Article Text
PDF
Abstract
Objectives To determine the accuracy of remote diagnosis of congenital heart disease (CHD) by real-time transmission of echocardiographic images via integrated services digital network (ISDN) lines, to assess the impact on patient management and examine cost implications.
Design Prospective comparison of echocardiograms on infants with suspected significant CHD performed as follows: (1) hands-on evaluation and echocardiogram by a paediatrician at a district general hospital (DGH) followed by (2) transmission of the echocardiogram via ISDN 6 with guidance from a paediatric cardiologist and finally (3) hands-on evaluation and echocardiogram by a paediatric cardiologist. The economic analysis compares the cost of patient care associated with the telemedicine service with a hypothetical control group.
Setting Neonatal units of three DGH and a UK regional paediatric cardiology unit.
Results Echocardiograms were transmitted on 124 infants. In five cases scans were inadequate for diagnosis. Of the remaining 119 tele-echocardiograms, a follow-up echocardiogram was performed on 109/119 (92%). Major CHD was diagnosed in 39/109 infants (36%) and minor CHD in 45 (41%). The tele-echo diagnosis was accurate in 96% of cases (κ=0.89). Unnecessary transfer to the regional unit was avoided in 93/124 patients (75%). Despite relatively high implementation costs, telemedicine care was substantially cheaper than standard care. Each DGH potentially saved money by utilising the telemedicine service (mean saving: £728/patient).
Conclusions CHD is accurately diagnosed by realtime transmission of echocardiograms performed by paediatricians under live guidance and interpretation by a paediatric cardiologist. Remote diagnosis and exclusion of CHD affects patient management and may be cost saving.
Statistics from Altmetric.com
Introduction
Despite its recent rapid expansion, there remains potential for further widespread application of telemedicine in many specialities,1 particularly those where imaging is crucial to diagnosis and communication with an expert is vital. Paediatric cardiology is just such a speciality, where structural heart disease forms the bulk of the neonatal workload in the context of a health service in which the majority of babies are born in hospitals remote from centres offering a paediatric cardiology service.2,3
Published literature provides evidence for the effectiveness of transmission of remote echocardiographic images via a telemedicine link in the diagnosis and management of children with suspected congenital heart disease (CHD).4,–,18 However, very few studies have achieved high follow-up.8,10 While earlier studies utilised two integrated services digital network lines (ISDN 2, bandwidth 128 kbps), it is now generally agreed that ISDN 6 (384 kbps) is the preferred bandwidth to transmit paediatric echocardiographic images.19,20 It is recognised that telemedicine applications are limited by a lack of long-term studies including cost analyses.21,22
What is already known on this topic
Remote diagnosis of congenital heart disease is possible using telemedicine links.
Telemedicine projects may be cost saving in the start-up phase but often become expensive as initial enthusiasm wanes.
What this study adds
The rigorous follow-up in this study strengthens the validation of telemedicine in the remote diagnosis and exclusion of congenital heart disease.
A long-standing, regional telemedicine programme may continue to be cost saving.
The aims of this study were to determine the diagnostic accuracy of real-time echocardiographic images transmitted across ISDN 6 lines, to assess the impact of the remote consultation on patient transfers and to perform a cost analysis of the telemedicine service.
Methods
This prospective study was performed between 1999 and 2006 at a regional paediatric cardiology unit (RPCU), serving a population of 1.7 million people with approximately 320 000 children (0–13 years). Telemedicine links (ISDN 6) were established between the RPCU in Belfast and three district general hospitals (DGHs) in Altnagelvin Area Hospital (ALT) 75 miles away, Craigavon Area Hospital (CAH) 35 miles away and Antrim Area Hospital (AAH) 20 miles away. There were 20 700, 19 900 and 18 400 live births at these hospitals, respectively, during the study period.
