All discharges (N = 5 032 254) were included in the analysis. ED indicates emergency department.
Emergency department (ED) use by diagnosis related group ranged from 22.4 visits/1000 discharges to 282.5 visits/1000 discharges. Readmission rates ranged from 7.6 (95% CI, 7.4-7.9) readmissions/1000 discharges to 875.7 (95% CI, 826.6-927.1) readmissions/1000 discharges. Eight data points are outside the limits of the x-axis.
The analysis was conducted on the 3 highest-volume medical conditions for both groups. ED indicates emergency department.
The analysis was conducted on the 3 highest-volume surgical conditions for both groups. ED indicates emergency department.
Vashi AA, Fox JP, Carr BG, et al. Use of hospital-based acute care among patients recently discharged from the hospital. JAMA. doi:10.1001/jama.2012.216219
eSupplement. Methods, results, and conclusion
eTable 1. Hospital-based, acute care utilization (emergency department use or readmission) within 30 days of index hospital discharge, according to the highest volume discharge diagnoses
eTable 2. Discharge diagnosis associated emergency department visits within 30 days of index hospital discharge
Customize your JAMA Network experience by selecting one or more topics from the list below.
Vashi AA, Fox JP, Carr BG, et al. Use of Hospital-Based Acute Care Among Patients Recently Discharged From the Hospital. JAMA. 2013;309(4):364–371. doi:10.1001/jama.2012.216219
Author Affiliations: Robert Wood Johnson Foundation Clinical Scholars Program (Drs Vashi, Ross, and Gross), Department of Emergency Medicine (Drs Vashi and D’Onofrio), Section of General Internal Medicine (Drs Ross and Gross), and Cancer Outcomes Policy and Effectiveness Research Center (Dr Gross), Yale University School of Medicine, and Yale Comprehensive Cancer Center (Dr Gross), New Haven, Connecticut; Department of Veterans Affairs/VA Connecticut Healthcare System, West Haven (Dr Vashi); Department of Surgery, Boonshoft School of Medicine, Wright State University, Dayton, Ohio (Dr Fox); Departments of Emergency Medicine, Biostatistics, and Epidemiology, University of Pennsylvania, Philadelphia (Dr Carr); Departments of Emergency Medicine and Health Policy, George Washington University, Washington, DC (Dr Pines); and Center for Outcomes Research and Evaluation, Yale–New Haven Hospital, New Haven, Connecticut (Dr Ross).
Importance Current efforts to improve health care focus on hospital readmission rates as a marker of quality and on the effectiveness of transitions in care during the period after acute care is received. Emergency department (ED) visits are also a marker of hospital-based acute care following discharge but little is known about ED use during this period.
Objectives To determine the degree to which ED visits and hospital readmissions contribute to overall use of acute care services within 30 days of discharge from acute care hospitals, to describe the reasons patients return for ED visits, and to describe these patterns among Medicare beneficiaries and those not covered by Medicare insurance.
Design, Setting, and Participants Prospective study of patients aged 18 years or older (mean age: 53.4 years) who were discharged between July 1, 2008, and September 31, 2009, from acute care hospitals in 3 large, geographically diverse states (California, Florida, and Nebraska) with data recorded in the Healthcare Cost and Utilization Project state inpatient and ED databases.
Main Outcome Measures The 3 primary outcomes during the 30-day period after hospital discharge were ED visits not resulting in admission (treat-and-release encounters), hospital readmissions from any source, and a combined measure of ED visits and hospital readmissions termed hospital-based acute care.
