Original Research Trends in Hospital Admission and Surgical Procedures Following ED visits for Diverticulitis Margaret B. Greenwood-Ericksen, MD, MPH*†‡ Joaquim M. Havens, MD‡§¶ Jiemin Ma, PhD|| Joel S. Weissman, PhD‡§ Jeremiah D. Schuur, MD, MHS†‡ *Brigham and Women’s Hospital, Department of Emergency Medicine, Boston, Massachusetts † Massachusetts General Hospital, Department of Emergency Medicine, Boston, Massachusetts ‡Harvard Medical School, Boston, Massachusetts §Brigham and Women’s Hospital, Center for Surgery and Public Health, Department of Surgery, Boston, Massachusetts ¶Brigham and Women’s Hospital, Division of Trauma, Burns and Surgical Critical Care, Boston, Massachusetts ||Surveillance and Health Services Research Program, American Cancer Society, Atlanta, Georgia Section Editor: Mark I. Langdorf, MD, MHPE Submission history: Submitted January 13, 2016; Revision received March 28, 2016; Accepted April 7, 2016 Electronically published June 13, 2016 Full text available through open access at http://escholarship.org/uc/uciem_westjem DOI: 10.5811/westjem.2016.4.29757 Introduction: Diverticulitis is a common diagnosis in the emergency department (ED). Outpatient management of diverticulitis is safe in selected patients, yet the rates of admission and surgical procedures following ED visits for diverticulitis are unknown, as are the predictive patient characteristics. Our goal is to describe trends in admission and surgical procedures following ED visits for diverticulitis, and to determine which patient characteristics predict admission. Methods: We performed a cross-sectional descriptive analysis using data on ED visits from 2006-2011 to determine change in admission and surgical patterns over time. The Nationwide Emergency Department Sample database, a nationally representative administrative claims dataset, was used to analyze ED visits for diverticulitis. We included patients with a principal diagnosis of diverticulitis (ICD-9 codes 562.11, 562.13). We analyzed the rate of admission and surgery in all admitted patients and in low-risk patients, defined as age <50 with no comorbidities (Elixhauser). We used hierarchical multivariate logistic regression to identify patient characteristics associated with admission for diverticulitis. Results: From 2006 to 2011 ED visits for diverticulitis increased by 21.3% from 238,248 to 302,612, while the admission rate decreased from 55.7% to 48.5% (-7.2%, 95% CI [–7.78 to -6.62]; p<0.001 for trend). The admission rate among low-risk patients decreased from 35.2% in 2006 to 26.8% in 2011 (-8.4%, 95% CI [–9.6 to –7.2]; p<0.001 for trend). Admission for diverticulitis was independently associated with male gender, comorbid illnesses, higher income and commercial health insurance. The surgical rate decreased from 6.5% in 2006 to 4.7% in 2011 (-1.8%, 95% CI [–2.1 to –1.5]; p<0.001 for trend), and among low-risk patients decreased from 4.0% to 2.2% (- 1.8%, 95% CI [–4.5 to –1.7]; p<0.001 for trend). Conclusion: From 2006 to 2011 ED visits for diverticulitis increased, while ED admission rates and surgical rates declined, with comorbidity, sociodemographic factors predicting hospitalization. Future work should focus on determining if these differences reflect increased disease prevalence, increased diagnosis, or changes in management. [West J Emerg Med. 2016;17(4):409-417.] Volume XVII, NO. 4 : July 2016 409 Western Journal of Emergency Medicine Diverticulitis Admission and Surgery Trends Greenwood-Ericksen et al. INTRODUCTION Colonic diverticular disease is increasingly prevalent in the developed world and affects more than half of the population over the age of 65 years.1 It is estimated that approximately 20% of patients with diverticulosis develop diverticulitis over the course of their lifetime.2 Diverticulitis frequently causes abdominal pain, which accounts for approximately 8% of U.S. emergency department (ED) visits.3 Approximately 300,000 patients are admitted to U.S. hospitals for diverticulitis each year, accounting for 1.5 million days of inpatient care per year.4,5 Treatment of diverticulitis is based on comorbidities and severity, with severe disease requiring admission and possible surgical intervention.6,7 A recent meta-analysis8 and prospective randomized control trial9 both demonstrate the safety of outpatient management with oral antibiotics for uncomplicated diverticulitis. In 2014 the American Society of Colon and Rectal Surgeons recommended outpatient management in selected patients with uncomplicated diverticulitis.7,10 Despite evidence to support outpatient management, the published literature has reported increased admission and surgical rates from the late 1990s to early 2000s.11,12 With the increasing prevalence of diverticular disease and the increasing role of the ED in management of acute conditions, we aimed to determine if there has