical social workers. Social workers prioritize providing services to trauma patients, their families, assault victims, homeless individuals, those with substance use disorders, and minors. They receive referrals via pager and from staff directly. Social work services provided to patients presenting only in the self-contained psychiatric emergency services were not included because of the way that those notes are entered into the EMR. Each note was given a title by the social worker with a brief description of the service provided. Content analysis was used to categorize the note titles (
25). We utilized an inductive analytic approach whereby codes were developed from the data (
26). One investigator (MM) read and coded the first 500 note titles to develop the data-driven initial codebook. The coding team (MM, DD, and LH) then met and discussed codes that were included in the final codebook. Subsequent notes were coded by using the codebook, and as new codes arose they were added to the codebook through an iterative process (
27). In addition to weekly discussions of codes, formal intercoder agreement was periodically assessed. Notes were randomly selected by the first author, who assessed agreement between use of codes and coders. If discrepancies arose, the coding team discussed coding choices until 100% agreement was reached.
Statistical Analysis
All analyses were done by using each patient’s first visit to the ED in the study period. We initially assessed the demographic, diagnostic, discharge, and visit count information for the entire cohort. Then we compared the characteristics of patients who received social work services during their ED visit and those who did not. Chi-square tests were used to assess significant differences for categorical variables, and two-tailed t tests were used for continuous variables.
Next, we delineated social work services provided to patients in the cohort by using the developed coding scheme. Patients could receive more than one service during their visit. We then determined predictors of social work services by estimating relative risk (RRs) and 95% confidence intervals (CIs) for receiving social work services by using unadjusted Poisson regression with robust standard error variance to determine univariate associations (
28). All predictors from univariate analysis that were significant at p<.20 were entered into a single Poisson regression model to predict receipt of social work services. Next, backward elimination was used to delete variables from the model that were not significant at p<.05 in order to determine adjusted RRs and CIs. Covariates included all variables described in
Table 1. The same analytic process was used to assess predictors of length of stay. Receipt of social work services was the main covariate of interest in the length-of-stay analysis. Thus we decided a priori to keep this variable in the model regardless of its significance in the univariate analysis. All analyses were done with SAS 9.3 and Stata 10.
Results
During the study period, 49,354 patients were treated in the ED (
Table 1) and included in analysis. Social workers provided at least one service to 18,532 (38%) patients. Compared with their population percentage in the county, whites and Asians were underrepresented in the ED patient population (76% versus 53% and 14% versus 9%, respectively), and African Americans (6% versus 22%), Hispanics (7% versus 8%), and American Indians (1% versus 2%) were overrepresented (
29).
Persons with one or more mental disorder diagnoses and persons with a substance use disorder diagnosis were overrepresented among patients seen by social workers, compared with their representation in the ED patient population. Injuries (N=9,523, 51%) and comorbid disorders (32%) were common among patients seen by social workers, as was TBI (N=4,077, 22%). Many patients seen by social workers were discharged home (42%).
More than one-quarter (26%) of patients seen by social workers had two or more ED visits in the study period (including the index visit), and 61% had at least one inpatient stay in the study period.
Most of the social work services provided were mental health services (54%), 31% were care coordination services, and 15% were material support or other referral (
Table 2). In addition to providing mental health services to patients, social workers provided crisis counseling to family or friends of critically ill or incapacitated patients (counted in mental health services).
Females, patients not discharged home, and patients arriving by emergency medical services (EMS) had a higher likelihood of receiving social work services compared with males, those discharged home, and those not arriving by EMS (
Table 3). An increase in the number of comorbid diagnoses was associated with an increased likelihood of receiving social work services. All known injury etiologies were associated with an increased likelihood of receiving social work services, as was having a TBI. Nonwhites and non-English speakers had a lower likelihood of receiving social work services.
Non-English speakers, patients not discharged home, and those with comorbid diagnoses had a higher risk of a longer ED stay (
Table 4). When social work services were provided, patients had an increased risk of a longer stay.
Discussion
There is a paucity of research on ED social work services despite the important role of social workers on interdisciplinary teams (
4). Previous studies have been limited by small samples or by a focus on specific populations (
6) or interventions (
30,
31). This study provides a comprehensive description of social work services as well as a detailed characterization of patients seen by ED social workers and an analysis of the contribution of social work services to ED length of stay. Social workers provided services to nearly 40% of patients seen in a large, urban, level 1 trauma center ED. Many patients triaged to social work services had complex psychiatric and general medical presentations, multiple service needs, comorbid diagnoses, and multiple ED and inpatient stays.
