Indications for echocardiography in the diagnosis of infective endocarditis in children

Indications for Echocardiography in the Dia nosis of Infective Endoca ril itis in Children Craig

A. Sable, MD, Jonathan J. Rome, MD, Gerard R. Martin, Kantilal M. Patel, PhD, and Sharon S. Karr, MD

MD,

The role of transthomcic echocardiography as a diagnostic tool in children suspected of having infective endocarditis (IE) has not been defined. We hypothesized is only useful in children in that echocardiograph whom there is high cr inical suspicion of IE based on hysical examination findings or persistentty positive II lood cultures. Echocardiographic reports and medical records of all inpatients (n = 133) from 1990 to 1992 who underwent echocardiography for suspected IE were reviewed. Fifty-nine of the 133 patients (44%) identified had either persistently positive blood cultures (n = 48), physical examination findings of IE (n = 20), or both (n = 9). The echocardiogram was positive in 7 of

these patients (12%) and negative in all 74 patients without positive physical findings or positive blood cultures = 0.003). A new or changing precordial murmur, embo G” ic phenomena, congestive heart failure, mechanical ventilation, and sitive blood cultures were prefor IE b univaridictive of positive ec r ocordiograms ate analysis. The presence of fever, immune creficiency, and central lines, alone or in combination, was not presitive echocardiogram. In the absence of dictive of a physical fin r ings or persistently positive blood cultures, echocardiography is a low-yield study and is unlikely to aid in the diagnosis of IE in children. (Am J Cardiol 1995;75:801-804)

I

METHODS

nfective endocarditis (IE) is relatively rare in childhood, with a prevalence ranging from 1 in 1,280 to 1 in 4,500 hospital admissions.‘s2 Because of its infrequent occurrence, TE has generally been studied in a retrospcctivc fashion.“.4 In recent years, 2-dimensional echocardiography and its role in the diagnosis of IE has been the subject of much investigation.5*” While initial reports7. ‘I focused on the echocardiographic features of IE, more recent studies have evaluated the sensitivity of 2-dimensional transthoracic cchocardiography in adults (42% to 76%)12 -I6 and children (52% to 70%)4.17 with known IE. Based on these studies, transthoracic echocardiography has been used with increasing frequency as a diagnostic tool in evaluating children for IE. The predictive value of a diagnostic test is dependent not only on the sensitivity and specificity of the test, but also on the prevalence of the disease in the population being studied. Therefore, it is essential to know prevalence data to determine the clinical utility of a diagnostic study.@ Such prevalence data are unknown for the population of children currently subjected to echocardiographic evaluation for IE. To evaluate the role of echoc‘ardioeraphy, we reviewed the charts of all patients at our institution who had transthoracic echocardiograms as part of a diagnostic evaluation for 1E to characterize the clinical fcatures and prevalence of IE in our population. We analyzed data to determine if we could identify clinical risk factors that would guide the appropriate use of echocardiography.

From the Departments of Cardiokxy and S*at.stlcs and Research, Childrer’s Nationcl Medicat Cerucr Wakinntaq D C Manuscr ct leccived Octooe: 7, 1994; revised manuscript received cnc occeatedjaruary ‘2, 1995 Address for repnts: Sharer; S. Kerr, k!C. Dcpcrtmer? cf Cardiolcgy, Chldrer;‘s National Medical Certel, I I 1 iV&gan Aven.~e V\w, Washington, D.C 200 IO