Study protocol
The study protocol received institutional ethics approval and informed parental consent was obtained prior to enrolment in the trial. Inclusion was limited to patients whom the attending paediatrician suspected had a major congenital heart defect and required an urgent cardiology consultation (figure 1). Before the advent of the telemedicine service all such patients were transferred to the RPCU for hands-on evaluation. Indeed this continues to be the case for those hospitals in our region without a telemedicine link. Step 1: initial clinical assessment and echocardiogram performed by the local paediatrician, reaching a provisional diagnosis (diagnosis 1). Step 2: having contacted the consultant paediatric cardiologist, a second echocardiogram was performed by the local paediatrician. This echocardiogram was transmitted in real time to the RPCU with guidance from the paediatric cardiologist. At this point a working diagnosis (diagnosis 2) and management plan were agreed. The management options were step 3a: transfer to the specialist centre (urgent or semi-urgent) or step 3b: continued management at the DGH. The final diagnosis (diagnosis 3) was determined following direct consultation with the paediatric cardiologist, including a hands-on echocardiogram, either at the RPCU after transfer (step 4a) or at a subsequent outpatient clinic at the DGH (step 4b) if not transferred. The hands-on consultation was not necessarily performed by the paediatric cardiologist present at the remote consultation. Diagnoses were classified as: major CHD, minor CHD or no CHD. Major CHD was defined as a defect likely to require medical or surgical treatment in the first year of life. Minor CHD was defined as a lesion unlikely to require medical or surgical treatment in the first year of life.
Study protocol. DGH, District General Hospital.
Cost analysis
The viewpoint for the cost analysis was that of the NHS as recommended by the National Institute for Health and Clinical Excellence (NICE). The two alternatives under comparison are the patient pathway in the presence of a telemedicine service and the hypothetical standard care pathway in the absence of a telemedicine service.
All relevant costs were identified, measured and compared for each of the care pathways (remote diagnosis via telemedicine versus standard care). Healthcare resource utilisation in undertaking the initial assessment and transfer of the patient from the DGH and any subsequent assessments were captured. The cost of such input was estimated using data derived from this study and expert panel opinion (in conjunction with best practice guidelines) on the staff requirements for the activities outlined in each care pathway. These data were then combined with unit cost data in pounds sterling as assigned by the PSSRU23)for such items as staff time, the cost of an echocardiogram and paediatric cardiology outpatient assessment. Transportation times included pick-up, drop-off and return journey: ALT required 400 min, CAH 240 min and AAH 210 min (Northern Ireland Ambulance Service, 2007, personal communication). Unit costs and quantities were reported separately in the interests of clarity. It was not necessary to discount costs as costing was undertaken using unit cost data for 2007. Sensitivity analyses were performed to address uncertainty in the parameters used in the cost analysis.
Statistical analysis
The main outcomes of interest were the level of diagnostic agreement of the initial and transmitted echocardiograms with the hands-on echocardiogram. These data were analysed using a co-efficient of agreement test (κ statistic). The accuracy of the initial echocardiogram with respect to the hands-on echocardiogram was directly compared with that of the transmitted echocardiogram by using McNemar's test. The data were analysed using Windows SPSS 15.0.
For the analysis of diagnostic accuracy, we have excluded those patients in whom a hands-on echocardiogram was not performed and the five patients deemed not suitable for analysis (n=109). For analysis of patient transfers, all patients have been included (n=124).
Results
Over an 8-year period, echocardiograms were transmitted on 124 patients suspected of having CHD (55 ALT, 28 CAH, 41 AAH). A total of 114 patients were followed up with a third echocardiogram with 109 patients suitable for analysis. In five patients major CHD was suspected but could not be confidently diagnosed or excluded (table 1). Each patient was transferred to the RPCU for assessment. The difficulty with these echocardiograms related, in all cases, to the acquisition of the appropriate images by the paediatrician rather than any degradation incurred in transmission.
Cases of inability to diagnose major congenital heart disease following telelink
CHD was diagnosed in 84/109 (77%) cases; 39 (36%) had major CHD and 45 (41%) had minor CHD (table 2). The initial echocardiogram by the paediatrician was accurate in 58% (63/109) of cases when compared with the hands-on echocardiogram (κ =0.14, 95% CI 0 to 0.31, sensitivity 57%, specificity 59%). The accuracy was significantly increased to 96% with the addition of the transmitted echocardiogram (105/109, κ=0.89, 95% CI 0.71 to 1.0, sensitivity 97%, specificity 96%).
