Utility of echocardiography in patients with suspected mitral valve prolapse

CLINICAL STUDIES

Utility of Echocardiography in Patients with Suspected Mitral Valve Prolapse WARREN Y. HERSHMAN, M.D., M.P.H., MARK A. MOSKOWITZ, M.D., KEITH I. MARTON, M.D., GARY J. BALADY, M.D. Boston, Massachusetts

PURPOSF~ Echocardiography has become a widely u ~ test since its introduction into clinical medicine in the early 1970s. Although it has frequently been p e r f o r m e d i n patients suspected of having n f i tral valve prolapse (MVP), its usefulness in this setting has not been systematically studied. To investigate the use and value of echocardiography in patients suspected of having MVP, we conducted a prospective study in which physicians were interviewed before and after ordering echocardiographic testing for patients in whom there was a suspicion of MVP. PATIENTS AND MEI~ODS: The study population i n cluded consecutive patients referred to the echocardiography laboratory at Boston University Medical Center because of suspected MVP between J a n u a r y 1 and December 31, 1987. Two standardized telephone interviews were conducted with the physician most responsible for ordering the echocardiogrnm_ The following information was obtained during the first interview, which was always conducted before the echocardiogram was performed: patient demographic and clinical d a t a ; the reason for ordering the echocardiogram; the physician's most likely clinical diagnosis; the physician's estimate of the likelihood that the patient had MVP; and the physician's proposed manAgement plans. After the referring physician received the echocardiographic results, a second interview was conducted to determine chgnges in the most likely clinical diagnosis and management plstns~ The impact of the echocardiogram on diagnosis and management was evaluated by comparing physician responses before and after reception of echocardiographic results. Receiver operating characteristic (ROC) curves were constructed to assess the physician's skills at distinguishing patients with echocardiographic-documented MVP from those without MVP. RESULTS:A total of 106 echocardiograms were ordered by 45 different physicians. More than 80% of all echocardiogrAms were ordered to address diagnostic or therapeutic concerns. On echocardiography, 47 (44%) patients were found to have MVP, six ( 6 % ) had mitral regurgitation without proFrom the Sections of General Internal Medicine and Cardiology, Evans Memorial Department of Clinical Research;Department of Medicine, The University Hospital, Boston University MedicalCenter; and the Department of Medicine, New England Deaconess Hospital, Boston, Massachusetts. This work was supported, in part, by Grant I F32 HLO7779-OIBt-5 (AHR-S) from the National Institutes of Health. Requests for reprints should be addressed to Warren Y. Hershman, M.D., M.P.H., General Internal Medicine Section, University Hospital, 720 Harrison Avenue, Suite t]08, Boston, MassachusettsO2118. Manuscript submitted December8, 1988, and accepted in revised form March 22, 1989. Dr. Marton's current addressis Pacific Presbyterian Hospital, San Francisco, California.

lapse, and 53 (50%) had normal results. On the basis of the ROC curve analysis, the physician's ability to d i ~ r l m l n n t , e between patients with and without echocardiographlc MVP varied significantly by physician specialty and practice setting. The echocardiographic results led to a change in diagnosis in 59 (56%) patients. A change in m a n . agement occurred in 29 (27%) patients, with 25 of these 29 changes (86%) related to the initiation or discontinuation of antibiotics. CONCLUSIONS:Echocardiography frequently alters diagnostic assessments and leads to therapeutic changes in some patients suspected of having MVP. However, the benefits of such changes have not yet been demonstrated. n recent years, we have witnessed an explosion in Isubsequent the introduction of new diagnostic technology. The rapid growth in the physician's armamentarium has raised concerns regarding the cost implications of these new modaiities and the settings in which they can be most effectively applied. A number of studies have documented that physicians collect and interpret medical data with great variability [1,2], and that an accurate diagnostic test may have a less than optimal clinical impact ff not employed appropriately

[3-5].

Since its introduction into clinical practice in the early 1970s, echocardiography has become a widely utilized test (more than 500,000 echocardiograms were ordered in 1986 on in-patients in non-federal hospitals) of moderate expense that is frequently performed in patients with suspected mitral valve prolapse (MVP) [6]. Nevertheless, the clinical utility of the echocardiogram has not been systematically evaluated in this setting. Previous investigations of this test have found that the echocardiogram often merely comrLrms the physician's clinical diagnosis [7,8]. Goldman et al [3] found echocardiography to have a limited impact on management decisions. However, these studies evaluated the use of echocardiography for multiple indications and contained few referrals for suspected MVP. On the basis of echocardiographic criteria, MVP is the most common cardiac valvular abnormality in the United States, with an estimated prevalence of 5% [9]. Detection has been considered worthwhile because of its reported association with a number of symptoms [10,11] and complications [12-14]. Echocardiography may be particularly useful if it can provide prognostic information [15] or ff it can help guide management decisions such as the use of antibiotic prophylaxis and beta-blocker therapy [16,17]. To investigate the use and value of echocardiography in patients suspected of having MVP, we have employed a prospective "before-after" design [18] to

October 1989 The American Journal of Medicine Volume 87

371

UTILITY OF ECHOCARDIOGRAPHVIN SUSPECTEDMVP / HERSHMAN ET AL

address the following primary questions: Why do physicians order an echocardiogram when MVP is suspected? How frequently and in what ways does echocardiography impact on diagnostic thinking and management plans? Does the use and yield of echocardiography vary by physician specialty and clinical setting? PATIENTS AND METHODS Patient Eligibility

Eligible patients included all those patients who were referred to the echocardiography laboratory at Boston University Medical Center because of suspected M V P between January 1 and December 31, 1987. This echocardiography laboratory serves both academic clinicians at the University Hospital and private practitioners in the surrounding community. Inclusion criteria required that the physician's suspicion of MVP be expressed in one of two ways. The diagnosis could be indicated explicitly by one of the following designations on the echocardiography request form: MVP, floppy valve, Barlow's syndrome, non-ejection click, click-murmur, or late systolic apical murmur. Additionally, a patient was also included if he or she had one of the following indications: chest pain, palpitations, dizziness, anxiety, fatigue, systolic murmur, or mitral regurgitation, and, if MVP was either the "most likely diagnosis" or the diagnosis the physician sought "to exclude" by ordering echocardiography. Patients who were less than 18 years of age or who had a previous echocardiographic-documented diagnosis of MVP were excluded. Data Collection Data were collected through an interview format that was pilot tested on 15 referring physicians to streamline the process and minimize interviewer bias. Subsequently, two standardized telephone interviews were conducted by one investigator (W.H.) with the physician most responsible for ordering the echocardiogram. During the firstinterview, always conducted before echocardiography was performed, the following information was obtained: patient demographic and clinical data; the physician's reason for ordering echocardiography; the physician's most likely clinicaldiagnosis; the physician's estimate of the likelihood that the patient had M V P ; and the physician's proposed management plans. After the referring physician received the echocardiographic results, a second interview was conducted to determine changes in the most likely clinicaldiagnosis and management plans. Physicians were informed of echocardiographic results within two weeks of the procedure and the post-echocardiographic interview was completed within the following seven-day period. Data Analysis

The impact of the echocardiogram was evaluated by comparing physician responses before and after receiving echocardiographic results.Physician estimates of the likelihood of M V P before the performance of echocardiography were given on a scale of 0 (not present) to 10 (certain). The diagnostic impact of the echocardiogram was examined by comparing the "most likely diagnosis" given before and after the procedure was performed. A change in diagnosis was attributed to echocardiography when the test provided 372

information that either directly led to a new postechocardiographic diagnosis or when it excluded the pre-echocardiographic "most likely diagnosis." Management plans before and after the echocardiogram were evaluated in a similar manner. Comparisons using frequency data were analyzed using the chi-square statistic. To assess the physician's skills at distinguishing patients with echocardiographic-documented MVP from those without MVP, we aggregated patients according to physician estimates of the pre-test probability of MVP. For each cut-off point (e.g., probability estimate), we computed sensitivity and specificity. Using these calculations, we constructed receiver operating characteristic (ROC) curves for each physician group [19,20]. The area under the ROC curve, a measure of the physician's ability to discriminate between the presence and absence of MVP independent of disease prevalence or cut-off criteria, was estimated using the method of Hanley and McNeil [21]. For several analyses, physicians were dichotomized along two axes to compare cardiologists versus noncardiologists and academic clinicians versus private practitioners. Because each physician was classified along both axes, he or she is a member of two groups (e.g., some cardiologists are academicians, whereas others are private practitioners). Accordingly, comparisons across axes (e.g., cardiologist versus academicians) were not performed.