‘/.A!VUL4t? rliART

ClSiASE/iCI

The computer data base of the echocardiography laboratory of the Department of Cardiology from April 1990 to December 1992 was searched for patients evaluated for IE. The hospital medical record computer and pathology records were also searched to ensure that no cases were missed. The total number of hospital admissions during this time was obtained. All echocardiograms were read by an experienced pediatric cardiologist. An echocardiogram was considered positive for IE if a vegetation was seen on a valve or in 1 of the cardiac chambers. Valve thickening alone was not sufficient to diagnose IE.‘” All echocardiogrdphic reports were reviewed. A study was excluded if it was (1) performed at another institution whose medical records were not available to us, (2) a follow-up on a patient previously diagnosed with IE, or (3) a repeat study of the same patient within 6 weeks of the initial study. The medical record for hospital admission in which the echocardiogram was recorded was reviewed for each patient included in the study. Additional patients were excluded from the final analysis if their medical records were not available or if upon reviewing their records it was determined that the echoCxdiOgmIn was not obtained primarily to diagnose IE. The following data were obtained from the medical record: Historical data-age, sex, congenital heart disease, acquired heart disease, history of cardiac surgery, a recent invasive procedure, cyanosis, prematurity (corrected age
IOCAKDIOGRAP~

“Y N

“ELM-RIG

‘NFCC’IVc

E~\~IXICARD!?!S

801

TAME I Historical

Data and the Association Number

Variable

TABLE II Physical groms

Present

2/15 4/29 t/38

(13) (14) (31

5/50 o/3

(10) (01

I/38 (8)

Examination

Variables

Variable

(%) with Positive Present

4/73 O/l 1 3/7 2/6 3/13 o/4

c2 kg l/l1

Echocardiograms Variable

and the Association

Variable

Fever kash New or changing murmur Embolic phenomena Congestive heart failure Focal neurologic findings

Echocardiograms

19)

Absent

p Value

(4) (3) (6) (41 (2) (5) (4)

0.179 0.040 NS NS 0.103 NS NS

5/l 18 3/104 6/95 2/48 2/83 7/130 4/95

5/85 16)

Number

Weight

Positive

(%) with Positive

Variable

Premature Ventilator Immune deficiency Central line Intensive care Acquired heart disease Congenital heart disease

With

With

Echocardip

1

Echocardiograms Variable

Absent

p Value

(5) (6) (3) (4) (3) (5)

NS NS 0.003 0.032 0.021 NS

3/60 7/122 4/126 5/l 27 4/120 7/129

(5) (01 (43) (33) (23) IO1

Positive

2-10

kg

210

kg

4/61

(7)

2/61

(3)

NS

Between April 1990 and December 1992, 10 patients met diagnostic criteria for IE based on criteria defined by the Mayo Clinic22 out of a total 31,836 hospital admissions (1 in 3,184 admissions). During that time, 164 of a total 11,300 echocardiograms (1.5%) were recorded for the diagnosis of IE. Twenty-seven studies (16%) were excluded: 15 were performed at an outside institution whose medical records were not available; 9 were obtained as follow-up for patients whose previous echocardiograms were positive for IE; and 3 were repeat studies performed in patients who had a negative echocardiogram for IE within the prior 6 weeks. Of the remaining 137 studies in 127 patients, medical records were avail-

able for 123 patients, and these patients were included for analysis. The medical record computer and pathology searches did not detect any additional patients with a diagnosis of IE. Evaluatian: All patients had a physical examination documented near the time of request for echocardiography: however, a spleen examination was documented in ~10% of patients. Fewer than 3 blood cultures were obtained in 14% of the study population. Sedimentation rate was not obtained in 81% of patients and urinalysis was not obtained in 68% of patients; therefore these variables were excluded from further statistical analysis. Clinical course: ?f the 133 echocardiographic studies performed, 7 of 10 patients diagnosed with IE were positive for vegetations. Medical management alone was given to 4 of 7 patients’and combined medical/surgical management was given to 3 of 7 patients. There was 1 death in each of these groups. Two of the 3 patients with negative echocardiograms received medical treatment and 1 received combined medical/surgical treatment. All 3 patients survived. ” Univariate analysis: By univariate analysis, risk factors that were associated with a statistically significant proportion of positive echocardiographic studies were new or changing murmur (p = 0.003), embolic phenomena (p = 0.03), congestive heart failure (p = 0.02), positive blood cultures (p = 0.009), and mechanical ventilation (p = 0.04). Analysis of the following risk factors were not statistically significant: sex, age, prematurity, presence in an intensive care unit, and length of hospitalization. It was unlikely that the 41 patients with congenital (n = 38) or acquired (n = 3) heart disease would have a positive cchocardiogram if they had a recent invasive procedure, history of heart surgery, or cyanosis. The results of Fisher’s exact test for each of the independent variables are listed in Tdbles I through III.

