Saturday, March 24, 2012

Interesting new paper: in this study, even interventionalists had a hard time differentiating non-ischemic ST elevation from STEMI

Reference:  Tran V, Huang HD, Diez JG, et al. Differentiating ST-elevation myocardial infarction from nonischemic ST-elevation in patients with chest pain. Am J Cardiol 2011;108(8):1096-101.


Here is a link to the pdf.

Here is a summary:

The authors (cardiologists) looked at the ECGs and records of 240 consecutive cath lab activations for STEMI (primary PCI, pPCI).  They excluded patients with LBBB or paced rhythms and others whose STE criteria did not meet guideline-based criteria for acute STEMI.  They did detailed chart reviews to determine if there was or was not actually a STEMI, in retrospect (looked at cath results, echos, biomarkers, etc.).  They then showed the ECGs to 7 experienced interventional cardiologists and asked them to interpret the ECGs and asked whether they would recommend immediate pPCI, assuming that the patient had appropriate ischemic symptoms.  They were blinded to outcome and other patient data.

Of 84 subjects, there were 40 patients with a true STEMI and 44 without (13 of whom had NonSTEMI).  Recommendations for immediate PCI varied widely, from 33% to 75%.  Sensitivities were 53% to 83% (mean, 71%), specificities 32% to 86% (mean 63%), PPV 52%-79% (mean 66%) and NPV 67% to 79% (mean 71%).  When the readers chose non-ischemic ST elevation, LVH was thought to be the cause in 6% to 31% and old MI/aneurysm in 10% to 26%.

They show some very interesting tracings which are very similar to many I have shown on this blog.

This adds to the growing evidence that even among the "experts", ST elevation is very difficult to diagnose.

A previous study by Turnipseed et al.(1) showed that early repolarization was frequently misdiagnosed as STEMI and vice versa.  Cardiologists performed slightly better than emergency physicians.  Jayroe et al. (2) distributed difficult ECGs to 15 "expert" electrocardiographers and found that they had difficulty distinguishing STEMI from nonischemic ST elevation.

How would emergency physicians perform?  Or paramedics?  This is unknown, but to some extent it is much more important to those of us on the front line, and I wouldn't be surprised if many spend more time studying ECGs and are more expert than interventionalists.  The interventionalist has to respond to a cath lab activation.  The paramedics and EPs need to actually make the decision to activate.  We should have a lot of motivation to learn the differences between the true positives and the false positives.

Christopher Watford posted some very interesting data below.  I'll post part of it here:

---Dr. Paul R. Hinchey, Austin-Travis County's medical director, had a presentation at the 2012 Gathering of Eagles conference (presentation is available online) where he discussed the practical implications of using Cardiology to overread EMS 12-Leads. His service had received complaints due to the number of "false activations" given by his Paramedics.

---Dr. Hinchey implemented standardized activation criteria and had a panel of 3 cardiologists over-read the EMS activations to determine appropriateness, his findings were as follows for 90 EMS STEMI alerts:

- 87% (74/90) Positive to at least 1 cardiologist
- 63% (57/90) "True" Positives (2 of 3 cardiologists agreed)
- 43% (39/90) Unanimous "True" Positive
- 37% (33/90) "False" Positives (2 of 3 agreement)
- 18% (16/90) Unanimous "False" Positive

Bottom line: there is a long way to go in getting better at interpreting ST elevation on the ECG.  Rules to help do so could prove very valuable.  Some such rules already exist, such as the early repol vs. anterior STEMI rule and the LV aneurysm rule.

Early repol ruleStrictly speaking, this was not studied in ECGs with: 1) T-wave inversion, 2) coved (upwardly convex) ST segments, or 3) LVH.  Here is the formula; there is an excel spreadsheet down the right side of this blog: (1.196 x STE at 60 ms after the J-point in V3 in mm) + (0.059 x computerized QTc) - (0.326 x R-wave Amplitude in V4 in mm).  A value greater than 23.4 is quite sensitive and specific for LAD occlusion.