Results The final cohort included 5 032 254 index hospitalizations among 4 028 555 unique patients. In the 30 days following discharge, 17.9% (95% CI, 17.9%-18.0%) of hospitalizations resulted in at least 1 acute care encounter. Of these 1 233 402 postdischarge acute care encounters, ED visits comprised 39.8% (95% CI, 39.7%-39.9%). For every 1000 discharges, there were 97.5 (95% CI, 97.2-97.8) ED treat-and-release visits and 147.6 (95% CI, 147.3-147.9) hospital readmissions in the 30 days following discharge. The number of ED treat-and-release visits ranged from a low of 22.4 (95% CI, 4.6-65.4) encounters per 1000 discharges for breast malignancy to a high of 282.5 (95% CI, 209.7-372.4) encounters per 1000 discharges for uncomplicated benign prostatic hypertrophy. Among the highest volume discharges, the most common reason patients returned to the ED was always related to their index hospitalization.
Conclusions and Relevance After discharge from acute care hospitals in 3 states, ED visits within 30 days were common among adults and accounted for 39.8% of postdischarge hospital-based acute care visits. Improving care transitions should focus not only on decreasing readmissions but also on ED visits.
Hospital readmissions within 30 days of discharge are common, costly, and often related to the index hospitalization.1-5 Increasingly, a hospital's readmission rate is being viewed as a marker of the quality of care provided to patients and the effectiveness of the discharge process as the patient's care is transitioned to the outpatient setting.6,7 Focusing solely on hospital readmissions, however, may be too narrow and provide an incomplete picture of the use of hospital-based acute care following discharge.
Emergency department (ED) visits are another important outcome following hospital discharge even when they do not result in hospital readmission. A return to the ED after hospital discharge often reflects poorly executed transitions in care and has the potential to result in fragmentation of care following discharge.8-10 A visit to the ED during the postdischarge period also is increasingly being interpreted by policy makers as an important measure in monitoring and evaluating innovative care delivery programs.11-14
Despite the clinical and policy relevance, patterns of emergency care use among recently discharged patients are poorly understood. Existing studies have been limited by failing to differentiate ED visits that result in readmission from those that result in discharge, or have tended to focus on the experience at a single institution, with a single payer, or with a specific condition.15-25 Consequently, clinicians and policy makers may be underestimating the extent of patients' hospital-based acute care needs after hospital discharge.
To address this knowledge gap, we studied hospital-based acute care encounters in the 30 days following hospital discharge. Using large, population-based, multipayer databases from 3 geographically dispersed states, we identified patients who were discharged from acute care hospitals to determine the degree to which ED visits (treat-and-release encounters) and hospital readmissions contribute to overall use of acute care services within 30 days of hospital discharge overall as well as within condition-specific subgroups, aimed to describe the clinical diagnoses leading to return ED treat-and-release visits, and aimed to describe utilization patterns among Medicare beneficiaries and those not covered by Medicare insurance.
We identified patients from California, Florida, and Nebraska with 2008-2009 data in the Healthcare Cost and Utilization Project (HCUP) state inpatient and ED databases. These states were selected for their geographic distribution, data quality, and chiefly because their databases contain unique patient identifiers that enable follow-up of patients over time and across the inpatient and ED settings. These data are collected at the state level and are made publicly available by the Agency for Healthcare Research and Quality.26,27
The inpatient databases include all inpatient discharges from short-term, acute care, nonfederal hospitals, including those patients admitted via the ED. In contrast, the ED databases are limited to treat-and-release encounters in which patients presented to the ED but were not admitted to the hospital.
Each hospital or ED discharge record includes sociodemographic, hospital, and clinical variables, as well as up to 25 diagnostic and 21 procedure codes based on International Classification of Diseases, Ninth Revision, Clinical Modification(ICD-9-CM) coding.28 Records across both the state ED and inpatient databases were linked using an encrypted patient-level identifier.
The 3 primary 30-day outcomes for this study were ED visits (defined as treat-and-release encounters), all-cause hospital readmissions, and a combined measure of ED treat-and-release visits and hospital readmissions termed hospital-based acute care. Using HCUP's patient identifiers, all ED treat-and-release visits were identified from state ED databases and all hospital readmissions (regardless of whether the readmission was to the discharging hospital) from state inpatient databases for the focus population.