802

At’?11 I5

failure, and focal findings on neurologic examination; laboratory test results-leukocytosis (leukocyte count >15,OOO/mm”), elevated band count (>lO’% bands), anemia (hemoglobin 2(1 mm/hour), and hematuria. The number of blood cultures obtained was also recorded. The patient’s final diagnosis at hospital discharge was noted. Statistical methods: Fisher’s exact test” was used to compare results of echocardiography in subgroups of patients by univariatc analysis. Patients were grouped by the presence or absence of 1 or >I risk factor. A multivariate linear logistic regression analysis2”*21was undertakeh to determine which variables had the greatest impact on the result of the echocardiogram. Odds ratios with 95% confidence intervals were calculated for each variable in the regression analysis. Age and weight were treated as continuous variables for this analysis. A significance level was defined as a probability of co.05 for all statistical analyses.

RESULTS

THk AMERICAN JOURNAL OF CARDIOLCGY”

VOL 75

I995

TABLE III

Laboratory

Data and the Association Number

Variable

TABLE IV Combinations Echocardiogroms

Present

Variable

Absent

l/21 O/31 6/48 2m

(0) (13) (7)

7/97 i/85 5/105

Voriables

Number

Present

6/20

(30)

7/59

(12)

3/47 t/26 l/22

16) (4) (5)

One or more of the following physical findings were present in 20 of the 133 patients: new or changing murmur, embolic phenomena, or congestive heart faihuc. Six of the 7 patients with positive echocardiograms were in this group (p
DISCUSSION Previous studies have shown that, once the diagnosis of IE is made, echocardiography plays an important role in the evaluation of complications such as abscess formation or vatve perforation.“.i4 Our results clearly show that echocardiography is only of value in the diagnosis of IE when there is a high level of clinical suspicion. Our retrospective analysis showed that every patient with IE had either a new or changing precordial murmur, embolic phenomena, congestive heart failure, or persistently positive blood cultures. This study found that only physical examination findings, mechanical ventilation, low weight, or persistent-

p Value NS NS NS NS 0.009 NS

(71 11) (5)

ond the Association

(“‘) with Positive

Variable

VAiVUlAi

Echocardiograms

5180 (6) 6/101 (6) 6/l 07 (6)

of Independent

New or changing murmur, embolic phenomeno, or congestive heort failure New or changing murmur, embolic phenomena, congestive heart failure, or positive blood culture Fever and central line Fever and immune deficiency Fever, central line, ond immune deficiency

Echocardiograms

(4) (4) (5)

ma 1127

Variable

Positive

(%) with Positive

Variable

Leukocytosis Elevoted band count Leukocytosis and elevoted band count Anemio Positive blood cultures Fungemia

With

With

Positive

Echocardiogroms Variable l/l

Absent 13 (1)

p Value
(01

0.003

4/86 (5) 6/l 07 (6) 6/l 11 (5)