been a change in hospital admission and surgery among ED patients with diverticulitis. This study analyzed data from a national allpayer hospital billing dataset to evaluate the prevalence, the rate of admission, and the rate of surgical intervention for patients with diverticulitis who presented to the ED. Additionally, we determined patient predictors of admission for patients. Specifically, we hypothesized that rates of admission and surgery have decreased in recent years. METHODS Study Design and Data Source We conducted a cross-sectional descriptive analysis using data on ED visits from 2006-2011 to determine change in admission and surgical patterns over time. Additionally, to determine patient predictors of admission, we performed a multiple variable logistic regression analysis, adjusting for patient comorbidity using the system developed by Elixhauser.13 This study was approved by the institutional review board at Brigham and Women’s Hospital. The Nationwide Emergency Department Sample (NEDS) was used for our analysis. NEDS is a U.S. administrative database that is part of the Healthcare Cost and Utilization Project.14 NEDS is a component of the Healthcare Cost and Utilization Project (HCUP) of the Agency for Healthcare Research and Quality (AHRQ); it is the largest all-payer ED database publicly available in the U.S.14 NEDS contains 26 to 29 million ED records per year from approximately 950 annually selected hospitals, which represents roughly a 20% stratified sample of hospital-based EDs in the U.S.14 NEDS uses a complex sampling design stratified by sampling weight, geographic region, trauma center designation, urban–rural status, teaching hospital status, and hospital ownership to allow for calculation of national estimates.14 Visit details available in NEDS include patient demographics, visit disposition (home, transfer to another facility, admitted to hospital, or expired), and up to 15 diagnoses from the final location (e.g. inpatient diagnoses are from the hospital bill while diagnoses for patients discharged from the ED are from the ED bill). By incorporating sampling weights provided in NEDS, we were able to generate national estimates for ED utilization at both hospital and visit level in the U.S. More detailed descriptions of NEDS can be found elsewhere.14 Study Population We included ED visits by adult patients, 18 years and older, who had an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code for diverticulitis of the large colon (562.11, 562.13) as their principal diagnosis. In sensitivity analyses, we included ED visits where diverticulitis was a secondary diagnosis and where the principal diagnosis was thought to be diverticulitisrelated, e.g. abdominal pain (Appendix A). We excluded patients with a disposition of neither discharge nor admission (left against medical advice, not admitted, destination unknown, or died in the ED; 0.38%). We also excluded hospitals with <10 cases (18.8% of hospitals; 1.1% of visits) because low hospital volumes result in unstable estimates of admission rates. We excluded patients with complicated diverticulitis as defined by the American College of Surgeons7 (i.e., peritonitis, obstruction, perforation and abscess) and those with sepsis or shock, because virtually all such patients should be admitted to the hospital from the ED (Appendix B). We defined low-risk patients as those with no Elixhauser comorbidities and as age less than 50, which is defined as “young” by the American College of Colon and Rectal Surgeons.7 Study Outcome and Variables The primary outcome of interest was hospital admission after ED visits. We classified patients as admitted if they were admitted to the hospital or transferred to an acute care hospital after the initial ED visit, because the decision to transfer a patient represents a similar use of hospital care rather than discharging the patient to outpatient management. Patients were classified as discharged if their disposition was “routine ED discharge,” “transfer to skilled nursing or intermediate care facility,” “home health care,” or “discharge or transfer to court or law enforcement.” An additional outcome of interest was the rate of surgical procedures; the surgical rate was calculated for all admitted patients and the low-risk sub-group. Data for outpatient, elective surgery were not available. We defined surgery as patients with at least one ICD-9 procedure code that indicated the patient had Western Journal of Emergency Medicine 410 Volume XVII, NO. 4 : July 2016 Greenwood-Ericksen et al. Diverticulitis Admission and Surgery Trends undergone a colectomy (45.7x or 45.94), a low anterior resection (48.6x), a colostomy (46.1x), an ileostomy (46.2x), a laparotomy (54.11 or 54.19), diagnostic laparoscopy (54.21), laparascopic lysis of adhesions (54.51), or percutaneous drainage (54.91). Data Analysis National