Author + information
- Received December 5, 2002
- Revision received January 17, 2003
- Accepted January 30, 2003
- Published online August 6, 2003.
- Andrew D Krahn, MD*,* (, )
- George J Klein, MD*,
- Raymond Yee, MD*,
- Jeffrey S Hoch, PhD† and
- Allan C Skanes, MD*
- ↵*Reprint requests and correspondence:
Dr. Andrew D. Krahn, London Health Sciences Center, University Campus, 339 Windermere Road, London, Ontario, Canada N6A 5A5.
Objectives We sought to assess the cost implications of two investigation strategies in patients with unexplained syncope.
Background Establishing a diagnosis in patients with unexplained syncope is complicated by infrequent and unpredictable events. The cost-effectiveness of immediate, prolonged monitoring as an alternative to conventional diagnostic strategies has not been studied.
Methods Sixty patients (age 66 ± 14 years; 33 males) with unexplained syncope and LV ejection fraction >35% were randomized to conventional testing with an external loop recorder, tilt and electrophysiologic (EP) testing, or prolonged monitoring with an implantable loop recorder with one-year monitoring. If patients remained undiagnosed after their assigned strategy, they were offered a crossover to the alternate strategy. Cost analysis of the two testing strategies was performed.
Results Fourteen of 30 patients who were being monitored were diagnosed at a cost of $2,731 ± $285 per patient and $5,852 ± $610 per diagnosis. In contrast, only six of 30 conventional patients were diagnosed (20% vs. 47%, p = 0.029), at a cost of $1,683 ± $505 per patient (p < 0.0001) and $8,414 ± $2,527 per diagnosis (p < 0.0001). After crossover, a diagnosis was obtained in 1 of 5 patients undergoing conventional testing, compared with 8 of 21 patients who completed monitoring (20% vs. 38%, p = 0.44). Overall, a strategy of monitoring followed by tilt and EP testing was associated with a diagnostic yield of 50%, at a cost of $2,937 ± $579 per patient and $5,875 ± $1,159 per diagnosis. Conventional testing followed by monitoring was associated with a diagnostic yield of 47%, at a greater cost of $3,683 ± $1,490 per patient (p = 0.013) and a greater cost per diagnosis ($7,891 ± $3,193, p = 0.002).
Conclusions A strategy of primary monitoring is more cost-effective than conventional testing in establishing a diagnosis in recurrent unexplained syncope.
The current approach to the investigation of patients with unexplained syncope involves short-term electrocardiographic monitoring or provocative testing with head-up tilt and electrophysiologic (EP) testing (1–5). Recent advances in long-term monitoring with an implantable loop recorder have provided an opportunity to obtain a correlation between symptom and rhythm in the majority of patients (6–13). This new technology is associated with a higher diagnostic yield than conventional testing but requires a minor invasive procedure and is associated with a significant initial cost. In addition, a strategy of prolonged monitoring may defer a diagnosis because it is dependent on the recurrence of symptoms potentially associated with morbidity and rarely mortality.
We performed a prospective, randomized trial of prolonged monitoring versus conventional testing in 60 patients undergoing cardiac testing for unexplained syncope (13). The current study reflects the cost analysis of the two diagnostic strategies.
Patients referred to the Arrhythmia Service at London Health Sciences Center for investigation of syncope were approached to participate in a prospective, randomized trial comparing two diagnostic approaches to syncope. The details of this trial have been reported previously (13). In summary, 60 patients with recurrent unexplained syncope or a single episode of syncope associated with injury that warranted cardiovascular investigation were enrolled in a prospective, randomized trial comparing conventional testing to prolonged monitoring with an implantable loop recorder (Table 1). Before enrollment, patients underwent a clinical assessment, including postural blood pressure testing, a minimum of 24 h of baseline ambulatory monitoring or in-patient telemetry, and a transthoracic echocardiogram. Additional neurologic or cardiovascular testing before enrollment performed by the referring physician was recorded but not mandated by the protocol. Patients were excluded if their left ventricular (LV) ejection fraction was <35%, if they were unlikely to survive for one year, or if they were unable to provide follow-up or give informed consent. Patients with a presentation typical of neurally mediated syncope at the baseline assessment were considered to have this diagnosis and were excluded.
Patients were randomized to a conventional investigation strategy, including a two- to four-week period of monitoring with an external loop recorder, followed by tilt-table and EP testing (14,15). Patients randomized to a strategy of prolonged monitoring underwent implantation of a Reveal implantable loop recorder (Medtronic, Minneapolis, Minnesota). The details of the implantable loop recorder have been reported (10,16–18). Monitored patients underwent follow-up for one year. For the purpose of analysis in the monitoring strategy, a diagnosis was defined as obtaining a correlation between symptom and rhythm in patients during spontaneous syncope or presyncope that resembled the symptoms experienced before enrollment. In the conventional arm, a diagnosis was defined according to standard published criteria for tilt and EP testing (5,14,19–23).
If the assigned strategy did not provide a diagnosis, patients were offered a crossover to the alternate strategy. An implantable loop recorder was offered to all conventional patients immediately after tilt and EP testing was negative. Tilt and EP testing were offered to monitored patients if a diagnosis was not obtained after one year of follow-up. All patients provided written, informed consent. The protocol was approved by the University of Western Ontario's Institutional Review Board.
Patients were seen one week after loop recorder implantation for wound assessment and to reinforce patient understanding of the activation process. Subsequent follow-up occurred at 1, 2, 3, 6, 9, and 12 months. Patients were seen immediately after a symptomatic event. Patients undergoing conventional testing were seen at enrollment and in conjunction with prescribed testing. If undiagnosed patients refused crossover, they were asked to contact the study nurse in the event of recurrence.
Before being enrolled in the study, all patients were interviewed and medical records were reviewed to determine previous health care resource utilization with respect to the investigation of syncope. After enrollment, resource use was charted according to patient outcome in the study protocol. The costs of investigations were calculated based on the Ontario Health Insurance Program fee schedule for technical and professional fees and also included an estimate of materials, labor, maintenance, and overhead for hospital-based investigations (Table 1). For example, the Holter monitoring cost was calculated by adding the material cost of $5.00, technical and labor fees of $57.45, hospital overhead of $17.24, and a professional fee of $45.90, for a total of $125.29 ($80.10 U.S. dollars; 82.09 Euros, converted October 20, 2002). No service charge was included for the Holter cost calculation, because the current service contract is included in the hospital overhead calculation. For neurologic testing, material and maintenance data were unavailable. To approximate the cost of neurologic testing, peer-reviewed research billing rates were used to calculate cost. Overall costs and cost per diagnosis were estimated based on the primary diagnostic strategy, crossover, and combined strategies. All costs were considered direct medical costs that were assessed from a societal perspective and expressed in 2002 Canadian dollars. The incremental cost-effectiveness ratio (ICER) was calculated as the difference between the total cost of the monitoring strategy and the conventional strategy divided by the difference in diagnostic efficacy between the monitoring strategy and conventional strategy. All measures were calculated per person.
Continuous variables were compared by the Student ttest. Categorical variables were compared by the chi-square test. A p value <0.05 was considered statistically significant.
Sixty patients with unexplained syncope were randomized in the study. There was no difference in baseline variables between the groups (Table 2). Structural heart disease was present in 23 patients (38%). Left ventricular function was well preserved in most patients, with an ejection fraction <50% in only six patients. Before enrollment, neurologic testing was performed in 36 patients (60%), with brain imaging in 25 (42%) and an electroencephalogram in 19 (32%). The cost of neurologic testing was $475 ± $295 per patient in the 36 patients who were tested and $285 ± $326 for all 60 patients (Table 3). The overall cost of testing before enrollment was $1,327 ± $503 (median $1,241), with a range of $507 to $2,466.
In the 30 patients randomized to a strategy of monitoring, a diagnosis was obtained in 14 (Fig. 1). This included bradycardia in 10 patients, supraventricular tachycardia in one patient, and oscillation of sinus rates in keeping with neurocardiogenic syncope in the remaining three patients. In 16 patients, symptoms did not recur during the 12-month follow-up period. One patient failed to appropriately activate the loop recorder after an episode of syncope, and no further episodes occurred after the failure. The cost of a primary strategy of monitoring was $2,731 ± $285, and the cost per diagnosis was $5,852 ± $610 (Table 3). In the 30 patients in the conventional arm, a diagnosis was obtained in only six patients (20% vs. 47% by monitoring, p = 0.029). One patient had symptomatic third-degree atrioventricular (AV) block while wearing an external loop recorder; two patients had a positive tilt test; and EP testing demonstrated poor AV node function in two patients and inducible ventricular tachycardia in one patient. The cost of the investigation with the conventional strategy was significantly less than investigation with the primary strategy of monitoring ($1,683 ± $505, p < 0.0001), but the cost per diagnosis was significantly greater ($8,414 ± $2,527, p < 0.0001). The ICER for an immediate strategy of monitoring was $3,930.