NS NS NS

o/74

1

ly positive blood cultures increased the likelihood of having a positive echocardiogram. Patients without these risk factors should be considered low risk for IE. The performance of a diagnostic study, such as echocardiography, in a patient population with a low disease prevalence, despite a high specificity, will result in a low positive predictive value and otentially a significant number of falsepositive studies.’ f: In addition to unnecessary expense, false-positive studies could result in more diagnostic procedures, such as transesophageal echocardiography (performed to better delineate the lesion), unnecessary courses of prolonged antibiotics, or failure to search for the correct diagnosis. Patients with high-risk factors should be considered for echocardiography. If a transthoracic echocardiogram is negative in this group of patients, one should consider performing a transesophageal echocardiogram because the sensitivity is significantly higher (94% to 96%) in adult studies.*’ 2s No similar pediatric studies have been published. Three patients in our series with negative transthoracic studies underwent transcsophageal echocardiography and one had a vegetation. Detection of vegetations by transthoracic echocardiography is especially difficult in patients with repaired complex heart discase26; all 3 patients in our study with negative transthoracic echocardiography had undergone repair of congenital heart disease.These detinitions of high- and low-risk patients probably should not bc applied to the premature infant. There have been multiple published cast reports regarding the nonclassic presentation of IE in low birth weight and premature infants.*“*s In support of this, multivariate analysis found an association between positive echocardiograms and decreasing weight. We therefore would have a higher level of suspicion in this group of patients.

I ItART DISEAS~/ECHOCARI)IORA?I

I” I’d I’tDIATKIC INFECllVE CNDOCARDIPS

803

Study ~imitalions: Although the prevalence of JE in our hospital population is consistent with previously published figures,3,29 the few patients with positive echocardiograms is a potential limitation of this study. In addition, we may have selection bias by including patients whose primary purpose for echocardiography was other than the diagnosis of IE, or by excluding patients with IE. We tried to minimize selection bias by an extensive chart review to exclude patients not being evaluated for IE and by excluding follow-up studies in patients who had several studies within a short period of time for the same illness. We also performed separate searchesof the medical records computer and pathology records which confirmed that all patients diagnosed with JE were referred for echocardiography and were included in the study.

1. Van Hare GF, Ben-Shachar G, I.icbman I. Boxerbaum B. Riemenschneider TA. Infective endccarditis in infants and children during the past 10 years: a decade of change. Am Hem J 19X4:107: 123%1240. 2. Sheller GF. Hawker RE, Celerma~er JM. lnfcctivc cndocarditis in childhcmd. Pediorr Cardid I986;6: I X3- I X6. 3. Siegel JD. Bacterial endcarditis in infants and children: incidcncc and pathogenesis. Pedwrr lnfecr Dis J lY83:2:S41--.S44. 4. Geva T. I+and M. Infective endwarditis in children with congenital heart dicease: the changing spectrum, lY65-85. Eur Hrarr J 1988:9:124&1249. 5. Dillon T. .Meyer RA. Korfhagen JC, Kaplan S, Chung KJ. Management of infw. live endocarditis using echocardiography. J Pediatr lYX0,96:552-558. 6. Williams KG. Bicrman I-Z. L.\c of echocardiography in evaluarica of general pediatric disaw. Prdiaw Rev I98 l:3: 137- 146. 7. Johnson CM, Rhodes KH. Pediatric endocardilis. Mayo Clin Proc 1982:.57:8t%Y4. 8. Byrd RF. Shelton ME, Wdson BII. Schillig S. Infective pcrivalvar abscess of the sonic ring: cchwardiographic features and clinical course. Am J Cardiol 1990:66:102-105. 9. Martin KP. Mcltrer RS, Chia BL, Stinson EB. Kakowki H. Popp RI.. Clinical