LV aneurysm rule:  There is one retrospective study (Smith SW. American Journal of Emergency Medicine 23(3):279-287, May 2005) showing that the T/QRS ratio is significantly greater in acute anterior STEMI than in old anterior MI with persistent ST Elevation. The best criterion for differentiating was the sum of STE in V1-V4 divided by the sum of the QRS's in V1-V4 (TV1+TV2+TV3+TV4 divided by QRSV1+QRSV2+QRSV3+QRSV4). If this value was greater than 0.22 vs. less than 0.22, then it is likely to be acute STEMI.  Another rule that was almost as good: if any one of the leads had a ratio of T to QRS greater than, vs. less than, 0.36, it was very likely to be STEMI.

References:

1. Turnipseed SD, Bair AE, Kirk JD, Diercks DB, Tabar P, Amsterdam EA. Electrocardiogram differentiation of benign early repolarization versus acute myocardial infarction by emergency physicians and cardiologists. Acad Emerg Med 2006;13(9):961-6.)

2. Jayroe JB, Spodick DH, Nikus K, et al. Differentiating ST elevation myocardial infarction and nonischemic causes of ST elevation by analyzing the presenting electrocardiogram. Am J Cardiol 2009;103(3):301-6.

4 comments:

  1. This is a longstanding position of personal faith for me. If you look at most of the published literature involving ECG interpretation, such as the decision rules you describe, it mainly comes from the world of emergency physicians and sometimes paramedics. With much respect to the cardiologists, they just aren't called upon as regularly to make the same types of clinical decisions using the ECG. Certainly there are many in EM who are also barely par with it -- the range is huge everywhere -- but I think the mean skill level there certainly tends to be higher.

    ReplyDelete
  2. Dr. Paul R. Hinchey, Austin-Travis County's medical director, had a presentation at the 2012 Gathering of Eagles conference (presentation is available online) where he discussed the practical implications of using Cardiology to overread EMS 12-Leads. His service had received complaints due to the number of "false activations" given by his Paramedics.

    Dr. Hinchey implemented standardized activation criteria and had a panel of 3 cardiologists overread the EMS activations to determine appropriateness, his findings were as follows for 90 EMS STEMI alerts:

    - 87% (74/90) Positive to at least 1 cardiologist
    - 63% (57/90) "True" Positives (2 of 3 cardiologists agreed)
    - 43% (39/90) Unanimous "True" Positive
    - 37% (33/90) "False" Positives (2 of 3 agreement)
    - 18% (16/90) Unanimous "False" Positive

    His takeaway was pretty much the same as yours. He did identify that the system needed at least 2 points of continuing education:

    1. QA/QI Feedback for EMS based on data quality, LBBB, LVH, early repol, and benign V1/V2 elevation
    2. QA/QI Feedback for the entire system based on the results from the cath

    What I've seen in my own system is that Paramedics are at least as accurate as our ED docs in terms of clinical context associated with the ECG. Where our system shines, I feel, is the strong QA/QI feedback loop that encourages provider education and a team approach, without which I do not think the system would perform as well as it does.

    ReplyDelete
  3. GREAT information with a number of important take-home points:
    i) There is tremendous inter- (and intra-) observer variability in the interpretation of ECGs - and as Steve & Christopher point out - this amazingly holds true for as fundamental an issue as determining whether to activate the cath lab!

    ii) Expertise may be obtained by a variety of providers - including cardiologists - ED physicians - EMS personnel - and other physician and non-physician groups. Given data from Dr. Hinchey that Christopher presents - it is hard to make a case for any one group as being "gold standard interpreters" - and I suspect there are true experts within each group of providers (as well as many within each group that are less than expert).

    iii) Tools (such as Steve Smith's formula with score) may help - and anything that helps is good. But nothing is perfect - not the ECG - not the interpretation by any interpreter - not the formula.

    iv) Sometimes a tincture of "time" may tell (serial tracings, troponins, clinical course, availability of prior tracings).

    v) In the best of hands - the correct decision can not always be made from the initial ECG - just like in the best of hands we'll never bat 1,000 in deciding which chest pain patients can go home.

    vi) Dr. Hinchey's 2 points of continuing education (in Christopher's comment) are true words of wisdom for continuing to improve the process.

    vii) THANKS to you all for your excellent and insightful comments.

    ReplyDelete