In addition, a condition-specific ED index (defined as the ratio of the ED visit rate to the readmission rate) was created to delineate how ED treat-and-release visits contribute to overall use at the condition level of hospital-based acute care after hospital discharge. A value of greater than 1 indicates that ED visits not requiring admission were more frequent than readmissions. Conversely, a value of less than 1 indicates patients were more frequently readmitted to the hospital than discharged from the ED.
The primary unit of analysis was the hospital discharge. Index hospitalizations were classified by diagnosis related group version 24 at the time of patient discharge. All index discharges were flagged as either medical or surgical based on HCUP's Grouper processing for diagnosis related group. Additional patient variables used for analysis were selected from the time of index discharge and included age, sex, race, and ethnicity (white, black, Hispanic, other, or missing), median income based on the patient's home zip code stratified into quartiles, and primary insurance payer status (private, Medicare, Medicaid, or other).
Information about reasons for postdischarge acute care encounters were collected from the diagnosis category associated with each encounter. Diagnosis categories are based on the Agency for Healthcare Research and Quality's Clinical Classification Software, which groups all ICD-9 diagnosis codes into clinically meaningful, mutually exclusive diagnosis categories. The Agency for Healthcare Research and Quality defined comorbidities based on ICD-9-CM codes and the methods of Elixhauser et al.29 These included dichotomous indicators for 29 comorbidities, which were summed to create a comorbidity score.
We calculated descriptive statistics for the sample. We defined the rate of ED treat-and-release visits, hospital readmissions, and overall hospital-based acute care as the total number of respective encounters (ie, ED visits, readmissions) within 30 days of index discharge divided by the total number of discharges from acute care hospitals. In this way, multiple visits to the ED or multiple readmissions by individual patients were uniquely captured. Similarly, we calculated condition-specific rates by dividing the total number of 30-day encounters among patients discharged within each diagnosis-related group (numerator) by the total number of patients discharged with that diagnosis-related group (denominator). All rates are expressed as the total number of encounters per 1000 discharges.
A sensitivity analysis using patient as the unit of analysis yielded similar results regarding the frequency of ED encounters (eSupplement, eTable 1, and eTable 2). Confidence intervals for all rates were calculated using a Poisson analysis. All analyses were conducted using SAS version 9.2 (SAS Institute Inc) and Stata version 10 (StataCorp). This study was considered exempt from review by the Yale University human investigations committee.
All California, Florida, and Nebraska hospital discharges between July 1, 2008 and September 31, 2009 (N = 6 735 565) were identified from the HCUP state inpatient databases for residents aged 18 years or older. From this population, discharges were sequentially excluded that had missing disposition data (n = 14 137) and those in which patients were discharged against medical advice (n = 89 280), died during their index hospitalization (n = 151 191), were transferred to another acute care facility (n = 94 737), were discharged after a hospitalization primarily requiring rehabilitation services (n = 102 068), or were missing a valid, encrypted patient identifier (n = 371 687).
Discharges by the same patient were excluded only if they occurred less than 31 days apart (n = 879 706) because these events would represent 1 of our primary outcomes. In addition, to avoid including rare or miscoded conditions, discharges were excluded when patients were discharged with a diagnosis that was reported in the inpatient database fewer than 100 times (n = 505). After all exclusions, 5 032 254 discharges remained among 4 028 555 unique patients.
The mean age of the patients in our cohort was 53.4 years, and they were well distributed across all age groups. Patients aged 65 years or older comprised 29.2% of the sample. The majority of patients were female (53.5%) and white (48.0%). The majority of patients had some form of insurance coverage (75.1%), with Medicare being the most common (29.9%), followed by private insurance (32.3%). Medicaid was the primary insurance for 15.3% of patients. Just more than one-quarter of patients (26.7%) had 0 comorbidities, while 21.5% had 3 or more comorbid conditions. Patients had been discharged after index hospitalization for 470 unique conditions. Of these, medical discharges (65.2%) were more common than surgical discharges (34.8%).
