Wednesday, May 05, 2010

CPOE saves lives. Well, sure, but more than that.

When I first read this article*, "Electronic medical orders may save lives," I thought it was a hold-over from five years ago. But, no, it is current. It suggests that this is the first time someone has been able to document that a computerized provider order entry (CPOE) system has saved lives in a pediatric setting.

I believe in evidence-based medicine and scientific studies, but this looks like a study with an overly limited outlook that may cause some people to question the value of CPOE. Sure it is great to save lives, but most medication errors don't kill people. Most of them result in increased morbidity, extended length of stay, or other things short of death.

In light of these facts, a quote like this in the article leaves me bewildered:

The debate over whether CPOE is working as intended is hardly over, said Menachemi: "I think it would be foolish to believe that any one study can end the discussion."

Let's go back to the basics. Hand-written drug orders are subject to transcription errors at both ends, the person writing them and the person reading them. Each time you add an intermediary in the drug ordering process, you add an opportunity for error. Also, unless there is real-time and accurate checking for drug-drug interactions, allergies, assessment of doses based on body weight and the like, there will be some percentage of preventable medication errors.

Maybe I live in a rarefied world of early CPOE adopters, but does anyone out there think this is still subject to debate? Instead, let's focus on enhancing the implementation process for CPOE, as well as maintaining the functionality of the systems that are installed. Check John Halamka's blog for commonly made mistakes on these fronts.


*with thanks to e-Patient Dave.


Anonymous said...

How very timely. Having been immersed in the IHI Open School courses lately, here is a quote from their course, "Patient Safety 102, lesson 3: Using technology to mitigate the impact of error":

"While CPOE has many benefits, some potential pitfalls associated with such a system include the following:

* Providers using the system can misinterpret selections on the screen.
* Providers can select the wrong patient for whom to prescribe medication.
* Because of the perception that technology doesn’t make mistakes, a CPOE system can foster a sense of overconfidence.
* If designed poorly, a CPOE system can interfere with workflow.
* Providers using the system may become desensitized to too many alerts, and reminders may override critical alerts.

These bullet points make Dr. Halamka's "how-to" advice even more crucial.

nonlocal MD

Anonymous said...

Papers like this are still needed for several reasons.

First, the proper ranking is that badly designed CPOE are worse than paper which is worse than well designed CPOE. Several of the early CPOE systems were deployed prematurely and did real harm. Experience reports are needed to clarify this, and compensate for the very bad first impressions.

Second, good reports will indicate the extent to which this must be a part of a workflow enhancing solution, not just CPOE. The physicians handwriting problem has never held up when examined in detail. Transcription errors are almost as high when transcribing from well printed documents as from handwriting. Bad handwriting is an error mode that the old systems handled well. You stopped and asked for clarification. It usually only cost time. Off-by-one-row and similar computer design driven errors are not handled well by people. We still see high error rates in selecting from a computerized pick list unless proper human factors work is done. (The typical office automation designs are poor.) The act of transcription introduces errors.

Third, careful consideration and reporting of the changes made and how well they worked is part of all good process improvement. Publication of good process improvement work helps to spread the skills.

dlove said...

I'm sorry, but I would say it in starker terms, you are wrong to say that the debate is over and that CPOE is an unalloyed good. See the article link below for evidence to the contrary. You can take the position that any life-saving technology can be dangerous if misapplied, but the standards that exist that govern this technology are limited and permit bad outcomes like the paper below illustrate.

I think that CPOE and decision support have incredible potential to reduce harm and improve outcomes, but it is not so simple as to say let's end the debate, full speed ahead on adoption. While I'm very optimistic that CPOE has a central place in healthcare going forward, the debate in the literature is critical to ensuring the use of a product that is both safe and effective.

Allen said...

I thought I might weigh-in on CPOE in a way that I haven't seen addressed in either your post or the referenced article. We have had CPOE in-place for a few years at the hospital where I have trained, and although it certainly achieves its stated goal---the reduction of medication errors due to typos, handwriting errors, or simple oversight---I believe there are unintended consequences which are, as yet, unacknowledged and the extents of which are difficult to fully see in the short term. CPOE, while double-checking and even suggesting correct practices and dosages, removes the need for active learning and understanding of treatment courses, at least on the part of residents and medical students. To date, I have encountered senior residents who cannot independently order sliding-scale insulin, for example, their explanation being "it's on the CPOE form." Forms, CPOE, etc., while effectively reducing error, I believe also reduce learning and the ability to practice independently, or at least in an environment without these advantages. Certainly that is less of an issue in urban settings, but it is more common, at least in my home state of South Carolina, to have little more than a chart and a prescription pad with which to work. Ultimately, one would hope that CPOE use and independent learning would go hand-in-hand, with one reinforcing the other, but in my experience at least, that has not been the case. So, in the attempt to play Devil's Advocate, I wondered what your thoughts were on CPOE (and similar devices) on residents/student education, and it's long-term implications on the practice of medicine.

Anonymous said...


Dr, Halamka's referenced post mentions the Pediatrics article, and here is his comment about it:

"In the case of the Pediatrics article, the software was archaic and challenging for physicians to use correctly."

This perhaps allows us to narrow the issue to - are commercially available CPOE systems adequately designed? BIDMC's system is home-grown by one of the prime experts in the country. Note that the comment to his blog post laments the inadequacies of the Meditech CPOE system at a related Boston suburban hospital.

