A 75-year-old woman with a history of hypertension and left-lung lobectomy for a carcinoid tumor 10 years ago presented with a 2-week history of progressive cough, dyspnea, and fatigue. Her heart rate was 159 beats per minute with an irregularly irregular rhythm, and her respiratory rate was 36 breaths per minute. Her blood pressure was 140/90 mm Hg. Examination revealed decreased breath sounds and dullness on percussion at the left lung base, jugular venous distention with a positive hepatojugular reflux sign, and an enlarged liver. Electrocardiography showed atrial fibrillation. Chest radiography (Figure 1) revealed enlargement of the cardiac silhouette, with a disproportionately increased transverse diameter, and an obscured left costophrenic angle. A radiograph taken 13 months earlier (Figure 1) had shown a normal cardiothoracic ratio.

EVALUATION OF PERICARDIAL EFFUSION

Pericardial effusion should be suspected in patients presenting with symptoms of impaired cardiac function such as fatigue, dyspnea, nausea, palpitations, lightheadedness, cough, and hoarseness. Patients may also present with chest pain, often decreased by sitting up and leaning forward and exacerbated by lying supine.

Physical examination may reveal distant heart sounds, an absent or displaced apical impulse, dullness and increased fremitus beneath the angle of the left scapula (the Ewart sign), pulsus paradoxus, and nonspecific findings such as tachycardia and hypotension. Jugular venous distention, hepatojugular reflux, and peripheral edema suggest impaired cardiac function.

A chest radiograph showing unexplained new symmetric cardiomegaly (which is often globe-shaped) without signs of pulmonary congestion¹ or with a left dominant pleural effusion is an indicator of pericardial effusion, as in our patient. Pericardial fluid may be seen outlining the heart between the epicardial and mediastinal fat, posterior to the sternum in a lateral view.

Other common causes of cardiomegaly include hypertension, congestive heart failure, valvular disease, cardiomyopathy, ischemic heart disease, and pulmonary disease.

Once pericardial effusion is suspected, the next step is to confirm its presence and determine its hemodynamic significance. Transthoracic echocardiography is the imaging test of choice to confirm effusion, as it can be done rapidly and in unstable patients.²

If transthoracic echocardiography is non­diagnostic but suspicion is high, further evaluation may include transesophageal echocardiography,³ computed tomography, or magnetic resonance imaging.

MAKING THE DIAGNOSIS

Pericardial effusion can occur as part of various diseases involving the pericardium, eg, acute pericarditis, myocarditis, autoimmune disease, postmyocardial infarction, malignancy, aortic dissection, and chest trauma. It can also be associated with certain drugs.

In our patient, echocardiography (Figure 2, Figure 3) demonstrated a large amount of pericardial fluid, and 820 mL of red fluid was aspirated by pericardiocentesis, resulting in relief of her respiratory symptoms. Subcostal two-dimensional echocardiography demonstrated rocking of the heart and intermittent right-ventricular collapse (watch video at www.ccjm.org). Flow cytometry demonstrated 10% kappa+ monoclonal cells. Bone marrow biopsy with immunohistochemical staining revealed infiltration by CD20+, CD5+, CD23+, and BCL1– cells, compatible with small lymphocytic lymphoma.

MALIGNANT PERICARDIAL EFFUSION

Pericardial disease can be the first manifestation of malignancy,⁴ more often when the patient presents with a large pericardial effusion or tamponade. Malignant tumors of the lung, breast, and esophagus—as well as lymphoma, leukemia, and melanoma—often spread to the pericardium directly or through the lymphatic vessels or bloodstream.⁴ In our patient, corticosteroid treatment was initiated, and echocardiography at a follow-up visit 2 months later showed no pericardial fluid.

A 75-year-old woman with a history of hypertension and left-lung lobectomy for a carcinoid tumor 10 years ago presented with a 2-week history of progressive cough, dyspnea, and fatigue. Her heart rate was 159 beats per minute with an irregularly irregular rhythm, and her respiratory rate was 36 breaths per minute. Her blood pressure was 140/90 mm Hg. Examination revealed decreased breath sounds and dullness on percussion at the left lung base, jugular venous distention with a positive hepatojugular reflux sign, and an enlarged liver. Electrocardiography showed atrial fibrillation. Chest radiography (Figure 1) revealed enlargement of the cardiac silhouette, with a disproportionately increased transverse diameter, and an obscured left costophrenic angle. A radiograph taken 13 months earlier (Figure 1) had shown a normal cardiothoracic ratio.

EVALUATION OF PERICARDIAL EFFUSION

Pericardial effusion should be suspected in patients presenting with symptoms of impaired cardiac function such as fatigue, dyspnea, nausea, palpitations, lightheadedness, cough, and hoarseness. Patients may also present with chest pain, often decreased by sitting up and leaning forward and exacerbated by lying supine.

Physical examination may reveal distant heart sounds, an absent or displaced apical impulse, dullness and increased fremitus beneath the angle of the left scapula (the Ewart sign), pulsus paradoxus, and nonspecific findings such as tachycardia and hypotension. Jugular venous distention, hepatojugular reflux, and peripheral edema suggest impaired cardiac function.

A chest radiograph showing unexplained new symmetric cardiomegaly (which is often globe-shaped) without signs of pulmonary congestion¹ or with a left dominant pleural effusion is an indicator of pericardial effusion, as in our patient. Pericardial fluid may be seen outlining the heart between the epicardial and mediastinal fat, posterior to the sternum in a lateral view.

Other common causes of cardiomegaly include hypertension, congestive heart failure, valvular disease, cardiomyopathy, ischemic heart disease, and pulmonary disease.

Once pericardial effusion is suspected, the next step is to confirm its presence and determine its hemodynamic significance. Transthoracic echocardiography is the imaging test of choice to confirm effusion, as it can be done rapidly and in unstable patients.²

If transthoracic echocardiography is non­diagnostic but suspicion is high, further evaluation may include transesophageal echocardiography,³ computed tomography, or magnetic resonance imaging.

MAKING THE DIAGNOSIS

Pericardial effusion can occur as part of various diseases involving the pericardium, eg, acute pericarditis, myocarditis, autoimmune disease, postmyocardial infarction, malignancy, aortic dissection, and chest trauma. It can also be associated with certain drugs.

In our patient, echocardiography (Figure 2, Figure 3) demonstrated a large amount of pericardial fluid, and 820 mL of red fluid was aspirated by pericardiocentesis, resulting in relief of her respiratory symptoms. Subcostal two-dimensional echocardiography demonstrated rocking of the heart and intermittent right-ventricular collapse (watch video at www.ccjm.org). Flow cytometry demonstrated 10% kappa+ monoclonal cells. Bone marrow biopsy with immunohistochemical staining revealed infiltration by CD20+, CD5+, CD23+, and BCL1– cells, compatible with small lymphocytic lymphoma.

MALIGNANT PERICARDIAL EFFUSION

Pericardial disease can be the first manifestation of malignancy,⁴ more often when the patient presents with a large pericardial effusion or tamponade. Malignant tumors of the lung, breast, and esophagus—as well as lymphoma, leukemia, and melanoma—often spread to the pericardium directly or through the lymphatic vessels or bloodstream.⁴ In our patient, corticosteroid treatment was initiated, and echocardiography at a follow-up visit 2 months later showed no pericardial fluid.

A 75-year-old woman with a history of hypertension and left-lung lobectomy for a carcinoid tumor 10 years ago presented with a 2-week history of progressive cough, dyspnea, and fatigue. Her heart rate was 159 beats per minute with an irregularly irregular rhythm, and her respiratory rate was 36 breaths per minute. Her blood pressure was 140/90 mm Hg. Examination revealed decreased breath sounds and dullness on percussion at the left lung base, jugular venous distention with a positive hepatojugular reflux sign, and an enlarged liver. Electrocardiography showed atrial fibrillation. Chest radiography (Figure 1) revealed enlargement of the cardiac silhouette, with a disproportionately increased transverse diameter, and an obscured left costophrenic angle. A radiograph taken 13 months earlier (Figure 1) had shown a normal cardiothoracic ratio.

EVALUATION OF PERICARDIAL EFFUSION

Pericardial effusion should be suspected in patients presenting with symptoms of impaired cardiac function such as fatigue, dyspnea, nausea, palpitations, lightheadedness, cough, and hoarseness. Patients may also present with chest pain, often decreased by sitting up and leaning forward and exacerbated by lying supine.

Physical examination may reveal distant heart sounds, an absent or displaced apical impulse, dullness and increased fremitus beneath the angle of the left scapula (the Ewart sign), pulsus paradoxus, and nonspecific findings such as tachycardia and hypotension. Jugular venous distention, hepatojugular reflux, and peripheral edema suggest impaired cardiac function.

A chest radiograph showing unexplained new symmetric cardiomegaly (which is often globe-shaped) without signs of pulmonary congestion¹ or with a left dominant pleural effusion is an indicator of pericardial effusion, as in our patient. Pericardial fluid may be seen outlining the heart between the epicardial and mediastinal fat, posterior to the sternum in a lateral view.

Other common causes of cardiomegaly include hypertension, congestive heart failure, valvular disease, cardiomyopathy, ischemic heart disease, and pulmonary disease.

Once pericardial effusion is suspected, the next step is to confirm its presence and determine its hemodynamic significance. Transthoracic echocardiography is the imaging test of choice to confirm effusion, as it can be done rapidly and in unstable patients.²

If transthoracic echocardiography is non­diagnostic but suspicion is high, further evaluation may include transesophageal echocardiography,³ computed tomography, or magnetic resonance imaging.

MAKING THE DIAGNOSIS

Pericardial effusion can occur as part of various diseases involving the pericardium, eg, acute pericarditis, myocarditis, autoimmune disease, postmyocardial infarction, malignancy, aortic dissection, and chest trauma. It can also be associated with certain drugs.

In our patient, echocardiography (Figure 2, Figure 3) demonstrated a large amount of pericardial fluid, and 820 mL of red fluid was aspirated by pericardiocentesis, resulting in relief of her respiratory symptoms. Subcostal two-dimensional echocardiography demonstrated rocking of the heart and intermittent right-ventricular collapse (watch video at www.ccjm.org). Flow cytometry demonstrated 10% kappa+ monoclonal cells. Bone marrow biopsy with immunohistochemical staining revealed infiltration by CD20+, CD5+, CD23+, and BCL1– cells, compatible with small lymphocytic lymphoma.

MALIGNANT PERICARDIAL EFFUSION

Pericardial disease can be the first manifestation of malignancy,⁴ more often when the patient presents with a large pericardial effusion or tamponade. Malignant tumors of the lung, breast, and esophagus—as well as lymphoma, leukemia, and melanoma—often spread to the pericardium directly or through the lymphatic vessels or bloodstream.⁴ In our patient, corticosteroid treatment was initiated, and echocardiography at a follow-up visit 2 months later showed no pericardial fluid.

A 39-year-old man presented with acute left hip pain and inability to bear weight following a minor trauma. The patient had a history of polycystic kidney disease and was on dialysis. Five years ago he had undergone bilateral nephrectomy and a renal transplantation that subsequently failed.

On examination, the active and passive range of motion of the left hip were limited due to pain. His serum laboratory values were:

• Parathyroid hormone 259.7 pmol/L (reference range 1.5–9.3)
• Calcium 2.32 mmol/L (1.15–1.32)
• Phosphate 3.26 mmol/L (0.8–1.45).

Computed tomography of the pelvis revealed an exophytic calcified lesion with multiple cystic spaces and fluid-fluid levels centered on the left pubis, extending medially into the right pubis and laterally into the left adductor muscle group. An acute pathologic fracture was documented in the left inferior pubic ramus (Figure 1). Other radiographic signs of long-standing hyperparathyroidism were present, including subperiosteal bone resorption at the radial side of the middle phalanges and the clavicle epiphysis.

The differential diagnosis of the pelvic lesion included giant cell tumor of bone with aneurysmal bone-cyst-like changes, osteitis fibrosa cystica, and, less likely, metastatic bone disease. Biopsy of the lesion showed clusters of osteoclast-type giant cells on a background of spindle cells and fibrous stroma that in this clinical context was consistent with the diagnosis of brown tumor (Figure 2).¹

BROWN TUMOR

Brown tumor has been reported in fewer than 2% of patients with primary hyperparathyroidism and in 1.5% to 1.7% of those with secondary hyperparathyroidism (ie, from chronic renal failure, malabsorption, vitamin D deficiency, or hypocalcemia).^2–4 An excess of parathyroid hormone increases the number and activity of osteoclasts, which are responsible for the lytic lesions. Brown tumor is the localized form of osteitis fibrosa cystica and is the most characteristic of the many skeletal changes that accompany secondary hyperparathyroidism.

Brown tumor is named for its color, which results from hemorrhages with accumulation of hemosiderin within the vascularized fibrous tissue. The tumor most commonly affects the pelvis, ribs, long-bone shafts, clavicle, and mandible.⁵ Clinical symptoms are nonspecific and depend on the size and location of the lesion.

Medical management of secondary hyperparathyroidism in dialysis patients involves some combination of phosphate binders (either calcium-containing or non-calcium-containing binders), calcitriol or synthetic vitamin D analogs, and a calcimimetic. Parathyroidectomy is required if drug therapy is ineffective. Surgical excision of brown tumor should be considered in patients who have large bone defects with spontaneous fracture risk or increasing pain. Our patient declined surgical intervention.

A 39-year-old man presented with acute left hip pain and inability to bear weight following a minor trauma. The patient had a history of polycystic kidney disease and was on dialysis. Five years ago he had undergone bilateral nephrectomy and a renal transplantation that subsequently failed.

On examination, the active and passive range of motion of the left hip were limited due to pain. His serum laboratory values were:

• Parathyroid hormone 259.7 pmol/L (reference range 1.5–9.3)
• Calcium 2.32 mmol/L (1.15–1.32)
• Phosphate 3.26 mmol/L (0.8–1.45).

Computed tomography of the pelvis revealed an exophytic calcified lesion with multiple cystic spaces and fluid-fluid levels centered on the left pubis, extending medially into the right pubis and laterally into the left adductor muscle group. An acute pathologic fracture was documented in the left inferior pubic ramus (Figure 1). Other radiographic signs of long-standing hyperparathyroidism were present, including subperiosteal bone resorption at the radial side of the middle phalanges and the clavicle epiphysis.

The differential diagnosis of the pelvic lesion included giant cell tumor of bone with aneurysmal bone-cyst-like changes, osteitis fibrosa cystica, and, less likely, metastatic bone disease. Biopsy of the lesion showed clusters of osteoclast-type giant cells on a background of spindle cells and fibrous stroma that in this clinical context was consistent with the diagnosis of brown tumor (Figure 2).¹

BROWN TUMOR

Brown tumor has been reported in fewer than 2% of patients with primary hyperparathyroidism and in 1.5% to 1.7% of those with secondary hyperparathyroidism (ie, from chronic renal failure, malabsorption, vitamin D deficiency, or hypocalcemia).^2–4 An excess of parathyroid hormone increases the number and activity of osteoclasts, which are responsible for the lytic lesions. Brown tumor is the localized form of osteitis fibrosa cystica and is the most characteristic of the many skeletal changes that accompany secondary hyperparathyroidism.

Brown tumor is named for its color, which results from hemorrhages with accumulation of hemosiderin within the vascularized fibrous tissue. The tumor most commonly affects the pelvis, ribs, long-bone shafts, clavicle, and mandible.⁵ Clinical symptoms are nonspecific and depend on the size and location of the lesion.

Medical management of secondary hyperparathyroidism in dialysis patients involves some combination of phosphate binders (either calcium-containing or non-calcium-containing binders), calcitriol or synthetic vitamin D analogs, and a calcimimetic. Parathyroidectomy is required if drug therapy is ineffective. Surgical excision of brown tumor should be considered in patients who have large bone defects with spontaneous fracture risk or increasing pain. Our patient declined surgical intervention.

A 39-year-old man presented with acute left hip pain and inability to bear weight following a minor trauma. The patient had a history of polycystic kidney disease and was on dialysis. Five years ago he had undergone bilateral nephrectomy and a renal transplantation that subsequently failed.

On examination, the active and passive range of motion of the left hip were limited due to pain. His serum laboratory values were:

• Parathyroid hormone 259.7 pmol/L (reference range 1.5–9.3)
• Calcium 2.32 mmol/L (1.15–1.32)
• Phosphate 3.26 mmol/L (0.8–1.45).

Computed tomography of the pelvis revealed an exophytic calcified lesion with multiple cystic spaces and fluid-fluid levels centered on the left pubis, extending medially into the right pubis and laterally into the left adductor muscle group. An acute pathologic fracture was documented in the left inferior pubic ramus (Figure 1). Other radiographic signs of long-standing hyperparathyroidism were present, including subperiosteal bone resorption at the radial side of the middle phalanges and the clavicle epiphysis.

The differential diagnosis of the pelvic lesion included giant cell tumor of bone with aneurysmal bone-cyst-like changes, osteitis fibrosa cystica, and, less likely, metastatic bone disease. Biopsy of the lesion showed clusters of osteoclast-type giant cells on a background of spindle cells and fibrous stroma that in this clinical context was consistent with the diagnosis of brown tumor (Figure 2).¹

BROWN TUMOR

Brown tumor has been reported in fewer than 2% of patients with primary hyperparathyroidism and in 1.5% to 1.7% of those with secondary hyperparathyroidism (ie, from chronic renal failure, malabsorption, vitamin D deficiency, or hypocalcemia).^2–4 An excess of parathyroid hormone increases the number and activity of osteoclasts, which are responsible for the lytic lesions. Brown tumor is the localized form of osteitis fibrosa cystica and is the most characteristic of the many skeletal changes that accompany secondary hyperparathyroidism.

Brown tumor is named for its color, which results from hemorrhages with accumulation of hemosiderin within the vascularized fibrous tissue. The tumor most commonly affects the pelvis, ribs, long-bone shafts, clavicle, and mandible.⁵ Clinical symptoms are nonspecific and depend on the size and location of the lesion.

