Etiology and Diagnosis of Systolic Murmurs in Adults

Transcription

1 CLINICAL RESEARCH STUDY Etiology and Diagnosis of Systolic Murmurs in Adults Steven McGee, MD Primary and Specialty Medical Care, Department of Veterans Affairs Medical Center, Seattle, Wash; Department of Medicine, University of Washington, Seattle. ABSTRACT BACKGROUND: It is unknown whether echocardiography can provide insights into the origin of systolic murmurs and the modern value of bedside cardiovascular diagnosis. METHODS: The author examined 376 inpatients and compared their physical findings to transthoracic echocardiography, exploring the associations between echocardiography and systolic murmurs and investigating the diagnostic accuracy of physical examination for pathologic murmurs. RESULTS: Four echocardiographic variables predict the presence of systolic murmurs: peak aortic velocity (P.001); mitral regurgitation severity (P.001); mitral valve E-point velocity (P.09); and absence of pericardial effusion (P.09). When diagnosing murmurs, the most helpful finding is its distribution on the chest wall with respect to the 3 rd left parasternal space, a landmark that distinguishes murmurs into 6 patterns. The apical-base pattern indicates increased aortic velocity (likelihood ratio [LR] 9.7; 95% confidence interval [CI]; ): a delayed carotid upstroke (LR 6.8; 95% CI; ); absent S2 (LR 12.7; 95% CI; ); and humming quality to the murmur (LR 8.5; 95% CI; ) further increase the probability of aortic valve disease. The broad apical murmur pattern suggests significant mitral regurgitation (LR 6.8; 95% CI; ); and the left lower sternal murmur pattern indicates significant tricuspid regurgitation (LR 8.4; 95% CI; ): additional bedside observations refine these diagnoses. Nonetheless, this study shows that some classic physical findings are no longer accurate, that physical examination cannot reliably distinguish severe aortic stenosis from less severe stenosis, and that classic physical findings, despite having proven value, are absent in many patients with significant cardiac lesions. CONCLUSIONS: In the diagnosis of systolic murmurs, physical examination has limitations but also unappreciated value. A simple system using onomatopoeia and classifying systolic murmurs into 1 of 6 patterns is diagnostically helpful Published by Elsevier Inc. The American Journal of Medicine (2010) 123, KEYWORDS: Heart auscultation; Heart murmurs; Physical diagnosis Funding: None. Conflict of Interest: There are no financial or personal relationships that could have inappropriately biased this work. Authorship: The author performed all aspects of the study and analysis of its results. Requests for reprints should be addressed to Steven McGee, MD, Seattle-Puget Sound VA Health Care System (111M), 1660 South Columbian Way, Seattle, WA address: steven.mcgee@med.va.gov In his Treatise on the Diseases of the Heart and Great Vessels, 1 published in 1832, just 11 years after Laennec s invention of the stethoscope, the British physician James Hope fully described the characteristics of systolic murmurs, attributing them to either abnormal forward flow over semilunar valves (eg, aortic or pulmonic valve) or regurgitation of blood from high pressure chambers into low pressure ones (eg, mitral or tricuspid regurgitation). His observations along with those of Austin Flint ( ), Graham Steell ( ), and others using phonocardiography during the 1950s and 1960s form virtually our entire knowledge base about systolic murmurs, including the classic teaching that pathologic systolic murmurs are identifiable by their location on the chest wall and by additional abnormalities of the neck veins, precordial pulsations, arteries, and heart tones. During recent decades the diagnosis and treatment of heart disease have changed significantly, and it is unknown whether these classic tenets remain true. For example, original observations were mostly of young patients with rheumatic and congenital heart disease and normal ejection fractions; modern patients, in contrast, /$ -see front matter 2010 Published by Elsevier Inc. doi: /j.amjmed

2 914 The American Journal of Medicine, Vol 123, No 10, October 2010 are frequently elderly and more often have ischemic, calcific, or degenerative heart disease with reduced ventricular function. Echocardiography has been widely applied to modern cardiovascular diagnosis, yet few if any studies providing additional insights into systolic murmurs have appeared. The purpose of this study is to carefully examine a series of hospitalized patients and compare their physical findings to transthoracic echocardiography, thereby investigating both the etiology of systolic murmurs and the modern diagnostic value of the bedside examination. METHODS CLINICAL SIGNIFICANCE Study Protocol Between 2001 and 2006, the author examined 409 inpatients at the Seattle Veterans Affairs Medical Center. These patients were a convenience sample, principally of non-intensive care unit patients undergoing echocardiography during their hospital stay. With only 14 exceptions, the author was unaware of the patient s diagnosis, indication for echocardiography, or echocardiographic results. Using a standardized form, the author recorded the patient s vital signs, arterial and venous pulsations (contour, velocity, waveforms), precordial pulsations (location, velocity, amplitude), heart tones (first, second, third, fourth, and extra heart sounds and their characteristics), and murmurs (systolic, diastolic, or both). Examination of the arteries, veins, and precordium preceded auscultation. The anterior chest from apex to clavicles was examined, and radiation of murmurs was completely described. Although most patients were examined in 3 positions (ie, supine, left lateral decubitus, and upright positions), reported findings refer only to the supine patient. Murmurs were defined as continuous sounds that persisted during inspiration and expiration, although their intensity could vary during the respiratory cycle. In contrast, continuous sounds that completely disappeared during inspiration or expiration were called rubs. All murmurs were characterized using onomatopoeia and conventional grading (Table 1) 2 and were sorted into 6 predetermined topographic patterns (Figure 1). All echocardiograms were interpreted by a cardiologist independent from the bedside examination. Echocardiographic regurgitation was significant only if moderate or severe. The study protocol was approved by the institutional review board at the Seattle-Puget Sound Veterans Affairs Medical Center. Four echocardiographic findings are associated with systolic murmurs: 1) aortic valve velocity, 2) mitral regurgitation severity, 3) absence of pericardial effusions, and 4) mitral valve E-point velocity. When diagnosing systolic murmurs, the most helpful physical finding is its distribution on the chest wall with respect to the 3 rd left parasternal space. Additional classic cardiovascular findings, although diagnostically accurate, are sometimes absent in patients with significant cardiac pathology. Statistical Methods Using echocardiographic parameters as the independent variable and either presence of systolic murmur or the specific murmur pattern as the dependent variable, the author conducted both univariate analysis (2-sided t-test for continuous variables; chi-squared test for categorical variables) and multivariate analysis (logistic regression with SPSS 13 software; SPSS Inc., Chicago, IL). In the multivariate analysis, variables entered the model if P.05 and remained if P.10. Diagnostic accuracy was expressed using likelihood ratios (LRs). 3 RESULTS There were 409 patients examined, aged 22 to 94 years (mean SD, years); 399 (97%) were men. The indications for echocardiography included assessments for structural heart disease (59%), progression of preexisting valvular disease (16%), source of arterial emboli (8%), suspected endocarditis (7%), or suspected pericardial disease (2%). Only 7% of the echocardiograms were ordered to diagnose unexplained murmurs. Thirty-three patients were excluded from analysis because they had diastolic or systolic/diastolic murmurs (n 18) or lacked complete echocardiograms (n 15), leaving 376 patients, 221 (59%) of whom had systolic murmurs. Associations between Echocardiography and Presence of Systolic Murmur As displayed in Table 2 (online), over 20 echocardiographic findings were associated with systolic murmurs, but after multivariate analysis only 4 variables remained: aortic valve (AV) peak velocity (P.001), E-point velocity (P.092), mitral regurgitation severity (P.001), and absence of pericardial effusion (P.090). In addition, tricuspid regurgitation severity was independently associated with the left lower sternal border murmur pattern (P.001). Pericardial effusions diminished the probability of all 6 murmur patterns (ie, 60% of patients without pericardial effusions had murmurs vs. only 25% of those with effusions; most effusions were clinically unimportant). Increasing E-point velocity also was associated with all 6 systolic murmur patterns, and its association persisted after excluding patients with increased AV velocity or mitral regurgitation (ie, in the 106 patients with AV velocity 1.8 m/sec and without mitral regurgitation, 43% with E-point velocity 0.9 m/sec had systolic murmurs vs. 21% with E-point velocity 0.9 m/sec, P.03).

3 McGee Etiology and Diagnosis of Systolic Murmurs 915 Table 1 Characteristic Timing of sound* Midsystolic Early systolic Long systolic Holosystolic Late systolic Quality Blowing Coarse Definitions of Physical Findings Definition Both S1 (lub) and S2 (dup) distinct: Lub shsh dup S1 indistinct, S2 distinct; gap before S2 Shshsh dup Pushsh dup S1 indistinct, S2 distinct; no gap before S2: Shshshshshdup Pushshshshdup S1 and S2 indistinct: Shshshshshsh ShshshshshshP Pushshshshsh PushshshshshP S1 distinct, S2 indistinct: Lub shshshp Pure high frequency, mimicked by the sounds ahahah or shshsh (sounds produced in the front of the mouth) Mixture of frequencies, mimicked by the sound of clearing the throat (produced farther back in the mouth) Intensity loudest grade recorded 1 Faint, heard with special effort 2 Recognized readily after placing the stethoscope on the chest wall 3 Very loud, but lacks palpable thrill Carotid upstroke Delayed Brisk Right ventricular (RV) rock Definite slow increase in carotid upstroke, occupying much of systole and different from the early tapping sensation of the normal carotid Unusually brisk carotid upstroke, different from the early tapping sensation of the normal carotid Simultaneous systolic retraction of the apical area and outward motion of the lower sternum (this precordial motion results from a dilated right ventricle at the apex ejecting blood into the dilated right atrium and liver near the sternum) *Characteristics of S1 and S2 are based on distinctness of heart sounds at the location where the murmur is loudest. All murmurs in this study lacked palpable thrills and were thus grade 3 or less. 2 Both carotids were examined at the anterior border of the sternocleidomastoid muscle in the mid neck, using the examiner s left thumb (on the patient s right carotid) or right fingertips (on the patient s left carotid). Correlations between Murmur and Echocardiography Peak aortic velocity, mitral regurgitation, and tricuspid regurgitation were the 3 principal echocardiographic variables associated with specific murmur patterns. Because many patients had combinations of these abnormalities, Figure 2 presents isolated lesions to simplify analysis. As AV peak velocity increases from 1.3 m/sec to 4 m/sec, the frequency of murmurs increases from 19% to 100%: the isolated base pattern appears at AV peak velocities of m/sec (see later), the isolated apical and small apicalbase patterns at m/sec, and the broad apical-base pattern at 2.1 m/sec. As mitral regurgitation increases from none to severe, the frequency of murmur increases from 29% to 100%: the broad apical pattern is the most common pattern, although some patients with severe regurgitation have the broad apical-base pattern and others with moderate regurgitation have the isolated apical pattern. As tricuspid regurgitation increases from none to severe, the frequency of murmur increases from 21% to 100%: the left lower sternal pattern is the only associated pattern. Table 3 summarizes the combinations of increased AV velocity, mitral regurgitation, and tricuspid regurgitation in all patients with each murmur pattern. Peak AV velocity is divided into 5 groups: 1.8 m/sec (normal AV velocity); m/sec (increased AV velocity without obstruction); and m/sec, m/sec, and 4 m/sec, velocities corresponding to mild, moderate, and severe aortic stenosis, respectively. Twenty percent of all patients had aortic stenosis (ie, AV velocity 2.5 m/sec), 20% had mitral regurgitation (moderate or worse), and 18% had tricuspid regurgitation (moderate or worse). Diagnostic Accuracy of Physical Examination All Patients. The most useful finding, applicable to all patients, is the specific murmur pattern detected. The broad apical-base pattern increases the probability of AV velocity 2.5 m/sec (LR 9.7), the broad apical pattern increases the probability of mitral regurgitation (LR 6.8), and the left lower sternal pattern increases the probability of tricuspid regurgitation (LR 8.4). The absence of systolic murmur greatly decreases the probability of AV velocity 2.5 m/sec (LR 0.05), but only slightly decreases the probability of regurgitation (LRs 0.4 and 0.6 for mitral and tricuspid regurgitation, respectively) (Table 4). Additional findings increasing the probability of AV velocity 2.5 m/sec are inaudible S2 at the left 2 nd parasternal space, radiation of the murmur to both sides of the neck and clavicles, humming murmur quality, delayed carotid upstroke, inaudible S1 at the apex, and coarse murmur quality (LRs , see Table 4). Additional findings decreasing the probability of AV velocity 2.5 m/sec (and thus arguing against aortic stenosis) include mid-systolic timing, absence of radiation into the neck, grade 1 intensity, and broad apical pattern (LRs ).

4 916 The American Journal of Medicine, Vol 123, No 10, October 2010 Figure 1 The 6 murmur patterns are distinguished by their distribution with respect to a key landmark, the 3 rd left parasternal interspace (indicated by on each drawing). Two patterns (broad apical-base pattern and small apical-base pattern, top 2 rows) extend above and below this landmark, usually to both sides of the sternum. The broad apicalbase pattern extends at least from the first right parasternal space to the apex (ie, 4 th intercostal space at the midclavicular line [MCL]), whereas the small apical base pattern does not extend this far. Three patterns are confined entirely below this landmark (left lower sternal pattern, broad apical pattern, and isolated apical pattern, 3 rd through 5 th rows); 1 pattern is confined entirely above this landmark (isolated base pattern, bottom row). The vertical line over the left chest in each drawing depicts the MCL. Additional findings increasing the probability of mitral regurgitation (LRs ) are a loud S2 at the left 2 nd parasternal space, unchanging murmur intensity despite an irregular rhythm, and chaotic pulse rhythm (ie, atrial fibrillation). Findings supporting a diagnosis of significant tricuspid regurgitation (LRs ) are early systolic out-

5 McGee Etiology and Diagnosis of Systolic Murmurs 917 Figure 2 Murmur patterns of isolated lesions. The figure presents the murmur patterns reflecting isolated aortic velocity (top row; 247 patients without significant mitral or tricuspid regurgitation), isolated mitral regurgitation (middle row; 174 patients with AV peak velocity 1.8 m/sec and without significant tricuspid regurgitation), and isolated tricuspid regurgitation (bottom row; 161 patients with AV velocity 1.8 m/sec and without significant mitral regurgitation). See text. ward movement in the neck veins (ie, CV wave), lower sternal pulsations, hepatic pulsations, and a right ventricular rock. All pathologic murmurs had long systolic or holosystolic timing or, in some patients with mitral regurgitation, late systolic timing. Patients with Broad Apical-base. The most common murmur pattern was the broad apical-base pattern (n 92, 25% of all patients). Ninety-eight percent of these patients had increased AV velocity ( 1.8 m/sec, Table 3): 28% had increased aortic flow without stenosis, 42% had mild or moderate aortic stenosis, and 28% had severe aortic stenosis. In patients with the broad apical-base pattern, several findings accurately detect aortic stenosis (ie, AV velocity 2.5 m/sec, Table 4): humming quality, delayed carotid upstroke, and inaudible S1 or S2; others indicate increased flow without stenosis (ie, AV velocity m/sec): mid-systolic timing, brisk carotid upstroke, grade 1 intensity, and blowing quality. Nonetheless, few findings accurately diagnose severe aortic stenosis (AV velocity 4 m/sec): only blowing quality and early-systolic timing decrease the probability of this diagnosis (LRs 0.1 for both). In patients with the broad apical-base pattern, a loud S2 increases the probability of mitral regurgitation (LR 3.3), a diagnosis otherwise difficult to make because the broad apical-base pattern circumscribes the broad apical pattern. Patients with Other Murmur s. Seventy-four percent of patients with the small apical-base pattern have modest increases in aortic velocity (AV velocity m/sec, Table 3), but aortic stenosis is rare. In patients with the isolated apical pattern, 35% have mildly increased aortic velocity and 28% have mitral regurgitation, but no

6 918 The American Journal of Medicine, Vol 123, No 10, October 2010 Table 3 Characteristics of Various s* Echocardiographic Findings All Patients Murmur No Murmur Broad Apical-base Small Apical-base Left Lower Sternal Broad Apical Isolated Apical Isolated Base n Aortic velocity (m/sec) (57) 129 (86) 2 (2) 4 (21) 12 (60) 28 (70) 11 (65) 23 (79) (23) 19 (13) 25 (27) 14 (74) 5 (25) 10 (25) 6 (35) 6 (21) (5) 2 (1) 13 (14) 0 2 (10) 1 (3) (8) 0 26 (28) 1 (5) 1 (5) 1 (3) (7) 0 26 (28) Mitral regurgitation 74 (20) 14 (9) 19 (21) 3 (16) 4 (22) 27 (63) 5 (28) 2 (7) Tricuspid regurgitation 65 (18) 18 (12) 13 (14) 2 (11) 13 (65) 15 (34) 2 (11) 2 (7) *All values n (%, of total in column). These aortic velocities correspond roughly to normal velocity ( 1.8 m/sec), increased flow without obstruction ( m/sec), mild aortic stenosis ( m/sec), moderate aortic stenosis ( m/sec), and severe aortic stenosis ( 4 m/sec). Regurgitation, moderate or worse. additional finding distinguishes these patients. The isolated base pattern is associated with modestly increased aortic velocity (21% have AV velocity ; Figure 2, Table 3), but many of these murmurs are probably extracardiac (eg, arteriovenous fistula or great artery stenosis), because they frequently radiate to distant sites over the neck or clavicle and because 53% of patients had either hemodialysis fistulae or asymmetry of the carotid or radial pulses. Frequency of Findings Inspection of the frequency of finding column (Table 4) indicates that many patients lack anything diagnostic other than a specific murmur pattern. For example, in patients with the broad apical-base pattern, a delayed carotid upstroke argues for AV velocity 2.5 m/sec (LR 6.3) and a brisk carotid upstroke argues against this diagnosis (LR 0.05), but 61% of patients with this pattern have a normal carotid upstroke, a finding of little diagnostic value (LR 0.7). Therefore, although classic findings have proven accuracy, they are frequently absent. DISCUSSION This study leads to 3 important conclusions. First, although it confirms the classic teaching that systolic murmurs are associated with increased AV velocity, mitral regurgitation, and tricuspid regurgitation, it identifies 2 additional variables associated with systolic murmurs: the absence of pericardial effusions and increased E-point velocity. Pericardial effusions (even if small) decrease the probability of systolic murmurs, probably just as pleural effusions impair transmission of lung sounds. The role of E-point velocity is less obvious because it measures early diastolic flow over the mitral valve. Increased E-point velocity may reflect elevated filling pressures, which tense the ventricular walls and render them more susceptible to vibrations producing sound, similar to the way a tense violin string is more likely to vibrate than a relaxed one. Second, systolic murmurs are distinguished into 6 diagnostic patterns, based on the murmur s distribution about the 3 rd left parasternal space (the landmark overlying both aortic and mitral valves). Inspection of the boundary surrounding all 6 murmur patterns suggests that the primary determinant of sound radiation is not necessarily the direction of blood flow but instead the orientation of bony thorax, specifically the left lower ribs, sternum, and clavicles (Figure 3). Increased flow across a semilunar valve or through a regurgitant leak generates vibrations in the ventricles, great arteries, or both, which depending on their location, amplitude, and ease of conduction to the bones of the body wall produce 1 of the 6 different murmur patterns. In fact, one of the best arguments that bone conduction and not direction of blood flow governs distribution of sound is the murmur of mitral regurgitation: in this lesion, blood flows from the left ventricle rightward and upward to the left atrium, yet the murmur radiates almost perpendicular to this, along the left lower ribs to the axilla. Finally, despite changing etiologies and treatments of heart disease, many classic physical findings remain accurate today (carotid upstroke, murmur radiation and quality, venous waveforms, and others). A new observation is the association between a loud S2 at the left base and significant mitral regurgitation. S2 may be loud in mitral regurgitation because of pulmonary hypertension, absence of a loud contiguous murmur obscuring S2 (ie, mitral regurgitation murmurs are confined below the 3 rd rib), a freely mobile aortic valve (ie, no calcific aortic disease) or, in patients with associated aortic disease, a shorter left ventricular ejection time (thus shortening the associated aortic murmur and revealing a loud S2).

7 McGee Etiology and Diagnosis of Systolic Murmurs 919 Table 4 Diagnostic Accuracy of Bedside Examination Finding Frequency of Finding (%) Likelihood Ratio (95% CI), for Detecting*: AV Peak Velocity 2.5 m/sec Mitral Regurgitation Tricuspid Regurgitation All patients Murmur pattern Broad apical-base pattern (6.7-14) 1.1 ( ) 0.8 ( ) Small apical-base pattern (0-1.6) 0.8 ( ) 0.6 ( ) Left lower sternal pattern ( ) 1.1 ( ) 8.4 ( ) Broad apical pattern ( ) 6.8 ( ) 2.5 ( ) Isolated apical pattern (0-1.9) 1.5 ( ) 0.6 ( ) Isolated base pattern (0-1.1) 0.3 ( ) 0.4 ( ) No murmur (0-0.2) 0.4 ( ) 0.6 ( ) Arterial pulse Carotid upstroke delayed (4-11.5) 2.0 ( ) 1.4 ( ) Carotid upstroke normal ( ) 0.9 ( ) 1.0 ( ) Carotid upstroke brisk (0-2.2) 0.6 ( ) 0.2 (0-2.8) Pulse rhythm Regular ( ) 0.5 ( ) 0.6 ( ) Chaotic ( ) 2.9 ( ) 3.0 ( ) Venous pulsation Early systolic outward movement (CV wave) (5.5-22) Precordial pulsation Lower sternal pulsation (4.1-38) Hepatic pulsation (3.3-44) RV rock ( ) Heart tones S1 inaudible, apex ( ) 1.4 ( ) 1.0 ( ) S2 inaudible, 2 nd left parasternal space ( ) 0.5 ( ) 1.4 ( ) S2 loud, 2 nd left parasternal space ( ) 4.7 ( ) 3.6 ( ) Patients with systolic murmurs Murmur radiation No radiation to neck ( ) 1.6 ( ) 1.5 ( ) Radiation to right clavicle, right neck, or both ( ) 0.7 ( ) 0.8 ( ) Radiation to clavicles and neck on both sides (4.5-34) 0.5 ( ) 0.4 ( ) Murmur quality Blowing quality ( ) 1.5 ( ) 1.4 ( ) Coarse quality ( ) 0.5 ( ) 0.5 ( ) Added humming quality ( ) 0.7 ( ) 1.3 ( ) Murmur timing Midsystolic (0-0.3) 0.4 ( ) 0.6 ( ) Early systolic ( ) 0.4 ( ) 0.4 ( ) Long systolic ( ) 1.8 ( ) 1.9 ( ) Holosystolic ( ) 2.0 ( ) 1.5 ( ) Late systolic (0-3.3) 3.9 (0.7-23) 0.9 ( ) Murmur grade Grade ( ) 0.6 ( ) 1.3 ( ) Grade (1-1.5) 0.9 ( ) 0.9 ( ) Grade ( ) 2.3 ( ) 1.1 ( ) If pulse rhythm irregular, murmur intensity same ( ) 2.5 ( ) 2.3 ( ) in the beat after a pause Patients with broad apical-base murmurs Arterial pulse Carotid upstroke delayed ( ) Carotid upstroke normal ( ) Carotid upstroke brisk (0-0.9)......

