Those drug-interaction warnings sure are irritating | Booster Shots | Los Angeles Times

Those drug-interaction warnings sure are irritating | Booster Shots | Los Angeles Times
Those drug-interaction warnings sure are irritating
10:30 AM, February 10, 2009

One can imagine how electronic drug-prescribing systems could be annoying to doctors — all those warnings about potential interactions and allergies and whatnot when all you really want to do is give a patient a drug he or she needs and be done with it. It’s probably easier to just ignore the blasted alerts and keep going.

That seems to be what’s happening. In a study published in the Feb. 9 issue of the Archives of Internal Medicine, most doctors simply shrugged off the warnings issued by their helpful electronic systems.

Out of almost a quarter-million medication safety alerts produced during the study period, the doctors involved accepted only 9.2% of the interaction warnings and 23% of the allergy warnings. In other words, they ignored more than 90% of the drug interaction alerts and more than 75% of the allergy alerts.

As a fellow practitioner of blog-snark I’m impressed with the slant of the article, that doctors treating patients ‘ignore the blasted alerts’ and ‘shrug[ged] off warnings’.  As a doctor who uses a very good EMR tool daily I can tell you that a terrific number of the medication alerts given by the system are of such exceedingly minimal value they need to be over-ridden for accurate and adequate patient treatment.

For instance, if a patient reports an allergy to ‘Demerol’ in our EMR, giving Zofran (an anti-emetic) yields an allergy interaction.  It’s a theoretical connection and not founded in reality, or practice.  Want to prescribe pain meds for a patient being admitted? We get an alert that the patient has already been prescribed pain meds, even though the earlier meds were given in the ED (only) and the order in question is for inpatient treatment.  Very helpful.

The LATimes article doesn’t indicate if any harm came to any patients (I certainly hope not), but the finding that practicing docs didn’t heed a zillion electronic drug warnings just means the electronic systems need to tell us when the sky is actually falling, and not squawk continuously without cause.


  1. It’s just bad system design.

    Just like all the monitors in the ER that constantly beep and alarm. Eventually, staff simply tune them out or shut them off… and thus nobody pays attention when something is really going haywire.

    I don’t know why engineers and programmers tolerate that kind of thing. An audio engineer would go absolutely insane at the elevated noise floor in most ERs. Keep the alerts pertinent, and people will be more likely to pay attention to them.

  2. This entire system is nothing more than a plaintiff attorney employment program. What, you gave Zofran to that patient with a documented allergy to Demerol??? You ignored the warning of the computer??? Sign this check please.

  3. My favorite: Patient is written for NS @100 for maintenance – want to give a 1L NS bolus? “May be identical order, please verify” — Verify this…

  4. Goatwhacker says:

    The most dangerous drug at my hospital is the deadly Fleet’s enema, which seems to trigger a drug interaction with everything else on the formulary.

  5. Disgruntled Internist says:

    In 20 years of medical practice, this is what I’ve learned about drug allergies and drug-drug interactions:

    1) If a patient is “allergic” to more than three separate classes of medications, they are more likely to benefit from a psychiatrist than an allergist.

    2) Most “allergies” are uncomfortable side effects and not true allergies, such as the so-called erythromycin allergy or amoxicillin allergy, which was actually stomach upset; the ACE-inhibitor allergy, which was a cough; the demerol allergy, which was a doped-up feeling; Cipro causing photosensitivity; Niacin causing flushing etc.

    Now that being said, even if it wasn’t a true allergy, a patient won’t be willing to take a medication he/she has had a bad experience with in the past. So it’s best not to administer or prescribe a medication the patient says is an allergy.

    3) For this reason, it’s important to ask the patient “what happened when you took drug X, Y or Z?” When you uncover this information, do everybody else a favor and dictate something like “The patient has no known drug allergies but is intolerant to amoxicillin, which caused nausea.” That way future physicians and pharmacists will not be misled.

    4) Many drug allergies that are supposed to be cross-reactive in theory are extremely rare in actual practice, such as the aforementioned demerol allergy/Zofran allergy, or my personal favorite, the Sulfa allergy/HCTZ allergy. In nearly 20 years of internal medicine practice, I’ve never seen a sulfa-allergic patient have any problem with HCTZ.

    5) There are plenty of druggies in “excruciating pain” who who claim to be allergic to everything except demerol; my guess is that allergy testing would show this to not be the case.

    6) Take the following deadly seriously: Any drug the patient says caused hives, throat swelling, tongue or lip swelling, or severe blisters.

    7) You know you’re in for a difficult patient encounter when the list of drug allergies includes a number of psychiatric medications and narcotics. Normal people don’t know whether they are allergic to “Lithium, Haldol, Seroquel, Depakote, Demerol, Methadone, Antabuse and Geodon” because they’ve never had any reason to take them.

