Traditional Approach.: Abdominal pain is one of the most frequent complaints in the ED and accounts for 4 to 8% of all visits.⁵ The typical ED evaluation of the patient with abdominal pain includes a thorough history and physical examination and the appropriate diagnostic tests. Within the short time frame of 2 to 4 hours, patients are given a provisional diagnosis and are either hospitalized or sent home.

Problem with Traditional Approach.: The ED evaluation is inadequate for many patients, and in 40% of patients the origin of abdominal pain is never determined. Acute appendicitis is the most common abdominal surgical emergency and illustrates the inadequacies of the traditional approach. The diagnosis of acute appendicitis is missed in 20 to 30% of cases (false-negative decisions). In addition, 20 to 30% of patients taken to surgery for acute appendicitis are found to have no abnormality (false-positive decisions).⁵ Whereas abdominal computed tomography (CT) imaging has been found to improve physician performance in identifying patients with acute appendicitis,^6–8 there is increasing concern for the amount of ionizing radiation exposure to patients from the increased use of CT.⁹

Observational Approach.: With observation, the ED evaluation can be extended from 2 to 4 hours to up to 23 hours for selected patients (Box 195-2). After the initial history and physical examination of the patient, the physician estimates the probability that the patient has appendicitis by use of validated risk stratification tools, such as the Alvarado score. Those patients deemed to have a low probability of disease are ideal candidates for the observational approach.¹⁰ Patients who are at greater risk of having their diagnosis missed, such as those who are immunocompromised, pregnant, or at the extremes of age, often benefit from observation. Elders in particular benefit from observation; surgical problems are often missed because of the atypical clinical findings that occur with older age.¹¹

During the period of observation, the patient is usually kept fasted and hydrated intravenously. Serial abdominal examinations are repeated at 4-hour intervals, and laboratory tests such as complete blood count and C-reactive protein level are repeated as appropriate. Imaging and consultations are also arranged during this time frame. Patients without appendicitis will experience improvement of their pain and have had completion of diagnostic workup and exclusion of surgical disease.¹² Patients are hospitalized if they have no improvement, worsening of their clinical findings, or surgical disease diagnosed by testing. In patients who do have appendicitis, signs and symptoms will continue or worsen.¹³

Physician decision-making improves with observation, and false-positive surgeries can nearly be eliminated. Intensive observation with serial physical examinations at least every 8 hours rather than once per day has been found to reduce the false-positive rate from 20 to 5%.13,14 The use of a period of observation can also help identify many patients whose diagnoses otherwise would be missed during the initial ED evaluation. Initially, many appendicitis patients have few clinical signs or symptoms of appendicitis on presentation, making diagnosis difficult.¹² Physicians delaying disposition decisions in questionable cases can avoid these false-negative decisions. Appendicitis patients have more signs and symptoms during short-term observation, whereas those without the disorder clear their signs and symptoms. Thus fewer patients have an unclear clinical picture after observation with fewer false-negative decisions.¹⁵ Missing of the diagnosis at the initial evaluation delays surgery up to 72 hours and doubles complications (perforation, abscess).¹¹ Physicians who observe selected patients with low probability of disease rather than discharge them after their initial evaluation can avoid most of these missed diagnoses and the resulting complications.

Traditional Approach.: The ED evaluation of chest pain is composed of two assessments: (1) the probability that the patient has an acute myocardial infarction (AMI) or acute coronary ischemia (ACI) and (2) the risk of the patient’s having a life-threatening event. These two factors determine the appropriate setting for further testing and monitoring. The probability of AMI is traditionally assessed in the ED with a directed history, physical examination, and electrocardiogram (ECG) and an initial measurement of cardiac biomarkers, such as creatine kinase MB fraction (CK-MB), cardiac troponin I or T, B-type natriuretic peptide (BNP), and myoglobin. Other markers, such as high-sensitivity C-reactive protein, myeloperoxidase, and D-dimer, are used less frequently.1–5,7,8,16–23

Patients with clear evidence of AMI on the ECG are potential candidates for immediate reperfusion therapy with either thrombolytics or emergency angioplasty. This subset is at high risk for life-threatening events and best managed in the coronary care setting with close monitoring. This approach has been shown to improve survival in patients with AMI.

