- Email: [email protected]
Syncope in children: Diagnostic tests have a high cost and low yield
SYNCOPE IN CHILDREN: DIAGNOSTIC TESTS HAVE A HIGH COST AND LOW YIELD LEONARD A. STEINBERG, MD, AND TIMOTHY K. KNILANS, MD
Objectives To assess the use, yield, and cost-effectiveness of diagnostic tests used in the evaluation of syncope in children. Study design A retrospective review of 169 pediatric patients presenting to a tertiary care center with new onset syncope was undertaken. Test results were considered diagnostic when an abnormal result correlated with the clinical diagnosis or a normal result was obtained during a syncopal episode. Costs were based on the hospital cost of testing for fiscal year 1999, using a relative value unit–based costing methodology and did not include professional fees or costs of hospitalization.
Results A total of 663 tests were performed at a cost of $180,128. Only 26 tests (3.9%) were diagnostic in 24 patients (14.2%). The average cost per patient was $1055, and the cost per diagnostic result was $6928. Echocardiograms, chest radiographs, cardiac catheterizations, electrophysiology studies, and serum evaluations were not diagnostic. Conclusions The evaluation of pediatric syncope remains expensive, and testing has a low diagnostic yield. An approach that focuses on the use of testing to verify findings from the history and physical examination or exclude life-threatening causes is justified. (J Pediatr 2005;146:355-8) he evaluation of syncope can be an expensive, often futile endeavor. In 1982, Kapoor et al1 reported the average cost of a syncope evaluation for adult patients at $2463 per patient, despite a diagnostic yield of only 11%. Although new diagnostic modalities have increased the yield of a syncope evaluation, the cost of the evaluation remains substantial and the diagnostic yield remains low.2,3 Literature on the efficacy and cost of diagnostic tests in the pediatric population is lacking. The aim of this study was to determine how diagnostic tests are used in a current pediatric practice and to report on the utility and cost-effectiveness of these tests.
T
METHODS Patients presenting to the hospital or clinics for evaluation of syncope between January 1, 1998, and December 31, 1999, were identified from hospital billing records. Subjects between 4 and 18 years of age were included if they underwent an initial evaluation for episodes characterized by transient loss of consciousness associated with loss of postural tone. Patients were excluded if their presenting episode was diagnosed as a seizure at the initial visit or if the episode was characteristic of a previously established cause of syncope. Hospital charts as well as cardiology and neurology clinic records were reviewed from the date of presentation through April 30, 2000, to identify the clinical impression of the cause of syncope and to determine what tests were performed on each patient. The result of From The Children’s Heart Center each test was compared with the clinical impression. A test was considered diagnostic only at St Vincent Hospital, Indianapolis, when an abnormal result was consistent with the clinical diagnosis or when a normal result Indiana; and Division of Cardiology, Children’s Hospital Medical Center, was obtained during a typical syncopal episode. Cincinnati, Ohio. Tests obtained at other institutions were included when test results were available Submitted Oct 26, 2003; revision and the test was performed specifically for the evaluation of syncope. Tests repeated during received Sep 20, 2004; accepted Oct 13, 2004. follow-up evaluations were included in the analysis if the diagnosis remained unknown or if Reprint requests: Leonard A. Steinadditional syncopal events had occurred. Diagnostic evaluations were excluded when the berg, MD, The Children’s Heart medical record indicated that they were done for other indications, even if syncope was the Center at St Vincent Hospital, Indianapolis, 8333 Naab Road, Suite 320, primary complaint. Indianapolis, IN 46260. The costs reported are hospital costs for testing only and do not reflect professional 0022-3476/$ - see front matter fees, the cost of clinic evaluations, or hospital admissions. Costs were calculated by using Copyright ª 2005 Elsevier Inc. All rights a relative value unit–based costing methodology and were based on hospital costs for 1999. reserved. In most instances, each diagnostic result counted toward the total cost and number of tests 10.1016/j.jpeds.2004.10.039 355
Table. Diagnostic results counted as a single test All electrocardiograms performed on the same day Tilt testing at baseline and with Isoproterenol Exercise testing with post–exertional orthostatic challenge All event monitor submissions for each patient Twenty-four hours of ambulatory monitoring (Holter, neurotelemetry) Commonly grouped evaluations of blood and body fluids
performed. Exceptions are noted in the Table. The diagnostic yield for each test was calculated as the number of diagnostic tests divided by the number of tests performed. The costeffectiveness for each test was defined as the cost per diagnostic test (the cost of all tests divided by the number of positive tests). Statistical significance was calculated by using a Student t test and a x2 test where appropriate.