Of all the hospitalizations in our study, 17.9% (95% CI, 17.9%-18.0%) resulted in at least 1 acute care encounter in the 30 days following discharge, 7.5% (95% CI, 7.5%-7.6%) of discharges were followed by at least 1 ED encounter, and 12.3% (95% CI, 12.3%-12.3%) by at least 1 readmission. For every 1000 discharges, there were 97.5 (95% CI, 97.2-97.8) ED treat-and-release visits and 147.6 (95% CI, 147.3-147.9) hospital readmissions in the 30 days following discharge (Figure 1). Visits to the ED comprised 39.8% (95% CI, 39.7%-39.9%) of the 1 233 402 postdischarge acute care encounters.
Approximately one-third of hospital-based acute care use occurred during the first 7 days following hospital discharge (35.3% [95% CI, 35.1%-35.4%] of ED visits and 31.9% [95% CI, 31.7%-32.0%] of readmissions), and more than half occurred during the first 14 days postdischarge (57.4% [95% CI, 57.2%-57.6%] of ED visits and 55.7% [95% CI, 55.5%-55.9%] of readmissions). About 57% (95% CI, 56.6%-57.0%) of the hospital readmissions were admitted through the ED.
Rates for visits to the ED, readmission, and use of hospital-based acute care varied for the most common (highest volume) medical and surgical conditions (Table 1). In aggregate, these highest volume medical and surgical conditions accounted for 40% of all index discharges.
Among the highest volume medical conditions, 30-day postdischarge ED treat-and-release visit rates were highest for digestive disorders (140.7 [95% CI, 138.1-143.3] encounters/1000 discharges) and psychosis (219.4 [95% CI, 217.2-221.5] encounters/1000 discharges). The highest overall use rates for hospital-based acute care were for heart failure (373.5 [95% CI, 370.0-377.0] encounters/1000 discharges) and psychosis (470.8 [95% CI, 467.7-474.0 encounters/1000 discharges).
Among the highest volume surgical conditions, the highest rates of ED treat-and-release visits were for complicated laparoscopic cholecystectomy (84.5 [95% CI, 81.3-87.8] encounters/1000 discharges) and complicated cesarean delivery (84.6 [95% CI, 82.2-87.0] encounters/1000 discharges). The highest rates of overall use rates for hospital-based acute care were for percutaneous coronary interventions with drug-eluting stent and major cardiovascular diagnosis (233.6 [95% CI, 228.5-238.8] encounters/1000 discharges) and complicated hip and femur procedures excluding major joint (241.7 [95% CI, 236.4-247.1] encounters/1000 discharges). Although patients returned to the ED for a variety of reasons, for the highest volume conditions, ED treat-and-release visits were always related to the index hospitalization (Table 2).
There was substantial variability in use rates of acute care across the 470 different index discharge conditions (Figure 2). The number of ED treat-and-release visits ranged from a low of 22.4 (95% CI, 4.6-65.4) encounters/1000 discharges for breast malignancy to a high of 282.5 (95% CI, 209.7-372.4) encounters/1000 discharges for uncomplicated benign prostatic hypertrophy. Conditions with the highest rates of ED visits were related to mental health, drug and alcohol abuse, and benign prostatic hypertrophy. Readmissions were lowest following vaginal deliveries, cesarean deliveries, and gynecological procedures and highest following admissions for false labor, chemotherapy, and malignancy-related hospitalizations, organ transplants, and threatened abortions.
Of the 470 index discharge conditions, 25.7% had higher ED revisit rates than readmission rates (ED index >1). The relationship between ED treat-and-release visits and readmissions, however, is condition specific. For example, the ED treat-and-release revisit rates were similar for seizures (197.6 [95% CI, 191.2-204.1] ED visits/1000 discharges) and headache (198.6 [95% CI, 190.8-206.7] ED visits/1000 discharges), whereas their respective readmission rates were considerably different (106.5 [95% CI, 100.8-112.4] and 184.5 [95% CI, 178.3-190.7] readmissions/1000 discharges).
