Reviews

Aortic stenosis (AS) is a common problem among aging patients,1 who often require surgical procedures. The medical consultant must determine whether the presence of a systolic murmur suggesting AS needs additional evaluation before the patient proceeds to surgery. This decision requires interpretation of cardiac murmurs, and understanding the natural history, pathophysiology, and risks of AS.

PATHOPHYSIOLOGY

Aortic stenosis is a progressive disease that leads to predictable impairment of cardiac responses to physiologic stresses of surgery. AS typically results from degenerative calcification or from a bicuspid aortic valve, both of which cause progressive constriction of left ventricular outflow.24 The heart compensates by left ventricular hypertrophy. Systolic ejection of blood across the stenotic valve requires more time than normal, leaving less time for diastolic refilling. Left ventricular hypertrophy creates a less compliant left ventricle that becomes dependent on left atrial contraction for optimal filling. Atrial fibrillation with loss of the atrial kick is particularly problematic for patients with AS and left ventricular hypertrophy. Thickened myocardium increases myocardial oxygen consumption and impairs myocardial perfusion. Myocardial oxygen demand in the hypertrophied ventricle results from increased systolic pressure on the ventricle, increased systolic contraction time, and increased muscle mass. Reduced capillary density in hypertrophied muscle, and diminished perfusion pressure because of a reduced aortic‐coronary pressure differential, impair myocardial perfusion. Shortened diastole allows less blood flow to the myocardium.

At rest, with a controlled heart rate and sinus rhythm to allow for left atrial contraction to enhance left ventricular filling, patients may tolerate significant AS. However, increased heart rate in response to physiologic stress reduces diastolic filling time, diminishes somewhat tenuous myocardial perfusion, and increases afterload.

Additionally, the left ventricle depends on adequate filling pressures; the hypertrophied ventricle is prone to reduced cardiac output because of reductions of preload caused by hypovolemia or venodilation. Venodilation has been a particular concern with epidural anesthesia, although recent studies suggest that this modality can be used safely.5 Many anesthetic agents reduce systemic blood pressure and thereby reduce the aortic‐coronary perfusion pressure gradient leading to reduced coronary blood flow. For surgical patients with significant AS, anesthetic management requires appropriate intravascular volume to optimize preload, heart rate control to allow adequate left ventricular filling along with time for coronary artery flow, and sufficient systemic blood pressure to maintain coronary artery blood flow.

IDENTIFYING AORTIC STENOSIS IN PREOPERATIVE PATIENTS AND JUDGING ITS SEVERITY

Many older patients are found to have a systolic murmur consistent with AS prior to surgery. The first step in evaluation is a detailed history to determine exercise capacity and to elicit any history of chest pain, heart failure symptoms, or syncope. A key question for the medical consultant is whether or not patients should have further evaluation of the murmur prior to surgery, typically starting with transthoracic echocardiography. Table 1 outlines echocardiographic criteria for grading AS severity. The history and physical exam inform the decision of whether to pursue echocardiography. Although it is not clear from the literature whether identification of AS by echocardiography improves outcomes (this question is unlikely to be addressed by randomized trials), anesthesiologists generally want to know if significant AS is present, as it impacts intraoperative monitoring and management. So the question then becomes the following: Can clinicians reliably exclude moderatesevere AS based on history and a careful cardiovascular exam?

For ruling in severe AS, effort syncope provides the highest positive predictive value; stenosis was found to be severe in all patients with a history of effort syncope in a sample of 67 patients with AS.6 The presence of a loud, late‐peaking systolic murmur or significant delay and decrease in the carotid upstroke, argue for severe AS.7 Etchells et al developed a simple decision rule for detecting moderatesevere AS (defined as an aortic valve area of 1.2 cm² or less, or a peak transvalvular gradient of 25 mmHg or more), based on a study of 162 inpatients who were examined by a senior medical resident and a general internist.8 If no murmur was heard over the right clavicle, AS was rare (1/69 [1.4%]; likelihood ratio (LR) 0.10 [95% confidence interval (CI) 0.020.44]). If there was a murmur radiating to the right clavicle with 3 to 4 associated findings (reduced second heart sound, reduced carotid volume, slow carotid upstroke, and murmur loudest in the second right intercostal space), moderatesevere AS was common (6/7 [86%]; LR 40 [95% CI 6.6239]).

Absence of radiation of a systolic murmur to the right carotid artery is a useful finding to exclude AS, with a negative likelihood ratio of 0.05 to 0.10.9 Although no single physical exam finding or combination of findings can reliably exclude hemodynamically significant AS when a systolic murmur radiates to the right neck, the combination of an early‐peaking, soft (grade 2 or less) systolic murmur, normal timing and upstroke of the carotids, and an audible aortic second sound substantially lessen the likelihood of severe AS. A recent study of 376 inpatients who underwent meticulous cardiac examination by a single investigator (blinded to the diagnosis in >96% of cases), followed by echocardiography, provides additional information about the operating characteristics of physical examination in determining the etiology of systolic murmurs.10 Murmurs heard diagonally across the chest from the right upper sternal border to the apex (broad apical‐base pattern) predicted increased aortic velocity that would be consistent with AS. Other findings that increased the likelihood of aortic valve disease included delayed carotid upstroke, absent second heart sound (S2), radiation to the clavicles and neck on both sides, and a humming quality to the murmur. This study concluded that the physical examination is not reliable in determining the severity of AS. While generally true, this study actually reveals that any pattern of murmur radiation other than the broad apical‐base pattern excluded severe AS entirely among 221 patients with murmurs, and excluded moderate AS in all but 3 of these patients.

A retrospective study of 3997 hip fracture patients evaluated 908 echocardiograms done to investigate cardiac murmurs detected during preoperative assessment.11 These echocardiograms detected 272 patients with AS that had not been previously diagnosed. Thirty patients had severe AS. Detection of AS prompted changes in anesthesia management. The authors argued for preoperative echocardiograms for all hip fracture patients in whom a murmur is detected.

In summary, no finding by history can exclude AS. However, if the murmur is not heard across the precordium and does not radiate to the clavicle or right neck, severe AS is very unlikely.10 For patients in whom the murmur suggests the possibility of severe AS, echocardiography is prudent.

PROGNOSIS OF ADVANCED AS

Symptomatic AS portends poor prognosis in the absence of aortic valve replacement. In a cohort of patients with severe AS who refused aortic valve replacement (AVR), patients survived a mean of 45 months after onset of angina, 27 months following onset of syncope, and only 11 months after the beginning of left heart failure.12 Recent studies further define the natural history of severe asymptomatic AS. A study of 128 consecutive patients with asymptomatic severe AS identified by echocardiography found 93% survival at 1 year, 91% at 2 years, and 87% at 4 years, suggesting a relatively benign prognosis.13 However, many patients developed symptoms during follow‐up and required aortic valve replacement. A larger study of 622 asymptomatic AS patients with aortic‐jet velocity greater than 4 m/s found that 82% of patients were free of cardiac symptoms after 1 year, but only 33% were free of cardiac symptoms or intervention at 5 years.14 Patients with asymptomatic, very severe AS, defined as peak aortic‐jet velocity of 5.0 m/s or greater have an even worse prognosis with an event‐free survival of 12% at 4 years and only 3% at 6 years.15

Although short‐term (1 to 5 years) prognosis for severe symptomatic AS is poor, and asymptomatic but severe AS also carries substantial risk, the major issue for the medical consultant evaluating patients prior to noncardiac surgery is the very short‐term perioperative risk imposed by AS. Put simply, will the patient survive surgery and the postoperative period of rehabilitation?

NONCARDIAC SURGERY AND AS

The evidence that AS increases risk of cardiac complications and cardiac death for patients undergoing noncardiac surgery is limited to retrospective studies. In the early 1960s, a retrospective study of cardiac risk among 766 patients found 10% mortality among 59 patients with an aortic valve abnormality.16 The 15 patients who underwent either intrathoracic or intra‐abdominal procedures did particularly poorly, with a mortality of 20%. As part of a large cohort study used to develop the first widely employed cardiac risk index for noncardiac surgery, Goldman et al found 13% (3/23 patients) cardiac mortality among patients with important valvular AS.17 In comparison, cardiac mortality among 978 patients without identified AS was 1.6% (16/978 patients).

More recent studies demonstrate lower perioperative mortality for AS patients. These studies are summarized in Table 2. A retrospective chart audit of all patients with AS who underwent noncardiac surgery, in Hamilton, Ontario, Canada between 1992 and 1994, identified 55 patients with a mean aortic valve area of 0.9 cm² and compared outcome to that of 55 randomly selected control patients.18 The investigators defined cardiac complications as onset of congestive heart failure, myocardial infarction within 7 postoperative days, dysrhythmias requiring cardioversion, unplanned or prolonged intensive care unit stay resulting from cardiac complications, and cardiac death. Cardiac complications occurred in 5 (9%) patients with AS and 6 (11%) control patients. There was 1 cardiac death among patients with AS.