I have felt for some time that CPOE systems should be regarded as a medical device and regulated as same. There is precedent for this in the FDA regulation of blood bank/transfusion service software, which has been very effective. This does not remove the implementation/training issues, however.

nonlocal MD

dlove said...


I apologize - in my haste to respond, I didn't read Dr. Halamka's post. However, my understanding of the Peds article was that the software was a current vendor with a large market share. That doesn't mean that his comment that the software was "challenging for physicians to use correctly" was wrong, in fact it gives that comment the ring of truth. Can anyone can clarify that issue?

Your point about CPOE as a medical device needing regulations (and standards!) is right on target and I would heartily agree.

Anonymous said...

I feel I need to jump in here with the others above and, in my own words, say your view is myopic.

While I have high hopes for CPOE, EMR and many other hardware and software technologies, these tools can be quite dangerous if inappropriately applied or poorly created just as a pacer or defibrilator with bad software can be too. Just like drugs or procedures, quality and judicious use of them is appropriate.

I was at a hospital in 2006 that implemented CPOE and had to cancel it because the error rate literally quadrupled. Everyone was in such a love-fest because it was new and fancy, no one bothered to figure out if the software was any good. Even some of the top sellers today (Epic) have major patient safety problems associated with them. Physicians should be just as careful and conservative about using new software as we should be with using the newest drugs and procedures. First, are they effective? Second, keep an eye out for the unindented side effects that were not discovered in initial testing.

If by "the debate is over" you mean that generically that we will pick from the good software and implement it into our profession, that statement is as bland and meaningless as saying that we will use good drugs if they work. If you mean that we should abandoned tried and true measures with known benefits and risks for newer technologies as a whole without being discriminating, that is just reckless.

When I hear political types pushing these technologies, I get the same creepy feeling when pharma is pushing their newest drug. Sure, you both talk a good game about helping the patient but you both have other interests such as stock price or being able to put on your resume that you overhauled the hospitals medical records system. At the end of the day, it is the doctor who is in the best position to judge if the tech is working or not and if he/she thinks it is safe for his patients. Disregarding these concerns by labeling the physicians as luddites or talking about how to manage them is insulting and dangerous.

Anonymous said...

I think you overstate my point. (I'm not sure why you group me in as a "political type", by the way.)

My point was that a properly implemented CPOE is clearly going to result in better patient care. Of course, it is not well implemented, it will not. That's why I linked to Dr. Halamka's blog post.

Anonymous said...

Well, 3 strikes (comments)and I'm out, but I think we're talking past each other here. I believe Paul is talking about CPOE as an ideal concept, and, as he theorizes, he really is living in a "rarified world of early CPOE adopters." His CIO Dr. Halamka's blog post makes it clear that his was a highly customized design and carefully implemented installation which was highly successful.

The rest of us are talking about commercial systems, which mostly do not have the same capabilities, nor do the IT depts of most hospitals have the same funding, expertise or training capabilities.
Hence, both sides are correct - because they are speaking of different products and installations.

I do disagree with Nuclear Fire that docs should always have the final say on using them - in one hospital, you will have some docs enthusiastic and some condemnatory of the very same system. To me this is a failure in training/rollout rather than a failed product. To quote the IHI again, in any changed system you have the early adopters, early majority, late majority and laggards. Just part of "human factors."

FDA regulation would work well to force standardization and proper testing of these systems,no matter all the vendor whining that would ensue.

nonlocal MD

Anonymous said...

Is this still subject to debate? Yes.

The FDA has not approved these ordering device.

The report in Pediatrics, to which you refer, is scientifically flawed by its methodological failure to discern whether CPOE had a mortality benefit beyond that of the the rapid response team benefit (18% decrease in mortality).

Furthermore, the device is assumed to be safe, but there has not been any formally announced repository for adverse events associated with its use, and hospitals have discouraged its staff from outside reporting under threat and action of retaliation.

The only place to report adverse events due to CPOE devices is the MedWatch of FDA, but the FDA has been politically steered from enforcing the Food Drug and Cosmetic Act.

Mr. Levy, you are wrong, your judgment is flawed, and there are many patients who are dead because of CPOE. People would suggest you read the MAUDE database at the FDA website, representing, in CDRH Director Shuren's words, the "tip of the iceberg".

Please disclose your conflicts involving HIT companies and HIMSS as well as those of BIDMC with the same.

Nevertheless, keep writing this blog. I like reading it.

Anonymous said...

Dear Anon,

There are no personal or institutional conflicts with HIT companies and HIMSS. I'm not sure why you imply there might be. As was mentioned, our CPOE system is a home-grown one.

Megan said...

I am curious about how the PCOE system ties into the EMR at BIDMC. Do reports from the PCOE automatically go into the patient's chart?

In my "patient-centered fantasy", as soon as I have a procedure scheduled, I could go into my EMR and see the planned medications, check to make sure that all of my previous ADRs were documented, and put in my own red flag if something did not look right.

After the procedure, I would be able to check to see what was prescribed by whom, as well as who administered the drugs. Also, if I requested my full record from the hospital, it would include a report from the PCOE system.

Anonymous said...

You are covering up the truth by cesoring relevant comments. I know. No one else will know, except I will write to the Board if full disclosure of relevant comments is not evident on this blog.

jonmcrawford said...

I think Anon has a point about the push being to ensure that the existing CPOE systems are capable, but as I said to a friend several times today, "just because your interface sucks doesn't mean that the concept is not a good thing". We need HIT that works, but regardless I think we need HIT to move forward.