Family Practice Vol. 19, No. 4, 422-425
© Oxford University Press 2002
Selections from Current Literature |
Horton hears a Who but no murmurs— does it matter?
Department of Family Medicine, Health Sciences Center L-4, 050, SUNY at Stony Brook, Stony Brook, NY 11794, USA.
Kopes-Kerr CP. Horton hears a Who but no murmurs—does it matter? Family Practice 2002; 19: 422–425.
Introduction to the cardiac auscultation problem
Schneiderman H. Cardiac ausculation and teaching rounds: how can cardiac ausculation be resuscitated?Am J Med 2001; 110: 233–235.[ISI][Medline]
A recent editorial in the American Journal of Medicine laments that cardiac auscultation skills appear to be heading for extinction. To reach this conclusion, one has to concur with two implicit assumptions: (i) at some time in the illustrious past of medicine, there was a moment when physicians were good at this skill and we have somehow declined from that peak; and (ii) that this matters—either to patients or to physicians in terms of outcomes—that somehow significant numbers of patients hover on the brink of catastrophe due to failure of auscultatory diagnosis. Let us examine for a moment some of the relevant literature to see if this is actually the case.
Background: what is the recorded history of cardiac auscultation skills?
Lembo NJ, Dell'Italia LJ, Crawford MH, O'Rourke RA. Bedside diagnosis of systolic murmurs.N Engl J Med 1988; 318: 1572–1578.[Abstract]
Tavel ME. Cardiac auscultation: a glorious past—but does it have a future?Circulation 1996; 93: 1250–1253.
The history of cardiac auscultation using modern standards of phonocardiographic evidence is quite checkered. The literature suggests that we were quite late to look at the ‘problem’ systematically, and then, ever since, have compiled a cumulative record of physician skillfulness that is significantly below mediocre. An early report from 1988 had to acknowledge that previously the diagnostic accuracy of bedside manoeuvres in the evaluation of patients with systolic murmurs had never been assessed objectively. Lembo et al. then evaluated 50 patients with documented systolic murmurs, compared all standard bedside examination techniques, and documented various manoeuvres that could enhance sensitivity and specificity for various specific diagnoses. They did not raise the question of whether patient outcomes would be significantly or measurably improved. A 1996 article by Dr Tavel acknowledged that there was no structured teaching of cardiac auscultation in 75% of American internal medicine programmes and 66% of cardiology programmes. He opined that atrophy of this discipline and skill was likely to spread to other basic history and physical exam skills. While he offered some thoughtful advice on how to improve teaching of this skill, he too neglected to address the issue of whether patient outcomes would be affected.
Modern clinical studies
Smythe JF, Teixeira OH, Vlad P, Demers PP, Feldman W. Initial evalaution of heart murmurs: are laboratory tests necessary?Pedatrics 1990; 86: 497–500.
Jost CHAet al. Echocardiography in the evaluation of systolic murmurs of unknown cause.Am J Med 2000; 108: 614–620.[ISI][Medline]
Gaskin PRet al. Clinical auscultation skills in pediatric residents.Pediatrics 2000; 105: 1184–1187.
Mangione S, Nieman LZ. Cardiac auscultatory skills of internal medicine and family practice trainees: a comparison of diagnostic proficiency.J Am Med Assoc 1997; 278: 717–722.[Abstract]
Oddone EZet al. Teaching cardiovascular examination skills: results from a randomized controlled trial.Am J Med 1993; 95: 389–396.[ISI][Medline]
Lok CE, Morgan CD, Ranganathan Net al. The accuracy and interobserver agreement in detecting the ‘Gallop Sounds’ by cardiac auscultation.Chest 1998; 114: 1283–1288.
Mangione S. Cardiac auscultatory skills of physiciansin-training: a comparison of three English-speaking countries.Am J Med 2001; 110: 210–216.[ISI][Medline]
Cardiologists' auscultatory exams
We have a reasonable sample of literature evaluating cardiac auscultation among both paediatric and adult patients by cardiologists, paediatricians and generalists (internists and family physicians). A 1990 article looked at paediatric cardiology practice in a prospective series of 161 patients, of which 98 were ‘innocent’ murmurs; only two were classified incorrectly on auscultation alone. Paediatric cardiologist auscultatory exams had a sensitivity of 96%, specificity of 95%, positive predictive value of 88% and negative predictive value of 98%. In a similar study among adult cardiologists, however, echocardiographic follow-up of cardiologists' examinations showed that the cardiologists misclassified up to 33% of innocent murmurs.
Paediatricians' auscultatory exams
In one study, 47 paediatric residents had their auscultatory exams assessed by using the cardiovascular patient simulator known as ‘Harvey’ to simulate five common lesions (VSD, ASD, pulmonic stenosis, combined aortic stenosis/insufficiency and innocent systolic ejection murmur). The diagnostic accuracy of the residents was only 33%, and the condition most often misdiagnosed was the most common one—innocent systolic ejection murmur.