8 920 The American Journal of Medicine, Vol 123, No 10, October 2010 Table 4 Continued Finding Frequency of Finding (%) Likelihood Ratio (95% CI), for Detecting*: AV Peak Velocity 2.5 m/sec Mitral Regurgitation Heart tones S1 inaudible, apex ( ) 0.7 ( )... S2 inaudible, 2 nd left parasternal space ( ) 0.2 (0-1.5)... S2 loud, 2 nd left parasternal space ( ) 3.3 ( )... Murmur characteristics Radiation to clavicles and neck on both sides ( ) No radiation lateral to mid clavicular line ( ) Blowing quality ( ) Added humming quality ( ) Midsystolic timing (0-0.6) Grade 1 intensity (0-0.7) CI confidence interval; RV rock right ventricular rock (definition in Table 1). *Values are likelihood ratio (95% confidence interval). Regurgitation, moderate or worse. The point estimate is statistically significant (ie, confidence interval excludes the value of 1.0). Tricuspid Regurgitation Figure 3 Boundary of murmur patterns. The 3 rd left parasternal space overlies both the aortic and mitral valves. If the ventricles vibrate sufficiently to produce sound, murmurs are generated below this landmark. Vibrations of the right ventricle produce the left lower sternal pattern, whereas those of the left ventricle produce the isolated apical pattern, larger apical patterns (left example of broad apical pattern, Figure 1) or, if of sufficient intensity, murmurs along the left ribs from sternum to axilla (right example of broad apical pattern, Figure 1). Should the great arteries vibrate sufficiently to make sound, the bones above this landmark vibrate and murmurs radiate from the upper sternum to clavicles and neck ( isolated base pattern). With increased velocity across the aortic valve, both the left ventricle (lower ribs) and great arteries (upper sternum and clavicles) vibrate, causing the apical-base pattern and its variations. This study also provides evidence supporting the hypothesis that observation of murmur intensity during irregular rhythms is diagnostically helpful. 4 After pauses in the heart rhythm (from atrial fibrillation or extrasystoles), the next ventricular beat has increased contractility, increased filling, and decreased afterload relative to prior beats. In patients with aortic flow murmurs, these hemodynamic changes increase aortic flow and thus murmur intensity. In regurgitant lesions, however, blood is flowing in 2 directions, and the diminished afterload increases aortic flow but leaves regurgitant volume and murmur intensity unchanged. 5 In contrast to classic teachings, mitral regurgitation is no longer associated with a brisk arterial pulse (the historical small water hammer pulse ), probably because modern patients are older and lack the supranormal ejection fractions and compliant vessels of younger, historical subjects. Also, the sustained apical impulse and displaced apical impulse are not specific for aortic stenosis or mitral regurgitation, respectively, because these findings, when found in study patients with murmurs, often signified alternative nonvalvular etiologies (eg, cardiomyopathy). Also, in contrast to descriptions of rheumatic aortic valve disease that emphasized murmurs located at the upper right sternum (the classic aortic area ), this study demonstrated that aortic valve murmurs radiate symmetrically above and below the 3 rd left parasternal space, in an oblique direction to both sides of the sternum, in a pattern sometimes resembling a sash worn over the right shoulder ( broad apical-base pattern). Finally, classifying the timing of murmurs using onomatopoeia (Table 1) proved useful, demonstrating holosystolic and long systolic murmurs as more significant than early systolic or mid-systolic ones. Onomatopoetic descriptors are also easier to convey to students than older terms (eg, diamond-shaped, crescendo-decrescendo ), probably because they communicate what clinicians actually hear, not what was seen on a phonocardiographic tracing. Limitations of this study include a population almost entirely of men; whether the female breast alters the radiation of the sound and the conclusions of this study is unknown. Also, patients consisted of a convenience sample and their examinations were necessarily detailed (about 20 minutes per exami-

