Prevalence of Pulmonary Embolism among Patients Hospitalized for Syncope

Paolo Prandoni, M.D., Ph.D., Anthonie W.A. Lensing, M.D., Ph.D., Martin H. Prins, M.D., Ph.D., Maurizio Ciammaichella, M.D., Marica Perlati, M.D., Nicola Mumoli, M.D., Eugenio Bucherini, M.D., Adriana Visonà, M.D., Carlo Bova, M.D., Davide Imberti, M.D., Stefano Campostrini, Ph.D., and Sofia Barbar, M.D., for the PESIT Investigators*




BACKGROUND 
The prevalence of pulmonary embolism among patients hospitalized for syncope is not well documented, and current guidelines pay little attention to a diagnostic workup for pulmonary embolism in these patients.

METHODS 
We performed a systematic workup for pulmonary embolism in patients admitted to 11 hospitals in Italy for a first episode of syncope, regardless of whether there were alternative explanations for the syncope. The diagnosis of pulmonary embolism was ruled out in patients who had a low pretest clinical probability, which was defined according to the Wells score, in combination with a negative d-dimer assay. In all other patients, computed tomographic pulmonary angiography or ventilation–perfusion lung scanning was performed.

RESULTS 
A total of 560 patients (mean age, 76 years) were included in the study. A diagnosis of pulmonary embolism was ruled out in 330 of the 560 patients (58.9%) on the basis of the combination of a low pretest clinical probability of pulmonary embolism and negative d-dimer assay. Among the remaining 230 patients, pulmonary embolism was identified in 97 (42.2%). In the entire cohort, the prevalence of pulmonary embolism was 17.3% (95% confidence interval, 14.2 to 20.5). Evidence of an embolus in a main pulmonary or lobar artery or evidence of perfusion defects larger than 25% of the total area of both lungs was found in 61 patients. Pulmonary embolism was identified in 45 of the 355 patients (12.7%) who had an alternative explanation for syncope and in 52 of the 205 patients (25.4%) who did not.

CONCLUSIONS 
Pulmonary embolism was identified in nearly one of every six patients hospitalized for a first episode of syncope. (Funded by the University of Padua; PESIT ClinicalTrials.gov number, NCT01797289.)

--------------------------------------------------------------------------------------------------------------------------

Syncope is defined as a transient loss of consciousness that has a rapid onset, short duration, and spontaneous resolution and is believed to be caused by temporary cerebral hypoperfusion. (1-3) According to current classifications, syncope can be neurally mediated (i.e., vasovagal, situational, or carotid-sinus syncope), can be caused by orthostatic hypotension (i.e., drug-induced hypotension or hypotension due to primary or secondary autonomic failure or due to volume depletion), or can have a cardiovascular origin (i.e., arrhythmias, structural cardiovascular diseases, or pulmonary embolism). (1)

Although pulmonary embolism is included in the differential diagnosis of syncope in most textbooks, rigorously designed studies to determine the prevalence of pulmonary embolism among patients hospitalized for syncope are lacking. Indeed, current international guidelines, including those from the European Society of Cardiology and the American Heart Association, pay little attention to establishing a diagnostic workup for pulmonary embolism in these patients. (1,2) Hence, when a patient is admitted to a hospital for an episode of syncope, pulmonary embolism — a potentially fatal disease that can be effectively treated — is rarely considered as a possible cause. In this study, we used a systematic diagnostic workup to assess the prevalence of pulmonary embolism in a large number of patients who were hospitalized for a first episode of syncope, regardless of whether there were potential alternative explanations for the syncope.


Methods
Study Design and Oversight
This was a cross-sectional study that was aimed at determining the prevalence of pulmonary embolism among patients older than 18 years of age who were hospitalized for a first episode of syncope. The study was designed by the first and last authors. The first author vouches for the completeness and accuracy of the data and analyses and for the fidelity of the study to the protocol. The protocol was approved by the institutional review board at each participating hospital. Syncope was defined as a transient loss of consciousness with rapid onset, short duration (i.e., <1 minute), and spontaneous resolution, with obvious causes such as epileptic seizure, stroke, and head trauma ruled out. (1-3)

All patients with syncope who visited the emergency department and were admitted to the medical ward of 1 of 11 participating general hospitals (2 academic and 9 nonacademic hospitals, each serving more than 100,000 inhabitants) were potentially eligible for enrollment in the study. Reasons for hospital admission were trauma related to falls, severe coexisting conditions, failure to identify an explanation for the syncope, or a high probability of cardiac syncope on the basis of the Evaluation of Guidelines in Syncope Study score. (4) Patients were excluded if they had had previous episodes of syncope, if they were receiving anticoagulation therapy, or if they were pregnant. All the patients provided written informed consent.

Study Assessments
All study assessments were completed within 48 hours after a patient was admitted to a hospital, as specified in the study protocol. All the patients were interviewed and evaluated by trained study physicians, who were investigators in the Pulmonary Embolism in Syncope Italian Trial (PESIT). The workup to be performed for each patient was prespecified in the study protocol and was based on the 2014 guidelines of the European Society of Cardiology. (5)

A medical history was obtained that included the presence of prodromal symptoms of autonomic activation (sweating, pallor, or nausea), the presence of known cardiac disease, recent bleeding, causes of volume depletion or venous pooling, and recent exposure to new or stronger hypotensive drugs or drugs that could potentially cause bradycardia or tachycardia. In addition, study physicians asked patients about symptoms (pain and swelling) in their legs and recorded the presence of risk factors for venous thromboembolism, including recent surgery, trauma, or infectious disease within the previous 3 months; ongoing hormonal treatment; prolonged immobilization of 1 week or longer; active cancer (i.e., recurrent or metastasized cancer or cancer that had been treated with chemotherapy or radiotherapy in the previous 6 months); and history of venous thromboembolism.

Patients were evaluated for the presence of arrhythmias, tachycardia (i.e., heart rate >100 beats per minute), valvular heart disease, hypotension (i.e., systolic blood pressure <110 mm Hg), autonomic dysfunction (as assessed by measuring blood pressure and pulse rate in the arms and legs with the patient in a supine and an upright position), tachypnea (i.e., respiratory rate >20 breaths per minute), and swelling or redness of the legs. All patients underwent chest radiography, electrocardiography, arterial blood gas testing, and routine blood testing that included a d-dimer assay. Further diagnostic workup included carotid sinus massage, tilt testing, echocardiography, and 24-hour electrocardiography recording, if applicable. Soon after hospital admission, patients received prophylaxis for venous thromboembolism, if indicated clinically. (6)

Ascertainment of Pulmonary Embolism
The presence or absence of pulmonary embolism was assessed with the use of a validated algorithm that was based on pretest clinical probability and the result of the d-dimer assay. (7) The d-dimer level was measured by the quantitative assay used routinely in each participating center; the cutoff for a positive result versus a negative result ranged between 250 and 500 μg per milliliter, depending on the manufacturer’s instructions. The pretest clinical probability of pulmonary embolism was defined according to the simplified Wells score, which classifies pulmonary embolism as being “likely” or “unlikely”(8).

In the patients who had a low (“unlikely”) pretest clinical probability and a negative d-dimer assay, no further testing was performed and a diagnosis of pulmonary embolism was ruled out. In patients who had a high (“likely”) pretest clinical probability, a positive d-dimer assay, or both, computed tomographic pulmonary angiography or ventilation–perfusion lung scanning (in the case of patients with severe renal impairment or allergy to contrast material) was performed. The criterion for the presence of pulmonary embolism was an intraluminal filling defect on computed tomography or a perfusion defect of at least 75% of a segment with corresponding normal ventilation. (9,10) In the event that a patient died before the completion of this diagnostic algorithm, an autopsy was requested. In patients with pulmonary embolism, the thrombotic burden was assessed by a central adjudication committee through identification of the most proximal location of the embolus on the computed tomographic scan or measurement of the severity of the perfusion defect on the ventilation–perfusion
lung scan. (11)

Statistical Analysis
On the basis of pilot data (6 of 50 patients who were admitted to a hospital for syncope had pulmonary embolism), we assumed a prevalence of pulmonary embolism of 10 to 15% among patients with a first episode of syncope. To obtain a two-sided 95% confidence interval of 2.5% for the prevalence of pulmonary embolism, we estimated that a sample size of 550 patients would be required. All participating centers were asked to enroll patients until the estimated sample size was reached.