RESULTS Patients Patients who met the inclusion criteria (n = 169; 99 female) ranged in age from 4.5 to 18.7 years (mean, 13.2 ± 3.6). The follow-up period ranged from 122 to 768 days (mean, 312 ± 129). Several patients had preexisting conditions that may have affected their evaluation. Three patients had congenital heart disease, and 1 patient had a family history of hypertrophic cardiomyopathy. Preexisting neurologic disorders were present in 9 patients, including 3 patients with a history of seizures. Depression or other psychiatric disorders were present in 8 patients. Two patients had chronic pain syndromes. A diagnosis was established in 128 patients (76%). Neurocardiogenic syncope—the leading cause—was diagnosed in 116 patients (68%) and accounted for 91% of all diagnoses. Twelve patients received other diagnoses. Three patients were diagnosed with a seizure disorder. Pseudoseizure, anxiety disorder, and psychogenic syncope were each diagnosed in 2 patients. One patient each was diagnosed with breath-holding spells, long QT syndrome, and exertional ventricular tachycardia. In 41 patients (24%), the diagnosis was unknown, unspecified, or disputed by different physicians.
Studies and Cost Overall, 663 tests were obtained at a total cost of $180,128. The contribution of each test toward the total cost and the number of each test performed are shown in Figure 1. The electrocardiogram was the most frequently ordered test. A total of 245 electrocardiograms were ordered, and all but 11 patients (93%) had at least 1 electrocardiogram. The cost per patient ranged from $0 to $10,686. The mean cost was $1055 per patient (median $506). The number 356
Steinberg and Knilans
Figure 1. Pie graph depicts relative contribution of each test to total cost of testing. Numbers of each test obtained are displayed in parentheses. Total cost of all tests was $180,128.
of tests per patient ranged from 0 (2 patients) to 23. An average of 3.9 tests was obtained per patient (median, 3). Twenty-six tests (3.9%) met the criteria for clinical diagnosis and were obtained in 24 subjects (14.2%). Three tests (2 ambulatory electrocardiograms and 1 event recording) were diagnostic by virtue of demonstrating the absence of a rhythm abnormality during a typical syncopal spell. Thus, a positive test result correlated with the clinical diagnosis in only 23 of 663 tests (3.5%) in 21 of the 169 patients (12.4%). In the remaining 107 patients in whom a clinical diagnosis was established, no test result was consistent with the diagnosis. The diagnostic yield and the cost per diagnostic result for each test are shown in Figure 2. Overall, $6928 was spent for each diagnostic result obtained. Tilt-testing had the highest diagnostic yield (55.5%), and loop memory cardiac event monitoring had the lowest cost per diagnostic result ($1613). Among studies with at least one diagnostic result, electrocardiography had the lowest diagnostic yield (0.4%) and highest cost per diagnostic test ($28,665). Echocardiograms, chest roentgenographs, cardiac catheterizations, electrophysiology studies, and evaluations of serum and body fluids were not diagnostic in any patient. Comparing patients with and without a diagnostic test result, there was no significant difference in the age or sex distribution between the two groups. Subjects with a diagnostic test result had, on average, more than twice the number of tests performed (7.0 vs 3.4, P < .01), at nearly 3 times the cost per patient ($2429 vs $840, P < .01). Patients with a preexisting condition were no more or less likely to obtain a diagnostic test result. However, patients with a preexisting condition had more tests performed (6.4 vs 3.5, P < .01) at a higher average cost ($2321 vs $858, P < .01).