A retrospective analysis of 108 patients with AS who underwent noncardiac surgery, at Erasmus Medical Center in The Netherlands between 1991 and 2000, provides insight regarding severity of stenosis and perioperative outcomes.19 Cardiac complications (cardiac death or nonfatal myocardial infarction within 30 days of surgery) occurred in 15/108 (14%) patients with AS, with the majority of these complications being cardiac deaths. A control group of 216 patients suffered a cardiac complication rate of 1.8%. Multivariate adjustment for other risk factors demonstrated an odds ratio of 5.2 (95% CI 1.617.0) for cardiovascular complication in patients with AS. Moderate AS was associated with 11% complication rate (10/92 patients), while severe stenosis was associated with 31% cardiac complications (5/16 patients). Table 3 summarizes cardiac risk among the patients in this study using the Revised Cardiac Risk Index.20

In contrast, the Mayo Clinic experience with severe AS (defined as an aortic valve area index <0.5 cm²/m² or mean transvalvular gradient >50 mmHg) suggested substantially lower complication rates among patients undergoing noncardiac surgery.21 In this series of 19 patients undergoing a variety of surgical procedures between 1988 and 1992, there were no intraoperative events, but 2 (11%) major postoperative events (1 myocardial infarction and 1 death related to multiorgan failure). The authors concluded that selected patients with severe AS could undergo noncardiac surgery with acceptable risk, and speculated that their experience of better outcomes was due to more aggressive intraoperative and postoperative monitoring and therapy, specifically prompt recognition and therapy of intraoperative hypotension.

A large database study identified 5149 patients undergoing noncardiac surgery, between 1996 and 2002, with a coexistent AS based on International Classification of Diseases, Ninth Revision (ICD‐9) discharge codes, and compared these patients to 10,284 controls.22 Acute myocardial infarction occurred more frequently among patients with AS (3.9% vs 2.0%, P < 0.001), but in‐hospital mortality was not more frequent (5.4% vs 5.7%). The association of perioperative nonfatal myocardial infarction persisted after adjustment for comorbidities. While the results of this study might be interpreted as showing no increase in perioperative mortality for patients with AS who are undergoing noncardiac surgery, there is no way to determine the severity of AS among study patients and endpoints were not uniformly sought, but rather, obtained by ICD‐9 reporting. A recent study of 30 patients with asymptomatic but severe AS, who underwent low‐ or intermediate‐risk noncardiac surgery, found that 30% of patients required intraoperative vasopressor use for hypotension, but there were no deaths, arrhythmias, or heart failure events.23

Summarizing evidence on noncardiac surgery for patients with AS, symptomatic AS is associated with an increased risk of adverse cardiac events in patients undergoing noncardiac surgery. Severe, asymptomatic AS increases risk of intraoperative hemodynamic instability and adverse perioperative cardiac outcomes, although mortality appears to be less than that associated with symptomatic AS.

ECHOCARDIOGRAPHY PRIOR TO NONCARDIAC SURGERY

There are no studies showing that preoperative echocardiograms lessen the perioperative risk for patients with AS. However, as noted earlier, physical examination alone is not adequate to determine the valvular abnormality causing a systolic murmur in many patients, nor is the exam accurate in determining severity of AS in many patients. Echocardiography clarifies both of these issues. Preoperative echocardiography should inform the approach to anesthesia and, for elective surgical procedures, should allow more accurate assessment of operative risk. Because aortic stenosis typically progresses in a relatively slow and steady fashion, demonstration of mild aortic stenosis by echocardiogram within the preceding few years is considered reassuring.

Emergent surgery (for example, exploratory laparotomy for a ruptured viscus) typically does not allow time for echocardiography prior to the procedure. If a previous echocardiogram is available, this may be useful in deciding the intensity of intraoperative monitoring. However, the presence of a suspicious systolic murmur should prompt careful hemodynamic monitoring and the anesthesiologist should be made aware of the suspicion of AS.

For patients with AS facing urgent surgery (for example, repair of a hip fracture), there is typically time to review previous echocardiograms and, if there has been no recent echocardiogram, it is reasonable to obtain one. The presence of severe AS by echocardiogram should prompt careful hemodynamic monitoring. Some anesthesiologists advocate the use of intraoperative transesophageal echocardiography (TEE) to monitor ventricular filling in patients with severe AS.2426 Intraoperative TEE provides real‐time assessment of the cause of left ventricular dysfunction and allows the anesthesiologist to manipulate hemodynamics to address the dysfunction. Intraoperative TEE prompted significant changes in therapy for 4 of 7 patients with AS in a larger cohort of noncardiac surgical patients monitored with TEE.27 A retrospective study of 123 intraoperative TEE examinations found an impact on management in 81% of patients undergoing noncardiac surgery, although only a small number of these patients had cardiac valvular abnormalities.28 Recent anesthesiology practice guidelines recommend that TEE be considered in patients who have cardiovascular pathology that might result in severe hemodynamic, pulmonary, or neurologic compromise.29 The anesthesiologist should decide potential utility of intraoperative TEE, but it is important that the consulting hospitalist be aware of this possible approach to hemodynamic monitoring. Intraoperative TEE requires specialized expertise and may not available in many hospitals.

For elective surgery, presence of a murmur suggestive of significant AS mandates echocardiography, unless there are study results available from the preceding year.30 Optimally, symptomatic AS should be addressed by aortic valve replacement prior to noncardiac surgery. For patients requiring semi‐urgent surgery but are deteriorating because of severe AS, temporizing percutaneous balloon valvuloplasty can be considered, but there are limited data and serious complication rates can be high.3133 Among 15 AS patients requiring noncardiac surgery but with a contraindication to valve replacement, 3 experienced ventricular perforation during percutaneous balloon valvuloplasty, with 1 death.31 In another series of 7 patients, there were no complications of the valvuloplasties, and all 7 patients underwent uncomplicated noncardiac surgery under general anesthesia thereafter.33

In the absence of interventions to improve cardiac hemodynamics, patients could proceed to necessary noncardiac surgery, understanding the high risk of mortality and morbidity (Table 2). These patients should have careful perioperative hemodynamic monitoring and could be considered for intraoperative TEE if available.

Patients with asymptomatic but severe AS can proceed to low‐ or moderate‐risk surgical procedures without further intervention, but with appropriate hemodynamic monitoring. Those patients with asymptomatic but severe AS needing high‐risk surgery should consider valve replacement prior to surgery. In addition, we believe most patients with severe AS should have a cardiologist involved in their perioperative care.

CONCLUSIONS

In summary, patients with suspected AS who require noncardiac surgery need thoughtful consideration by the medical consultant. Careful cardiac examination should be performed on all patients prior to noncardiac surgery. If there is no precordial murmur radiating to the right carotid artery or right clavicle, and if there are no other signs (eg, delayed or reduced carotid upstroke, or absent or distant second heart sound) or symptoms (eg, history of angina, congestive heart failure, or exertional syncope or presyncope), then echocardiography performed for the purpose of discovering AS is not necessary. The majority of patients with a suggestive systolic murmur should be evaluated with echocardiography to provide more accurate prognostic estimates and to guide hemodynamic management during the operation. Patients with severe symptomatic AS are at particularly high risk of cardiac complications, and aortic valve replacement should take priority if the noncardiac surgery can be delayed.

Aortic stenosis (AS) is a common problem among aging patients,1 who often require surgical procedures. The medical consultant must determine whether the presence of a systolic murmur suggesting AS needs additional evaluation before the patient proceeds to surgery. This decision requires interpretation of cardiac murmurs, and understanding the natural history, pathophysiology, and risks of AS.

PATHOPHYSIOLOGY

Aortic stenosis is a progressive disease that leads to predictable impairment of cardiac responses to physiologic stresses of surgery. AS typically results from degenerative calcification or from a bicuspid aortic valve, both of which cause progressive constriction of left ventricular outflow.24 The heart compensates by left ventricular hypertrophy. Systolic ejection of blood across the stenotic valve requires more time than normal, leaving less time for diastolic refilling. Left ventricular hypertrophy creates a less compliant left ventricle that becomes dependent on left atrial contraction for optimal filling. Atrial fibrillation with loss of the atrial kick is particularly problematic for patients with AS and left ventricular hypertrophy. Thickened myocardium increases myocardial oxygen consumption and impairs myocardial perfusion. Myocardial oxygen demand in the hypertrophied ventricle results from increased systolic pressure on the ventricle, increased systolic contraction time, and increased muscle mass. Reduced capillary density in hypertrophied muscle, and diminished perfusion pressure because of a reduced aortic‐coronary pressure differential, impair myocardial perfusion. Shortened diastole allows less blood flow to the myocardium.

At rest, with a controlled heart rate and sinus rhythm to allow for left atrial contraction to enhance left ventricular filling, patients may tolerate significant AS. However, increased heart rate in response to physiologic stress reduces diastolic filling time, diminishes somewhat tenuous myocardial perfusion, and increases afterload.

Additionally, the left ventricle depends on adequate filling pressures; the hypertrophied ventricle is prone to reduced cardiac output because of reductions of preload caused by hypovolemia or venodilation. Venodilation has been a particular concern with epidural anesthesia, although recent studies suggest that this modality can be used safely.5 Many anesthetic agents reduce systemic blood pressure and thereby reduce the aortic‐coronary perfusion pressure gradient leading to reduced coronary blood flow. For surgical patients with significant AS, anesthetic management requires appropriate intravascular volume to optimize preload, heart rate control to allow adequate left ventricular filling along with time for coronary artery flow, and sufficient systemic blood pressure to maintain coronary artery blood flow.