The prevalence of pulmonary embolism and the associated 95% confidence interval were calculated for the entire group of patients and for relevant subgroups. To compare the baseline characteristics between patients with and those without pulmonary embolism, we used the chisquare test for categorical variables and Student’s t-test for continuous variables. Odds ratios with 95% confidence intervals were calculated with the use of logistic regression. The 95% confidence intervals and P values were calculated according to the normal approximation of the binomial distribution. No adjustments were made for multiple testing. All calculations were performed with the use of SPSS software, version 22.0 (SPSS).

Results
Patients
From March 2012 through October 2014, a total of 2584 patients visited the emergency departments of the 11 study hospitals (see the Supplementary Appendix, available with the full text of this article at NEJM.org) because of syncope. A total of 1867 of the 2584 patients (mean age, 54 years; range, 16 to 79) were either not admitted to the hospital or declined hospitalization. Of the 717 patients (27.7%) who were admitted, 157 (21.9%) were excluded from the study because they were receiving ongoing anticoagulation therapy (118 patients, 82 of whom were receiving it for atrial fibrillation and 36 for other reasons), had had previous episodes of syncope (35 patients), or did not provide informed consent (4 patients). Hence, 560 patients with a first episode of syncope were included in the study. The main demographic and clinical characteristics of the patients are provided in Table 2. Most of the patients were elderly (>75% were ≥70 years of age). Clinical evidence suggested an explanation for syncope other than pulmonary embolism in 355 of the 560 patients (63.4%).

Prevalence of Pulmonary Embolism
In 330 of the 560 patients (58.9%), a diagnosis of pulmonary embolism was ruled out on the basis of the combination of low pretest clinical probability of pulmonary embolism and a negative d-dimer assay. Of the remaining 230 patients, 135 (58.7%) had a positive d-dimer assay only, 3 (1.3%) had a high pretest clinical probability of pulmonary embolism only, and 92 (40.0%) had both. In 229 of these patients, either computed tomography or ventilation–perfusion lung scanning was performed; in the case of 1 patient who died before objective testing could be performed, an autopsy was performed after permission had been obtained. Pulmonary embolism was diagnosed in 72 of the 180 patients (40.0%) who underwent computed tomography and in 24 of the 49 patients (49.0%) who underwent ventilation–perfusion scanning (see the Supplementary Appendix) and was the cause of death of the 1 patient in whom an autopsy was performed. Hence, pulmonary embolism was confirmed in 97 of the patients who had a positive d-dimer assay, a high pretest clinical probability, or both (42.2%; 95% confidence interval [CI], 35.8 to 48.6). In the entire cohort, the prevalence of pulmonary embolism was 17.3% (95% CI, 14.2 to 20.5).

Thrombotic Burden 
Among the 72 patients in whom pulmonary embolism was detected by computed tomography, the most proximal location of the embolus was a main pulmonary artery in 30 patients (41.7%), a lobar artery in 18 patients (25.0%), a segmental artery in 19 patients (26.4%), and a subsegmental artery in 5 patients (6.9%). Among the 24 patients in whom pulmonary embolism was detected by ventilation–perfusion lung scanning, the perfusion defect involved more than 50% of the area of both lungs in 4 patients (16.7%), 26 to 50% of the area of both lungs in 8 patients (33.3%), and 1 to 25% of the area of both lungs in the remaining 12 patients (50.0%). In the 1 patient who died, pulmonary embolism involved both main pulmonary arteries.

Additional Observations
Pulmonary embolism was detected in 52 of the 205 patients who had syncope of undetermined origin (25.4%; 95% CI, 19.4 to 31.3) and in 45 of the 355 patients who were regarded as having a potential alternative explanation for syncope (12.7%; 95% CI, 9.2 to 16.1). Of the latter 45 patients, 31 (68.9%) had a lobar or more proximal location of the thrombus on computed tomography or a perfusion defect of more than 25% of the area of both lungs on ventilation–perfusion scanning. The prevalence of tachypnea was higher among the patients with pulmonary embolism than among the patients without pulmonary embolism (occurring in 45.4% vs. 7.1% of the patients), as were the prevalences of tachycardia (in 33.0% vs. 16.2%), hypotension (in 36.1% vs. 22.9%), clinical signs or symptoms of deep-vein thrombosis (in 40.2% vs. 4.5%), previous venous thromboembolism (in 11.3% vs. 4.3%), and active cancer (in 19.6% vs. 9.9%). Of the 97 patients with pulmonary embolism, 24 (24.7%) had no clinical manifestations of the diagnosis, including tachypnea, tachycardia, hypotension, or clinical signs or symptoms of deep-vein thrombosis.


Discussion
Our study used a systematic workup for pulmonary embolism in a large series of patients who were hospitalized for a first episode of syncope and showed a high prevalence of pulmonary embolism among these patients; pulmonary embolism was confirmed in approximately one of every six patients (17.3%). Although the prevalence of pulmonary embolism was highest among patients who presented with syncope of undetermined origin (25% of patients), almost 13% of patients with potential alternative explanations for syncope had pulmonary embolism. Not surprisingly, patients with dyspnea, tachycardia, hypotension, or clinical signs or symptoms of deep-vein thrombosis were more likely to have pulmonary embolism, as were those with active cancer. However, the proportion of patients who did not have these features yet had an objective confirmation of pulmonary embolism was not negligible.

The unexpectedly high prevalence of pulmonary embolism among our patients with syncope contrasts with that reported elsewhere. (12-17) It should be noted, however, that in the few contemporary studies that involved patients presenting with syncope, diagnostic testing for pulmonary embolism was performed only in selected subgroups, which may have resulted in a potential underestimation of the prevalence of this vascular disorder. In contrast, our study involved consecutive patients, all of whom underwent a guidelines-based workup for pulmonary embolism (5) regardless of whether another explanation was suggested clinically. Our study also involved multiple centers, and the results across the centers were consistent, with the prevalence of pulmonary embolism ranging from 15 to 20% across centers.

Some methodologic issues in our study require comment. First, patients were included in the study if they were admitted to a medical ward after being examined in the emergency department for syncope, which was defined as full loss of consciousness for less than 1 minute, followed by spontaneous, complete resolution. As a consequence, this study did not include patients who were cared for on an ambulatory basis or patients who visited the emergency department but for whom hospitalization was not considered necessary. Second, syncope is a diagnostic challenge, because the diagnosis is based largely on the history of the patient, which could be supported by observations of bystanders who are usually not medically trained. In addition, there is often uncertainty about the causal relationship between an identified disorder (such as a self-terminating arrhythmia) and the episode of syncope. Third, all participating hospitals used a standardized protocol for the diagnostic workup of syncope that was based on international guidelines (1,2) but a specific workup was not mandated by the study protocol.

In addition, the study protocol specified that a diagnosis of pulmonary embolism should not affect the usual workup for syncope. Fourth, diagnostic imaging for pulmonary embolism was performed only in patients who had an elevated d-dimer level or a high pretest clinical probability of pulmonary embolism. Nevertheless, well-conducted clinical studies have shown conclusively that pulmonary embolism is highly unlikely in patients who have a low pretest clinical probability and a negative d-dimer assay. (7,8,18-22) Fifth, the study protocol did not mandate objective confirmation of deep vein thrombosis in symptomatic patients; thus, we are not aware of the rate of this complication among patients who reported pain or swelling in their legs. However, none of the patients who were included in the study spontaneously reported these symptoms or visited the emergency department because of these symptoms. Sixth, the search for other causes of syncope was left to the discretion of the attending physicians. Hence, other causes of syncope may have been underreported. This may have been partly responsible for the fact that a definite cause of the syncope could not be determined in 205 patients. Seventh, pulmonary embolism is unlikely in patients who have had multiple episodes of syncope and in patients who are receiving anticoagulation therapy; therefore, these patients were excluded from our study, and accordingly, our study results are not applicable to such patients. Finally, we did not collect information on treatment decisions and patient follow-up after completion of the diagnostic algorithm for pulmonary embolism because this was not a study objective.

Syncope is generally expected to occur in patients with pulmonary embolism if they have a sudden obstruction of the most proximal pulmonary arteries that leads to a transient depression in cardiac output. (23-25) In 49 of the 73 patients (67.1%) in our cohort who had pulmonary embolism that was diagnosed according to findings from computed tomography or autopsy, the most proximal location of the embolus was a main pulmonary artery or a lobar artery. Similarly, among the 24 patients who were assessed with ventilation–perfusion scanning, the perfusion defect was larger than 25% of the total lung area in 12 patients (50.0%). These findings suggest that, in at least half of the patients with pulmonary embolism in our study, the extent of thrombosis was large enough to produce an abrupt obstruction of the blood flow that would be likely to result in a sudden loss of consciousness. However, in approximately 40% of the patients, the extent of pulmonary vascular obstruction was smaller. Because there was no standard approach to the evaluation of syncope, a number of patients with small pulmonary emboli may have had syncope that was associated with another condition that was missed. However, other mechanisms may be involved in the occurrence of syncope once a pulmonary embolism has developed, such as vasodepressor or cardioinhibitory mechanisms. (26-28)

In addition, when a clot dislodges from the venous system and lodges in the pulmonary circulation, it may induce arrhythmias when it passes through the heart. Hence, even smaller clots could be a potential cause of syncope. Studies addressing the mechanisms that trigger syncope in patients who have limited obstruction of the pulmonary arteries are warranted. In conclusion, among patients who were hospitalized for a first episode of syncope and who were not receiving anticoagulation therapy, pulmonary embolism was confirmed in 17.3% (approximately one of every six patients). The rate of pulmonary embolism was highest among those who did not have an alternative explanation for syncope.

