The incidence of diabetic ketoacidosis (DKA) increased during the COVID-19 pandemic but estimates from low-resource settings are limited. We examined the odds of DKA among emergency department (ED) visits in the Los Angeles County Department of Health Services (DHS) (1) during the COVID-19 pandemic compared to the pre-COVID era, (2) without active COVID infections, and (3) stratified by effect modifiers to identify impacted sub-groups. We estimated the odds of DKA from 400,187 ED visits pre-COVID era (March 2019–Feb 2020) and 320,920 ED visits during the COVID era (March 2020–Feb 2021). Our base model estimated the odds of DKA based on the COVID era. Additional specifications stratified by effect modifiers, controlled for confounders, and limited to visits without confirmed COVID-19 disease. After adjusting for triage acuity and interaction terms for upper respiratory infections and payor, the odds of DKA during the COVID era were 27% higher compared to the pre-COVID era (95%CI 14–41%, p < 0.001). In stratified analyses, visits with private payors had a 112% increased odds and visits with Medicaid had a 20% increased odds of DKA during the COVID era (95%CI 7–36%, p = 0.003). We identified increased odds of DKA during the COVID pandemic, robust to a variety of specifications. We found differential effects by the payor; with increased odds during COVID for privately-insured patients.

STUDY OBJECTIVE:To describe characteristics and outcomes of coronavirus disease (COVID-19) patients with new supplemental oxygen requirements discharged from a large public urban emergency department (ED) with supplemental oxygen.METHODS:This observational case series describes the characteristics and outcomes of 360 consecutive COVID-19 patients with new supplemental oxygen requirements discharged from a large urban public ED between April 2020 and March 2021 with supplemental oxygen. Primary outcomes included 30-day survival and 30-day survival without unscheduled inpatient admission. Demographic and clinical data were collected through a structured chart review.RESULTS:Among 360 patients with COVID-19 discharged from the ED with supplemental oxygen, 30-day survival was 97.5% (95% confidence interval (CI) 95.3 to 98.9%; n=351), and 30-day survival without unscheduled admission was 81.1% (95% CI 76.7 to 85.0%; n=292). A sensitivity analysis incorporating worst-case-scenario for 12 patients without complete follow-up 30 days after index visit yields 30-day survival of 95.5% (95% CI 92.5 to 97.2%; n=343), and 30-day survival without unscheduled admission of 78.9% (95% CI 74.3 to 83.0%; n=284). Among study patients, 32.2% (n=116) had a nadir ED oxygen saturation of <90%, among these 30-day survival was 97.4% (95% CI 92.6 to 99.4%; n=113), and 30-day survival without unscheduled admission was 76.7% (95% CI 68.8 to 84.1%; n=89).CONCLUSION:COVID-19 patients with new supplemental oxygen requirements discharged from the ED had survival comparable to COVID-19 ED patients with mild exertional hypoxia treated with supplemental oxygen in other settings, and this held true when the analysis was restricted to patients with nadir ED index visit oxygen saturations <90%. Discharge of select COVID-19 patients with supplemental oxygen from the ED may provide a viable alternative to hospitalization, particularly when inpatient capacity is limited.

3.26).Those with a ''severe'' PHQ-9 scores were 9 times (aOR 2.3-35.3)more likely to be daily drinkers. ConclusionsScreening and identification of people with depression and substance use disorders in the ED of a large national hospital in Kenya is feasible.This offers an opportunity for brief intervention and referral to further treatment.

Background:Emergency medicine (EM) physicians sometimes respond to critical events outside the emergency department. To prepare for these complex cases-typically called "rapid responses" (RRs)-EM residents receive simulation-based training involving four practice tasks and three exam tasks during a 1-day session. Cognitive load (CL) theory describes how humans function with limited working memories to perform complex tasks. RRs are expected to generate high levels of CL, but the profile of CL across providers and RR cases is not well understood. In this study, we analyzed resident's CL during RR training. We hypothesized variations in CL across individual and case and that exam cases would cause higher CLs than practice cases.Methods:Residents anonymously self-reported CL levels after each case using the Paas scale, a single-item, 9-point scale from "very, very low CL" to "very, very high CL." To examine case-based differences in CL, data were rescaled by individual residents. "High CL" was defined as a score of 9/9.Results:Among 18 residents participating, CLs ranged from 4 to 9, with median of 7 and interquartile range of 7-8. While many cases showed bell curve-like distributions of CLs, one case-a bleeding tracheostomy-showed a rightward skew reflecting higher levels of CL. No significant difference was found in CL between practice and exam cases. There were 20 reports (16.5%) of "high" CL with variation across residents (0/7 [0%] to 5/6 [83.3%] cases) and across cases (1/18 [5.6%) to 8/18 [44.4%]).Conclusions:The CL that EM residents experienced did show considerable interpersonal and intercase variation, but there was no significant difference between practice and exam cases. These results highlight several questions about how to optimally design future training, including how best to balance low and high CL training cases and which cases may require further training.