Five of the 16 patients who remained undiagnosed after one year of monitoring consented to a crossover to conventional testing. Because of the previous monitoring, patients did not undergo short-term monitoring with an external loop recorder. Tilt testing was negative in all five patients. Electrophysiologic testing induced sustained AV node reentrant tachycardia associated with hypotension in one patient. This patient underwent slow pathway ablation.
Twenty-one of 24 patients with negative conventional testing underwent implantation of a loop recorder. A diagnosis was obtained in eight patients, and symptoms did not recur in the remaining 13 patients during 12 months of follow-up. In the eight patients who were diagnosed, bradycardia was noted in four and tachycardia in two. Two patients had sinus rhythm recorded during syncope, with phasic motion artifact on the recorded signal suggestive of seizure activity (24). Neurologic consultation led to anticonvulsant therapy, with resolution of symptoms. The cost and cost per diagnosis in the crossover patients is shown in Table 2.
By combining the primary strategy with crossover, the overall likelihood of being diagnosed with a strategy of monitoring was 43% (22 of 51) compared with 20% (7 of 35) with conventional testing (p = 0.026). Combining the primary strategy with crossovers allowed the overall diagnostic yield for each strategy and the cost of each approach to be calculated (Fig. 2). The overall diagnostic yield was comparable between strategies (50% with primary monitoring vs. 47% with primary conventional testing), but the cost of testing was reduced from $3,683 ± $1,490 in the conventional group to $2,937 ± $579 in the monitored group (p = 0.013). Furthermore, the cost per diagnosis was reduced from $7,891 ± $3,193 in the conventional group to $5,875 ± $1,159 in the monitored group (p = 0.002). This resulted in a negative ICER ($−22,380), implying actual cost savings with the strategy of monitoring.
The cost per diagnosis for individual tests was calculated. Tilt testing was most cost effective at $2,901 per diagnosis, followed by the implantable loop recorder at $6,158 per diagnosis. External loop recorders ($10,061/diagnosis) and EP testing ($10,700/diagnosis) were associated with much greater average costs.
We have shown that a primary monitoring approach to patients with unexplained syncope undergoing cardiovascular testing is both diagnostically efficient and cost effective. Despite the relatively high cost and minimally invasive nature of an implantable loop recorder, primary monitoring provided a diagnostic approach that yielded more diagnoses than conventional testing, with favorable cost-effectiveness. The initial cost of monitoring was greater than that of conventional testing but was justified in light of its greater yield. Performance of conventional tests leads to a high incidence of crossover to the monitored approach, thereby increasing cost.
This finding does not negate the role of conventional testing in selected patients when the yield is likely to be greater than that seen in the current study. Patients capable of using an external loop recorder should undergo a trial with this device before an implanted device is inserted, because of its non-invasive nature. Similarly, tilt-table testing may have a role in select individuals when neurocardiogenic syncope is suspected. This is borne out by the cost per diagnosis of individual tests, supporting the role for tilt testing even if the yield is relatively low. We excluded patients with a clinical presentation suggesting a high probability of neurocardiogenic syncope, because we felt it was not appropriate to consider invasive testing in the context of a strong clinical diagnosis. This is reflected in the age of the patients in the current study. The results of the current study cannot be generalized to other populations with syncope. This is particularly true in patients with poor ventricular function who were excluded. Electrophysiologic testing plays a much greater role in these patients because of concern of life-threatening ventricular arrhythmias, the potential risk of sudden death during recurrent symptoms, and the greater yield of EP testing in this population (25–30).
The main driver of cost in the conventional arm was EP testing, which has a diminishing role in patients with syncope (22). Although it retains a role in patients with structural heart disease, emerging evidence suggests there is an unacceptable recurrence and sudden death rate after negative EP testing in patients with syncope in both ischemic and non-ischemic cardiomyopathy (25–30). The current study and several others have demonstrated that EP testing has a low yield in patients with preserved LV function and should be considered only for selected patients when an inducible abnormality is likely or when it is clinically unacceptable to wait for a recurrent event. In the latter population, prolonged monitoring appears to be the approach of choice after non-invasive testing. Clearly, the order of testing influences both the diagnostic yield and cost-effectiveness, as demonstrated by the high cost per diagnosis of EP testing in the current study.