References: http://www.nejm.org/doi/full/10.1056/NEJMoa1602172#t=article


Questions:

1. True or False: In this study, the diagnosis of pulmonary embolism was ruled out in patients who had low pretest probability (defined by Wells score) in combination with a negative D-dimer.  

2. True or False: In this study the Pulmonary Embolism Rule-Out Criteria rule or “PERC” rule was also used to rule out pulmonary embolism.

3. First time syncope patients who could not be ruled out for PE initially, underwent which of the following diagnostic tests?

1. Computed tomographic pulmonary angiography
2. Ventilation-perfusion lung scanning
3. Chest X-ray
4. a and b
5. None of the above

4. In this study of hospitalized first time syncope patients, pulmonary embolism was identified in  _____% of the admitted cohort.

1. 1.2%
2. 5.3%
3. 10.2%
4. 17.3%
5. 25%

5. True or False: First time syncope patients who were discharged to home from the emergency department were NOT included in the study population.

Cardiac Arrest: A Treatment Algorithm for Emergent Invasive Cardiac Procedures in the Resuscitated Comatose Patient

Tanveer Rab, MD, Karl B. Kern, MD, Jacqueline E. Tamis-Holland, MD, Timothy D. Henry, MD, Michael McDaniel, MD, Neal W. Dickert, MD, PhD, Joaquin E. Cigarroa, MD, Matthew Keadey, MD, Stephen Ramee, MD, on behalf of the Interventional Council, American College of Cardiology

Abstract:

Patients who are comatose after cardiac arrest continue to be a challenge, with high mortality. Although there is an American College of Cardiology Foundation/American Heart Association Class I recommendation for performing immediate angiography and percutaneous coronary intervention (when indicated) in patients with ST-segment elevation myocardial infarction, no guidelines exist for patients without ST-segment elevation. Early introduction of mild therapeutic hypothermia is an established treatment goal. However, there are no established guidelines for risk stratification of patients for cardiac catheterization and possible percutaneous coronary intervention, particularly in patients who have unfavorable clinical features in whom procedures may be futile and affect public reporting of mortality. An algorithm is presented to improve the risk stratification of these severely ill patients with an emphasis on consultation and evaluation of patients prior to activation of the cardiac catheterization laboratory.

Over the past 30 years, significant advances have been made in resuscitation therapy for cardiac arrest victims, with improved survival and neurological outcomes (12) . The vast majority of adult cardiac arrests are associated with obstructive coronary artery disease (3) . Emergent coronary revascularization in appropriate patients, coupled with therapeutic hypothermia (TH) and hemodynamic support, has continued to improve outcomes (45). Therefore, the standard practice in many centers is to emergently activate the cardiac catheterization laboratory (CCL) in patients presenting with cardiac arrest, the majority being out-of-hospital cardiac arrests (OHCAs). This is particularly true in cardiac arrest patients with ST-segment elevation myocardial infarction (STEMI). Although the 2013 American College of Cardiology Foundation (ACCF)/American Heart Association (AHA) guidelines for the management of STEMI (6) have a Class I recommendation for performing immediate angiography and percutaneous coronary intervention (PCI) in comatose patients with STEMI after OHCA when indicated, there are no guidelines for comatose cardiac arrest patients without ST-segment elevation on electrocardiogram (STE). In patients with OHCA, 64% will be comatose, and the neurological status on presentation has a dramatic effect on subsequent mortality and mortality (7) . Mortality in post-cardiac arrest patients with STEMI who are awake and undergo successful PCI is only 5%, but it increases to 50% if patients are comatose (7).

Although PCI can offer important benefits to resuscitated patients who remain comatose, current quality metrics and public reporting programs have not recognized the expected high mortality rate in this population and may deincentivize appropriate care. Whereas door-to-balloon time (D2B) in OHCA patients is excluded from core measures, hospital and operator mortality are key performance metrics and are not excluded. In addition, insurance programs offer hospitals quality improvement programs with significant financial reward if the adjusted mortality rate after PCI is <1%. Therefore, public reporting of adverse outcomes in this high-risk population without adequate risk adjustment, coupled with financial incentives for hospitals with low PCI mortality, has created a significant misalignment of goals. There is concern in the interventional community that this may lead to risk-averse behavior, resulting in suboptimal care by not providing early cardiac catheterization to appropriate patients.

Risk Stratification

Early risk stratification of patients with OHCA in the emergency room and the recommendations for early angiography vary considerably amongst providers and institutions. Many regional STEMI systems include automatic activation of the CCL for all STEMI and OHCA patients. The role of first responders, emergency room doctors, and noncardiologists focuses on the process, rather than the appropriateness of the activation. Although many patients have improved outcomes with an early invasive approach, some patient subsets may not derive a benefit and may experience excess risk.

A strategy to reliably identify patients who benefit from early angiography and those who benefit from compassionate supportive care is clearly needed. An algorithm may assist front line clinicians in identifying appropriate cardiac arrest patients for emergent cardiac catheterization. Recently, our European colleagues published a comprehensive review delineating their approach to OHCA patients (8). Although our approach addresses the care of OHCA patients in the United States, we hope that continued universal dialogue and research will accelerate improved outcomes in this critically ill population. We propose an algorithm to best risk stratify cardiac arrest patients who are comatose on presentation for emergent CCL activation for coronary angiography and possible intervention.

Algorithm for Risk Stratification of Comatose Cardiac Arrest Patients

ACT = assessment, consultation, transport; CCL = cardiac catheterization laboratory; CPR = cardiopulmonary resuscitation; ECG = electrocardiography; LV = left ventricular; OHCA = out-of-hospital cardiac arrest; PCI = percutaneous coronary intervention; ROSC = return of spontaneous circulation; STEMI = ST-segment elevation myocardial infarction; TH = therapeutic hypothermia; TTM = targeted temperature management; VF = ventricular fibrillation.

Explanation of the Algorithm

The principal purpose of this algorithm is to provide an easily implementable aid in identifying appropriate care for all comatose survivors of cardiac arrest and to identify patients who are unlikely to receive substantial benefit from an early invasive approach.

Cardiac arrest, return of spontaneous circulation, and the comatose patient

This algorithm focuses on patients who have experienced OHCA and have achieved return of spontaneous circulation (ROSC), but remain comatose. Although an initial shockable rhythm, such as ventricular tachycardia (VT) or ventricular fibrillation (VF), improves the likelihood of ROSC (691011 and of a favorable outcome (12) , nonshockable rhythms may also be caused by coronary artery occlusion (12).

Successfully resuscitated comatose patients represent a heterogeneous population with a baseline survival rate of only 25%. With hypothermia and PCI, survival improves to 60%, with favorable neurological outcomes achieved in 86% of survivors (341012131415 ( Table 1 ). However, the presence of certain unfavorable features reduces the likelihood of a good outcome.

Targeted temperature management with mild TH and coronary angiography post-cardiac arrest

Early initiation of targeted temperature management (TTM) is critical and should neither delay nor interfere with an early invasive approach. TTM is the active control of systemic body temperature to limit tissue injury after ischemia-reperfusion conditions occurring from cardiac arrest. The use of mild TH has been demonstrated to improve survival and neurological outcomes when combined with PCI in patients with OHCA who remain comatose on presentation. One nonrandomized report found an associated 20% increase in mortality rate with every hour of delay in initiating cooling (12). In 2002, 2 randomized clinical trials found that lowering body temperature to 32°C to 34°C for 12 to 24 h in those still comatose after being resuscitated from VF OHCA improved survival and neurological function of survivors (1626). Recently, 2 other randomized clinical trials of TTM in post-resuscitated patients have found equally impressive survival rates, whether cooled to 33°C versus 36°C (24) or whether initiated in the field or after arrival at the hospital (29). During the decade after the original reports of TTM efficacy in post-cardiac arrest patients, clinical cohort studies signaled that the combination of early coronary angiography and TTM might produce the best outcomes in the resuscitated, but unconscious, critically ill population. There are now a total of 28 cohort studies of post-arrest STEMI patients who were comatose upon hospital arrival and therefore received TTM and coronary angiography. A summary of these data shows a survival to hospital discharge rate of 60%, with 86% of such survivors being neurologically intact ( Table 1 ).