DISCUSSION Multiple studies have documented the limited utility of laboratory testing in adult populations with syncope, particularly when not guided by the history and physical examination.4-6 Limited data are available in children, but other small studies have reported a high cost and low yield.3,7 The Journal of Pediatrics March 2005
Our study confirms this finding, as demonstrated in Figure 2. The disparity in the numbers of each test performed is displayed in the denominator of the fraction above each test. The cost per diagnostic tests reflects the number of diagnostic tests, the number of tests performed, and the expense of the test. We chose to include subjects with preexisting cardiac conditions. The incidence of known heart disease in our subjects was 1.8%, whereas the incidence often reported in the general population is 0.7% to 1.0%. When these patients were excluded, the number of tests per patient (3.4) and the cost of testing per patient ($858) remained high. Exclusion of these patients did not significantly affect the diagnostic yield of testing. Loop memory cardiac event monitoring was the most cost-effective test in this study. However, only 1 monitor of 13 yielded a diagnostic result, demonstrating sinus rhythm during a syncopal spell. No pathologic rhythms were recorded. In adult populations, the high yield and cost-effectiveness of cardiac event monitors have been well documented.8-11 Possible explanations for a lower yield in children include a lower incidence of cardiac disease, intolerance to long-term continuous monitoring, inability to properly activate a monitoring device, and sampling error. Electrocardiograms were the most frequently ordered test, accounting for more than one third of all tests obtained. Only one electrocardiogram was diagnostic. Other studies in syncopal children have reported the low diagnostic yield of electrocardiography.3,7 However, it is a highly sensitive test for detecting heart disease. Ritter et al12 found that the history, physical examination, and electrocardiography were 96% sensitive for cardiac causes of syncope. In that study, the electrocardiogram was abnormal in 16 of 22 patients with cardiac syncope and was the only indicator of disease in 5 of those cases. Despite its low specificity, we and other authors recommend its routine use in the evaluation of syncopal children.12-14 Tilt-testing accounted for almost half of all diagnostic results. However, the worth of this test has recently been questioned. Reports on the specificity of tilt-testing in children range from 48% to 100%.15-20 Thus, a positive tilttest may not reliably exclude other conditions. Furthermore, tilt-testing is rarely required to diagnose neurocardiogenic syncope.13,21 In our study, 87% of patients believed to have neurocardiogenic syncope were diagnosed without the benefit of a diagnostic test result. This group includes 4 patients with a negative tilt-test. Excluding tilt-tests, only 2.2% of all other tests produced diagnostic results. Echocardiography accounted for 14% of the total cost of testing but did not produce a single diagnostic result. In the Ritter et al12 review of 480 syncopal children, 322 echocardiograms were obtained, but none revealed a pertinent abnormality not already suspected on the basis of the history, physical findings, or electrocardiogram. The evaluation of the syncopal patient is often complicated by conflicting objectives. One aim—to keep the Syncope In Children: Diagnostic Tests Have A High Cost And Low Yield
Figure 2. On upper graph, diagnostic yield for each test and overall yield are depicted as the percentage of tests that were diagnostic. The fraction above each test reveals numbers of diagnostic tests in the numerator and the total number of tests obtained in the denominator. Bottom graph displays cost per diagnostic test (cost of test 3 number of tests ordered/number of diagnostic results). EKG, electrocardiogram; XT, exercise test; tilt, head-up tilt-test; EEG, electroencephalogram; NI, neuroimaging (head CT or MRI); NT, neurotelemetry; all, all studies.
evaluation simple, cost-effective, and minimally invasive or noninvasive—often conflicts with the other—to exclude morbid conditions. Patients in this study with a diagnostic test result had significantly more tests performed. A likely explanation is concern over the severity of the problem and the perceived need for a definitive diagnosis confirmed by objective data. In our experience, an extensive workup that returns multiple negative results and no diagnostic answers often increases parental anxiety. This report may assist pediatricians in reassuring families and may help to dissuade them from embarking on a costly and unproductive evaluation. The data from this study do not support specific guidelines for evaluation of the syncopal child. In our own practice, we perform a detailed search for clues to suggest underlying cardiovascular disease. The family history is directed at identifying arrhythmogenic syndromes, sudden death (including drowning), or congenital heart disease. The personal history addresses prior symptoms potentially referable to the cardiovascular system, particularly exertional symptoms. Questions regarding the syncopal episodes focus on precipitating factors and prodromal symptoms. A careful cardiac examination is performed. All patients receive an electrocardiogram. When the history, physical examination, or electrocardiogram suggests the possibility of cardiac disease, further 357
testing is obtained. This usually includes an echocardiogram but depends on the specific nature of the positive finding. Event monitors are reserved for patients with associated palpitations or those with recurrent unexplained syncope. A treadmill test is used when patients have exertional symptoms. All patients with exertional syncope or structural heart disease undergo cardiac catheterization and electrophysiologic testing unless a definitive cardiac cause for syncope has already been established. Other tests are almost never used unless the history or physical reveals findings that would be elucidated by the specific result. We do not use tilt-testing to establish a diagnosis of neurocardiogenic syncope. Instead, we use it infrequently to reassure anxious families of the diagnosis or to provoke episodes of psychogenic syncope.
LIMITATIONS This study is subject to several limitations. The group may not be representative of patients evaluated in a primary care setting or other tertiary care centers. The costs of individual tests may be different at other centers and the yield of tests may differ. Also, the study does not account for the validity of the clinical impression. In some patients, test results were abnormal but no diagnosis was established. Since the effectiveness of each test was judged against each patient’s diagnosis, the identification of a diagnosis in these subjects may have altered the results. Another limitation is the inability to account for the clinical value of a negative test result. For example, a normal echocardiogram may have significant value in a patient with exertional syncope. However, our study implies that most negative results offer little additional value because there is a high pretest likelihood that the result will be nondiagnostic. Finally, this study is affected by the lack of standardized criteria to determine when specific tests should be applied. The use of such criteria could have drastically altered the diagnostic yield and cost-effectiveness of testing. Unfortunately, there is no available literature to support broad, uniform criteria for the use of specific testing in syncopal children. The recommendations made above are consistent with, though not directly supported by, the results of our study.