Generalists' auscultatory exams
A 1997 study compared the cardiac auscultatory proficiency of 453 physicians in training and 88 medical students. All subjects listened to recordings of 12 distinct cardiac events directly recorded from patients; 1 point was assigned to each event identified correctly. The trainees' cumulative scores ranged from 0 to 7 for both internal medicine and family practice residents (median, 2.5 and 2.0, respectively). Internal medicine residents had the highest cumulative adjusted scores for the six extra sounds and for all 12 cardiac events tested, but on average the internal medicine and family practice residents recognized only 20% of all cardiac events. The number of correct identifications improved little with years of training and was not significantly higher than the number identified by medical students. In another study using Harvey, 56 internal medicine interns rotating through an eight-session physical diagnosis course as part of a cardiology rotation were randomized with respect to the use of the Harvey simulator. There were similar, moderate improvements in diagnostic ability with the simulator, but at no time did the proportion of correct responses exceed 64%. For example, the diagnosis of mitral regurgitation improved from 42 to 54%, mitral stenosis from 8 to 23%, and aortic regurgitation from 46 to 58%.
Other cardiac sounds
A study of gallops (S3 and S4) analysed the reliability of two cardiologists, one general internist, three senior and two junior postgraduate internal medicine trainees. Physician findings of an S3 or S4 had a positive predictive value of only 51 and 71%, respectively. There was no trend in the accuracy of interobserver agreement related to the level of observer experiences. Agreement between observers and the phonocardiographic gold standard in correctly identifying S4 and S3 sounds was not much higher than chance alone.
Finally, the most recent and comprehensive assessment of cardiac auscultation skills of physicians appeared this year in the American Journal of Medicine. The authors tested the cardiac auscultatory skills of 314 internal medicine residents (189 from the USA, 89 from Canada and 36 from England) from 14 different programmes. All subjects were asked to listen by stethophones to 12 pre-recorded cardiac events and to answer a multiple-choice questionnaire. Mean identification scores for the 12 cardiac events ranged from 0 to 58% for American trainees, 0 to 58% for Canadians and 0 to 42% for British trainees. The authors conclude that auscultatory proficiency was poor in all three countries.
Analysis
Is there a problem? If so, have we defined it correctly? What is it worth doing about it?
At first glance, the above data might appear to suggest there is a rather serious problem in general cardiac auscultation skills. The astute observer, however, will quickly dismiss that conclusion as premature since none of the investigators documented a ‘problem’ in terms of outcomes. They looked only at a rather unsatisfactory process. As an alternative mode of analysis, we might want to ask: are patients dying or suffering worse outcomes because physicians cannot diagnose their cardiac sounds accurately? On this, we have no data. Literature's very silence on this question, however, may suggest an answer. It does not appear that anyone has found an outcome problem since no reports have emerged despite our ever intensifying medico-legal scrutiny of such issues. In fact, the very data just presented actually suggest that more rigorous training in cardiac auscultation may be completely pointless. There is no evidence whatsoever of a significant learning effect or of any actual achievement of ‘satisfactory’ proficiency. When even cardiologists miss 33% of the easy ones, there seems little reason to believe that primary care physicians are ever likely to do well enough for it to matter. These relatively depressing data describe only the most common and most significant cardiac findings and fail to justify any presumption that we should be searching out, or letting ourselves be confused further by even less common or more obscure varieties of cardiac sounds. All of our history of medical education in this respect has simply failed, if you consider the correct identification of cardiac sounds to be important. This failure does not appear to be due to a lack of diligent effort. Rather it persists despite steadily intensifying effort with ever-improving technology. Perhaps, it is just time to acknowledge failure and just move on to something worthwhile.
When is cardiac auscultation worthwhile in primary care and what should we be seeking to learn?
Carabello BA, Crawford FA. Valvular heart disease.N Engl J Med 1997; 337: 32–41.
I make a modest proposal. In view of our conspicuous, but apparently benign failure to identify cardiac sounds correctly, we should simply give up listening to the hearts of patients without symptoms. We are not very good at it. We have enough to do anyway. Some of what we have to do is really important so that valuable time should not be wasted on pursuit of either the unfeasible or the impossible. At the very least, we should refrain from much investment in this skill until some data are published that demonstrate that our deficiency in this area is actually remediable by time, effort and training, and ultimately is a cost-effective use of time and resources.