9 McGee Etiology and Diagnosis of Systolic Murmurs 921 nation): results may not apply to uncooperative or sicker patients, noisy environments, or more cursory examinations. Even so, traditional cardiovascular examinations have always been performed systematically in quiet settings, and the time required is similar to that for a complete echocardiogram. Finally, examinations were conducted by a single observer, raising the possibility of poor reproducibility, but many studies have previously shown good interobserver agreement for most cardiovascular findings. 6 In conclusion, the main causes of distinct systolic murmur patterns are increased aortic velocity, mitral regurgitation, and tricuspid regurgitation. Pericardial effusions, even if small, diminish the frequency of all murmur patterns, and increased mitral valve E-point velocity further increases the likelihood of systolic sound. When diagnosing systolic murmurs, the most important physical finding is the distribution of sound on the chest wall with respect to the 3 rd left parasternal space. Additional classic cardiovascular findings refine these observations, but they are sometimes absent, thus illustrating both the value of the bedside examination and its limits. ACKNOWLEDGMENT This material is the result of work supported by use of facilities at the VA Puget Sound Health Care System, Seattle, Washington. References 1. Hope J. Treatise on Diseases of the Heart and Great Vessels, 1st edn. London: William Kidd; Constant J, Lippschutz EJ. Diagramming and grading heart sounds and murmurs. Am Heart J. 1965;70(3): Simel DL, Samsa GP, Matchar DB. Likelihood ratios with confidence: sample size estimation for diagnostic test studies. J Clin Epidemiol. 1991;44(8): Henke RP, March HW, Hultgren HN. An aid to identification of the murmur of aortic stenosis with atypical localization. Am Heart J. 1960; 60(3): Karliner JS, O Rourke RA, Kearney DJ, Shabetai R. Haemodynamic explanation of why the murmur of mitral regurgitation is independent of cycle length. Br Heart J. 1973;35: McGee SR. Reliability of physical findings. In: McGee SR, ed. Evidence-based Physical Diagnosis, 2nd edn. Philadelphia: W. B. Saunders; 2007:28-46.

10 921.e1 The American Journal of Medicine, Vol 123, No 10, October 2010 Table 2 Systolic Murmur Associations with Echocardiography (online only) Variable Murmur Present (n 226) Murmur Absent (n 165) P Value Age (years) BMI Blood pressure (mm Hg) Systolic Diastolic Pulse rate (beats per minute) Respiratory rate (breaths per minute) Velocities (m/sec) Aortic valve peak * Left ventricular outflow tract Pulmonary artery peak Mitral valve E-point * Mitral valve A point Regurgitation severity (0-7) Mitral regurgitation * Tricuspid regurgitation Ejection fraction (1-9) Dimensions (cm) Aortic root Left atrium Left ventricle, diastolic Left ventricle, systolic IVS thickness PW thickness Other (%) Aortic sclerosis Prosthetic aortic valve Mitral annular calcification Mitral valve thickening Mitral stenosis Prosthetic mitral valve Mitral valve prolapse Ruptured chordae Dilated left atrium Segmental hypo/akinesis Global dysfunction Left ventricular hypertrophy Dilated right atrium Dilated right ventricle Pericardial effusion * BMI body mass index; IVS interventricular septum; PW posterior wall. *Variables significant in the multivariate model (, odds ratio, P-value): peak AV velocity (2.246, 9.452,.001), E-point velocity (0.918, 2.504,.092), mitral regurgitation severity (0.361, 1.435,.001), pericardial effusion ( 1.279, 0.278,.090); constant for the model was Regurgitation severity: 0 none, 1 trace, 2 trace/mild, 3 mild, 4 mild/moderate, 5 moderate, 6 moderate/severe, 7 severe. Ejection fraction: 1 20%, %, %, %, %, %, % (normal), 8 60%, 9 65%.