CONCLUSIONS The evaluation of syncope remains costly, and the majority of tests demonstrate a low diagnostic yield. Often, a diagnosis can be established without a positively correlating test result. Since most tests are unlikely to produce diagnostic results, an approach that focuses on the use of testing primarily to corroborate findings from the history and physical examination and to identify life-threatening diseases may be warranted. The results of this study are consistent with previous findings in both children and adults. These data should provide reassurance to parents and medical caretakers
358
Steinberg and Knilans
that a limited and directed approach to diagnostic testing best serves the majority of pediatric patients presenting with syncope.
REFERENCES 1. Kapoor WN, Karpf M, Maher Y, Miller RA, Levey GS. Syncope of unknown origin: the need for a more cost-effective approach to its diagnosis evaluation. JAMA 1982;247:2687-91. 2. Calkins H, Byrne M, el-Atassi R, Kalbfleisch S, Langberg JJ, Morady F. The economic burden of unrecognized vasodepressor syncope. Am J Med 1993;95:473-9. 3. Gordon TA, Moodie DS, Passalacqua M, Sterba R, Rothner AD, Erenberg G, et al. A retrospective analysis of the cost-effective workup of syncope in children. Cleve Clin J Med 1987;54:391-4. 4. Kapoor WN. Evaluation and outcome of patients with syncope. Medicine (Baltimore) 1990;69:160-75. 5. Day SC, Cook EF, Funkenstein H, Goldman L. Evaluation and outcome of emergency room patients with transient loss of consciousness. Am J Med 1982;73:15-23. 6. Eagle KA, Black HR. The impact of diagnostic tests in evaluating patients with syncope. Yale J Biol Med 1983;56:1-8. 7. Lerman-Sagie T, Lerman P, Mukamel M, Blieden L, Mimouni M. A prospective evaluation of pediatric patients with syncope. Clin Pediatr (Phila) 1994;33:67-70. 8. Linzer M, Pritchett EL, Pontinen M, McCarthy E, Divine GW. Incremental diagnostic yield of loop electrocardiographic recorders in unexplained syncope. Am J Cardiol 1990;66:214-9. 9. Cumbee SR, Pryor RE, Linzer M. Cardiac loop ECG recording: a new noninvasive diagnostic test in recurrent syncope. South Med J 1990;83:39-43. 10. Brown AP, Dawkins KD, Davies JG. Detection of arrhythmias: use of a patient-activated ambulatory electrocardiogram device with a solid-state memory loop. Br Heart J 1987;58:251-3. 11. Fogel RI, Evans JJ, Prystowsky EN. Utility and cost of event recorders in the diagnosis of palpitations, presyncope, and syncope. Am J Cardiol 1997;79: 207-8. 12. Ritter S, Tani LY, Etheridge SP, Williams RV, Craig JE, Minich LL. What is the yield of screening echocardiography in pediatric syncope? Pediatrics 2000;105:E58. 13. Steinberg LA, Knilans TK. Costs and utility of tests in the evaluation of the pediatric patients with syncope. Prog Pediatr Cardiol 2001;13:139-49. 14. Driscoll DJ, Jacobsen SJ, Porter CJ, Wollan PC. Syncope in children and adolescents. J Am Coll Cardiol 1997;29:1039-45. 15. Fouad FM, Sitthisook S, Vanerio G, Maloney J III, Okabe M, Jaeger F, et al. Sensitivity and specificity of the tilt table test in young patients with unexplained syncope. Pacing Clin Electrophysiol 1993;16:394-400. 16. Berkowitz JB, Auld D, Hulse JE, Campbell RM. Tilt table evaluation for control pediatric patients: comparison with symptomatic patients. Clin Cardiol 1995;18:521-5. 17. Alehan D, Lenk M, Ozme S, Celiker A, Ozer S. Comparison of sensitivity and specificity of tilt protocols with and without isoproterenol in children with unexplained syncope. Pacing Clin Electrophysiol 1997;20: 1769-76. 18. Alehan D, Celiker A, Ozme S. Head-up tilt test: a highly sensitive, specific test for children with unexplained syncope. Pediatr Cardiol 1996;17: 86-90. 19. Grubb BP, Temesy-Armos PN, Samoil D, Wolfe DA, Hahn H, Elliott L. Tilt table testing in the evaluation and management of athletes with recurrent exercise-induced syncope. Med Sci Sports Exerc 1993;25:24-8. 20. Lewis DA, Zlotocha J, Henke L, Dhala A. Specificity of head-up tilt testing in adolescents: effect of various degrees of tilt challenge in normal control subjects. J Am Coll Cardiol 1997;30:1057-60. 21. Levine MM. Neurally mediated syncope in children: results of tilt testing, treatment, and long-term follow-up. Pediatr Cardiol 1999;20:331-5.