Echocardlography Patients were studied in the left lateral decubitus position using a Hewlett-Packard Model 77020 (Andover, Massachusetts)phased array sector scanner. Mmode, two-dimensional, and Doppler-flow studies were obtained in the standard fashion [22] and were recorded on half-inch VHS videotape. All echocardiograms were of diagnostic quality and each study was assessed for the presence of abnormalities. Each study was interpreted by one of three independent, experienced echocardiographers who emp l o y e d u n i f o r m d i a g n o s t i c c r i t e r i a for M V P . Specifically, MVP was said to be present if there were superior displacement of the mitral valve leaflets above the annular hinge points in at least the parasternal long-axis view, or both the parasternal long-axis and apical four-chamber views [23,24]. The mitral valve was also evaluated for evidence of leaflet redundancy, which was said to be present if the leaflet appeared thickened, with prominent floppiness and infolding n o t e d p a r t i c u l a r l y on the p a r a s t e r n a l short-axis view [25]. Estimates of mitral regurgitation were graded on a scale of 1+ (mild) to 4+ (severe), based on the depth of the systolic regurgitant jet recorded in the left atrium using pulsed-Doppler mapping [26]. The atrium was divided into fourths, with 1+ being limited to the area just above the mitral valve and 4+ meaning that a regurgitant jet extended into the most superior quarter of the left atrium furthest from the mitral valve. RESULTS

Eligibility criteria were fulfilled in 133 patients, of whom 23 did not keep their appointment at the echocardiography laboratory (no shows). Four additional patients could not be included because a single referring physician declined to participate in the study.

October1989 The American Journal of Medicine Volume 87

UTILITY OF ECHOCARDIOGRAPHYIN SUSPECTED MVP / HERSHMAN ET AL

The analysis is therefore based on 106 consecutive echocardiograms in which physician interviews were completed before and after echocardiography. These 106 echocardiograms were requested by 45 different physicians whose median age was 50 years and ranged from 26 to 68 years. House officers or fellows ordered 10% of the echocardiograms, whereas academic clinicians and private practitioners ordered 30% and 60% of echocardiograms, respectively. Table I displays the physician breakdown by specialty. No individual physician accounted for greater than 10% of the echocardiograms ordered. Among the 106 patients referred, 76% were female and 95% were out-patients. The median age was 35 years, with 75% of all patients studied being less than 44 years of age. Of these patients, 30% had an abnormal physical examination (click or systolic murmur, or both) but were asymptomatic, 27% had symptoms but a normal cardiac examination, while 43% exhibited a combination of physical findings and symptoms. When assessing physical findings alone, we found 32% of patients had a systolic murmur and click, 26% had a murmur only, 15% had a click only, and 27% had normal examination results. Chest pain (42%) and palpitations (36%) were the most frequently cited symptoms. Three patients were suspected of having a habitus suggestive of Marfan's syndrome. Although physicians often have more than one concern in mind when requesting an echocardiogram, the most important reasons for ordering this procedure are displayed in Table II. More than 80% of all echocardiograms were ordered to address diagnostic or therapeutic concerns. On echocardiography, 44% of patients were found to have MVP, 6% had mitral regurgitation without prolapse, and 50% had normal examination results. Among the 47 patients with an echocardiogram demonstrating MVP, 25 had isolated prolapse, whereas four had associated mitral regurgitation, 12 had leaflet redundancy, and six had both redundancy and regurgitation in addition to the prolapse. Although 52 of 53 patients with echocardiographic

"~

.6

CL

.4

i ~-

TABLE I Characteristics of Physicians Ordering Echocardiograms

Specialty Internists Cardiologists Othersubspecialists

29 (64) 8 (18) 8 (18)

77 (73) 16 (15) 13 (12)

Total

45 (100)

106 (100)

TABLE II Physicians' Most Important Reasons for Ordering the Echocardiograms

cians' ability to discriminate between MVP and other diagnoses.