Medical management of secondary hyperparathyroidism in dialysis patients involves some combination of phosphate binders (either calcium-containing or non-calcium-containing binders), calcitriol or synthetic vitamin D analogs, and a calcimimetic. Parathyroidectomy is required if drug therapy is ineffective. Surgical excision of brown tumor should be considered in patients who have large bone defects with spontaneous fracture risk or increasing pain. Our patient declined surgical intervention.

In his physician-coming-of-age novel House of God, published in 1978, Dr. Steven Bergman (aka Sam Shem) presented rules for an intern’s survival devised by the senior resident, the Fat Man. Rule X was that there is no fever if you don’t check the patient’s temperature, implying that if the physician is unaware of an elevated temperature, no “fever workup” is warranted. A fever workup back then was not just a few keystrokes to order a chest x-ray, complete blood cell count, and blood cultures. The intern had to go to the bedside, awaken and examine the patient, draw the blood, perhaps transport the blood samples to the lab, do a urinalysis, and take the patient to the radiology department to get the chest x-ray. There was often little thought to the intern’s action; a fever in the hospital automatically meant there needed to be a fever workup.

A covering senior resident might have gotten the same notification of a fever, quickly reviewed the chart, gone to the bedside, and assessed whether a bacterial infection was likely enough to warrant the time and annoyance of a full fever workup. As supervising faculty, I will accept that assessment from a senior resident in June more willingly than from an intern in July. Tests and physical findings must be evaluated in context, taking into consideration the patient as well as the skill and experience of the physician.

So how should we react to guidelines that seem to be based on the premise that a positive finding will result in reflexive ordering of additional tests or initiating a therapeutic intervention, and thus should be avoided by all of us—young intern and senior cardiologist alike?

In this issue of the Journal, Dr. Aldo Schenone et al discuss the management of the asymptomatic patient who has carotid artery stenosis. They put into perspective the risks and benefits of medical or surgical intervention as initially defined by several landmark trials, noting how those conclusions should now be modified by knowledge of the efficacy of current medical therapy.

The US Preventive Services Task Force (USPSTF)¹ has recommended against screening for asymptomatic carotid artery stenosis in the general population, noting the limited sensitivity (71%) and specificity (98%) of auscultation to diagnose significant stenosis and lumping it with other ineffective screening tests. In other words, we should not examine asymptomatic patients for carotid bruits, just as we should not look for the fever because finding it could lead to additional testing and potentially unnecessary therapy.

But there are broader implications when a bruit is discovered, beyond simply trodding the algorithmic path toward stenting or endarterectomy. A bruit can suggest occult atherosclerotic disease that warrants medical attention, even if traditional risk factors for atherosclerosis are not prominent. Its discovery can be a wake-up call to the patient (and physician) that the hackneyed admonitions to eat healthy, lose weight, and exercise are actually relevant. Its discovery may lead to medical intervention with a potent statin or with a more aggressive target for blood pressure control. It may color the interpretation of the patient’s described vague arm tingling when bowling.

I may well be misleading myself, but I am more comfortable in dealing with whatever oddities I discover on a physical examination than not doing the examination at all. I’d rather know about the bruit (or the fever) and then think about our options. The stethoscope indeed has limited test reliability, but the real action takes place between its earpieces; the bruit is merely the catalyst for thought. There must be a guideline somewhere that says that a thoughtful, informed, commonsense evaluation is a useful contributor to patient care.

In his physician-coming-of-age novel House of God, published in 1978, Dr. Steven Bergman (aka Sam Shem) presented rules for an intern’s survival devised by the senior resident, the Fat Man. Rule X was that there is no fever if you don’t check the patient’s temperature, implying that if the physician is unaware of an elevated temperature, no “fever workup” is warranted. A fever workup back then was not just a few keystrokes to order a chest x-ray, complete blood cell count, and blood cultures. The intern had to go to the bedside, awaken and examine the patient, draw the blood, perhaps transport the blood samples to the lab, do a urinalysis, and take the patient to the radiology department to get the chest x-ray. There was often little thought to the intern’s action; a fever in the hospital automatically meant there needed to be a fever workup.

A covering senior resident might have gotten the same notification of a fever, quickly reviewed the chart, gone to the bedside, and assessed whether a bacterial infection was likely enough to warrant the time and annoyance of a full fever workup. As supervising faculty, I will accept that assessment from a senior resident in June more willingly than from an intern in July. Tests and physical findings must be evaluated in context, taking into consideration the patient as well as the skill and experience of the physician.

So how should we react to guidelines that seem to be based on the premise that a positive finding will result in reflexive ordering of additional tests or initiating a therapeutic intervention, and thus should be avoided by all of us—young intern and senior cardiologist alike?

In this issue of the Journal, Dr. Aldo Schenone et al discuss the management of the asymptomatic patient who has carotid artery stenosis. They put into perspective the risks and benefits of medical or surgical intervention as initially defined by several landmark trials, noting how those conclusions should now be modified by knowledge of the efficacy of current medical therapy.

The US Preventive Services Task Force (USPSTF)¹ has recommended against screening for asymptomatic carotid artery stenosis in the general population, noting the limited sensitivity (71%) and specificity (98%) of auscultation to diagnose significant stenosis and lumping it with other ineffective screening tests. In other words, we should not examine asymptomatic patients for carotid bruits, just as we should not look for the fever because finding it could lead to additional testing and potentially unnecessary therapy.

But there are broader implications when a bruit is discovered, beyond simply trodding the algorithmic path toward stenting or endarterectomy. A bruit can suggest occult atherosclerotic disease that warrants medical attention, even if traditional risk factors for atherosclerosis are not prominent. Its discovery can be a wake-up call to the patient (and physician) that the hackneyed admonitions to eat healthy, lose weight, and exercise are actually relevant. Its discovery may lead to medical intervention with a potent statin or with a more aggressive target for blood pressure control. It may color the interpretation of the patient’s described vague arm tingling when bowling.

I may well be misleading myself, but I am more comfortable in dealing with whatever oddities I discover on a physical examination than not doing the examination at all. I’d rather know about the bruit (or the fever) and then think about our options. The stethoscope indeed has limited test reliability, but the real action takes place between its earpieces; the bruit is merely the catalyst for thought. There must be a guideline somewhere that says that a thoughtful, informed, commonsense evaluation is a useful contributor to patient care.

In his physician-coming-of-age novel House of God, published in 1978, Dr. Steven Bergman (aka Sam Shem) presented rules for an intern’s survival devised by the senior resident, the Fat Man. Rule X was that there is no fever if you don’t check the patient’s temperature, implying that if the physician is unaware of an elevated temperature, no “fever workup” is warranted. A fever workup back then was not just a few keystrokes to order a chest x-ray, complete blood cell count, and blood cultures. The intern had to go to the bedside, awaken and examine the patient, draw the blood, perhaps transport the blood samples to the lab, do a urinalysis, and take the patient to the radiology department to get the chest x-ray. There was often little thought to the intern’s action; a fever in the hospital automatically meant there needed to be a fever workup.

A covering senior resident might have gotten the same notification of a fever, quickly reviewed the chart, gone to the bedside, and assessed whether a bacterial infection was likely enough to warrant the time and annoyance of a full fever workup. As supervising faculty, I will accept that assessment from a senior resident in June more willingly than from an intern in July. Tests and physical findings must be evaluated in context, taking into consideration the patient as well as the skill and experience of the physician.

So how should we react to guidelines that seem to be based on the premise that a positive finding will result in reflexive ordering of additional tests or initiating a therapeutic intervention, and thus should be avoided by all of us—young intern and senior cardiologist alike?

In this issue of the Journal, Dr. Aldo Schenone et al discuss the management of the asymptomatic patient who has carotid artery stenosis. They put into perspective the risks and benefits of medical or surgical intervention as initially defined by several landmark trials, noting how those conclusions should now be modified by knowledge of the efficacy of current medical therapy.

The US Preventive Services Task Force (USPSTF)¹ has recommended against screening for asymptomatic carotid artery stenosis in the general population, noting the limited sensitivity (71%) and specificity (98%) of auscultation to diagnose significant stenosis and lumping it with other ineffective screening tests. In other words, we should not examine asymptomatic patients for carotid bruits, just as we should not look for the fever because finding it could lead to additional testing and potentially unnecessary therapy.

But there are broader implications when a bruit is discovered, beyond simply trodding the algorithmic path toward stenting or endarterectomy. A bruit can suggest occult atherosclerotic disease that warrants medical attention, even if traditional risk factors for atherosclerosis are not prominent. Its discovery can be a wake-up call to the patient (and physician) that the hackneyed admonitions to eat healthy, lose weight, and exercise are actually relevant. Its discovery may lead to medical intervention with a potent statin or with a more aggressive target for blood pressure control. It may color the interpretation of the patient’s described vague arm tingling when bowling.

I may well be misleading myself, but I am more comfortable in dealing with whatever oddities I discover on a physical examination than not doing the examination at all. I’d rather know about the bruit (or the fever) and then think about our options. The stethoscope indeed has limited test reliability, but the real action takes place between its earpieces; the bruit is merely the catalyst for thought. There must be a guideline somewhere that says that a thoughtful, informed, commonsense evaluation is a useful contributor to patient care.

Carotid artery disease that is asymptomatic poses a dilemma: Should the patient undergo revascularization (surgical carotid endarterectomy or percutaneous stenting) or receive medical therapy alone?

On one hand, because one consequence of carotid atherosclerosis—ischemic stroke—can be devastating or deadly, many physicians and patients would rather “do something,” ie, proceed with surgery. Furthermore, several randomized trials^1–4 found carotid endarterectomy superior to medical therapy.

On the other hand, these trials were conducted in the 1990s. Surgery has improved since then, but so has medical therapy. And if we re-examine the data from the trials in terms of the absolute risk reduction and number needed to treat, as opposed to the relative risk reduction, surgery may appear less beneficial.

Needed is a way to identify patients who would benefit from surgery and those who would more likely be harmed. Research in that direction is ongoing.

Here, we present a simple algorithmic approach to managing asymptomatic carotid artery stenosis based on the patient’s age, sex, and life expectancy. Our approach is based on a review of the best available evidence.

UP TO 8% OF ADULTS HAVE STENOSIS

Stroke is the third largest cause of death in the United States and the leading cause of disability.⁵ From 10% to 15% of strokes are associated with carotid artery stenosis.^6,7

The prevalence of asymptomatic carotid disease, defined as stenosis greater than 50%, ranges from 4% to 8% in adults.⁸

However, major societies recommend against screening for carotid stenosis in the general population.^9–12 Similarly, the US Preventive Services Task Force also discourages the use of carotid auscultation as screening in the general population (Table 1).¹³ Generally, cases of asymptomatic carotid stenosis are diagnosed by ultrasonography after the patient’s physician happens to hear a bruit during a routine examination, during a preoperative assessment, or after the patient suffers a transient ischemic attack or stroke on the contralateral side.

CLASS II RECOMMENDATIONS FOR SURGERY OR STENTING

There are well-established guidelines for managing symptomatic carotid disease,¹⁴ based on evidence from the North American Symptomatic Carotid Endarterectomy Trial¹⁵ and the European Carotid Surgery Trial,¹⁶ both from 1998. But how to manage asymptomatic carotid disease remains uncertain.

If stenosis of the internal carotid artery is greater than 70% on ultrasonography, computed tomography, or magnetic resonance imaging, and if the risk of perioperative stroke and death is low (< 3%), current guidelines¹⁴ give carotid endarterectomy a class IIa recommendation (ie, evidence is conflicting, but the weight of evidence is in favor), and they give prophylactic carotid artery stenting with optimal medical treatment a class IIb recommendation (efficacy is less well established).⁵

But medical management has improved, and new data suggest that this improvement may override the minimal net benefit of intervention in some patients.¹⁷ Some authors suggest that it is best to use patient characteristics and imaging features to guide treatment.¹⁸

EVIDENCE TO SUPPORT CAROTID REVASCULARIZATION

Three major trials (Table 2) published nearly 20 years ago provide the foundation of the current guidelines:

• the Endarterectomy for Asymptomatic Carotid Atherosclerosis Study (ACAS)¹
• the Asymptomatic Carotid Surgery Trial (ACST)^2,3
• the Veterans Affairs (VA) Cooperative Study.⁴

A Cochrane review of these trials,¹⁹ where medical therapy consisted only of aspirin and little use of statin therapy, found that carotid endarterectomy reduced the rate of perioperative stroke or death or any subsequent stroke in the next 3 years by 31% (relative risk 69%, 95% confidence interval [CI] 0.57–0.83). “Perioperative” was defined as the period from randomization until 30 days after surgery in the surgical group and an equivalent period in the medical group.

Moreover, carotid endarterectomy reduced the rate of disabling or fatal nonperioperative stroke by 50% compared with medical management alone.^1,2,19 Patients who had contralateral symptomatic disease or who had undergone contralateral carotid endarterectomy seemed to benefit more from the procedure than those who had not.¹⁹

Also, the ACST investigators found that revascularization was associated with a reduction in contralateral strokes (which occurred in 39 vs 64 patients, P = .01) independent of contralateral symptoms or contralateral carotid endarterectomy.^2,3 The exact mechanism is unknown but could be related to better blood pressure control and risk factor modification after carotid endarterectomy.

Another factor supporting revascularization is that the outcomes of revascularization have improved over time. In 2010, the Carotid Revascularization Endarterectomy Versus Stenting Trial (CREST)²⁰ reported a 30-day periprocedural incidence of death or stroke of only 1.4%, compared with 2.9% in the earlier landmark trials.

Stenting is a noninferior alternative

For patients who have asymptomatic stenosis greater than 80% on color duplex ultrasonography and a risk of stroke or death during carotid endarterectomy that is prohibitively high (> 3%), carotid stenting has proved to be a noninferior alternative.^21,22

The Stenting and Angioplasty With Protection of Patients With High Risk for Endarterectomy (SAPPHIRE) trial²¹ reported a risk of death, stroke, or myocardial infarction of about 5% at 30 days and 10% at 1 year after stenting. A recent observational study revealed lower perioperative complication rates, with a risk of death or stroke of about 3%, which satisfy current guideline requirements.²³

To be deemed at high surgical risk and therefore eligible for the SAPPHIRE trial,²¹ patients had to have clinically significant cardiac disease, severe pulmonary disease, contralateral carotid occlusion, contralateral laryngeal-nerve palsy, recurrent stenosis after carotid endarterectomy, previous radical neck surgery or radiation therapy to the neck, or age greater than 80.

EVIDENCE AGAINST CAROTID REVASCULARIZATION

Although carotid revascularization has evidence to support it, further interpretation of the data may lessen its apparent benefits.

Small absolute benefit, high number needed to treat

If we compare the relative risk reduction for the outcome of perioperative death or any stroke over 5 years (30% to 50%) vs the absolute risk reduction (4% to 5.9%), revascularization seems less attractive.¹⁹

Relative risk reduction in death or stroke with carotid surgery is 30%–50%; absolute risk reduction is 4%–5.9%

The benefit may be further diminished if we consider only strokes related to large vessels, since up to 45% of strokes in patients with carotid disease are lacunar or cardioembolic.²⁴ Assessing for prevention of large-vessel stroke using the ACAS data, the benefit of carotid endarterectomy for prevention of stroke is further decreased to a 3.5% absolute risk reduction, and the number needed to treat for 2 years increases from 62 to 111.^24,25 Nevertheless, revascularization is necessary in appropriately selected patients, as a cerebrovascular event can cause life-altering changes to a patient’s cognitive, emotional, and physical condition.²⁶

Medical therapy—and surgery—are evolving

The optimal medical management used in the landmark studies was significantly different from what is currently recommended. The ACAS trial¹⁸ used only aspirin as optimal medical management, with no mention of statins. In the ACST trial,^2,3 the use of statins increased over time, from 7% to 11% at the beginning of the trial to 80% to 82% at the end.

On the other hand, the ACAS¹ surgeons were required to have an excellent safety record to participate. This might have compromised the trial’s validity or our ability to generalize its conclusions.

Recent data from Abbott¹⁷ suggested a loss of a statistically significant surgical advantage in prevention of ipsilateral stroke and transient ischemic attack from the early 1990s. This is most likely explained by improved medical therapy, since there was a 22% increase in baseline proportion of patients receiving antiplatelet therapy from 1985 to 2007, with 60% of patients taking antihypertensive drugs and 30% of patients taking lipid-lowering drugs. Moreover, since 2001, the annual rates of ipsilateral stroke in patients receiving medical management alone fell below those of patients who underwent carotid endarterectomy in the ACAS trial.

The analysis by Abbott¹⁷ has major limitations: inclusion of small studies, many crossover patients, and heterogeneity. In support of this allegation, a small trial (33 patients) reported a risk of stroke ipsilateral to an asymptomatic carotid stenosis as low as 0.34% per year.²⁵ Even when contrasting the outcomes of medical therapy against those of current carotid endarterectomy, in which the rate of perioperative stroke and death have fallen to 0.88% to 1.7%,^17,27,28 there is concern that the risk associated with surgery may outweigh the long-term benefit.

Flaws in the landmark trials

Beyond the debate of the questionable benefit of revascularization, well-defined flaws in the landmark trials weaken or limit their influence on current treatment guidelines and protocols for deciding whether to revascularize.

No significant benefit was found for patients over age 75.^2,3 This was thought to be due to decreased life expectancy, since the benefit from revascularization becomes significant after 3 years from intervention.^1–3 Also, studies have shown that increasing age is associated with a higher risk of perioperative stroke and death.^20,21

Women showed no benefit at 5 years and only a trend toward benefit at 10 years (P = .05),² likely from a higher rate of periprocedural strokes.

Blacks and Hispanics were underrepresented in the landmark studies,¹⁹ while one observational study reported a higher incidence of in-hospital stroke after carotid endarterectomy in black patients (6.6%) than in white patients (2%).²⁹

When associated with contralateral carotid occlusion, carotid endarterectomy carries a higher risk of perioperative stroke or death.^23,30,31

Carotid revascularization failed to reduce the risk of death—the total number of deaths within 10 years was not significantly reduced by immediate carotid endarterectomy compared with deferring the procedure.²

EVIDENCE SUPPORTING OPTIMAL MEDICAL MANAGEMENT

Optimal medical therapy mainly consists of antiplatelet therapy, blood pressure management, diabetic glycemic control, and statin therapy along with lifestyle changes including smoking cessation, exercise, and weight loss (Table 3).⁹ Detailed recommendations are provided in the American Heart Association/American Stroke Association guidelines for primary prevention of stroke.³²

Antiplatelet therapy has been shown to reduce the incidence of stroke by 25%. There is no added benefit in combining antiplatelet agents unless the patient has concomitant symptomatic coronary artery disease, recent coronary stenting, or severe peripheral artery disease.^33,34

Blood pressure control can reduce the incidence of stroke by 30% to 40%, and recent data suggest that drugs working on the renin-angiotensin system offer more benefit than beta-blockers for the same reduction in blood pressure.^34,35

Diabetic glycemic control is supported, as higher hemoglobin A_1c and fasting glucose values are associated with higher relative risk of stroke.^32,36,37 However, the stroke rate does not differ significantly between patients receiving intensive therapy and those receiving standard therapy.³⁴

Statins actually shrink carotid plaques and reduce the risk of stroke by 15% for each 10% reduction in low-density lipoprotein cholesterol. It is estimated that statin therapy confers a 30% relative risk reduction of stroke over 20 years.^34,38–41

Smoking increases the overall risk of stroke by 150%, making its cessation mandatory.⁴²

HIGH-RISK FEATURES FOR STROKE IN ASYMPTOMATIC CAROTID STENOSIS

Studies have tried to identify risk factors for stroke, so that patients at high risk could undergo revascularization and benefit from it. However, no well-defined high-risk features have yet been described that would identify patients who would benefit from early surgery.

For instance, no correlation has been found between age, sex, diabetes mellitus, lipid levels, or smoking and progression of disease.⁴³ In contrast, having either contralateral symptomatic carotid disease or contralateral total occlusion translated into a higher ipsilateral stroke risk.¹⁸ And in several studies, the 5-year risk of ipsilateral stroke was as high as 16.2% for those with 60% to 99% stenosis.^1,2,18,24,43

Features of the plaque itself

More recently, there has been a focus on plaque evaluation to predict outcomes.

Statins shrink carotid plaques and reduce the risk of stroke by 15% for each 10% reduction in LDL-C

Percent stenosis. An increased risk of death or stroke has been reported with higher degrees of stenosis or plaque progression.^44,45 The gross annual risk of ipsilateral stroke increases from 1.5% with stenosis of 60% to 70%, to 4.2% with stenosis of 71% to 90%, and to 7% with stenosis of 91% to 99%. Nevertheless, current data are insufficient to determine whether there is increasing benefit from surgery with increasing degree of stenosis in asymptomatic carotid disease.^1,3,24,44

Plaque progression translates to a 7.2% absolute increase in the incidence of stroke (1.1% if the plaque is stable vs 8.3% if the plaque is progressing). Interestingly, plaque progression to greater than 80% stenosis results in worse outcomes (relative risk 3.4, 95% CI 1.5–7.8) compared with the same level of stenosis without recent progression.³³

Intimal wall thickening of more than 1.15 mm confers a hazard ratio for stroke of 3 (95% CI 1.48–6.11).⁴⁶

Increased echolucency also confers a hazard ratio for stroke of 3 (95% CI 1.4–8.0).⁴⁶

A low gray-scale median (a surrogate of plaque composition) and plaque area have been identified as independent predictors of ipsilateral events.⁴⁴

Embolic signals on transcranial Doppler ultrasonography (Figure 1) have been associated with a hazard ratio for stroke of 2.54 over 2 years.⁴⁷

Carotid plaques predominantly composed of lipid-rich necrotic cores carry a higher risk of stroke (hazard ratio 7.2, 95% CI 1.12–46.20).⁴⁸

High tensile stress (circumferential wall tension divided by the intima-media thickness), and fibrous cap thickening (< 500 µm) predict plaque rupture.⁴⁹

Plaque ulceration. The risk of stroke increases with worsening degree of plaque ulceration: 0.4% per year for type A ulcerated plaques (small minimal excavations) compared with 12.5% for type B (large obvious excavations) and type C (multiple cavities or cavernous).⁵⁰

Low cerebrovascular reactivity. Perfusion studies such as cerebrovascular reactivity evaluate changes in cerebral blood flow in response to a stimulus such as inhaled carbon dioxide, breath-holding, or acetazolamide. This may provide a useful index of cerebral vascular function. For instance, low reactivity has been associated with ipsilateral ischemic events (odds ratio 14.4, 95% CI 2.63–78.74, P = .0021).^51,52 Silvestrini et al⁵³ reported that the incidence of ipsilateral cerebrovascular ischemic events was 4.1% per year in patients who had normal cerebral vasoreactivity during breath-holding, vs 13.9% in those with low cerebral reactivity.

BEST MEDICAL THERAPY, ALONE OR COMBINED WITH REVASCULARIZATION

For carotid revascularization to be a viable option for asymptomatic carotid stenosis, the morbidity and mortality rates associated with the operation must be less than the incidence of neurologic events in patients who do not undergo the operation.⁵⁴ An important caveat is that the longer a patient survives after carotid endarterectomy, the greater the potential benefit, since the adverse consequences of surgery are generally limited to the perioperative period.¹⁹

The current evidence regarding medical management of asymptomatic carotid stenosis suggests that the rate of ipsilateral stroke is now lower than it was in the control groups in the landmark trials.^{2,3,17,45,47,55,56} Ultimately, adherence to current best medical management takes priority over the decision to revascularize. The best current medical therapy includes, but is not limited to, antithrombotic therapy, statin therapy, blood pressure control, diabetes management, smoking cessation, and lifestyle changes (Table 3).

As noted above, stroke risk seems variable in the asymptomatic population according to the presence or absence of risk factors. Yet no well-defined “high-risk stroke profile” has been identified. Therefore, a patient-by-patient decision based on best available evidence should identify patients who may benefit from carotid revascularization. If asymptomatic carotid stenosis of 70% to 99% is found, factors that favor revascularization are male sex, younger age, and longer life expectancy (Figure 2).

For those with intermediate or high-risk surgical features, uncertainty exists in management since no studies have compared revascularization against medical management only in this group of patients.¹ However, data from high-risk cohorts had high enough complication rates in both intervention arms to question the benefit of revascularization over medical therapy.^20,21 Therefore, the individual perioperative risk of stroke, myocardial infarction, and death must be weighed against the potential benefit of revascularization for each patient.

If revascularization is pursued, studies have demonstrated that carotid artery stenting is not inferior to endarterectomy^15,16 in high-surgical-risk patients. However, the revascularization approach must be tailored to the patient profile, since stenting demonstrated a lower risk of periprocedural myocardial infarction but a higher risk of stroke compared with endarteretomy.²⁰

Finally, the current acceptable risks of perioperative stroke and death must be revised if revascularization is elected. Current data suggest that a lower threshold—around 1.4%—can be used.²⁰ Moreover, further guidelines must determine the impact of adding myocardial infarction to the tolerable perioperative risks, since it has been excluded from main trials and guidelines.²⁰

Carotid artery disease that is asymptomatic poses a dilemma: Should the patient undergo revascularization (surgical carotid endarterectomy or percutaneous stenting) or receive medical therapy alone?

On one hand, because one consequence of carotid atherosclerosis—ischemic stroke—can be devastating or deadly, many physicians and patients would rather “do something,” ie, proceed with surgery. Furthermore, several randomized trials^1–4 found carotid endarterectomy superior to medical therapy.

On the other hand, these trials were conducted in the 1990s. Surgery has improved since then, but so has medical therapy. And if we re-examine the data from the trials in terms of the absolute risk reduction and number needed to treat, as opposed to the relative risk reduction, surgery may appear less beneficial.

Needed is a way to identify patients who would benefit from surgery and those who would more likely be harmed. Research in that direction is ongoing.

Here, we present a simple algorithmic approach to managing asymptomatic carotid artery stenosis based on the patient’s age, sex, and life expectancy. Our approach is based on a review of the best available evidence.

UP TO 8% OF ADULTS HAVE STENOSIS

Stroke is the third largest cause of death in the United States and the leading cause of disability.⁵ From 10% to 15% of strokes are associated with carotid artery stenosis.^6,7

The prevalence of asymptomatic carotid disease, defined as stenosis greater than 50%, ranges from 4% to 8% in adults.⁸

However, major societies recommend against screening for carotid stenosis in the general population.^9–12 Similarly, the US Preventive Services Task Force also discourages the use of carotid auscultation as screening in the general population (Table 1).¹³ Generally, cases of asymptomatic carotid stenosis are diagnosed by ultrasonography after the patient’s physician happens to hear a bruit during a routine examination, during a preoperative assessment, or after the patient suffers a transient ischemic attack or stroke on the contralateral side.

CLASS II RECOMMENDATIONS FOR SURGERY OR STENTING

There are well-established guidelines for managing symptomatic carotid disease,¹⁴ based on evidence from the North American Symptomatic Carotid Endarterectomy Trial¹⁵ and the European Carotid Surgery Trial,¹⁶ both from 1998. But how to manage asymptomatic carotid disease remains uncertain.

If stenosis of the internal carotid artery is greater than 70% on ultrasonography, computed tomography, or magnetic resonance imaging, and if the risk of perioperative stroke and death is low (< 3%), current guidelines¹⁴ give carotid endarterectomy a class IIa recommendation (ie, evidence is conflicting, but the weight of evidence is in favor), and they give prophylactic carotid artery stenting with optimal medical treatment a class IIb recommendation (efficacy is less well established).⁵

But medical management has improved, and new data suggest that this improvement may override the minimal net benefit of intervention in some patients.¹⁷ Some authors suggest that it is best to use patient characteristics and imaging features to guide treatment.¹⁸

EVIDENCE TO SUPPORT CAROTID REVASCULARIZATION

Three major trials (Table 2) published nearly 20 years ago provide the foundation of the current guidelines:

• the Endarterectomy for Asymptomatic Carotid Atherosclerosis Study (ACAS)¹
• the Asymptomatic Carotid Surgery Trial (ACST)^2,3
• the Veterans Affairs (VA) Cooperative Study.⁴

A Cochrane review of these trials,¹⁹ where medical therapy consisted only of aspirin and little use of statin therapy, found that carotid endarterectomy reduced the rate of perioperative stroke or death or any subsequent stroke in the next 3 years by 31% (relative risk 69%, 95% confidence interval [CI] 0.57–0.83). “Perioperative” was defined as the period from randomization until 30 days after surgery in the surgical group and an equivalent period in the medical group.

Moreover, carotid endarterectomy reduced the rate of disabling or fatal nonperioperative stroke by 50% compared with medical management alone.^1,2,19 Patients who had contralateral symptomatic disease or who had undergone contralateral carotid endarterectomy seemed to benefit more from the procedure than those who had not.¹⁹

Also, the ACST investigators found that revascularization was associated with a reduction in contralateral strokes (which occurred in 39 vs 64 patients, P = .01) independent of contralateral symptoms or contralateral carotid endarterectomy.^2,3 The exact mechanism is unknown but could be related to better blood pressure control and risk factor modification after carotid endarterectomy.

Another factor supporting revascularization is that the outcomes of revascularization have improved over time. In 2010, the Carotid Revascularization Endarterectomy Versus Stenting Trial (CREST)²⁰ reported a 30-day periprocedural incidence of death or stroke of only 1.4%, compared with 2.9% in the earlier landmark trials.

Stenting is a noninferior alternative

For patients who have asymptomatic stenosis greater than 80% on color duplex ultrasonography and a risk of stroke or death during carotid endarterectomy that is prohibitively high (> 3%), carotid stenting has proved to be a noninferior alternative.^21,22

The Stenting and Angioplasty With Protection of Patients With High Risk for Endarterectomy (SAPPHIRE) trial²¹ reported a risk of death, stroke, or myocardial infarction of about 5% at 30 days and 10% at 1 year after stenting. A recent observational study revealed lower perioperative complication rates, with a risk of death or stroke of about 3%, which satisfy current guideline requirements.²³

To be deemed at high surgical risk and therefore eligible for the SAPPHIRE trial,²¹ patients had to have clinically significant cardiac disease, severe pulmonary disease, contralateral carotid occlusion, contralateral laryngeal-nerve palsy, recurrent stenosis after carotid endarterectomy, previous radical neck surgery or radiation therapy to the neck, or age greater than 80.

EVIDENCE AGAINST CAROTID REVASCULARIZATION

Although carotid revascularization has evidence to support it, further interpretation of the data may lessen its apparent benefits.

Small absolute benefit, high number needed to treat

If we compare the relative risk reduction for the outcome of perioperative death or any stroke over 5 years (30% to 50%) vs the absolute risk reduction (4% to 5.9%), revascularization seems less attractive.¹⁹

Relative risk reduction in death or stroke with carotid surgery is 30%–50%; absolute risk reduction is 4%–5.9%

The benefit may be further diminished if we consider only strokes related to large vessels, since up to 45% of strokes in patients with carotid disease are lacunar or cardioembolic.²⁴ Assessing for prevention of large-vessel stroke using the ACAS data, the benefit of carotid endarterectomy for prevention of stroke is further decreased to a 3.5% absolute risk reduction, and the number needed to treat for 2 years increases from 62 to 111.^24,25 Nevertheless, revascularization is necessary in appropriately selected patients, as a cerebrovascular event can cause life-altering changes to a patient’s cognitive, emotional, and physical condition.²⁶

Medical therapy—and surgery—are evolving

The optimal medical management used in the landmark studies was significantly different from what is currently recommended. The ACAS trial¹⁸ used only aspirin as optimal medical management, with no mention of statins. In the ACST trial,^2,3 the use of statins increased over time, from 7% to 11% at the beginning of the trial to 80% to 82% at the end.

On the other hand, the ACAS¹ surgeons were required to have an excellent safety record to participate. This might have compromised the trial’s validity or our ability to generalize its conclusions.

Recent data from Abbott¹⁷ suggested a loss of a statistically significant surgical advantage in prevention of ipsilateral stroke and transient ischemic attack from the early 1990s. This is most likely explained by improved medical therapy, since there was a 22% increase in baseline proportion of patients receiving antiplatelet therapy from 1985 to 2007, with 60% of patients taking antihypertensive drugs and 30% of patients taking lipid-lowering drugs. Moreover, since 2001, the annual rates of ipsilateral stroke in patients receiving medical management alone fell below those of patients who underwent carotid endarterectomy in the ACAS trial.

The analysis by Abbott¹⁷ has major limitations: inclusion of small studies, many crossover patients, and heterogeneity. In support of this allegation, a small trial (33 patients) reported a risk of stroke ipsilateral to an asymptomatic carotid stenosis as low as 0.34% per year.²⁵ Even when contrasting the outcomes of medical therapy against those of current carotid endarterectomy, in which the rate of perioperative stroke and death have fallen to 0.88% to 1.7%,^17,27,28 there is concern that the risk associated with surgery may outweigh the long-term benefit.

Flaws in the landmark trials

Beyond the debate of the questionable benefit of revascularization, well-defined flaws in the landmark trials weaken or limit their influence on current treatment guidelines and protocols for deciding whether to revascularize.

No significant benefit was found for patients over age 75.^2,3 This was thought to be due to decreased life expectancy, since the benefit from revascularization becomes significant after 3 years from intervention.^1–3 Also, studies have shown that increasing age is associated with a higher risk of perioperative stroke and death.^20,21

Women showed no benefit at 5 years and only a trend toward benefit at 10 years (P = .05),² likely from a higher rate of periprocedural strokes.

Blacks and Hispanics were underrepresented in the landmark studies,¹⁹ while one observational study reported a higher incidence of in-hospital stroke after carotid endarterectomy in black patients (6.6%) than in white patients (2%).²⁹

When associated with contralateral carotid occlusion, carotid endarterectomy carries a higher risk of perioperative stroke or death.^23,30,31

Carotid revascularization failed to reduce the risk of death—the total number of deaths within 10 years was not significantly reduced by immediate carotid endarterectomy compared with deferring the procedure.²

EVIDENCE SUPPORTING OPTIMAL MEDICAL MANAGEMENT

Optimal medical therapy mainly consists of antiplatelet therapy, blood pressure management, diabetic glycemic control, and statin therapy along with lifestyle changes including smoking cessation, exercise, and weight loss (Table 3).⁹ Detailed recommendations are provided in the American Heart Association/American Stroke Association guidelines for primary prevention of stroke.³²

Antiplatelet therapy has been shown to reduce the incidence of stroke by 25%. There is no added benefit in combining antiplatelet agents unless the patient has concomitant symptomatic coronary artery disease, recent coronary stenting, or severe peripheral artery disease.^33,34

Blood pressure control can reduce the incidence of stroke by 30% to 40%, and recent data suggest that drugs working on the renin-angiotensin system offer more benefit than beta-blockers for the same reduction in blood pressure.^34,35

Diabetic glycemic control is supported, as higher hemoglobin A_1c and fasting glucose values are associated with higher relative risk of stroke.^32,36,37 However, the stroke rate does not differ significantly between patients receiving intensive therapy and those receiving standard therapy.³⁴

Statins actually shrink carotid plaques and reduce the risk of stroke by 15% for each 10% reduction in low-density lipoprotein cholesterol. It is estimated that statin therapy confers a 30% relative risk reduction of stroke over 20 years.^34,38–41

Smoking increases the overall risk of stroke by 150%, making its cessation mandatory.⁴²

HIGH-RISK FEATURES FOR STROKE IN ASYMPTOMATIC CAROTID STENOSIS

Studies have tried to identify risk factors for stroke, so that patients at high risk could undergo revascularization and benefit from it. However, no well-defined high-risk features have yet been described that would identify patients who would benefit from early surgery.

For instance, no correlation has been found between age, sex, diabetes mellitus, lipid levels, or smoking and progression of disease.⁴³ In contrast, having either contralateral symptomatic carotid disease or contralateral total occlusion translated into a higher ipsilateral stroke risk.¹⁸ And in several studies, the 5-year risk of ipsilateral stroke was as high as 16.2% for those with 60% to 99% stenosis.^1,2,18,24,43

Features of the plaque itself

More recently, there has been a focus on plaque evaluation to predict outcomes.

Statins shrink carotid plaques and reduce the risk of stroke by 15% for each 10% reduction in LDL-C

Percent stenosis. An increased risk of death or stroke has been reported with higher degrees of stenosis or plaque progression.^44,45 The gross annual risk of ipsilateral stroke increases from 1.5% with stenosis of 60% to 70%, to 4.2% with stenosis of 71% to 90%, and to 7% with stenosis of 91% to 99%. Nevertheless, current data are insufficient to determine whether there is increasing benefit from surgery with increasing degree of stenosis in asymptomatic carotid disease.^1,3,24,44

Plaque progression translates to a 7.2% absolute increase in the incidence of stroke (1.1% if the plaque is stable vs 8.3% if the plaque is progressing). Interestingly, plaque progression to greater than 80% stenosis results in worse outcomes (relative risk 3.4, 95% CI 1.5–7.8) compared with the same level of stenosis without recent progression.³³

Intimal wall thickening of more than 1.15 mm confers a hazard ratio for stroke of 3 (95% CI 1.48–6.11).⁴⁶

Increased echolucency also confers a hazard ratio for stroke of 3 (95% CI 1.4–8.0).⁴⁶

A low gray-scale median (a surrogate of plaque composition) and plaque area have been identified as independent predictors of ipsilateral events.⁴⁴

Embolic signals on transcranial Doppler ultrasonography (Figure 1) have been associated with a hazard ratio for stroke of 2.54 over 2 years.⁴⁷

Carotid plaques predominantly composed of lipid-rich necrotic cores carry a higher risk of stroke (hazard ratio 7.2, 95% CI 1.12–46.20).⁴⁸

High tensile stress (circumferential wall tension divided by the intima-media thickness), and fibrous cap thickening (< 500 µm) predict plaque rupture.⁴⁹

Plaque ulceration. The risk of stroke increases with worsening degree of plaque ulceration: 0.4% per year for type A ulcerated plaques (small minimal excavations) compared with 12.5% for type B (large obvious excavations) and type C (multiple cavities or cavernous).⁵⁰

Low cerebrovascular reactivity. Perfusion studies such as cerebrovascular reactivity evaluate changes in cerebral blood flow in response to a stimulus such as inhaled carbon dioxide, breath-holding, or acetazolamide. This may provide a useful index of cerebral vascular function. For instance, low reactivity has been associated with ipsilateral ischemic events (odds ratio 14.4, 95% CI 2.63–78.74, P = .0021).^51,52 Silvestrini et al⁵³ reported that the incidence of ipsilateral cerebrovascular ischemic events was 4.1% per year in patients who had normal cerebral vasoreactivity during breath-holding, vs 13.9% in those with low cerebral reactivity.

BEST MEDICAL THERAPY, ALONE OR COMBINED WITH REVASCULARIZATION

For carotid revascularization to be a viable option for asymptomatic carotid stenosis, the morbidity and mortality rates associated with the operation must be less than the incidence of neurologic events in patients who do not undergo the operation.⁵⁴ An important caveat is that the longer a patient survives after carotid endarterectomy, the greater the potential benefit, since the adverse consequences of surgery are generally limited to the perioperative period.¹⁹

The current evidence regarding medical management of asymptomatic carotid stenosis suggests that the rate of ipsilateral stroke is now lower than it was in the control groups in the landmark trials.^{2,3,17,45,47,55,56} Ultimately, adherence to current best medical management takes priority over the decision to revascularize. The best current medical therapy includes, but is not limited to, antithrombotic therapy, statin therapy, blood pressure control, diabetes management, smoking cessation, and lifestyle changes (Table 3).

As noted above, stroke risk seems variable in the asymptomatic population according to the presence or absence of risk factors. Yet no well-defined “high-risk stroke profile” has been identified. Therefore, a patient-by-patient decision based on best available evidence should identify patients who may benefit from carotid revascularization. If asymptomatic carotid stenosis of 70% to 99% is found, factors that favor revascularization are male sex, younger age, and longer life expectancy (Figure 2).

For those with intermediate or high-risk surgical features, uncertainty exists in management since no studies have compared revascularization against medical management only in this group of patients.¹ However, data from high-risk cohorts had high enough complication rates in both intervention arms to question the benefit of revascularization over medical therapy.^20,21 Therefore, the individual perioperative risk of stroke, myocardial infarction, and death must be weighed against the potential benefit of revascularization for each patient.

If revascularization is pursued, studies have demonstrated that carotid artery stenting is not inferior to endarterectomy^15,16 in high-surgical-risk patients. However, the revascularization approach must be tailored to the patient profile, since stenting demonstrated a lower risk of periprocedural myocardial infarction but a higher risk of stroke compared with endarteretomy.²⁰

Finally, the current acceptable risks of perioperative stroke and death must be revised if revascularization is elected. Current data suggest that a lower threshold—around 1.4%—can be used.²⁰ Moreover, further guidelines must determine the impact of adding myocardial infarction to the tolerable perioperative risks, since it has been excluded from main trials and guidelines.²⁰

Carotid artery disease that is asymptomatic poses a dilemma: Should the patient undergo revascularization (surgical carotid endarterectomy or percutaneous stenting) or receive medical therapy alone?

On one hand, because one consequence of carotid atherosclerosis—ischemic stroke—can be devastating or deadly, many physicians and patients would rather “do something,” ie, proceed with surgery. Furthermore, several randomized trials^1–4 found carotid endarterectomy superior to medical therapy.

On the other hand, these trials were conducted in the 1990s. Surgery has improved since then, but so has medical therapy. And if we re-examine the data from the trials in terms of the absolute risk reduction and number needed to treat, as opposed to the relative risk reduction, surgery may appear less beneficial.

Needed is a way to identify patients who would benefit from surgery and those who would more likely be harmed. Research in that direction is ongoing.

Here, we present a simple algorithmic approach to managing asymptomatic carotid artery stenosis based on the patient’s age, sex, and life expectancy. Our approach is based on a review of the best available evidence.

UP TO 8% OF ADULTS HAVE STENOSIS

Stroke is the third largest cause of death in the United States and the leading cause of disability.⁵ From 10% to 15% of strokes are associated with carotid artery stenosis.^6,7

The prevalence of asymptomatic carotid disease, defined as stenosis greater than 50%, ranges from 4% to 8% in adults.⁸

However, major societies recommend against screening for carotid stenosis in the general population.^9–12 Similarly, the US Preventive Services Task Force also discourages the use of carotid auscultation as screening in the general population (Table 1).¹³ Generally, cases of asymptomatic carotid stenosis are diagnosed by ultrasonography after the patient’s physician happens to hear a bruit during a routine examination, during a preoperative assessment, or after the patient suffers a transient ischemic attack or stroke on the contralateral side.

CLASS II RECOMMENDATIONS FOR SURGERY OR STENTING

There are well-established guidelines for managing symptomatic carotid disease,¹⁴ based on evidence from the North American Symptomatic Carotid Endarterectomy Trial¹⁵ and the European Carotid Surgery Trial,¹⁶ both from 1998. But how to manage asymptomatic carotid disease remains uncertain.

If stenosis of the internal carotid artery is greater than 70% on ultrasonography, computed tomography, or magnetic resonance imaging, and if the risk of perioperative stroke and death is low (< 3%), current guidelines¹⁴ give carotid endarterectomy a class IIa recommendation (ie, evidence is conflicting, but the weight of evidence is in favor), and they give prophylactic carotid artery stenting with optimal medical treatment a class IIb recommendation (efficacy is less well established).⁵

But medical management has improved, and new data suggest that this improvement may override the minimal net benefit of intervention in some patients.¹⁷ Some authors suggest that it is best to use patient characteristics and imaging features to guide treatment.¹⁸

EVIDENCE TO SUPPORT CAROTID REVASCULARIZATION

Three major trials (Table 2) published nearly 20 years ago provide the foundation of the current guidelines:

• the Endarterectomy for Asymptomatic Carotid Atherosclerosis Study (ACAS)¹
• the Asymptomatic Carotid Surgery Trial (ACST)^2,3
• the Veterans Affairs (VA) Cooperative Study.⁴

A Cochrane review of these trials,¹⁹ where medical therapy consisted only of aspirin and little use of statin therapy, found that carotid endarterectomy reduced the rate of perioperative stroke or death or any subsequent stroke in the next 3 years by 31% (relative risk 69%, 95% confidence interval [CI] 0.57–0.83). “Perioperative” was defined as the period from randomization until 30 days after surgery in the surgical group and an equivalent period in the medical group.

Moreover, carotid endarterectomy reduced the rate of disabling or fatal nonperioperative stroke by 50% compared with medical management alone.^1,2,19 Patients who had contralateral symptomatic disease or who had undergone contralateral carotid endarterectomy seemed to benefit more from the procedure than those who had not.¹⁹

Also, the ACST investigators found that revascularization was associated with a reduction in contralateral strokes (which occurred in 39 vs 64 patients, P = .01) independent of contralateral symptoms or contralateral carotid endarterectomy.^2,3 The exact mechanism is unknown but could be related to better blood pressure control and risk factor modification after carotid endarterectomy.

Another factor supporting revascularization is that the outcomes of revascularization have improved over time. In 2010, the Carotid Revascularization Endarterectomy Versus Stenting Trial (CREST)²⁰ reported a 30-day periprocedural incidence of death or stroke of only 1.4%, compared with 2.9% in the earlier landmark trials.

Stenting is a noninferior alternative

For patients who have asymptomatic stenosis greater than 80% on color duplex ultrasonography and a risk of stroke or death during carotid endarterectomy that is prohibitively high (> 3%), carotid stenting has proved to be a noninferior alternative.^21,22

The Stenting and Angioplasty With Protection of Patients With High Risk for Endarterectomy (SAPPHIRE) trial²¹ reported a risk of death, stroke, or myocardial infarction of about 5% at 30 days and 10% at 1 year after stenting. A recent observational study revealed lower perioperative complication rates, with a risk of death or stroke of about 3%, which satisfy current guideline requirements.²³

To be deemed at high surgical risk and therefore eligible for the SAPPHIRE trial,²¹ patients had to have clinically significant cardiac disease, severe pulmonary disease, contralateral carotid occlusion, contralateral laryngeal-nerve palsy, recurrent stenosis after carotid endarterectomy, previous radical neck surgery or radiation therapy to the neck, or age greater than 80.

EVIDENCE AGAINST CAROTID REVASCULARIZATION

Although carotid revascularization has evidence to support it, further interpretation of the data may lessen its apparent benefits.

Small absolute benefit, high number needed to treat

If we compare the relative risk reduction for the outcome of perioperative death or any stroke over 5 years (30% to 50%) vs the absolute risk reduction (4% to 5.9%), revascularization seems less attractive.¹⁹

Relative risk reduction in death or stroke with carotid surgery is 30%–50%; absolute risk reduction is 4%–5.9%

The benefit may be further diminished if we consider only strokes related to large vessels, since up to 45% of strokes in patients with carotid disease are lacunar or cardioembolic.²⁴ Assessing for prevention of large-vessel stroke using the ACAS data, the benefit of carotid endarterectomy for prevention of stroke is further decreased to a 3.5% absolute risk reduction, and the number needed to treat for 2 years increases from 62 to 111.^24,25 Nevertheless, revascularization is necessary in appropriately selected patients, as a cerebrovascular event can cause life-altering changes to a patient’s cognitive, emotional, and physical condition.²⁶

Medical therapy—and surgery—are evolving

The optimal medical management used in the landmark studies was significantly different from what is currently recommended. The ACAS trial¹⁸ used only aspirin as optimal medical management, with no mention of statins. In the ACST trial,^2,3 the use of statins increased over time, from 7% to 11% at the beginning of the trial to 80% to 82% at the end.

On the other hand, the ACAS¹ surgeons were required to have an excellent safety record to participate. This might have compromised the trial’s validity or our ability to generalize its conclusions.

Recent data from Abbott¹⁷ suggested a loss of a statistically significant surgical advantage in prevention of ipsilateral stroke and transient ischemic attack from the early 1990s. This is most likely explained by improved medical therapy, since there was a 22% increase in baseline proportion of patients receiving antiplatelet therapy from 1985 to 2007, with 60% of patients taking antihypertensive drugs and 30% of patients taking lipid-lowering drugs. Moreover, since 2001, the annual rates of ipsilateral stroke in patients receiving medical management alone fell below those of patients who underwent carotid endarterectomy in the ACAS trial.

The analysis by Abbott¹⁷ has major limitations: inclusion of small studies, many crossover patients, and heterogeneity. In support of this allegation, a small trial (33 patients) reported a risk of stroke ipsilateral to an asymptomatic carotid stenosis as low as 0.34% per year.²⁵ Even when contrasting the outcomes of medical therapy against those of current carotid endarterectomy, in which the rate of perioperative stroke and death have fallen to 0.88% to 1.7%,^17,27,28 there is concern that the risk associated with surgery may outweigh the long-term benefit.

Flaws in the landmark trials

Beyond the debate of the questionable benefit of revascularization, well-defined flaws in the landmark trials weaken or limit their influence on current treatment guidelines and protocols for deciding whether to revascularize.

No significant benefit was found for patients over age 75.^2,3 This was thought to be due to decreased life expectancy, since the benefit from revascularization becomes significant after 3 years from intervention.^1–3 Also, studies have shown that increasing age is associated with a higher risk of perioperative stroke and death.^20,21

Women showed no benefit at 5 years and only a trend toward benefit at 10 years (P = .05),² likely from a higher rate of periprocedural strokes.

Blacks and Hispanics were underrepresented in the landmark studies,¹⁹ while one observational study reported a higher incidence of in-hospital stroke after carotid endarterectomy in black patients (6.6%) than in white patients (2%).²⁹

When associated with contralateral carotid occlusion, carotid endarterectomy carries a higher risk of perioperative stroke or death.^23,30,31

Carotid revascularization failed to reduce the risk of death—the total number of deaths within 10 years was not significantly reduced by immediate carotid endarterectomy compared with deferring the procedure.²

EVIDENCE SUPPORTING OPTIMAL MEDICAL MANAGEMENT

Optimal medical therapy mainly consists of antiplatelet therapy, blood pressure management, diabetic glycemic control, and statin therapy along with lifestyle changes including smoking cessation, exercise, and weight loss (Table 3).⁹ Detailed recommendations are provided in the American Heart Association/American Stroke Association guidelines for primary prevention of stroke.³²

Antiplatelet therapy has been shown to reduce the incidence of stroke by 25%. There is no added benefit in combining antiplatelet agents unless the patient has concomitant symptomatic coronary artery disease, recent coronary stenting, or severe peripheral artery disease.^33,34

Blood pressure control can reduce the incidence of stroke by 30% to 40%, and recent data suggest that drugs working on the renin-angiotensin system offer more benefit than beta-blockers for the same reduction in blood pressure.^34,35

Diabetic glycemic control is supported, as higher hemoglobin A_1c and fasting glucose values are associated with higher relative risk of stroke.^32,36,37 However, the stroke rate does not differ significantly between patients receiving intensive therapy and those receiving standard therapy.³⁴

Statins actually shrink carotid plaques and reduce the risk of stroke by 15% for each 10% reduction in low-density lipoprotein cholesterol. It is estimated that statin therapy confers a 30% relative risk reduction of stroke over 20 years.^34,38–41

Smoking increases the overall risk of stroke by 150%, making its cessation mandatory.⁴²

HIGH-RISK FEATURES FOR STROKE IN ASYMPTOMATIC CAROTID STENOSIS

Studies have tried to identify risk factors for stroke, so that patients at high risk could undergo revascularization and benefit from it. However, no well-defined high-risk features have yet been described that would identify patients who would benefit from early surgery.

For instance, no correlation has been found between age, sex, diabetes mellitus, lipid levels, or smoking and progression of disease.⁴³ In contrast, having either contralateral symptomatic carotid disease or contralateral total occlusion translated into a higher ipsilateral stroke risk.¹⁸ And in several studies, the 5-year risk of ipsilateral stroke was as high as 16.2% for those with 60% to 99% stenosis.^1,2,18,24,43

Features of the plaque itself

More recently, there has been a focus on plaque evaluation to predict outcomes.

Statins shrink carotid plaques and reduce the risk of stroke by 15% for each 10% reduction in LDL-C

Percent stenosis. An increased risk of death or stroke has been reported with higher degrees of stenosis or plaque progression.^44,45 The gross annual risk of ipsilateral stroke increases from 1.5% with stenosis of 60% to 70%, to 4.2% with stenosis of 71% to 90%, and to 7% with stenosis of 91% to 99%. Nevertheless, current data are insufficient to determine whether there is increasing benefit from surgery with increasing degree of stenosis in asymptomatic carotid disease.^1,3,24,44

Plaque progression translates to a 7.2% absolute increase in the incidence of stroke (1.1% if the plaque is stable vs 8.3% if the plaque is progressing). Interestingly, plaque progression to greater than 80% stenosis results in worse outcomes (relative risk 3.4, 95% CI 1.5–7.8) compared with the same level of stenosis without recent progression.³³

Intimal wall thickening of more than 1.15 mm confers a hazard ratio for stroke of 3 (95% CI 1.48–6.11).⁴⁶

Increased echolucency also confers a hazard ratio for stroke of 3 (95% CI 1.4–8.0).⁴⁶

A low gray-scale median (a surrogate of plaque composition) and plaque area have been identified as independent predictors of ipsilateral events.⁴⁴

Embolic signals on transcranial Doppler ultrasonography (Figure 1) have been associated with a hazard ratio for stroke of 2.54 over 2 years.⁴⁷

Carotid plaques predominantly composed of lipid-rich necrotic cores carry a higher risk of stroke (hazard ratio 7.2, 95% CI 1.12–46.20).⁴⁸

High tensile stress (circumferential wall tension divided by the intima-media thickness), and fibrous cap thickening (< 500 µm) predict plaque rupture.⁴⁹

Plaque ulceration. The risk of stroke increases with worsening degree of plaque ulceration: 0.4% per year for type A ulcerated plaques (small minimal excavations) compared with 12.5% for type B (large obvious excavations) and type C (multiple cavities or cavernous).⁵⁰

Low cerebrovascular reactivity. Perfusion studies such as cerebrovascular reactivity evaluate changes in cerebral blood flow in response to a stimulus such as inhaled carbon dioxide, breath-holding, or acetazolamide. This may provide a useful index of cerebral vascular function. For instance, low reactivity has been associated with ipsilateral ischemic events (odds ratio 14.4, 95% CI 2.63–78.74, P = .0021).^51,52 Silvestrini et al⁵³ reported that the incidence of ipsilateral cerebrovascular ischemic events was 4.1% per year in patients who had normal cerebral vasoreactivity during breath-holding, vs 13.9% in those with low cerebral reactivity.

BEST MEDICAL THERAPY, ALONE OR COMBINED WITH REVASCULARIZATION

For carotid revascularization to be a viable option for asymptomatic carotid stenosis, the morbidity and mortality rates associated with the operation must be less than the incidence of neurologic events in patients who do not undergo the operation.⁵⁴ An important caveat is that the longer a patient survives after carotid endarterectomy, the greater the potential benefit, since the adverse consequences of surgery are generally limited to the perioperative period.¹⁹

The current evidence regarding medical management of asymptomatic carotid stenosis suggests that the rate of ipsilateral stroke is now lower than it was in the control groups in the landmark trials.^{2,3,17,45,47,55,56} Ultimately, adherence to current best medical management takes priority over the decision to revascularize. The best current medical therapy includes, but is not limited to, antithrombotic therapy, statin therapy, blood pressure control, diabetes management, smoking cessation, and lifestyle changes (Table 3).

As noted above, stroke risk seems variable in the asymptomatic population according to the presence or absence of risk factors. Yet no well-defined “high-risk stroke profile” has been identified. Therefore, a patient-by-patient decision based on best available evidence should identify patients who may benefit from carotid revascularization. If asymptomatic carotid stenosis of 70% to 99% is found, factors that favor revascularization are male sex, younger age, and longer life expectancy (Figure 2).

For those with intermediate or high-risk surgical features, uncertainty exists in management since no studies have compared revascularization against medical management only in this group of patients.¹ However, data from high-risk cohorts had high enough complication rates in both intervention arms to question the benefit of revascularization over medical therapy.^20,21 Therefore, the individual perioperative risk of stroke, myocardial infarction, and death must be weighed against the potential benefit of revascularization for each patient.

If revascularization is pursued, studies have demonstrated that carotid artery stenting is not inferior to endarterectomy^15,16 in high-surgical-risk patients. However, the revascularization approach must be tailored to the patient profile, since stenting demonstrated a lower risk of periprocedural myocardial infarction but a higher risk of stroke compared with endarteretomy.²⁰

Finally, the current acceptable risks of perioperative stroke and death must be revised if revascularization is elected. Current data suggest that a lower threshold—around 1.4%—can be used.²⁰ Moreover, further guidelines must determine the impact of adding myocardial infarction to the tolerable perioperative risks, since it has been excluded from main trials and guidelines.²⁰

When you finish reading this column … on second thought, stop now and read the Oct. 17, 2015, opinion piece titled “Overselling Breast-Feeding.” You will discover a well-researched and thoughtfully crafted article by Courtney Jung, a political science professor at the University of Toronto, in which she dares to carefully dissect one of our most revered sacred cows. The result is a convincing argument for rethinking how we present and promote breastfeeding. I won’t attempt to reconstruct her rationale. You can read it for yourself. But, I suspect that if you spend any part of your day trying to help new parents navigate the choppy waters of those first 6 months, you will find what she has to say strikes more than a few familiar chords.

Like most of you, what I learned about breastfeeding came as on the job training. Marilyn and I started our family while I was still in medical school, giving me the advantage of having watched the process bump along twice before I found myself on the frontline of private practice. I had been taught in school about all the advantages breast milk, but it didn’t take long in the real world to discover that breastfeeding could have a dark side.

I had to become a chameleon. I needed to be strong advocate for the advantages of breast milk and support new mothers as they tried to match the American Academy of Pediatrics’ guidelines. However, there were situations in which despite everyone’s best efforts, the handwriting on the wall said, “This isn’t working.” Then it was time to change my colors and convincingly convey the new truth that even a baby that isn’t breastfed is going to be fine. That a woman who doesn’t breastfeed can and will be a mother every bit as good as one who doesn’t breastfed her baby for 6 months or a year.

©Jupiterimages/thinkstockphotos.com

The tension between the party line and reality became so great that in frustration I decided to write my third book about breastfeeding. The result was “The Maternity Leave Breastfeeding Plan” (New York: Simon and Schuster, 2002). The watered-down title was chosen by the publisher. The subtitle, “How to Enjoy Nursing for 3 Months and Go Back to Work Guilt-Free,” was a better reflection of my message that there can be some serious challenges to breastfeeding and not to worry if it doesn’t work. Surprisingly, it found itself on a La Leche League list of recommended books – that is until someone in the organization actually read it.

Although I had always harbored doubts that many of the studies purporting to show the advantages of breastfeeding were poorly controlled, in 2002, I couldn’t find any data to support my concerns. But over the last decade those studies have begun to emerge and Professor Jung has found them and included them in her new book, “Lactivism: How Feminists and Fundamentalists, Hippies and Yuppies, and Physicians and Politicians Made Breastfeeding Big Business and Bad Policy” (New York: Basic Books, 2015).

It will be interesting to see how her observations play to the wider audience it deserves. The discussions may be lively and heated, and public opinion may shift a bit. But what won’t change is that those of us who deal with mothers and babies in a very personal way will still have to struggle with promoting a good product that isn’t always easy to obtain.

Breast milk is good … but it isn’t always better or best.

Dr. Wilkoff practiced primary care pediatrics in Brunswick, Maine, for nearly 40 years. He has authored several books on behavioral pediatrics including “How to Say No to Your Toddler.”

When you finish reading this column … on second thought, stop now and read the Oct. 17, 2015, opinion piece titled “Overselling Breast-Feeding.” You will discover a well-researched and thoughtfully crafted article by Courtney Jung, a political science professor at the University of Toronto, in which she dares to carefully dissect one of our most revered sacred cows. The result is a convincing argument for rethinking how we present and promote breastfeeding. I won’t attempt to reconstruct her rationale. You can read it for yourself. But, I suspect that if you spend any part of your day trying to help new parents navigate the choppy waters of those first 6 months, you will find what she has to say strikes more than a few familiar chords.

Like most of you, what I learned about breastfeeding came as on the job training. Marilyn and I started our family while I was still in medical school, giving me the advantage of having watched the process bump along twice before I found myself on the frontline of private practice. I had been taught in school about all the advantages breast milk, but it didn’t take long in the real world to discover that breastfeeding could have a dark side.

I had to become a chameleon. I needed to be strong advocate for the advantages of breast milk and support new mothers as they tried to match the American Academy of Pediatrics’ guidelines. However, there were situations in which despite everyone’s best efforts, the handwriting on the wall said, “This isn’t working.” Then it was time to change my colors and convincingly convey the new truth that even a baby that isn’t breastfed is going to be fine. That a woman who doesn’t breastfeed can and will be a mother every bit as good as one who doesn’t breastfed her baby for 6 months or a year.

©Jupiterimages/thinkstockphotos.com

The tension between the party line and reality became so great that in frustration I decided to write my third book about breastfeeding. The result was “The Maternity Leave Breastfeeding Plan” (New York: Simon and Schuster, 2002). The watered-down title was chosen by the publisher. The subtitle, “How to Enjoy Nursing for 3 Months and Go Back to Work Guilt-Free,” was a better reflection of my message that there can be some serious challenges to breastfeeding and not to worry if it doesn’t work. Surprisingly, it found itself on a La Leche League list of recommended books – that is until someone in the organization actually read it.

Although I had always harbored doubts that many of the studies purporting to show the advantages of breastfeeding were poorly controlled, in 2002, I couldn’t find any data to support my concerns. But over the last decade those studies have begun to emerge and Professor Jung has found them and included them in her new book, “Lactivism: How Feminists and Fundamentalists, Hippies and Yuppies, and Physicians and Politicians Made Breastfeeding Big Business and Bad Policy” (New York: Basic Books, 2015).

It will be interesting to see how her observations play to the wider audience it deserves. The discussions may be lively and heated, and public opinion may shift a bit. But what won’t change is that those of us who deal with mothers and babies in a very personal way will still have to struggle with promoting a good product that isn’t always easy to obtain.

Breast milk is good … but it isn’t always better or best.

Dr. Wilkoff practiced primary care pediatrics in Brunswick, Maine, for nearly 40 years. He has authored several books on behavioral pediatrics including “How to Say No to Your Toddler.”

When you finish reading this column … on second thought, stop now and read the Oct. 17, 2015, opinion piece titled “Overselling Breast-Feeding.” You will discover a well-researched and thoughtfully crafted article by Courtney Jung, a political science professor at the University of Toronto, in which she dares to carefully dissect one of our most revered sacred cows. The result is a convincing argument for rethinking how we present and promote breastfeeding. I won’t attempt to reconstruct her rationale. You can read it for yourself. But, I suspect that if you spend any part of your day trying to help new parents navigate the choppy waters of those first 6 months, you will find what she has to say strikes more than a few familiar chords.

Like most of you, what I learned about breastfeeding came as on the job training. Marilyn and I started our family while I was still in medical school, giving me the advantage of having watched the process bump along twice before I found myself on the frontline of private practice. I had been taught in school about all the advantages breast milk, but it didn’t take long in the real world to discover that breastfeeding could have a dark side.

I had to become a chameleon. I needed to be strong advocate for the advantages of breast milk and support new mothers as they tried to match the American Academy of Pediatrics’ guidelines. However, there were situations in which despite everyone’s best efforts, the handwriting on the wall said, “This isn’t working.” Then it was time to change my colors and convincingly convey the new truth that even a baby that isn’t breastfed is going to be fine. That a woman who doesn’t breastfeed can and will be a mother every bit as good as one who doesn’t breastfed her baby for 6 months or a year.

©Jupiterimages/thinkstockphotos.com

The tension between the party line and reality became so great that in frustration I decided to write my third book about breastfeeding. The result was “The Maternity Leave Breastfeeding Plan” (New York: Simon and Schuster, 2002). The watered-down title was chosen by the publisher. The subtitle, “How to Enjoy Nursing for 3 Months and Go Back to Work Guilt-Free,” was a better reflection of my message that there can be some serious challenges to breastfeeding and not to worry if it doesn’t work. Surprisingly, it found itself on a La Leche League list of recommended books – that is until someone in the organization actually read it.

Although I had always harbored doubts that many of the studies purporting to show the advantages of breastfeeding were poorly controlled, in 2002, I couldn’t find any data to support my concerns. But over the last decade those studies have begun to emerge and Professor Jung has found them and included them in her new book, “Lactivism: How Feminists and Fundamentalists, Hippies and Yuppies, and Physicians and Politicians Made Breastfeeding Big Business and Bad Policy” (New York: Basic Books, 2015).

It will be interesting to see how her observations play to the wider audience it deserves. The discussions may be lively and heated, and public opinion may shift a bit. But what won’t change is that those of us who deal with mothers and babies in a very personal way will still have to struggle with promoting a good product that isn’t always easy to obtain.

Breast milk is good … but it isn’t always better or best.

Dr. Wilkoff practiced primary care pediatrics in Brunswick, Maine, for nearly 40 years. He has authored several books on behavioral pediatrics including “How to Say No to Your Toddler.”

Imagine yourself in a small community hospital standing at the bedside of a tiny preemie waiting for the neonatal transport team to return your call for help.

With one eye shifting between the clock and the oximeter, you have the other one looking out the window hoping that the predicted snow and freezing rain will hold out for another hour. You have done everything you can do, but clearly it’s not going to be enough to rescue this little person who had the misfortune of exiting the birth canal several months too early.

You have been able to insert an umbilical vein catheter and miraculously have threaded an endotracheal tube into a trachea that looked no bigger than a piece of spaghetti, or maybe you have failed and the nurses are taking turns bagging. The transport team returns your call for help and with apologies reports that they are tied up with a similar scenario further south; they predict that it may be an hour and a half before they will be able to get back to their hospital, which is a half hour down the road from you.

They suggest some things that you have already done. Should you wait for more skilled hands and their equipment or transport the patient yourself and get on the road before it becomes a skating rink? There is an antique transport isolette gathering dust in the storage room down the hall, and the local fire department ambulance crew with whom you are on a first-name basis is always ready to help. Is it time to gather the troops and tell them, “Let’s roll!” ?

If you have ever lived through a similar scenario, you may find a recent study interesting (Ann Intern Med. 2015;163[9]:681-90). What these investigators found was that for adults who had suffered major trauma, stroke, respiratory failure, and acute myocardial infarction, those who were transported by crews with basic life support (BLS) skills had significantly better long-term survival and neurologic outcomes than did those victims transported by crews with advanced life support (ALS) skills.

In the flurry of comments that circulated following the release of the study were a few questions about the methodology, but most commentators were searching for an explanation. Was critical time lost by the ALS crews doing stuff when the better course of action would have been to get the ambulance rolling to the hospital and more definitive care? Does the temptation to do things because you can do them sometimes cloud the decision-making process?

Although I have lived the scenario I described, it is less likely to happen now. Backup teams from other institutions may be activated. The teams are so well equipped and trained that the gaps between their capabilities and the neonatal intensive care unit have narrowed, but there is no question that they remain and are significant.

The other thing that hasn’t changed is the weather here in Maine. While we have beautiful summers that prompt us to put “Vacationland” on our license plates, our winters are a challenge. In addition to the patient’s condition and the availability of resources, the decision of whether to invest time in stabilization or get moving toward the referral center also must include the risk to the patient and staff who will be traveling on weather-threatened roads.

On the other hand, we can’t ignore the elephant that occasionally finds its way into the room when decisions are made about how thoroughly a critically ill patient is stabilized and how speedily he is transferred. And, that ponderous pachyderm is the hot potato factor and sometimes answers to its acronym, NIMBY (“not in my back yard”). You know as well as I do that despite the Emergency Medical Treatment and Active Labor Act (EMTALA) regulations, there are cases when a patient is hustled out the door without being appropriately stabilized primarily to avoid having that patient die in the referring hospital. We must continue to ask ourselves if we have done everything that we can do to stabilize the patient before we say, “Let’s roll!”

William G. Wilkoff, M.D., practiced primary care pediatrics in Brunswick, Maine, for nearly 40 years. He has authored several books on behavioral pediatrics including “How to Say No to Your Toddler.”

Imagine yourself in a small community hospital standing at the bedside of a tiny preemie waiting for the neonatal transport team to return your call for help.

With one eye shifting between the clock and the oximeter, you have the other one looking out the window hoping that the predicted snow and freezing rain will hold out for another hour. You have done everything you can do, but clearly it’s not going to be enough to rescue this little person who had the misfortune of exiting the birth canal several months too early.

You have been able to insert an umbilical vein catheter and miraculously have threaded an endotracheal tube into a trachea that looked no bigger than a piece of spaghetti, or maybe you have failed and the nurses are taking turns bagging. The transport team returns your call for help and with apologies reports that they are tied up with a similar scenario further south; they predict that it may be an hour and a half before they will be able to get back to their hospital, which is a half hour down the road from you.

They suggest some things that you have already done. Should you wait for more skilled hands and their equipment or transport the patient yourself and get on the road before it becomes a skating rink? There is an antique transport isolette gathering dust in the storage room down the hall, and the local fire department ambulance crew with whom you are on a first-name basis is always ready to help. Is it time to gather the troops and tell them, “Let’s roll!” ?

If you have ever lived through a similar scenario, you may find a recent study interesting (Ann Intern Med. 2015;163[9]:681-90). What these investigators found was that for adults who had suffered major trauma, stroke, respiratory failure, and acute myocardial infarction, those who were transported by crews with basic life support (BLS) skills had significantly better long-term survival and neurologic outcomes than did those victims transported by crews with advanced life support (ALS) skills.

In the flurry of comments that circulated following the release of the study were a few questions about the methodology, but most commentators were searching for an explanation. Was critical time lost by the ALS crews doing stuff when the better course of action would have been to get the ambulance rolling to the hospital and more definitive care? Does the temptation to do things because you can do them sometimes cloud the decision-making process?

Although I have lived the scenario I described, it is less likely to happen now. Backup teams from other institutions may be activated. The teams are so well equipped and trained that the gaps between their capabilities and the neonatal intensive care unit have narrowed, but there is no question that they remain and are significant.

The other thing that hasn’t changed is the weather here in Maine. While we have beautiful summers that prompt us to put “Vacationland” on our license plates, our winters are a challenge. In addition to the patient’s condition and the availability of resources, the decision of whether to invest time in stabilization or get moving toward the referral center also must include the risk to the patient and staff who will be traveling on weather-threatened roads.

On the other hand, we can’t ignore the elephant that occasionally finds its way into the room when decisions are made about how thoroughly a critically ill patient is stabilized and how speedily he is transferred. And, that ponderous pachyderm is the hot potato factor and sometimes answers to its acronym, NIMBY (“not in my back yard”). You know as well as I do that despite the Emergency Medical Treatment and Active Labor Act (EMTALA) regulations, there are cases when a patient is hustled out the door without being appropriately stabilized primarily to avoid having that patient die in the referring hospital. We must continue to ask ourselves if we have done everything that we can do to stabilize the patient before we say, “Let’s roll!”

William G. Wilkoff, M.D., practiced primary care pediatrics in Brunswick, Maine, for nearly 40 years. He has authored several books on behavioral pediatrics including “How to Say No to Your Toddler.”

Imagine yourself in a small community hospital standing at the bedside of a tiny preemie waiting for the neonatal transport team to return your call for help.

With one eye shifting between the clock and the oximeter, you have the other one looking out the window hoping that the predicted snow and freezing rain will hold out for another hour. You have done everything you can do, but clearly it’s not going to be enough to rescue this little person who had the misfortune of exiting the birth canal several months too early.

You have been able to insert an umbilical vein catheter and miraculously have threaded an endotracheal tube into a trachea that looked no bigger than a piece of spaghetti, or maybe you have failed and the nurses are taking turns bagging. The transport team returns your call for help and with apologies reports that they are tied up with a similar scenario further south; they predict that it may be an hour and a half before they will be able to get back to their hospital, which is a half hour down the road from you.

They suggest some things that you have already done. Should you wait for more skilled hands and their equipment or transport the patient yourself and get on the road before it becomes a skating rink? There is an antique transport isolette gathering dust in the storage room down the hall, and the local fire department ambulance crew with whom you are on a first-name basis is always ready to help. Is it time to gather the troops and tell them, “Let’s roll!” ?

If you have ever lived through a similar scenario, you may find a recent study interesting (Ann Intern Med. 2015;163[9]:681-90). What these investigators found was that for adults who had suffered major trauma, stroke, respiratory failure, and acute myocardial infarction, those who were transported by crews with basic life support (BLS) skills had significantly better long-term survival and neurologic outcomes than did those victims transported by crews with advanced life support (ALS) skills.

In the flurry of comments that circulated following the release of the study were a few questions about the methodology, but most commentators were searching for an explanation. Was critical time lost by the ALS crews doing stuff when the better course of action would have been to get the ambulance rolling to the hospital and more definitive care? Does the temptation to do things because you can do them sometimes cloud the decision-making process?

Although I have lived the scenario I described, it is less likely to happen now. Backup teams from other institutions may be activated. The teams are so well equipped and trained that the gaps between their capabilities and the neonatal intensive care unit have narrowed, but there is no question that they remain and are significant.

The other thing that hasn’t changed is the weather here in Maine. While we have beautiful summers that prompt us to put “Vacationland” on our license plates, our winters are a challenge. In addition to the patient’s condition and the availability of resources, the decision of whether to invest time in stabilization or get moving toward the referral center also must include the risk to the patient and staff who will be traveling on weather-threatened roads.

On the other hand, we can’t ignore the elephant that occasionally finds its way into the room when decisions are made about how thoroughly a critically ill patient is stabilized and how speedily he is transferred. And, that ponderous pachyderm is the hot potato factor and sometimes answers to its acronym, NIMBY (“not in my back yard”). You know as well as I do that despite the Emergency Medical Treatment and Active Labor Act (EMTALA) regulations, there are cases when a patient is hustled out the door without being appropriately stabilized primarily to avoid having that patient die in the referring hospital. We must continue to ask ourselves if we have done everything that we can do to stabilize the patient before we say, “Let’s roll!”

William G. Wilkoff, M.D., practiced primary care pediatrics in Brunswick, Maine, for nearly 40 years. He has authored several books on behavioral pediatrics including “How to Say No to Your Toddler.”

CLINICAL CASE FROM 2009
JF, a 64-year-old man with a 30-year history of type 2 diabetes managed with basal and rapid-acting prandial insulin, started peritoneal dialysis using icodextrin dialysis solution. Since starting dialysis, JF has experienced persistently elevated blood glucose readings (in the high 200 mg/dL to high 300 mg/dL range) using his Accu-Chek Compact glucometer purchased in 2008. In response, JF has been taking higher doses of rapid-acting insulin with meals and for correction, with two-to-three-hour postprandial blood glucose readings persistently elevated (in the high 200s). JF has no fevers, chills, abdominal pain, or other signs/symptoms of infection. Urine ketone testing is negative.

Yesterday, JF’s pre-lunch blood glucose registered at 380 mg/dL on his glucometer, and he took a dose of rapid-acting insulin that was double what he would have taken prior to starting dialysis. About 90 minutes after lunch, JF felt weak and diaphoretic and became unresponsive, with seizure-like activity. His wife called the paramedics; when they arrived, JF’s fingerstick glucose level was 28 mg/dL (using a One Touch Ultra glucometer).

JF was treated acutely with IV dextrose and then transported to a nearby hospital. During his hospitalization, his blood glucose level was maintained in the mid-100 to high-200 mg/dL range, with approximately 50% lower doses of rapid-acting insulin with meals. Hospital work-up revealed no evidence of secondary causes of hyperglycemia. EEG was negative.

Further investigation determined that JF’s Accu-Chek Compact glucometer used GDH-PQQ methodology, which is unable to distinguish between the blood glucose level and the maltose metabolite of icodextrin contained in the peritoneal dialysis solution—leading to falsely elevated glucose results. JF switched to a different glucometer that did not use test strips containing the GDH-PQQ method, allowing for more accurate blood glucose readings and no recurrent episodes of severe hypoglycemia.

Continue for biochemistry of glucose measurements >>

BIOCHEMISTRY OF GLUCOSE MEASUREMENTS
In 1964, Ernie Adams invented Dextrostix, a paper strip that developed varying shades of color proportional to the glucose concentration. In 1970, Anton Clemens developed the first glucometer, the Ames Reflectance Meter (ARM), to detect reflected light from a Dextrostix. The ARM weighed 3 lb and cost $650.¹

Modern glucometers analyze whole blood using both an enzymatic reaction and a detector. The enzyme is packaged in a dehydrated state contained in a disposable strip. The glucose in the patient’s blood rehydrates and reacts with enzymes in the strip to produce a detectable product.¹

The gold standard for measuring glucose is isotope dilution mass spectrometry; however, this is not commonly performed in clinical laboratories. The accuracy of glucometers is most commonly assessed by comparing the glucometer result to a venous plasma sample collected at the same time and analyzed by a clinical laboratory using multi-analyte automated instrumentation.¹

The two main types of commercially available glucometers are the glucose oxidase (GO) and glucose dehydrogenase (GDH) systems. The GO meters utilize the GO enzyme to catalyze the oxidation of glucose into gluconic acid. The oxidation reaction produces electrons that generate current proportional to the glucose level in the test sample.^1-3

With GDH glucometers, several different enzymes can catalyze glucose oxidation, including nicotinamide adenine dinucleotide (GDH-NAD), flavin adenine dinucleotide (GDH-FAD), pyrroloquinoline quinone (GDH-PQQ), or mutant glucose dehydrogenase PQQ (Mut Q-GDH).^2,4,5

Measurement of glucose using the hexokinase enzyme is considered more accurate than both the GO and GDH systems and is commonly used in clinical laboratories. However, the cost of this system is more than that of the commercially available glucometers, and thus it is not widely available.²

Continue for performance requirements for glucometer systems >>

PERFORMANCE REQUIREMENTS FOR GLUCOMETER SYSTEMS
There is no single standard for glucometer accuracy. Per Guideline 15197, issued by the International Organization for Standardization (ISO) in 2013, the minimum criteria for accuracy is at least 95% of blood glucose results within ± 15 mg/dL of the reference value at blood sugar concentrations < 100 mg/dL and within ± 15% at blood sugar concentrations ≥ 100 mg/dL.6 For OTC glucometers, the FDA has recommended that at least 95% of measurements fall within ± 15% and at least 99% of measurements fall within ± 20% of reference values across the entire claimed range of the glucometer system.⁷

The ISO and FDA both recommend that industry test glucometer accuracy using glucose levels ranging from ≤ 50 mg/dL to ≥ 400 mg/dL.^6,7 They also recommend evaluating blood glucose accuracy at different hematocrit levels and assessing accuracy in the presence of interfering substances, such as acetaminophen, ibuprofen, salicylate, sodium, ascorbic acid, bilirubin, creatinine, dopamine, maltose, xylose, galactose, hemoglobin, heparin, L-dopa, methyldopa, triglycerides, cholesterol, sugar alcohols, and uric acid.^6,7 The FDA additionally recommends testing glucometer accuracy in the presence of temperature extremes, humidity, and different altitudes.⁷

Currently, the premarket evaluation of glucometers is a one-time procedure that is typically conducted by the manufacturer. Not all available glucometers currently comply with the less stringent ISO accuracy standards from 2003, and most currently available glucometer systems fail to meet the more stringent accuracy criteria outlined by the ISO in 2013 and the FDA in 2014. Furthermore, there can be inconsistency in the measurement quality between different test strip lots, adding another variable to assessing glucometer accuracy.⁶

Continue for variables affecting glucometer accuracy >>

VARIABLES AFFECTING GLUCOMETER ACCURACY
Patient and environmental factors
Both patient and environmental factors can interfere with obtaining accurate glucometer results. These include sampling errors, improper storage of test strips, inadequate amount of blood applied to the test strip, improper meter coding, and altitude.¹

Temperature extremes and humidity can denature, inactivate, or prematurely rehydrate enzymes and proteins within the test strip.¹ GO meters can overestimate glucose levels at low temperatures, while GDH meters can produce unpredictable results in increased humidity.¹ The detector portion of the meter is composed of electronics and should be protected from temperature extremes and excessive moisture as well.¹

In high altitude, both GO and GDH meters can produce unreliable results, with a tendency to overestimate blood glucose levels.⁸ Another variable confounding the accuracy of glucometer readings at high altitude is the potential for secondary polycythemia, which can result in underestimation of glucose levels.^8,9

Physiologic factors
Physiologic factors that can cause inaccurate glucometer results include hypoxia, abnormal pH, hyperuricemia, jaundice, polycythemia, anemia, peripheral vascular disease, and hypotension resulting in poor perfusion.^1,7,9

Elevated oxygen tension in patients receiving oxygen therapy can falsely lower glucometer results for GO meters, while hypoxia can falsely elevate glucose results for these meters.^1,3

Low pH (< 6.95), such as in diabetic ketoacidosis, falsely lowers glucose readings in GO meters, while a high pH falsely elevates glucose readings.^1,10 Elevated serum uric acid (> 10-16 mg/dL) and elevated total bilirubin concentration (> 20 mg/dL) can cause overestimation of blood glucose levels due to electrochemical interaction at the electrode site in GDH-PQQ meters.¹¹

Polycythemia can result in underestimation of glucose levels, and glucose levels can be overestimated in the setting of anemia.⁹ In anemia, the reduced red blood cell volume results in less displacement of plasma, causing more glucose molecules to be available to react with the enzyme contained in the test strip.¹²

Despite manufacturers’ claims that glucometers are reliable to a hematocrit range of 20% to 25%, clinically significant errors of greater than 20% were observed when the hematocrit level dropped below 34%, which can present challenges if glucometers are used in the ICU.¹³ Mathematical formulas to correct point-of-care glucometer measurements based on the hematocrit level have been proposed and have demonstrated effectiveness in decreasing the incidence of hypoglycemia in critically ill patients treated with insulin.¹²

Medications
Drugs that most commonly interfere with glucometer measurements include acetaminophen (especially at a serum concentration > 8 mg/dL), ascorbic acid, maltose, galactose, and xylose.^1,11 Acetaminophen and ascorbic acid consume peroxide, resulting in falsely lowered blood glucose readings in GO meters. In GDH meters, direct oxidation can occur at the electrode site in the presence of acetaminophen and ascorbic acid, resulting in falsely elevated glucose levels.^6,9,12

Maltose, galactose, and xylose are nonglucose sugars found in certain drug and biologic formulations, such as icodextrin peritoneal dialysis solution, certain immunoglobulins (Octagam 5%, WinRho SDF Liquid, Vaccinia Immune Globulin Intravenous [Human], and HepGamB), Orencia, and BEXXAR radioimmunotherapy agent.¹⁴

The GDH-PQQ meters cannot distinguish between glucose and nonglucose sugars, resulting in either undetected hypoglycemia or a falsely elevated glucose result (up to 3 to 15 times higher than corresponding laboratory results), which can lead to inappropriate medication dosing that results in potential hypoglycemia, coma, or death.¹⁴ Laboratory-based blood glucose assays, the GO, and most GDH-FAD, GDH-NAD, Mut Q-GDH, and hexokinase test strips do not have the potential for cross-reactivity from sugars other than glucose.^4,14

It should be noted that in the United States, most GDH-PQQ test strips are no longer manufactured for home glucose testing. However, it is important to review the product insert contained in the test strip box for verification of the specific enzymatic methodology used in the test strip.^4,5

Continue for the conclusion >>

CONCLUSION
Multiple factors affect the accuracy of currently available glucometers. Consideration of patient comorbidities, medication use, operational technique, and the conditions under which test strips are stored is important when utilizing glucometer data to make medication adjustments in diabetes management. It is important to refer to specific glucometer and test strip manufacturer device labeling to help select the appropriate glucometer for a particular patient.

The case presentation from 2009, involving falsely elevated blood glucose readings in a patient using a GDH-PQQ meter while receiving icodextrin peritoneal dialysis solution, highlights the importance of background knowledge of glucometer operational mechanisms. For a full list of test strips that are compatible with icodextrin peritoneal dialysis solution, please see the Country-Specific Glucose Monitor List at www.glucosesafety.com.⁵

Examples of specific GO meters include the OneTouch Ultra, iBGStar, and ReliOn meters. Although the GO meters do not cross-react with icodextrin, these meters should be avoided in patients receiving supplemental oxygen, due to the potential for falsely lowered readings.

The GDH-FAD, GDH-NAD, and Mut Q-GDH test strips may be used in patients receiving icodextrin peritoneal dialysis solution and those receiving supplemental oxygen.^3,5 Examples of GDH-FAD meters include most currently available FreeStyle meters, Bayer Contour meters, and One Touch Verio meters. The Precision Xtra meter uses GDH-NAD test strips. Most Accu-Chek meters currently use Mut Q-GDH test strips.

REFERENCES
1. Tonyushkina K, Nichols JH. Glucose meters: a review of technical challenges to obtaining accurate results. J Diabetes Sci Technol. 2009;3(4):971-980.
2. Floré KMJ, Delanghe JR. Analytical interferences in point-of-care testing glucometers by icodextrin and its metabolites: an overview. Peritoneal Dial Int. 2009;29(4):377-383.
3. Tang Z, Louie RF, Lee JH, et al. Oxygen effects on glucose meter measurements with glucose dehydrogenase- and oxidase-based test strips for point-of-care testing. Crit Care Med. 2001;29(5):1062-1070.
4. Olansky L. Finger-stick glucose monitoring: issues of accuracy and specificity. Diabetes Care. 2010;33(4):948-949.
5. Baxter Healthcare Corporation. Country-specific glucose monitor list, 2015. www.glucosesafety.com/us/pdf/Glucose_Monitor_List.pdf. Accessed November 18, 2015.
6. Freckmann G, Schmid C, Baumstark A, et al. Analytical performance requirements for systems for self-monitoring of blood glucose with focus on system accuracy: relevant differences among ISO 15197:2003, ISO 15197: 2013, and current FDA recommendations. J Diabetes Sci Technol. 2015;9(4):885-894.
7. FDA. Self-Monitoring Blood Glucose Test Systems for Over-The-Counter Use: Draft Guidance for Industry and Food and Drug Administration Staff (2014). www.fda.gov/downloads/medicaldevices/deviceregulationandguidance/guidancedocuments/ucm380327.pdf. Accessed November 18, 2015.
8. Olateju T, Begley J, Flanagan D, Kerr D. Effects of simulated altitude on blood glucose meter performance: implications for in-flight blood glucose monitoring. J Diabetes Sci Technol. 2012;6(4):867-874.
9. Rao LV, Jakubiak F, Sidwell JS, et al. Accuracy evaluation of a new glucometer with automated hematocrit measurement and correction. Clinica Chimica Acta. 2005;356(1-2):178-183.
10. Tang Z, Du X, Louie RF, Kost GJ. Effects of pH on glucose measurements with handheld glucose meters and a portable glucose analyzer for point-of-care testing. Arch Pathol Lab Med. 2000;124:577-582.
11. Eastham JH, Mason D, Barnes DL, Kollins J. Prevalence of interfering substances with point-of-care glucose testing in a community hospital. Am J Health Syst Pharm. 2009;66: 167-170.
12 Pidcoke HF, Wade CE, Mann EA, et al. Anemia causes hypoglycemia in ICU patients due to error in single-channel glucometers: methods of reducing patient risk. Crit Care Med. 2010;38(2):471-476.
13. Mann EA, Pidcoke HF, Salinas J, et al. Accuracy of glucometers should not be assumed. Am J Crit Care. 2007;16(6):531-532.
14. FDA. FDA Public Health Notification: Potentially Fatal Errors with GDH-PQQ Glucose Monitoring Technology (2009). www.fda.gov/MedicalDevices/Safety/AlertsandNotices/PublicHealthNotifications/ucm176992.htm. Accessed November 18, 2015.

CLINICAL CASE FROM 2009
JF, a 64-year-old man with a 30-year history of type 2 diabetes managed with basal and rapid-acting prandial insulin, started peritoneal dialysis using icodextrin dialysis solution. Since starting dialysis, JF has experienced persistently elevated blood glucose readings (in the high 200 mg/dL to high 300 mg/dL range) using his Accu-Chek Compact glucometer purchased in 2008. In response, JF has been taking higher doses of rapid-acting insulin with meals and for correction, with two-to-three-hour postprandial blood glucose readings persistently elevated (in the high 200s). JF has no fevers, chills, abdominal pain, or other signs/symptoms of infection. Urine ketone testing is negative.

Yesterday, JF’s pre-lunch blood glucose registered at 380 mg/dL on his glucometer, and he took a dose of rapid-acting insulin that was double what he would have taken prior to starting dialysis. About 90 minutes after lunch, JF felt weak and diaphoretic and became unresponsive, with seizure-like activity. His wife called the paramedics; when they arrived, JF’s fingerstick glucose level was 28 mg/dL (using a One Touch Ultra glucometer).

JF was treated acutely with IV dextrose and then transported to a nearby hospital. During his hospitalization, his blood glucose level was maintained in the mid-100 to high-200 mg/dL range, with approximately 50% lower doses of rapid-acting insulin with meals. Hospital work-up revealed no evidence of secondary causes of hyperglycemia. EEG was negative.