Introduction: To describe the impact of COVID-19 on a large, urban emergency department (ED) in Los Angeles, California, we sought to estimate the effect of the novel coronavirus 2019 (COVID-19) and "safer-at-home" declaration on ED visits, patient demographics, and diagnosis-mix compared to prior years. Methods: We used descriptive statistics to compare ED volume and rates of admission for patients presenting to the ED between January and early May of 2018, 2019, and 2020. Results: Immediately after California's "safer-at-home" declaration, ED utilization dropped by 11,000 visits (37%) compared to the same nine weeks in prior years. The drop affected patients regardless of acuity, demographics, or diagnosis. Reductions were observed in the number of patients reporting symptoms often associated with COVID-19 and all other complaints. After the declaration, higher acuity, older, male, Black, uninsured or non-Medicaid, publicly insured, accounted for a disproportionate share of utilization. Conclusion: We show an abrupt, discontinuous impact of COVID-19 on ED utilization with a slow return as safer-at-home orders have lifted. It is imperative to determine how this reduction will impact patient outcomes, disease control, and the health of the community in the medium and long terms.

Acute stroke is one of the leading causes of death in the United States, with an estimated annual cost of $34 billion and a death from stroke occurring approximately every 4 minutes.1 Early access to specialized stroke care is critically important in the treatment of suspected acute stroke, but access to dedicated stroke centers is unequal and varies geographically.2 Among the different levels of hospitals providing stroke care, the comprehensive stroke center (CSC) is the highest designation: CSCs utilize multidisciplinary stroke teams to provide the most advanced stroke care, including therapies like endovascular clot retrieval that are not available at most non-CSC hospitals.3 Some patients with acute strokes may benefit from transport by emergency medical services (EMS) directly to CSC—even if this transport requires bypassing a closer, non-CSC stroke hospital—but identifying which patients is an area of ongoing investigation. In many systems of stroke care, if transport to a CSC is expected to take less than a certain number of minutes, the patient is taken to the CSC even if a non-CSC hospital is closer.4-6 Transport time–based decisions like these, however, could be significantly impacted during times of heavy traffic: road congestion from traffic is a serious problem in major U.S. cities, with commuters losing on average 97 hours a year due to traffic, at a total cost of $87 billion.7 In Los Angeles County (LAC)—the most populous county in the United States—EMS routing protocols specify that patients with suspected large-vessel occlusion stroke should be transported directly to a CSC if the expected transport time is < 30 minutes.8 We hypothesized the interaction of a constant transport threshold with variable resistance to transport from traffic would result not only in areas which always or never have CSC access but also geographic “watershed” areas which only intermittently have CSC access within 30 minutes, depending on traffic conditions. We performed a prospective geospatial analysis of driving times to CSCs in LAC during various traffic conditions at the United States Census Block Groups (CBG) level. For each CBG, we identified the closest CSC and determined driving time from the centroid of the CBG to that CSC using the Google Maps Distance Matrix API with the “best-guess” parameter for traffic time estimation.9 For each route, the transit time was estimated 12 times during nonholiday weekdays over the course of 2 weeks in early December 2018, including three times each during morning and evening “rush hours” (06:00–09:30 and 14:30–19:00, respectively), midday between rush hours, and after the evening rush hour.10 Based on the ensemble of transit times, we defined each CBG as always having CSC access if travel times were consistently below 30 minutes, never having CSC access if travel times were consistently above 30 minutes, or only intermittently having CSC access if travel times included at least one trip above and one trip below 30 miuntes. To account for potential time savings of ambulances using lights and sirens, we performed a sensitivity analysis using a conservative 10-minute buffer, which raised the transport cutoff to 40 minutes. To examine population-level differences between CBGs by their access status, we tested whether demographic factors differed significantly between CBGs that always, never, or only intermittently met the 30-minute transit time threshold. This study was deemed exempt from institutional review board (IRB) approval by the IRB at the Keck School of Medicine of USC. Among the 6,415 CBGs in LAC, transport time was successfully estimated 12 times for 6,413 (99.97%) CBGs and 11 times for the remaining two. Transport time ranged from less than 1 minute to over 131 minutes, with a median time of 14.3 minutes (IQR = 9.6 to 20.6 minutes). Figure S1 in Data Supplement S1 (available as supporting information in the online version of this paper, which is available at http://onlinelibrary.wiley.com/doi/10.1111/acem.13909/full) shows the distributions of transport times for all 12 repetitions. Collectively, 5,036 (78.5%) CBGs with a combined 2010 population of 7,585,283 (76.7%) always had CSC access within 30 minutes, 225 (3.5%) with a combined population of 479,352 (4.9%) never had access, and 1,154 (18.0%) with a combined population of 1,809,874 (18.4%) only intermittently had CSC access. Figure 1 shows a choropleth map of the CBGs in LAC colored by their access to a CSC within 30 minutes of transport time and shows that areas that intermittently have access exist on the boundaries between always and never groups, but also as distinct groups within the densely populated center of LAC; these areas include notably a large area in South-Central Los Angeles with a population of over 1.13 million inhabitants that is located in between several CSCs. In the sensitivity analysis simulating potential transport with lights and sirens, 774 (67.1%) CBGs went from intermittently to always having access and 39 (17.3%) went from