The Journal of Pediatrics March 2005
Objectives To assess the use, yield, and cost-effectiveness of diagnostic tests used in the evaluation of syncope in children. Study design A retrospective review of 169 pediatric patients presenting to a tertiary care center with new onset syncope was undertaken. Test results were considered diagnostic when an abnormal result correlated with the clinical diagnosis or a normal result was obtained during a syncopal episode. Costs were based on the hospital cost of testing for fiscal year 1999, using a relative value unit–based costing methodology and did not include professional fees or costs of hospitalization.
Results A total of 663 tests were performed at a cost of $180,128. Only 26 tests (3.9%) were diagnostic in 24 patients (14.2%). The average cost per patient was $1055, and the cost per diagnostic result was $6928. Echocardiograms, chest radiographs, cardiac catheterizations, electrophysiology studies, and serum evaluations were not diagnostic. Conclusions The evaluation of pediatric syncope remains expensive, and testing has a low diagnostic yield. An approach that focuses on the use of testing to verify findings from the history and physical examination or exclude life-threatening causes is justified. (J Pediatr 2005;146:355-8) he evaluation of syncope can be an expensive, often futile endeavor. In 1982, Kapoor et al1 reported the average cost of a syncope evaluation for adult patients at $2463 per patient, despite a diagnostic yield of only 11%. Although new diagnostic modalities have increased the yield of a syncope evaluation, the cost of the evaluation remains substantial and the diagnostic yield remains low.2,3 Literature on the efficacy and cost of diagnostic tests in the pediatric population is lacking. The aim of this study was to determine how diagnostic tests are used in a current pediatric practice and to report on the utility and cost-effectiveness of these tests.
T
METHODS Patients presenting to the hospital or clinics for evaluation of syncope between January 1, 1998, and December 31, 1999, were identified from hospital billing records. Subjects between 4 and 18 years of age were included if they underwent an initial evaluation for episodes characterized by transient loss of consciousness associated with loss of postural tone. Patients were excluded if their presenting episode was diagnosed as a seizure at the initial visit or if the episode was characteristic of a previously established cause of syncope. Hospital charts as well as cardiology and neurology clinic records were reviewed from the date of presentation through April 30, 2000, to identify the clinical impression of the cause of syncope and to determine what tests were performed on each patient. The result of From The Children’s Heart Center each test was compared with the clinical impression. A test was considered diagnostic only at St Vincent Hospital, Indianapolis, when an abnormal result was consistent with the clinical diagnosis or when a normal result Indiana; and Division of Cardiology, Children’s Hospital Medical Center, was obtained during a typical syncopal episode. Cincinnati, Ohio. Tests obtained at other institutions were included when test results were available Submitted Oct 26, 2003; revision and the test was performed specifically for the evaluation of syncope. Tests repeated during received Sep 20, 2004; accepted Oct 13, 2004. follow-up evaluations were included in the analysis if the diagnosis remained unknown or if Reprint requests: Leonard A. Steinadditional syncopal events had occurred. Diagnostic evaluations were excluded when the berg, MD, The Children’s Heart medical record indicated that they were done for other indications, even if syncope was the Center at St Vincent Hospital, Indianapolis, 8333 Naab Road, Suite 320, primary complaint. Indianapolis, IN 46260. The costs reported are hospital costs for testing only and do not reflect professional 0022-3476/$ - see front matter fees, the cost of clinic evaluations, or hospital admissions. Costs were calculated by using Copyright ª 2005 Elsevier Inc. All rights a relative value unit–based costing methodology and were based on hospital costs for 1999. reserved. In most instances, each diagnostic result counted toward the total cost and number of tests 10.1016/j.jpeds.2004.10.039 355
Table. Diagnostic results counted as a single test All electrocardiograms performed on the same day Tilt testing at baseline and with Isoproterenol Exercise testing with post–exertional orthostatic challenge All event monitor submissions for each patient Twenty-four hours of ambulatory monitoring (Holter, neurotelemetry) Commonly grouped evaluations of blood and body fluids
performed. Exceptions are noted in the Table. The diagnostic yield for each test was calculated as the number of diagnostic tests divided by the number of tests performed. The costeffectiveness for each test was defined as the cost per diagnostic test (the cost of all tests divided by the number of positive tests). Statistical significance was calculated by using a Student t test and a x2 test where appropriate.