estimates of ED visits, admission rates and surgical rates for diverticulitis were estimated accounting for NEDS’s complex sampling design and sampling weights. We tested the trend in admission and surgical rates from 2006 ̶ 2011 by logistic regression modeling by calendar year. The admission rate was defined as the number of patients admitted or transferred to another hospital, divided by the number of ED visits. The surgical rate was defined as the number of patients who underwent a surgical procedure, divided by the number of ED visits. Additionally, we determined an inpatient surgical rate in all admitted patients and in low-risk, admitted patients. The inpatient surgical rate was defined as the number of patients who underwent a surgical procedure, divided by the number of admitted patients. As our study population is a subset of NEDS, we applied subset analysis methods as recommended by AHRQ to obtain correct variance estimates for these descriptive statistics. Patient predictors include age at time of visit, gender, insurance status (private, Medicare, Medicaid, self-pay/ no charge, and other), median household income (quartile within the patient’s home ZIP code), and comorbid illness. We adjusted for comorbity using the system developed by Elixhauser. For each ED visit, we created dummy variables for each comorbidity cluster defined by Elixhauser, based on secondary diagnosis codes 13 and also created three dummy variables for additional conditions identified as likely to increase the chance of admission for diverticulitis that are not included in Elixhauser (“GI symptom,” “GI disease,” “disease severity”). For example, leukocytosis and acute renal failure are examples of diagnoses grouped under “disease severity,” that would increase a patient’s risk of being admitted with a principal diagnosis of diverticulitis, while benign prostatic hypertrophy is not. To determine these diagnoses, one author (MBG-E) reviewed all secondary codes on patients admitted with diverticulitis and flagged those that would increase the likelihood of admission. Independently, a surgical expert (JMH) reviewed the codes, and disagreements were resolved by discussion (Appendix C). Statistical Analyses We report descriptive statistics and compare trends across years using chi-square tests for trend. To account for patient clustering within EDs and the associated clustering of care patterns for admission and surgery, we created hierarchical multivariate logistic regression models using validated analytical methods used by Centers for Medicare and Medicaid Services for analyzing administrative claims to determine morbidity and readmission.15 The models included patient and hospital characteristics as covariates. As suggested by the HCUP, sampling weights were not used in multilevel modeling. All analyses were done in SAS 9.3 (SAS, Cary, NC). RESULTS In 2011 there were 302,612 ED visits for diverticulitis. Mean patient age was 58 years, the majority were female (56.7%), with the plurality having private insurance (43.7%), presenting to metropolitan non-teaching hospitals (50.1%), and being located in the southern region of the U.S. (41%; Table 1). ED visits increased by 21.1% from 2006 to 2011 (Figure 1). From 2006 to 2011, admission rates decreased from 55.7% to 48.5% (-7.2%, 95% CI [–7.78 to -6.62]; test for trend, p<0.001 (Figure 1 and Table 2). The rate of surgery decreased from 6.5% in 2006 to 4.7% Table 1. Patient and hospital characteristics for diverticulitis emergency department visits, 2011. Characteristics N% Mean age (SD) 66,656 57.6 (0.06) Female 37,760 56.7 Insurance Medicare Medicaid Private insurance Self-pay/no charge Other 23,264 5,568 29,078 6,585 2,017 35.0 8.4 43.7 9.9 3.0 Income Lowest quartile Second quartile Third quartile Highest quartile 15,490 15,843 17,433 16,604 23.7 24.2 26.7 25.4 Region Northeast Midwest South West 13,442 12,755 27,331 13,128 20.2 19.1 41.0 19.7 Teaching status Metropolitan, non-teaching Metropolitan, teaching Non-metropolitan 33,400 22,655 10,601 50.1 34.0 15.9 Emergency department volume <20,000 7,087 10.6 20,000-49,999 26,591 39.9 <50,000 32,978 49.5 Income, quartile of the median household income of the patient’s home ZIP code. Region, as defined by the U.S. Census Bureau. Volume XVII, NO. 4 : July 2016 411 Western Journal of Emergency Medicine DFiivgeruticruelit1is.ANdmaitsisoionn aanldTSurregenrydTirnendEs mergency Department Visits and AdmGriesesniwoonodR-Earictkessenfeotral. Diverticulitis, 2006 – 2011. 350,000 60% 300,000 50% 250,000 40% ED Visits Admit Rate 200,000 150,000 30% 100,000 20% 50,000 10% ED visits 2006 2007 2008 2009 Year 2010 0% 2011 Overall admission rate Low-risk admission rate *See Table 2 for values.Figure 1. National trend in emergency department (ED) diagnosis and admission rates for diverticulitis, 2006-2011. *See Table 2 for values. FFoor arllatrlelndtsr,epn