Depending on the evaluating physician’s threshold, chest pain patients without definite evidence of AMI at initial evaluation are generally admitted to the hospital to confirm or to exclude (“rule in” or “rule out”) ACI or AMI. Patients who are judged to have low probability of ACI after the initial evaluation are released home for outpatient follow-up.

Problem with Traditional Approach.: The poor performance of initial diagnostic testing makes the evaluation of chest pain highly dependent on clinical judgment. The initial ECG is diagnostic in only 50% of AMI patients,²⁴ and the initial CK-MB measurement has a sensitivity of only 35%.²⁵ Reliance on physician judgment and initial ED testing has resulted in as many as 2 to 5% of AMI patients being discharged home with inadvertent reassurance.^1,26 AMI patients not identified at the initial evaluation and released from the ED have up to a 25% risk of poor outcome.¹ Fear of inadvertently releasing AMI patients has led many emergency physicians to err on the side of admitting a large number of patients who do not have AMI or unstable angina. As a result of the increased sensitivity of a more liberal admission policy, two thirds of patients admitted for chest pain have a noncardiac cause of their symptoms.¹ Costs range in the billions of dollars for these negative evaluation hospitalizations.⁴ Despite this liberal admission policy, missed AMI remains one of the leading causes of malpractice suits against emergency physicians.²⁷

Observational Approach.: The emergency physician can use the observation unit to extend the evaluation of selected patients with chest pain (Box 195-3). When it is used principally for such a purpose, the observation unit has been termed a chest pain unit. The chest pain unit has been successful in improving the sensitivity and specificity of the evaluation process.^28,29 Patients with low risk of AMI are transferred to the observation unit. Patients unsuitable for observation unit evaluation include those who have a high to moderate probability of acute myocardial ischemia, unstable vital signs, electrocardiographic findings of AMI, or persistent or recurring chest pain consistent with unstable angina.

The physician identifies patients with low probability for ACI by use of one or more risk stratification tools. Risk stratification based on classic risk factors alone has been shown to be a poor predictor of short-term outcome.³⁰ Risk stratification based on electrocardiographic findings is more reliable. The Brush ECG criteria classify as low risk those without ST segment elevation or depression, T wave inversion or strain, new (or presumed new) Q waves, left bundle branch block, or paced rhythm.³¹ Another useful risk stratification tool is the Goldman protocol, which uses history, physical examination, and electrocardiographic findings to classify patients into high (>70%), moderate, or low (>7%) risk.³² Another tool is the acute cardiac ischemia time-insensitive predictive instrument, which uses age and gender of the patient, presence or absence of chest pain, and electrocardiographic findings to assign a probability of acute ischemia.³³

Patients admitted to the observation unit are first evaluated to rule out a myocardial infarction. They are serially tested with cardiac markers and ECGs. CK-MB estimation at 0, 3, and 6 hours after presentation has 100% sensitivity, 98% specificity, and 100% negative predictive value in the detection of AMI.28,29 Other useful serum cardiac markers are the troponins (I and T) and myoglobin. Patients who present more than 24 hours after symptom onset have negative CK-MB and myoglobin testing findings, but troponin T or I remains positive for up to 6 days. Patients are monitored with continuous ECG monitors equipped with dysrhythmia alarms and memory storage capabilities. Continuous electrocardiographic ST segment monitoring can detect dynamic ST segment changes indicative of ischemia, which, when present, indicate an increased likelihood for an adverse cardiac event.³⁴ Continuous electrocardiographic monitoring also helps detect arrhythmias that may be associated with acute coronary syndrome.

