 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
From the Department of Pediatrics, University of Kansas Medical Center, Kansas City, Kansas
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Methods. Seventy-six consecutive patients (mean age: 10.0; standard deviation: 6.5 years) in a pediatric cardiology outpatient clinic were studied. One pediatric cardiologist examined the patients with his acoustic stethoscope (AS); a second examined them within a few minutes using a remote TS. A nurse placed the TS chest piece as directed by the remote examiner via intercom, but neither video examination nor conversation with the parent/patient were permitted. Examiners independently recorded the stethoscope findings for all heart sounds, all murmurs, and heart disease (present/absent). TS accuracy was indexed using the 
statistic for TS/AS agreement and for TS agreement with auscultatory findings predicted from echocardiographic (echo) studies (N = 49).
Results. TS/AS agreement was satisfactory for presence/absence of heart disease ( = 0.63) and for organic, functional, vibratory, diastolic aortic, and diastolic pulmonic murmurs ( range: 0.65–0.75). For other specific murmurs and all heart sounds, TS/AS agreement was either unsatisfactory ( 
0.60) or indeterminate because prevalence was 0. TS-AS agreement improved when the TS was used by the more-experienced TS examiner and with patients at least 5 years of age. When the older children were examined by the more TS-experienced examiner, the TS-echo comparison yielded = 0.90, raw agreement = 0.96, sensitivity = 0.94, and specificity = 1.00.
Conclusions. In pediatric patients, a narrow-bandwidth telephonic stethoscope can accurately distinguish between functional and organic murmurs and thus can detect heart disease. Accuracy is greatest when the instrument is used by an experienced examiner with patients at least 5 years of age.
Key Words: pediatric cardiology • heart disease • telemedicine • instrumentation • electronic stethoscope
Abbreviations: TS, telephonic stethoscope • echo, echocardiography, echocardiogram • AS, acoustic stethoscope • CP, consensus prevalence • HR, heart rate
In the past decade, there has been a steep increase in pediatric telemedical research, and pediatric telecardiology has developed rapidly within this field. Clinically satisfactory diagnostic reliability and accuracy have been demonstrated for wide-bandwidth transmission of digital echocardiograms (echo) in infants1–3 and digital electronic cardiostethoscopy in older children.4,5
A variety of low-cost, very-narrow-bandwidth analog "telephonic" stethoscopes (TS) have recently come into use.6 These use ordinary telephone service, which would seem to provide significant advantages of cost and convenience over the wider-bandwidth models. However, it is not clear that any TS provides satisfactory auscultation. No quantitative clinical tests have been reported, yet TS utility may fall short for reasons related to electronic stethoscopy generally and TS specifications in particular.7,8
The aim of this study was to determine whether a TS can satisfy the accuracy requirements for clinical telecardiostethoscopy within a pediatric population. On the basis of previous studies, it was expected that the TS would perform better when used with older children5,9 and when used by the examiner who is more experienced with the telemedical equipment.10
The clinical test of a telediagnostic instrument typically focuses on its agreement with results obtained by the traditional modality1,4—in this case, the acoustic stethoscope (AS). Accordingly, we evaluated the TS in an unselected series of children who were referred to the pediatric cardiology clinic. We compared the independent auscultatory findings of 2 pediatric cardiologists, one using the TS, the other, his preferred AS, with these roles being reversed for a portion of the patient sample. To test further the TS findings regarding innocent and pathologic murmurs, we compared the TS results with those obtained independently by echo on the same patients.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Equipment
As described by the manufacturer, the TS (ATI, Eden Prairie, MN) consists of a Littman Classic II adult-size chest piece with 11 cm of tubing terminating in a microphone that feeds into a control box. The box modulates the signal and passes it into an ordinary telephone line. At the receiving end, the modulated TS signal is fed from the consulting physician’s telephone into a control box that demodulates the signal, selectively amplifies its lower frequencies, and plays the signal into a pair of headphones (Koss PRO480, Milwaukee, WI). The manufacturer does not specify the TS’s dynamic range or dynamic sensitivity but does claim satisfactory frequency response in the range 20 Hz to 250 Hz (bell) and 20 Hz to 550 Hz (diaphragm). The maximum transmitted frequency (550 Hz) falls far short of 1000 Hz, which is the highest frequency component of aortic and mitral regurgitant murmurs.11,12
The patients were examined in a pediatric cardiology outpatient examination room. One examiner performed the regularly scheduled examination using his preferred AS. The other (TS) examiner was located on the same floor in a separate room that is used for regular pediatric telemedicine consultations with public schools. The TS signal was communicated over the telephone, whereas voice communication was by a separate direct-wire intercom. The pediatric cardiologist who played the role of remote consultant (TS examiner) used the intercom to direct the study nurse’s placement of the TS chest piece and the switch from diaphragm to bell. All TS examinations were done with the patient in the supine position. There was no time limit on the remote auscultation.
