For example, that 150 bpm rate might be perfectly appropriate for a patient doing strenuous exercise. It would just be an exercise-induced sinus tachycardia. In fact, in such a patient, it would be a problem if the ventricular rate did not respond that way!
On the other hand, a 150 bpm rate might be caused by a ventricular arrhythmia. It could be provoking symptoms and might advance to faster and more dangerous even life-threatening rates.
Sometimes a rapid ventricular rate is just a ventricular response to an atrial tachycardia. However, if we as clinicians need an ECG along with our judgment, training, and experience to discern which rhythm disorder is which, how does an ICD know what arrhythmias to treat?
ICDs were designed to treat dangerous ventricular tachyarrhythmias, such as ventricular tachycardia (VT) and ventricular fibrillation (VF). ICDs tend to see cardiac activity strictly in terms of programmable rate ranges. So how does an ICD know when a 150 bpm rate is a supraventricular tachycardia (SVT) or a VT?
The distinction is an important one, because ICD therapy is designed to treat VT and VF. Shocking an SVT may or may not convert the rhythm, but SVT is not life-threatening. An ICD should not be wasting its battery energy by delivering therapy for SVTs, which do not require treatment. This saves device service life, but more importantly, spares the patient painful and unnecessary shocks. The industry euphemism for ICD shocks for SVTs is called inappropriate therapy.
However, there is something even more important to the patient than avoiding inappropriate therapy. It s assuring appropriate therapy. Discriminators not only prevent unnecessary shocks, they assure that patients get the life-saving therapy they need for VT.
SVT discrimination is not about treating VF. When a patient experiences VF, therapy is delivered as quickly as possible. SVT discrimination only applies to VT, which in itself can be a life-threatening arrhythmia. However, when we talk about device therapy, VF remains rightfully in a class by itself.
The challenge for the device inventors has been to find a way to minimize or prevent inappropriate therapy while assuring appropriate therapy for every VT. But how does a device know what s what?
Put in simple terms, the ICD does it the same way we clinicians do. It looks at the ECG and asks itself some simple questions. In fact, SVT discrimination features in modern ICDs are really nothing more than the questions we would pose when analyzing an ECG.
If you had a patient with a rapid ventricular rate, you would want to check out the ECG to see what the atria are doing with respect to the ventricles. If you have an atrial rate that is going even faster than the ventricular rate, then the patient has atrial fibrillation (AF), and the ventricular rate is likely to be the heart s attempt to keep pace with the atria. This strongly suggests you have an SVT with a rapid ventricular response.
A 1:1 response, that is, every ventricular beat linked to an atrial event, would make you think that this might be a sinus tachycardia. You would want to find out what the patient was doing when the ECG was taken. A sinus tachycardia for a patient doing a treadmill test is different than a sinus tachycardia in a little old lady bedridden in a nursing home.
If you see that the ventricles are going much faster than the atria, then it is clear that the atria aren t driving the rate. This would have to be VT.
ICDs do the same thing by comparing the heart s intrinsic atrial rate to the ventricular rate. They do this by sensing atrial data and concurrent ventricular data and making flow-chart type decisions: is A>V? Is A

However, while most clinicians would check out the relationship of atrial and ventricular rates first and foremost, we all can imagine (and maybe even have treated) cases where the patient had AF…but also a simultaneous VT. So how does the ICD know when a suspicious arrhythmia is really SVT and not VT?
That s where multiple discriminators come into play. Modern ICDs offer several discrimination algorithms which can be combined, the same way a clinician might check out several items on an ECG before rendering his decision as to the type of arrhythmic activity going on. One of the things we look at is the shape of the QRS. When an atrial beat conducts naturally to the ventricle, it gives the QRS a characteristic normal shape. But when a beat originates within the ventricle, the QRS has a wider morphology (Figure 2). If a clinician saw that these ventricular beats had an aberrant shape, he would know that this was not ventricular response to the atrium, but a true VT.