IDENTIFYING AORTIC STENOSIS IN PREOPERATIVE PATIENTS AND JUDGING ITS SEVERITY

Many older patients are found to have a systolic murmur consistent with AS prior to surgery. The first step in evaluation is a detailed history to determine exercise capacity and to elicit any history of chest pain, heart failure symptoms, or syncope. A key question for the medical consultant is whether or not patients should have further evaluation of the murmur prior to surgery, typically starting with transthoracic echocardiography. Table 1 outlines echocardiographic criteria for grading AS severity. The history and physical exam inform the decision of whether to pursue echocardiography. Although it is not clear from the literature whether identification of AS by echocardiography improves outcomes (this question is unlikely to be addressed by randomized trials), anesthesiologists generally want to know if significant AS is present, as it impacts intraoperative monitoring and management. So the question then becomes the following: Can clinicians reliably exclude moderatesevere AS based on history and a careful cardiovascular exam?

For ruling in severe AS, effort syncope provides the highest positive predictive value; stenosis was found to be severe in all patients with a history of effort syncope in a sample of 67 patients with AS.6 The presence of a loud, late‐peaking systolic murmur or significant delay and decrease in the carotid upstroke, argue for severe AS.7 Etchells et al developed a simple decision rule for detecting moderatesevere AS (defined as an aortic valve area of 1.2 cm² or less, or a peak transvalvular gradient of 25 mmHg or more), based on a study of 162 inpatients who were examined by a senior medical resident and a general internist.8 If no murmur was heard over the right clavicle, AS was rare (1/69 [1.4%]; likelihood ratio (LR) 0.10 [95% confidence interval (CI) 0.020.44]). If there was a murmur radiating to the right clavicle with 3 to 4 associated findings (reduced second heart sound, reduced carotid volume, slow carotid upstroke, and murmur loudest in the second right intercostal space), moderatesevere AS was common (6/7 [86%]; LR 40 [95% CI 6.6239]).

Absence of radiation of a systolic murmur to the right carotid artery is a useful finding to exclude AS, with a negative likelihood ratio of 0.05 to 0.10.9 Although no single physical exam finding or combination of findings can reliably exclude hemodynamically significant AS when a systolic murmur radiates to the right neck, the combination of an early‐peaking, soft (grade 2 or less) systolic murmur, normal timing and upstroke of the carotids, and an audible aortic second sound substantially lessen the likelihood of severe AS. A recent study of 376 inpatients who underwent meticulous cardiac examination by a single investigator (blinded to the diagnosis in >96% of cases), followed by echocardiography, provides additional information about the operating characteristics of physical examination in determining the etiology of systolic murmurs.10 Murmurs heard diagonally across the chest from the right upper sternal border to the apex (broad apical‐base pattern) predicted increased aortic velocity that would be consistent with AS. Other findings that increased the likelihood of aortic valve disease included delayed carotid upstroke, absent second heart sound (S2), radiation to the clavicles and neck on both sides, and a humming quality to the murmur. This study concluded that the physical examination is not reliable in determining the severity of AS. While generally true, this study actually reveals that any pattern of murmur radiation other than the broad apical‐base pattern excluded severe AS entirely among 221 patients with murmurs, and excluded moderate AS in all but 3 of these patients.

A retrospective study of 3997 hip fracture patients evaluated 908 echocardiograms done to investigate cardiac murmurs detected during preoperative assessment.11 These echocardiograms detected 272 patients with AS that had not been previously diagnosed. Thirty patients had severe AS. Detection of AS prompted changes in anesthesia management. The authors argued for preoperative echocardiograms for all hip fracture patients in whom a murmur is detected.

In summary, no finding by history can exclude AS. However, if the murmur is not heard across the precordium and does not radiate to the clavicle or right neck, severe AS is very unlikely.10 For patients in whom the murmur suggests the possibility of severe AS, echocardiography is prudent.

PROGNOSIS OF ADVANCED AS

Symptomatic AS portends poor prognosis in the absence of aortic valve replacement. In a cohort of patients with severe AS who refused aortic valve replacement (AVR), patients survived a mean of 45 months after onset of angina, 27 months following onset of syncope, and only 11 months after the beginning of left heart failure.12 Recent studies further define the natural history of severe asymptomatic AS. A study of 128 consecutive patients with asymptomatic severe AS identified by echocardiography found 93% survival at 1 year, 91% at 2 years, and 87% at 4 years, suggesting a relatively benign prognosis.13 However, many patients developed symptoms during follow‐up and required aortic valve replacement. A larger study of 622 asymptomatic AS patients with aortic‐jet velocity greater than 4 m/s found that 82% of patients were free of cardiac symptoms after 1 year, but only 33% were free of cardiac symptoms or intervention at 5 years.14 Patients with asymptomatic, very severe AS, defined as peak aortic‐jet velocity of 5.0 m/s or greater have an even worse prognosis with an event‐free survival of 12% at 4 years and only 3% at 6 years.15

Although short‐term (1 to 5 years) prognosis for severe symptomatic AS is poor, and asymptomatic but severe AS also carries substantial risk, the major issue for the medical consultant evaluating patients prior to noncardiac surgery is the very short‐term perioperative risk imposed by AS. Put simply, will the patient survive surgery and the postoperative period of rehabilitation?

NONCARDIAC SURGERY AND AS

The evidence that AS increases risk of cardiac complications and cardiac death for patients undergoing noncardiac surgery is limited to retrospective studies. In the early 1960s, a retrospective study of cardiac risk among 766 patients found 10% mortality among 59 patients with an aortic valve abnormality.16 The 15 patients who underwent either intrathoracic or intra‐abdominal procedures did particularly poorly, with a mortality of 20%. As part of a large cohort study used to develop the first widely employed cardiac risk index for noncardiac surgery, Goldman et al found 13% (3/23 patients) cardiac mortality among patients with important valvular AS.17 In comparison, cardiac mortality among 978 patients without identified AS was 1.6% (16/978 patients).

More recent studies demonstrate lower perioperative mortality for AS patients. These studies are summarized in Table 2. A retrospective chart audit of all patients with AS who underwent noncardiac surgery, in Hamilton, Ontario, Canada between 1992 and 1994, identified 55 patients with a mean aortic valve area of 0.9 cm² and compared outcome to that of 55 randomly selected control patients.18 The investigators defined cardiac complications as onset of congestive heart failure, myocardial infarction within 7 postoperative days, dysrhythmias requiring cardioversion, unplanned or prolonged intensive care unit stay resulting from cardiac complications, and cardiac death. Cardiac complications occurred in 5 (9%) patients with AS and 6 (11%) control patients. There was 1 cardiac death among patients with AS.

A retrospective analysis of 108 patients with AS who underwent noncardiac surgery, at Erasmus Medical Center in The Netherlands between 1991 and 2000, provides insight regarding severity of stenosis and perioperative outcomes.19 Cardiac complications (cardiac death or nonfatal myocardial infarction within 30 days of surgery) occurred in 15/108 (14%) patients with AS, with the majority of these complications being cardiac deaths. A control group of 216 patients suffered a cardiac complication rate of 1.8%. Multivariate adjustment for other risk factors demonstrated an odds ratio of 5.2 (95% CI 1.617.0) for cardiovascular complication in patients with AS. Moderate AS was associated with 11% complication rate (10/92 patients), while severe stenosis was associated with 31% cardiac complications (5/16 patients). Table 3 summarizes cardiac risk among the patients in this study using the Revised Cardiac Risk Index.20

In contrast, the Mayo Clinic experience with severe AS (defined as an aortic valve area index <0.5 cm²/m² or mean transvalvular gradient >50 mmHg) suggested substantially lower complication rates among patients undergoing noncardiac surgery.21 In this series of 19 patients undergoing a variety of surgical procedures between 1988 and 1992, there were no intraoperative events, but 2 (11%) major postoperative events (1 myocardial infarction and 1 death related to multiorgan failure). The authors concluded that selected patients with severe AS could undergo noncardiac surgery with acceptable risk, and speculated that their experience of better outcomes was due to more aggressive intraoperative and postoperative monitoring and therapy, specifically prompt recognition and therapy of intraoperative hypotension.

A large database study identified 5149 patients undergoing noncardiac surgery, between 1996 and 2002, with a coexistent AS based on International Classification of Diseases, Ninth Revision (ICD‐9) discharge codes, and compared these patients to 10,284 controls.22 Acute myocardial infarction occurred more frequently among patients with AS (3.9% vs 2.0%, P < 0.001), but in‐hospital mortality was not more frequent (5.4% vs 5.7%). The association of perioperative nonfatal myocardial infarction persisted after adjustment for comorbidities. While the results of this study might be interpreted as showing no increase in perioperative mortality for patients with AS who are undergoing noncardiac surgery, there is no way to determine the severity of AS among study patients and endpoints were not uniformly sought, but rather, obtained by ICD‐9 reporting. A recent study of 30 patients with asymptomatic but severe AS, who underwent low‐ or intermediate‐risk noncardiac surgery, found that 30% of patients required intraoperative vasopressor use for hypotension, but there were no deaths, arrhythmias, or heart failure events.23

Summarizing evidence on noncardiac surgery for patients with AS, symptomatic AS is associated with an increased risk of adverse cardiac events in patients undergoing noncardiac surgery. Severe, asymptomatic AS increases risk of intraoperative hemodynamic instability and adverse perioperative cardiac outcomes, although mortality appears to be less than that associated with symptomatic AS.