After evaluation to exclude AMI, the patient is evaluated for possible acute coronary syndrome. This may be performed before release of the patient from the observation unit or at follow-up evaluation within 72 hours of discharge. The most common testing modality used is exercise stress testing.³⁵ Patients who obtain their target heart rate without electrocardiographic evidence of ischemia can be released home. They have an annual mortality rate of less than 1%.³⁵ The performance of exercise testing depends on the ability of the patient to exercise adequately, gender (women have higher false-positive rates), interpretability of the resting ECG, and availability of the test. Other testing modalities include stress echocardiography, technetium (^99mTc) sestamibi scanning, and cardiac CT angiography.

Approximately one third of ED patients with chest pain are candidates for observation, with 80 to 85% released home after observation.¹ This reduces the hospitalization rate from 60 to 70% down to 40 to 50%.¹ The cost of evaluation with observation is half that of traditional evaluation with hospitalization.¹ The safety and cost-effectiveness of the observation approach have been confirmed in four randomized clinical trials.^36–39 Payers expect clinicians to be cost-effective and to use observation for evaluation of patients with chest pain when it is appropriate. The Centers for Medicaid and Medicare Services began audits in 2010 by Recovery Audit Contractors to deny payment for patients with low probability of AMI chest pain who were admitted to the hospital rather than evaluated in the lower cost outpatient observation unit. The use of observation units for chest pain evaluation is thus becoming the standard in U.S. hospitals.

Traditional Approach.: The primary objectives for the treatment of deep venous thrombosis (DVT) are to prevent pulmonary embolism, to reduce morbidity, and to prevent or to minimize the risk for development of postphlebitic syndrome. Patients with suspected DVT are usually hospitalized when diagnostic testing is unavailable in the ED or the diagnosis has been confirmed and further management is required. Traditionally, anticoagulation with unfractionated heparin has been administered by continuous intravenous infusion for 5 to 7 days while oral anticoagulation is instituted.⁴⁰ Meta-analyses of randomized trials of unfractionated heparin and low-molecular-weight heparin (LMWH) showed that they were similar, with a risk of recurrent DVT of 4%, a risk of pulmonary embolism of 2%, and a risk of major bleeding of 3%.^41,42 The anticoagulant response to this treatment varies markedly among patients, and therefore the dosage must be monitored by coagulation profiles.⁴³

Problem with Traditional Approach.: New modalities for investigation and treatment have made it possible to manage DVT on an outpatient basis with lower costs than with hospitalization. There are new, less invasive investigative tests and newer therapeutic agents that do not require monitoring of coagulation profiles.

Observational Approach.: The role of the observation unit in the management of patients thought to have DVT is for diagnostic testing as well as for initiation of therapy with LMWH and patient education. Patients often present during the night or on weekends when definitive tests for DVT (e.g., Doppler ultrasonography) are not available. The patient may have a positive D-dimer test finding, which requires a confirmatory definitive test because of its poor specificity.⁴⁴ In these circumstances, the patient can be anticoagulated for the short term with one dose of LMWH (enoxaparin, 1 mg/kg twice daily) until the diagnosis can be clarified. If the diagnosis is confirmed, the patient can be admitted or treated as an outpatient on the basis of hospital protocol. Patients considered for outpatient management are instructed in how to administer the medication. They are educated about DVT and its complications and the possible side effects of the LMWH. Appropriate follow-up evaluation is also arranged before discharge. With this approach, patients spend 67% less time in the hospital and have greater physical activity and social functioning than their standard heparin cohorts do.⁴⁵ Outpatient management is not recommended if the patient has proven or suspected concomitant pulmonary embolism, significant comorbidities, extensive iliofemoral DVT, active bleeding, renal failure, or poor follow-up compliance. LMWH is administered by subcutaneous injection in doses adjusted for the patient’s weight, without laboratory monitoring.

