Comments on the use of
"interpretative" software for
computer/machine reading of ECG / EKG recordings
by James W. Grier, Department of Biological Sciences,
North Dakota State University
Fargo, ND 58102-3400, USA
Virtually all ECG machines, whether "interpretive" or "non-interpretive", provide basic measurements of the ECG trace (such as P, PQ, QRS, QT/QTc, P/R/T axes, HR, and others). However, only "interpretive" ones do diagnostic "interpretation". That can provide insights or serve as a teaching tool in some circumstances. But it also poses several problems. Machine/computer diagnoses are frequently wrong. At best, they are distracting and can be a nuisance, in my opinion.
Computers are notorious for misdiagnosing ECGs and they ALL do it to some extent. I think that not having a computer interpretation is better than getting a wrong one. Automatic, machine-generated interpretations are potentially dangerous and can put persons in jeopardy, both by missing medical conditions that might be present and by identifying conditions that do not exist, that is, by creating a higher frequency of false negatives and/or false positives than would occur when read by a trained, experienced cardiologist. All machine interpretations must be over-read by a physician anyway, so it seems best that a physician should read them properly in the first place. If a non-cardiologist physician is not sufficiently trained and experienced to read ECGs themselves (and might do worse or at least no better than a computer), they should obtain the assistance of a cardiologist in confirming their readings or actually doing the readings.
Machine interpretation of ECGs can be distracting. Proper readings of ECGs should be systematic and thorough, as taught by the various texts for ECG reading (see references below). A machine interpretation draws one's attention to what the computer says so that one has to determine whether the machine is correct or not, plus one can easily overlook other things that should be considered, or otherwise not read the ECG properly.
Also, the computer is unable to consider the patient's/subject's history and background. (Athletes, for example, have ECGs that can confuse even cardiologists and are very likely to confuse computers.) Many persons have relatively unique ECG patterns, almost like fingerprints, and knowledgeable, experienced humans are far better than machines at sorting through the variety and considering individual situations.
Even the best interpretive systems come with acknowledgements and descriptions of the potential problems of automatic interpretation. For an example, using the Welch Allyn interpretive software manual, click here.
Thus, I personally prefer (both for my own use and in classes with students) and recommend NOT using interpretive machines/computers. If you have such a system, I recommend turning off the interpretive function if possible or at least ignoring it. Medical, diagnostic interpretation of ECGs should be done only by an experienced, qualified human such as a cardiologist or oneself if you know what you are doing. Otherwise, get help from someone who does have the background and experience.
References and links concerning machine interpretation of ECGs:
- A classic, large-scale study of computer vs human interpretation of ECGs (New England Journal of Medicine, 1991), click here. The study showed that (1) computer interpretation was considerably less accurate than cardiologists and (2) some computer interpretation programs were much better than others and a few even approached human accuracy, but there was much variation in accuracy among different computer programs. How well an interpretive system does depends on which one it is ... and they ALL make mistakes.
- A more recent study of the role of computerized interpretation in errors by non-cardiologist physicians vs cardiologists (JAMIA, 2003), click here. This study yielded mixed results. Computer interpretation provided general improvement in readings by non-cardiologist resident physicians when the computer was correct, but the non-cardiologist doctors were also easily led astray when the computer was wrong.
- Guidelines for standardization and interpretation of ECG (JACC, 2007), click here
- International Society for Computerized Electrocardiology, click here.
General references and further information:
These books can be ordered through any book store or online, such as Amazon.com. Note: Also be sure to see the relevant sections on the heart in whatever basic college biology or human anatomy & physiology textbook that you happen to have.
The following is only a selection of ECG / EKG books currently available, but I have found these to be very helpful and useful. I usually like to have more than one book on any particular topic. If one were to have only one introductory book for ECGs, I would recommend Dubin 2000, Garcia and Holtz 2003, or Thaler 2003. These three are very different from each other, but all are good as first or only introductory books for learning how to interpret ECGs. If you want to go deeper into the subject and have a lot more actual ECG examples to look at and puzzle over, then Garcia and Holtz 2001, Garcia and Miller 2004, and Green and Chiaramida 2003, are good; they are more appropriate as subsequent books, not first or only ones. There are also many more books available, as can be seen on the link above to Amazon.com.
