ABSTRACT
BACKGROUND: It is not known whether delivering inpatient care earlier to patients boarding in the emergency department (ED) by a hospitalist-led team can decrease length of stay (LOS). OBJECTIVE: To study the association between care provided by a hospital medicine ED Boarder (EDB) service and LOS. DESIGN, SETTING, AND PARTICIPANTS: Retrospective cross-sectional study (July 1, 2016 to June 30, 2018) conducted at a single, large, urban academic medical center. Patients admitted to general medicine services from the ED were included. EDB patients were defined as those waiting for more than two hours for an inpatient bed. Patients were categorized as covered EDB, noncovered EDB, or nonboarder. INTERVENTION: The hospital medicine team provided continuous care to covered EDB patients waiting for an inpatient bed. PRIMARY OUTCOME AND MEASURES: The primary outcome was median hospital LOS defined as the time period from ED arrival to hospital departure. Secondary outcomes included ED LOS and 30-day ED readmission rate. RESULTS: There were 8,776 covered EDB, 5,866 noncovered EDB, and 2,026 nonboarder patients. The EDB service covered 59.9% of eligible patients and 62.9% of total boarding hours. Median hospital LOS was 4.76 (interquartile range [IQR] 2.90-7.22) days for nonboarders, 4.92 (IQR 3.00-8.03) days for covered EDB patients, and 5.11 (IQR 3.16-8.34) days for noncovered EDB (P < .001). Median ED LOS for nonboarders was 5.6 (IQR 4.2-7.5) hours, 20.7 (IQR 15.8-24.9) hours for covered EDB, and 10.1 (IQR 7.9-13.8) hours for noncovered EDB (P < .001). There was no difference in 30-day ED readmission rates. CONCLUSION: Admitted patients who were not boarders had the shortest LOS. Among boarded patients, coverage by a hospital medicine-led EDB service was associated with a reduced hospital LOS.
ABSTRACT
OBJECTIVE: To conduct a systematic review of published cardiac risk indices relevant to patients undergoing noncardiac surgery and to provide clinically meaningful recommendations to physicians regarding the use of these indices. METHODS: A literature search of articles published from January 1, 1999, through December 28, 2018, was conducted in Ovid (MEDLINE), PubMed, Embase, CINAHL, and Web of Science. Publications describing models predicting risk of cardiac complications after noncardiac surgery were included and citation chaining was used to identify additional studies for inclusion. RESULTS: Eleven risk indices involving 2,910,297 adult patients were included in this analysis. Studies varied in size, population, quality, risk of bias, outcome event definitions, risk factors identified, index outputs, accuracy, and clinical usefulness. Studies considered 6 to 83 variables to develop their models. Among the identified models, the factors with the highest predictiveness for adverse cardiac outcomes included congestive heart failure, type of surgery, creatinine, diabetes, history of stroke or transient ischemic attack, and emergency surgery. Substantial data from the large studies also supports advancing age, American Society of Anesthesiology physical status classification, functional status, and hypertension as additional risks. CONCLUSION: The risk indices identified generally fell into two groups - those with higher accuracy for predicting a narrow range of cardiac outcomes and those with lower accuracy for predicting a broader range of cardiac outcomes. Using one index from each group may be the most clinically useful approach. Risk factors identified varied widely among studies. In addition to judicious use of predictive indices, reasoned clinical judgment remains indispensable in assessing perioperative cardiac risk.