Personnel and Study Protocol
Because of a wide age difference between the study’s 2 pediatric cardiologists, a standard hearing test was performed, which showed that both had normal or better-than-normal hearing across the standard audiometric frequencies. The cardiologists had unequal experience with electronic stethoscopy: the more-experienced had used an electronic stethoscope at least weekly for 8 years, whereas the other, less-experienced, had used one regularly for 3 years followed by 5 years without practice.
The TS examination was done either immediately before or immediately after the hands-on AS examination. The TS examiner was completely blind, neither seeing nor knowing anything about the patient beyond what was learned from the auscultation. The examiners’ roles were reversed for the last 30 patients. For using each examiner as the reference standard for the TS, it was necessary that the examiners have equal auscultatory skills with the AS. This equality was shown by their nearly perfect agreement when each examiner used the AS independently to detect heart disease in an unselected series of patients in a previous study.5
All findings were recorded on a comprehensive cardiac auscultation data-recording form that required a response to every item (Fig 1). Examiners did not discuss definitions of items before implementing the protocol.
|
statistic ( = [PO – PE] ÷ [1 – PE]).13 is preferable to raw agreement because of the correction for chance, and it summarizes in a single number the clinical utility of a diagnostic modality independent of sensitivity and specificity. ranges from approximately 0 (pure chance agreement) to 1.0 (perfect agreement beyond chance). For clinical purposes, > 0.6 is "substantial" and "clinically useful," whereas > 0.8 is "almost perfect."14 is the most widely used agreement statistic in biomedical studies15 and has current applications in pediatrics and pediatric cardiology.16–18 is not interpretable if consensus prevalence (CP; the proportion of cases for which the examiners agree that the target finding is present) = 0.0 or 1.00 or if the 2 examiners’ diagnostic biases are significantly different from one another by McNemar’s test.19 For none of the analyses were examiner biases different (all P > 0.05), but for several analyses, CP = 0.0, as reported. Sensitivity and specificity were computed for TS versus AS with the latter standing as the reference standard.
The assessment of stethoscope accuracy with respect to echo findings (reference standard) was defined as the stethoscope’s ability to diagnose heart disease solely on the basis of the presence of an organic murmur. For this purpose, 1 of the authors blindly reviewed the latest echos (available for 49 [64%] of the patients) and identified those lesions that could be expected to produce an audible organic murmur. In this review, he followed Roldan et al20 in ignoring minimal jets and minimal transvalvar gradients. Sensitivity, specificity, and agreement () were then calculated, pitting the actual auscultatory finding (organic murmur present/absent) against the echo projection (audible organic murmur expected/not expected).
Results for continuous variables are reported as mean ± standard deviationn-1 and were analyzed using t test and Pearson correlation coefficient (r). The hypothesis that = 0.0 was assessed by z test, whereas differences between s were assessed using Fleiss’s test.13 Statistical significance was inferred when P .05. All statistics were computed using SPSS version 9.0 (SPSS, Chicago, IL).
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
Accuracy: TS Versus AS
Individual s were computed for agreement between the TS and AS (TS-AS) for each variable listed in the data-recording form excepting heart rate and diagnosis (Table 2). For 7 of the variables, TS-AS was indeterminate because CP = 0.0. For all 15 of the remaining variables, TS-AS agreement was significantly greater than chance (all P < .02), and TS-AS was clinically useful (>0.6) for 8 of these items: 1) detecting murmur and detecting specific murmurs, including the overarching classes 2) functional and 3) organic and 3 specific types: 4) vibratory, 5) diastolic aortic, and 6) diastolic pulmonic; 7) detecting heart disease; and 8) ordering an echo. For these 8 items, median sensitivity was 0.81 (range: 0.63–0.93) and median specificity was 0.87 (range: 0.71–1.00). In contrast, for both of the assessable heart sounds (S2, S3) and both assessable clicks (pulmonic, aortic), TS-AS 0.5—not clinically useful. Agreement was also unsatisfactory (TS-AS < 0.4) for systolic regurgitant and systolic ejection murmurs. Disregarding the specific type of systolic murmur, however, TS-AS agreement on presence of systolic murmur was found to be clinically useful (TS-AS = 0.66; not shown in Table 2).
|
= 0.97). All results reported for detection of heart disease apply equally to ordering an echo, and the latter is therefore not reported separately.