Congenital heart disease diagnosed by telemedicine
In four cases minor errors were made in the transmitted scan diagnosis shown in table 2 (three false negative, one false positive). However, in each of these cases additional congenital heart defects were correctly identified by the transmitted scan.
Ninety five (87%) patients were neonates at the time of the transmitted echocardiogram (median age: 3 days, IQR: 1–12 days, range 0–342 days). The median time to follow-up from the transmitted echocardiogram was 9.5 days (IQR: 3–22.5 days). The mean consultation time was 19.8 min (range 9–44 min).
Outcomes
Ninety three patients (93/124, 75%) did not require transfer from the DGH to the RPCU as a result of the telemedicine consultation. The financial impact of telemedicine for each hospital is shown in table 3. Seventeen patients were urgently transferred to the regional unit following the transmitted scan. Fourteen patients were transferred semi-urgently to the regional unit within 48 h of the initial referral (table 2). In 11 cases a false positive diagnosis of a duct dependent CHD was made by the paediatrician in the DGH. In all 11 cases significant CHD was ruled out following the transmitted echocardiogram. Seven patients had persistent pulmonary hypertension of the newborn and in four cases coarctation of the aorta was disproved.
Total cost (per patient) for each care pathway (by DGH) by category of cost
Cost
The cost of the initial assessment (figure 1, step 1, £376/patient) is common to both pathways and has been discounted from the cost analysis. In the absence of telemedicine all patients would be transferred to the RPCU for evaluation (figure 1, steps 3a and 4a). It was estimated that a consultant paediatrician, specialist registrar and nurse (band 5) were required for approximately 90 min. A second echocardiogram was performed at this time (£180). The same assumption was made for those receiving a telemedicine consultation but only a proportion of patients from each DGH were transferred: ALT 33%, CAH 21%, AAH 17% (table 1). Patients who were diagnosed by telemedicine but did not require transfer (step 3b) were electively reviewed at an outpatient clinic at the DGH (step 4b). This assessment cost £386 including an echocardiogram.
The cost of providing the telemedicine service (for the three DGH and RPCU) was a combination of hardware costs, ISDN 6 rental, ISDN maintenance and calls (£9.50/call). The one-off installation of the ISDN 6 lines (£297) was treated as sunk. The professional time of the consultation was based upon two consultants and nurse (band 5). The relative costs per patient in both telemedicine and standard care are summarised in table 3.
Discussion
High diagnostic accuracy is possible using a telemedicine link to transmit images obtained with the assistance of live guidance by a paediatric cardiologist. The results also demonstrate the importance of an expert interpretation of the echocardiographic images for accurate diagnosis. In the context of an acutely ill baby, this facility of a readily available, expert opinion that alters patient management, demonstrates a clinical benefit. There has been a modest increase in diagnostic accuracy in this study with 384 kbps bandwidth in comparison with our previous study at 128 kbps bandwidth (96% vs 91%).4 However, it is not possible to attribute this increase to bandwidth increase alone. Other contributing factors include more sophisticated videoconferencing codecs and software algorithms and improved echocardiographic skills on the part of the referring paediatrician. No clinically significant diagnostic errors resulted from telemedicine consultations when compared with hands-on echocardiogram.
In this series, 124 video conferences were performed during the 8-year study period. This is a lower transmission rate in comparison with other studies. However, in this study, only 23% of patients did not have CHD; 36% of patients had defects requiring surgical or medical treatment including 9% with a duct dependent lesion. This confirms that in our study referrals were made appropriately in the face of well-founded concern regarding CHD, rather than on a routine basis. Although previous studies have researched the feasibility and financial implications of diagnosing CHD by telemedicine, no studies have validated the accuracy of echocardiograms, transmitted via ISDN 6, by following up each patient with a hands-on echocardiogram. Two studies have achieved high follow-up rates with a combination of reviewing the original videotape and hands-on echocardiogram.9,11 The only way to ensure the development of an effective body of knowledge on the impact of telemedicine is by an objective validation and evaluation programme.1 In this study 92% of patients were followed up by hands-on evaluation. We believe that this step is important to confidently assess the accuracy of echocardiography transmitted via telemedicine links.