ECHOCARDIOGRAPHY PRIOR TO NONCARDIAC SURGERY

There are no studies showing that preoperative echocardiograms lessen the perioperative risk for patients with AS. However, as noted earlier, physical examination alone is not adequate to determine the valvular abnormality causing a systolic murmur in many patients, nor is the exam accurate in determining severity of AS in many patients. Echocardiography clarifies both of these issues. Preoperative echocardiography should inform the approach to anesthesia and, for elective surgical procedures, should allow more accurate assessment of operative risk. Because aortic stenosis typically progresses in a relatively slow and steady fashion, demonstration of mild aortic stenosis by echocardiogram within the preceding few years is considered reassuring.

Emergent surgery (for example, exploratory laparotomy for a ruptured viscus) typically does not allow time for echocardiography prior to the procedure. If a previous echocardiogram is available, this may be useful in deciding the intensity of intraoperative monitoring. However, the presence of a suspicious systolic murmur should prompt careful hemodynamic monitoring and the anesthesiologist should be made aware of the suspicion of AS.

For patients with AS facing urgent surgery (for example, repair of a hip fracture), there is typically time to review previous echocardiograms and, if there has been no recent echocardiogram, it is reasonable to obtain one. The presence of severe AS by echocardiogram should prompt careful hemodynamic monitoring. Some anesthesiologists advocate the use of intraoperative transesophageal echocardiography (TEE) to monitor ventricular filling in patients with severe AS.2426 Intraoperative TEE provides real‐time assessment of the cause of left ventricular dysfunction and allows the anesthesiologist to manipulate hemodynamics to address the dysfunction. Intraoperative TEE prompted significant changes in therapy for 4 of 7 patients with AS in a larger cohort of noncardiac surgical patients monitored with TEE.27 A retrospective study of 123 intraoperative TEE examinations found an impact on management in 81% of patients undergoing noncardiac surgery, although only a small number of these patients had cardiac valvular abnormalities.28 Recent anesthesiology practice guidelines recommend that TEE be considered in patients who have cardiovascular pathology that might result in severe hemodynamic, pulmonary, or neurologic compromise.29 The anesthesiologist should decide potential utility of intraoperative TEE, but it is important that the consulting hospitalist be aware of this possible approach to hemodynamic monitoring. Intraoperative TEE requires specialized expertise and may not available in many hospitals.

For elective surgery, presence of a murmur suggestive of significant AS mandates echocardiography, unless there are study results available from the preceding year.30 Optimally, symptomatic AS should be addressed by aortic valve replacement prior to noncardiac surgery. For patients requiring semi‐urgent surgery but are deteriorating because of severe AS, temporizing percutaneous balloon valvuloplasty can be considered, but there are limited data and serious complication rates can be high.3133 Among 15 AS patients requiring noncardiac surgery but with a contraindication to valve replacement, 3 experienced ventricular perforation during percutaneous balloon valvuloplasty, with 1 death.31 In another series of 7 patients, there were no complications of the valvuloplasties, and all 7 patients underwent uncomplicated noncardiac surgery under general anesthesia thereafter.33

In the absence of interventions to improve cardiac hemodynamics, patients could proceed to necessary noncardiac surgery, understanding the high risk of mortality and morbidity (Table 2). These patients should have careful perioperative hemodynamic monitoring and could be considered for intraoperative TEE if available.

Patients with asymptomatic but severe AS can proceed to low‐ or moderate‐risk surgical procedures without further intervention, but with appropriate hemodynamic monitoring. Those patients with asymptomatic but severe AS needing high‐risk surgery should consider valve replacement prior to surgery. In addition, we believe most patients with severe AS should have a cardiologist involved in their perioperative care.

CONCLUSIONS

In summary, patients with suspected AS who require noncardiac surgery need thoughtful consideration by the medical consultant. Careful cardiac examination should be performed on all patients prior to noncardiac surgery. If there is no precordial murmur radiating to the right carotid artery or right clavicle, and if there are no other signs (eg, delayed or reduced carotid upstroke, or absent or distant second heart sound) or symptoms (eg, history of angina, congestive heart failure, or exertional syncope or presyncope), then echocardiography performed for the purpose of discovering AS is not necessary. The majority of patients with a suggestive systolic murmur should be evaluated with echocardiography to provide more accurate prognostic estimates and to guide hemodynamic management during the operation. Patients with severe symptomatic AS are at particularly high risk of cardiac complications, and aortic valve replacement should take priority if the noncardiac surgery can be delayed.

Aortic stenosis (AS) is a common problem among aging patients,1 who often require surgical procedures. The medical consultant must determine whether the presence of a systolic murmur suggesting AS needs additional evaluation before the patient proceeds to surgery. This decision requires interpretation of cardiac murmurs, and understanding the natural history, pathophysiology, and risks of AS.

PATHOPHYSIOLOGY

Aortic stenosis is a progressive disease that leads to predictable impairment of cardiac responses to physiologic stresses of surgery. AS typically results from degenerative calcification or from a bicuspid aortic valve, both of which cause progressive constriction of left ventricular outflow.24 The heart compensates by left ventricular hypertrophy. Systolic ejection of blood across the stenotic valve requires more time than normal, leaving less time for diastolic refilling. Left ventricular hypertrophy creates a less compliant left ventricle that becomes dependent on left atrial contraction for optimal filling. Atrial fibrillation with loss of the atrial kick is particularly problematic for patients with AS and left ventricular hypertrophy. Thickened myocardium increases myocardial oxygen consumption and impairs myocardial perfusion. Myocardial oxygen demand in the hypertrophied ventricle results from increased systolic pressure on the ventricle, increased systolic contraction time, and increased muscle mass. Reduced capillary density in hypertrophied muscle, and diminished perfusion pressure because of a reduced aortic‐coronary pressure differential, impair myocardial perfusion. Shortened diastole allows less blood flow to the myocardium.

At rest, with a controlled heart rate and sinus rhythm to allow for left atrial contraction to enhance left ventricular filling, patients may tolerate significant AS. However, increased heart rate in response to physiologic stress reduces diastolic filling time, diminishes somewhat tenuous myocardial perfusion, and increases afterload.

Additionally, the left ventricle depends on adequate filling pressures; the hypertrophied ventricle is prone to reduced cardiac output because of reductions of preload caused by hypovolemia or venodilation. Venodilation has been a particular concern with epidural anesthesia, although recent studies suggest that this modality can be used safely.5 Many anesthetic agents reduce systemic blood pressure and thereby reduce the aortic‐coronary perfusion pressure gradient leading to reduced coronary blood flow. For surgical patients with significant AS, anesthetic management requires appropriate intravascular volume to optimize preload, heart rate control to allow adequate left ventricular filling along with time for coronary artery flow, and sufficient systemic blood pressure to maintain coronary artery blood flow.

IDENTIFYING AORTIC STENOSIS IN PREOPERATIVE PATIENTS AND JUDGING ITS SEVERITY

Many older patients are found to have a systolic murmur consistent with AS prior to surgery. The first step in evaluation is a detailed history to determine exercise capacity and to elicit any history of chest pain, heart failure symptoms, or syncope. A key question for the medical consultant is whether or not patients should have further evaluation of the murmur prior to surgery, typically starting with transthoracic echocardiography. Table 1 outlines echocardiographic criteria for grading AS severity. The history and physical exam inform the decision of whether to pursue echocardiography. Although it is not clear from the literature whether identification of AS by echocardiography improves outcomes (this question is unlikely to be addressed by randomized trials), anesthesiologists generally want to know if significant AS is present, as it impacts intraoperative monitoring and management. So the question then becomes the following: Can clinicians reliably exclude moderatesevere AS based on history and a careful cardiovascular exam?

For ruling in severe AS, effort syncope provides the highest positive predictive value; stenosis was found to be severe in all patients with a history of effort syncope in a sample of 67 patients with AS.6 The presence of a loud, late‐peaking systolic murmur or significant delay and decrease in the carotid upstroke, argue for severe AS.7 Etchells et al developed a simple decision rule for detecting moderatesevere AS (defined as an aortic valve area of 1.2 cm² or less, or a peak transvalvular gradient of 25 mmHg or more), based on a study of 162 inpatients who were examined by a senior medical resident and a general internist.8 If no murmur was heard over the right clavicle, AS was rare (1/69 [1.4%]; likelihood ratio (LR) 0.10 [95% confidence interval (CI) 0.020.44]). If there was a murmur radiating to the right clavicle with 3 to 4 associated findings (reduced second heart sound, reduced carotid volume, slow carotid upstroke, and murmur loudest in the second right intercostal space), moderatesevere AS was common (6/7 [86%]; LR 40 [95% CI 6.6239]).