    8) Take drug-drug interactions seriously if you’re dealing with coumadin, theophylline, dilantin, digoxin and cyclosporine. There are others, so don’t think this list is exhaustive.

    9) Take penicillin allergies, aspirin allergies, contrast-dye, sulfa and latex allergies seriously.

    10) Thank the pharmacist for bringing this sort of stuff to your attention instead of yelling at her/him. That pharmacist just might have saved your patient’s butt (and yours, too).


    I participated in the above study. Most of the alerts in the VA EMR were noise. And very irritating. And the worst part was that even if there was a significant drug-drug interaction, if it was somewhat esoteric, the EMR did not tell you what the interaction was. You still had to go to Epocrates and check drug-drug interactions or call the pharmacy. I override most of my alerts still to this day cuz most are BS. I know the patient is on amio, I prescribed it. I know he’s on simva, I prescribed that too. Do they potentially interact, sure, but I’m not stopping them dagnabit.

  7. Welcome to my world! I’m a pharmacist & this occurs about 50 times per hour – all day long!

    Disgruntled Internist had great suggestions. A good history with documentation will spare you lots of calls.

    I have seen the sulfa/hctz interaction. It was an “almost” Stevens Johnson symdrome which was very unpleasant. She couldn’t take furosemide, bumetamide or dorsamide either – badness!

    The most annoying “allergy” problems from patients are when they say they’re allergic to sulfa & can’t take this eye drop with a sulfite preservaative -they are different!!! Sometimes its a good thing hospitalized patients can’t read the drug info – TMI.

    Fortunately, the pharmacy computer is filtered so only the highest level of alerts come through, but it still occupies a lot of my day. Just shows how complicated therapies can get.

  8. My favorite are the allergies to “steroids” and “adrenaline” — Oh my god — if you get them accidentally we’ll have to give you STEROIDS AND ADRENALINE!!!!

  9. I had a patient allergic to Glucose one time.

    Yep… that’s right… “allergic” to the common currency of cellular metabolism (no… she really wasn’t… she was just screaming that to the staff)

    I guess she could have been a space alien (and she was weird enough that I’d almost believe it)

  10. For the better part of 15 years I have sold, installed, trained and supported a highly integrated practice management and EMR system. The drug interaction aspects have evolved over the years based on both mandated requirements (government funding and subsidies) and the feedback from users.

    The earliest implementation provided warnings for drug-drug interactions and explicitly noted allergies. This has evolved to also provide warnings for drug-disease states, dosage errors based on age and weight and medication to food allergies.

    User feedback has been closely monitored. From the outset, our software has always tried to signal alerts with a severity level clearly displayed. Depending on the severity and prevalence of reactions, management of the alerts range from a simple click to dismiss the alert to an absolute requirement that the user sign off on the management of a severe or possibly life threatening interaction.

    The default set of alerts upon installation proved to be annoying to users and many felt that these alerts were sometimes very bizarre. There is no question that if alerts are simply an annoyance, they become more noise than signal. Once this threshold is crossed, the likelihood that some significant problem is overlook becomes much greater.

    The solution to this comes as no surprise… It is a trip back to the drawing board and provide a mechanism to allow users to adjust the alerts in such a way as to adjust the signal to noise level according to their comfort level. The initial installation has all alerts on by default.

    A set of preferences in the software allows users to determine how alerts will be displayed. For each of the alert types (drug-allergy, drug-drug, dosage, drug-disease interaction and drug-caused disease) there is a matrix of alert types vs severity.

    Within the preferences, some of the alert matrix items can not be turned off for both patient safety and legal/liability reasons.

    For each patient, any management that has been applied to a medication or combination is stored in the patient’s record and this management can be applied automatically on subsequent prescription renewals. (The alert is displayed but can be dismissed with a return keystroke.)

    “Favourite” managements can be saved at both clinic wide and individual user levels. For the severe “favourites” there is a 6 month expiry limit after which they must be reviewed and explicitly re-instated.

  11. Personally my favorite is when I get the alert that the Aspirin, Plavix, and lovenox I’m giving to my MI patient increase his risk of bleeding. Never would have figured that one out on my own.

  12. If you are using a CPOE system, and find that it’s too “noisy” with its drug-drug interactions, it can be adjusted! I’m a pharmacist working on the installation of a CPOE system at a large academic medical center and we have made a point of MINIMIZING the number of alerts that physicians see when ordering. The alerts you see are most likely driven by a commercially available drug database (ie. Multum or First Data Bank). Although you can’t change the stupid interactions that are built in the database (NS and oxytocin? Come on.) you can selectively deactivate whole levels of interactions or even single interactions. If the people in your IT department aren’t clinicians (and most aren’t), they have no idea what you’re going through. Don’t assume they know, ask them to fix it!