Conversely, readmission rates were similar following septicemia (220.7 [95% CI, 217.5-223.9] readmissions/1000 discharges), coronary bypass (221.7 [95% CI, 204.9-239.5] readmissions/1000 discharges), and complicated kidney and ureter procedures (221.0 [95% CI, 208.7-233.7] readmissions/1000 discharges), whereas their respective rates of ED treat-and-release visits varied (81.6 [95% CI, 79.6-83.5], 101.2 [95% CI, 90.0-113.5], and 140.1 [95% CI, 130.4-150.3] ED visits/1000 discharges).
Patterns of hospital-based acute care use for the 3 most common medical (Figure 3) and surgical (Figure 4) conditions varied between patients with Medicare coverage and those without Medicare coverage. The overall use of hospital-based acute care for patients with Medicare coverage was 288.9 (95% CI, 288.2-289.7) encounters per 1000 discharges, whereas among patients without Medicare coverage, the rate was 212.1 (95% CI, 211.7-212.7) encounters per 1000 discharges. Patients with Medicare coverage returned to the ED (92.0 [95% CI, 91.6-92.5] ED visits/1000 discharges) at similar rates as patients without Medicare coverage (101.6 [95% CI, 101.3-102.0] ED visits/1000 discharges), whereas their respective readmission rates were considerably different (196.9 [95% CI, 196.3-197.5] and 110.6 [95% CI, 110.2-111.0] readmissions/1000 discharges).
Our population-based study of more than 4 million adult patients in 3 states demonstrated high rates of hospital-based acute care use following medical and surgical inpatient discharges. Nearly 18% of hospitalizations resulted in at least 1 acute care encounter within the 30 days following discharge. Our study adds to prior work on hospital readmission rates by showing that ED treat-and-release visits account for nearly 40% of all hospital-based acute care use during the postdischarge period. Focusing solely on readmissions would have missed nearly half a million ED treat-and-release encounters in these 3 states and substantially underestimated acute care use following medical and surgical inpatient discharges. These ED visits are likely to result in fragmented care following discharge and consequently contribute to duplication of services, conflicting care recommendations, medication errors, patient distress, or higher costs.9
While the optimal role of the ED during the postdischarge period has yet to be clearly defined, high and varying rates of ED use suggest there is potential to improve acute care delivery. Some ED encounters involve critically ill, high-risk patients who will require readmission. On the other hand, we found several conditions with high ED indices, meaning patients with these conditions were much more likely to be treated and released from the ED than readmitted. Because many of these patients presented to the ED for reasons related to their index admission, anticipating patient needs and developing an appropriate care plan prior to hospital discharge may help prevent some of these likely low-acuity visits. Similarly, given that patients hospitalized for reasons related to mental illness and drug and alcohol abuse had especially high rates of return to the ED, there must be consideration of how acute care can be best delivered and targeted to this population outside of hospitals.
We also found that patterns of use varied by Medicare insurance status, which likely reflects the unique needs of an older adult population. Our results may be helpful to those who are targeting care transition interventions to older populations with the intention of decreasing ED visits and readmissions.6,30 The policies of the Centers for Medicare & Medicaid Services that are directed at reducing use and cost after hospital discharge also should consider the implications of accounting for postdischarge ED use and hospital readmissions, both for patients and for discriminating hospital quality. However, it is important to note that ED use after discharge is not synonymous with a lapse in quality. In the design of interventions to reduce high rates of acute care use after hospital discharge, these descriptive results should be used to inform future research and to identify the underlying, modifiable patient factors and system failures that increase risk.
Our finding of high rates of ED visits during the period after hospital discharge has important policy implications. Current reform efforts will encourage increasing participation in new payment and delivery models that directly and indirectly incentivize avoiding use of costly and possibly avoidable acute care.31,32 However, policies that incentivize reducing readmissions may result in unintended consequences. For example, patient care could shift to EDs and observation units in which emergency physicians may be encouraged to avoid readmitting patients. Hence, ED use may be a useful, patient-centered metric to track as part of the efforts to decrease hospital readmission. Even if readmission rates decrease, high or increasing rates of ED use during the period after hospital discharge may reflect shortcomings in access to and delivery of care during the transition from hospital to home.