never to intermittently having access. Figure S2 in Data Supplement S1 shows the resulting choropleth map of CSC access and shows that the core of the intermittent access group in South-Central Los Angeles continues to exist. Figure S3 in Data Supplement S1 shows the proportions of CBGs that intermittently, always, or never have CSC access assuming a range of transport cutoffs between 5 and 90 minutes. Demographic comparisons by CSC access showed that intermittent access CBGs have the largest Hispanic population share (59.3 compared to 44.6 and 39.5 for the always and never access groups, p < 0.001), the largest black population share (16.7 compared to 7.0 and 11.7, p < 0.001), the largest share with less than a high school diploma (32.0 compared to 21.1 and 19.0, p < 0.001), and the largest share living below the poverty line (12.2 compared to 9.4 and 9.5, p < 0.001). The sizeable differences in the demographic makeup of the intermittent access group compared to the always and never groups supports their treatment as a distinct population with the potential for distinct health needs. Tables S1–S5 in Data Supplement S1 show the results of the full demographic analyses considering both models involving two states of access (either always or intermittently vs. never having access) or three states of access, with and without the consideration of a 10-minute buffer. Taken together, these results suggest that CSC access within a set time threshold is neither stable nor spatially equal, leading to three distinct classes of areas that always, never, or only intermittently have access to CSCs. To our knowledge this is the first study that shows that the existence of a distinct group of areas whose access to a CSC within a time threshold is intermittent and dependent on traffic.2 Additionally, we identified significant demographic differences between areas with different types of CSC access, suggesting that areas with limited CSC access may also have significant socioeconomic disadvantage. Within LAC, the most unexpected and novel finding of our analysis was the identification of groups of areas with intermittent CSC access located within the urban core of the city of Los Angeles. These areas likely exist because of a combination of inner-city traffic and the lack of CSCs in the socioeconomically disadvantaged neighborhoods of South Central and East Los Angeles. If policy makers wanted to invest in improving access to CSCs, the logical place to focus might be the northern areas. due to their relative geographic isolation. However, while the combined population of the areas that never have CSC access is ~479,000 inhabitants, the population of the large South Central area with intermittent 30-minutes access is over twice as large at ~1.14 million inhabitants. Given these population differences, development might be better prioritized within urban Los Angeles, despite higher concentrations of CSCs in this area. Beyond LAC, our findings demonstrate the value of incorporating data on traffic patterns and treating access to a CSC as a three-state model (always, never, or intermittently having access) as opposed to a two-state model (having or not having access) when discussing the most effective and equitable ways to invest scarce stroke care resources. The impact of traffic is salient across most urban settings in the United States, and the majority of the most populated counties in the United States are, like LAC, mixes of high-density urban areas with more rural areas that collectively have heavy traffic burdens.7 This study did have several limitations including the use of predicted travel times as opposed to measured travel times. Analysis of actual EMS transport times to CSCs from different areas would be useful in validating the conclusions we reach here. Additionally, our estimates of predicted travel time were obtained within a single 2-week period so it is possible that small variations in travel times outside of the study period might alter these results. However, data from the California Department of Transportation suggest that traffic patterns and travel times remained very stable over multiple preceding quarters so we believe that the possibility of this is quite low.10 Finally, we did not incorporate models of air ambulance transport or transport to CSCs outside of LAC. Air ambulances could increase CSC access in the more geographically isolated areas that currently never have CSC access but the utility of air ambulances in a congested urban environment is likely very limited. Two neighboring counties also have CSCs, but the effects of including these out-of-county CSCs would likely be minimal and confined to the outer edges of LAC. In summary, we found that groups of areas that only intermittently have access to CSCs depending on traffic do exist in a major U.S. county—including a highly unexpected region within an urban area—and that areas with decreased CSC access also were more socioeconomically disadvantaged. These findings suggest a potential new public health planning tool to inform capital-intensive decisions, such as where to build a new CSC, that weighs not only the importance of distance from a CSC but also barriers due to traffic congestion to help optimize the distribution of a region’s stroke resources. Data Supplement S1. Supplemental material. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Objective: Opioid use and the risk of opioid overdose are growing public health concerns for college-aged adults. Naloxone can temporarily reverse opioid overdoses, but only if easily accessible. On most college campuses, "blue light" phones (BLPs)-call boxes topped with a blue light-offer visible access to emergency services. We hypothesized that BLPs would provide potential naloxone access points. Participants: A major university campus in Los Angeles, CA. Methods: BLP locations were obtained using Google Maps, and the area of campus within a set distance to each BLP calculated. To model effects of loss or diversion, we simulated the random loss of various BLPs. Results: Placing naloxone kits at the 59 BLP locations could provide access within 100 m to 91.5% of the campus. With loss of half of the BLPs, campus access remained above 70%. Conclusions: Naloxone at BLP locations could be accessed from almost all campus areas.

INTRODUCTION:Detroit, Michigan, is among the leading United States cities for per-capita homicide and violent crime. Hospital- and community-based intervention programs could decrease the rate of violent-crime related injury but require a detailed understanding of the locations of violence in the community to be most effective.METHODS:We performed a retrospective geospatial analysis of all violent crimes reported within the city of Detroit from 2009-2015 comparing locations of crimes to locations of major hospitals. We calculated distances between violent crimes and trauma centers, and applied summary spatial statistics.RESULTS:Approximately 1.1 million crimes occurred in Detroit during the study period, including approximately 200,000 violent crimes. The distance between the majority of violent crimes and hospitals was less than five kilometers (3.1 miles). Among violent crimes, the closest hospital was an outlying Level II trauma center 60% of the time.CONCLUSION:Violent crimes in Detroit occur throughout the city, often closest to a Level II trauma center. Understanding geospatial components of violence relative to trauma center resources is important for effective implementation of hospital- and community-based interventions and targeted allocation of resources.

Background: Trauma is a major cause of death and disability in the United States, and significant disparities exist in access to care, especially in non-urban settings. From 2007 to 2017 New Mexico expanded its trauma system by focusing on building capacity at the hospital level. Methods: We conducted a geospatial analysis at the census block level of access to a trauma center in New Mexico within 1 h by ground or air transportation for the years 2007 and 2017. We then examined the characteristics of the population with access to care. A multiple logistic regression model assessed for remaining disparities in access to trauma centers in 2017. Results: The proportion of the population in New Mexico with access to a trauma center within 1 h increased from 73.8% in 2007 to 94.8% in 2017. The largest increases in access to trauma care within 1 h were found among American Indian/Alaska Native populations (AI/AN) (35.2%) and people living in suburban areas (62.9%). In 2017, the most rural communities (aOR 58.0), communities on an AI/AN reservation (aOR 25.6), communities with a high proportion of Hispanic/Latino persons (aOR 8.4), and a high proportion of elderly persons (aOR 3.2) were more likely to lack access to a trauma center within 1 h. Conclusion: The New Mexico trauma system expansion significantly increased access to trauma care within 1 h for most of New Mexico, but some notable disparities remain. Barriers persist for very rural parts of the state and for its sizable American Indian community. (C) 2019 Elsevier Inc. All rights reserved.