Number of Physicians

Percentof Physicians

63 22

59 21

13

12

2

2

To improvediagnosticcertainty To decideon treatment To reassurethe patient To estimateprognosis

6

Other Total

106

6

100

findings of MVP or pure mitral regurgitation were given a matching post-test clinical diagnosis by their physicians (one patient with mild MVP diagnosed by echocardiography was recorded as having a normal heart), patients with normal echocardiographic results were given a variety of clinical designations. Among these 53 patients with normal study results, 29 were labeled "normal." The remaining 24 were diagnosed with the following conditions: anxiety (14), angina (three), vestibular disease (two), musculoskeletal strain (two), supraventricular tachycardia (two), and pleurisy (one). The ROC curves are shown in Figure 1. The ROC areas and standard errors are 0.864 (+ 0.074) for cardi-

i

"" /

.2q

/

/.,'~'.~ /,/~

O---OCardiologist ~Noncardiologist

em-ician-

i

Figure 1. ROCcurves that depict physi-

Echocardlograms Ordered(%)

~s S~''*

[i ~

0

Physicians Participating (%)

~,--APrivate I .2

I .4

I .6

I .8

I 1

False Positive Rate October 1989 The American Journal of Medicine Volume 87

373

UTILITY OF ECHOCARDIOGRAPHYIN SUSPECTEDMVP / HERSHMAN ET AL TABLE III Overall Impact of Echocardlogram Numberof Percentage of Echocardlograms Echocardlograms No changein diagnosisor management Changein diagnosisonly Changein managementonly Changein diagnosisand management

Total

35

33

42

40

17

16

12

106

11

100

TABLE IV Impact of the Echecardlegram on Management Numberof Percentageof Echocardlograms Echccardlcgrams Changein management Instituteantibiotic prophylaxis Discontinueantibiotic prophylaxis Obtaincardiacconsultation Institutebeta-blockertherapy Cardiaccatheterization No change

29 18

27

7 2

l

1 74

73

ologists versus 0.539 (± 0.066) for non-cardiologists and 0.821 (~- 0.079) for academic clinicians versus 0.422 (± 0.092) for private practitioners. A perfect test would have an area of 1.0. These findings suggest that cardiologists and academic clinicians demonstrated significantly greater skill at discriminating between patients with and without echocardiographic-documented MVP, than did non-cardiologists and private practitioners, respectively. Nevertheless, even among the cardiologists and academic groups, high sensitivities (greater than 85%) were associated with falsepositive rates ranging from 35% to 40%. The impact of the echocardiogram on diagnosis and management is depicted in Table III. A change in the most likely diagnosis occurred in a total of 59 patients (56%). This change in diagnosis was more likely to occur among private physicians (65%) versus academic clinicians (34%, p --- 0.004) and among non-cardiologists (63%) versus cardiologists (12%, p <0.001). A change in management (Table IV) occurred in 29 (27%) patients, with 86% of these changes related to the initiation or discontinuation of antibiotic prophylaxis. The cardiac catheterization involved a patient with severe mitral regurgitation associated with MVP who ultimately required surgery. The likelihood that a change in management would occur was not associated with physician specialty, practice setting, or clinical variables.

tional "before~after" study. This has allowed us to describe echocardiographic usage and to assess more proximal outcomes such as the diagnostic and management impact of the procedure. Diagnostic Impact

The ability of physicians to discriminate between patients with and without echocardiographic MVP based on clinical assessment was limited and varied significantly by physician specialty and practice setting. Cardiologists and academic clinicians appeared to exhibit greater skill at this task than non-cardiologists and private practitioners, respectively. However, because we lack an accurate estimate of the percentage of all patients clinically suspected of having MVP who were referred for echocardiography, we should interpret these data with caution. Some of the difference between physician groups may reflect different thresholds for ordering the test or patient selection bias (referral versus primary care patient base) and therefore may not be a pure index of discriminatory skills. Given the limited and varying discriminating power of clinical assessment, it is not surprising that we found echocardiography to frequently (56%) lead to a change in the physician's most likely diagnosis. As expected, these changes were more common among physicians who were less able to predict MVP based on clinical evaluation (private practitioners, non-cardiologists). Although our results support the observations of McDonald et al [27], that the echocardiogram exerts greater influence on diagnosis than management, the type and magnitude of diagnostic impact differ. We found a greater impact on changes in "most likely diagnosis," whereas MacDonald's group found frequent changes in the physician's probability estimate of or confidence in a given diagnosis before and after echocardiography. This result likely reflects two important differences in these studies. First, whereas McDonald et al [27] examined multiple indications for echocardiography, our study focused solely on patients suspected of having 1VFVPwhere the task is frequently to separate patients who have MVP from those who have a normal heart. Second, whereas their study evaluated the use of the echocardiogram by cardiologists, only 15% of echocardiograms ordered in our study were requested by this specialty group. These differences clearly highlight the need to account for both clinical indications and physician characteristics when assessing the impact of a diagnostic test.