Absence of radiation of a systolic murmur to the right carotid artery is a useful finding to exclude AS, with a negative likelihood ratio of 0.05 to 0.10.9 Although no single physical exam finding or combination of findings can reliably exclude hemodynamically significant AS when a systolic murmur radiates to the right neck, the combination of an early‐peaking, soft (grade 2 or less) systolic murmur, normal timing and upstroke of the carotids, and an audible aortic second sound substantially lessen the likelihood of severe AS. A recent study of 376 inpatients who underwent meticulous cardiac examination by a single investigator (blinded to the diagnosis in >96% of cases), followed by echocardiography, provides additional information about the operating characteristics of physical examination in determining the etiology of systolic murmurs.10 Murmurs heard diagonally across the chest from the right upper sternal border to the apex (broad apical‐base pattern) predicted increased aortic velocity that would be consistent with AS. Other findings that increased the likelihood of aortic valve disease included delayed carotid upstroke, absent second heart sound (S2), radiation to the clavicles and neck on both sides, and a humming quality to the murmur. This study concluded that the physical examination is not reliable in determining the severity of AS. While generally true, this study actually reveals that any pattern of murmur radiation other than the broad apical‐base pattern excluded severe AS entirely among 221 patients with murmurs, and excluded moderate AS in all but 3 of these patients.

A retrospective study of 3997 hip fracture patients evaluated 908 echocardiograms done to investigate cardiac murmurs detected during preoperative assessment.11 These echocardiograms detected 272 patients with AS that had not been previously diagnosed. Thirty patients had severe AS. Detection of AS prompted changes in anesthesia management. The authors argued for preoperative echocardiograms for all hip fracture patients in whom a murmur is detected.

In summary, no finding by history can exclude AS. However, if the murmur is not heard across the precordium and does not radiate to the clavicle or right neck, severe AS is very unlikely.10 For patients in whom the murmur suggests the possibility of severe AS, echocardiography is prudent.

PROGNOSIS OF ADVANCED AS

Symptomatic AS portends poor prognosis in the absence of aortic valve replacement. In a cohort of patients with severe AS who refused aortic valve replacement (AVR), patients survived a mean of 45 months after onset of angina, 27 months following onset of syncope, and only 11 months after the beginning of left heart failure.12 Recent studies further define the natural history of severe asymptomatic AS. A study of 128 consecutive patients with asymptomatic severe AS identified by echocardiography found 93% survival at 1 year, 91% at 2 years, and 87% at 4 years, suggesting a relatively benign prognosis.13 However, many patients developed symptoms during follow‐up and required aortic valve replacement. A larger study of 622 asymptomatic AS patients with aortic‐jet velocity greater than 4 m/s found that 82% of patients were free of cardiac symptoms after 1 year, but only 33% were free of cardiac symptoms or intervention at 5 years.14 Patients with asymptomatic, very severe AS, defined as peak aortic‐jet velocity of 5.0 m/s or greater have an even worse prognosis with an event‐free survival of 12% at 4 years and only 3% at 6 years.15

Although short‐term (1 to 5 years) prognosis for severe symptomatic AS is poor, and asymptomatic but severe AS also carries substantial risk, the major issue for the medical consultant evaluating patients prior to noncardiac surgery is the very short‐term perioperative risk imposed by AS. Put simply, will the patient survive surgery and the postoperative period of rehabilitation?

NONCARDIAC SURGERY AND AS

The evidence that AS increases risk of cardiac complications and cardiac death for patients undergoing noncardiac surgery is limited to retrospective studies. In the early 1960s, a retrospective study of cardiac risk among 766 patients found 10% mortality among 59 patients with an aortic valve abnormality.16 The 15 patients who underwent either intrathoracic or intra‐abdominal procedures did particularly poorly, with a mortality of 20%. As part of a large cohort study used to develop the first widely employed cardiac risk index for noncardiac surgery, Goldman et al found 13% (3/23 patients) cardiac mortality among patients with important valvular AS.17 In comparison, cardiac mortality among 978 patients without identified AS was 1.6% (16/978 patients).

More recent studies demonstrate lower perioperative mortality for AS patients. These studies are summarized in Table 2. A retrospective chart audit of all patients with AS who underwent noncardiac surgery, in Hamilton, Ontario, Canada between 1992 and 1994, identified 55 patients with a mean aortic valve area of 0.9 cm² and compared outcome to that of 55 randomly selected control patients.18 The investigators defined cardiac complications as onset of congestive heart failure, myocardial infarction within 7 postoperative days, dysrhythmias requiring cardioversion, unplanned or prolonged intensive care unit stay resulting from cardiac complications, and cardiac death. Cardiac complications occurred in 5 (9%) patients with AS and 6 (11%) control patients. There was 1 cardiac death among patients with AS.

A retrospective analysis of 108 patients with AS who underwent noncardiac surgery, at Erasmus Medical Center in The Netherlands between 1991 and 2000, provides insight regarding severity of stenosis and perioperative outcomes.19 Cardiac complications (cardiac death or nonfatal myocardial infarction within 30 days of surgery) occurred in 15/108 (14%) patients with AS, with the majority of these complications being cardiac deaths. A control group of 216 patients suffered a cardiac complication rate of 1.8%. Multivariate adjustment for other risk factors demonstrated an odds ratio of 5.2 (95% CI 1.617.0) for cardiovascular complication in patients with AS. Moderate AS was associated with 11% complication rate (10/92 patients), while severe stenosis was associated with 31% cardiac complications (5/16 patients). Table 3 summarizes cardiac risk among the patients in this study using the Revised Cardiac Risk Index.20

In contrast, the Mayo Clinic experience with severe AS (defined as an aortic valve area index <0.5 cm²/m² or mean transvalvular gradient >50 mmHg) suggested substantially lower complication rates among patients undergoing noncardiac surgery.21 In this series of 19 patients undergoing a variety of surgical procedures between 1988 and 1992, there were no intraoperative events, but 2 (11%) major postoperative events (1 myocardial infarction and 1 death related to multiorgan failure). The authors concluded that selected patients with severe AS could undergo noncardiac surgery with acceptable risk, and speculated that their experience of better outcomes was due to more aggressive intraoperative and postoperative monitoring and therapy, specifically prompt recognition and therapy of intraoperative hypotension.

A large database study identified 5149 patients undergoing noncardiac surgery, between 1996 and 2002, with a coexistent AS based on International Classification of Diseases, Ninth Revision (ICD‐9) discharge codes, and compared these patients to 10,284 controls.22 Acute myocardial infarction occurred more frequently among patients with AS (3.9% vs 2.0%, P < 0.001), but in‐hospital mortality was not more frequent (5.4% vs 5.7%). The association of perioperative nonfatal myocardial infarction persisted after adjustment for comorbidities. While the results of this study might be interpreted as showing no increase in perioperative mortality for patients with AS who are undergoing noncardiac surgery, there is no way to determine the severity of AS among study patients and endpoints were not uniformly sought, but rather, obtained by ICD‐9 reporting. A recent study of 30 patients with asymptomatic but severe AS, who underwent low‐ or intermediate‐risk noncardiac surgery, found that 30% of patients required intraoperative vasopressor use for hypotension, but there were no deaths, arrhythmias, or heart failure events.23

Summarizing evidence on noncardiac surgery for patients with AS, symptomatic AS is associated with an increased risk of adverse cardiac events in patients undergoing noncardiac surgery. Severe, asymptomatic AS increases risk of intraoperative hemodynamic instability and adverse perioperative cardiac outcomes, although mortality appears to be less than that associated with symptomatic AS.

ECHOCARDIOGRAPHY PRIOR TO NONCARDIAC SURGERY

There are no studies showing that preoperative echocardiograms lessen the perioperative risk for patients with AS. However, as noted earlier, physical examination alone is not adequate to determine the valvular abnormality causing a systolic murmur in many patients, nor is the exam accurate in determining severity of AS in many patients. Echocardiography clarifies both of these issues. Preoperative echocardiography should inform the approach to anesthesia and, for elective surgical procedures, should allow more accurate assessment of operative risk. Because aortic stenosis typically progresses in a relatively slow and steady fashion, demonstration of mild aortic stenosis by echocardiogram within the preceding few years is considered reassuring.

Emergent surgery (for example, exploratory laparotomy for a ruptured viscus) typically does not allow time for echocardiography prior to the procedure. If a previous echocardiogram is available, this may be useful in deciding the intensity of intraoperative monitoring. However, the presence of a suspicious systolic murmur should prompt careful hemodynamic monitoring and the anesthesiologist should be made aware of the suspicion of AS.