Age-group comparisons were made for the 7 clinically useful variables (other than ordering an echo) identified in Table 2 (Fig 2A). None of the differences between the 2 age groups was statistically significant; and for both age groups, TS-AS was in the clinically useful range for vibratory and diastolic aortic murmurs. However, for the younger children, TS-AS was generally lower than for the older children; and, of particular concern, for the younger children, TS-AS was not in the clinically useful range for functional murmur, organic murmur, or heart disease. TS-AS was computed for the more-experienced versus the less-experienced TS examiner (Fig 2B). The difference was marginally significant for presence of murmur (P = .083) and was significant for presence of organic murmur (P = .037). Moreover, in all possible comparisons, the more-experienced examiner’s TS-AS exceeded that of the less-experienced. All of the more-experienced examiner’s TS-AS were in the clinically useful range, whereas for the less-experienced examiner, this was true only for the diastolic pulmonic murmur.
|
was computed for audible organic murmur (present/absent by stethoscope vs expected/not expected by echo). In the present series, 33 (67%) of 49 patients who had echos were expected to present with audible organic murmurs: including multiple diagnoses, the echo revealed 2 tricuspid valve regurgitation, 1 tricuspid valve prolapse, 10 pulmonary valve stenosis or regurgitation, 1 bicuspid aortic valve, 14 aortic valve/subvalvar stenosis or regurgitation, 4 mitral valve stenosis or regurgitation, 1 mitral valve prolapse, 1 atrial septal defect, 10 ventricular septal defect, and 2 patent ductus arteriosus. For the remaining 16 of (33%) 49 patients, audible organic murmur was not expected on the basis of the echo.
Figure 3 shows sensitivity, specificity, and TS-Echo and AS-Echo for all patients and for the age and experience subgroups. When all patients were included in the analysis (Fig 3A), the TS did not show clinically useful accuracy (TS-Echo = 0.55), whereas the AS achieved highly satisfactory accuracy (AS-Echo = 0.77). The difference between these Ks was marginally significant (P = .083).
|
TS-Echo = 0.29, AS-Echo = 0.87; P = .024). When used with older patients, however, the TS had satisfactory accuracy comparable to that of the AS (TS-Echo = 0.67, AS-Echo = 0.74).
In Fig 3C, it is seen that the TS in the hands of the less-experienced TS examiner was significantly less accurate than when he was using the AS (TS-Echo = 0.32, AS-Echo = 0.85; P = .024). In contrast, the more-experienced TS examiner’s TS accuracy was comparable to his AS accuracy, with both being in the clinically useful range (TS-Echo = 0.66, AS-Echo = 0.62). When used optimally (ie, by the more-experienced examiner with older patients; not shown in Fig 3), the TS had near-perfect accuracy, with TS-Echo = 0.90, sensitivity = 0.94, and specificity = 1.00.
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
statistic to quantify agreement corrected for chance. The technical acoustics and subjective "feel" of electronic stethoscopes are inherently different from traditional air-conduction acoustic instruments.21–23 Aside from the characteristics of the TS itself, the telephone service on which it depends has 3 major limitations, including the telephone’s narrow dynamic (loudness) range, its narrow dynamic resolution (capture of variation in loudness), and its reduced frequency response below 100 Hz. These limitations may interfere with grading of murmurs or nuanced appreciation of signs such as S3 gallop, split sounds, and innocent murmurs, all of which are soft or low frequency.21,24,25 The problem of the TS’s dynamic response has been partially addressed by selective amplification. The low-frequency problem has been addressed by elevating all frequencies before telephonic transmission, then restoring them on reception (bandwidth modulation). These solutions, coupled with the manufacturer’s other adaptations for electronic stethoscopy (eg, modified chest piece, use of headphones instead of the familiar air-conduction binaural), may yield unfamiliar acoustics that would, at the very least, exacerbate the difficulties and extend the time required to master telemedical equipment.10,26,27 Despite these potential problems, however, the analog TS is shown here to have clinical capabilities remarkably similar to those of the wide-bandwidth digital instrument.5
TS Accuracy Versus AS
The TS, used completely blindly, showed clinically useful agreement with an independent hands-on AS in 1) detecting and subclassifying heart murmurs as functional or organic, thus enabling the examiner to recognize the presence of heart disease; and 2) further recognizing vibratory, diastolic aortic and diastolic pulmonic murmurs. In these respects, the TS performed at least as well as the wide-bandwidth instrument.5
The TS had relatively weak agreement with the AS in specifically distinguishing systolic ejection and systolic regurgitant murmurs. This unsatisfactory TS-AS agreement on these specific systolic murmurs stands in contrast to the wide-bandwidth digital electronic stethoscope, which agreed well with the AS on presence of systolic regurgitant murmurs.5 The low TS-AS agreement on systolic murmurs shown here also cannot be explained by any limitation in the AS, which is highly reliable in detecting and classifying systolic murmurs in children5 and in adults.25,28
The TS was poor at detecting S3 and abnormalities in S2. This is in keeping with findings from the wide-bandwidth instrument,5 and it confirms a recognized lack of interexaminer agreement concerning S3 and S2 in acoustic stethoscopy.28–31 The difficulty with S3 may result in part from its transitory nature, whereas disagreement on the presence of S2 abnormalities may result from interexaminer differences in perception of subtle changes in A2–P2 interval—perhaps no more than 0.02 seconds.32,33
Stethoscope Accuracy Relative to Echo
The results for TS versus echo clearly agree with the TS-AS results in suggesting that the TS be used only for children at least 5 years of age, and then only by an experienced TS operator. The age limitation would not be problematic for remote examinations of neonates or infants: auscultation is known to be of reduced value for these young patients,34 for whom tele-echocardiography is the well-established modality of choice.3 The 5-year minimum age for TS accuracy permits its use in telemedical consultations involving elementary- and secondary-school students, who are the chief focus of some telemedicine programs.35 However, the TS used in this study does not seem to be satisfactory for the examination of children ages 1 to 4 years who present for evaluation of murmur.