The cost of the strategy of prolonged monitoring was predominantly influenced by the cost of the implantable monitor. The cost estimate in this relatively small patient population reflected the cost of device implantation, follow-up, and removal, but it did not take into account the infrequent complication of infection seen with all implanted devices. Prophylactic antibiotics were administered to all patients, in keeping with our practice in pacemaker implants. Rare complications such as device migration, erosion, and intolerable pain were not experienced. These events are so infrequent that they would be unlikely to influence the overall estimate of the cost of monitoring.
The cost of investigations before being enrolled in the study reflected the common clinical practice of performing low-yield cardiovascular and neurologic testing before being referred to a cardiologist or electrophysiologist. Neurologic testing was performed in 60% of patients. The costs in the current study were similar to those seen in our previous study that evaluated costs in referred patients (31,32)and considerably less than those reported by Calkins et al. (33).
The current trial enrolled patients with recurrent, unexplained syncope that warranted referral to a cardiologist. The results may not be applicable to other patient groups with syncope, particularly patients who present with an initial event that may benefit from non-invasive conventional monitoring. Crossover in the current study was offered but not mandated, because we thought that patients may not be willing to pursue a diagnosis any further after having negative test results in the primary arm. This was the case in very few of the patients for whom the conventional approach failed, but in the majority of those who underwent a year of negative monitoring. This reflects a practical approach that is in keeping with usual clinical practice. Efficacy and cost analysis convincingly demonstrated the benefit of monitoring in the primary strategy, with consistent findings in those who crossed over.
Electrophysiologic testing is often withheld in patients with syncope and normal LV function. We chose to include EP testing because it has been upheld as the test of last resort in patients with refractory symptoms and no diagnosis (5,14,22,23). Excluding this would have reduced the sensitivity of the diagnostic protocol and would have withheld this form of testing from the 38% of study patients who had some form of structural heart disease without a marked reduction in LV function.
Finally, the costs are calculated in Canadian dollars, reflecting Ontario-based fee codes. Although the cost of health care varies between systems in other countries, the relative cost of the two strategies is likely to be similar. The relative cost of invasive, non-invasive, and out-patient testing between health care systems may have a minor influence on the costs of the two strategies.
A strategy of prolonged monitoring is a more cost effective and efficient diagnostic approach than conventional testing in patients with recurrent, unexplained syncope with preserved LV function. The strategy of primary monitoring significantly reduced cost by $2,016 (p = 0.002).
The authors thank Bonnie Spindler for her assistance in patient recruitment, education, and data collection, and Medtronic for providing loop recorders for the 30 patients who underwent primary monitoring.
☆ This study was supported by grant NA3397 from the Heart and Stroke Foundation of Ontario. Dr. Krahn is a Research Scholar of the Heart and Stroke Foundation of Canada. Drs. Klein and Yee are consultants to Medtronic.
- incremental cost-effectiveness ratio
- left ventricular
- Received December 5, 2002.
- Revision received January 17, 2003.
- Accepted January 30, 2003.
- American College of Cardiology Foundation
- Brown A.P.,
- Dawkins K.D.,
- Davies J.G.
- REVEAL Investigators,
- Krahn A.D.,
- Klein G.J.,
- Yee R.,
- Takle-Newhouse T.,
- Norris C.
- Krahn A.D.,
- Klein G.J.,
- Yee R.,
- Skanes A.C.
- Klein G.J.,
- Gersh B.J.,
- Yee R.
- Krahn A.D.,
- Klein G.J.,
- Norris C.,
- Yee R.
- REVEAL Investigators,
- Seidl K.,
- Rameken M.,
- Breunung S.,
- et al.
- Benditt D.G.,
- Ferguson D.W.,
- Grubb B.P.,
- et al.
- Zipes D.P.,
- DiMarco J.P.,
- Gillette P.C.,
- et al.
- Andrews N.P.,
- Fogel R.I.,
- Pelargonio G.,
- Evans J.J.,
- Prystowsky E.N.
- LeLorier P.,
- Krahn A.D.,
- Klein G.J.,
- Yee R.,
- Skanes A.C.
- Link M.S.,
- Costeas X.F.,
- Griffith J.L.,
- Colburn C.D.,
- Estes N.A.,
- Wang P.J.
- Mittal S.,
- Iwai S.,
- Stein K.M.,
- Markowitz S.M.,
- Slotwiner D.J.,
- Lerman B.B.