Among adolescents and young adults, opioid use disorder and opioid overdose represent a public health crisis. Naloxone kits can be used by non-medical bystanders to reverse opioid overdoses, but only if they know where to find the kits when needed. Therefore, optimal deployment of these lifesaving kits requires visibility and ease of access in highly populated areas ideally in a location already associated with an emergency response. On most college campuses, networks of "blue light" phones (BLPs)—emergency call boxes topped by a blue light—are designed to offer visible points of access to emergency services. We hypothesized that positioning naloxone kits at existing BLP locations would provide excellent spatial coverage on a college campus. We performed a geospatial analysis of accessibility to BLPs on a single college campus in Los Angeles, CA, as a locale for the potential co-location of naloxone kits. BLP locations were obtained using Google Maps and the iPhone "Mark My Location" feature. Buffers of various radii were created around each BLP, and the percent of campus covered was calculated. To model coverage under conditions of potential loss or diversion, we calculated campus coverage after simulating the random loss of over 25% of kits (n=1000 simulations). The relative contributions of each BLP location to overall access was assessed and a subgroup analysis was performed to consider potential access to naloxone kits with optimized smaller sets of BLPs. This study was deemed exempt from institutional review board review. We identified and mapped 58 BLPs on campus; over 75% of the campus was within 70m of a BLP, and over 99% was within 149m. Assuming an access range of 100m, positioning lifesaving kits at these BLPs would provide access to 91.5% of the campus, with complete coverage of on-campus residence halls, dining areas, and main walkways. (Figure 1) In simulations which randomly removed 15 (25.9%, 43 remaining) BLPs, median coverage at 100m was 85.6% of campus (IQR 84.2-86.8%). Subgroup analysis identified a key set of 26 BLP locations which could still provide access within 100m for 85.2% of the campus. Making use of a well-known resource already used in times of emergencies, networks of existing BLPs could be used to position naloxone kits, with only 26 kits needed to ensure 85% coverage on this college campus within a 100m radius. Access to naloxone kits deployed at all points of the BLP network would be resistant to loss or diversion with approximately 85% coverage maintained with loss of 25% of kits. Further work is needed to incorporate physical accessibility of BLP locations and explore BLP networks to co-locate naloxone kits on other college campuses.

Study objective: The effect of urgent cares on local emergency department (ED) patient volumes is presently unknown. In this paper, we aimed to assess the change in low-acuity ED utilization at 2 academic medical centers in relation to patient proximity to an affiliated urgent care. Methods: We created a geospatial database of ED visits occurring between April 2016 and March 2018 to 2 academic medical centers in an integrated health care system, geocoded by patient home address. We used logistic regression to characterize the relationship between the likelihood of patients visiting the ED for a low-acuity condition, based on ED discharge diagnosis, and urgent care center proximity, defined as living within 1 mile of an open urgent care center, for each of the academic medical centers in the system, adjusting for spatial, temporal, and patient factors. Results: We identified a statistically significant reduction in the likelihood of ED visits for low-acuity conditions by patients living within 1 mile of an urgent care center at 1 of the 2 academic medical centers, with an adjusted odds ratio of 0.87 (95% confidence interval 0.78 to 0.98). There was, however, no statistically significant reduction at the other affiliated academic medical center. Further analysis showed a statistically significant temporal relationship between time since urgent care center opening and likelihood of a low-acuity ED visit, with approximately a 1% decrease in the odds of a low-acuity visit for every month that the proximal urgent care center was open (odds ratio 0.99; 95% confidence interval 0.985 to 0.997). Conclusion: Although further research is needed to assess the factors driving urgent care centers' variable influence on low-acuity ED use, these findings suggest that in similar settings urgent care center development may be an effective strategy for health systems hoping to decrease ED utilization for low-acuity conditions at academic medical centers.