Impact on Management

In a substantial number of patients (27%), the echocardiogram was associated with a change in management. Our finding that most changes were associated with the use of antibiotic prophylaxis and not with the use of beta-blocker therapy or the ordering of other tests is consistent with the report of Retchin et al [28]. COMMENTS Our estimate of the magnitude of management impact The ideal approach to assess the usefulness of echo- is greater than that found by Goldman et al [3], who cardiography in patients suspected of having mitral employed cardiology reviewers. valve prolapse would be to perform a randomized trial ~ Although we have documented m a n a g e m e n t (of echocardiography versus no echocardiography) changes associated with echocardiography, we have and measure patient health outcomes. However, given not addressed the appropriateness and value of such the overriding ethical and logistical problems in pur- interventions. The cardiac consultation and cathetersuing such a design, we have conducted an observaization changes were made for patients with angina or 374

October 1989 The American Journal of Medicine Volume 87

UTILITY OF ECHOCARDIOGRAPHYIN SUSPECTEDMVP / HERSHMANET AL

significant regurgitation and appeared quite reasonable and justifiable. However, the institution of betablocker therapy should, arguably, be made on the basis of symptoms or cardiac rhythm abnormalities, or both, rather than echocardiographic criteria. In addition, the routine use of echocardiography to guide antibiotic prophylaxis is arguable. The American Heart Association recommends that prophylaxis for patients with MVP be given to those with mitral regurgitation [17]. Data supporting this statement are based on auscultatory, and not Doppler, findings of mitral regurgitation. It remains to be determined whether antibiotic prophylaxis should also be recommended when the Doppler echocardiogram demonstrates mitral regurgitation but auscultation falls to reveal a systolic murmur. Furthermore, it has been reported that the attributable risk of endocarditis in patients with MVP is actually very low [29,30], and prophylaxis may not be cost-effective [31,32].

problem in using the echocardiogram as a diagnostic "gold standard" for MVP. This investigation was designed to assess the diagnostic and management impact of echocardiography, not its validity and reliability. T h o u g h i n t r a - o b s e r v e r and inter-observer variability among echocardiographers was not measured, this should not affect the primary findings of this study--the impact of echocardiography on diagnostic thinking and management plans. Nevertheless, to date there are no uniformly accepted criteria, although several have been proposed. Although some criteria have demonstrated good intra-observer and inter-observer reliability [35], the natural history and clinical significance of various findings based on physical examination and echocardiography need further delineation. For many echocardiographic criteria of MVP, the prevalence in the general population is too high and the risk of significant complications appears too low to justify the label of abnormal or disease [23,24,36-39]. Other Clinical Contributions In conclusion, we have described the use and impact Beyond the issues of diagnosis and management are of echocardiography in patients suspected of having two other potential impacts of the echocardiogram. MVP. Although the test frequently alters diagnostic The test may, in some instances, alter a physician's assessments and leads to therapeutic changes in some probability assessment of and confidence in a particu- patients, the benefits of such changes have not yet lar diagnosis without leading to a change in the diag- been demonstrated. nostic category. Although we did not directly measure this endpoint, our t'mding that the echocardiogram led A C K N O W L E D G M E N T to a change in management without a change in diag- We are grateful to Mary Jane Birmingham and GiuseppeSarno for technical assisnosis in 11% of patients supports this hypothesis. This tance, to Diane Lobel for coordinating data collection, and to Leanne Gitell for leads to the question: How much information and manuscript preparation. what level of confidence are appropriate in clinical REFERENCES decision-making? A second potential use of the echo- 1. Komaroff AL: Variability in medical practice. J Gen Intern Med 1986; 1: 531cardiogram is to provide patient reassurance. In our 534. study, physicians ordered echocardiography for this 2. Schroeder SA, Kenders K, Cooper JK, Remme rE: Use of laboratory tests and primary purpose in 12% of patients. Although we do pharmaceuticals. Variation among physicians and effect of cost audit on subseuse. JAMA 1973; 225: 969-973. not have patient responses, physicians believed that quent 3. Goldman L, Cohn PF, Mudge GH, etal:Clinical utility and management impact of patient reassurance was a major benefit from the echo- M-mode echocardiography. Am J Meal 1983; 75: 49-56. cardiogram in a similar percentage of patients. How- 4. Marton KI, Sox HC, Wasson J, Duisenberg CE: The clinical value of the upper ever, it has been shown that patients suspected of gastrointestinal tract roentganogram series. Arch Intern Med 1980; 140: 191having MVP are likely to develop significant function- 5.195. Fineborg HV, Bauman R, Sosman M: Computerized cranial tomography: effect al disability, independent of the echocardiographic on diagnostic and therapeutic plans. JAMA ]977; 238: 224-227. findings [33]. Clearly, patient reassurance is an impor- 6. National Center for Health Statistics, Graham D: Detailed diagnoses and procetant and a major function of the physician, and diag- dures h-om short-stay hospitals, United States, 1986. Vital and Health Statistics. Series 13, No. 95. DHHS Pub. no (PHS) 88-1756. Public Health Service. Washingnostic tests can be useful in achieving this end [34]. ton: U.S. Government Printing Office, 1988. Nevertheless, it is possible that appropriate education 7. Markiewicz W, Puled B, Hammerman H, Greif Z, Hir J, RissE: Contribution of Mand counseling could minimize the need to order the mode echocardiography to cardiac diagnosis. An assessment in 1,000 successive patients. Am J Med 1978; 65: 803-807. echocardiogram solely for this purpose. Limitations

