Cost-Effectiveness of TEE in Atrial Fibrillation


Synopsis: TEE-guided early cardioversion is a reasonable cost saving alternative to conventional therapy for hospitalized patients with atrial fibrillation for more than two days and no contraindications to cardioversion.

Source: Seto TB, et al. J Am Coll Cardiol 1997;29: 122-130.
Considerable data now suggest that new onset atrial fibrillation is an urgent therapeutic issue, not only to prevent stroke, but to restore atrial function before a refractory atrial myopathy occurs. In this regard, delays for anticoagulation may help perpetuate atrial fibrillation. Thus, Seto et al evaluated the cost-effectiveness of transesophageal echocardiography (TEE)-guided early cardioversion as compared to conventional therapy with warfarin for one month and to a third strategy with transthoracic echocardiography (TTE) done before TEE.

In both the strategies employing TEE, if no atrial thrombi are detected, cardioversion is performed immediately. All three strategies started anticoagulation and continued it for one month after cardioversion. The study was based on a computer decision model using patient data from published reports and a prototypic 70-year-old patient admitted to a hospital with atrial fibrillation of more than two days in duration who is a candidate for elective cardioversion.

One reason for the TTE/TEE strategy comparison was to detect underlying pathology such as mitral stenosis which might eliminate the patient from immediate cardioversion. The only complication of warfarin included in the model was major hemorrhage; the only complication of cardioversion was stroke; and the only complication of TEE was esophageal perforation. Effectiveness was measured as quality-adjusted-life-years (QALY). Cost was based on true costs, not charges. Finally, sensitivity analyses were done by varying baseline probabilities, utilities (quality of life), and costs over wide ranges.

The TEE strategy had the lowest cost ($2774 vs conventional therapy at $3070 and TTE/TEE at $3106). Effectiveness did not differ among the three strategies (8.49 vs 8.48 and 8.48 QALY, respectively). Sensitivity analyses showed that the TEE strategy always dominated the TTE/TEE strategy because of the increased cost of the TTE without any evidence of added benefit. The sensitivity analyses showed variable dominance of the TEE strategy vs. conventional therapy; this was based upon the relative risk of thromboembolism with a negative TEE vs. one month of warfarin.

As long as the relative risk of stroke with TEE-guided cardioversion is less than 0.95, the TEE strategy is superior, but if the RR is greater than 0.95, the TEE strategy remains less costly, but also less effective. The strengths of these two strategies also depend on the probability of hemorrhage on one month of warfarin but are less sensitive to the risks of TEE and other factors.

The conventional strategy also included a TTE. If this test is eliminated because of uncertain benefit, the cost of conventional therapy drops below that of TEE ($2750 vs 2774), but the TEE strategy remains more effective.

The authors conclude, on the basis of their decision-analytic model and published data, that TEE-guided early cardioversion is a reasonable cost saving alternative to conventional therapy for hospitalized patients with atrial fibrillation for more than two days and no contraindications to cardioversion. This strategy is especially useful in patients with an increased risk of hemorrhage from warfarin. Also, the data strongly suggest that the use of TTE adds little to any strategy and should be eliminated in future clinical trials.


The results of this computer analysis are very seductive: cost savings with less risk of stroke and hemorrhage using the TEE-guided strategy. Also, physicians dislike managing patients on warfarin, and this strategy cuts the warfarin time in half. Computer-model studies like this seem ideal. They are less costly than clinical trials, don’t subject human subjects to risk, and are evidence-based. It is likely that managed care companies will implement new policies based upon such studies. However, without meaningful tort reform, they are a trap for physicians. We are held to the community practice standard, and conventional therapy still rules.

Also, because of peculiarities in ERISA laws that our trial-lawyer-dominated Congress won’t change, physicians are still liable, and the managed care organizations that implement new policies are not liable.

Sure, the risk of stroke has been zero in some small trials of TEE-guided therapy, but I can’t believe that all atrial thrombi in all atrial fibrillation patients can be detected by TEE—and it only takes one stroke to make your life miserable with a malpractice suit. Sure, the thrombi missed would be small, but small thrombi to the brain can cause disabling consequences.

Consider the professional cellist I saw who went into atrial fibrillation and permanently lost hearing in one ear. She could no longer be employed in an orchestra, so this was a career-ending event. Fortunately, all this happened before I saw her. Community standards could gradually change, but do you want to be the vanguard?

On the other hand, this computer analysis probably minimizes the risk of the conventional approach. These were hospitalized patient data where efficient anticoagulation, tight INR control, and expeditious cardioversion were assumed. In many situations, the patients are outpatients. Warfarin therapy is extended for 6-8 weeks because of trouble adjusting the INR, the logistics of visits, and a lackadaisical attitude by physicians and patients about the urgency of an "elective" cardioversion.

Thus, real patients are often exposed to warfarin for longer than a month; this leads to an increased risk of hemorrhage and delayed cardioversion. The latter issue is difficult to evaluate, but there is growing evidence that the longer the patient is in atrial fibrillation, the less likely it is that cardioversion will be successful.

Some physicians have expressed concern about an atrial myopathy occurring due to prolonged fibrillation, but data are scant. Thus, we don’t know if delaying cardioversion for 2-4 weeks is a major problem, but it is a concern.

Also not addressed in this study is the role of drug therapy or spontaneous cardioversions. There is considerable debate about the risk-benefit ratio of starting type I or III antiarrhythmic agents, yet many physicians still do so. These agents may expose patients to the risk of premature cardioversions under conventional therapy, not to mention their own adverse effects.

Finally, the analysis suggests that we do away with TTE in atrial fibrillation patients. Although TTE probably adds little to the well-characterized clinical trial patient, in less experienced clinical hands it may be of value to eliminate cardioversion as an option for some patients, thus saving money and reducing risk. Also, the detection of mitral valve disease and other entities, which may be missed in some clinical situations, is of great benefit to the patient.

In conclusion, this is a provocative study that should cause us to critically reexamine our strategies for the management of atrial fibrillation. If large clinical trials of this issue are not forthcoming, I recommend careful consideration of the recommendations and the cautious adoption of them in subgroups of patients where they make the most sense (i.e., patients at high risk of hemorrhage). In this way, the community standard of practice can be slowly changed if initial outcomes are good.