The International Liaison Committee on Resuscitation included the following statement in their 2010 International Consensus on Cardiopulmonary Resuscitation and Emergency Cardiovascular Care Science with Treatment Recommendations, “Therapeutic hypothermia is recommended in combination with primary PCI, and should be started as early as possible, preferably before initiation of PCI” (3839) , and the AHA 2010 Guidelines for Cardiopulmonary Resuscitation and Emergency Cardiovascular Care states, “angiography and/or PCI need not preclude or delay other therapeutic strategies including therapeutic hypothermia” (39) . Within the last few years, both the European Society of Cardiology 2012 (40) and the ACCF/AHA 2013 (6) STEMI guidelines included a Class I recommendation for the use of TTM for STEMI patients who are resuscitated from cardiac arrest but remain comatose on arrival at the hospital (1437). There are numerous choices for cooling post-arrest patients, including simple ice packs, intravenous cold saline (1 to 2 l), surface temperature-regulating devices, intranasal spray devices, and intravascular catheter-based systems. Although some are more convenient, none have been shown superior to the others for patient outcomes (4142).

The application of TTM, particularly the maintenance of hypothermia, in those undergoing coronary angiography and PCI, has raised concerns about possible increased bleeding, particularly from vascular access sites, and the potential to increase stent thrombosis. Excess bleeding, seen in earlier studies of extreme hypothermia (<28°C) has not been seen with the TTM recommended in cardiac arrest (32°C to 36°C) (3443) . In the largest report to date, the Resuscitation Outcomes Consortium reported a severe bleeding incidence of 2.7% (106 of 3,981) among all OHCA survivors admitted to the hospital, with similar rates among those receiving TTM (2.7%; 42 of 1,566), those receiving early coronary angiography (3.8%; 29 of 765), and those receiving reperfusion therapy (3.1%; 22 of 705).

There have been 3 reports of increased early stent thrombosis in patients treated post-arrest with PCI while simultaneously being cooled (4445). All were relatively small series, with a total of 15 of 110 (13.6%) patients having an acute or subacute stent thrombosis. Possible mechanisms include increased platelet activation, poor absorption of antiplatelet agents, multiorgan failure with altered metabolism of antiplatelet/antithrombotic agents, and procoagulant effects. A fourth report found no increase in stent thrombosis among the post-cardiac arrest population receiving both PCI and TTM (2 of 77 = 2.6% vs. 30 of 1,377 = 2.2% in their nonarrested STEMI patients) (46) . The observed increase in stent thrombosis did not adversely affect long-term outcomes (4748).

12-lead electrocardiogram

A 12-lead electrocardiogram (ECG) should be performed within 10 min of arrival to identify patients who benefit from emergent angiography. This should be undertaken simultaneously with initiation of TH.

STEMI on the ECG

This defines the Class I recommendation for emergent catheterization laboratory activation in the ACCF/AHA guidelines (6) . There is substantial evidence demonstrating efficacy of early angiography and PCI in OHCA patients with STEMI (341314495051 . Nonetheless, if multiple unfavorable resuscitation features are present, the benefit/futility ratio of proceeding to the catheterization laboratory should be carefully considered.

No STEMI on the ECG

The presence of an identifiable culprit vessel is found in 33% of patients without STEMI. Approximately 70% of these culprit vessels are occluded (52) . Hence, emergent cardiac catheterization to define a possible ischemic culprit and to perform revascularization if indicated should be considered in these patients. There is great clinical variability in the management of these patients and a lack of consensus about the best approach to risk stratification and the role of early revascularization.

The acronym ACT implies assessment for unfavorable resuscitation features, a multidisciplinary teamconsultation , including the interventional cardiologist, and urgent transport to the CCL, once a decision is made to proceed with coronary angiography.

Unfavorable resuscitation features

The presence of unfavorable resuscitation features that adversely affect the procedural risk/survival benefit of PCI must be considered prior to reaching a decision to proceed with coronary angiography, especially when multiple unfavorable features are present.

Patients with unfavorable resuscitation features are less likely to benefit from coronary intervention. In these cases, individualized care and interventional cardiology consultation are strongly recommended. Multidisciplinary team members should include physicians from the emergency department, critical care unit, neurology, cardiology, and interventional cardiology. All of the following features are relative, and are not absolute predictors of poor outcomes:

• Unwitnessed arrest. Extended time without systemic circulation prior to the resuscitation effort is associated with a decreased ROSC rate (62) and decreased survival-to-discharge rate 63646566 . When successful ROSC and survival are achieved after unwitnessed arrest, the rate of favorable neurological outcome is less than in those with witnessed arrest (65).
• Initial rhythm non-VF. Although the presence of an initial shockable rhythm, such as VT or VF, improves the likelihood of ROSC (691011 and of a favorable outcome (12) after PCI to an acute culprit artery stenosis, severe coronary artery stenosis may also be present among patients with nonshockable rhythms (12) . Nonshockable rhythms are associated with worse short- and long-term outcomes 62636467) . Patients with an initial nonshockable rhythm that transforms into a shockable rhythm fare worse than those presenting with an initial shockable rhythm (68).
• No bystander cardiopulmonary resuscitation. Recent studies have confirmed that the lack of bystander cardiopulmonary resuscitation (CPR) is associated with poor long-term outcomes 626364 . A recent meta-analysis reported that bystanders witnessed 53% of cardiac arrests, but only 32% of cardiac arrests received CPR. Survival was 16.1% in those who received bystander CPR versus 3.9% in those who did not (2).
• Longer than 30 min to ROSC. Early data from in-hospital cardiac arrest patients found that when the resuscitation efforts exceeded 30 min, the survival to discharge was markedly decreased (69) . More recent data from OHCA has demonstrated similar results (12) . Kamatsu et al. (70) found the mean time to ROSC for those with favorable (Cerebral Performance Category 1 or 2) neurological function post-arrest to be 18 ± 15 min compared with 47 ± 18 min for those with unfavorable (Cerebral Performance Category 3, 4, or 5) neurological function (70) . Multiple logistic regression analysis showed a significant relationship with the time interval from receipt of the emergency call (911) to ROSC. A longer time to ROSC correlated with poor neurological outcome (odds ratio [OR]: 0.86; 95% confidence interval [CI]: 0.81 to 0.92; p < 0.001)(70).
•  Ongoing CPR. Whereas short CPR duration (e.g., <16 min) correlates with a favorable prognosis, continuous or ongoing CPR for >30 min, especially in the presence of unwitnessed arrest, has been shown to significantly reduce the chance of survival. Additionally, studies have shown that the duration of CPR is an independent predictor of poorer functional status after OHCA (71).
• Evidence of unresponsive hypoperfusion and microcirculatory failure (pH and lactate levels). Cardiac arrest leads to systemic hypoperfusion with a low flow state and microcirculatory failure resulting in tissue ischemia, anaerobic metabolism, and the development of lactic acidosis (72) . Normal lactate levels are <1 mmol/l and correspond to a pH of 7.40. A lactate level of 7 mmol/l corresponds to a pH of 7.2 (72) and suggests a very poor prognosis post-resuscitation. Lactic acidosis is independently associated with a 3-fold increase in mortality (72) secondary to multiorgan failure, including severe anoxic brain injury with poor neurological outcome (73).
• In the PROCAT (Parisian Region Out of hospital Cardiac ArresT) study with 435 cardiac arrest patients, there were 264 nonsurvivors, of whom 112 had a lactate level >7 mmol/l (332) . In post-cardiac arrest patients, a pH <7.2 reflects severe acidemia, with increased risk of left ventricular dysfunction (7475) and poor neurological recovery, whereas those with a pH >7.2 had a >3-fold chance of neurological recovery (7677).
• Severe lactic acidosis is present when the lactate level is >18 mmol/l, corresponding to a pH of 7.0(72) . In the CHEER (Refractory Cardiac Arrest Treated With Mechanical CPR, Hypothermia, ECMO and Early Reperfusion) trial, 14 of the 26 enrolled patients with cardiac arrest did not survive, and their deaths were associated with a pH of 6.8 (78).
• Age >85 years. Although age alone is not an exclusion criterion, it is a poor prognostic indicator and should be carefully assessed (including physiological age vs. true age) before an emergent cardiac catheterization is undertaken. Although controversial, age needs to be considered, as studies suggest a worse outcome with advanced age, particularly in octogenarians (311497779) . Of 179 post-cardiac arrest patients >75 years of age treated with TH/TTM and PCI, only 33% survived to discharge and only 28% attained good functional recovery (77) . In a large registry of patients >85 years of age, 60% failed to achieve ROSC and the mortality rate was 90% (80) . A recent abstract from a Danish registry reported a successful resuscitation rate of only 25% in octogenarians (mean age 85 years) after cardiac arrest, compared with 40% among younger patients. Those octogenarians who were successfully resuscitated had a 30-day survival of 19% (compared with 45% for younger patients). However, most who survived in both groups (75% and 85%, respectively) had good functional status (81) . These studies illustrate the substantial effect of age on survival, but did not address the specific effect of an early invasive approach. Given the potential challenges of delineating futility among high-risk patients, we recommend careful assessment of those >85 years of age, prior to emergent activation of CCL.
• End-stage renal disease on hemodialysis. Compared with the general population, patients with end-stage renal disease on dialysis are at increased risk for sudden cardiac arrest. Myocardial ischemia secondary to coronary artery disease is the primary cause of cardiac arrest. In addition to the usual triggers of cardiac arrest, hemodialysis patients may have electrolyte derangements resulting from fluid shifts or changes in pH, leading to the arrest. Survival rates for dialysis patients with cardiac arrest are dismal, with <15% of dialysis patients alive at 1 year 828384 . Of 729 patients who experienced cardiac arrest while in the hemodialysis unit, 310 (42.5%) were alive at 24 h, with only 80 survivors (11%) at 6 months (83) . There were 110 cardiac arrests at outpatient dialysis centers in Seattle, Washington, between 1990 and 2004. Only 51 patients (46%) were alive at 24 h, 26 (24%) survived to hospital discharge, and only 16 (15%) were alive at 1 year (84) . Recent reviews report mortality in excess of 60% in the first 48 h, with a 1-year mortality of 87% 858687.
• Noncardiac causes. Patients with cardiac arrest due to drugs, drowning, choking, acute stroke, respiratory failure, terminal cancer, and trauma are typically not appropriate candidates for emergent cardiac catheterization. Although many of these patients may have a reasonable prognosis, it is unlikely to be enhanced by early angiography.