Our study should be viewed in the context of several limitations. Our data were derived from only 3 states; however, they were representative of the state population and the states chosen were large and geographically diverse. Moreover, in aggregate these states account for approximately 17% of hospitalizations in the United States.
Second, because this study focused on hospital-based acute care visits, we only measured acute care that occurred in the ED or inpatient hospital setting. Patients placed in observation status and visits to physician offices or other ambulatory care sites were not included. It is also possible that some patients visited urgent care centers or walk-in clinics, as opposed to the ED, that are not captured in the HCUP databases, particularly in the state of California where Kaiser has a substantial presence. Hence, our results may actually underestimate the use of acute care after hospital discharge.
Third, we were not able to identify those patients who died following hospitalization. Fourth, as with any large database analysis, inherent limitations related to the use of claims-based administrative data has the potential for errors in recording diagnoses, and thus misclassification of encounters. While such errors are possible, HCUP data are highly accurate, rigorously tested, and widely used to estimate diagnoses and visit frequency.15,28-30
In addition, we used diagnoses from administrative data as a proxy for the reason for return visit. Using the reason for visit may be a more accurate representation of why patients return; however, chief complaints or similar data are not available in HCUP data.
In conclusion, hospital-based acute care encounters are frequent among patients recently discharged from an inpatient setting. An improved understanding of how the ED setting is best used in the management of acute care needs—particularly for patients recently discharged from the hospital—is an important component of the effort to improve care transitions. The use of hospital readmissions as a lone metric for postdischarge health care quality may be incomplete without considering the role of the ED. Just as the Patient Protection and Affordable Care Act requires the development of programs to reduce readmissions, further initiatives are necessary to understand the drivers of postdischarge ED use and the clinical and financial efficiency associated with providing such acute care in the ED.
Corresponding Author: Anita A. Vashi, MD, MPH, Yale University, 333 Cedar St, SHM-1E-61, New Haven, CT 06520 (firstname.lastname@example.org).
Author Contributions: Drs Vashi and Gross had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Vashi, Fox, D’Onofrio, Pines, Ross, Gross.
Acquisition of data: Vashi, Fox.
Analysis and interpretation of data: Vashi, Fox, Carr, Ross.
Drafting of the manuscript: Vashi, Fox.
Critical revision of the manuscript for important intellectual content: Vashi, Carr, D’Onofrio, Pines, Ross, Gross.
Statistical analysis: Vashi, Fox.
Obtained funding: Gross.
Study supervision: Carr, D’Onofrio, Ross, Gross.
Conflict of Interest Disclosures: The authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Carr reported being a senior policy advisor for the Office of the Assistant Secretary for Preparedness and Response. Dr Pines reported working as a senior advisor for the Centers for Medicare & Medicaid Services Innovation Center during the writing of this article. Dr Ross reported receiving funding from the Centers for Medicare & Medicaid Services to develop and maintain performance measures that are used for public reporting; and receiving funding from the Pew Charitable Trusts to examine regulatory issues at the US Food and Drug Administration. Drs Ross and Gross reported receiving funding as collaborators on the Yale University Open Access project, which is facilitating the objective analysis of Medtronic clinical trial data. Drs Ross and Gross are members of a scientific advisory board for FAIR Health Inc. No other disclosures were reported.
Funding/Support: Dr Carr is supported by career development award K08 AG032886 from the Agency for Healthcare Research and Quality. Dr Ross is supported by the National Institute on Aging and by the American Federation for Aging Research through the Paul B. Beeson Career Development Award Program.
Role of the Sponsor: The funding organizations had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.
Disclaimer: The views expressed in this article are those of the authors and do not reflect the official policy of the US Air Force, Department of Defense, Department of Veterans Affairs, Department of Health and Human Services, or the US government.
Create a personal account or sign in to: