Cleveland Clinic Journal of Medicine

The physical examination used to be a foundation of clinical practice, but it is under assault. A specialist in inpatient medicine here at Cleveland Clinic decried the inefficiency of time spent by residents performing and documenting the examination. How often, he asked, does the examination actually change the diagnostic workup? Academic provocateurs have done sensitivity and specificity analyses on components of the physical examination and found them to be imperfect. Imperfect, yes, but I would argue not worthless.

One reason for the imperfection is that skills of examination are not always appropriately emphasized during training, and then are not utilized in practice (especially in a 10-minute visit). Several published studies describe the attrition of examination skills. While doing teaching rounds as a visiting professor, I have found that some residents and medical students have difficulty distinguishing a murmur of aortic sclerosis from one of aortic insufficiency or detecting epitrochlear adenopathy.

The structured physical examination still fulfills a legitimate need. When approaching a patient with an ill-defined, potentially multisystem disorder, the examination should provide an initial “staging” of the disease process that contributes to the construction of the differential diagnosis, and thus refines the ordering of specific tests.

Adenopathy, enlarged lacrimal glands, and retinal exudates can all be asymptomatic and, if detected, may affect the differential diagnosis. Yet how many examinations focus attention on these areas? We often teach (and we were taught) the physical examination as a rote skill to be performed in toto. But the examination is not a static procedure. It needs to be tailored to the patient in front of us, in a reiterative, reflective manner. I am more likely to find adenopathy or an abdominal aneurysm if I am specifically looking for it, as opposed to performing a perfunctory examination. I am less likely to be struck by a pronounced second heart sound if I am not considering pulmonary hypertension as a possible explanation for the patient’s dyspnea.

I believe that a reflective physical examination is effective and valuable. Besides, our patients actually expect (and deserve) to be carefully examined.

The physical examination used to be a foundation of clinical practice, but it is under assault. A specialist in inpatient medicine here at Cleveland Clinic decried the inefficiency of time spent by residents performing and documenting the examination. How often, he asked, does the examination actually change the diagnostic workup? Academic provocateurs have done sensitivity and specificity analyses on components of the physical examination and found them to be imperfect. Imperfect, yes, but I would argue not worthless.

One reason for the imperfection is that skills of examination are not always appropriately emphasized during training, and then are not utilized in practice (especially in a 10-minute visit). Several published studies describe the attrition of examination skills. While doing teaching rounds as a visiting professor, I have found that some residents and medical students have difficulty distinguishing a murmur of aortic sclerosis from one of aortic insufficiency or detecting epitrochlear adenopathy.

The structured physical examination still fulfills a legitimate need. When approaching a patient with an ill-defined, potentially multisystem disorder, the examination should provide an initial “staging” of the disease process that contributes to the construction of the differential diagnosis, and thus refines the ordering of specific tests.

Adenopathy, enlarged lacrimal glands, and retinal exudates can all be asymptomatic and, if detected, may affect the differential diagnosis. Yet how many examinations focus attention on these areas? We often teach (and we were taught) the physical examination as a rote skill to be performed in toto. But the examination is not a static procedure. It needs to be tailored to the patient in front of us, in a reiterative, reflective manner. I am more likely to find adenopathy or an abdominal aneurysm if I am specifically looking for it, as opposed to performing a perfunctory examination. I am less likely to be struck by a pronounced second heart sound if I am not considering pulmonary hypertension as a possible explanation for the patient’s dyspnea.

I believe that a reflective physical examination is effective and valuable. Besides, our patients actually expect (and deserve) to be carefully examined.

The physical examination used to be a foundation of clinical practice, but it is under assault. A specialist in inpatient medicine here at Cleveland Clinic decried the inefficiency of time spent by residents performing and documenting the examination. How often, he asked, does the examination actually change the diagnostic workup? Academic provocateurs have done sensitivity and specificity analyses on components of the physical examination and found them to be imperfect. Imperfect, yes, but I would argue not worthless.

One reason for the imperfection is that skills of examination are not always appropriately emphasized during training, and then are not utilized in practice (especially in a 10-minute visit). Several published studies describe the attrition of examination skills. While doing teaching rounds as a visiting professor, I have found that some residents and medical students have difficulty distinguishing a murmur of aortic sclerosis from one of aortic insufficiency or detecting epitrochlear adenopathy.

The structured physical examination still fulfills a legitimate need. When approaching a patient with an ill-defined, potentially multisystem disorder, the examination should provide an initial “staging” of the disease process that contributes to the construction of the differential diagnosis, and thus refines the ordering of specific tests.

Adenopathy, enlarged lacrimal glands, and retinal exudates can all be asymptomatic and, if detected, may affect the differential diagnosis. Yet how many examinations focus attention on these areas? We often teach (and we were taught) the physical examination as a rote skill to be performed in toto. But the examination is not a static procedure. It needs to be tailored to the patient in front of us, in a reiterative, reflective manner. I am more likely to find adenopathy or an abdominal aneurysm if I am specifically looking for it, as opposed to performing a perfunctory examination. I am less likely to be struck by a pronounced second heart sound if I am not considering pulmonary hypertension as a possible explanation for the patient’s dyspnea.

I believe that a reflective physical examination is effective and valuable. Besides, our patients actually expect (and deserve) to be carefully examined.

In detecting and evaluating pleural effusion, technology has not replaced clinical skills. Yet, despite centuries of lore, data are limited on the role of the physical examination and on its accuracy compared with other noninvasive tests such as conventional chest radiography or ultrasonography.

The following is an overview of the value of the clinical history and physical examination in detecting pleural effusion and a brief review of the available information regarding its accuracy compared with other diagnostic methods.

POTENTIAL CAUSES ARE MANY

The pleurae consist of two membranes that protect the lungs, allow them to move, contribute to their shape, and prevent the alveoli at the pleural surface from becoming overdistended. Between the visceral pleura (covering the lung) and the parietal pleura (covering the diaphragm and the chest wall) is the pleural space.

In healthy adults, the pleural space contains an estimated 5 to 10 mL of pleural fluid (0.1 mg/kg body weight).¹ Pleural effusion is an accumulation of an abnormal amount of fluid in the pleural space.

Although the potential causes are many, the most common are congestive heart failure, pneumonia (40% of patients hospitalized with pneumonia have pleural effusion),^2,3 cancer, and pulmonary embolism.⁴

Because many diseases affecting different organs can cause a pleural effusion, we cannot overemphasize the importance of a thorough history and physical examination to uncover clues that will help identify its cause and narrow the diagnostic workup. For example, significant weight loss and cachexia could be due to cancer, and joint, skin, or eye symptoms could be due to a connective tissue disorder.

A thorough review of the patient’s medications is mandatory, since several medications (eg, amiodarone [Cordarone], methotrexate [Rheumatrex, Trexall], and nitrofurantoin [Macrobid]) can be associated with exudative effusions. In addition, the patient’s occupational history must be ascertained, since exposure to asbestos can raise the suspicion of a malignant disease of the pleura such as mesothelioma.

SYMPTOMS ARE NEITHER SENSITIVE NOR SPECIFIC

The symptoms of pleural effusion are neither sensitive nor specific, and many patients have manifestations of the underlying process but not of the effusion itself. The most common symptoms directly related to effusion are cough, dyspnea, and pleuritic chest pain.⁵

Cough. Many patients with a pleural effusion have a dry, nonproductive cough, a consequence of inflammation of the pleurae or compression of the bronchial walls. Although this symptom is rarely helpful in diagnosing a pleural effusion, if accompanied by purulent sputum it suggests pneumonia, and if complicated by hemoptysis it suggests cancer or pulmonary embolism.

Dyspnea is a consequence of a combination of a restrictive lung defect, a ventilation-perfusion mismatch, and a decrease in cardiac output. Although large pleural effusions reduce lung volume and are generally associated with dyspnea, the symptoms may be out of proportion to the size of the effusion, and patients with small to moderate effusions may also have shortness of breath if their baseline lung function is poor.²

Chest pain accompanying a pleural effusion suggests inflammation of the parietal pleura,⁶ but could be due to cancer in the chest wall and ribs—or to a benign disease of the thoracic wall such as rib fracture or costo-chondritis.

Pain of pleural origin can remain localized to the adjacent area of the chest, but sometimes it is referred to other areas. If the diaphragmatic pleura is involved, the pain is in many cases referred to the ipsilateral shoulder.⁵ Pain may also be referred to the abdomen.

Pleuritic chest pain is described as being worse with deep inspiration or when lying down. It is common in patients with pulmonary embolism, parapneumonic effusion, or viral pleurisy, but it can also occur in patients with pneumothorax or pericarditis. A dull, aching chest pain may be due to an underlying pleural malignancy.⁷

PHYSICAL EXAMINATION: LONG TRADITION, FEW DATA

Our knowledge of the role of physical examination in detecting pleural effusion is still based mostly on expert opinion and on small case series.^8,9

Patterson et al¹¹ prospectively compared physical examination (including auscultation, percussion, and tactile fremitus) with bedside ultrasonography and found that physical examination had a lower sensitivity (53% vs 80%, respectively) but a similar specificity (71%).

The physical findings are related to the volume of fluid in the pleural effusion and its effects on the chest wall, diaphragm, and lungs. Physical findings are generally normal if less than 300 mL of fluid is present, whereas large effusions (> 1,500 mL) can be associated with significant asymmetry of chest expansion and bulging of intercostal spaces.

Inspection

Although inspection of the chest is not very helpful in detecting a pleural effusion, it can provide other relevant information such as the respiratory rate and the breathing position adopted by the patient (patients with a large pleural effusion may have orthopnea); it can also reveal abnormalities in the shape of the thorax such as the increased anteroposterior diameter (“barrel shape”) seen in patients with chronic obstructive pulmonary disease.¹⁸

In addition, by inspection we can assess the expansion of the thorax. The utility of inspecting chest expansion to detect lung restriction was first noted by Laennec¹⁹ in 1821. A simple method of evaluating chest expansion is to place a measuring tape around the chest at the level of the nipples to compare the circumference at end-inspiration and at end-expiration.²⁰ In the absence of emphysema, the difference should be at least 2 inches. An expansion of 1.5 inches or less is considered abnormal.²¹ More relevant to pleural effusion than the amount of overall chest expansion is whether the expansion is symmetrical, which we can assess by palpation.

Palpation

Signs of pleural effusion that can be detected by palpation include asymmetric chest expansion and asymmetric tactile fremitus.

Other signs. Palpation of the chest can also help in detecting underlying disease of the thorax sometimes associated with pleurisy or pleural effusions. Chest wall tumors or skin abscesses may be related to underlying empyema, localized tenderness may be associated with rib fractures or costochondritis, and crepitus may be due to subcutaneous emphysema.

The chest can be percussed directly with the tips of the fingers of one hand or indirectly by placing a third finger against the surface to be percussed. There are two main techniques used to detect pleural effusions: comparative percussion and auscultatory percussion.

Since other conditions such as consolidation of the lung and atelectasis can also be associated with dullness to percussion, some authors advocate percussion in the lateral supine position to detect a shift in the dullness that would indicate movement of fluid in the chest.²⁵

The sensitivity of comparative chest percussion and its accuracy related to the size of the effusion are unknown. Kalantri et al¹⁰ found that dullness to percussion had a positive predictive value of only 55% but a negative predictive value of 97%, suggesting that the absence of the sign is very helpful in ruling out an effusion.

According to classic textbook descriptions,²⁶ percussive sounds penetrate a maximum of 6 cm (2 cm of body wall thickness and 4 cm of lung), and at least 500 mL of fluid must be present in order to be able to detect an effusion by physical examination.^2,8 Most of these descriptions are based on original studies done in cadavers more than 100 years ago.

The auscultatory percussion technique was first described by Laennec and used to delineate the size of several organs by placing the stethoscope directly above the structure to be outlined, followed by percussion from the periphery towards the organ of interest. The original technique was subsequently modified for the examination of the chest by Guarino.²⁷

Although auscultatory percussion was used initially to try to detect lung lesions, masses and consolidations, Guarino and Guarino¹² found this technique to be highly effective in detecting pleural effusion. In a prospective blinded study in 118 patients, this method was highly (95%) sensitive and 100% specific in detecting pleural effusion, even in patients with obesity, pneumonia, or other pleural abnormalities. Of note, their findings suggested that auscultatory percussion can detect as little as 50 mL of pleural fluid.

Bohadana et al¹³ compared auscultatory and conventional percussion with chest radiographic findings in 281 patients. They found that auscultatory percussion was 100% sensitive for detecting large pleural effusions.

However, when Bourke et al¹⁴ compared conventional and auscultatory percussion in 21 patients with abnormal radiographs, both methods had low sensitivity (15.4% vs 19.2%) but high specificity (97.3% vs 85.1%, respectively). It is important to mention that in this series only a few patients had a pleural effusion.

McDermott et al¹⁶ compared conventional and auscultatory percussion in detecting pleural effusion in 14 hospitalized patients, using ultrasonography instead of chest radiography as the gold standard for comparison. The findings on auscultatory percussion correlated better with the findings on ultrasonography than did those on conventional percussion. The authors gave no information about sensitivity or specificity.

Kalantri et al¹⁰ found that auscultatory percussion had a sensitivity of 58% and a specificity of 85%.

Auscultation

Originally described by Laennec (who invented the stethoscope),¹⁹ auscultation is perhaps the physical examination technique most used to detect pleural effusion.

Lichtenstein et al¹⁵ performed a study of auscultation in critically ill patients and found it to have a very low sensitivity (42%) but a higher specificity (90%), with an overall diagnostic accuracy of 60%. Of note, compared with chest radiography, auscultatory findings had similar sensitivity but higher accuracy.

Absent or diminished breath sounds strongly suggest an effusion.¹⁹

Egophonism. Laennec also described egophonism as a pathognomonic sign associated with a moderate degree of effusion. The word egophony comes from the Greek “ego,” which means goat; it is used to describe the change in the pronounced sound of E to A. The mechanism responsible for finding this sign in massive pleural effusions is probably upward displacement and compression or consolidation of the lung at the top of the effusion.

However, if the effusion is small, the consolidation will not be large enough to produce this sign. Similarly, other lung conditions associated with large consolidations may produce egophonism without a pleural effusion. Little is known about the predictive value of this sign, and significant interob-server variability needs to be taken into account.²⁸

Pleural rub. Pleural effusions that result from any disease that causes direct inflammation of the pleurae can be associated with a pleural rub. This sound, classically described as rubbing of unoiled leather, is pathognomonic of pleural disease but not of pleural effusion. In fact, a pleural rub will disappear once an effusion develops. Little is known about the accuracy of this finding; in the study by Kalantri et al it had a very low sensitivity (5%) but a very high specificity (99%).¹⁰ The differential diagnosis includes pleuritis, pneumonia, mesothelioma, and tumors that metastasize to the pleura.

In detecting and evaluating pleural effusion, technology has not replaced clinical skills. Yet, despite centuries of lore, data are limited on the role of the physical examination and on its accuracy compared with other noninvasive tests such as conventional chest radiography or ultrasonography.

The following is an overview of the value of the clinical history and physical examination in detecting pleural effusion and a brief review of the available information regarding its accuracy compared with other diagnostic methods.

POTENTIAL CAUSES ARE MANY

The pleurae consist of two membranes that protect the lungs, allow them to move, contribute to their shape, and prevent the alveoli at the pleural surface from becoming overdistended. Between the visceral pleura (covering the lung) and the parietal pleura (covering the diaphragm and the chest wall) is the pleural space.

In healthy adults, the pleural space contains an estimated 5 to 10 mL of pleural fluid (0.1 mg/kg body weight).¹ Pleural effusion is an accumulation of an abnormal amount of fluid in the pleural space.

Although the potential causes are many, the most common are congestive heart failure, pneumonia (40% of patients hospitalized with pneumonia have pleural effusion),^2,3 cancer, and pulmonary embolism.⁴

Because many diseases affecting different organs can cause a pleural effusion, we cannot overemphasize the importance of a thorough history and physical examination to uncover clues that will help identify its cause and narrow the diagnostic workup. For example, significant weight loss and cachexia could be due to cancer, and joint, skin, or eye symptoms could be due to a connective tissue disorder.

A thorough review of the patient’s medications is mandatory, since several medications (eg, amiodarone [Cordarone], methotrexate [Rheumatrex, Trexall], and nitrofurantoin [Macrobid]) can be associated with exudative effusions. In addition, the patient’s occupational history must be ascertained, since exposure to asbestos can raise the suspicion of a malignant disease of the pleura such as mesothelioma.

SYMPTOMS ARE NEITHER SENSITIVE NOR SPECIFIC

The symptoms of pleural effusion are neither sensitive nor specific, and many patients have manifestations of the underlying process but not of the effusion itself. The most common symptoms directly related to effusion are cough, dyspnea, and pleuritic chest pain.⁵

Cough. Many patients with a pleural effusion have a dry, nonproductive cough, a consequence of inflammation of the pleurae or compression of the bronchial walls. Although this symptom is rarely helpful in diagnosing a pleural effusion, if accompanied by purulent sputum it suggests pneumonia, and if complicated by hemoptysis it suggests cancer or pulmonary embolism.

Dyspnea is a consequence of a combination of a restrictive lung defect, a ventilation-perfusion mismatch, and a decrease in cardiac output. Although large pleural effusions reduce lung volume and are generally associated with dyspnea, the symptoms may be out of proportion to the size of the effusion, and patients with small to moderate effusions may also have shortness of breath if their baseline lung function is poor.²

Chest pain accompanying a pleural effusion suggests inflammation of the parietal pleura,⁶ but could be due to cancer in the chest wall and ribs—or to a benign disease of the thoracic wall such as rib fracture or costo-chondritis.

Pain of pleural origin can remain localized to the adjacent area of the chest, but sometimes it is referred to other areas. If the diaphragmatic pleura is involved, the pain is in many cases referred to the ipsilateral shoulder.⁵ Pain may also be referred to the abdomen.

Pleuritic chest pain is described as being worse with deep inspiration or when lying down. It is common in patients with pulmonary embolism, parapneumonic effusion, or viral pleurisy, but it can also occur in patients with pneumothorax or pericarditis. A dull, aching chest pain may be due to an underlying pleural malignancy.⁷

PHYSICAL EXAMINATION: LONG TRADITION, FEW DATA

Our knowledge of the role of physical examination in detecting pleural effusion is still based mostly on expert opinion and on small case series.^8,9

Patterson et al¹¹ prospectively compared physical examination (including auscultation, percussion, and tactile fremitus) with bedside ultrasonography and found that physical examination had a lower sensitivity (53% vs 80%, respectively) but a similar specificity (71%).

The physical findings are related to the volume of fluid in the pleural effusion and its effects on the chest wall, diaphragm, and lungs. Physical findings are generally normal if less than 300 mL of fluid is present, whereas large effusions (> 1,500 mL) can be associated with significant asymmetry of chest expansion and bulging of intercostal spaces.

Inspection

Although inspection of the chest is not very helpful in detecting a pleural effusion, it can provide other relevant information such as the respiratory rate and the breathing position adopted by the patient (patients with a large pleural effusion may have orthopnea); it can also reveal abnormalities in the shape of the thorax such as the increased anteroposterior diameter (“barrel shape”) seen in patients with chronic obstructive pulmonary disease.¹⁸

In addition, by inspection we can assess the expansion of the thorax. The utility of inspecting chest expansion to detect lung restriction was first noted by Laennec¹⁹ in 1821. A simple method of evaluating chest expansion is to place a measuring tape around the chest at the level of the nipples to compare the circumference at end-inspiration and at end-expiration.²⁰ In the absence of emphysema, the difference should be at least 2 inches. An expansion of 1.5 inches or less is considered abnormal.²¹ More relevant to pleural effusion than the amount of overall chest expansion is whether the expansion is symmetrical, which we can assess by palpation.

Palpation

Signs of pleural effusion that can be detected by palpation include asymmetric chest expansion and asymmetric tactile fremitus.

Other signs. Palpation of the chest can also help in detecting underlying disease of the thorax sometimes associated with pleurisy or pleural effusions. Chest wall tumors or skin abscesses may be related to underlying empyema, localized tenderness may be associated with rib fractures or costochondritis, and crepitus may be due to subcutaneous emphysema.

The chest can be percussed directly with the tips of the fingers of one hand or indirectly by placing a third finger against the surface to be percussed. There are two main techniques used to detect pleural effusions: comparative percussion and auscultatory percussion.

Since other conditions such as consolidation of the lung and atelectasis can also be associated with dullness to percussion, some authors advocate percussion in the lateral supine position to detect a shift in the dullness that would indicate movement of fluid in the chest.²⁵

The sensitivity of comparative chest percussion and its accuracy related to the size of the effusion are unknown. Kalantri et al¹⁰ found that dullness to percussion had a positive predictive value of only 55% but a negative predictive value of 97%, suggesting that the absence of the sign is very helpful in ruling out an effusion.

According to classic textbook descriptions,²⁶ percussive sounds penetrate a maximum of 6 cm (2 cm of body wall thickness and 4 cm of lung), and at least 500 mL of fluid must be present in order to be able to detect an effusion by physical examination.^2,8 Most of these descriptions are based on original studies done in cadavers more than 100 years ago.

The auscultatory percussion technique was first described by Laennec and used to delineate the size of several organs by placing the stethoscope directly above the structure to be outlined, followed by percussion from the periphery towards the organ of interest. The original technique was subsequently modified for the examination of the chest by Guarino.²⁷

Although auscultatory percussion was used initially to try to detect lung lesions, masses and consolidations, Guarino and Guarino¹² found this technique to be highly effective in detecting pleural effusion. In a prospective blinded study in 118 patients, this method was highly (95%) sensitive and 100% specific in detecting pleural effusion, even in patients with obesity, pneumonia, or other pleural abnormalities. Of note, their findings suggested that auscultatory percussion can detect as little as 50 mL of pleural fluid.

Bohadana et al¹³ compared auscultatory and conventional percussion with chest radiographic findings in 281 patients. They found that auscultatory percussion was 100% sensitive for detecting large pleural effusions.

However, when Bourke et al¹⁴ compared conventional and auscultatory percussion in 21 patients with abnormal radiographs, both methods had low sensitivity (15.4% vs 19.2%) but high specificity (97.3% vs 85.1%, respectively). It is important to mention that in this series only a few patients had a pleural effusion.

McDermott et al¹⁶ compared conventional and auscultatory percussion in detecting pleural effusion in 14 hospitalized patients, using ultrasonography instead of chest radiography as the gold standard for comparison. The findings on auscultatory percussion correlated better with the findings on ultrasonography than did those on conventional percussion. The authors gave no information about sensitivity or specificity.

Kalantri et al¹⁰ found that auscultatory percussion had a sensitivity of 58% and a specificity of 85%.

Auscultation

Originally described by Laennec (who invented the stethoscope),¹⁹ auscultation is perhaps the physical examination technique most used to detect pleural effusion.

Lichtenstein et al¹⁵ performed a study of auscultation in critically ill patients and found it to have a very low sensitivity (42%) but a higher specificity (90%), with an overall diagnostic accuracy of 60%. Of note, compared with chest radiography, auscultatory findings had similar sensitivity but higher accuracy.

Absent or diminished breath sounds strongly suggest an effusion.¹⁹

Egophonism. Laennec also described egophonism as a pathognomonic sign associated with a moderate degree of effusion. The word egophony comes from the Greek “ego,” which means goat; it is used to describe the change in the pronounced sound of E to A. The mechanism responsible for finding this sign in massive pleural effusions is probably upward displacement and compression or consolidation of the lung at the top of the effusion.

However, if the effusion is small, the consolidation will not be large enough to produce this sign. Similarly, other lung conditions associated with large consolidations may produce egophonism without a pleural effusion. Little is known about the predictive value of this sign, and significant interob-server variability needs to be taken into account.²⁸

Pleural rub. Pleural effusions that result from any disease that causes direct inflammation of the pleurae can be associated with a pleural rub. This sound, classically described as rubbing of unoiled leather, is pathognomonic of pleural disease but not of pleural effusion. In fact, a pleural rub will disappear once an effusion develops. Little is known about the accuracy of this finding; in the study by Kalantri et al it had a very low sensitivity (5%) but a very high specificity (99%).¹⁰ The differential diagnosis includes pleuritis, pneumonia, mesothelioma, and tumors that metastasize to the pleura.

In detecting and evaluating pleural effusion, technology has not replaced clinical skills. Yet, despite centuries of lore, data are limited on the role of the physical examination and on its accuracy compared with other noninvasive tests such as conventional chest radiography or ultrasonography.

The following is an overview of the value of the clinical history and physical examination in detecting pleural effusion and a brief review of the available information regarding its accuracy compared with other diagnostic methods.

POTENTIAL CAUSES ARE MANY

The pleurae consist of two membranes that protect the lungs, allow them to move, contribute to their shape, and prevent the alveoli at the pleural surface from becoming overdistended. Between the visceral pleura (covering the lung) and the parietal pleura (covering the diaphragm and the chest wall) is the pleural space.

In healthy adults, the pleural space contains an estimated 5 to 10 mL of pleural fluid (0.1 mg/kg body weight).¹ Pleural effusion is an accumulation of an abnormal amount of fluid in the pleural space.

Although the potential causes are many, the most common are congestive heart failure, pneumonia (40% of patients hospitalized with pneumonia have pleural effusion),^2,3 cancer, and pulmonary embolism.⁴

Because many diseases affecting different organs can cause a pleural effusion, we cannot overemphasize the importance of a thorough history and physical examination to uncover clues that will help identify its cause and narrow the diagnostic workup. For example, significant weight loss and cachexia could be due to cancer, and joint, skin, or eye symptoms could be due to a connective tissue disorder.

A thorough review of the patient’s medications is mandatory, since several medications (eg, amiodarone [Cordarone], methotrexate [Rheumatrex, Trexall], and nitrofurantoin [Macrobid]) can be associated with exudative effusions. In addition, the patient’s occupational history must be ascertained, since exposure to asbestos can raise the suspicion of a malignant disease of the pleura such as mesothelioma.

SYMPTOMS ARE NEITHER SENSITIVE NOR SPECIFIC

The symptoms of pleural effusion are neither sensitive nor specific, and many patients have manifestations of the underlying process but not of the effusion itself. The most common symptoms directly related to effusion are cough, dyspnea, and pleuritic chest pain.⁵

Cough. Many patients with a pleural effusion have a dry, nonproductive cough, a consequence of inflammation of the pleurae or compression of the bronchial walls. Although this symptom is rarely helpful in diagnosing a pleural effusion, if accompanied by purulent sputum it suggests pneumonia, and if complicated by hemoptysis it suggests cancer or pulmonary embolism.

Dyspnea is a consequence of a combination of a restrictive lung defect, a ventilation-perfusion mismatch, and a decrease in cardiac output. Although large pleural effusions reduce lung volume and are generally associated with dyspnea, the symptoms may be out of proportion to the size of the effusion, and patients with small to moderate effusions may also have shortness of breath if their baseline lung function is poor.²

Chest pain accompanying a pleural effusion suggests inflammation of the parietal pleura,⁶ but could be due to cancer in the chest wall and ribs—or to a benign disease of the thoracic wall such as rib fracture or costo-chondritis.

Pain of pleural origin can remain localized to the adjacent area of the chest, but sometimes it is referred to other areas. If the diaphragmatic pleura is involved, the pain is in many cases referred to the ipsilateral shoulder.⁵ Pain may also be referred to the abdomen.

Pleuritic chest pain is described as being worse with deep inspiration or when lying down. It is common in patients with pulmonary embolism, parapneumonic effusion, or viral pleurisy, but it can also occur in patients with pneumothorax or pericarditis. A dull, aching chest pain may be due to an underlying pleural malignancy.⁷

PHYSICAL EXAMINATION: LONG TRADITION, FEW DATA

Our knowledge of the role of physical examination in detecting pleural effusion is still based mostly on expert opinion and on small case series.^8,9

Patterson et al¹¹ prospectively compared physical examination (including auscultation, percussion, and tactile fremitus) with bedside ultrasonography and found that physical examination had a lower sensitivity (53% vs 80%, respectively) but a similar specificity (71%).

The physical findings are related to the volume of fluid in the pleural effusion and its effects on the chest wall, diaphragm, and lungs. Physical findings are generally normal if less than 300 mL of fluid is present, whereas large effusions (> 1,500 mL) can be associated with significant asymmetry of chest expansion and bulging of intercostal spaces.

Inspection

Although inspection of the chest is not very helpful in detecting a pleural effusion, it can provide other relevant information such as the respiratory rate and the breathing position adopted by the patient (patients with a large pleural effusion may have orthopnea); it can also reveal abnormalities in the shape of the thorax such as the increased anteroposterior diameter (“barrel shape”) seen in patients with chronic obstructive pulmonary disease.¹⁸

In addition, by inspection we can assess the expansion of the thorax. The utility of inspecting chest expansion to detect lung restriction was first noted by Laennec¹⁹ in 1821. A simple method of evaluating chest expansion is to place a measuring tape around the chest at the level of the nipples to compare the circumference at end-inspiration and at end-expiration.²⁰ In the absence of emphysema, the difference should be at least 2 inches. An expansion of 1.5 inches or less is considered abnormal.²¹ More relevant to pleural effusion than the amount of overall chest expansion is whether the expansion is symmetrical, which we can assess by palpation.

Palpation

Signs of pleural effusion that can be detected by palpation include asymmetric chest expansion and asymmetric tactile fremitus.

Other signs. Palpation of the chest can also help in detecting underlying disease of the thorax sometimes associated with pleurisy or pleural effusions. Chest wall tumors or skin abscesses may be related to underlying empyema, localized tenderness may be associated with rib fractures or costochondritis, and crepitus may be due to subcutaneous emphysema.

The chest can be percussed directly with the tips of the fingers of one hand or indirectly by placing a third finger against the surface to be percussed. There are two main techniques used to detect pleural effusions: comparative percussion and auscultatory percussion.

Since other conditions such as consolidation of the lung and atelectasis can also be associated with dullness to percussion, some authors advocate percussion in the lateral supine position to detect a shift in the dullness that would indicate movement of fluid in the chest.²⁵

The sensitivity of comparative chest percussion and its accuracy related to the size of the effusion are unknown. Kalantri et al¹⁰ found that dullness to percussion had a positive predictive value of only 55% but a negative predictive value of 97%, suggesting that the absence of the sign is very helpful in ruling out an effusion.

According to classic textbook descriptions,²⁶ percussive sounds penetrate a maximum of 6 cm (2 cm of body wall thickness and 4 cm of lung), and at least 500 mL of fluid must be present in order to be able to detect an effusion by physical examination.^2,8 Most of these descriptions are based on original studies done in cadavers more than 100 years ago.

The auscultatory percussion technique was first described by Laennec and used to delineate the size of several organs by placing the stethoscope directly above the structure to be outlined, followed by percussion from the periphery towards the organ of interest. The original technique was subsequently modified for the examination of the chest by Guarino.²⁷

Although auscultatory percussion was used initially to try to detect lung lesions, masses and consolidations, Guarino and Guarino¹² found this technique to be highly effective in detecting pleural effusion. In a prospective blinded study in 118 patients, this method was highly (95%) sensitive and 100% specific in detecting pleural effusion, even in patients with obesity, pneumonia, or other pleural abnormalities. Of note, their findings suggested that auscultatory percussion can detect as little as 50 mL of pleural fluid.

Bohadana et al¹³ compared auscultatory and conventional percussion with chest radiographic findings in 281 patients. They found that auscultatory percussion was 100% sensitive for detecting large pleural effusions.

However, when Bourke et al¹⁴ compared conventional and auscultatory percussion in 21 patients with abnormal radiographs, both methods had low sensitivity (15.4% vs 19.2%) but high specificity (97.3% vs 85.1%, respectively). It is important to mention that in this series only a few patients had a pleural effusion.

McDermott et al¹⁶ compared conventional and auscultatory percussion in detecting pleural effusion in 14 hospitalized patients, using ultrasonography instead of chest radiography as the gold standard for comparison. The findings on auscultatory percussion correlated better with the findings on ultrasonography than did those on conventional percussion. The authors gave no information about sensitivity or specificity.

Kalantri et al¹⁰ found that auscultatory percussion had a sensitivity of 58% and a specificity of 85%.

Auscultation

Originally described by Laennec (who invented the stethoscope),¹⁹ auscultation is perhaps the physical examination technique most used to detect pleural effusion.

Lichtenstein et al¹⁵ performed a study of auscultation in critically ill patients and found it to have a very low sensitivity (42%) but a higher specificity (90%), with an overall diagnostic accuracy of 60%. Of note, compared with chest radiography, auscultatory findings had similar sensitivity but higher accuracy.

Absent or diminished breath sounds strongly suggest an effusion.¹⁹

Egophonism. Laennec also described egophonism as a pathognomonic sign associated with a moderate degree of effusion. The word egophony comes from the Greek “ego,” which means goat; it is used to describe the change in the pronounced sound of E to A. The mechanism responsible for finding this sign in massive pleural effusions is probably upward displacement and compression or consolidation of the lung at the top of the effusion.

However, if the effusion is small, the consolidation will not be large enough to produce this sign. Similarly, other lung conditions associated with large consolidations may produce egophonism without a pleural effusion. Little is known about the predictive value of this sign, and significant interob-server variability needs to be taken into account.²⁸

Pleural rub. Pleural effusions that result from any disease that causes direct inflammation of the pleurae can be associated with a pleural rub. This sound, classically described as rubbing of unoiled leather, is pathognomonic of pleural disease but not of pleural effusion. In fact, a pleural rub will disappear once an effusion develops. Little is known about the accuracy of this finding; in the study by Kalantri et al it had a very low sensitivity (5%) but a very high specificity (99%).¹⁰ The differential diagnosis includes pleuritis, pneumonia, mesothelioma, and tumors that metastasize to the pleura.

In the February 2008 issue, the article “Infective endocarditis prophylaxis before dental procedures: new guidelines spark controversy” by Dr. Alice Kim and Dr. Thomas Keys (pages 89–92) contained a typographical error. In Table 2, “Antibiotic prophylactic regimens” on page 91, the dose of azithromycin or clarithromycin in adults was incorrect. It should be 500 mg.

In the February 2008 issue, the article “Infective endocarditis prophylaxis before dental procedures: new guidelines spark controversy” by Dr. Alice Kim and Dr. Thomas Keys (pages 89–92) contained a typographical error. In Table 2, “Antibiotic prophylactic regimens” on page 91, the dose of azithromycin or clarithromycin in adults was incorrect. It should be 500 mg.

In the February 2008 issue, the article “Infective endocarditis prophylaxis before dental procedures: new guidelines spark controversy” by Dr. Alice Kim and Dr. Thomas Keys (pages 89–92) contained a typographical error. In Table 2, “Antibiotic prophylactic regimens” on page 91, the dose of azithromycin or clarithromycin in adults was incorrect. It should be 500 mg.

Heart-brain medicine is dedicated to furthering our understanding of the interaction between the body’s neurologic and cardiovascular systems. As discussed previously,¹ the advent of subspecialization in health care delivery has led to significant advances in the care of patients with acute disease or acute exacerbations of chronic disease. While these advances have led to improved outcomes, we were reminded several times this past year how difficult it is to further improve outcomes using the “silo”-based, highly subspecialized approach that has yielded results in the past.

The 2007 Bakken Heart-Brain Summit, held last June in Cleveland, further demonstrated real progress in our understanding of the importance of heart-brain interactions in health and disease. A series of presentations—highlighted by the Bakken Lecture given by Peter Shapiro, MD, an investigator with the SAD­HART trial—reviewed the effect of psychiatric disorders on the incidence of cardiovascular disease and its consequences. These presentations by leaders in the field (many of which are summarized in the pages that follow) offer irrefutable evidence of the following:

• Patients with depression and heart disease have worse outcomes than patients with heart disease with­out depression²
• Patients with depression have decreased vagal tone³
• Patients with coronary artery disease (CAD) can be safely treated with and respond to antidepressants.⁴

These data were complemented by a keynote presentation by Kevin Tracey, MD, whose elegant work over the past many years has demonstrated a link between vagal tone and inflammation.⁵ His most recent data have shown that the vagus has direct input into the inflammatory state of macrophages in the spleen. The effect is mediated via vagal innervation of the spleen and the α7 subunit of the nicotinic receptor expressed on the cell surface of the resident macrophages.^6,7 The relevance of vagally mediated modulation of systemic inflammation has been shown in sepsis and more recently by our group in left ventricular remodeling following acute myocardial infarction.

‘RECONNECTING THE BODY’ TO IMPROVE OUTCOMES

The continuing emergence of the link between psychiatric and neurocontrol of systemic inflammation offers an undeveloped strategy for further improving outcomes in patients with cardiovascular disease. One of our interests in pursuing heart-brain medicine is to reconnect the body and exploit the physiologic interplay between the heart and brain to improve patient outcomes.¹ Given the disappointments over the past year for new therapies like cholesteryl ester transfer protein inhibitors⁸ and vascular cell adhesion molecule (VCAM) inhibitors, strategies that have a singular organ or cellular target focus, now may be the time for exploiting multisystem approaches for modulating disease states such as CAD, congestive heart failure, and arrhythmia.

The potential consequences of these pathways are profound and include the following:

• A physiologic mechanism for the increased incidence of myocardial infarction observed with medications that have anticholinergic properties and potentially decrease autonomic tone
• Worse outcomes in patients with CAD and depression
• An increased incidence of CAD in patients with psychiatric disorders that in themselves may be associated with decreased vagal tone, as well as in patients on long-term drug therapies that alter parasympathetic tone
• Increased incidences of CAD, myocardial infarction, and death in patients with PTSD.

Clearly the underlying science of heart-brain medicine is fascinating and needs to be pursued vigorously. While the science is ongoing, the need to translate what we know to the bedside has never been greater, given the prevalence of CAD, chronic heart failure, and psychiatric and mood disorders, as well as the likelihood of an increasing incidence of PTSD in light of the Iraq war and terrorist threats.

Multiple studies have been performed to position the field for a trial to test whether treating depression leads to improved outcomes in patients with CAD. We know that patients with depression have decreased vagal tone based on decreased heart rate variability; we know that CAD can be safely treated with selective serotonin reuptake inhibitors; and we know that this patient population is more effectively treated with medications. There was a clear sentiment among faculty and attendees of the 2006 Bakken Heart-Brain Summit that the next step in the clinical science of heart disease and neurologic state is in fact a clinical trial to test the efficacy of this approach. Unfortunately, funding for such a trial from the pharmaceutical industry or government agencies is lacking. The Bakken Heart-Brain Institute is working diligently to secure private financing of such a trial from those with personal interests in moving this field forward. We hope to be able to commence such a trial in the near future. We believe the successful initiation of a multicenter trial not only will demonstrate new avenues for improving outcomes in millions of patients but will validate the concept and usher in a new age of cooperative medicine among multiple disciplines.

As we discussed last year,¹ both the need for and the future of heart-brain medicine are great. The advances seen over the past year and those being pursued in basic and clinical science laboratories throughout the world are very exciting. We thank those colleagues who attended the 2007 Bakken Heart-Brain Summit, and we hope you can join us June 4–5, 2008, in Cleveland to continue this exciting pursuit.

Heart-brain medicine is dedicated to furthering our understanding of the interaction between the body’s neurologic and cardiovascular systems. As discussed previously,¹ the advent of subspecialization in health care delivery has led to significant advances in the care of patients with acute disease or acute exacerbations of chronic disease. While these advances have led to improved outcomes, we were reminded several times this past year how difficult it is to further improve outcomes using the “silo”-based, highly subspecialized approach that has yielded results in the past.

The 2007 Bakken Heart-Brain Summit, held last June in Cleveland, further demonstrated real progress in our understanding of the importance of heart-brain interactions in health and disease. A series of presentations—highlighted by the Bakken Lecture given by Peter Shapiro, MD, an investigator with the SAD­HART trial—reviewed the effect of psychiatric disorders on the incidence of cardiovascular disease and its consequences. These presentations by leaders in the field (many of which are summarized in the pages that follow) offer irrefutable evidence of the following:

• Patients with depression and heart disease have worse outcomes than patients with heart disease with­out depression²
• Patients with depression have decreased vagal tone³
• Patients with coronary artery disease (CAD) can be safely treated with and respond to antidepressants.⁴

These data were complemented by a keynote presentation by Kevin Tracey, MD, whose elegant work over the past many years has demonstrated a link between vagal tone and inflammation.⁵ His most recent data have shown that the vagus has direct input into the inflammatory state of macrophages in the spleen. The effect is mediated via vagal innervation of the spleen and the α7 subunit of the nicotinic receptor expressed on the cell surface of the resident macrophages.^6,7 The relevance of vagally mediated modulation of systemic inflammation has been shown in sepsis and more recently by our group in left ventricular remodeling following acute myocardial infarction.

‘RECONNECTING THE BODY’ TO IMPROVE OUTCOMES

The continuing emergence of the link between psychiatric and neurocontrol of systemic inflammation offers an undeveloped strategy for further improving outcomes in patients with cardiovascular disease. One of our interests in pursuing heart-brain medicine is to reconnect the body and exploit the physiologic interplay between the heart and brain to improve patient outcomes.¹ Given the disappointments over the past year for new therapies like cholesteryl ester transfer protein inhibitors⁸ and vascular cell adhesion molecule (VCAM) inhibitors, strategies that have a singular organ or cellular target focus, now may be the time for exploiting multisystem approaches for modulating disease states such as CAD, congestive heart failure, and arrhythmia.

The potential consequences of these pathways are profound and include the following:

• A physiologic mechanism for the increased incidence of myocardial infarction observed with medications that have anticholinergic properties and potentially decrease autonomic tone
• Worse outcomes in patients with CAD and depression
• An increased incidence of CAD in patients with psychiatric disorders that in themselves may be associated with decreased vagal tone, as well as in patients on long-term drug therapies that alter parasympathetic tone
• Increased incidences of CAD, myocardial infarction, and death in patients with PTSD.

Clearly the underlying science of heart-brain medicine is fascinating and needs to be pursued vigorously. While the science is ongoing, the need to translate what we know to the bedside has never been greater, given the prevalence of CAD, chronic heart failure, and psychiatric and mood disorders, as well as the likelihood of an increasing incidence of PTSD in light of the Iraq war and terrorist threats.

Multiple studies have been performed to position the field for a trial to test whether treating depression leads to improved outcomes in patients with CAD. We know that patients with depression have decreased vagal tone based on decreased heart rate variability; we know that CAD can be safely treated with selective serotonin reuptake inhibitors; and we know that this patient population is more effectively treated with medications. There was a clear sentiment among faculty and attendees of the 2006 Bakken Heart-Brain Summit that the next step in the clinical science of heart disease and neurologic state is in fact a clinical trial to test the efficacy of this approach. Unfortunately, funding for such a trial from the pharmaceutical industry or government agencies is lacking. The Bakken Heart-Brain Institute is working diligently to secure private financing of such a trial from those with personal interests in moving this field forward. We hope to be able to commence such a trial in the near future. We believe the successful initiation of a multicenter trial not only will demonstrate new avenues for improving outcomes in millions of patients but will validate the concept and usher in a new age of cooperative medicine among multiple disciplines.

As we discussed last year,¹ both the need for and the future of heart-brain medicine are great. The advances seen over the past year and those being pursued in basic and clinical science laboratories throughout the world are very exciting. We thank those colleagues who attended the 2007 Bakken Heart-Brain Summit, and we hope you can join us June 4–5, 2008, in Cleveland to continue this exciting pursuit.

Heart-brain medicine is dedicated to furthering our understanding of the interaction between the body’s neurologic and cardiovascular systems. As discussed previously,¹ the advent of subspecialization in health care delivery has led to significant advances in the care of patients with acute disease or acute exacerbations of chronic disease. While these advances have led to improved outcomes, we were reminded several times this past year how difficult it is to further improve outcomes using the “silo”-based, highly subspecialized approach that has yielded results in the past.

The 2007 Bakken Heart-Brain Summit, held last June in Cleveland, further demonstrated real progress in our understanding of the importance of heart-brain interactions in health and disease. A series of presentations—highlighted by the Bakken Lecture given by Peter Shapiro, MD, an investigator with the SAD­HART trial—reviewed the effect of psychiatric disorders on the incidence of cardiovascular disease and its consequences. These presentations by leaders in the field (many of which are summarized in the pages that follow) offer irrefutable evidence of the following:

• Patients with depression and heart disease have worse outcomes than patients with heart disease with­out depression²
• Patients with depression have decreased vagal tone³
• Patients with coronary artery disease (CAD) can be safely treated with and respond to antidepressants.⁴

These data were complemented by a keynote presentation by Kevin Tracey, MD, whose elegant work over the past many years has demonstrated a link between vagal tone and inflammation.⁵ His most recent data have shown that the vagus has direct input into the inflammatory state of macrophages in the spleen. The effect is mediated via vagal innervation of the spleen and the α7 subunit of the nicotinic receptor expressed on the cell surface of the resident macrophages.^6,7 The relevance of vagally mediated modulation of systemic inflammation has been shown in sepsis and more recently by our group in left ventricular remodeling following acute myocardial infarction.

‘RECONNECTING THE BODY’ TO IMPROVE OUTCOMES

The continuing emergence of the link between psychiatric and neurocontrol of systemic inflammation offers an undeveloped strategy for further improving outcomes in patients with cardiovascular disease. One of our interests in pursuing heart-brain medicine is to reconnect the body and exploit the physiologic interplay between the heart and brain to improve patient outcomes.¹ Given the disappointments over the past year for new therapies like cholesteryl ester transfer protein inhibitors⁸ and vascular cell adhesion molecule (VCAM) inhibitors, strategies that have a singular organ or cellular target focus, now may be the time for exploiting multisystem approaches for modulating disease states such as CAD, congestive heart failure, and arrhythmia.

The potential consequences of these pathways are profound and include the following:

• A physiologic mechanism for the increased incidence of myocardial infarction observed with medications that have anticholinergic properties and potentially decrease autonomic tone
• Worse outcomes in patients with CAD and depression
• An increased incidence of CAD in patients with psychiatric disorders that in themselves may be associated with decreased vagal tone, as well as in patients on long-term drug therapies that alter parasympathetic tone
• Increased incidences of CAD, myocardial infarction, and death in patients with PTSD.

Clearly the underlying science of heart-brain medicine is fascinating and needs to be pursued vigorously. While the science is ongoing, the need to translate what we know to the bedside has never been greater, given the prevalence of CAD, chronic heart failure, and psychiatric and mood disorders, as well as the likelihood of an increasing incidence of PTSD in light of the Iraq war and terrorist threats.

Multiple studies have been performed to position the field for a trial to test whether treating depression leads to improved outcomes in patients with CAD. We know that patients with depression have decreased vagal tone based on decreased heart rate variability; we know that CAD can be safely treated with selective serotonin reuptake inhibitors; and we know that this patient population is more effectively treated with medications. There was a clear sentiment among faculty and attendees of the 2006 Bakken Heart-Brain Summit that the next step in the clinical science of heart disease and neurologic state is in fact a clinical trial to test the efficacy of this approach. Unfortunately, funding for such a trial from the pharmaceutical industry or government agencies is lacking. The Bakken Heart-Brain Institute is working diligently to secure private financing of such a trial from those with personal interests in moving this field forward. We hope to be able to commence such a trial in the near future. We believe the successful initiation of a multicenter trial not only will demonstrate new avenues for improving outcomes in millions of patients but will validate the concept and usher in a new age of cooperative medicine among multiple disciplines.

As we discussed last year,¹ both the need for and the future of heart-brain medicine are great. The advances seen over the past year and those being pursued in basic and clinical science laboratories throughout the world are very exciting. We thank those colleagues who attended the 2007 Bakken Heart-Brain Summit, and we hope you can join us June 4–5, 2008, in Cleveland to continue this exciting pursuit.

Depression’s association with incident coronary artery disease (CAD) and recurrent cardiac events became established 10 to 20 years ago. Efforts in the past decade have focused on the specific effects of treating depression in patients with CAD—whether such treatment is beneficial and, if so, exactly how it exerts its benefits. This article briefly surveys the current evidence on these questions after reviewing how we got interested in depression in CAD in the first place.

THE EMERGENCE OF DEPRESSION AS A CARDIAC RISK FACTOR

The shift from a focus on Type A behavior

Not long ago, Type A behavior pattern was the psychosocial variable of greatest research interest as a contributor to CAD. Just 26 years ago, a National Institutes of Health consensus development conference anointed Type A behavior pattern as a CAD risk factor.¹ Five years later, one of the landmark studies in psychosomatic medicine—the Recurrent Coronary Prevention Project—showed that Type A behavior modification, added to usual cardiac care in post–myocardial infarction (MI) patients, not only reduced patients’ Type A behavior but also reduced the rate of reinfarction and death.²

But that was the high-water mark for Type A behavior in CAD research. The focus soon shifted, especially after the publication of a 1987 review by Booth-Kewley and Friedman showing that larger and later studies found less and less impressive effects of Type A on cardiac outcomes.³ This same review pointed out the cumulative evidence indicating that depression might be the most important psychological factor associated with coronary disease.³

An explosion of research on depression in CAD

In the 10 years following the review by Booth-Kewley and Friedman, there was an explosion of study about depression in CAD.⁴ This resulted in what is fair to call a consensus on several key points about this relationship:

• Depression is associated with an approximate 1.5-­fold to twofold increase in the risk for incident CAD.^5–8
• Depression is associated with about a threefold to fourfold increase in the risk of recurrent cardiac events and death in patients with CAD, including patients with a new diagnosis, those with acute coronary events, and those who have undergone revascularization procedures.^9–13
• Several biobehavioral mechanisms are plausible candidates as mediators of the mind-body relationship linking depression and coronary disease. These include abnormal platelet function, autonomic function, inflammatory processes, and nonadherence to therapy.^4,14
• Depression is extremely common in CAD, affecting about 15% to 20% of patients, and is a serious illness in its own right, even apart from its effects on cardiac outcomes.^{9–11,15–17}

In light of these observations, the obvious research questions are whether treating depression in patients with CAD helps, and if so, what it helps with—the depression itself, the pathophysiology and outcomes of CAD, or both. These questions have been the increasing focus of the past 10 years.

DOES DEPRESSION THERAPY IN CAD PATIENTS IMPROVE DEPRESSION IN THESE PATIENTS?

The short answer to this question is an almost unqualified yes. Even setting aside the literature on tricyclic antidepressants (which is an old literature but impressive in its own right in its systematic working through of issues of efficacy and the delineation and management of adverse effects^18–25), we have at least half a dozen studies showing that depression treatment helps to relieve depression in patients with CAD with reasonable safety and efficacy. Some are open-label, small-scale studies, while others are more rigorously designed and controlled, but the overall conclusion is unambiguous.^26–34

Roose, Glassman, and colleagues were among the first to describe the effects of antidepressants other than tricyclics in cardiac patients.^26–29 They demonstrated the safety profile of bupropion, but did not report on its efficacy.²⁶ They demonstrated safety but found rather low efficacy of fluoxetine in doses up to 60 mg/day in markedly depressed inpatients, many of whom had a “melancholic” profile (early-morning waking, positive diurnal mood variation, guilt, anhedonia, poor appetite).^27,28 In a randomized double-blind trial, these same researchers subsequently demonstrated paroxetine to be at least as effective as the tricyclic agent nortriptyline and to have excellent tolerability at doses up to 40 mg/day.²⁹

Strik et al published an early study of the efficacy of depression treatment in 54 patients with major depression after a first MI.³⁰ Fluoxetine demonstrated superiority over placebo with respect to the percentage of patients achieving a clinical response (48% vs 26%; P = .05) (clinical response was defined as a ≥ 50% reduction in the Hamilton Depression Rating Scale [HAM-D] score), but fluoxetine did not have a statistically significant effect on HAM-D symptom ratings except in the subset of patients with mild symptoms to start with. This is somewhat counterintuitive, and to be contrasted with the results of SADHART.

The Sertraline Antidepressant Heart Attack Randomized Trial (SADHART), conducted in depressed patients following MI or unstable angina, is well known.^31,32 Patients with a recent acute coronary syndrome (acute MI in 74%; unstable angina in 26%) were randomized within 30 days of the coronary event to sertraline or placebo (following a 2-week placebo run-in period for all patients). Sertraline was associated with superior scores on the Clinical Global Impression Improvement Scale, particularly among patients with recurrent depression and more severe depression, but its effect on HAM-D scores was not significantly better than that of placebo. As opposed to the finding of Strik et al, the biggest difference in response was among patients with more severe depression symptoms rather than those with mild symptoms to begin with.

The Enhancing Recovery in Coronary Heart Disease Patients (ENRICHD) trial tested the hypothesis that psychosocial intervention aimed at depression and low levels of social support would improve cardiac prognosis in post-MI patients.³³ In this large randomized study (N = 2,481), cognitive behavior therapy exerted a modestly significant effect in reducing symptoms of depression as compared with usual medical care. Most patients in the intervention arm underwent 6 to 10 sessions of individual and/or group therapy over 6 months, and their HAM-D scores improved by approximately 10 points from baseline to 6-month follow-up. However, patients in the usual-care arm also had substantial improvements (almost 9 points) in HAM-D scores at 6-month follow-up.

The Canadian Cardiac Randomized Evalution of Antidepressant and Psychotherapy Efficacy (CREATE) used a 2 x 2 factorial design to assess interpersonal psychotherapy and antidepressant therapy (citalopram) for depression in patients with stable CAD.³⁴ Citalopram was more effective than placebo in reducing depression symptoms and in achieving response and remission. The mean decline in HAM-D scores was more than 3 points greater in citalopram recipients than in placebo recipients. Interpersonal psychotherapy was no more effective than clinical management.

In none of the studies reviewed above was the benefit of active treatment very powerful—response rates were between 50% and 60%, and remission rates were much lower.

In the ENRICHD trial, cognitive behavior therapy–based psychosocial intervention did not result in lower rates of recurrent MI or mortality compared with usual medical care, but the (nonrandomized) use of selective serotonin reuptake inhibitors (SSRIs) by some patients in the study was associated with a 42% reduction in the risk of death.^33,35

Likewise, depression intervention had no significant effect on cardiac outcome in SADHART, although this 369-patient study was not powered to demonstrate such a benefit.³¹ Deaths were reduced by more than 50% with sertraline compared with placebo, but there were only 5 and 2 deaths in the placebo and sertraline groups, respectively. The point estimate for sertraline’s effect on major adverse cardiac events was a 23% reduction in events (ie, relative risk of 0.77), but the 95% confidence interval corresponding to this relative risk was 0.51 to 1.16, indicating a lack of statistical significance.

Still, this 23% reduction from SADHART was suggestive—certainly enough to interest the cardiologists associated with the study. Together with the ENRICHD trial findings and results from case-control studies indicating that SSRI therapy reduces the risk of incident MI,^36,37 the SADHART findings have encouraged other investigators to suggest additional studies of the effects of antidepressant therapy on CAD outcomes.^38,39

Notably, in both the ENRICHD trial⁴⁰ and SADHART (unpublished data, manuscript in preparation), patients who recovered from depression (regardless of treatment assignment) had better long­term survival, and this was also true in a long-term longitudinal study of patients following coronary artery bypass graft (CABG) surgery.¹² This suggests that interventions to promote recovery from depression should be useful in improving cardiac prognosis, but it does not prove it. It may be that patients whose depression improves are in some way healthier, regardless of their depression intervention.

As noted above, data from observational case-control studies of patients admitted to coronary care units suggest that SSRI therapy reduces incident MI.^36,37 On the other hand, a study of mortality among patients undergoing CABG surgery revealed worse outcomes in those taking SSRIs than in those who were not.⁴¹ Because this study was observational and not randomized, its findings must be interpreted with caution. The effect observed could be due to an adverse effect of SSRI treatment, an adverse effect of depression, or some other mechanism.

DOES DEPRESSION THERAPY HAVE A BENEFICIAL EFFECT ON INTERMEDIARY MECHANISMS LINKING DEPRESSION TO CORONARY EVENTS?

The answer to this question depends on the specific mechanism being considered.

Platelet activation. In the case of platelet activation, the answer may be yes. In a randomized study of depressed patients with ischemic heart disease, paroxetine but not nortriptyline reduced elevated biomarkers of platelet activation.⁴² In a substudy of SADHART, blood levels of sertraline and desmethylsertraline were inversely correlated with platelelet activation.⁴³ Moreover, serotonin reuptake inhibitors appear to reduce platelet activation in proportion to their affinity for the serotonin transporter.^37,44

Heart rate variability. There is little evidence that depression therapy influences heart rate variability. In SADHART, for instance, sertraline and placebo did not differ in their effects on heart rate variability.³¹

Nonadherence to CAD therapy. It is clear that nonadherence to therapy is more common in depressed patients with cardiac disease than in their nondepressed counterparts^45,46 and that poor adherence is associated with worse cardiac outcomes.⁴⁷ But no study in depressed patients has yet demonstrated that depression treatment per se results in improved adherence. One study has demonstrated, however, that adherence tends to “travel with” depression over the course of treatment: as symptoms of depression declined, adherence improved.⁴⁸

FUTURE DIRECTIONS

In the future, a more efficient way to improve cardiac outcomes associated with depression may be to target interventions directly at intermediary mechanisms rather than at depression itself. For example, if depression is robustly associated with a deleterious effect on platelet function that heightens the risk of thrombus formation, it might be helpful to optimize antiplatelet therapy in patients with depression, independent of the depression treatment. Similarly, anti-inflammatory treatments might have added benefit in those cardiac patients who are depressed, because these patients tend to have abnormally elevated inflammatory activity, which is associated with worse outcomes. These hypotheses would need to be specifically tested in randomized controlled trials, of course.

Because depression is associated with smoking, a recent study of a 3-month smoking cessation intervention in patients admitted to a coronary care unit provides an instructive example.⁴⁹ At 2-year follow-up, significantly more patients had continuously abstained from smoking in the intervention group than in a usual-care control group (33% vs 9%, respectively), and significantly fewer patients had died in the intervention group compared with the control group (2.8% vs 12%, respectively).

Another desirable objective is the development of treatments that are more robust in their effects on depression for patients with CAD than the interventions tested so far. Higher rates of response and remission of depression would be highly desirable in their own right. Moreover, only with more potent interventions, whose effects separate more robustly from those seen with placebo or usual care, is it likely that depression treatments themselves could affect cardiac outcomes.

Depression’s association with incident coronary artery disease (CAD) and recurrent cardiac events became established 10 to 20 years ago. Efforts in the past decade have focused on the specific effects of treating depression in patients with CAD—whether such treatment is beneficial and, if so, exactly how it exerts its benefits. This article briefly surveys the current evidence on these questions after reviewing how we got interested in depression in CAD in the first place.

THE EMERGENCE OF DEPRESSION AS A CARDIAC RISK FACTOR

The shift from a focus on Type A behavior

Not long ago, Type A behavior pattern was the psychosocial variable of greatest research interest as a contributor to CAD. Just 26 years ago, a National Institutes of Health consensus development conference anointed Type A behavior pattern as a CAD risk factor.¹ Five years later, one of the landmark studies in psychosomatic medicine—the Recurrent Coronary Prevention Project—showed that Type A behavior modification, added to usual cardiac care in post–myocardial infarction (MI) patients, not only reduced patients’ Type A behavior but also reduced the rate of reinfarction and death.²

But that was the high-water mark for Type A behavior in CAD research. The focus soon shifted, especially after the publication of a 1987 review by Booth-Kewley and Friedman showing that larger and later studies found less and less impressive effects of Type A on cardiac outcomes.³ This same review pointed out the cumulative evidence indicating that depression might be the most important psychological factor associated with coronary disease.³

An explosion of research on depression in CAD

In the 10 years following the review by Booth-Kewley and Friedman, there was an explosion of study about depression in CAD.⁴ This resulted in what is fair to call a consensus on several key points about this relationship:

• Depression is associated with an approximate 1.5-­fold to twofold increase in the risk for incident CAD.^5–8
• Depression is associated with about a threefold to fourfold increase in the risk of recurrent cardiac events and death in patients with CAD, including patients with a new diagnosis, those with acute coronary events, and those who have undergone revascularization procedures.^9–13
• Several biobehavioral mechanisms are plausible candidates as mediators of the mind-body relationship linking depression and coronary disease. These include abnormal platelet function, autonomic function, inflammatory processes, and nonadherence to therapy.^4,14
• Depression is extremely common in CAD, affecting about 15% to 20% of patients, and is a serious illness in its own right, even apart from its effects on cardiac outcomes.^{9–11,15–17}

In light of these observations, the obvious research questions are whether treating depression in patients with CAD helps, and if so, what it helps with—the depression itself, the pathophysiology and outcomes of CAD, or both. These questions have been the increasing focus of the past 10 years.

DOES DEPRESSION THERAPY IN CAD PATIENTS IMPROVE DEPRESSION IN THESE PATIENTS?

The short answer to this question is an almost unqualified yes. Even setting aside the literature on tricyclic antidepressants (which is an old literature but impressive in its own right in its systematic working through of issues of efficacy and the delineation and management of adverse effects^18–25), we have at least half a dozen studies showing that depression treatment helps to relieve depression in patients with CAD with reasonable safety and efficacy. Some are open-label, small-scale studies, while others are more rigorously designed and controlled, but the overall conclusion is unambiguous.^26–34

Roose, Glassman, and colleagues were among the first to describe the effects of antidepressants other than tricyclics in cardiac patients.^26–29 They demonstrated the safety profile of bupropion, but did not report on its efficacy.²⁶ They demonstrated safety but found rather low efficacy of fluoxetine in doses up to 60 mg/day in markedly depressed inpatients, many of whom had a “melancholic” profile (early-morning waking, positive diurnal mood variation, guilt, anhedonia, poor appetite).^27,28 In a randomized double-blind trial, these same researchers subsequently demonstrated paroxetine to be at least as effective as the tricyclic agent nortriptyline and to have excellent tolerability at doses up to 40 mg/day.²⁹

Strik et al published an early study of the efficacy of depression treatment in 54 patients with major depression after a first MI.³⁰ Fluoxetine demonstrated superiority over placebo with respect to the percentage of patients achieving a clinical response (48% vs 26%; P = .05) (clinical response was defined as a ≥ 50% reduction in the Hamilton Depression Rating Scale [HAM-D] score), but fluoxetine did not have a statistically significant effect on HAM-D symptom ratings except in the subset of patients with mild symptoms to start with. This is somewhat counterintuitive, and to be contrasted with the results of SADHART.

The Sertraline Antidepressant Heart Attack Randomized Trial (SADHART), conducted in depressed patients following MI or unstable angina, is well known.^31,32 Patients with a recent acute coronary syndrome (acute MI in 74%; unstable angina in 26%) were randomized within 30 days of the coronary event to sertraline or placebo (following a 2-week placebo run-in period for all patients). Sertraline was associated with superior scores on the Clinical Global Impression Improvement Scale, particularly among patients with recurrent depression and more severe depression, but its effect on HAM-D scores was not significantly better than that of placebo. As opposed to the finding of Strik et al, the biggest difference in response was among patients with more severe depression symptoms rather than those with mild symptoms to begin with.

The Enhancing Recovery in Coronary Heart Disease Patients (ENRICHD) trial tested the hypothesis that psychosocial intervention aimed at depression and low levels of social support would improve cardiac prognosis in post-MI patients.³³ In this large randomized study (N = 2,481), cognitive behavior therapy exerted a modestly significant effect in reducing symptoms of depression as compared with usual medical care. Most patients in the intervention arm underwent 6 to 10 sessions of individual and/or group therapy over 6 months, and their HAM-D scores improved by approximately 10 points from baseline to 6-month follow-up. However, patients in the usual-care arm also had substantial improvements (almost 9 points) in HAM-D scores at 6-month follow-up.

The Canadian Cardiac Randomized Evalution of Antidepressant and Psychotherapy Efficacy (CREATE) used a 2 x 2 factorial design to assess interpersonal psychotherapy and antidepressant therapy (citalopram) for depression in patients with stable CAD.³⁴ Citalopram was more effective than placebo in reducing depression symptoms and in achieving response and remission. The mean decline in HAM-D scores was more than 3 points greater in citalopram recipients than in placebo recipients. Interpersonal psychotherapy was no more effective than clinical management.

In none of the studies reviewed above was the benefit of active treatment very powerful—response rates were between 50% and 60%, and remission rates were much lower.

In the ENRICHD trial, cognitive behavior therapy–based psychosocial intervention did not result in lower rates of recurrent MI or mortality compared with usual medical care, but the (nonrandomized) use of selective serotonin reuptake inhibitors (SSRIs) by some patients in the study was associated with a 42% reduction in the risk of death.^33,35

Likewise, depression intervention had no significant effect on cardiac outcome in SADHART, although this 369-patient study was not powered to demonstrate such a benefit.³¹ Deaths were reduced by more than 50% with sertraline compared with placebo, but there were only 5 and 2 deaths in the placebo and sertraline groups, respectively. The point estimate for sertraline’s effect on major adverse cardiac events was a 23% reduction in events (ie, relative risk of 0.77), but the 95% confidence interval corresponding to this relative risk was 0.51 to 1.16, indicating a lack of statistical significance.

Still, this 23% reduction from SADHART was suggestive—certainly enough to interest the cardiologists associated with the study. Together with the ENRICHD trial findings and results from case-control studies indicating that SSRI therapy reduces the risk of incident MI,^36,37 the SADHART findings have encouraged other investigators to suggest additional studies of the effects of antidepressant therapy on CAD outcomes.^38,39

Notably, in both the ENRICHD trial⁴⁰ and SADHART (unpublished data, manuscript in preparation), patients who recovered from depression (regardless of treatment assignment) had better long­term survival, and this was also true in a long-term longitudinal study of patients following coronary artery bypass graft (CABG) surgery.¹² This suggests that interventions to promote recovery from depression should be useful in improving cardiac prognosis, but it does not prove it. It may be that patients whose depression improves are in some way healthier, regardless of their depression intervention.

As noted above, data from observational case-control studies of patients admitted to coronary care units suggest that SSRI therapy reduces incident MI.^36,37 On the other hand, a study of mortality among patients undergoing CABG surgery revealed worse outcomes in those taking SSRIs than in those who were not.⁴¹ Because this study was observational and not randomized, its findings must be interpreted with caution. The effect observed could be due to an adverse effect of SSRI treatment, an adverse effect of depression, or some other mechanism.

DOES DEPRESSION THERAPY HAVE A BENEFICIAL EFFECT ON INTERMEDIARY MECHANISMS LINKING DEPRESSION TO CORONARY EVENTS?

The answer to this question depends on the specific mechanism being considered.

Platelet activation. In the case of platelet activation, the answer may be yes. In a randomized study of depressed patients with ischemic heart disease, paroxetine but not nortriptyline reduced elevated biomarkers of platelet activation.⁴² In a substudy of SADHART, blood levels of sertraline and desmethylsertraline were inversely correlated with platelelet activation.⁴³ Moreover, serotonin reuptake inhibitors appear to reduce platelet activation in proportion to their affinity for the serotonin transporter.^37,44

Heart rate variability. There is little evidence that depression therapy influences heart rate variability. In SADHART, for instance, sertraline and placebo did not differ in their effects on heart rate variability.³¹

Nonadherence to CAD therapy. It is clear that nonadherence to therapy is more common in depressed patients with cardiac disease than in their nondepressed counterparts^45,46 and that poor adherence is associated with worse cardiac outcomes.⁴⁷ But no study in depressed patients has yet demonstrated that depression treatment per se results in improved adherence. One study has demonstrated, however, that adherence tends to “travel with” depression over the course of treatment: as symptoms of depression declined, adherence improved.⁴⁸

FUTURE DIRECTIONS

In the future, a more efficient way to improve cardiac outcomes associated with depression may be to target interventions directly at intermediary mechanisms rather than at depression itself. For example, if depression is robustly associated with a deleterious effect on platelet function that heightens the risk of thrombus formation, it might be helpful to optimize antiplatelet therapy in patients with depression, independent of the depression treatment. Similarly, anti-inflammatory treatments might have added benefit in those cardiac patients who are depressed, because these patients tend to have abnormally elevated inflammatory activity, which is associated with worse outcomes. These hypotheses would need to be specifically tested in randomized controlled trials, of course.

Because depression is associated with smoking, a recent study of a 3-month smoking cessation intervention in patients admitted to a coronary care unit provides an instructive example.⁴⁹ At 2-year follow-up, significantly more patients had continuously abstained from smoking in the intervention group than in a usual-care control group (33% vs 9%, respectively), and significantly fewer patients had died in the intervention group compared with the control group (2.8% vs 12%, respectively).

Another desirable objective is the development of treatments that are more robust in their effects on depression for patients with CAD than the interventions tested so far. Higher rates of response and remission of depression would be highly desirable in their own right. Moreover, only with more potent interventions, whose effects separate more robustly from those seen with placebo or usual care, is it likely that depression treatments themselves could affect cardiac outcomes.

Depression’s association with incident coronary artery disease (CAD) and recurrent cardiac events became established 10 to 20 years ago. Efforts in the past decade have focused on the specific effects of treating depression in patients with CAD—whether such treatment is beneficial and, if so, exactly how it exerts its benefits. This article briefly surveys the current evidence on these questions after reviewing how we got interested in depression in CAD in the first place.

THE EMERGENCE OF DEPRESSION AS A CARDIAC RISK FACTOR

The shift from a focus on Type A behavior

Not long ago, Type A behavior pattern was the psychosocial variable of greatest research interest as a contributor to CAD. Just 26 years ago, a National Institutes of Health consensus development conference anointed Type A behavior pattern as a CAD risk factor.¹ Five years later, one of the landmark studies in psychosomatic medicine—the Recurrent Coronary Prevention Project—showed that Type A behavior modification, added to usual cardiac care in post–myocardial infarction (MI) patients, not only reduced patients’ Type A behavior but also reduced the rate of reinfarction and death.²

But that was the high-water mark for Type A behavior in CAD research. The focus soon shifted, especially after the publication of a 1987 review by Booth-Kewley and Friedman showing that larger and later studies found less and less impressive effects of Type A on cardiac outcomes.³ This same review pointed out the cumulative evidence indicating that depression might be the most important psychological factor associated with coronary disease.³

An explosion of research on depression in CAD

In the 10 years following the review by Booth-Kewley and Friedman, there was an explosion of study about depression in CAD.⁴ This resulted in what is fair to call a consensus on several key points about this relationship:

• Depression is associated with an approximate 1.5-­fold to twofold increase in the risk for incident CAD.^5–8
• Depression is associated with about a threefold to fourfold increase in the risk of recurrent cardiac events and death in patients with CAD, including patients with a new diagnosis, those with acute coronary events, and those who have undergone revascularization procedures.^9–13
• Several biobehavioral mechanisms are plausible candidates as mediators of the mind-body relationship linking depression and coronary disease. These include abnormal platelet function, autonomic function, inflammatory processes, and nonadherence to therapy.^4,14
• Depression is extremely common in CAD, affecting about 15% to 20% of patients, and is a serious illness in its own right, even apart from its effects on cardiac outcomes.^{9–11,15–17}

In light of these observations, the obvious research questions are whether treating depression in patients with CAD helps, and if so, what it helps with—the depression itself, the pathophysiology and outcomes of CAD, or both. These questions have been the increasing focus of the past 10 years.

DOES DEPRESSION THERAPY IN CAD PATIENTS IMPROVE DEPRESSION IN THESE PATIENTS?

The short answer to this question is an almost unqualified yes. Even setting aside the literature on tricyclic antidepressants (which is an old literature but impressive in its own right in its systematic working through of issues of efficacy and the delineation and management of adverse effects^18–25), we have at least half a dozen studies showing that depression treatment helps to relieve depression in patients with CAD with reasonable safety and efficacy. Some are open-label, small-scale studies, while others are more rigorously designed and controlled, but the overall conclusion is unambiguous.^26–34

Roose, Glassman, and colleagues were among the first to describe the effects of antidepressants other than tricyclics in cardiac patients.^26–29 They demonstrated the safety profile of bupropion, but did not report on its efficacy.²⁶ They demonstrated safety but found rather low efficacy of fluoxetine in doses up to 60 mg/day in markedly depressed inpatients, many of whom had a “melancholic” profile (early-morning waking, positive diurnal mood variation, guilt, anhedonia, poor appetite).^27,28 In a randomized double-blind trial, these same researchers subsequently demonstrated paroxetine to be at least as effective as the tricyclic agent nortriptyline and to have excellent tolerability at doses up to 40 mg/day.²⁹

Strik et al published an early study of the efficacy of depression treatment in 54 patients with major depression after a first MI.³⁰ Fluoxetine demonstrated superiority over placebo with respect to the percentage of patients achieving a clinical response (48% vs 26%; P = .05) (clinical response was defined as a ≥ 50% reduction in the Hamilton Depression Rating Scale [HAM-D] score), but fluoxetine did not have a statistically significant effect on HAM-D symptom ratings except in the subset of patients with mild symptoms to start with. This is somewhat counterintuitive, and to be contrasted with the results of SADHART.

The Sertraline Antidepressant Heart Attack Randomized Trial (SADHART), conducted in depressed patients following MI or unstable angina, is well known.^31,32 Patients with a recent acute coronary syndrome (acute MI in 74%; unstable angina in 26%) were randomized within 30 days of the coronary event to sertraline or placebo (following a 2-week placebo run-in period for all patients). Sertraline was associated with superior scores on the Clinical Global Impression Improvement Scale, particularly among patients with recurrent depression and more severe depression, but its effect on HAM-D scores was not significantly better than that of placebo. As opposed to the finding of Strik et al, the biggest difference in response was among patients with more severe depression symptoms rather than those with mild symptoms to begin with.

The Enhancing Recovery in Coronary Heart Disease Patients (ENRICHD) trial tested the hypothesis that psychosocial intervention aimed at depression and low levels of social support would improve cardiac prognosis in post-MI patients.³³ In this large randomized study (N = 2,481), cognitive behavior therapy exerted a modestly significant effect in reducing symptoms of depression as compared with usual medical care. Most patients in the intervention arm underwent 6 to 10 sessions of individual and/or group therapy over 6 months, and their HAM-D scores improved by approximately 10 points from baseline to 6-month follow-up. However, patients in the usual-care arm also had substantial improvements (almost 9 points) in HAM-D scores at 6-month follow-up.

The Canadian Cardiac Randomized Evalution of Antidepressant and Psychotherapy Efficacy (CREATE) used a 2 x 2 factorial design to assess interpersonal psychotherapy and antidepressant therapy (citalopram) for depression in patients with stable CAD.³⁴ Citalopram was more effective than placebo in reducing depression symptoms and in achieving response and remission. The mean decline in HAM-D scores was more than 3 points greater in citalopram recipients than in placebo recipients. Interpersonal psychotherapy was no more effective than clinical management.

In none of the studies reviewed above was the benefit of active treatment very powerful—response rates were between 50% and 60%, and remission rates were much lower.

In the ENRICHD trial, cognitive behavior therapy–based psychosocial intervention did not result in lower rates of recurrent MI or mortality compared with usual medical care, but the (nonrandomized) use of selective serotonin reuptake inhibitors (SSRIs) by some patients in the study was associated with a 42% reduction in the risk of death.^33,35

Likewise, depression intervention had no significant effect on cardiac outcome in SADHART, although this 369-patient study was not powered to demonstrate such a benefit.³¹ Deaths were reduced by more than 50% with sertraline compared with placebo, but there were only 5 and 2 deaths in the placebo and sertraline groups, respectively. The point estimate for sertraline’s effect on major adverse cardiac events was a 23% reduction in events (ie, relative risk of 0.77), but the 95% confidence interval corresponding to this relative risk was 0.51 to 1.16, indicating a lack of statistical significance.

Still, this 23% reduction from SADHART was suggestive—certainly enough to interest the cardiologists associated with the study. Together with the ENRICHD trial findings and results from case-control studies indicating that SSRI therapy reduces the risk of incident MI,^36,37 the SADHART findings have encouraged other investigators to suggest additional studies of the effects of antidepressant therapy on CAD outcomes.^38,39

Notably, in both the ENRICHD trial⁴⁰ and SADHART (unpublished data, manuscript in preparation), patients who recovered from depression (regardless of treatment assignment) had better long­term survival, and this was also true in a long-term longitudinal study of patients following coronary artery bypass graft (CABG) surgery.¹² This suggests that interventions to promote recovery from depression should be useful in improving cardiac prognosis, but it does not prove it. It may be that patients whose depression improves are in some way healthier, regardless of their depression intervention.

As noted above, data from observational case-control studies of patients admitted to coronary care units suggest that SSRI therapy reduces incident MI.^36,37 On the other hand, a study of mortality among patients undergoing CABG surgery revealed worse outcomes in those taking SSRIs than in those who were not.⁴¹ Because this study was observational and not randomized, its findings must be interpreted with caution. The effect observed could be due to an adverse effect of SSRI treatment, an adverse effect of depression, or some other mechanism.

DOES DEPRESSION THERAPY HAVE A BENEFICIAL EFFECT ON INTERMEDIARY MECHANISMS LINKING DEPRESSION TO CORONARY EVENTS?

The answer to this question depends on the specific mechanism being considered.

Platelet activation. In the case of platelet activation, the answer may be yes. In a randomized study of depressed patients with ischemic heart disease, paroxetine but not nortriptyline reduced elevated biomarkers of platelet activation.⁴² In a substudy of SADHART, blood levels of sertraline and desmethylsertraline were inversely correlated with platelelet activation.⁴³ Moreover, serotonin reuptake inhibitors appear to reduce platelet activation in proportion to their affinity for the serotonin transporter.^37,44

Heart rate variability. There is little evidence that depression therapy influences heart rate variability. In SADHART, for instance, sertraline and placebo did not differ in their effects on heart rate variability.³¹

Nonadherence to CAD therapy. It is clear that nonadherence to therapy is more common in depressed patients with cardiac disease than in their nondepressed counterparts^45,46 and that poor adherence is associated with worse cardiac outcomes.⁴⁷ But no study in depressed patients has yet demonstrated that depression treatment per se results in improved adherence. One study has demonstrated, however, that adherence tends to “travel with” depression over the course of treatment: as symptoms of depression declined, adherence improved.⁴⁸

FUTURE DIRECTIONS

In the future, a more efficient way to improve cardiac outcomes associated with depression may be to target interventions directly at intermediary mechanisms rather than at depression itself. For example, if depression is robustly associated with a deleterious effect on platelet function that heightens the risk of thrombus formation, it might be helpful to optimize antiplatelet therapy in patients with depression, independent of the depression treatment. Similarly, anti-inflammatory treatments might have added benefit in those cardiac patients who are depressed, because these patients tend to have abnormally elevated inflammatory activity, which is associated with worse outcomes. These hypotheses would need to be specifically tested in randomized controlled trials, of course.

Because depression is associated with smoking, a recent study of a 3-month smoking cessation intervention in patients admitted to a coronary care unit provides an instructive example.⁴⁹ At 2-year follow-up, significantly more patients had continuously abstained from smoking in the intervention group than in a usual-care control group (33% vs 9%, respectively), and significantly fewer patients had died in the intervention group compared with the control group (2.8% vs 12%, respectively).

Another desirable objective is the development of treatments that are more robust in their effects on depression for patients with CAD than the interventions tested so far. Higher rates of response and remission of depression would be highly desirable in their own right. Moreover, only with more potent interventions, whose effects separate more robustly from those seen with placebo or usual care, is it likely that depression treatments themselves could affect cardiac outcomes.

CARDIAC CASE PRESENTATION

A 53-year-old woman, a malpractice lawyer, with a history of mitral valve prolapse was diagnosed with severe mitral regurgitation and referred for mitral valve repair.

History and examination

The patient had no other cardiac history. She reported jogging 2 to 3 miles daily and playing tennis regularly, but over the past few months she had become more fatigued during her jogs, to the point that she occasionally had to reduce her pace and even shorten the duration of her runs.

On her initial visit, she expressed surprise regarding the severity of her mitral valve disease, as she had always been healthy. She seemed somewhat nervous but appropriately concerned about the impending surgery, and questioned whether she would be able to return to her previous level of activity. She also mentioned that she hoped the timing of the surgery would permit her to attend her son’s college graduation in 9 weeks.

Her medical history was notable for mitral valve prolapse. She had a history of panic attacks, for which she occasionally took alprazolam. There was no family history of cardiac disease. She did not use tobacco and occasionally consumed alcohol. A review of systems was negative.

Her physical examination was unremarkable except for a grade 4/6 holosystolic murmur at the apex that radiated to the axilla, which was consistent with the mitral regurgitation.

A transthoracic echocardiogram demonstrated mitral regurgitation that extended through the left atrium back into the pulmonary veins. The left ventricular ejection fraction was 50%, which is considered low-normal. The degree of mitral regurgitation was 4+. No other significant valvular disease was observed.

An electrocardiogram revealed a normal sinus rhythm. Per our routine, the patient underwent cardiac catheterization, which showed normal coronary arteries.

An uncomplicated repair, but slow recovery

The mitral valve repair was performed without complications. The course in the intensive care unit was uncomplicated, and the patient was quickly extubated and transferred to a regular nursing floor.

On the nursing floor, controlling the patient’s pain was difficult. She refused to use her incentive spirometer and initially refused to ambulate or even move from her bed to a chair. She was quite tearful.

A postoperative transthoracic echocardiogram revealed a satisfactorily repaired mitral valve with no mitral regurgitation. Her ejection fraction decreased to 40%, which is not unusual after mitral valve surgery.

Her hospital course was notable for an episode of shortness of breath and tachycardia. Sinus tachycardia was evident on review of the telemetry strips. A repeat echocardiogram showed no changes compared with the prior postoperative echocardiogram. Spiral computed tomography was negative for pulmonary embolism.

Pain control remained difficult. The patient expressed concern about the postoperative decrease in her ejection fraction; she was reassured that a decrease in ejection fraction was not unusual, but she remained tearful. The family expressed concern because the patient “wasn’t acting like herself,” and her ambulation and use of her incentive spirometer continued to be minimal, which had the potential to hamper her recovery and rehabilitation. For these reasons, a psychiatric consult was requested and the patient was seen prior to discharge from the hospital on postoperative day 6.

Wound check at 1 week postdischarge

A routine wound check was performed 1 week after discharge, at which time the patient was still reporting pain that was more severe than would be expected at her postoperative stage. She reported concern about drainage from the incision. She said that she was unable to do much walking or stair climbing, and she reported sleeping in the guest bedroom on the first floor of her house because she was unable to negotiate the stairs to her bedroom.

A check of the wound showed minimal serous drainage at the inferior aspect and was consistent with normal wound healing. The slow progress of her recovery was a concern, as was the possible contribution of her anxiety to this slow progress, so we kept our psychiatric colleagues informed about the patient’s recovery.

Follow-up at postoperative week 4

At the follow-up visit at postoperative week 4, the patient reported still being in pain, although the pain had improved, and complained of constant fatigue and shortness of breath that prevented her from returning to work. She had been discharged on lisinopril and admitted to occasional medication noncompliance. She said that if she did not improve dramatically and quickly, she would not be able to attend her son’s graduation.

We considered the possibility of new ischemia, a large pleural effusion, postpericardiotomy syndrome, constrictive pericarditis, or a mitral valve leak as potential causes of her symptoms. A chest radiograph was obtained, which demonstrated a small left pleural effusion, and an echocardiogram showed that her ejection fraction remained at 40% and the mitral valve repair remained intact. The patient had a psychiatric visit scheduled later on the day of this follow-up visit and was referred to the cardiac rehabilitation program, to start on week 6 of her postoperative care.

At the time of the first psychiatric consult, postoperative day 6, the patient’s chart was reviewed, detailing her presentation and hospital course as described above. The chart confirmed one episode of “panic” following surgery while the patient was on telemetry, showing only sinus tachycardia. This episode was successfully treated with 1 mg of lorazepam. She expressed a fear of “losing it,” which is how she characterized her panicky state during the hospital stay, punctuated by the feeling that she was not in control. The nursing staff reported that she was distressed and irritable. Her husband also confirmed that the patient “was not herself.”

Her baseline functioning was high; she is a partner in a law firm and is customarily “in control.” Before the interview began, the patient had several questions ready, including how quickly she would heal, how soon she could return to work and resume her normal activities, the reason for her low ejection fraction despite having mitral valve surgery, and whether or not she would be able to attend her son’s graduation. Even though she knew the psychiatry consult had been ordered, she was not very receptive to it at first and was more focused on her physical symptoms.

Psychiatric history

Her psychiatric history was significant for fear of heights and panic attacks, but she had been able to conquer each. She overcame performance anxiety in high school and was able to be a successful malpractice attorney, deliberating cases in court. She had never seen a psychiatrist or mental health professional, and had never been on psychotropic medications, although for the past couple of years she had been using 0.5 mg of alprazolam to treat flight anxiety. She admitted to postpartum depression that lasted about 2 months; no treatment was sought at the time, and the depression resolved.

Family and personal history

Her mother was a teacher and a “professional worrier,” and her father is a retired lawyer. She reported resolving to “suck it up” during times of adversity during childhood, but her childhood was otherwise unremarkable. She is an only child and finished at the top of her class at law school.

Review of symptoms

An assessment of depressive symptoms using the mnemonic SIGECAPS (disturbance of sleep; disturbance of interest; presence of guilt; disturbance of energy, concentration, or appetite; increased or decreased psychomotor activity; ideas of suicide) elicited low energy levels, decreased concentration, and a “slowed down” feeling. The WART (withdrawal, anhedonia, rumination, tearfulness) scale, used to assess depressive symptoms in the medically ill, showed the patient to be withdrawn and tearful at times.

Mental status examination

The patient was polite, professionally courteous, and sitting up in bed. Her vital signs were stable (heart rate, 70 beats per minute; blood pressure, 122/72 mm Hg) and her mood was “fine,” although she had many concerns about her physical health. Her affect was serious, constricted, and controlling. Her thought process was linear and organized, and her thought content revealed no psychosis, suicidal ideations, or overt hopelessness. She admitted that she was slightly anxious and overwhelmed, and that this anxiety precipitated her “panic” on telemetry and tearfulness, but she believed (and asked for assurance) that this level of anxiety was normal following surgery.

Diagnosis and recommendations

By the end of the consultation, we were able to make a series of recommendations. We arrived at a diagnosis of adjustment disorder with anxious features, and we agreed to treat her with alprazolam at a dose of 0.5 mg twice daily as needed. We provided education about mood and anxiety disorders in cardiac patients. We explained that her postpartum depression was a risk factor for future depression. We discussed coping strategies and relaxation techniques, and scheduled a follow-up appointment with her primary care physician for further monitoring of her mood and anxiety.

One week postdischarge

The cardiology team communicated with us after her wound check at postdischarge week 1. At this time, she was still having pain and was concerned about excessive wound drainage even though it was found to be minimal. The cardiology team was concerned because her progress was slow and she appeared anxious and tense. A follow-up psychiatry consultation was arranged for the patient’s next postoperative visit.

Follow-up at postoperative week 4

At her scheduled psychiatric visit at postoperative week 4, the patient was a little surprised to see the fellow, as she expected to see the staff psychiatrist. She appeared tense and frustrated, was fixated on her echocardiogram and her physical symptoms, and reported that she was not yet back to work. She was preoccupied with her son’s graduation that was coming up and wondered if she would be able to attend and celebrate it.

We administered the Patient Health Questionnaire depression scale, and the patient’s score of 11 indicated moderate depression. Treatment options, including psychotherapy and pharmacotherapy with a selective serotonin reuptake inhibitor (SSRI), were reviewed with the patient. A call to the cardiology team revealed that her ejection fraction was fairly typical for a patient who has had a mitral valve repair but that the continued fatigue was not normal, leading us to suspect that depression may be the actual cause of her fatigue. She remarked, “Let’s see how the cardiac rehabilitation program goes and then we’ll talk about medications for depression.”

Cardiac rehabilitation at postoperative week 6

The patient was entered into the cardiac rehabilitation program, and she was administered a Short Form–36 (SF-36) health survey, which showed a low mental summary score and a low physical component summary score (low scores connote worse health and/or more disability). She was referred to the psychiatrist at the cardiovascular behavioral health clinic for further assessment of her mood as she commenced the cardiac rehabilitation program.

To explore management options in this case and discuss the insights it provides into heart-brain interactions, the case presentation was followed by an interactive discussion (moderated by Dr. James B. Young, Department of Cardiovascular Medicine and Chairman, Division of Medicine at Cleveland Clinic) between the physicians who presented the case and the Heart-Brain Summit audience.

Dr. James Young: Let’s begin by considering whether there were some red flags that may have been apparent up front to predict that this patient might have been challenging in the postoperative period. I think one red flag was the diagnosis of mitral valve prolapse itself, which has been known to occur in type A personalities, who tend to exhibit catecholamine excess and sympathetic nervous system arousal that activates the autonomic nervous system.

Also, I’d be interested to know a few more findings from the patient’s physical examination. Was she thin? Did she have a narrow anteroposterior diameter? Did she have pectus excavatum? Did she have arachnodactyly tendencies? These are important characteristics that might have flagged the anxiety up front, as psychosomatic manifestations of patients with mitral valve prolapse were identified—and hotly debated—20 to 30 years ago. Although the link between mitral valve prolapse and personality type has fallen out of favor in cardiology circles, it clearly seems to describe this patient. The history of anxiety, panic, and possibly agoraphobia has been well described in patients with mitral valve prolapse and excematous degeneration.

I’d like to pose the following questions to the audience. What do you do with this patient now? Do you push medication therapy? Do you push psychotherapy? What is the next step?

Comment from audience: You haven’t excluded the post-pump syndrome. This patient is very bright and it wouldn’t take much of an insult to impair her sufficiently so that she would interpret the world in a different way. From my point of view, she needs sophisticated neuropsychological testing soon.

Dr. Young: That’s a good point. We know that cardiopulmonary bypass is associated with difficulties and problems that have been underreported in the past.

Comment from audience: The last thing that this patient wants to admit or even allude to is a psychological problem. She is the last one who’s going to even hint at it, which makes it very easy to miss. Look at how she reacted when she heard that there was a psychiatrist in the room. These patients are not necessarily well disposed to completing screening tests because they recognize that somebody is trying to identify a psychological problem. I don’t know that I have the answer, but I think that we should avoid browbeating ourselves for the problem.

Dr. Young: I want to mention the cultural anthropology of physicians and how it affects our approach to treatment. I like being a cardiologist because I write prescriptions for drugs that have proven to be useful, such as beta-blockers and ACE inhibitors, among others. From this experience came my earlier question, “Should we give this patient a drug?” The cardiologist’s focus—perhaps excessive focus—on pharmacologic solutions may not be the best way to approach this patient. You allude to some important issues about screening a patient for diseases that can be more easily treated.

Comment from audience: I have seen such situations as a result of drug interactions; many of our patients are on multiple drugs when they leave the hospital. The other issue to consider is sleep deprivation, with or without sleep apnea.

Dr. Young: Many complications, particularly in patients with heart failure, are related to disordered sleep, which certainly causes some heart-brain dysfunction. What about the drugs?

Dr. Thomas Callahan: We considered the effects of her medications, which included an ACE inhibitor and her analgesics. We also considered the lingering effects of anesthesia or other medications that she might have been receiving.

Dr. Young: Remember, she was reporting considerable pain. I suspect that she was on a cocktail of pain medications that might have been contributing to her difficulties.

Comment from audience: Morphine’s effects tend to be stronger in women than in men. The other issue is the 10% drop in ejection fraction after the surgery. This patient may be thinking, “Why did I go through all of this if my ejection fraction is going to be worse?”

Dr. Callahan: A drop in the ejection fraction, especially after mitral valve repair, is common. We often address it with patients preoperatively, but perhaps not with everyone, and perhaps not clearly enough.

Dr. Young: Also, this is an example of a patient who had heart failure going into the operation, but “heart failure” would be the worst term to use with this particular patient. An ejection fraction of 50% is not normal for a patient with 4+ mitral regurgitation and, as Dr. Callahan suggested, when you take away the mitral regurgitation, you dump a little more load on the left ventricle, and the ejection fraction will go down. We see this all the time, although I admit that cardiologists or cardiac surgeons don’t necessarily do the best job of discussing these subtleties with patients. Something we can take away from this case is a sense of the importance of improving our communications with patients about what they might expect postoperatively, although it still needs to be tailored to the individual patient. If this patient had understood the pathophysiology behind the drop in ejection fraction, it may have helped her. Other patients, on the other hand, may not require detailed conversations about this phenomenon.

Comment from audience: It was mentioned several times that the husband said the patient was not herself. Did you interact with the husband and the son to get a sense of the long-term dynamics of this family? It seems that there may have been some issues with the family dynamics.

Dr. Ubaid Khokhar: That’s a good question, although no underlying dynamics seemed apparent. The husband and son’s primary concern was that the patient’s previous characteristics of perfectionism and always being “in control” were so much in contrast with the tearful episodes she was having now. “She is not the same,” is how they kept phrasing it. However, there were no other significant changes—no rumination about suicide, no overt unwillingness to go along with treatment, or anything like that.

Comment from audience: I believe strongly that this patient was depressed, although she did not admit it. She had four of the five symptoms. She did not admit to a depressed mood but was tearful, which you reported at every postoperative visit. This is a sign of depression. We know very well that anxiety and depression often are present in tandem, especially in patients with high baseline anxiety. When they have more stress in their lives, they tend to get depressed.

I agree with the preceding comments that drug interactions are a potential worry; however, a few of the SSRIs have favorable drug-drug interaction profiles. I would urge this patient to try SSRI therapy. If she rejected this by responding, “I’m not depressed,” you could point out that SSRIs work very well for anxiety. Alprazolam is not a good medication for anxiety because it has a very short half-life, which can leave patients with an increase in anxious feelings after the medication is cleared from their system but before their next dose.

In addition to SSRI therapy as a first-line approach, I would try stress management, biofeedback, or even psychosupportive therapy that relies on patient education to help this patient understand her condition and take back control.

Our initial approach with this patient was the path of least resistance. Very good points have been made by the discussants and members of the audience. This patient was attached to alprazolam because it was the only psychotropic medication that she had ever taken. For this reason, she was discharged on alprazolam even though it wasn’t the ideal medication. As pointed out by the audience, the patient was quite resistant to the concept of having depression superimposed on a history of anxiety. In the cardiac rehabilitation setting she was again reassured by the exercise physiologists that her heart was doing well. A cardiologist personally reviewed the echocardiographic reports and films with the patient, pointed out the absence of unusual abnormalities with her heart, and suggested that something else was causing her symptoms. This direct explanation and reassurance from the cardiologist facilitated the patient’s ability to entertain depression as a comorbid condition.

At the visit with the psychiatrist in the cardiac rehabilitation program, the patient finally accepted that her lack of confidence could also be a symptom of depression. We repeated the Patient Health Questionnaire, which still showed moderate depression, and we started her on an SSRI, citalopram. About 3 weeks later, she began to regain her confidence, and she was able to attend and host her son’s graduation. By 8 weeks after the start of antidepressant therapy, a repeat Patient Health Questionnaire showed no evidence of depression.

Her progress, both physically and emotionally, was quite pronounced during the 12-week cardiac rehabilitation program. Her physical stamina improved, her fatigue abated, and her sense of confidence was restored. She successfully returned to work and her family concurred that she had returned to her “old self.” She benefited from the stress management and lifestyle seminars that were offered in the cardiac rehabilitation program, and her exit SF-36 scores were much improved. The patient pleasantly surprised us all by taking the initiative of forming a monthly women’s support group for coping with heart surgery.

She completed a 9-month course of the SSRI, with the depression in full remission, and has continued to follow up with her cardiologist and her exercise regimen.