For many years, no device was capable of making this sort of assessment. An ICD could count intervals or measure rates, but it could not compare morphologies until about a decade ago. The technical challenge was further complicated by the fact that there is no generic conducted QRS morphology. Every patient is unique, and a conducted QRS in Mrs. Jones may look quite different than the same cardiac event in Mr. Smith. However, it has been observed that QRS morphology is relatively consistent within an individual patient.
ICDs with Morphology Discrimination are able to compare QRS complexes against a patient-specific template. In this way, the ICD can determine if the QRS complex looks like a VT or SVT, based on the same principle we use when we scope out an ECG. Yet the ICD does it more rigorously.
The ICD stores in its memory a QRS template of a beat known to be conducted. When properly programmed, an ICD with Morphology Discrimination will always have a recent snapshot of a conducted QRS complex in memory. The ICD is actually capable of periodically updating this template automatically, since QRS morphology can change over time, due to disease progression, drug interactions, and other factors. Once it has a pattern of how a conducted QRS beat should look, it is able to compare suspicious ventricular activity against the template and rate complexes as matches or non-matches. If the ventricular activity is a match, then the activity is presumed to be an SVT, which would inhibit therapy delivery. If the QRS complexes do not resemble the template of a conducted beat, then a non-match is declared, VT is diagnosed, and therapy would be delivered.
For patients with paroxysmal or persistent AF, AV rate branch (A>V) can only give us part of the picture. Combine this with Morphology Discrimination, and the device can determine if the QRS beat was conducted from the atrium or arose within the ventricle. The result: if the Morphology Discrimination gets a match (the beat was indeed conducted), therapy is inhibited. If the Morphology Discrimination can t match the morphology, then the beat is ventricular in origin, which means the arrhythmia is a VT and requires therapy.
Another good example of the utility of Rate Branch plus Morphology Discrimination occurs when a patient has a ventricular rate of 180 beats a minute where V=A. Most of us would recognize this as a possible sinus tachycardia. But what if it is actually VT with 1:1 retrograde conduction? In my experience, that occurs in about 15% of such scenarios, but how can an ICD make the determination? The device does it the same way we would: by looking first at the relationship of atrial to ventricular rates and then by looking at QRS morphology (Figure 3).

Another thing clinicians check out on the ECG is the regularity of the ventricular rhythm. By nature, AF is an erratic, irregular rhythm and ventricular response to AF tends to be irregular as well. VT tends to be fast but fairly regular, so you won t see wildly fluctuating R-R intervals. However, erratic R-R intervals are pretty typical for rapid ventricular response to AF.
The ICD uses a special algorithm called Interval Stability to analyze the patient s intrinsic activity. The device looks at a series of R-R intervals on the rhythm strip and compares the difference between intervals (known as a delta) against a programmable delta value. (By programmable, I mean that the clinician can adjust the ICD to determine what delta value is most appropriate. The default value of the delta is 80 ms.) If the R-R interval changes by more than the delta value over a programmed number of intervals, then the rhythm is determined to be irregular. This means that AF is involved and the arrhythmia is an SVT (no therapy delivered).
Likewise, if the R-R interval changes less than the delta value over the programmed number of intervals, then the arrhythmia is regular, which means it is a VT and therapy is required. (See Figure 4 for an example of Interval Stability in action.)

Another area of interest when trying to determine from a tracing whether an arrhythmia is VT or SVT is to check out how quickly the arrhythmia started. An arrhythmia that abruptly commences is far more likely to be a VT than an SVT, which tends to start gradually and warm up.
The ICD does this as well, using a Sudden Onset algorithm. The Sudden Onset algorithm has a programmable delta value that might be set to something like 100 ms. When an arrhythmia occurs, the ICD starts to see how quickly the change from non-tachy intervals to tachy intervals occurred. If the change occurred more quickly than the programmable delta, Sudden Onset is diagnosed, meaning the device decides it must be a VT. This will launch therapy delivery. On the other hand, if the change from non-tachy state to tachy state occurred more slowly than the programmed delta, then the ICD decides that the arrhythmia is likely an SVT and therapy is inhibited.
While SVT discriminators are powerful tools, they should be used with careful consideration.
Before you go overboard with SVT discrimination, how often has the patient been shocked? SVT discriminators are very important in patients who get a lot of shocks or who are very distressed by shocks (even if they only have a few). SVT discrimination is less important for those patients who are never or rarely shocked.
Before you select discriminators, what do you know about the documented arrhythmias in this particular patient? If the VT can be irregular, you should avoid Interval Stability. If VT sometimes starts gradually and accelerates, Sudden Onset may not work as well. In other words, match the discriminator to the patient.
Patients with known AF should not use Rate Branch without Morphology Discrimination, since Rate Branch can overlook a VT that occurs simultaneously with AF.
Does the patient get VTs that rapidly accelerate into VF? In such cases, you want to keep the device very likely to delivery therapy and probably won t want to use too many (if any) SVT discriminators.
Has the patient received inappropriate shocks? You may have to download information from the device to know for sure. If inappropriate therapy has occurred, then you should use SVT discrimination.
If SVT discrimination has been programmed on for a while, check the device diagnostics to see how it has worked. Nothing succeeds like success, so stick with the SVT discriminators that have worked well previously.