Further investigation determined that JF’s Accu-Chek Compact glucometer used GDH-PQQ methodology, which is unable to distinguish between the blood glucose level and the maltose metabolite of icodextrin contained in the peritoneal dialysis solution—leading to falsely elevated glucose results. JF switched to a different glucometer that did not use test strips containing the GDH-PQQ method, allowing for more accurate blood glucose readings and no recurrent episodes of severe hypoglycemia.

Continue for biochemistry of glucose measurements >>

BIOCHEMISTRY OF GLUCOSE MEASUREMENTS
In 1964, Ernie Adams invented Dextrostix, a paper strip that developed varying shades of color proportional to the glucose concentration. In 1970, Anton Clemens developed the first glucometer, the Ames Reflectance Meter (ARM), to detect reflected light from a Dextrostix. The ARM weighed 3 lb and cost $650.¹

Modern glucometers analyze whole blood using both an enzymatic reaction and a detector. The enzyme is packaged in a dehydrated state contained in a disposable strip. The glucose in the patient’s blood rehydrates and reacts with enzymes in the strip to produce a detectable product.¹

The gold standard for measuring glucose is isotope dilution mass spectrometry; however, this is not commonly performed in clinical laboratories. The accuracy of glucometers is most commonly assessed by comparing the glucometer result to a venous plasma sample collected at the same time and analyzed by a clinical laboratory using multi-analyte automated instrumentation.¹

The two main types of commercially available glucometers are the glucose oxidase (GO) and glucose dehydrogenase (GDH) systems. The GO meters utilize the GO enzyme to catalyze the oxidation of glucose into gluconic acid. The oxidation reaction produces electrons that generate current proportional to the glucose level in the test sample.^1-3

With GDH glucometers, several different enzymes can catalyze glucose oxidation, including nicotinamide adenine dinucleotide (GDH-NAD), flavin adenine dinucleotide (GDH-FAD), pyrroloquinoline quinone (GDH-PQQ), or mutant glucose dehydrogenase PQQ (Mut Q-GDH).^2,4,5

Measurement of glucose using the hexokinase enzyme is considered more accurate than both the GO and GDH systems and is commonly used in clinical laboratories. However, the cost of this system is more than that of the commercially available glucometers, and thus it is not widely available.²

Continue for performance requirements for glucometer systems >>

PERFORMANCE REQUIREMENTS FOR GLUCOMETER SYSTEMS
There is no single standard for glucometer accuracy. Per Guideline 15197, issued by the International Organization for Standardization (ISO) in 2013, the minimum criteria for accuracy is at least 95% of blood glucose results within ± 15 mg/dL of the reference value at blood sugar concentrations < 100 mg/dL and within ± 15% at blood sugar concentrations ≥ 100 mg/dL.6 For OTC glucometers, the FDA has recommended that at least 95% of measurements fall within ± 15% and at least 99% of measurements fall within ± 20% of reference values across the entire claimed range of the glucometer system.⁷

The ISO and FDA both recommend that industry test glucometer accuracy using glucose levels ranging from ≤ 50 mg/dL to ≥ 400 mg/dL.^6,7 They also recommend evaluating blood glucose accuracy at different hematocrit levels and assessing accuracy in the presence of interfering substances, such as acetaminophen, ibuprofen, salicylate, sodium, ascorbic acid, bilirubin, creatinine, dopamine, maltose, xylose, galactose, hemoglobin, heparin, L-dopa, methyldopa, triglycerides, cholesterol, sugar alcohols, and uric acid.^6,7 The FDA additionally recommends testing glucometer accuracy in the presence of temperature extremes, humidity, and different altitudes.⁷

Currently, the premarket evaluation of glucometers is a one-time procedure that is typically conducted by the manufacturer. Not all available glucometers currently comply with the less stringent ISO accuracy standards from 2003, and most currently available glucometer systems fail to meet the more stringent accuracy criteria outlined by the ISO in 2013 and the FDA in 2014. Furthermore, there can be inconsistency in the measurement quality between different test strip lots, adding another variable to assessing glucometer accuracy.⁶

Continue for variables affecting glucometer accuracy >>

VARIABLES AFFECTING GLUCOMETER ACCURACY
Patient and environmental factors
Both patient and environmental factors can interfere with obtaining accurate glucometer results. These include sampling errors, improper storage of test strips, inadequate amount of blood applied to the test strip, improper meter coding, and altitude.¹

Temperature extremes and humidity can denature, inactivate, or prematurely rehydrate enzymes and proteins within the test strip.¹ GO meters can overestimate glucose levels at low temperatures, while GDH meters can produce unpredictable results in increased humidity.¹ The detector portion of the meter is composed of electronics and should be protected from temperature extremes and excessive moisture as well.¹

In high altitude, both GO and GDH meters can produce unreliable results, with a tendency to overestimate blood glucose levels.⁸ Another variable confounding the accuracy of glucometer readings at high altitude is the potential for secondary polycythemia, which can result in underestimation of glucose levels.^8,9

Physiologic factors
Physiologic factors that can cause inaccurate glucometer results include hypoxia, abnormal pH, hyperuricemia, jaundice, polycythemia, anemia, peripheral vascular disease, and hypotension resulting in poor perfusion.^1,7,9

Elevated oxygen tension in patients receiving oxygen therapy can falsely lower glucometer results for GO meters, while hypoxia can falsely elevate glucose results for these meters.^1,3

Low pH (< 6.95), such as in diabetic ketoacidosis, falsely lowers glucose readings in GO meters, while a high pH falsely elevates glucose readings.^1,10 Elevated serum uric acid (> 10-16 mg/dL) and elevated total bilirubin concentration (> 20 mg/dL) can cause overestimation of blood glucose levels due to electrochemical interaction at the electrode site in GDH-PQQ meters.¹¹

Polycythemia can result in underestimation of glucose levels, and glucose levels can be overestimated in the setting of anemia.⁹ In anemia, the reduced red blood cell volume results in less displacement of plasma, causing more glucose molecules to be available to react with the enzyme contained in the test strip.¹²

Despite manufacturers’ claims that glucometers are reliable to a hematocrit range of 20% to 25%, clinically significant errors of greater than 20% were observed when the hematocrit level dropped below 34%, which can present challenges if glucometers are used in the ICU.¹³ Mathematical formulas to correct point-of-care glucometer measurements based on the hematocrit level have been proposed and have demonstrated effectiveness in decreasing the incidence of hypoglycemia in critically ill patients treated with insulin.¹²

Medications
Drugs that most commonly interfere with glucometer measurements include acetaminophen (especially at a serum concentration > 8 mg/dL), ascorbic acid, maltose, galactose, and xylose.^1,11 Acetaminophen and ascorbic acid consume peroxide, resulting in falsely lowered blood glucose readings in GO meters. In GDH meters, direct oxidation can occur at the electrode site in the presence of acetaminophen and ascorbic acid, resulting in falsely elevated glucose levels.^6,9,12

Maltose, galactose, and xylose are nonglucose sugars found in certain drug and biologic formulations, such as icodextrin peritoneal dialysis solution, certain immunoglobulins (Octagam 5%, WinRho SDF Liquid, Vaccinia Immune Globulin Intravenous [Human], and HepGamB), Orencia, and BEXXAR radioimmunotherapy agent.¹⁴

The GDH-PQQ meters cannot distinguish between glucose and nonglucose sugars, resulting in either undetected hypoglycemia or a falsely elevated glucose result (up to 3 to 15 times higher than corresponding laboratory results), which can lead to inappropriate medication dosing that results in potential hypoglycemia, coma, or death.¹⁴ Laboratory-based blood glucose assays, the GO, and most GDH-FAD, GDH-NAD, Mut Q-GDH, and hexokinase test strips do not have the potential for cross-reactivity from sugars other than glucose.^4,14

It should be noted that in the United States, most GDH-PQQ test strips are no longer manufactured for home glucose testing. However, it is important to review the product insert contained in the test strip box for verification of the specific enzymatic methodology used in the test strip.^4,5

Continue for the conclusion >>

CONCLUSION
Multiple factors affect the accuracy of currently available glucometers. Consideration of patient comorbidities, medication use, operational technique, and the conditions under which test strips are stored is important when utilizing glucometer data to make medication adjustments in diabetes management. It is important to refer to specific glucometer and test strip manufacturer device labeling to help select the appropriate glucometer for a particular patient.

The case presentation from 2009, involving falsely elevated blood glucose readings in a patient using a GDH-PQQ meter while receiving icodextrin peritoneal dialysis solution, highlights the importance of background knowledge of glucometer operational mechanisms. For a full list of test strips that are compatible with icodextrin peritoneal dialysis solution, please see the Country-Specific Glucose Monitor List at www.glucosesafety.com.⁵

Examples of specific GO meters include the OneTouch Ultra, iBGStar, and ReliOn meters. Although the GO meters do not cross-react with icodextrin, these meters should be avoided in patients receiving supplemental oxygen, due to the potential for falsely lowered readings.

The GDH-FAD, GDH-NAD, and Mut Q-GDH test strips may be used in patients receiving icodextrin peritoneal dialysis solution and those receiving supplemental oxygen.^3,5 Examples of GDH-FAD meters include most currently available FreeStyle meters, Bayer Contour meters, and One Touch Verio meters. The Precision Xtra meter uses GDH-NAD test strips. Most Accu-Chek meters currently use Mut Q-GDH test strips.

REFERENCES
1. Tonyushkina K, Nichols JH. Glucose meters: a review of technical challenges to obtaining accurate results. J Diabetes Sci Technol. 2009;3(4):971-980.
2. Floré KMJ, Delanghe JR. Analytical interferences in point-of-care testing glucometers by icodextrin and its metabolites: an overview. Peritoneal Dial Int. 2009;29(4):377-383.
3. Tang Z, Louie RF, Lee JH, et al. Oxygen effects on glucose meter measurements with glucose dehydrogenase- and oxidase-based test strips for point-of-care testing. Crit Care Med. 2001;29(5):1062-1070.
4. Olansky L. Finger-stick glucose monitoring: issues of accuracy and specificity. Diabetes Care. 2010;33(4):948-949.
5. Baxter Healthcare Corporation. Country-specific glucose monitor list, 2015. www.glucosesafety.com/us/pdf/Glucose_Monitor_List.pdf. Accessed November 18, 2015.
6. Freckmann G, Schmid C, Baumstark A, et al. Analytical performance requirements for systems for self-monitoring of blood glucose with focus on system accuracy: relevant differences among ISO 15197:2003, ISO 15197: 2013, and current FDA recommendations. J Diabetes Sci Technol. 2015;9(4):885-894.
7. FDA. Self-Monitoring Blood Glucose Test Systems for Over-The-Counter Use: Draft Guidance for Industry and Food and Drug Administration Staff (2014). www.fda.gov/downloads/medicaldevices/deviceregulationandguidance/guidancedocuments/ucm380327.pdf. Accessed November 18, 2015.
8. Olateju T, Begley J, Flanagan D, Kerr D. Effects of simulated altitude on blood glucose meter performance: implications for in-flight blood glucose monitoring. J Diabetes Sci Technol. 2012;6(4):867-874.
9. Rao LV, Jakubiak F, Sidwell JS, et al. Accuracy evaluation of a new glucometer with automated hematocrit measurement and correction. Clinica Chimica Acta. 2005;356(1-2):178-183.
10. Tang Z, Du X, Louie RF, Kost GJ. Effects of pH on glucose measurements with handheld glucose meters and a portable glucose analyzer for point-of-care testing. Arch Pathol Lab Med. 2000;124:577-582.
11. Eastham JH, Mason D, Barnes DL, Kollins J. Prevalence of interfering substances with point-of-care glucose testing in a community hospital. Am J Health Syst Pharm. 2009;66: 167-170.
12 Pidcoke HF, Wade CE, Mann EA, et al. Anemia causes hypoglycemia in ICU patients due to error in single-channel glucometers: methods of reducing patient risk. Crit Care Med. 2010;38(2):471-476.
13. Mann EA, Pidcoke HF, Salinas J, et al. Accuracy of glucometers should not be assumed. Am J Crit Care. 2007;16(6):531-532.
14. FDA. FDA Public Health Notification: Potentially Fatal Errors with GDH-PQQ Glucose Monitoring Technology (2009). www.fda.gov/MedicalDevices/Safety/AlertsandNotices/PublicHealthNotifications/ucm176992.htm. Accessed November 18, 2015.

CLINICAL CASE FROM 2009
JF, a 64-year-old man with a 30-year history of type 2 diabetes managed with basal and rapid-acting prandial insulin, started peritoneal dialysis using icodextrin dialysis solution. Since starting dialysis, JF has experienced persistently elevated blood glucose readings (in the high 200 mg/dL to high 300 mg/dL range) using his Accu-Chek Compact glucometer purchased in 2008. In response, JF has been taking higher doses of rapid-acting insulin with meals and for correction, with two-to-three-hour postprandial blood glucose readings persistently elevated (in the high 200s). JF has no fevers, chills, abdominal pain, or other signs/symptoms of infection. Urine ketone testing is negative.

Yesterday, JF’s pre-lunch blood glucose registered at 380 mg/dL on his glucometer, and he took a dose of rapid-acting insulin that was double what he would have taken prior to starting dialysis. About 90 minutes after lunch, JF felt weak and diaphoretic and became unresponsive, with seizure-like activity. His wife called the paramedics; when they arrived, JF’s fingerstick glucose level was 28 mg/dL (using a One Touch Ultra glucometer).

JF was treated acutely with IV dextrose and then transported to a nearby hospital. During his hospitalization, his blood glucose level was maintained in the mid-100 to high-200 mg/dL range, with approximately 50% lower doses of rapid-acting insulin with meals. Hospital work-up revealed no evidence of secondary causes of hyperglycemia. EEG was negative.

Further investigation determined that JF’s Accu-Chek Compact glucometer used GDH-PQQ methodology, which is unable to distinguish between the blood glucose level and the maltose metabolite of icodextrin contained in the peritoneal dialysis solution—leading to falsely elevated glucose results. JF switched to a different glucometer that did not use test strips containing the GDH-PQQ method, allowing for more accurate blood glucose readings and no recurrent episodes of severe hypoglycemia.

Continue for biochemistry of glucose measurements >>

BIOCHEMISTRY OF GLUCOSE MEASUREMENTS
In 1964, Ernie Adams invented Dextrostix, a paper strip that developed varying shades of color proportional to the glucose concentration. In 1970, Anton Clemens developed the first glucometer, the Ames Reflectance Meter (ARM), to detect reflected light from a Dextrostix. The ARM weighed 3 lb and cost $650.¹

Modern glucometers analyze whole blood using both an enzymatic reaction and a detector. The enzyme is packaged in a dehydrated state contained in a disposable strip. The glucose in the patient’s blood rehydrates and reacts with enzymes in the strip to produce a detectable product.¹

The gold standard for measuring glucose is isotope dilution mass spectrometry; however, this is not commonly performed in clinical laboratories. The accuracy of glucometers is most commonly assessed by comparing the glucometer result to a venous plasma sample collected at the same time and analyzed by a clinical laboratory using multi-analyte automated instrumentation.¹

The two main types of commercially available glucometers are the glucose oxidase (GO) and glucose dehydrogenase (GDH) systems. The GO meters utilize the GO enzyme to catalyze the oxidation of glucose into gluconic acid. The oxidation reaction produces electrons that generate current proportional to the glucose level in the test sample.^1-3

With GDH glucometers, several different enzymes can catalyze glucose oxidation, including nicotinamide adenine dinucleotide (GDH-NAD), flavin adenine dinucleotide (GDH-FAD), pyrroloquinoline quinone (GDH-PQQ), or mutant glucose dehydrogenase PQQ (Mut Q-GDH).^2,4,5

Measurement of glucose using the hexokinase enzyme is considered more accurate than both the GO and GDH systems and is commonly used in clinical laboratories. However, the cost of this system is more than that of the commercially available glucometers, and thus it is not widely available.²

Continue for performance requirements for glucometer systems >>

PERFORMANCE REQUIREMENTS FOR GLUCOMETER SYSTEMS
There is no single standard for glucometer accuracy. Per Guideline 15197, issued by the International Organization for Standardization (ISO) in 2013, the minimum criteria for accuracy is at least 95% of blood glucose results within ± 15 mg/dL of the reference value at blood sugar concentrations < 100 mg/dL and within ± 15% at blood sugar concentrations ≥ 100 mg/dL.6 For OTC glucometers, the FDA has recommended that at least 95% of measurements fall within ± 15% and at least 99% of measurements fall within ± 20% of reference values across the entire claimed range of the glucometer system.⁷

The ISO and FDA both recommend that industry test glucometer accuracy using glucose levels ranging from ≤ 50 mg/dL to ≥ 400 mg/dL.^6,7 They also recommend evaluating blood glucose accuracy at different hematocrit levels and assessing accuracy in the presence of interfering substances, such as acetaminophen, ibuprofen, salicylate, sodium, ascorbic acid, bilirubin, creatinine, dopamine, maltose, xylose, galactose, hemoglobin, heparin, L-dopa, methyldopa, triglycerides, cholesterol, sugar alcohols, and uric acid.^6,7 The FDA additionally recommends testing glucometer accuracy in the presence of temperature extremes, humidity, and different altitudes.⁷

Currently, the premarket evaluation of glucometers is a one-time procedure that is typically conducted by the manufacturer. Not all available glucometers currently comply with the less stringent ISO accuracy standards from 2003, and most currently available glucometer systems fail to meet the more stringent accuracy criteria outlined by the ISO in 2013 and the FDA in 2014. Furthermore, there can be inconsistency in the measurement quality between different test strip lots, adding another variable to assessing glucometer accuracy.⁶

Continue for variables affecting glucometer accuracy >>

VARIABLES AFFECTING GLUCOMETER ACCURACY
Patient and environmental factors
Both patient and environmental factors can interfere with obtaining accurate glucometer results. These include sampling errors, improper storage of test strips, inadequate amount of blood applied to the test strip, improper meter coding, and altitude.¹

Temperature extremes and humidity can denature, inactivate, or prematurely rehydrate enzymes and proteins within the test strip.¹ GO meters can overestimate glucose levels at low temperatures, while GDH meters can produce unpredictable results in increased humidity.¹ The detector portion of the meter is composed of electronics and should be protected from temperature extremes and excessive moisture as well.¹