For patients with AS facing urgent surgery (for example, repair of a hip fracture), there is typically time to review previous echocardiograms and, if there has been no recent echocardiogram, it is reasonable to obtain one. The presence of severe AS by echocardiogram should prompt careful hemodynamic monitoring. Some anesthesiologists advocate the use of intraoperative transesophageal echocardiography (TEE) to monitor ventricular filling in patients with severe AS.2426 Intraoperative TEE provides real‐time assessment of the cause of left ventricular dysfunction and allows the anesthesiologist to manipulate hemodynamics to address the dysfunction. Intraoperative TEE prompted significant changes in therapy for 4 of 7 patients with AS in a larger cohort of noncardiac surgical patients monitored with TEE.27 A retrospective study of 123 intraoperative TEE examinations found an impact on management in 81% of patients undergoing noncardiac surgery, although only a small number of these patients had cardiac valvular abnormalities.28 Recent anesthesiology practice guidelines recommend that TEE be considered in patients who have cardiovascular pathology that might result in severe hemodynamic, pulmonary, or neurologic compromise.29 The anesthesiologist should decide potential utility of intraoperative TEE, but it is important that the consulting hospitalist be aware of this possible approach to hemodynamic monitoring. Intraoperative TEE requires specialized expertise and may not available in many hospitals.

For elective surgery, presence of a murmur suggestive of significant AS mandates echocardiography, unless there are study results available from the preceding year.30 Optimally, symptomatic AS should be addressed by aortic valve replacement prior to noncardiac surgery. For patients requiring semi‐urgent surgery but are deteriorating because of severe AS, temporizing percutaneous balloon valvuloplasty can be considered, but there are limited data and serious complication rates can be high.3133 Among 15 AS patients requiring noncardiac surgery but with a contraindication to valve replacement, 3 experienced ventricular perforation during percutaneous balloon valvuloplasty, with 1 death.31 In another series of 7 patients, there were no complications of the valvuloplasties, and all 7 patients underwent uncomplicated noncardiac surgery under general anesthesia thereafter.33

In the absence of interventions to improve cardiac hemodynamics, patients could proceed to necessary noncardiac surgery, understanding the high risk of mortality and morbidity (Table 2). These patients should have careful perioperative hemodynamic monitoring and could be considered for intraoperative TEE if available.

Patients with asymptomatic but severe AS can proceed to low‐ or moderate‐risk surgical procedures without further intervention, but with appropriate hemodynamic monitoring. Those patients with asymptomatic but severe AS needing high‐risk surgery should consider valve replacement prior to surgery. In addition, we believe most patients with severe AS should have a cardiologist involved in their perioperative care.

CONCLUSIONS

In summary, patients with suspected AS who require noncardiac surgery need thoughtful consideration by the medical consultant. Careful cardiac examination should be performed on all patients prior to noncardiac surgery. If there is no precordial murmur radiating to the right carotid artery or right clavicle, and if there are no other signs (eg, delayed or reduced carotid upstroke, or absent or distant second heart sound) or symptoms (eg, history of angina, congestive heart failure, or exertional syncope or presyncope), then echocardiography performed for the purpose of discovering AS is not necessary. The majority of patients with a suggestive systolic murmur should be evaluated with echocardiography to provide more accurate prognostic estimates and to guide hemodynamic management during the operation. Patients with severe symptomatic AS are at particularly high risk of cardiac complications, and aortic valve replacement should take priority if the noncardiac surgery can be delayed.

Delirium is a syndrome of disturbance of consciousness, with reduced ability to focus, sustain, or shift attention, that occurs over a short period of time and fluctuates over the course of the day.1 It encompasses a variety of cognitive, behavioral, and psychological symptoms including inattention, short‐term memory loss, sleep disturbances, agitated behaviors, delusions, and visual hallucinations.2 Delirium complicates the care of 70% to 80% of mechanically ventilated patients in intensive care units (ICUs).3 Of 13 million patients aged 65 and older hospitalized in 2002, 10% to 52% had delirium at some point during their admission.4, 5

Patients experiencing delirium have a higher probability of death during their hospital stay, adjusted for age, gender, race, and comorbidities.3, 6, 7 They are more vulnerable to hospital‐acquired complications leading to prolonged ICU and hospital stay, new institutionalization, and higher healthcare costs.3, 6, 7 Even with such a range of poor outcomes, the rates of delirium recognition are low,8 resulting in inadequate management.9 There has been considerable growth in the number of articles published on delirium in recent years. Therefore, it is of value to provide a state‐of‐the‐art summary of robust evidence in the field to healthcare personnel and delirium investigators.

We systematically reviewed the literature to identify published systematic evidence reviews (SERs), which evaluated the evidence on delirium risk factors, diagnosis, pathogenesis, prevention, treatment, and outcomes. We then summarized the data from the methodologically sound SERs to provide the reader with a clinically oriented summary of delirium literature for patient care. We also identify current gaps in delirium literature, and present future directions for delirium investigators to design studies that will enhance delirium care.

DATA SOURCES AND REVIEW METHODS

The domains of risk factors, diagnosis, pathophysiology, prevention, treatment, and outcomes were selected a priori to capture all relevant SERs regarding delirium based on the framework suggested by the American Delirium Society task force.10 To maximize article retrieval, a 3‐step search strategy was applied. First, we searched the electronic database utilizing OVID Medline, PubMed, the Cochrane Library, and Cumulative Index of Nursing and Allied Health Literature (CINAHL) using the following delirium‐specific search terms: delirium, confusion, agitation, mental status change, inattention, encephalopathy, organic mental disorders, and disorientation. We combined the above terms with the following study design terms: technical report, systematic evidence review, systematic review, meta‐analysis, editorial, and clinical reviews. We limited our search to human subjects. We excluded studies that: a) enrolled patients aged <18; b) enrolled patients with current or past Diagnostic and Statistical Manual of Mental Disorders (DSM) Axis I psychotic disorders; c) did not have standardized delirium evaluation; d) evaluated alcohol or substance abuse‐related delirium; e) did not use a systematic search method for identifying delirium‐related articles; and f) evaluated delirium sub‐types. We searched articles published from January 1966 through April 2011. Second, a manual search of references of the retrieved papers plus an Internet search using Google Scholar was conducted to find additional SERs. Titles and abstracts were screened by 2 reviewers (B.A.K., M.Z.). Authors of the included studies were contacted as necessary. Third, a library professional at the Indiana University School of Medicine independently performed a literature search, and those results were compared with our search to retrieve any missing SERs.

The methodological quality of each SER was independently assessed by 2 reviewers (B.A.K., M.Z.) using the United States Preventive Services Task Force (USPSTF) Critical Appraisal for SER.11 This scale assesses parameters that are critical to the scientific credibility of an SER and categorizes the SER as poor, fair, or good (Table 1). The 2 reviewers (B.A.K., M.Z.) used a data extraction form to record the following information from each SER: primary author, publication year, number and type of studies, number of participants and their mean age, study population, method for delirium diagnosis, risk factors, preventive and therapeutic interventions, and outcomes. Any disagreement between reviewers in SER selection, data extraction, or SER appraisal was resolved through discussion with a third reviewer (M.A.B.). The conflicting findings among SERs were resolved by consensus and by including the findings from a good SER over a fair SER.

RESULTS

Our search yielded 76,060 potential citations, out of which we identified 38 SERs meeting our inclusion criteria (Table 2). Figure 1 outlines our search strategy. Based on the USPSTF criteria, 22 SERs graded as good or fair provided the data to establish our review.

Figure 1

DISCUSSION AND CLINICAL IMPLICATIONS

Our study identified age, cognitive impairment, depression, and mepiridine use for analgesia as risk factors for delirium in surgical patients. Drugs with anticholinergic properties were implicated in delirium development in both medical and surgical patients. The CAM has the best available data to be used as a diagnostic tool for delirium. Multicomponent interventions to prevent delirium occurrence are effective in a non‐cognitively impaired population, and low‐dose haloperidol prophylaxis decreases delirium duration and severity without affecting delirium incidence in hip surgery patients. There is no advantage of using atypical antipsychotics over haloperidol in treating delirium, and low‐dose haloperidol is as effective as a higher dose without unwarranted extrapyramidal side effects. Delirium carries a poor prognosis with an increased risk of death, institutionalization, and dementia.

Hospitals may benefit from implementing multicomponent strategies, focusing on at‐risk elderly medical and surgical patients, administered by a multidisciplinary team to reduce delirium incidence. For ICU physicians and administrators, development of sedation guidelines minimizing the use of benzodiazepines will decrease the risk of delirium development.

A structured approach in diagnosing delirium is required to maximize identification. Use of the CAM, based on best available data is recommended. However, the length of time in doing the CAM (more than 10 minutes with the requisite mental status examination) and insensitivity in nonexpert hands suggest a need for alternative screening tools. Haloperidol should be the preferred first‐line pharmacological therapy for delirium, with atypical antipsychotics reserved for patients with contraindications to haloperidol or those who are refractory to therapy with haloperidol. Figure 2 delineates a clinical model for delirium management derived from the findings in the Results section.

Figure 2

FUTURE RESEARCH DIRECTIONS

We identified multiple areas without clear guidelines that could provide opportunities for future research. A role for routine delirium screening can be clarified through a well‐designed delirium screening trial investigating the benefits of delirium screening, coupled with a multicomponent intervention versus usual care. Use of pharmacotherapy in delirium prevention needs to be explored further in a large randomized trial, with 3 arms to compare typical antipsychotics, atypical antipsychotics, and placebo in patients at risk for delirium with a primary outcome of delirium incidence. In regard to delirium treatment, a large randomized trial to compare haloperidol with atypical antipsychotics, with a placebo arm focusing not only on delirium duration and severity, but also on long‐term outcomes such as rehospitalizations, institutionalization, cognitive impairment, and mortality, is warranted. Figure 3 points out potential areas for researchers to investigate hypotheses generated by our review and thereby improve delirium care.