RESULTS Patients Patients who met the inclusion criteria (n = 169; 99 female) ranged in age from 4.5 to 18.7 years (mean, 13.2 ± 3.6). The follow-up period ranged from 122 to 768 days (mean, 312 ± 129). Several patients had preexisting conditions that may have affected their evaluation. Three patients had congenital heart disease, and 1 patient had a family history of hypertrophic cardiomyopathy. Preexisting neurologic disorders were present in 9 patients, including 3 patients with a history of seizures. Depression or other psychiatric disorders were present in 8 patients. Two patients had chronic pain syndromes. A diagnosis was established in 128 patients (76%). Neurocardiogenic syncope—the leading cause—was diagnosed in 116 patients (68%) and accounted for 91% of all diagnoses. Twelve patients received other diagnoses. Three patients were diagnosed with a seizure disorder. Pseudoseizure, anxiety disorder, and psychogenic syncope were each diagnosed in 2 patients. One patient each was diagnosed with breath-holding spells, long QT syndrome, and exertional ventricular tachycardia. In 41 patients (24%), the diagnosis was unknown, unspecified, or disputed by different physicians.
Studies and Cost Overall, 663 tests were obtained at a total cost of $180,128. The contribution of each test toward the total cost and the number of each test performed are shown in Figure 1. The electrocardiogram was the most frequently ordered test. A total of 245 electrocardiograms were ordered, and all but 11 patients (93%) had at least 1 electrocardiogram. The cost per patient ranged from $0 to $10,686. The mean cost was $1055 per patient (median $506). The number 356
Steinberg and Knilans
Figure 1. Pie graph depicts relative contribution of each test to total cost of testing. Numbers of each test obtained are displayed in parentheses. Total cost of all tests was $180,128.
of tests per patient ranged from 0 (2 patients) to 23. An average of 3.9 tests was obtained per patient (median, 3). Twenty-six tests (3.9%) met the criteria for clinical diagnosis and were obtained in 24 subjects (14.2%). Three tests (2 ambulatory electrocardiograms and 1 event recording) were diagnostic by virtue of demonstrating the absence of a rhythm abnormality during a typical syncopal spell. Thus, a positive test result correlated with the clinical diagnosis in only 23 of 663 tests (3.5%) in 21 of the 169 patients (12.4%). In the remaining 107 patients in whom a clinical diagnosis was established, no test result was consistent with the diagnosis. The diagnostic yield and the cost per diagnostic result for each test are shown in Figure 2. Overall, $6928 was spent for each diagnostic result obtained. Tilt-testing had the highest diagnostic yield (55.5%), and loop memory cardiac event monitoring had the lowest cost per diagnostic result ($1613). Among studies with at least one diagnostic result, electrocardiography had the lowest diagnostic yield (0.4%) and highest cost per diagnostic test ($28,665). Echocardiograms, chest roentgenographs, cardiac catheterizations, electrophysiology studies, and evaluations of serum and body fluids were not diagnostic in any patient. Comparing patients with and without a diagnostic test result, there was no significant difference in the age or sex distribution between the two groups. Subjects with a diagnostic test result had, on average, more than twice the number of tests performed (7.0 vs 3.4, P < .01), at nearly 3 times the cost per patient ($2429 vs $840, P < .01). Patients with a preexisting condition were no more or less likely to obtain a diagnostic test result. However, patients with a preexisting condition had more tests performed (6.4 vs 3.5, P < .01) at a higher average cost ($2321 vs $858, P < .01).
DISCUSSION Multiple studies have documented the limited utility of laboratory testing in adult populations with syncope, particularly when not guided by the history and physical examination.4-6 Limited data are available in children, but other small studies have reported a high cost and low yield.3,7 The Journal of Pediatrics March 2005
Our study confirms this finding, as demonstrated in Figure 2. The disparity in the numbers of each test performed is displayed in the denominator of the fraction above each test. The cost per diagnostic tests reflects the number of diagnostic tests, the number of tests performed, and the expense of the test. We chose to include subjects with preexisting cardiac conditions. The incidence of known heart disease in our subjects was 1.8%, whereas the incidence often reported in the general population is 0.7% to 1.0%. When these patients were excluded, the number of tests per patient (3.4) and the cost of testing per patient ($858) remained high. Exclusion of these patients did not significantly affect the diagnostic yield of testing. Loop memory cardiac event monitoring was the most cost-effective test in this study. However, only 1 monitor of 13 yielded a diagnostic result, demonstrating sinus rhythm during a syncopal spell. No pathologic rhythms were recorded. In adult populations, the high yield and cost-effectiveness of cardiac event monitors have been well documented.8-11 Possible explanations for a lower yield in children include a lower incidence of cardiac disease, intolerance to long-term continuous monitoring, inability to properly activate a monitoring device, and sampling error. Electrocardiograms were the most frequently ordered test, accounting for more than one third of all tests obtained. Only one electrocardiogram was diagnostic. Other studies in syncopal children have reported the low diagnostic yield of electrocardiography.3,7 However, it is a highly sensitive test for detecting heart disease. Ritter et al12 found that the history, physical