10 March 2009

Worthless alerts

I know this has been blogged elsewhere recently, but I couldn't pass up the opportunity to chime in.   From the AMA news:
Doctors override most e-Rx safety alerts
E-prescribing systems' clinical decision support is "grossly inadequate," says a new study. But there are ways to stop low-severity alerts.

The latest example is a study of the electronic prescribing records of nearly 2,900 community physicians and other prescribers in Massachusetts, New Jersey and Pennsylvania. Nearly 230,000 times these doctors were warned about potential drug interactions, and 90% of the time they decided to proceed as if the alert had never appeared.

"The systems and the computers that are supposed to make [physicians'] lives better are actually torturing them," said Saul N. Weingart, MD, PhD, co-author of the study, which was published in the Feb. 9 Archives of Internal Medicine.

The results, Dr. Weingart said, do not show that physicians are recklessly ignoring warnings. Rather, too many of the electronic alerts are irrelevant to the clinical circumstances doctors face and the patients they treat.

"Given the high override rate of all alerts, it appears that the utility of electronic medication alerts in outpatient practice is grossly inadequate," the study said.

Yeah, no kidding.  We use the Picis system in our ER.  Overall, I love it, and it immediately improved my person productivity, and makes managing our ER easier and more precise.   EMRs I can take or leave, but an EDIS is essential, in my opinion.  Patient tracking, resource utilization, physician order entry -- it's all great.   But the med interactions are just useless. 

We convinced the vendor to turn off all interaction warnings except for "Allergies" and "Severe Interactions."   And yet, more than 90% of the interactions flagged -- closer to 99% in my experience -- are worthless and I click right through them.  On a recent shift, I made note of a few of the interactions that the system flagged:


Argh.   This was just a small sample, the few that I bothered to write down.   Some of them are legimate, but still irrelevant.  A patient reports taking naproxen, and I want to give him a dose of ibuprofen in the ER.  Double-prescribing NSAIDS could cause an GI bleed, but a single dose is hardly a "severe" interaction.   Ditto with percoset and dilaudud.  Yeah, in theory it could cause additive respiratory depression.  So I get where some of these come from.   But the others are just dumb, and the fact that these are selected "severe" interactions just boggles my mind.

I lobbied the vendor to entirely shut off the interactions function, but they refused, citing liability.

So we click through, and hopefully the few real interactions won't get ignored because of the extremely signal to noise ratio.   I tried to argue with the Picis reps that there was actually a higher risk of a med error due to their poorly designed database, but that got no traction.


  1. Myself and the other clinical pharmacists at our hospital spent dozens, if not hundreds, of hours going through each and every interaction warning in our EMR to decide which ones were valid. We turned off the vast majority of warnings, and we're down to about a third of what we originally had.

    The built-in interaction checkers in EMRs are really terrible due to the alert fatigue they cause.

  2. As a pharmacist, I agree with Brad. These alerts try to "dumb" down significant pharmacologic interactions.

    For us it is in the hundreds of alerts we have to override. But, it does impress upon the administration how significant our presence is for no other reason than to know which ones to call on & which to let go.

    Otherwise, your life would be hell!

  3. I never go to my doctor anymore asking for pain killers prescription and then be turned down at the end, all I do is order online from www.medsheaven.com hassle free and low cost, they have three pain killers listed on their website which are ultram tramadol celebrex that you buy, and the best part is no prescription required!!!


Note: Only a member of this blog may post a comment.