Figure 3

To our knowledge, our SER presents the first summary of SERs in delirium. Prior to this review, Michaud et al9 and National Institute for Health and Clinical Excellence48 published delirium guidelines, but in both of these guidelines, evidence was collected from a multitude of studies ranging in methodology from scientific review and meta‐analysis to observational studies, and the majority of recommendations were based on expert opinion. On the contrary, our review was limited to rigorously conducted SERs; hence, we utilized the highest level, critically appraised evidence to provide guidance to clinicians and researchers.

Limitations include a diverse group of studies with a heterogeneous population of patients, preventing pooling of results. We did not review each individual study included in the 38 SERs. We excluded non‐English language SERs, studies evaluating delirium subtypes, alcohol or substance abuse‐related delirium, or delirium associated with psychiatric disorders. As we only reviewed SERs, some notable studies not included in the SERs may have been missed.

CONCLUSION

Delirium among hospitalized patients is a common syndrome with a significant burden to the healthcare system and society. The field of delirium has seen considerable advances in diagnosis, prevention, and treatment over the last decade. Even with this advancement, there are still areas of uncertainty, such as: the benefits and costs of delirium screening; the benefits and harms of single or combined pharmacological agents for delirium prevention and treatment; the development of a set of reliable biomarkers for delirium diagnosis, prognosis, and response to therapy; the long‐term effect of delirium‐specific therapeutics on patients' cognitive, physical, and psychological functions; and the relationship between delirium and the development of Alzheimer's disease. As our understanding of delirium's impact on patients and healthcare improves, delirium should be identified as an indicator of poor long‐term prognosis, and should prompt immediate and effective evidence‐based management strategies, like any other critical illness.

Delirium is a syndrome of disturbance of consciousness, with reduced ability to focus, sustain, or shift attention, that occurs over a short period of time and fluctuates over the course of the day.1 It encompasses a variety of cognitive, behavioral, and psychological symptoms including inattention, short‐term memory loss, sleep disturbances, agitated behaviors, delusions, and visual hallucinations.2 Delirium complicates the care of 70% to 80% of mechanically ventilated patients in intensive care units (ICUs).3 Of 13 million patients aged 65 and older hospitalized in 2002, 10% to 52% had delirium at some point during their admission.4, 5

Patients experiencing delirium have a higher probability of death during their hospital stay, adjusted for age, gender, race, and comorbidities.3, 6, 7 They are more vulnerable to hospital‐acquired complications leading to prolonged ICU and hospital stay, new institutionalization, and higher healthcare costs.3, 6, 7 Even with such a range of poor outcomes, the rates of delirium recognition are low,8 resulting in inadequate management.9 There has been considerable growth in the number of articles published on delirium in recent years. Therefore, it is of value to provide a state‐of‐the‐art summary of robust evidence in the field to healthcare personnel and delirium investigators.

We systematically reviewed the literature to identify published systematic evidence reviews (SERs), which evaluated the evidence on delirium risk factors, diagnosis, pathogenesis, prevention, treatment, and outcomes. We then summarized the data from the methodologically sound SERs to provide the reader with a clinically oriented summary of delirium literature for patient care. We also identify current gaps in delirium literature, and present future directions for delirium investigators to design studies that will enhance delirium care.

DATA SOURCES AND REVIEW METHODS

The domains of risk factors, diagnosis, pathophysiology, prevention, treatment, and outcomes were selected a priori to capture all relevant SERs regarding delirium based on the framework suggested by the American Delirium Society task force.10 To maximize article retrieval, a 3‐step search strategy was applied. First, we searched the electronic database utilizing OVID Medline, PubMed, the Cochrane Library, and Cumulative Index of Nursing and Allied Health Literature (CINAHL) using the following delirium‐specific search terms: delirium, confusion, agitation, mental status change, inattention, encephalopathy, organic mental disorders, and disorientation. We combined the above terms with the following study design terms: technical report, systematic evidence review, systematic review, meta‐analysis, editorial, and clinical reviews. We limited our search to human subjects. We excluded studies that: a) enrolled patients aged <18; b) enrolled patients with current or past Diagnostic and Statistical Manual of Mental Disorders (DSM) Axis I psychotic disorders; c) did not have standardized delirium evaluation; d) evaluated alcohol or substance abuse‐related delirium; e) did not use a systematic search method for identifying delirium‐related articles; and f) evaluated delirium sub‐types. We searched articles published from January 1966 through April 2011. Second, a manual search of references of the retrieved papers plus an Internet search using Google Scholar was conducted to find additional SERs. Titles and abstracts were screened by 2 reviewers (B.A.K., M.Z.). Authors of the included studies were contacted as necessary. Third, a library professional at the Indiana University School of Medicine independently performed a literature search, and those results were compared with our search to retrieve any missing SERs.

The methodological quality of each SER was independently assessed by 2 reviewers (B.A.K., M.Z.) using the United States Preventive Services Task Force (USPSTF) Critical Appraisal for SER.11 This scale assesses parameters that are critical to the scientific credibility of an SER and categorizes the SER as poor, fair, or good (Table 1). The 2 reviewers (B.A.K., M.Z.) used a data extraction form to record the following information from each SER: primary author, publication year, number and type of studies, number of participants and their mean age, study population, method for delirium diagnosis, risk factors, preventive and therapeutic interventions, and outcomes. Any disagreement between reviewers in SER selection, data extraction, or SER appraisal was resolved through discussion with a third reviewer (M.A.B.). The conflicting findings among SERs were resolved by consensus and by including the findings from a good SER over a fair SER.

RESULTS

Our search yielded 76,060 potential citations, out of which we identified 38 SERs meeting our inclusion criteria (Table 2). Figure 1 outlines our search strategy. Based on the USPSTF criteria, 22 SERs graded as good or fair provided the data to establish our review.

Figure 1

DISCUSSION AND CLINICAL IMPLICATIONS

Our study identified age, cognitive impairment, depression, and mepiridine use for analgesia as risk factors for delirium in surgical patients. Drugs with anticholinergic properties were implicated in delirium development in both medical and surgical patients. The CAM has the best available data to be used as a diagnostic tool for delirium. Multicomponent interventions to prevent delirium occurrence are effective in a non‐cognitively impaired population, and low‐dose haloperidol prophylaxis decreases delirium duration and severity without affecting delirium incidence in hip surgery patients. There is no advantage of using atypical antipsychotics over haloperidol in treating delirium, and low‐dose haloperidol is as effective as a higher dose without unwarranted extrapyramidal side effects. Delirium carries a poor prognosis with an increased risk of death, institutionalization, and dementia.

Hospitals may benefit from implementing multicomponent strategies, focusing on at‐risk elderly medical and surgical patients, administered by a multidisciplinary team to reduce delirium incidence. For ICU physicians and administrators, development of sedation guidelines minimizing the use of benzodiazepines will decrease the risk of delirium development.

A structured approach in diagnosing delirium is required to maximize identification. Use of the CAM, based on best available data is recommended. However, the length of time in doing the CAM (more than 10 minutes with the requisite mental status examination) and insensitivity in nonexpert hands suggest a need for alternative screening tools. Haloperidol should be the preferred first‐line pharmacological therapy for delirium, with atypical antipsychotics reserved for patients with contraindications to haloperidol or those who are refractory to therapy with haloperidol. Figure 2 delineates a clinical model for delirium management derived from the findings in the Results section.

Figure 2

FUTURE RESEARCH DIRECTIONS

We identified multiple areas without clear guidelines that could provide opportunities for future research. A role for routine delirium screening can be clarified through a well‐designed delirium screening trial investigating the benefits of delirium screening, coupled with a multicomponent intervention versus usual care. Use of pharmacotherapy in delirium prevention needs to be explored further in a large randomized trial, with 3 arms to compare typical antipsychotics, atypical antipsychotics, and placebo in patients at risk for delirium with a primary outcome of delirium incidence. In regard to delirium treatment, a large randomized trial to compare haloperidol with atypical antipsychotics, with a placebo arm focusing not only on delirium duration and severity, but also on long‐term outcomes such as rehospitalizations, institutionalization, cognitive impairment, and mortality, is warranted. Figure 3 points out potential areas for researchers to investigate hypotheses generated by our review and thereby improve delirium care.

Figure 3

To our knowledge, our SER presents the first summary of SERs in delirium. Prior to this review, Michaud et al9 and National Institute for Health and Clinical Excellence48 published delirium guidelines, but in both of these guidelines, evidence was collected from a multitude of studies ranging in methodology from scientific review and meta‐analysis to observational studies, and the majority of recommendations were based on expert opinion. On the contrary, our review was limited to rigorously conducted SERs; hence, we utilized the highest level, critically appraised evidence to provide guidance to clinicians and researchers.

Limitations include a diverse group of studies with a heterogeneous population of patients, preventing pooling of results. We did not review each individual study included in the 38 SERs. We excluded non‐English language SERs, studies evaluating delirium subtypes, alcohol or substance abuse‐related delirium, or delirium associated with psychiatric disorders. As we only reviewed SERs, some notable studies not included in the SERs may have been missed.