Dubin, D. 2000. (6th edition) Rapid Interpretation of EKG's. COVER Publ. Co. Tampa, FL.368 pp. ISBN 0-912912-06-5. A classic and very popular book for learning ECG's. It uses a fairly unique interactive, page by page fill-in-the-blanks self-test system throughout the book.
T.B. Garcia and N.E. Holtz. 2001. 12-lead ECG: The Art of Interpretation. Jones & Bartlett Publ. Sudbury, MA.236 pp. ISBN 0-7637-1284-1. (Also see the next two.) This book is a fairly complete teaching and reference text that includes three levels of difficulty: basic, intermediate, and advanced. The levels of difficulty are mixed together by topic but color coded so the reader can cover what is appropriate to their level and not get lost in more difficult topics. It is profusely illustrated and includes many examples of actual ECGs. The book has a horizontal, rather than standard vertical format, which makes it handy for including full-scale ECGs, but it is awkward to read.
T.B. Garcia and N.E. Holtz. 2003. Introduction to 12-lead ECG: The Art of Interpretation. Jones & Bartlett Publ. Sudbury, MA.536 pp. ISBN 0-7637-1961-7. Essentially the basic material from their more complete 2001 book. The first 72 pages, all basic material, of both books are the same. After the first 72 pages, this book then distills the subsequent material with reworking and additional illustrations. It is a less bulky book, at less than half the thickness and weight of the 2001 book. It shares with the 2001 book the same horizontal and awkward format (but also many of the full-scale ECGs).
T.B. Garcia and G.T. Miller. 2004. Arrhythmia Recognition: The Art of Interpretation. Jones & Bartlett Publ. Sudbury, MA.633 pp. ISBN 0-7637-2246-4. This book is by the same first author as the two above, but it is focused on arrhythmia topics. It begins with much of the same basic material, so any of these three books would be equally useful for introductory topics, but even that material has been reworked and improved slightly (it is good even in the 2001 book, but even better here). I would regard this book as a companion to either of the other two, although it would probably be best with the more thorough 2001 version for the more advanced topics (of which much in the arrhythmia book would be).
J.M. Green and A.J. Chiaramida. 2003. 12-lead EKG Confidence: Step-by-Step to Mastery. Lippincott Williams & Wilkins. Philadelphia, PA. 435 pp. ISBN 0-7817-3921-7. A useful book including many examples with their interpretations. The writing is perhaps less readable at the introductory level than the others in this list and I would not recommend it as the first or only book. But it is good for the examples and a somewhat different approach, after using one or more of the other books to begin with.
Thaler, M.S. 2003. The Only EKG Book You'll Ever Need. Lippincott Williams & Wilkins. Philadelphia, PA. 317 pp. ISBN 0-7817-3921-7. A very readable, concise, and helpful introduction. It is well illustrated with segments that pertain to the points at hand, although it lacks examples of complete 12-lead ECGs.
Internet web sites ...
There are a large number of web sites currently available on the topic of ECG / EKG, including many instructional ones and several that provide numerous examples of actual ECGs. A quick search on Google will turn up lots of them!
An introduction to ECGs that I wrote, click here.
Here is a sample of others that I have found useful and interesting --
Basics of ECG -- including a lot of excellent introduction to the heart itself, plus extensive ECG information
An online EKG course
ECG Learning Center tutorial
A guide to reading and understanding the EKG
Interpreting ECG -- another great site, including an excellent SUMMARY table (scroll down through the page)
www.12LeadECG.com -- an extensive, very useful site; designed to accompany the books by Garcia et al. but accessible for everyone
A large collection of ECGs with examples of most of the abnormal conditions
Interactive EKG quizzer (starts with a simple quiz; scroll down in it for additional quizzes)
Some challenging ECG cases, with explanations
... and there are many more, including links within the sites above. You can continue searching on your own.
To contact the author, e-mail: email@example.com
Author's web page: James W. Grier
Department of Biological Sciences, NDSU
Last update: 6/16/2008