In the meantime, we should concern ourselves with the available literature that tells us when cardiac auscultation matters—when patients have symptoms. The implicit assumption in assigning a priority to correct identification of cardiac sounds, particularly murmurs, is that this is the best way to diagnose cardiac disease, particularly cardiac valvular disease. An excellent recent review of cardiac valvular disease supports this view; even when there is definite organic valvular disease present, there is no need for specific diagnosis, intervention or therapy until patients become symptomatic.
For the lesion of aortic stenosis there is no medical therapy. Surgical intervention is not indicated until there are classic symptoms, i.e. surgical intervention would not be warranted on the basis of the aortic valvular gradient alone, no matter how severe. If the diagnosis is made timely after the onset of symptoms, surgical therapy is quite effective in prolonging life. There is no need to listen for a murmur until a patient becomes symptomatic with dyspnoea, angina or syncope. There is no value and probably much harm to be done by listening to asymptomatic persons for the murmur of aortic stenosis; the principle harm will be in the misclassifying of an innocent murmur, typically calcific aortic stenosis or a functional murmur, as a potentially significant lesion.
For aortic regurgitation, the value of medical therapy with nifedipine is unclear. Surgical intervention is not warranted until there are typical symptoms of angina or CHF. The decision for surgical intervention is made on the basis of the ‘55 rule’—surgery should be performed before the ejection fraction falls below 55% or the end-systolic dimension exceeds 55 mm.
For mitral stenosis, there is no medical therapy and no need for surgical intervention until there are symptoms. Patients with significant mitral stenosis will have symptoms of left ventricular CHF—dyspnoea on exertion, orthopnea and PND; they may also have haemoptysis, hoarseness and symptoms of right ventricular CHF. For the asymptomatic patient in sinus rhythm, prophylaxis against endocarditis is the only medical therapy indicated. With mild symptoms, diuretic therapy will lower left atrial pressure. When symptoms are more than mild, balloon valvotomy provides excellent results, with the alternative of open commissurotomy, valve reconstruction or mitral valve replacement for more difficult cases.
For chronic, non-ischaemic mitral regurgitation, there may be left ventricular damage before symptoms develop. There is no apparent benefit from long-term vasodilator therapy. Surgery is indicated when ‘more than mild symptoms’ develop or when the ejection fraction falls from a hyperdynamic fraction in the 65% range back below 60% or when the left ventricular end-systolic dimension exceeds 45 mm.
These then are the most common significant organic lesions as well as the ones that are easiest to recognize clinically. None of them appear to be associated with improved clinical outcomes merely by detection of the murmur in the pre-symptomatic stage.
Conclusion
There never appears to have been an era when cardiac auscultation was more effective than it is today. Why are we lately bemoaning the problem with increasing intensity? The reason, I believe, is simply that we are defining the problem incorrectly. A good part of this is nothing more than our vain attempt to perpetuate the myth of the ‘Renaissance’ clinician. Another part is a simple, but critical failure to consider the time costs and other externalities of any clinical practice featuring routine auscultation of hearts. The real problem is that we are just going to hear too many sounds. The literature makes it clear that we do not and will not know what they are. Given the situation of an uncertain physician confronted with an unrecognized sound, there arises the immediate potential for an unwarranted, aggressive manhunt for cardiac disease at the expense of patients' peace of mind and their wallets—all this without serious suggestion in the literature that there is any benefit to be derived from doing so.
Routine auscultation of the heart in asymptomatic patients leads only to a dreadful waste of time and resources and the creation of a morass of false-positive auscultation. What we need is a selection criterion that will limit our positive findings to a group of patients more likely to benefit. This simple selection criterion should be the presence of cardiovascular symptoms (however defined, including fatigue, dyspnoea, chest pain, syncope, etc.). When our careful histories have determined that any one of these is present to a significant degree, then we should listen to the heart. Those patients who have cardiac sounds beyond our normal experience should undergo further investigation, probably most cost-efficiently with an echocardiogram. Conserving our time in this way will pay ample dividends: physicians will have more time to devote to more important issues, among them better histories; physicians will spend less time confused and troubled by the sounds they hear; patients will spend less time confused and troubled by the sounds their physicians hear; and patients will end up with more time and money to enjoy life. Now isn't that worth thinking about?
No more routine cardiac auscultation of asymptomatic patients. Be not depressed if your cardiac auscultatory skills are not that sharp—no one else's are either. Rejoice that it does not seem to matter. Get back to real medicine. Pay attention to histories and symptoms. Before heeding anyone's exhortation to pay more attention to murmurs in asymptomatic patients, ask them for some data showing that it will improve your patients' outcomes.
And Horton should go on listening for Who's, not murmurs.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. B. Mahnke, A. Nowalk, D. Hofkosh, J. R. Zuberbuhler, and Y. M. Law Comparison of Two Educational Interventions on Pediatric Resident Auscultation Skills Pediatrics, May 1, 2004; 113(5): 1331 - 1335. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||