examination, and electrocardiography were 96% sensitive for cardiac causes of syncope. In that study, the electrocardiogram was abnormal in 16 of 22 patients with cardiac syncope and was the only indicator of disease in 5 of those cases. Despite its low specificity, we and other authors recommend its routine use in the evaluation of syncopal children.12-14 Tilt-testing accounted for almost half of all diagnostic results. However, the worth of this test has recently been questioned. Reports on the specificity of tilt-testing in children range from 48% to 100%.15-20 Thus, a positive tilttest may not reliably exclude other conditions. Furthermore, tilt-testing is rarely required to diagnose neurocardiogenic syncope.13,21 In our study, 87% of patients believed to have neurocardiogenic syncope were diagnosed without the benefit of a diagnostic test result. This group includes 4 patients with a negative tilt-test. Excluding tilt-tests, only 2.2% of all other tests produced diagnostic results. Echocardiography accounted for 14% of the total cost of testing but did not produce a single diagnostic result. In the Ritter et al12 review of 480 syncopal children, 322 echocardiograms were obtained, but none revealed a pertinent abnormality not already suspected on the basis of the history, physical findings, or electrocardiogram. The evaluation of the syncopal patient is often complicated by conflicting objectives. One aim—to keep the Syncope In Children: Diagnostic Tests Have A High Cost And Low Yield
Figure 2. On upper graph, diagnostic yield for each test and overall yield are depicted as the percentage of tests that were diagnostic. The fraction above each test reveals numbers of diagnostic tests in the numerator and the total number of tests obtained in the denominator. Bottom graph displays cost per diagnostic test (cost of test 3 number of tests ordered/number of diagnostic results). EKG, electrocardiogram; XT, exercise test; tilt, head-up tilt-test; EEG, electroencephalogram; NI, neuroimaging (head CT or MRI); NT, neurotelemetry; all, all studies.
evaluation simple, cost-effective, and minimally invasive or noninvasive—often conflicts with the other—to exclude morbid conditions. Patients in this study with a diagnostic test result had significantly more tests performed. A likely explanation is concern over the severity of the problem and the perceived need for a definitive diagnosis confirmed by objective data. In our experience, an extensive workup that returns multiple negative results and no diagnostic answers often increases parental anxiety. This report may assist pediatricians in reassuring families and may help to dissuade them from embarking on a costly and unproductive evaluation. The data from this study do not support specific guidelines for evaluation of the syncopal child. In our own practice, we perform a detailed search for clues to suggest underlying cardiovascular disease. The family history is directed at identifying arrhythmogenic syndromes, sudden death (including drowning), or congenital heart disease. The personal history addresses prior symptoms potentially referable to the cardiovascular system, particularly exertional symptoms. Questions regarding the syncopal episodes focus on precipitating factors and prodromal symptoms. A careful cardiac examination is performed. All patients receive an electrocardiogram. When the history, physical examination, or electrocardiogram suggests the possibility of cardiac disease, further 357
testing is obtained. This usually includes an echocardiogram but depends on the specific nature of the positive finding. Event monitors are reserved for patients with associated palpitations or those with recurrent unexplained syncope. A treadmill test is used when patients have exertional symptoms. All patients with exertional syncope or structural heart disease undergo cardiac catheterization and electrophysiologic testing unless a definitive cardiac cause for syncope has already been established. Other tests are almost never used unless the history or physical reveals findings that would be elucidated by the specific result. We do not use tilt-testing to establish a diagnosis of neurocardiogenic syncope. Instead, we use it infrequently to reassure anxious families of the diagnosis or to provoke episodes of psychogenic syncope.
LIMITATIONS This study is subject to several limitations. The group may not be representative of patients evaluated in a primary care setting or other tertiary care centers. The costs of individual tests may be different at other centers and the yield of tests may differ. Also, the study does not account for the validity of the clinical impression. In some patients, test results were abnormal but no diagnosis was established. Since the effectiveness of each test was judged against each patient’s diagnosis, the identification of a diagnosis in these subjects may have altered the results. Another limitation is the inability to account for the clinical value of a negative test result. For example, a normal echocardiogram may have significant value in a patient with exertional syncope. However, our study implies that most negative results offer little additional value because there is a high pretest likelihood that the result will be nondiagnostic. Finally, this study is affected by the lack of standardized criteria to determine when specific tests should be applied. The use of such criteria could have drastically altered the diagnostic yield and cost-effectiveness of testing. Unfortunately, there is no available literature to support broad, uniform criteria for the use of specific testing in syncopal children. The recommendations made above are consistent with, though not directly supported by, the results of our study.