In high altitude, both GO and GDH meters can produce unreliable results, with a tendency to overestimate blood glucose levels.⁸ Another variable confounding the accuracy of glucometer readings at high altitude is the potential for secondary polycythemia, which can result in underestimation of glucose levels.^8,9

Physiologic factors
Physiologic factors that can cause inaccurate glucometer results include hypoxia, abnormal pH, hyperuricemia, jaundice, polycythemia, anemia, peripheral vascular disease, and hypotension resulting in poor perfusion.^1,7,9

Elevated oxygen tension in patients receiving oxygen therapy can falsely lower glucometer results for GO meters, while hypoxia can falsely elevate glucose results for these meters.^1,3

Low pH (< 6.95), such as in diabetic ketoacidosis, falsely lowers glucose readings in GO meters, while a high pH falsely elevates glucose readings.^1,10 Elevated serum uric acid (> 10-16 mg/dL) and elevated total bilirubin concentration (> 20 mg/dL) can cause overestimation of blood glucose levels due to electrochemical interaction at the electrode site in GDH-PQQ meters.¹¹

Polycythemia can result in underestimation of glucose levels, and glucose levels can be overestimated in the setting of anemia.⁹ In anemia, the reduced red blood cell volume results in less displacement of plasma, causing more glucose molecules to be available to react with the enzyme contained in the test strip.¹²

Despite manufacturers’ claims that glucometers are reliable to a hematocrit range of 20% to 25%, clinically significant errors of greater than 20% were observed when the hematocrit level dropped below 34%, which can present challenges if glucometers are used in the ICU.¹³ Mathematical formulas to correct point-of-care glucometer measurements based on the hematocrit level have been proposed and have demonstrated effectiveness in decreasing the incidence of hypoglycemia in critically ill patients treated with insulin.¹²

Medications
Drugs that most commonly interfere with glucometer measurements include acetaminophen (especially at a serum concentration > 8 mg/dL), ascorbic acid, maltose, galactose, and xylose.^1,11 Acetaminophen and ascorbic acid consume peroxide, resulting in falsely lowered blood glucose readings in GO meters. In GDH meters, direct oxidation can occur at the electrode site in the presence of acetaminophen and ascorbic acid, resulting in falsely elevated glucose levels.^6,9,12

Maltose, galactose, and xylose are nonglucose sugars found in certain drug and biologic formulations, such as icodextrin peritoneal dialysis solution, certain immunoglobulins (Octagam 5%, WinRho SDF Liquid, Vaccinia Immune Globulin Intravenous [Human], and HepGamB), Orencia, and BEXXAR radioimmunotherapy agent.¹⁴

The GDH-PQQ meters cannot distinguish between glucose and nonglucose sugars, resulting in either undetected hypoglycemia or a falsely elevated glucose result (up to 3 to 15 times higher than corresponding laboratory results), which can lead to inappropriate medication dosing that results in potential hypoglycemia, coma, or death.¹⁴ Laboratory-based blood glucose assays, the GO, and most GDH-FAD, GDH-NAD, Mut Q-GDH, and hexokinase test strips do not have the potential for cross-reactivity from sugars other than glucose.^4,14

It should be noted that in the United States, most GDH-PQQ test strips are no longer manufactured for home glucose testing. However, it is important to review the product insert contained in the test strip box for verification of the specific enzymatic methodology used in the test strip.^4,5

Continue for the conclusion >>

CONCLUSION
Multiple factors affect the accuracy of currently available glucometers. Consideration of patient comorbidities, medication use, operational technique, and the conditions under which test strips are stored is important when utilizing glucometer data to make medication adjustments in diabetes management. It is important to refer to specific glucometer and test strip manufacturer device labeling to help select the appropriate glucometer for a particular patient.

The case presentation from 2009, involving falsely elevated blood glucose readings in a patient using a GDH-PQQ meter while receiving icodextrin peritoneal dialysis solution, highlights the importance of background knowledge of glucometer operational mechanisms. For a full list of test strips that are compatible with icodextrin peritoneal dialysis solution, please see the Country-Specific Glucose Monitor List at www.glucosesafety.com.⁵

Examples of specific GO meters include the OneTouch Ultra, iBGStar, and ReliOn meters. Although the GO meters do not cross-react with icodextrin, these meters should be avoided in patients receiving supplemental oxygen, due to the potential for falsely lowered readings.

The GDH-FAD, GDH-NAD, and Mut Q-GDH test strips may be used in patients receiving icodextrin peritoneal dialysis solution and those receiving supplemental oxygen.^3,5 Examples of GDH-FAD meters include most currently available FreeStyle meters, Bayer Contour meters, and One Touch Verio meters. The Precision Xtra meter uses GDH-NAD test strips. Most Accu-Chek meters currently use Mut Q-GDH test strips.

REFERENCES
1. Tonyushkina K, Nichols JH. Glucose meters: a review of technical challenges to obtaining accurate results. J Diabetes Sci Technol. 2009;3(4):971-980.
2. Floré KMJ, Delanghe JR. Analytical interferences in point-of-care testing glucometers by icodextrin and its metabolites: an overview. Peritoneal Dial Int. 2009;29(4):377-383.
3. Tang Z, Louie RF, Lee JH, et al. Oxygen effects on glucose meter measurements with glucose dehydrogenase- and oxidase-based test strips for point-of-care testing. Crit Care Med. 2001;29(5):1062-1070.
4. Olansky L. Finger-stick glucose monitoring: issues of accuracy and specificity. Diabetes Care. 2010;33(4):948-949.
5. Baxter Healthcare Corporation. Country-specific glucose monitor list, 2015. www.glucosesafety.com/us/pdf/Glucose_Monitor_List.pdf. Accessed November 18, 2015.
6. Freckmann G, Schmid C, Baumstark A, et al. Analytical performance requirements for systems for self-monitoring of blood glucose with focus on system accuracy: relevant differences among ISO 15197:2003, ISO 15197: 2013, and current FDA recommendations. J Diabetes Sci Technol. 2015;9(4):885-894.
7. FDA. Self-Monitoring Blood Glucose Test Systems for Over-The-Counter Use: Draft Guidance for Industry and Food and Drug Administration Staff (2014). www.fda.gov/downloads/medicaldevices/deviceregulationandguidance/guidancedocuments/ucm380327.pdf. Accessed November 18, 2015.
8. Olateju T, Begley J, Flanagan D, Kerr D. Effects of simulated altitude on blood glucose meter performance: implications for in-flight blood glucose monitoring. J Diabetes Sci Technol. 2012;6(4):867-874.
9. Rao LV, Jakubiak F, Sidwell JS, et al. Accuracy evaluation of a new glucometer with automated hematocrit measurement and correction. Clinica Chimica Acta. 2005;356(1-2):178-183.
10. Tang Z, Du X, Louie RF, Kost GJ. Effects of pH on glucose measurements with handheld glucose meters and a portable glucose analyzer for point-of-care testing. Arch Pathol Lab Med. 2000;124:577-582.
11. Eastham JH, Mason D, Barnes DL, Kollins J. Prevalence of interfering substances with point-of-care glucose testing in a community hospital. Am J Health Syst Pharm. 2009;66: 167-170.
12 Pidcoke HF, Wade CE, Mann EA, et al. Anemia causes hypoglycemia in ICU patients due to error in single-channel glucometers: methods of reducing patient risk. Crit Care Med. 2010;38(2):471-476.
13. Mann EA, Pidcoke HF, Salinas J, et al. Accuracy of glucometers should not be assumed. Am J Crit Care. 2007;16(6):531-532.
14. FDA. FDA Public Health Notification: Potentially Fatal Errors with GDH-PQQ Glucose Monitoring Technology (2009). www.fda.gov/MedicalDevices/Safety/AlertsandNotices/PublicHealthNotifications/ucm176992.htm. Accessed November 18, 2015.

ANSWER
The correct interpretation is coarse atrial fibrillation with a rapid ventricular response and left-axis deviation.

Coarse atrial fibrillation is evidenced by the irregularly irregular rhythm with a normal QRS duration and flutter/fibrillation waves arising from the atria. Rapid ventricular response is defined as a ventricular response > 100 beats/min (seen in this case). Finally, an R-wave axis between –30° and –90° is indicative of left-axis deviation.

Correcting the patient’s hypokalemia and hypomagnesemia resulted in a return to normal sinus rhythm. At one-year follow-up, he had had no further episodes of atrial fibrillation.

ANSWER
The correct interpretation is coarse atrial fibrillation with a rapid ventricular response and left-axis deviation.

Coarse atrial fibrillation is evidenced by the irregularly irregular rhythm with a normal QRS duration and flutter/fibrillation waves arising from the atria. Rapid ventricular response is defined as a ventricular response > 100 beats/min (seen in this case). Finally, an R-wave axis between –30° and –90° is indicative of left-axis deviation.

Correcting the patient’s hypokalemia and hypomagnesemia resulted in a return to normal sinus rhythm. At one-year follow-up, he had had no further episodes of atrial fibrillation.

ANSWER
The correct interpretation is coarse atrial fibrillation with a rapid ventricular response and left-axis deviation.

Coarse atrial fibrillation is evidenced by the irregularly irregular rhythm with a normal QRS duration and flutter/fibrillation waves arising from the atria. Rapid ventricular response is defined as a ventricular response > 100 beats/min (seen in this case). Finally, an R-wave axis between –30° and –90° is indicative of left-axis deviation.

Correcting the patient’s hypokalemia and hypomagnesemia resulted in a return to normal sinus rhythm. At one-year follow-up, he had had no further episodes of atrial fibrillation.

ANSWER
The radiograph shows a slightly displaced fracture of the distal fourth metacarpal head. No other injuries are present.

The patient’s hand was left in the splint, and orthopedic evaluation was obtained. 

ANSWER
The radiograph shows a slightly displaced fracture of the distal fourth metacarpal head. No other injuries are present.

The patient’s hand was left in the splint, and orthopedic evaluation was obtained. 

ANSWER
The radiograph shows a slightly displaced fracture of the distal fourth metacarpal head. No other injuries are present.

The patient’s hand was left in the splint, and orthopedic evaluation was obtained. 

Clinical question: Should steroids be used as adjunctive therapy for patients with community-acquired pneumonia?

Bottom line: Moderate-quality to high-quality evidence suggests that steroids, when added to antibiotics and usual care, can improve outcomes in the treatment of community-acquired pneumonia (CAP). Benefits include reduced hospital length of stay, decreased time to clinical stability, and lower rates of mechanical ventilation and acute respiratory distress syndrome. Steroids may also play a role in preventing deaths, especially in patients with severe CAP; however, the certainty of this evidence is not as clear. Given varying treatment regimens used in the individual studies, the appropriate steroid formulation, dose, and duration of steroids cannot be elucidated from the current set of data. (LOE = 1a)

Reference: Siemieniuk RAC, Meade MO, Alonso-Coello P, et al. Corticosteroid therapy for patients hospitalized with community-acquired pneumonia. Ann Intern Med. 2015;163(7):519-528.

Study design: Meta-analysis (randomized controlled trials)

Funding source: Self-funded or unfunded

Allocation: Uncertain

Setting: Inpatient (any location)

Synopsis: These authors searched MEDLINE, EMBASE, and the Cochrane Register to find randomized controlled trials that compared the use of steroids with placebo in adults with CAP. Two reviewers independently evaluated studies for eligibility, extracted data, and assessed the included studies for risk of bias. Five of the 13 included studies, whose population made up 70% of the total sample population, had low risk of bias. The treatment groups in the individual studies received different steroid preparations, routes of administration, dosages, and duration of treatment. All groups otherwise received antibiotics and usual care for CAP.

High-quality evidence showed that the use of steroids decreased hospital length of stay by 1 day (3 studies: mean difference: -1.0 day; 95% CI -1.79 to -0.21 days) and decreased time to clinical stability by 1.22 days (5 studies, mean difference: -1.22 days; -2.08 to -0.35 days). Moderate-quality evidence showed that the use of steroids decreased the need for mechanical ventilation (5 studies: relative risk [RR] = 0.45; 0.26-0.79) and the incidence of acute respiratory distress syndrome (4 studies: RR = 0.24; 0.10-0.56).

Finally, data from the 12 trials that assessed all-cause mortality revealed a trend toward decreased risk of death in the steroid group. The difference between the two groups for this end point became statistically significant when only the trials that met the criteria for severe pneumonia were included (6 studies: RR = 0.39; 0.20-0.77). Although steroid use, not surprisingly, increased the risk of significant hyperglycemia (6 studies: RR = 1.49;1.01-2.19), there were no differences detected in the rates of gastrointestinal bleeds, severe neuropsychiatric complications, or rehospitalizations.

Dr. Kulkarni is an assistant professor of hospital medicine at Northwestern University in Chicago.

Clinical question: Should steroids be used as adjunctive therapy for patients with community-acquired pneumonia?

Bottom line: Moderate-quality to high-quality evidence suggests that steroids, when added to antibiotics and usual care, can improve outcomes in the treatment of community-acquired pneumonia (CAP). Benefits include reduced hospital length of stay, decreased time to clinical stability, and lower rates of mechanical ventilation and acute respiratory distress syndrome. Steroids may also play a role in preventing deaths, especially in patients with severe CAP; however, the certainty of this evidence is not as clear. Given varying treatment regimens used in the individual studies, the appropriate steroid formulation, dose, and duration of steroids cannot be elucidated from the current set of data. (LOE = 1a)

Reference: Siemieniuk RAC, Meade MO, Alonso-Coello P, et al. Corticosteroid therapy for patients hospitalized with community-acquired pneumonia. Ann Intern Med. 2015;163(7):519-528.

Study design: Meta-analysis (randomized controlled trials)

Funding source: Self-funded or unfunded

Allocation: Uncertain

Setting: Inpatient (any location)

Synopsis: These authors searched MEDLINE, EMBASE, and the Cochrane Register to find randomized controlled trials that compared the use of steroids with placebo in adults with CAP. Two reviewers independently evaluated studies for eligibility, extracted data, and assessed the included studies for risk of bias. Five of the 13 included studies, whose population made up 70% of the total sample population, had low risk of bias. The treatment groups in the individual studies received different steroid preparations, routes of administration, dosages, and duration of treatment. All groups otherwise received antibiotics and usual care for CAP.

High-quality evidence showed that the use of steroids decreased hospital length of stay by 1 day (3 studies: mean difference: -1.0 day; 95% CI -1.79 to -0.21 days) and decreased time to clinical stability by 1.22 days (5 studies, mean difference: -1.22 days; -2.08 to -0.35 days). Moderate-quality evidence showed that the use of steroids decreased the need for mechanical ventilation (5 studies: relative risk [RR] = 0.45; 0.26-0.79) and the incidence of acute respiratory distress syndrome (4 studies: RR = 0.24; 0.10-0.56).

Finally, data from the 12 trials that assessed all-cause mortality revealed a trend toward decreased risk of death in the steroid group. The difference between the two groups for this end point became statistically significant when only the trials that met the criteria for severe pneumonia were included (6 studies: RR = 0.39; 0.20-0.77). Although steroid use, not surprisingly, increased the risk of significant hyperglycemia (6 studies: RR = 1.49;1.01-2.19), there were no differences detected in the rates of gastrointestinal bleeds, severe neuropsychiatric complications, or rehospitalizations.

Dr. Kulkarni is an assistant professor of hospital medicine at Northwestern University in Chicago.

Clinical question: Should steroids be used as adjunctive therapy for patients with community-acquired pneumonia?

Bottom line: Moderate-quality to high-quality evidence suggests that steroids, when added to antibiotics and usual care, can improve outcomes in the treatment of community-acquired pneumonia (CAP). Benefits include reduced hospital length of stay, decreased time to clinical stability, and lower rates of mechanical ventilation and acute respiratory distress syndrome. Steroids may also play a role in preventing deaths, especially in patients with severe CAP; however, the certainty of this evidence is not as clear. Given varying treatment regimens used in the individual studies, the appropriate steroid formulation, dose, and duration of steroids cannot be elucidated from the current set of data. (LOE = 1a)

Reference: Siemieniuk RAC, Meade MO, Alonso-Coello P, et al. Corticosteroid therapy for patients hospitalized with community-acquired pneumonia. Ann Intern Med. 2015;163(7):519-528.

Study design: Meta-analysis (randomized controlled trials)

Funding source: Self-funded or unfunded

Allocation: Uncertain

Setting: Inpatient (any location)

Synopsis: These authors searched MEDLINE, EMBASE, and the Cochrane Register to find randomized controlled trials that compared the use of steroids with placebo in adults with CAP. Two reviewers independently evaluated studies for eligibility, extracted data, and assessed the included studies for risk of bias. Five of the 13 included studies, whose population made up 70% of the total sample population, had low risk of bias. The treatment groups in the individual studies received different steroid preparations, routes of administration, dosages, and duration of treatment. All groups otherwise received antibiotics and usual care for CAP.

High-quality evidence showed that the use of steroids decreased hospital length of stay by 1 day (3 studies: mean difference: -1.0 day; 95% CI -1.79 to -0.21 days) and decreased time to clinical stability by 1.22 days (5 studies, mean difference: -1.22 days; -2.08 to -0.35 days). Moderate-quality evidence showed that the use of steroids decreased the need for mechanical ventilation (5 studies: relative risk [RR] = 0.45; 0.26-0.79) and the incidence of acute respiratory distress syndrome (4 studies: RR = 0.24; 0.10-0.56).

Finally, data from the 12 trials that assessed all-cause mortality revealed a trend toward decreased risk of death in the steroid group. The difference between the two groups for this end point became statistically significant when only the trials that met the criteria for severe pneumonia were included (6 studies: RR = 0.39; 0.20-0.77). Although steroid use, not surprisingly, increased the risk of significant hyperglycemia (6 studies: RR = 1.49;1.01-2.19), there were no differences detected in the rates of gastrointestinal bleeds, severe neuropsychiatric complications, or rehospitalizations.

Dr. Kulkarni is an assistant professor of hospital medicine at Northwestern University in Chicago.