Other comorbidities or contraindications to aggressive treatment

The effect on post-arrest outcomes of other comorbidities, such as advanced dementia, chronic ventilator dependence, respiratory failure, severe frailty and disability, and other multisystem illnesses are not well characterized within the published data. However, these conditions are likely to result in a poor outcome post-resuscitation and need to be taken into consideration. Moreover, many patients with these conditions have care plans and do not wish aggressive treatment, including resuscitation. If it becomes apparent during the evaluation that a patient did not want to be resuscitated, an invasive approach should not be undertaken (77).

Immediate Coronary Angiography in Patients Without STE on ECG

The majority of patients resuscitated from cardiac arrest do not have STE on post-arrest ECG (31230568889) . Although there is strong evidence to support immediate coronary angiography and PCI in patients with resuscitated cardiac arrest and STEMI, data supporting immediate coronary angiography in patients without STE is less clear.

Approximately one-fourth of patients without STE have an acute occlusion (2531525558) and nearly 60% have significant obstructive lesions (31430315556) ( Table 2 ). Clinical and electrocardiographic characteristics are poor predictors of the presence of an acutely occluded vessel. In 1 report (53) of 84 patients with cardiac arrest referred for coronary angiography, the presence of chest pain preceding the arrest and the presence of STE on ECG were the only independent predictors of an acute occlusion (OR: 4.0; 95% CI: 1.3 to 10.1; p = 0.016; and OR: 4.3; 95% CI: 1.6 to 2.0; p = 0.004, respectively). However, the positive and negative predictive values associated with the presence of 1 of these 2 factors were only 0.63 and 0.74, respectively. If both variables were present, the positive and negative predictive values were 0.87 and 0.61, respectively. More importantly, 11% of patients with an acute coronary occlusion did not have STE. These data suggest that coronary angiography remains the “gold standard” for the identification of a culprit artery that may benefit from early revascularization.

Observational studies have demonstrated that patients resuscitated from cardiac arrest and referred for early coronary angiography and/or PCI have better outcomes, as compared with patients who are conservatively treated post-arrest ( Table 3 ). In all but 1 of these reports, the examined population included patients with and without STE, which limits the ability to assess the prognostic effect of early angiography specifically in those patients without STE on ECG. The study by Hollenbeck et al. (25) is the only report to examine the effect of an early invasive strategy on neurological outcome and in-hospital survival in a group of patients resuscitated from cardiac arrest who did not have STE (25) on ECG. In this study of 269 comatose patients after cardiac arrest due to VF or VT, 122 patients (45%) underwent “early” cardiac catheterization (defined as a procedure performed immediately after hospital admission or during hypothermia treatment). As compared with late or no catheterization, early cardiac catheterization was associated with a lower adjusted OR for in-hospital mortality (OR: 0.35; 95% CI: 0.18 to 0.70; p = 0.003). Furthermore, long-term survival and a favorable neurological outcome on follow-up were significantly higher in the group of patients referred for early catheterization (60.0% vs. 40.4%, p = 0.005; and 60.0% vs. 39.7%, p = 0.004, respectively).

The data summarized in Table 3 support early coronary angiography in patients after cardiac arrest, irrespective of the presence or absence of STE. Although these results imply improved outcomes among patients referred for cardiac catheterization, they should be interpreted with caution due to the observational nature of the studies. Observational reports are frequently confounded and seldom adequately control for all factors affecting physicians’ decisions regarding management. Randomized controlled trials of early coronary angiography versus no or late coronary angiography in patients without STE after cardiac arrest are needed. Until then, we recommend proceeding with coronary angiography in appropriate patients. The principal goal of angiography is to define the coronary anatomy and identify culprit lesions that require urgent PCI. Coronary anomalies would also be noted. If angiography is considered, it should be done early after hospital presentation. In the study by Hollenbeck et al. (25) , early angiography was defined as a procedure within the first 24 h. We favor a quicker time to angiography, ideally as soon as possible after the initial triage and assessment for unfavorable features (as outlined in our algorithm), with an emphasis on emergent consultation by a multidisciplinary team. Patients unlikely to benefit from coronary angiography and possible PCI should be identified early and should not proceed to the catheterization laboratory. In patients referred for angiography, the decision to proceed with PCI should be on the basis of the angiographic findings, coupled with the hemodynamic and electrical status of the patient.

The Challenges of Public Reporting

Although several states have been publicly reporting PCI outcomes for years, the passage of the Patient Protection and Affordable Care Act in 2010 increased the focus on public reporting and quality improvement. The National Quality Forum recently endorsed risk-adjusted total in-hospital PCI mortality and 30-day all-cause risk-standardized PCI mortality with STEMI and/or cardiogenic shock for public reporting (91).

Although paved with noble intentions, public reporting of mortality can have unintended consequences, possibly promoting risk-averse behaviors that negatively affect the patients who potentially have the most to gain from the procedure (92) . Patients with OHCA and ROSC have an approximately 10-fold higher mortality rate than non–cardiac arrest patients with STEMI (5) . Furthermore, most of the mortality in this population is due to neurological complications or multiorgan failure, despite receiving appropriate care. Moreover, current risk modeling does not adequately adjust for these extremes of risk, and high volumes of cardiac arrest patients can adversely affect individual and institutional outcomes. This is particularly important in the context of lower-volume centers, where patients who are appropriately treated for OHCA can have an outsized effect on mortality rates. Public reporting inadvertently places clinicians in the difficult situation of having to choose between what may be in their patient’s interest and what may be best for their own quality metrics or for their hospital’s reported outcomes.

Importantly, there is little evidence that public reporting of mortality improves outcomes in PCI, especially for patients with OHCA. In fact, several studies evaluating the effect of public reporting on PCI mortality in New York, Massachusetts, and Pennsylvania suggest that risk-averse behaviors related to public reporting may actually negatively affect patient outcomes. The 3 public reporting states rank 42nd, 48th, and 50th for utilization of PCI for acute myocardial infarction, a guideline-supported indication (93) . Furthermore, the adjusted mortality for patients presenting with STEMI is 35% higher in states with public reporting compared with those without public reporting. This is, in part, related to lower utilization of angiography and PCI in patients with STEMI (61.8% vs. 68%; OR: 0.73; 95% CI: 0.59 to 0.89; p = 0.002), including patients with either cardiogenic shock or cardiac arrest (41.5% vs. 46.7%; OR: 0.79; 95% CI: 0.64 to 0.98; p = 0.03) compared with states that do not publicly report mortality outcomes. This lower utilization of revascularization and higher mortality for cardiac arrest patients in states with public reporting was echoed in an analysis of 84,121 patients from the National Inpatient Sample database (94) . Interestingly, in Massachusetts, the rates of PCI were similar to those in other nonreporting states prior to public reporting, but began to diverge after public reporting was implemented, strongly implicating public reporting in the decline in optimal care. Finally, being identified as a “negative outlier” in risk-adjusted mortality in Massachusetts has been associated with a significant decline in predicted mortality in subsequent years, suggesting that risk-averse behaviors led to the exclusion of critically ill patients (95).