CONCLUSION

Delirium among hospitalized patients is a common syndrome with a significant burden to the healthcare system and society. The field of delirium has seen considerable advances in diagnosis, prevention, and treatment over the last decade. Even with this advancement, there are still areas of uncertainty, such as: the benefits and costs of delirium screening; the benefits and harms of single or combined pharmacological agents for delirium prevention and treatment; the development of a set of reliable biomarkers for delirium diagnosis, prognosis, and response to therapy; the long‐term effect of delirium‐specific therapeutics on patients' cognitive, physical, and psychological functions; and the relationship between delirium and the development of Alzheimer's disease. As our understanding of delirium's impact on patients and healthcare improves, delirium should be identified as an indicator of poor long‐term prognosis, and should prompt immediate and effective evidence‐based management strategies, like any other critical illness.

Delirium is a syndrome of disturbance of consciousness, with reduced ability to focus, sustain, or shift attention, that occurs over a short period of time and fluctuates over the course of the day.1 It encompasses a variety of cognitive, behavioral, and psychological symptoms including inattention, short‐term memory loss, sleep disturbances, agitated behaviors, delusions, and visual hallucinations.2 Delirium complicates the care of 70% to 80% of mechanically ventilated patients in intensive care units (ICUs).3 Of 13 million patients aged 65 and older hospitalized in 2002, 10% to 52% had delirium at some point during their admission.4, 5

Patients experiencing delirium have a higher probability of death during their hospital stay, adjusted for age, gender, race, and comorbidities.3, 6, 7 They are more vulnerable to hospital‐acquired complications leading to prolonged ICU and hospital stay, new institutionalization, and higher healthcare costs.3, 6, 7 Even with such a range of poor outcomes, the rates of delirium recognition are low,8 resulting in inadequate management.9 There has been considerable growth in the number of articles published on delirium in recent years. Therefore, it is of value to provide a state‐of‐the‐art summary of robust evidence in the field to healthcare personnel and delirium investigators.

We systematically reviewed the literature to identify published systematic evidence reviews (SERs), which evaluated the evidence on delirium risk factors, diagnosis, pathogenesis, prevention, treatment, and outcomes. We then summarized the data from the methodologically sound SERs to provide the reader with a clinically oriented summary of delirium literature for patient care. We also identify current gaps in delirium literature, and present future directions for delirium investigators to design studies that will enhance delirium care.

DATA SOURCES AND REVIEW METHODS

The domains of risk factors, diagnosis, pathophysiology, prevention, treatment, and outcomes were selected a priori to capture all relevant SERs regarding delirium based on the framework suggested by the American Delirium Society task force.10 To maximize article retrieval, a 3‐step search strategy was applied. First, we searched the electronic database utilizing OVID Medline, PubMed, the Cochrane Library, and Cumulative Index of Nursing and Allied Health Literature (CINAHL) using the following delirium‐specific search terms: delirium, confusion, agitation, mental status change, inattention, encephalopathy, organic mental disorders, and disorientation. We combined the above terms with the following study design terms: technical report, systematic evidence review, systematic review, meta‐analysis, editorial, and clinical reviews. We limited our search to human subjects. We excluded studies that: a) enrolled patients aged <18; b) enrolled patients with current or past Diagnostic and Statistical Manual of Mental Disorders (DSM) Axis I psychotic disorders; c) did not have standardized delirium evaluation; d) evaluated alcohol or substance abuse‐related delirium; e) did not use a systematic search method for identifying delirium‐related articles; and f) evaluated delirium sub‐types. We searched articles published from January 1966 through April 2011. Second, a manual search of references of the retrieved papers plus an Internet search using Google Scholar was conducted to find additional SERs. Titles and abstracts were screened by 2 reviewers (B.A.K., M.Z.). Authors of the included studies were contacted as necessary. Third, a library professional at the Indiana University School of Medicine independently performed a literature search, and those results were compared with our search to retrieve any missing SERs.

The methodological quality of each SER was independently assessed by 2 reviewers (B.A.K., M.Z.) using the United States Preventive Services Task Force (USPSTF) Critical Appraisal for SER.11 This scale assesses parameters that are critical to the scientific credibility of an SER and categorizes the SER as poor, fair, or good (Table 1). The 2 reviewers (B.A.K., M.Z.) used a data extraction form to record the following information from each SER: primary author, publication year, number and type of studies, number of participants and their mean age, study population, method for delirium diagnosis, risk factors, preventive and therapeutic interventions, and outcomes. Any disagreement between reviewers in SER selection, data extraction, or SER appraisal was resolved through discussion with a third reviewer (M.A.B.). The conflicting findings among SERs were resolved by consensus and by including the findings from a good SER over a fair SER.

RESULTS

Our search yielded 76,060 potential citations, out of which we identified 38 SERs meeting our inclusion criteria (Table 2). Figure 1 outlines our search strategy. Based on the USPSTF criteria, 22 SERs graded as good or fair provided the data to establish our review.

Figure 1

DISCUSSION AND CLINICAL IMPLICATIONS

Our study identified age, cognitive impairment, depression, and mepiridine use for analgesia as risk factors for delirium in surgical patients. Drugs with anticholinergic properties were implicated in delirium development in both medical and surgical patients. The CAM has the best available data to be used as a diagnostic tool for delirium. Multicomponent interventions to prevent delirium occurrence are effective in a non‐cognitively impaired population, and low‐dose haloperidol prophylaxis decreases delirium duration and severity without affecting delirium incidence in hip surgery patients. There is no advantage of using atypical antipsychotics over haloperidol in treating delirium, and low‐dose haloperidol is as effective as a higher dose without unwarranted extrapyramidal side effects. Delirium carries a poor prognosis with an increased risk of death, institutionalization, and dementia.

Hospitals may benefit from implementing multicomponent strategies, focusing on at‐risk elderly medical and surgical patients, administered by a multidisciplinary team to reduce delirium incidence. For ICU physicians and administrators, development of sedation guidelines minimizing the use of benzodiazepines will decrease the risk of delirium development.

A structured approach in diagnosing delirium is required to maximize identification. Use of the CAM, based on best available data is recommended. However, the length of time in doing the CAM (more than 10 minutes with the requisite mental status examination) and insensitivity in nonexpert hands suggest a need for alternative screening tools. Haloperidol should be the preferred first‐line pharmacological therapy for delirium, with atypical antipsychotics reserved for patients with contraindications to haloperidol or those who are refractory to therapy with haloperidol. Figure 2 delineates a clinical model for delirium management derived from the findings in the Results section.

Figure 2

FUTURE RESEARCH DIRECTIONS

We identified multiple areas without clear guidelines that could provide opportunities for future research. A role for routine delirium screening can be clarified through a well‐designed delirium screening trial investigating the benefits of delirium screening, coupled with a multicomponent intervention versus usual care. Use of pharmacotherapy in delirium prevention needs to be explored further in a large randomized trial, with 3 arms to compare typical antipsychotics, atypical antipsychotics, and placebo in patients at risk for delirium with a primary outcome of delirium incidence. In regard to delirium treatment, a large randomized trial to compare haloperidol with atypical antipsychotics, with a placebo arm focusing not only on delirium duration and severity, but also on long‐term outcomes such as rehospitalizations, institutionalization, cognitive impairment, and mortality, is warranted. Figure 3 points out potential areas for researchers to investigate hypotheses generated by our review and thereby improve delirium care.

Figure 3

To our knowledge, our SER presents the first summary of SERs in delirium. Prior to this review, Michaud et al9 and National Institute for Health and Clinical Excellence48 published delirium guidelines, but in both of these guidelines, evidence was collected from a multitude of studies ranging in methodology from scientific review and meta‐analysis to observational studies, and the majority of recommendations were based on expert opinion. On the contrary, our review was limited to rigorously conducted SERs; hence, we utilized the highest level, critically appraised evidence to provide guidance to clinicians and researchers.

Limitations include a diverse group of studies with a heterogeneous population of patients, preventing pooling of results. We did not review each individual study included in the 38 SERs. We excluded non‐English language SERs, studies evaluating delirium subtypes, alcohol or substance abuse‐related delirium, or delirium associated with psychiatric disorders. As we only reviewed SERs, some notable studies not included in the SERs may have been missed.

CONCLUSION

Delirium among hospitalized patients is a common syndrome with a significant burden to the healthcare system and society. The field of delirium has seen considerable advances in diagnosis, prevention, and treatment over the last decade. Even with this advancement, there are still areas of uncertainty, such as: the benefits and costs of delirium screening; the benefits and harms of single or combined pharmacological agents for delirium prevention and treatment; the development of a set of reliable biomarkers for delirium diagnosis, prognosis, and response to therapy; the long‐term effect of delirium‐specific therapeutics on patients' cognitive, physical, and psychological functions; and the relationship between delirium and the development of Alzheimer's disease. As our understanding of delirium's impact on patients and healthcare improves, delirium should be identified as an indicator of poor long‐term prognosis, and should prompt immediate and effective evidence‐based management strategies, like any other critical illness.