The Role of Social and Clinical Determinants in the Frequent Utilization of Emergency Departments Open Access
Downloadable ContentDownload PDF
Objective: Patients who utilize Emergency Department (ED) services at least four times per year, also referred to as frequent utilizers of EDs (FUEDs) comprise a high-cost-high-need vulnerable group of patients. Persistent ED users, people who use the ED 4 or more times for two consecutive years are a subset of FUEDs with the greatest needs and poorest health outcomes. FUEDs are less than 10% of all ED patients but account for a quarter of ED visits. Much of the research to date has focused on the clinical conditions and high use of health care services of FUEDs, much less is known about the impact of social determinants of health (SDH) on FUEDs. In this dissertation, I use the WHO CSDH framework to examine the relationship of social and clinical determinants and FUEDs’ unmet social and clinical needs, total and preventable ED visits, and persistent ED use. Study Design: Analysis of claims data merged with interview data from George Washington University Hospital (GWUH) Frequent User Study was used in a retrospective cohort study design. I used multivariate linear, negative binomial and logistic regression respectively to model the three main study outcomes: (1) number of unmet needs; (2) total and preventable ED visits; and (3) persistent ED use, as a function of structural SDH (education, employment, income), intermediary SDH (homelessness, food insecurity, social support) and clinical determinants (access to care, clinical illness severity, care continuity, behavioral health status) of health.Population Studied: 474 DC Medicaid beneficiaries, 18-64 years, who attended GWUH between October 2015 – 2016 and had 4 or more ED visits in a calendar year.Principal Findings: In Aim 1, 24 unmet needs grouped into 4 unique domains based on a factor analysis: material resources, health care management and literacy, employment and family-related needs. FUEDs experienced an average of 5.7 unmet needs and had a high burden of mental illness (60%). 80% reported material resource needs, such as housing, food, and transportation. 56% reported health care management and literacy needs (HCML) including dental (30%), medical care needs (22%) and psychiatric needs (20%). 45% reported employment related needs. When the number of unmet needs was modeled as a function of social (SDH) and clinical determinants of health, SDH contributed to 37% of the variance in the number of unmet needs while clinical determinants contributed less (18%). Poor access to food and shelter (β 3.12, CI 1.78,4.45), and poor behavioral health status (β 1.58, CI 0.49, 2.66) were associated with increased number of unmet needs. Social support (β -0.88, CI -1.24, -0.52) and higher income levels (β -1.92, CI -3.22, -0.63) were associated with decreased number of unmet needs. Aim 2 revealed that in one year, FUEDs had an average of 16.2 total ED visits and 30% of these ED visits were preventable. 60% of FUEDs had a diagnosed mental illness but there was significant underutilization of psychiatric services (mean 1.2, SD 4.4). Multivariate analyses revealed that increased physical illness severity, poor behavioral health status, inadequate food and shelter were associated with an increased number of total and preventable ED visits. For example, a one unit increase in the level of social support was associated with a 6% decrease in the number of total (IRR 0.94, CI 0.88, 1.01) ED visits and 14% decrease in preventable (IRR 0.86, CI 0.77, 0.96) ED visits. Physical illness severity and care continuity had a significant positive interaction effect upon both total (IRR 0.90, CI 0.85 0.96) and preventable ED utilization (IRR 0.89, CI 0.81, 0.97). Aim 3 showed that persistent FUEDs are sicker than non-persistent FUEDs. Greater duration of unemployment (OR 1.83) and cumulative homelessness (OR 2.03) were significantly associated with persistent ED use, however adjusting for access to public assistance renders their effects non-significant. Public assistance is strongly and positively associated with increased odds of persistent ED use (OR 2.24), and may buffer the influence of cumulative homelessness and severe behavioral illness on the odds of persistent ED use. Greater physical illness severity is strongly associated with persistent ED use (OR 3.11). Conclusion: Frequent ED use will not change substantially without addressing SDH such as low social support and lack of adequate food and housing that significantly contribute to the unmet needs, frequent and persistent ED utilization. Behavioral health is a significant clinical determinant of unmet needs and total and preventable ED utilization among FUEDs. Clinical determinants such as increased clinical illness severity are significant drivers of persistent ED use, with social determinants having lesser influence. Policy Implications: This research is directly applicable to DC Medicaid Care Coordination initiatives whereby patients with multiple chronic illnesses can receive a care coordination benefit. Some of the clinical problems cannot be coordinated effectively in this high need vulnerable population unless social adversities are addressed. Curbing ED use by FUEDs will require a two-pronged approach: adequate and coordinated ambulatory care for those most at risk as well as attention to specific social determinants they experience. Data from this research can inform state based Accountable Health Community and Health Home initiatives. FUEDs will benefit from tailored interventions that address food and housing needs and provide social support, especially in a patient population that has a high burden of mental illness. A data infrastructure to assess a comprehensive array of social needs alongside clinical needs should be incorporated at EDs and primary care sites for such high-need high-cost patients. Information from such a data infrastructure can also be used to increase alignment of Medicaid services with other safety net agencies to enable resource allocation efficiencies. This information can also be used to better prioritize public resources.