Given these limitations in public reporting of PCI mortality, especially in patients at extreme risk, such as OHCA patients, we endorse the recommendations set forth in the scientific statement from the AHA:

“OHCA cases should be tracked but not publicly reported or used for overall PCI performance ranking, which would allow accountability for their management but would not penalize high-volume cardiac resuscitation centers (CRCs) for following the 2010 AHA Guidelines for CPR and ECC. Until an adequate risk adjustment model is created to account for the numerous out-of-hospital and in-hospital variables that impact survival more than the performance of PCI, we believe that categorizing OHCA STEMI-PCI cases separately from other STEMI-PCI cases should occur. These patients should not be included in public reporting” (5).

Ethical Issues

Ethical challenges are unavoidable in the care of acutely ill patients resuscitated from OHCA. Although these challenges cannot be eliminated, there are ways to maximize ethical decision-making regarding angiography/PCI in this context.

First, there is an ethical imperative for rigorous research to inform decisions, particularly given heterogeneous practice patterns and varied standards of care. Randomized trials for treatment of cardiac arrest and other critical illnesses are at times controversial, largely due to ethical challenges regarding informed consent (9697) . However, regulations exist to facilitate these trials under an exception from informed consent (9899) . These regulations balance the need for research with important protections for patients and communities, and trials conducted under these regulations have resulted in significant insights and improvements in cardiac arrest care(124).

Second, decision-making in treatment of OHCA requires confronting issues of futility. The proposed algorithm highlights factors that may help to define when angiography/PCI is most likely futile. For example, a combination of comorbidities, advanced age, and prolonged ischemia (as indicated by severe lactic acidosis or long resuscitative efforts) may signify a high enough chance of multiorgan failure or anoxic brain injury that the incremental benefit of restoring coronary perfusion is truly minimal. However, these predictors are imperfect, and it is unknown how many unfavorable resuscitation features result in futility. Moreover, there are no established thresholds for what chance of a favorable outcome warrants aggressive treatment. If an intervention improves expected survival with good neurological status from 10% to 20%, many physicians or patients would likely not consider it futile, although the prognosis remains dismal. In contrast, an improvement from 1% to 2% represents the same relative benefit, but is likely futile.

Determinations of futility problematically involve quantitative and qualitative assessments with marked heterogeneity among providers. Most importantly, they require judging the value of different outcomes 100101102103 . In cardiac arrest, these judgments must be made without discussing the issues directly with the patient. Three key ethical implications must be emphasized in the context of futility in OHCA: 1) first, the unavoidable need for clinical judgment; 2) the need for better data and prognostic tools; and 3) the need for transparent discussion at the practice and policy levels about what characterizes appropriate or futile care. These discussions should substantially inform policies regarding reporting practices and quality metrics.

Finally, post-arrest care must involve assessing the patient’s likely preferences. A key component of pausing to individualize care is an attempt to contact family or other proxy decision-makers to assess whether aggressive treatment is consistent with the patient’s values or preferences. If pre-existing advance directives or do not resuscitate orders are revealed, they should be respected. Perhaps most importantly, direct and honest communication with proxy decision-makers is essential to preserving transparency and trust, and maximizing compatibility of decisions with patients’ values and goals.

Conclusions

1. We propose an easily implementable algorithm to identify resuscitated comatose patients after cardiac arrest who are appropriate candidates for emergent coronary angiography.

2. Urgent consultation and evaluation by a multidisciplinary team, including the interventional cardiologist, should occur before the patient is transferred to the CCL.

3. Early initiation of TTM is strongly recommended.

4. We emphasize our viewpoint and explicitly recommend without reservation that PCI outcomes in cardiac arrest patients not be included in public reporting. A national platform for tracking outcomes of cardiac arrest patients undergoing PCI is needed and should distinguish patients with and without ST-segment elevation.

5. Randomized controlled trials of early PCI in post–cardiac arrest patients without ST-segment elevation are needed.

Full Article/References: http://content.onlinejacc.org

 Virtual Journal Club August 2016: CME Questions

1.     For comatose patients who suffer out of hospital cardiac arrest (OHCA) and return of spontaneous circulation (ROSC) with ST elevation MI (STEMI), the recommendation to proceed to percutaneous coronary intervention is ________.

 a. Class I

 b. Class II

 c. Class III

 d. Class IV

 e. Classless

2.     True of False: According to these authors, current quality metrics and public reporting programs, as well as financial incentives, have not recognized expected high mortality rate in out of hospital cardiac arrest patients. This creates a misalignment of goals and a potential ethical conflict.

3.     True or False: These authors propose an algorithm to help risk stratify cardiac arrest patients who are comatose on presentation.

4.     True or False: excess bleeding has been seen in more recent studies of targeted temperature management (TTM) at a ranges of 32 degrees to 36 degrees Celsius.

5.     Which of the following are listed as unfavorable resuscitation features in comatose out of hospital cardiac arrest patients who obtain return of spontaneous circulation:

      a.     Un-witnessed arrest
b.     Initial rhythm non-VF
c.     No bystander CPR
d.     Longer then 30 minutes to ROSC
e.     Ongoing CPR  (greater than 30 minutes)
f.      pH less than 7.2
g.     Lactate greater than 7
h.     Age greater than 85 years
i.      End stage renal disease on hemodialysis
j.      Non-cardiac causes
k.     All of the above

Ketamine as Rescue Treatment for Difficult-to-Sedate Severe Acute Behavioral Disturbance in the Emergency Department

Geoffrey Kennedy Isbister, MD, FACEM, Leonie A. Calver, PhD, Michael A. Downes, MBBS, FACEM, Colin B. Page, MBBS
Annals of Emergency Medicine
Volume 67, Issue 5, Pages 581-587.e1 (May 2016)
DOI: 10.1016/j.annemergmed.2015.11.028

Study objective
We investigate the effectiveness and safety of ketamine to sedate patients with severe acute behavioral disturbance who have failed previous attempts at sedation.

Methods
This was a prospective study of patients given ketamine for sedation who had failed previous sedation attempts. Patients with severe acute behavioral disturbance requiring parenteral sedation were treated with a standardized sedation protocol including droperidol. Demographics, drug dose, observations, and adverse effects were recorded. The primary outcome was the number of patients who failed to sedate within 120 minutes of ketamine administration or requiring further sedation within 1 hour.

Results
Forty-nine patients from 2 hospitals were administered rescue ketamine during 27 months; median age was 37 years (range 20-82 years); 28 were men. Police were involved with 20 patients. Previous sedation included droperidol (10 mg; 1), droperidol (10+10 mg; 33), droperidol (10+10+5 mg; 1), droperidol (10+10+10 mg; 11), and combinations of droperidol and benzodiazepines (2) and midazolam alone (1). The median dose of ketamine was 300 mg (range 50 to 500 mg). Five patients (10%; 95% confidence interval 4% to 23%) were not sedated within 120 minutes or required additional sedation within 1 hour. Four of 5 patients received 200 mg or less. Median time to sedation postketamine was 20 minutes (interquartile range 10 to 30 minutes; 2 to 500 minutes). Three patients (6%) had adverse effects, 2 had vomiting, and a third had a transient oxygen desaturation to 90% after ketamine that responded to oxygen.

Conclusion
Ketamine appeared effective and did not cause obvious harm in this small sample and is a potential option for patients who have failed previous attempts at sedation. A dose of 4 to 5 mg/kg is suggested, and doses less than 200 mg are associated with treatment failure.

Introduction
The sedation of agitated and aggressive patients in the emergency department (ED) and other acute care areas is a major problem for health care workers. Patients with acute behavioral disturbance may respond to verbal de-escalation or oral sedation, but a substantial proportion of this group requires parenteral sedation and mechanical restraint.1,2,3,4,5 The majority of these patients will be sedated or tranquilized with parenteral sedation with an antipsychotic or benzodiazepine.1,3,4,5,6 In a recent study, only 8% of patients were not sedated with 1 or 2 doses of droperidol, and only 3% after 3 doses.6 However, this small number of patients with acute behavioral disturbance who remain difficult to sedate despite multiple doses of parenteral medications are highly problematic.7 Although such patients are uncommon, they cause a significant disruption and danger to the ED and consume time and resources required for other patients.