Despite our findings, several questions remain. Discrepancies may exist between stated plans and clinical actions. Because most interventions involved antibiotic prophylaxis that was usually performed before dental procedures, the medical chart was not a helpful source to confirm such actions. Also, physicians ordering a particular test are likely to believe in its efficacy, and subconscious bias may influence management plans prior to obtaining echocardiographic results. Furthermore, diagnostic and therapeutic decisions are often based on multiple sources of data, and, therefore, it may be difficult to attribute subsequent management changes to a single test. These two biases are inherent to the study design and difficult to measure or control; these biases tend to favor the test being evaluated (echocardiography) and lead to overestimates of impact. Most importantly, there is a major

8. Grimmer SFM, Sndall H, Hill JD: Diagnostic contribution of echocardiography. Lancet 1982; I: 440-441. 9. SavageDD, Garrison PJ, Devereaux RB, etal:Mitral valve prolapse in thegeneral population. I. Epidemiologic features: The Framingham Study. Am Heart J 1983; 106: 57]-576. 10. Barlow JB, Pocock WA: The problem of nonejection systolic clicks and associated mitral systolic murmurs: emphasis on the billowing mitral leaflet syndrome. Am Heart J 1975; 90: 636-655. 11. Markiewicz W, Stoner J, London E, Hunt SA, Popp RL: Mitralvalveprolapsein one hundred presumably healthy young females. Circulation 1976; 53: 464-473. 12. Mills P, Rose J, Hollingsworth J, Amara I, Craige E: Long-term prognosis of mitral valve prolapse. N Engl J Med 1977; 297: 13-18. 13. Clemens JD, Horowitz RI, Jaffee CC, Feinstein AR, Stanton BF: A controlled evaluation of the risk of endocarditis in persons with mib'al valve prolapse. N EngJJ Med 1981; 307:776-781. 14. McKinsey DS, Ratts TE, Bisno AL: Underlying cardiac lesions in adults with infective endocarditis. Am J Med 1987; 82: 681-688. 15. Nishimura RA, McGoan MD, Shub C, Miller FA, Ilstrup DM; Tajik AJ: Echocardiographically documented mitral valve prolapse. Long-term follow-up of 237 patients. N Engl J Med 1985; 313: 1305-1309. 16. Jeresaty RM: Mitral valve prolapse: an update. JAMA 1985; 254: 793-795. 17. American Heart Association Committee on Prevention of Bacterial Endocarditis: Prevention of bacterial endocarditis. Circulation 1984; 70: 1123A-1127A. 18. Guyatt GH, Tugwell PX, Feeny DH, Drummond MF, Haynes RB: The role of