Objective: Identifying communities at high risk of stroke is an important step in improving systems of stroke care. Stroke is known to show spatial clustering at the state and county levels, but it is not known if clusters are present within city boundaries. Methods: We performed a geospatial analysis of the prevalence of stroke within 500 major cities in the United States using the Centers for Disease Control and Prevention 500 Cities Project. For each city, we calculated the Moran's I statistic, which looks for evidence of spatial clustering, and used Monte Carlo simulation to assess for clustering significance. Results: The mean overall crude prevalence of self-reported history of stroke at the city level was 2.8% (IQR 2.4-3.2%). Monte Carlo simulations of spatial patterns of stroke were successfully performed for 497 cities, of which 136 (27.3%) showed significant spatial clustering at the neighborhood level. All nine cities with more than one million inhabitants in 2010 showed significant spatial clustering. Conclusions: This is the first study to demonstrate that stroke shows clustering at the neighborhood level within many major cities in the United States and within all of the largest cities. Understanding where stroke clusters exist within cities can form the basis of optimizing emergency medical services deployment and improving systems of stroke care. (c) 2019 Elsevier Inc. All rights reserved.

Abstract Objective: Pre-stationing naloxone, a competitive antagonist that can reverse the effects of opioid overdose, in public spaces may expedite antidote delivery. Our study aimed to determine the feasibility of bystander-assisted overdose treatment using pre-stationed naloxone. Methods: Convenience sample of bystanders in Cambridge, Massachusetts in April 2017. Subjects assisted a simulated patient described as unconscious. Subjects interacted with simulated EMS dispatch to locate a nearby box, unlock it, and administer naloxone. Results: Fifty participants completed the simulation. Median time from simulated ambulance dispatch to naloxone administration was 189 seconds, and from arrival at patient side to administration 61 seconds. All but one participant (98.0%) correctly administered naloxone. Subjects' comfort with administration and willingness to provide medical care increased from before to after the trial. Comfort in administering naloxone varied significantly with level of previous training prior to, but not following, study participation. Conclusions: Bystanders are willing and able to access pre-stationed naloxone and administer it to a simulated patient in a public space. Public access naloxone stations may be a useful tool to reduce time to naloxone administration, particularly in areas where opioid overdoses are clustered.

INTRODUCTION:The epidemic of opioid use disorder and opioid overdose carries extensive morbidity and mortality and necessitates a multi-pronged, community-level response. Bystander administration of the opioid overdose antidote naloxone is effective, but it is not universally available and requires consistent effort on the part of citizens to proactively carry naloxone. An alternate approach would be to position naloxone kits where they are most needed in a community, in a manner analogous to automated external defibrillators. We hypothesized that opioid overdoses would show geospatial clustering within a community, leading to potential target sites for such publicly deployed naloxone (PDN).METHODS:We performed a retrospective chart review of 700 emergency medical service (EMS) runs that involved opioid overdose or naloxone administration in Cambridge, Massachusetts, between October 16, 2016 and May 10, 2017. We used geospatial analysis to examine for clustering in general, and to identify specific clusters amenable to PDN sites.RESULTS:Opioid-related emergency medical services (EMS) runs in Cambridge, Massachusetts (MA), exhibit significant geospatial clustering, and we identified three clusters of opioid-related EMS runs in Cambridge, MA, with distinct characteristics. Models of PDN sites at these clusters show that approximately 40% of all opioid-related EMS runs in Cambridge, MA, would be accessible within 200 meters of PDN sites placed at cluster centroids.CONCLUSION:Identifying clusters of opioid-related EMS runs within a community may help to improve community coverage of naloxone, and strongly suggests that PDN could be a useful adjunct to bystander-administered naloxone in stemming the tide of opioid-related death.

We developed a nontargeted diabetes screening program in a rural Indian Health Service emergency department in Shiprock, New Mexico to measure the proportion of previously undiagnosed diabetes and prediabetes, and to assess glycemic control among patients with known disease. Of 924 patients screened in the emergency department between May and July 2017, 28.8% screened positive for previously undiagnosed diabetes or prediabetes; among patients with known disease, the median hemoglobin A1c was 8.2%. Of the newly identified patients, 54.9% attended follow-up.

Community Action Programs Inter-City, Inc. (CAPIC) is a private, non-profit corporation founded in 1967 to eradicate the root causes of poverty and support families and individuals with complex social needs in the greater Boston area, specifically in the communities of Chelsea, Revere and Winthrop. The emergency department (ED) at the Massachusetts General Hospital (MGH) also serves these communities, and formed a collaborative academic-community partnership with CAPIC to provide analysis of ED visit data to help develop targeted, evidence-based interventions for a variety of health needs.