During this study there were five cases in which major CHD was suspected following the transmitted scan but a confident diagnosis could not be established. These cases highlight the importance of recognising the limitations of this technique. If there is a clinical suspicion of major CHD that cannot be confidently diagnosed or excluded, following a tele-echocardiogram, then transfer to a RPCU for hands-on assessment should not be delayed.
Costs
The aim of this study was not to determine if telemedicine was cost effective compared to standard care, but rather, to determine if, given that no differences in diagnostic accuracy were detected, was telemedicine less costly. Each DGH made a potential saving through involvement in the telemedicine programme. The DGH furthest from the paediatric cardiology centre utilised the tele-echocardiography service most often and was most cost saving in comparison with the other two DGHs. The key cost driver in standard care was the cost of ambulance transfer from the DGH to RPCU. The cost per patient for a teleconsultation was also a cost driver.
Education
The value of telemedicine in education has been well documented if not fully researched.3,9,12,13 Our experience of the educational benefit of real-time transmission of echocardiograms has been very positive. Having performed echocardiograms under the guidance of a paediatric cardiologist, paediatricians report an improvement in their ability to interpret echocardiograms and have been observed to gain confidence in echocardiography. A survey of the opinions of consultant paediatricians in our region was conducted relating to the telemedicine programme. As part of this survey paediatricians were asked whether they found the telemedicine service useful and if they felt reassured by the facility. Using a Likert scale (5=strongly agree, 1=strongly disagree) the mean response to (A) useful was 4.5/5 (SD 0.82) and (B) reassured was 4.2/5 (SD 1.09).
We believe that real-time, live guidance of the paediatrician by the paediatric cardiologist has advantages compared to a ‘store and forward’ protocol, in both acquiring quality images and in providing educational support. Often the inexperienced sonographer obtains images that they believe are not adequate. However, it is a difficulty interpreting the images that is the major contributing factor to poor image acquisition. During the videoconference, the paediatric cardiologist is able to highlight relevant structures as they appear and describe what manoeuvres are necessary to view other structures. The authors advise frequent use of the telemedicine service and attendance at an echocardiography skills course to facilitate quality remote consultations.
Future
One of the main barriers to increased use of telemedicine is the initial start-up cost. However, running costs for ISDN 6 lines are also significant (£1188 per annum + call costs). As predicted, the cost of videoconferencing codecs suitable for transmission of echocardiograms has decreased significantly since the beginning of this study and are now available for approximately £2000. Over the past few years, the possibility of using the internet as a means of transmission has been explored with some success. We are currently evaluating transmission of images via a symmetric digital subscriber line (SDSL) which affords a 2 Mbps bandwidth. The rental for this connection is currently £150 per month but is predicted to reduce to approximately £60 per month within 2 years (Questmark, 2007, personal communication). Such running costs should enable our telemedicine service to be more cost effective.
We envisage a hub and spoke model for the provision of paediatric cardiology in our region. Telemedicine will be a valuable aid in supporting paediatricians with a special interest in cardiology. A fourth DGH has recently installed a telemedicine link and we plan to establish links with the remaining three DGH providing acute paediatric services in Northern Ireland.
Conclusions
The results of this study clearly demonstrate that CHD is accurately diagnosed by real-time transmission of echocardiograms performed by paediatricians, under live guidance and interpretation by a paediatric cardiologist. This study also suggests that a long-standing, regional telemedicine programme may continue to be cost effective. Savings are predominantly made from the avoidance of costly inter-hospital transfers. Telemedicine has become an integral part of our regional paediatric cardiology service.
Acknowledgments
The authors would like to thank the consultants from Altnagelvin Area Hospital, Craigavon Area Hospital and Antrim Area Hospital for their participation in the telemedicine programme. The authors would like to acknowledge the advice provided by Sam McMaster of Questmark Ltd. The authors would also like to acknowledge the advice and help of Mr Chris Cardwell (Department of Epidemiology and Public Health, Queen's University Belfast) in analysing the data and interpreting the statistics for this study.
References
Footnotes
-
Funding Funding for this research was provided by the Royal Belfast Hospital for Sick Children.
-
Competing interests None.
-
Ethics approval The Royal Group of Hospitals Research Ethics Committee approved this study.
-
Provenance Not commissioned; externally peer reviewed.
-
Patient consent Obtained.