October 1989 The American Journal of Medicine Volume 87

375

UTILITY OF ECHOCARDIOGRAPHYIN SUSPECTEDMVP / HERSHMAN ET AL before-after studies of therapeutic impact in the evaluation of diagnostic technologies. J Chronic Dis 1986; 39: 295-304. 19. Berwick DM, Thibideau LA: Receiver operating characteristic analysis of diagnostic skill. Med Care 1983; 21: 876-885. 20. Weinstein MC, FinebergHV, ElsteinAS: Clinical decision analysis. Philadelphia: WB Saunders and Company, 1984; 113. 21. Hanley JA, McNeil BJ: The meaning and use of the area under a receiver operating characteristic curve. Radiology 1982; 143: 29-36. ~..The echocardiographic examination. In: Feigenbaum H, ed. Echocardiography. Philadelphia: Lea and Febiger, 1986; 50-127. 23. Levine RA, Triu~ MO, Harrisan P, Weyman AE: The relationship of mitral annular shape to the diagnosis of MVP. Circulation 1987; 75: 756-767. 24. Levine RA, Stathogiannis E, Newell JB, Harrigan P, Weyman AE: Reconsideration of echocardiographic standards for MVP: lack of association between leaflet displacement isolated to the apical four chamber view and independent echocardiographic evidence of abnormality. J Am Coil Cardiol 1988; 11: 1010-]019. 25. WeymanAE: Cross-sectionalechocardiography. Philadelphia:Lea and Febiger, 1982; 173. 29. Adhar GC, AbbasiAS, NandaNC: Dopplerechocardiography in the assessment of mitral regurgitation and mitral valve prolapse. In: Nanda NC, ed. Doppler echocardiography. New York: Ignku-Shoin. 1985; 188-210. 27. McDonald IG, Guyatt GH, Gutman JM, Jelinek M, Fox P, Daly J: The contribution of a non-invasive test to clinical care: the impact of echocardiography on diagnosis, management and patient anxiety. J Clin Epidemio11988;41: 151-161. 28. Retchin SM, Waush RA, Fletcher RH: The impact of echocardiog~'aphicresults on treatment decisions for patients suspected of mitral valve prolapse. Mad Care

376

October 1989 The American Journal of Medicine Volume 87

1987; 25: 652-657. 29. Retchin SM, Fletcher RH, Waugh RA: Endocarditis in mitral valve prolapse: what do we mean by "risk"? Int J Cardiol 1984; 5: 653-659° 30. MacMahon SW, Roberts JK, Kramer-Fox R, Zucker DM, Roberts RB, Devereaux RB: Mitral valve prolapse and infective endocarditis. Am Heart J 1987; 113: 1291-1297. 31. Bor DH, Himmelstein DU: Endocarditis prophylaxis for patients with mitral valve prolapse. A quantitative analysis. Am J Meal 1984; 76: 711-717. 32. Clemens JD, Ransohoff DF: A quantitativeassessment of pre-dental antibiotic prophylaxis for patientswith mitralvalve prolapse. J Chron Dis 1984; 37: 531-544. 33. Retchin SM, Retcher RH, EarpJA, Lamson N, Waugh PA: Mitra$valve prolapse: disease or illness? Arch Intern Meal 1986; 146: 1081-1084. 34. Sox HC, Marsulies I, Sox CH: Psychologically mediated effects of diagnostic tests. Ann Intern Med 1981; 95: 680-685. 35. Alpert MA. Carney RJ, Munuswamy K, et al: Observer variation in the echocardiographic diagnosis of mitral valve prolapse. Am Heart J 1986; 111: 1123-1129. 36. Savage DD, Devereux RB, Garrison RJ, et al: Mitral valve prolapse in the general population. 2. Clinical features; The Framingham Study. Am Heart J 1983; 108: 566-581. 37. Devereux RB, Kramer-Fox R, Brown WT, et al: Relation between clinical features of the mitral valve prolapse syndrome and echocardiographically documented mitral valve prolapse. J Am Coil Cardiol 1986; 8: 763-772. 38. Retchin SM, Retcher RH, Earp J, Lamson N, Waush RA: Mitral valve prolapse: disease or illness. Arch Intern Mad 1986; 146: 1081-1084. 39. Perloff JK: Evolvingconcepts of mitral valve prolapse. N EngiJ Mad 1982; 307: 369-370.