Littmann’s Sign

In 2007, Littmann and colleagues presented a novel EKG indicator of hyperkalemia based on the computerized “double counting” of heart rate. Over a 13-year period they identified 33 cases in which the GE-Marquette computerized 12SL EKG interpretation algorithm “double counted” or “near double counted” the actual heart rate seen on electrocardiogram. All 33 patients had hyperkalemia (between 5.3-8.8 mEq/L K+) as confirmed by serology taken within two hours of the double-counted EKG recording.

Littmann’s sign can be seen in the following EKG, initially presented here as a study in hyperkalemia in September of 2010.

A 65 year-old Caucasian man with sepsis and HHNKC; the potassium was 7.7mEq/L. The heart rate is 72 bpm; the computer counts 137 bpm.

Although the GE algorithm is proprietary and unavailable for analysis, they argue that, “The QRS width and axis measurement by the interpretation software suggested that, on many occasions, the computer recognized the T waves as being the QRS complexes….” (p.586)

Littmann et al. conclude stating, “Although interpretation software double counting of heart rate appears to be quite specific for hyperkalemia, its sensitivity is almost certainly very low.” (p.586)

While this research represents an intriguing new insight, it leaves many open questions. How were these 33 EKG cases identified? Were certain populations screened for this anomaly? How many total EKGs were reviewed in the course of this investigation?

Littmann claims that double counting “appears to be quite specific” presumably because all 33 EKGs were associated with underlying hyperkalemia. Yet without insight into the methods used to assemble this case series, I remain skeptical that these investigators did not succumb to confirmation bias in the process of collecting their EKGs. The existence of non-hyperkalemic double counting is not discussed in this publication and no differential diagnosis is presented regarding alternative etiologies.

In fact, concerning alternative etiologies and the specificity of double QRS counting, it turns out that this phenomenon has been extensively described in the pacemaker literature for over a decade. A Pubmed search using the combined terms “double counting QRS” produces numerous references. (Al-Ahmad A, Barold SS, Boriani G)

I reviewed 22 pacemaker EKGs collected over a 10 month period, recorded pre-hospitaly using the same GE-Marquette algorithm. This is what I found:

Looking further through ~50 EKGs collected over the same 10 months I found this as well:

In the first pair of EKGs, the computer appears to be confusing the pacemaker spike for a QRS complex; this is clearly “double counting of the QRS.”

The second pair is more complicated. This is not technically “double counting” of the QRS; there is not a 1:1 relationship between each QRS and a particular artifactual deflection. This is over-counting with a coincidental but not precise 2:1 relationship between the total count made by the computer and the total number of true QRS complexes. The interpretation of “Atrial Fibrillation” supports this in that it suggests that numerous complexes are counted during the first 5 seconds and then much fewer in the last 5. Although Littmann’s sign is not technically present here, the 12-lead EKG must be scrutinized to reach this conclusion. This seems problematic given Littmann’s stated goal of elucidating an “objective”, and “easily identifiable” indicator. (p.586)

No clinical data is available on these EKGs and it cannot be proven that these patients did not in fact have elevated potassium.

While producing these two hypothetical “false positives” hardly constitutes significant evidence, it is interesting that such counterpoints to Littmann’s specificity claims were so easily identified. Of further concern is the distinction between perfect double counting as seen in the pacemaker case and “near double counting” as seen in both the hyperkalemia case from Sept. 2010 and artifact case. What bearing this has on Littmann’s argument remains unclear.

Although a novel indicator, I am not sure that there is as yet persuasive evidence for what exactly Littmann’s sign indicates or how often this indication is specific for any one underlying clinical etiology.

References

Littmann L, Brearley WD, Taylor L, Monroe MH. Double counting of heart rate by interpretation software: a new electrocardiographic sign of severe hyperkalemia. Am J Emerg Med 2007;25:584-90.

Tomcsányi J, Wágner V, Bózsik B. Littmann sign in hyperkalemia: double counting of heart rate. Am J Emerg Med. 2007 Nov;25(9):1077-8.

Al-Ahmad A, Wang PJ, Homoud MK, Estes NA 3rd, Link MS. Frequent ICD shocks due to double sensing in patients with bi-ventricular implantable cardioverter defibrillators. J Interv Card Electrophysiol. 2003 Dec;9(3):377-81.

Barold SS, Herweg B, Gallardo I. Double counting of the ventricular electrogram in biventricular pacemakers and ICDs. Pacing Clin Electrophysiol. 2003 Aug;26(8):1645-8.

Boriani G, Biffi M, Frabetti L, Parlapiano M, Galli R, Branzi A, Magnani B. Cardioverter-defibrillator oversensing due to double counting of ventricular tachycardia electrograms. Int J Cardiol. 1998 Sep 1;66(1):91-5.

Advertisements

4 responses

  1. Christopher

    I had a trauma patient the other day with a relatively low-voltage QRS in lead II with a T-wave amplitude at least as high as the QRS (due to his axis). This caused my LP12, which uses the GE Marquette 12SL, to double count the complexes. I switched to Lead I and had no problem.

    Upon arrival at the hospital the nurse asked me if my patient was tachycardic when I had him, which surprised me as his rate had been in the 60’s and 70’s for the duration of my patient contact. Upon looking at their Philips monitor, I noticed that it too was double counting in Lead II.

    I switched their device to Lead I and the problem was again solved.

    The software algorithms for QRS detection/counting are usually separate from the axis determination and representative beat classifications. I believe they use a model of rapid deflections of a certain duration and amplitude change, and thus would be fooled by sufficiently rapid deflections at near-physiologic distances from one another.

    March 18, 2012 at 4:01 pm

    • Thank you for your comments, Christopher, and I very much appreciate your perspective on this.

      My intuitions regarding the algorithmic QRS classification criteria are similar in that it seemed that deflection amplitude and duration rather than axis of deflection had to be the essential indicators. Your insight here carries much more weight than mine. Regarding duration as a criteria, the duration of the pacemaker deflections in the LP12 EKG I presented seem unusually long (~0.03ms), and I had wondered if this was playing a part in the mistaking of PM spike for endogenous depolarization. The PM EKG I present is not normal in either amplitude or duration of PM spike.

      Also of interest, the rhythm strips from the pacemaker EKG do not make the same error as the 12-lead interpretation– I am referring now to the EKG rhythm strip HR, not a HR from NIBP or SPO2.

      As you describe, however, I have several times noted that the hospital rhythm monitors were prone to double counting. I have seen it with a-flutter, where the HR reflected the flutter rate, and also with movement artifact– the RN humorously pointed to the blinking “VT” alarm and remarked, “V-Tach, Parkinsons type.”

      If you happened to run a 12-lead on your trauma case, was the 12-lead also in error, or just the rhythm strip?

      Thank you again for your comments and consideration!

      March 18, 2012 at 4:31 pm

      • Christopher

        12-Lead had the correct heart rate interestingly enough. But I believe the LP12 operates differently when all 12-leads are placed as the frequency filter changes. I think this altered the QRS/T-wave amplitudes enough to get it right. They were only subtly different in the rhythm strip.

        March 18, 2012 at 4:50 pm

      • Thanks for the follow up! And it is a good point you make about the filter!

        March 18, 2012 at 4:52 pm

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s