The opioid epidemic in the United States carries significant morbidity and mortality and requires a coordinated response among emergency providers, outpatient providers, public health departments, and communities. Anecdotally, providers across the spectrum of care at Massachusetts General Hospital (MGH) in Boston, MA have noticed that Charlestown, a community in northeast Boston, has been particularly impacted by the opioid epidemic and needs both emergency and longer-term resources. We hypothesized that geospatial analysis of the home addresses of patients presenting to the MGH emergency department (ED) with opioid-related emergencies might identify "hot spots" of opioid-related healthcare needs within Charlestown that could then be targeted for further investigation and resource deployment. Here, we present a geospatial analysis at the United States census tract level of the home addresses of all patients who presented to the MGH ED for opioid-related emergency visits between 7/1/2012 and 6/30/2015, including 191 visits from 100 addresses in Charlestown, MA. Among the six census tracts that comprise Charlestown, we find a 9.5-fold difference in opioid-related ED visits, with 45% of all opioid-related visits from Charlestown originating in tract 040401. The signal from this census tract remains strong after adjusting for population differences between census tracts, and while this tract is one of the higher utilizing census tracts in Charlestown of the MGH ED for all cause visits, it also has a 2.9-fold higher rate of opioid-related visits than the remainder of Charlestown. Identifying this hot spot of opioid-related emergency needs within Charlestown may help re-distribute existing resources efficiently, empower community and ED-based physicians to advocate for their patients, and serve as a catalyst for partnerships between MGH and local community groups. More broadly, this analysis demonstrates that EDs can use geospatial analysis to address the emergency and longer-term health needs of the communities they are designed to serve.

Study Objectives: Emergency department (ED) patients are disproportionately impacted by poverty and other social stressors.Research in selected populations has found food insecurity to be an important social determinant of health.While studies have shown that ED patients are more likely to be food insecure than the general population, little research has examined the relationship between food insecurity and frequency of ED use.Methods: We surveyed a random sample of ED patients at an urban, public hospital from November 2016-April 2017.Eligible patients were: 18 years old, clinically stable, not arrested or incarcerated, spoke English or Spanish and had not already participated in the survey.RAs verbally administered a 20-40 minute survey covering a wide range of health-related topics.Frequent ED use was defined as selfreport of 4 ED visits in the past 12 months, including the current visit.Food insecurity was defined as responding positively to any of four food insecurity questions from the U.S. Department of Agriculture Adult Food Security Module: running out of food before getting money to buy more, food not lasting until having money to buy more, being unable to afford balanced meals, and eating less than they felt they should have due to financial concerns, all in the past 12 months.We performed statistical testing for bivariate relationships between food insecurity and frequent ED use with chi-square and Kruskal-Wallis tests.Multivariable logistic regression controlling for age, race/ethnicity, sex, and self-reported overall health was conducted to better assess the effect of food insecurity on frequent ED use.Results: 1157 of 1412 eligible patients participated (81.9%).Mean age was 48 years, 41.7% were female, 53.5% were Hispanic/Latino, 22.1% were white, and 29.0% were black.One-third (31.8%) reported frequent ED use and 51.1% reported food insecurity.Rates of food insecurity were higher among frequent vs. non-frequent ED users, 60.5% vs. 46.7%(p<.001).Differences by question were: 47.1% vs. 35.4% worried about food running out (p<.001), 46.5% vs. 30.4% food not lasting (p<.001), 49.6% vs. 32.9%being unable to afford balanced meals (p<.001), and 38.2% vs. 22.6% eating less than they felt they should (p<.001).In multivariable logistic regression analyses, food insecurity continued to be a significant predictor of frequent ED use.Conclusions: ED patients in this study had high rates of food insecurity.Food insecurity was significantly associated with frequent ED use.Future research will include analysis of the pathways through which food insecurity may be related to frequent ED use and will examine whether the relationship between food insecurity and ED use is stronger for patients with certain "food-sensitive" illnesses (eg, diabetes).These early findings suggest that food insecurity may be important to consider in studies of and interventions for frequent ED users.