Circulation 36,825 • Volume 17, No. 3 • Fall 2002

Computerized Anesthesia Record: Value Recognized, Terminology Scrutinized

David W. Edsall, MD

To the Editor

I am pleased to see the endorsement by the APSF of computerized records for anesthesia in the APSF Newsletter. As the second anesthesia organization to do so (behind the STA in 1994 or 1995), I hope this will finally give the boost needed for the industry. The industry should be allowed to market these two endorsements. The more important news is the effort toward the data dictionary. This is a difficult task that is long overdue. If successful, this should really get the ball rolling toward dramatically improved care. I hope the participants can keep focused on the definitions (which go far beyond outcomes) and not on the structure or the coding issues. Let the companies each come up with their own “better” way of doing that. However, I am still distressed to see the APSF referring to the technology as “automated records.” No other branch of medicine does this.

They should be called computer-based medical records (CBPRs) or computerized patient records (CPRs). Although parts of the records are automatically recorded, the clinician manually annotates the vast majority of the fields in the record (>70%). Personnel names, drug name, dose route, intubation and spinal information, incision times, tourniquet pressure and time, descriptive notes about the condition of patient, and the state of the surgery are just some of those fields. Furthermore, the rare (1 in 300 anesthesia minutes) piece of automatic data that is erroneous can be annotated in a few seconds. All of this annotation takes far less time than accurately recording the rest of the monitor data by hand. The record is far from automatic. That “A” word and any acronyms using it, like ARK, cause a disservice to the promotion of the technology.

This name for the technology is not just an issue of semantics. It is sales, marketing, and politics. The very same newsletter in which you have endorsed the technology shows the results of your semantics. Namely, the commonly and incorrectly expressed view of Dr. Metz1 that computer records are legally risky proves this. This unsubstantiated fear is promulgated by the use of the term automatic, which implies that the data on the record are out of the control of the clinician. Ergo, this creates the false assumption of legal risk for showing some bad vital signs that did not occur. In my view, this issue of terminology for the technology and the lack of the data dictionary are the two main reasons (out of five) why the industry has had so many failed companies. As a result, many more patients suffer bad outcomes such as emesis and myocardial infarction now than would have occurred had we adopted the technology a decade ago. If that statement is not true, why endorse the technology as a patient safety issue?

If the industry and the APSF would use the more accurate term “anesthesia computer records (ACR)” or some phrase like “computerized patient record for anesthesia (CPRA),” one might not have to keep beating down this dead horse. This issue has hung on since my former department started using computer records in 1988. There have been many white papers, textbook comments, and expert panels at the ASA, STA, and other meetings that have squashed this notion of increased legal risk and yet it hangs on. I have spoken to no JD or JD/anesthesiologist who thinks CPRAs are legally risky. I would wager that Dr. Metz has never done a case with a computerized record keeper and used that record as the legal record. One-day test trials, without proper training, in which the record serves as a duplicate to the hand record, do not count. Insurance companies like St. Paul and Preferred Physicians Mutual encourage their use. Why, if they are legally risky? Maybe the APSF could sponsor a multi-company statement from the insurance industry endorsing the use of CPRAs.

The true legal implications of the ACR are that they are more accurate and more credible than the handwritten record and, therefore, present a stronger legal case for the clinician. Be mindful that they are not 100% accurate but, as shown by every study that has looked at the issue, they are more accurate than the manual record. Ask yourself which of these two scenarios you would rather be in.

You’re on the witness stand defending a record that shows not one blood pressure below 100 systolic. Yet, a surgeon and an OR nurse testified that they heard alarms and saw NIBP pressure below 70 several times. “And by the way, Doctor, why did you give ephedrine when there was no hypotension?” “We know, Doctor, that you lied on the record and you did not deal quickly or effectively with this long and gross hypotension! What else are you covering up, Doctor?” You might as well write the check.

In the second case, you have a record that shows that there was one blood pressure below 70. This occurred 2.5 minutes after one of several blood pressures in the 90s systolic. The record shows that you treated it with 5 mg of ephedrine within 1 minute of its occurrence and then another dose of 10 mg for a blood pressure of 78 systolic. All further blood pressures were above 90. Furthermore, your database shows hundreds of cases of similar age and ASA class with this degree of hypotension with no negative result. (Actually, if your database is like mine was in 1988 when I started to measure hypotension in the department, it is close to 17%. Until you measure and address the issue with the department, you cannot reduce it to 2% as we did by 1993, but that is another story). So you respond to the pushy lawyer, “Why do you postulate that a bad outcome occurred here?” You choose where you want to sit if the courtroom is your worry.

Both these scenarios assume all values are accurate. In fact, my analysis of literally millions of automatically recorded cuff pressures indicates that the value is correct in the vast majority of cases. But, let’s assume that the value is an artifact. Those who fear this issue only assume that the low value is an artifact. As in Dr. Metz’s discussion, it does not include the more likely possibility that the value is a true value. If the value is an artifact, is it really that bad to have a value of 60/40 one time with a note saying the patient is shivering? This note is written on the spot within a minute or two and takes less than 10 seconds. You can prove when it was written to avoid the charge of fudging the record. I don’t think any plaintiff’s attorney would be able to make much headway against that. Sure, it would be better to never have the artifact recorded in the first place, but to allow it to be erased would also allow the true value to be erased. This is the very problem created by the handwritten record. Credibility is lost. The only solution is to reduce the artifact rate. Now that we can measure the artifact rate, maybe those rates will be published for different monitors and we can buy monitors with lower artifact rates. This information has been available from EPRI. The monitoring industry will be motivated to improve that good rate and possibly produce a six-sigma monitor (3 failures in 1 million recordings).

Hopefully enough people will soon see that this worry about legal and artifact issues is worth the improved hypotension, hypoxia, tachycardia, hypertension, emesis, and cost of anesthesia or almost anything else one chooses to measure. My former department experienced those improvements from 1988 to 1996. As the saying goes, “If you don’t measure it, you can’t improve it.” That statement is not quite right either. What is more true is that if you do measure and report on an issue, your department will pay attention to that issue and, as a result, it almost assuredly will improve. Let’s help move off this legal issue by changing the terminology to CPRAs or ACRs. If we don’t, I predict we will have many more mock M&M panels and white papers in an attempt to correct the false impression created by that “A” word terminology. Will the APSF formally endorse another term for the technology? I think it will take a statement like this to change the history of using this incorrect and misleading term. This should be the first standard term dealt with in the dictionary. The term “Automated Record” is a poor alias or a term for a machine that has not yet been developed.

David W. Edsall, MD
Albany, New York

Disclaimer: David W. Edsall, MD does not have any financial relationship to any computerized anesthesia record manufacturer.


  1. Metz S. Legal implications of automated records. APSF Newsletter 2001-02; 16:59.