There is limited evidence available on the effective management of patients with acute behavioral disturbance when standard approaches to sedation with sedating antipsychotics and benzodiazepines have failed. Clinical practice guidelines do not cover the treatment options to manage repeatedly failed sedation. A number of agents have been suggested, including barbiturates, propofol, sedating antihistamines (diphenhydramine or promethazine), dexmedetomidine, and ketamine,7,8, 9,10 but there is little evidence to support one over another. Some clinicians may simply opt to intubate the patient, which is a last resort, resource intensive, and fraught with potential complications. To our knowledge, no previous studies have explored an alternative management for failed sedation of acute behavioral disturbance.

The success of ketamine for the sedation of out-of-hospital patients11, 12 and those retrieved with psychiatric illness13 suggests it may be a useful medication for difficult-to-sedate patients in the ED. We hypothesized that ketamine may be a safe and useful agent for the management of difficult-to-sedate patients in the ED.

Goals of This Investigation
The aim of the study was to investigate the effectiveness and safety of ketamine in severely agitated and aggressive patients in the ED when other parenteral sedation had failed on at least 2 occasions.

Study Design and Setting
This was a subgroup analysis of difficult-to-sedate patients with severe acute behavioral disturbance, included from the Droperidol or Midazolam1 (DORM II) study,6 a prospective observational study of ED patients with acute behavioral disturbance who required parenteral sedation and physical restraint. Ethics approval was obtained from the Hunter New England Area Health Service Human Research Ethics Committee and the Metro South Health Service District Human Research Ethics Committee to cover all hospitals involved. Patients required immediate sedation for patient and staff safety, and because of the lack of the patients’ decision making capacity, consent was waived because medical treatment was provided as a duty of care.

Both this analysis and DORM II were observational studies of a clinical protocol in which ketamine (this analysis) or droperidol (DORM II) was administered as part of that protocol. They were not clinical trials, so a clinical trials notification was not required. Although this analysis used a subgroup of patients from the DORM II study, it was not simply a retrospective review of the DORM II data. Ketamine was introduced by clinicians at the 2 hospitals in December 2011, with the intention of prospectively analyzing patients who received ketamine.

Patients were included in this analysis from 2 adult metropolitan hospitals of the 6 hospitals involved in the DORM II study.6 These 2 hospitals were used because they are teaching hospitals that have clinical toxicology services providing advice on difficult-to-sedate patients, and the clinician investigators at these 2 hospitals made a decision to use ketamine if droperidol failed. In addition, both hospitals had the highest recruitment rate to DORM II, and in one there was consecutive recruitment of all cases. The first hospital is a medium-sized urban hospital with a tertiary toxicology unit and drug and alcohol service, and has twice the number of patients presenting with acute behavioral disturbance compared with most EDs.5 The second hospital is a large tertiary adult referral hospital with 60,000 presentations each year, has a dedicated toxicology service, and has a similar number of presentations with acute behavioral disturbance.

Selection of Participants
Patients (>16 years) with acute behavioral disturbance were recruited as part of the DORM II study from the 2 EDs if they required physical restraint and parenteral sedation and had a score of 2 to 3 on the Sedation Assessment Tool (Table E1, available online at http://www.annemergmed.com).14
Patients who could be settled (ie, Sedation Assessment Tool score 0 or 1) with verbal de-escalation or oral medication were excluded. Any patients who then remained agitated and aggressive after their initial sedation and received ketamine as additional sedation were included in this analysis. A standardized protocol was used for all patients with acute behavioral disturbance in the 2 EDs, which recommended two 10-mg doses of droperidol. However, in a minority of cases the protocol was not followed, so either fewer doses of droperidol were given or benzodiazepines were used. In some cases, after consultation with the on-call clinical toxicologist a third dose of droperidol was given before ketamine. The aim was to review ketamine as a rescue treatment, so we included all cases in which there was an initial failure of sedative medication. Patients were included from December 2011 until February 2014.

Interventions
In the DORM II study, all patients with acute behavioral disturbance were managed with a standardized protocol that included the administration of intramuscular sedation and the assessment of the level of sedation or agitation with the Sedation Assessment Tool.6, 14 The protocol recommended starting with a 10-mg dose of intramuscular droperidol for parenteral sedation, and if the patient is not sedated within 15 minutes a second dose of 10 mg droperidol was given. Patients who were not sedated after 30 minutes were discussed with the on-call clinical toxicologist at each site, and from December 2011 ketamine was used in the majority of cases as the third-line agent. The clinical toxicologists decided on an intramuscular dose of ketamine of 4 to 6 mg/kg, based on intramuscular use in other settings.10,15,16 Rarely, droperidol was not used as the first-line agent. This usually occurred when new junior or training medical staff did not consult the clinical toxicologist. In one case, only a single dose of droperidol was given before ketamine.

All patients were observed in a critical care area and had pulse rate, pulse oximetry, respiratory rate, and blood pressure recorded every 5 minutes for the first 20 minutes and then every 30 minutes for the next 2 to 4 hours. The level of agitation or sedation was measured with the Sedation Assessment Tool score.14 This score is used routinely in both EDs and assesses the degree of agitation or sedation as a score of 3 (physically violent) to –3 (unconscious).

Data Collection and Processing
All observations were recorded on a purpose-designed acute behavioral disturbance chart by the treating medical and nursing staff, which was part of the patient’s medical record. It included patient demographics (age and sex), reason for presentation, method of arrival, drug administration details (dose and timing), sedation scores, vital signs (pulse rate, blood pressure, respiratory rate, and oxygen saturations), and adverse effects. This information was then entered into a relational database (Microsoft Access 2010; Microsoft, Redmond, WA) by a single researcher who collected all cases for DORM II daily for data entry. This researcher identified all cases of ketamine administration prospectively as they were entered into the database.

Primary Data Analysis
The primary outcome for the study was the number of patients who failed to achieve sedation within 120 minutes of ketamine administration or required further sedation within 1 hour of ketamine. A number of other outcomes were included: (1) the time to sedation from the initial onset of acute behavioral disturbance, defined as a decrease in the Sedation Assessment Tool score by 2 levels or a score of zero or less; (2) the time to sedation after the administration of ketamine; and (3) any adverse effects (airway obstruction, oxygen saturation less than 90%, respiratory rate less than 12 breaths/min, new-onset arrhythmia, and systolic blood pressure less than 90 mm Hg). In addition, the change in pulse rate and blood pressure after ketamine administration was measured. Continuous variables were summarized as medians, interquartile ranges (IQRs), and ranges, and dichotomous outcomes were reported with 95% confidence intervals. All analyses and graphics were conducted with GraphPad Prism (version 6.03; GraphPad Software, San Diego, CA).

Results
There were 1,296 patients sedated as part of the DORM II protocol at the 2 hospitals during the 27-month period. Of these, 53 patients (4%) received ketamine as part of their sedation. Four of these patients received ketamine without any previous sedation, leaving 49 patients who received ketamine after the failure of previous parenteral sedation (Figure). Three of the 4 patients were sedated by ketamine, 2 at 10 minutes and 1 at 20 minutes. The fourth patient received only 30 mg ketamine and then a further 30 mg 85 minutes later and did not settle for 5 hours. One of the 4 patients had a dystonic reaction after droperidol, but none had any adverse effects after ketamine.

Flowchart of the patients recruited to DORM II and those included in this subgroup analysis. An estimate of the total number of patients with acute behavioral sedation is included according to the number reported in the DORM study and then randomized to parenteral sedation.
Of the 49 patients, there were 28 men (57%) and the median age was 37 years (range 20 to 82 years). Police were required to assist with the transport of 20 patients to the hospital (41%). Droperidol alone was administered before ketamine in 46 patients (10 mg [1], 10+10 mg [33], 10+10+10 mg [11], and 10+10+5 mg [1]), whereas a combination of droperidol, diazepam, and midazolam was given in 2 patients and midazolam alone in 1 (Table). The median dose of ketamine used was 300 mg (IQR 200 to 400 mg; range 50 to 500 mg).