CONCLUSIONS The evaluation of syncope remains costly, and the majority of tests demonstrate a low diagnostic yield. Often, a diagnosis can be established without a positively correlating test result. Since most tests are unlikely to produce diagnostic results, an approach that focuses on the use of testing primarily to corroborate findings from the history and physical examination and to identify life-threatening diseases may be warranted. The results of this study are consistent with previous findings in both children and adults. These data should provide reassurance to parents and medical caretakers
358
Steinberg and Knilans
that a limited and directed approach to diagnostic testing best serves the majority of pediatric patients presenting with syncope.
REFERENCES 1. Kapoor WN, Karpf M, Maher Y, Miller RA, Levey GS. Syncope of unknown origin: the need for a more cost-effective approach to its diagnosis evaluation. JAMA 1982;247:2687-91. 2. Calkins H, Byrne M, el-Atassi R, Kalbfleisch S, Langberg JJ, Morady F. The economic burden of unrecognized vasodepressor syncope. Am J Med 1993;95:473-9. 3. Gordon TA, Moodie DS, Passalacqua M, Sterba R, Rothner AD, Erenberg G, et al. A retrospective analysis of the cost-effective workup of syncope in children. Cleve Clin J Med 1987;54:391-4. 4. Kapoor WN. Evaluation and outcome of patients with syncope. Medicine (Baltimore) 1990;69:160-75. 5. Day SC, Cook EF, Funkenstein H, Goldman L. Evaluation and outcome of emergency room patients with transient loss of consciousness. Am J Med 1982;73:15-23. 6. Eagle KA, Black HR. The impact of diagnostic tests in evaluating patients with syncope. Yale J Biol Med 1983;56:1-8. 7. Lerman-Sagie T, Lerman P, Mukamel M, Blieden L, Mimouni M. A prospective evaluation of pediatric patients with syncope. Clin Pediatr (Phila) 1994;33:67-70. 8. Linzer M, Pritchett EL, Pontinen M, McCarthy E, Divine GW. Incremental diagnostic yield of loop electrocardiographic recorders in unexplained syncope. Am J Cardiol 1990;66:214-9. 9. Cumbee SR, Pryor RE, Linzer M. Cardiac loop ECG recording: a new noninvasive diagnostic test in recurrent syncope. South Med J 1990;83:39-43. 10. Brown AP, Dawkins KD, Davies JG. Detection of arrhythmias: use of a patient-activated ambulatory electrocardiogram device with a solid-state memory loop. Br Heart J 1987;58:251-3. 11. Fogel RI, Evans JJ, Prystowsky EN. Utility and cost of event recorders in the diagnosis of palpitations, presyncope, and syncope. Am J Cardiol 1997;79: 207-8. 12. Ritter S, Tani LY, Etheridge SP, Williams RV, Craig JE, Minich LL. What is the yield of screening echocardiography in pediatric syncope? Pediatrics 2000;105:E58. 13. Steinberg LA, Knilans TK. Costs and utility of tests in the evaluation of the pediatric patients with syncope. Prog Pediatr Cardiol 2001;13:139-49. 14. Driscoll DJ, Jacobsen SJ, Porter CJ, Wollan PC. Syncope in children and adolescents. J Am Coll Cardiol 1997;29:1039-45. 15. Fouad FM, Sitthisook S, Vanerio G, Maloney J III, Okabe M, Jaeger F, et al. Sensitivity and specificity of the tilt table test in young patients with unexplained syncope. Pacing Clin Electrophysiol 1993;16:394-400. 16. Berkowitz JB, Auld D, Hulse JE, Campbell RM. Tilt table evaluation for control pediatric patients: comparison with symptomatic patients. Clin Cardiol 1995;18:521-5. 17. Alehan D, Lenk M, Ozme S, Celiker A, Ozer S. Comparison of sensitivity and specificity of tilt protocols with and without isoproterenol in children with unexplained syncope. Pacing Clin Electrophysiol 1997;20: 1769-76. 18. Alehan D, Celiker A, Ozme S. Head-up tilt test: a highly sensitive, specific test for children with unexplained syncope. Pediatr Cardiol 1996;17: 86-90. 19. Grubb BP, Temesy-Armos PN, Samoil D, Wolfe DA, Hahn H, Elliott L. Tilt table testing in the evaluation and management of athletes with recurrent exercise-induced syncope. Med Sci Sports Exerc 1993;25:24-8. 20. Lewis DA, Zlotocha J, Henke L, Dhala A. Specificity of head-up tilt testing in adolescents: effect of various degrees of tilt challenge in normal control subjects. J Am Coll Cardiol 1997;30:1057-60. 21. Levine MM. Neurally mediated syncope in children: results of tilt testing, treatment, and long-term follow-up. Pediatr Cardiol 1999;20:331-5.
The Journal of Pediatrics March 2005