utility of two dimensional cchocardiography in infective cndcarditls. r\m J Cnrdro/ lY80;46:379 -3X5. 10. Hricker JT. Istwn LA. Huhta JC. Gutgcwll IIP. Echwxrdiographic evaluation of endocarditis in children. C/in Prdiarr 19X.5:24:3 12-3 17. 11. Webb Kavey R, Frank DM. Hyrum CJ, Blackman MS, Sondheimer IIM, Bow EL. Two-dimensional echocardiographic assessment of infective endwarditis in children. Am J Db Child lY83; 137:851-X56. 12. Lutz EM, Roberts: RB. Dcvereux RB. F’riem LM. Relation between the prw encc of echocartliographic vegetauon? and the complication rate in infective endo. cardiris. &n Ilcurr J 1986;l 12:107-l 13. 13. Santilippo Al, Picard MH, Newell JB. Rows E, Davidoff R, Thomas JD, Weyman AE. Echocardiopraphic assessment of patients with infectious endwarditis: praliction of risk for complications. J Am Co/l Cardiol 1991:1X:1 191-l 199. 14. Stafford WJ, Petch I. Radford DJ. Vegetations in infective endocarditis: chnical relevance and diagnosis by cross sectional echocardiography. Hr Ilenrr J IYX5: 53:Rl&.313. 15. Jaffe WM. Morgan DE, Pearlman AS. Otto CM. Infective endcmrditir, 19X3. 1988: echocardiograph~c findmgr and factors intluencing morbidity and morwlity. J Am Co// Crrdiol IYYO; 15: 1227-1233. 16. Buda AJ. Zxz RJ. LeMire BS. Hach DS. Prognostic ~igniticancc of vegeladons detwted by two-dimensional echocardiography in infective endocarditis. Am //mrrJ 1986:112:12Y1-12%. 17. Parru F. Bowa 1. Romcro J, Huron I.. Quero b. Brim J, Vellibrc D. Infectious endwarditis in children. Prdiafr c‘ordiol 1990: 11:77-X1. 18. G‘arwn A Jr, Hricker JT. McNamara IX;. The Science and Practice of Pediatric Cardiology. Philadelphia: Lea & Febiger. 1990: 142-145. 19. Armitage P. Statistical Methods in .Medical Kesearch. \;cw York: John Wiley &Sons. lY71:13S-138. 20. SAS Insntutc. Inc. SAS/STAT Users Guide, Verwn 6. 4th ed. vol 2. Cay, NC: SAS Institute. IyyO:lO7l-I 126. 21. Ilosmer SW. Lencshow S. Applied Iagistic Rcgr&on. New York: John Wiley &Sons. lY89:141. 22. Van Re)n CF. Ixvy BS. Arbeit KU. Friedland G. Crumpackcr CS. Infcctwe endwarditis: an analysis based on snict case definitions. Ann lnrcrn Med 19X l:94: 505-51X. 23. Erhcl R. Rohmann S, Drexler M, Mohr-Kahaly S, Meyer J. Diagnostic value of uan~sophageal echocardiognphy in infective endcrarditis (ahstr). Ciwrrlorion lYX6:74~suppl II):II-55. 24. Shapiro SM. Bayer AS. ‘I‘ransesophageal and Doppler echocardiography in the diagnosis and management of infective cndocarditis. Chcsf 1991:100:112.-I 130. 25. Pedcrsen WR, Walker M, Olson JD. G&xl F. Lange HW. Daniel JA. Rogers I, Lange T. Kane 41. .Mconey MR. Goldenberg IF. Value of tranrcwphageal echocardiography as an adjunct to transthoracic echocardiography in evaluation of narive and prosthetic valve endcrarditis. Chew 199 I: loO:351-356. 26. Saiman L, Prince A, Gersony WM. Pediatric infective endocarditis in the modcm era. J Pediow IYY3; 122:817-X53. 27. Jackson W, Pinsky W. Neonatal cwdiac infections. In: Long WA. ed. Fetal and Neonatal Cardiology. Philadelphia: WB Saundas. 1990:571-577. 28. Awadallah S.4, Kavey RW. Byrum CJ. Smith FC, Kveseli, DA. Blackman MS. The changing panem of infective endwarditi:, in children. Am J Cardiol lYYl;68: 90 94. Infect Urz J I9Yw2; I I: 29. Baltimore RS. Infective endwarditis in children. Prdwtr %7-912.

804

A+?11 15, i 99.5

Acknowledgment We appreciate the critical review of the study by Thomas J. Hougen, MD. We thank Elaine Skurow, RDCS, RCT, Michele Volgy, CCVT, Laurie Chauvin, RCPT, Laura Hoye, RDCS, RDMS, and Alicia Bennett, RDCS, RDMS, for their expertise in recording the echocardiograms.

It 1E NV\ERICAN !OUKNAL OF CARDIO!OCSY*

VOI.. 75