Table

Five patients (10%; 95% confidence interval 4% to 23%) were not sedated within 120 minutes (1), required additional sedation within 1 hour (1), or both (3). The doses administered in these 5 patients were 100, 150, 150, 200, and 400 mg. One of the 5 patients receiving 200 mg ketamine remained severely agitated for 12 hours (overnight) and was given no further sedation. Unfortunately, the clinical toxicologist was not notified about this patient again until the following morning.
The median time to sedation from the onset of acute behavioral disturbance was 60 minutes (IQR 40 to 140 minutes; range 20 to 540 minutes). The median time to sedation postketamine was 20 minutes (IQR 10 to 30 minutes; range 2 to 500 minutes). Three patients were resedated with ketamine 4 to 24 hours after the initial dose.

There were adverse effects in 3 patients after ketamine (6%). Two patients had vomiting (one treated with intramuscular metoclopramide 10 mg) and the third had an episode of oxygen desaturation to 90% without airway obstruction 40 minutes after ketamine, which immediately responded to oxygen, with no further problems. No patient had laryngeal spasm. One patient developed hypotension before ketamine but after droperidol.

There were 43 patients who had a preadministration systolic blood pressure with a median of 130 mm Hg (IQR 115 to 146 mm Hg; range 100 to 195 mm Hg). Blood pressures were measured a median of 15 minutes (2 to 120 minutes) postadministration, with a median change in systolic blood pressure of +5 mm Hg (IQR –3 to 23 mm Hg; range –47 to 38 mm Hg). No patients were hypotensive and 3 patients had a systolic blood pressure greater than 180 mm Hg (181, 183, and 198 mm Hg), but all had blood pressure greater than 140 mm Hg before ketamine (181, 149, and 174 mm Hg). There were 45 patients with pre– and post–pulse rate measurements with a median change of 0 beats/min (IQR –13 to 11 beats/min; range –60 to 30 beats/min).

Limitations
Our study is limited by its sample size. Although ketamine administration was associated with no serious adverse events, larger samples would be required to reliably confirm its safety profile. The study had a relatively small heterogenous patient sample and the inability to prevent the occasional variation in the treatment protocol. Although the majority of patients had 2 administrations of droperidol before ketamine, this was not always the case, with some receiving 1 or 3 doses and some receiving benzodiazepines. However, the aim was to assess ketamine as a rescue medication, not only after droperidol.

There was also some variability in the timing between droperidol doses and ketamine administration. This meant that it is not possible to determine whether the ultimate sedation of the patient was a result of the ketamine, delayed response to the initial medication (mainly droperidol), or both. There is some support for the sedation being due to ketamine because the median total time to sedation (from the initial onset of acute behavioral sedation) was 55 minutes, which is much longer than the median time in the DORM II study of 20 minutes.6
However, the median time to sedation after ketamine administration in this analysis was 20 minutes, the same as the median time to sedation in DORM II, suggesting that the sedation was due to ketamine.

Another potential limitation was that we present a subgroup analysis of patients from the DORM II study, and therefore the outcomes and data collection were designed to assess droperidol and not ketamine. However, after the introduction of ketamine in the 2 hospitals, the investigators planned a priori to prospectively assess the safety and effectiveness of ketamine by using the DORM II study infrastructure.

The study environment of the ED is a limitation of our study, and caution should be exercised in regard to generalizing it to areas that do not have ready access to critical care monitoring and medical staff. It would not be appropriate for ketamine to be used for acute behavioral disturbance on general wards or in psychiatric settings. However, recent reports have demonstrated the safety and effectiveness of ketamine for sedation in other environments, such as out-of-hospital transport, as well as in retrieval of psychotic patients.11,12,13,17,18

Discussion
This study reports the clinical use of ketamine in 49 patients with severe acute behavioral disturbance who could not be sedated with high-dose droperidol or, in a few cases, droperidol and benzodiazepines. Only 10% of the 49 patients could not be sedated within 2 hours or required additional sedation, which is only a very small proportion of the initial 1,296 patients who required sedation. There were only 3 adverse effects, 2 minor and the other easily treated with oxygen.

The major reason for failure of ketamine appeared to be the use of smaller doses, with 4 of 5 patients receiving 200 mg or less. The aim was to use 4 to 6 mg/kg, but in some cases staff decided to administer only half of the dose because of concerns about oversedation. However, larger doses of ketamine are not necessarily associated with oversedation, although they are associated with other adverse effects such as emergence phenomena.19

Although “nightmares” and recovery agitation are commonly reported in adult patients when ketamine has been administered for other indications, this was not reported in our series.16,9 It is possible that it was difficult to distinguish agitation associated with ketamine from the agitation already present in the patient. However, in the majority of cases no more medication was given to the patient after ketamine, and the patient settled for more than an hour and in most cases woke normally. The other concern with ketamine is the well-reported increase in blood pressure and pulse rate after administration. However, in our study there were only minor increases in both after administration. This may be due to an excess of endogenous sympathomimetic substances being present because of the agitation, therefore limiting further release of endogenous monoamines by ketamine. Hopper et al also reported only minor increases in blood pressure and pulse rate.

Benzodiazepines are the other obvious choice for difficult-to-sedate patients. However, there is increasing evidence that benzodiazepines alone 3,21,22 and combinations of benzodiazepines and antipsychotics are associated with higher rates of adverse effects.1,6 Both the DORM and DORM II studies found a higher rate of adverse effects with airway obstruction, oxygen desaturation, and hypotension than antipsychotics alone.1,6  A systematic review found that the addition of a benzodiazepine to haloperidol for psychosis-induced aggression provided no additional benefit but was associated with increased risk of harm.23 The additional use of ketamine after droperidol for difficult-to-sedate patients with severe acute behavioral disturbance appears to provide a safer option in this patient group compared with the combination of droperidol and benzodiazepines.

There is one other recent study of ketamine in the ED by Hopper et al.20 This study reviewed 32 patients given ketamine for acute agitation from 459 patients given ketamine in the ED during a 7-year period. Ketamine was administered intravenously and intramuscularly, and in almost half of the patients it was given without previous sedative medication. The median dose administered intramuscularly was 200 mg, with an IQR of 150 to 200 mg. There was a much higher failure rate in this study, with 16 of 32 patients (50%) requiring further sedation within an hour. The results of this study support our suggestion that lower doses of ketamine are associated with a higher failure rate. In addition, Hopper et al20 reported a similar low adverse event rate.

The major difference between our study and that by Hopper et al20 was that ketamine was not administered as part of a standardized sedation protocol. Our study was undertaken as part of a much larger study of more than 1,000 patients requiring parenteral sedation. The majority of the patients were sedated with droperidol, which was safe and effective.6 We then focused on the small group of patients who were not sedated, and this study now provides evidence that ketamine is an appropriate third-line agent to be used in these patients. There is one recent study of 5 adolescents (aged 14 to 18 years) who were sedated with intramuscular or intravenous ketamine, which again had similar outcomes. Ketamine was administered initially in some cases or after other attempts had failed.

There are an increasing number of studies of ketamine use in the out-of-hospital setting for agitated and difficult-to-sedate patients.11,12,17,18 These include a variety of reasons to sedate, most commonly trauma, and a mixture of intravenous and intramuscular use. These studies provide further support for the safety of ketamine. However, a significant proportion of patients receiving out-of-hospital ketamine have to be intubated on arrival to the ED.12

Ketamine appears to be a reasonable third-line agent in the sedation of patients with acute behavioral disturbance. The recommended dose is 4 to 6 mg/kg, which should be administered intramuscularly in a critical care area. Further research is required to define its use in settings outside of the ED for severe behavioral disturbance.

The authors acknowledge the staff of the Calvary Mater Newcastle and Princess Alexandra Emergency Departments.

Appendix




References: http://www.annemergmed.com/article/S0196-0644(15)01562-0/fulltext#sec1




_________________________________________


Virtual Journal Club June 2016 Test Questions (CME available through Allina CME)

1.   1  This article notes that evidence is limited regarding the use of a second line agent when sedating antipsychotics and benzodiazepines have failed. Which of the following agents have been suggested:

a.      barbiturates

b.     propofol

c.      sedating antihistamines

d.     dexmedetomidate

e.      ketamine

f.       all of the above

2.     2 True or False: The aim of this study was to investigate the effectiveness and safety of ketamine in severely agitated and aggressive patients in the ED when other parenteral sedation had failed on at least 2 occasions. 

3.     3 This study was a subgroup analysis of the  __________ study.

a.      STORM I

b.     DORM II

c.      FORM III

d.     NORM IV

4.     4 The dose of Ketamine used was ___ to ___ mg/kg administered intramuscularly.

a.      1, 2

b.     2, 4

c.      3, 9

d.     4, 6   

e.      0, Infinity

5.     5 The following were reported as adverse reactions to ketamine in this study:

                  

a.      Vomiting

b.     Oxygen desaturation

c.      Laryngeal spasm

d.     Hypotension

e.      a and c