Outpatient testing with venous compression ultrasonography has become readily available.⁴⁶ It is both sensitive and specific for the diagnosis of proximal (femoropopliteal) DVT.⁴⁷ When repeated compression ultrasonography is compared with impedance plethysmography, compression ultrasonography is superior in detection of DVT.⁴⁶ It has been proved to be a safe method of deciding when to administer anticoagulation.⁴⁴ The D-dimer assay has also been shown to be a useful adjunct to compression ultrasonography in outpatient testing.

Traditional Approach.: Most patients with upper gastrointestinal bleeding are admitted to the hospital after initial ED assessment and stabilization.

Problem with Traditional Approach.: Upper gastrointestinal bleeding is a common and potentially life-threatening condition with an overall mortality rate of 6 to 10%.⁴⁸ However, most cases of upper gastrointestinal bleeding are self-limited, and 80% of patients have only one bleeding episode.⁴⁹

Observational Approach.: Not all patients with upper gastrointestinal bleeding do poorly, suggesting that outpatient management is possible if patients at high risk for further bleeding can be identified. Prognostic indicators include the patient’s age, heart rate, systolic blood pressure, orthostatic changes in blood pressure or pulse, color of stool or emesis, anticoagulant use, and comorbid conditions.⁵⁰ In an attempt to refine diagnostic accuracy, risk assessment, and disposition, several scoring systems have been developed. Some practitioners use hemodynamic stability, intensity of bleeding, and underlying health status as predictors of rebleeding, need for surgery, and mortality.⁵¹ Some use a period of observation with early endoscopy to identify the patient who can be discharged early. Patients found to have clean-based ulcers at endoscopy have a rebleeding rate of less than 2% and virtually never require urgent intervention for recurrent bleeding and can be released. Use of this approach has been proved to be both safe and cost-effective; a prospective clinical trial demonstrated that 24% of patients can avoid hospitalization, with cost savings of $990 per patient.⁵²

Traditional Approach.: Syncope is caused by a spectrum of disease entities. ED evaluation includes a thorough history, physical examination, and 12-lead ECG. Patients with evidence of possible myocardial ischemia or a cardiac cause of their syncope are usually admitted to the hospital because cardiac syncope has a high risk of death (up to one third will have a poor outcome).⁵³ Those with concomitant heart failure have a 25% mortality rate at 30 days.⁵⁴ On the other hand, patients with noncardiac syncope have a low risk of adverse events (1%) and can be managed as outpatients.⁵³

Problem with Traditional Approach.: Attempts to exclude a possible cardiac cause of syncope usually result in 25 to 40% of patients being hospitalized for further evaluation and management.⁵⁴ The traditional ED evaluation identifies only 50% of patients with a serious cause of their syncope,⁵⁵ and this has often resulted in a liberal admission policy; however, one study found that only 12% of patients had a serious cause of syncope that justified hospitalization.⁵⁶ It is estimated that a third of patients admitted after their ED evaluation have very low risk of an adverse event (<2%) and would be appropriate for outpatient observation evaluation.⁵⁷

Observational Approach.: The use of observation for selected patients with syncope reduces unnecessary hospitalizations (Box 195-4). Patients with a cardiac syncope have a poor prognosis and need to be identified. These patients often do not have chest pain as a symptom, but they may have ischemic changes on the ECG, arm or shoulder pain, or prior history of exercise-induced angina. A “rule out myocardial infarction” evaluation with cardiac monitoring, serial ECGs, and enzyme measurement may be the only way to identify these patients. Prolonged electrocardiographic monitoring can point to a specific cause in up to one fifth of patients, with half of all abnormalities detected in the first 24 hours.⁵⁸

The challenge for the emergency physician is to risk stratify patients into very low risk, who can be discharged home; low risk, who are appropriate for outpatient observation; and moderate to high risk, who are appropriate for acute care hospitalization. Many factors have been found to correlate with adverse outcomes and should be considered by the clinician in the risk stratification, but attempts to create simple high-reliability decision rules have not been successful.⁵⁹