Our results with the telephonic stethoscope—particularly its excellent agreement with echo—reinforce and extend the consensus over many studies over many years that the clinical examination can provide highly satisfactory detection of heart disease in children beyond early childhood without recourse to immediate echo.9,36
Limitations of the Study
This study’s limited sample size precluded assessment of the TS’s accuracy in detecting several specifically abnormal signs that have low prevalence in the clinic population and therefore were not represented in the current sample. For the analyses of the subgroups, a larger sample would have added some precision to the estimates of and sensitivity and specificity. However, the sample was sufficient from a purely statistical point of view in that it yielded enough power to detect significant and significant differences between s.
The composition of the clinic sample used in this study is not representative of any school-age population for which the TS might be used as a cardiac screening tool. This limitation precluded calculation of useful positive and negative predictive values, but the instrument’s sensitivity and specificity remain valid regardless of variation in sample composition.
Several features of the study that would be altered in a more thoroughgoing field study may be mentioned. That the TS was equipped with an adult-size chest piece may account for some of its diagnostic inaccuracy with the younger children. Some TS inaccuracy almost certainly derived from the study protocol’s requirement that the TS examiner not view the child and hence not assess skin color, operative scars, precordial impulse, and respiratory rate. The study nurse’s initial lack of experience with the equipment, which is a recognized hazard in telemedicine,37 may also have contributed to TS inaccuracy. Similarly, the TS examiner did not discuss the amount of pressure exerted on the chest piece by the study nurse, yet the influence of chest piece pressure on appreciation of low-frequency murmurs is well established.38 Finally, the study protocol did not allow TS examination of the patient in the sitting position, which likely accounts for the TS’s failure to detect any of the venous hums identified by the AS in 8.0% of the patients. No cases of heart disease were missed by the TS because of this omission. The venous hum, however, is a potentially important exclusionary finding, and it therefore should be included in future instrument-validation studies with children.
These limitations all lead to underestimation of the TS’s accuracy for ordinary telemedical applications. That the TS performed well despite these indicates the instrument’s fundamental usefulness. To provide a fair test of the instrument’s accuracy, the TS examiner was blinded to all information except what was gathered by auscultation. This blinding was in accordance with the classic requirements of Feinstein et al39 for establishing the diagnostic accuracy of cardiac auscultation. Under ordinary use, when the TS examiner has full access to history and other physical findings, we expect the TS to perform even better than was shown here.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
We are grateful to Kenneth Goertz, MD, for his generous professional contribution as 1 of the project’s 2 cardiologists; to Mari Platt, RN, for steadfast service as study nurse and study manager; and to Christopher Budig for arranging the experimental setup.
|
FOOTNOTES |
|---|
Reprint requests to (L.F.M.) Department of Pediatrics, University of Kansas Medical Center, Kansas City, KS 66160-7330. E-mail: lmattiol{at}kumc.edu
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. P. Finley, A. E. Warren, G. P. Sharratt, and M. Amit Assessing Children's Heart Sounds at a Distance With Digital Recordings Pediatrics, December 1, 2006; 118(6): 2322 - 2325. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2003 American Academy of Pediatrics. All rights reserved. |
||
|
||
5 years (A, solid bars; N = 56) and for the more-TS-experienced examiner (B, solid bars; N = 40). Greater experience was associated with significantly better TS-AS agreement on presence of organic murmur (P = .037).