The educational and professional goals of ED social workers align with sophisticated use of social worker time and go beyond provision of tangible resources or simple discharge planning (
32). Formal social work education extends social workers’ training beyond these tasks and equips social workers with biopsychosocial assessment skills, proficiency in brief psychosocial intervention, and clinical case management or care coordination skills. To date, limited data have been available to elucidate the role of social work in the ED or the extent to which social workers’ skills are utilized. This study identified the major social work services provided to patients in the ED: mental health services, care coordination, and material support and other referral services. The findings suggest an important role for social workers in this setting.
Probably because of illness complexity and severity and the related high level of patient need, as well as patient flow and prioritization in the trauma center ED, social workers were more likely to see patients who were not discharged home, arrived by EMS, and had comorbid diagnoses or an injury. Nonwhite patients and patients who were not English speakers were less likely to receive social work services. Racial disparities in ED medical care not related to social work have been noted (
33–
35). In the ED, it is likely that a combination of factors, rather than a lower level of need, contributes to the lower likelihood of receiving social work services, including inadequate and subjective referral and triage procedures to social work and staff difficulty using interpretation services.
The ED setting poses particular challenges for triage to social workers because of the fast-paced nature of the medical care and unpredictable arrival of new patients. Providers need easily accessible and rapidly applicable triage algorithms to consistently refer all patients with social work needs. Data from this study can be used as a guide for trauma centers to begin to develop these algorithms. Other factors to consider include examination of existing triage procedures and data elements and inclusion of priority populations to meet social work’s social justice mission. Universal screening methods for alcohol use disorder (
36) and methods that use EMRs to screen inpatients for psychiatric comorbidities and identify those at risk of posttraumatic stress disorder have been successful and would serve as useful models for ED social work (
37,
38).
Social workers require time to complete assessments, provide counseling and referrals, and organize complicated discharge plans for patients with multifaceted needs, including those with comorbid diagnoses, severe illness, and injury. Given the complex patient populations served and the types of services provided, a marginal increase in length of ED stay was expected. Other possible contributors to the increased length of stay include lack or limited availability of community resources.
With trained social workers in place, the ED can be a critical assessment and intervention point for the mental health needs of patients. With the passage of the ACA, we can expect at least initial increases in ED use. In an Oregon study, newly insured Medicaid patients used more ED services than uninsured patients (
39), and other studies have shown that ED visits continue to rise despite health care reform (
40). Insurance acquisition is only one factor affecting ED utilization (
39,
40). Therefore, the ED may be an important setting to assist newly insured patients under the ACA to access primary care and help uninsured patients sign up and gain access to care. Social workers’ care coordination role will be critical to achieving the ACA’s primary goals (
41). Care coordination in the ED will be particularly important for low-income individuals and individuals with mental disorders, who will likely continue to use the ED when primary care facilities are closed or difficult to access. Expanding social work capacity and roles in the ED may be crucial to support ACA mandates for coordinated care.
Although this study has provided valuable information about provision of social work services in the ED, it had a number of limitations related to using large administrative databases. For some patients, psychiatric diagnoses were assigned by ED medical providers rather than by mental health specialists. This may have affected the accuracy of some ICD-9-CM diagnostic categories, which could have resulted in low observed rates of psychiatric diagnosis. Also, several demographic and clinical characteristics, such as insurance status and income, were not assessed. Therefore, we could not determine the impact of unmeasured sociodemographic factors on receipt of social work services and length of stay. Also, this study was done at a single, urban, academic trauma center with a dedicated psychiatric emergency service and thus may not be generalizable to settings that are not trauma centers. Social work services provided to patients seen only in the psychiatric emergency service were not included in analyses because of the nature of the way those notes are entered, which likely resulted in an underestimation of number of mental health services provided. Finally, this study did not explore the effects of preexisting diagnoses or visit history on receipt of social work services or length of ED stay or cost-effectiveness of services. Future studies should incorporate these characteristics.
Conclusions
Social workers in trauma center EDs are providing mental health services, care coordination, and material support and other referral services to patients with multifaceted needs resulting from complex mental and substance use disorder diagnoses, trauma, and comorbid diagnoses. Women, patients not discharged home, those with comorbid diagnoses, and those with injuries had an increased likelihood of receiving social work services. Although social work services marginally increased length of ED stay, social workers provided multiple services to patients with complex presentations. To meet increasing demands of the ACA for care coordination, there may be a need to enhance and expand social work services in this critical intervention setting and develop triage algorithms to meet the social work needs of all patients. Future studies could examine the cost-benefits of providing social work services and the impact of these services on care coordination.