The Windows Vista Effect



For those of you who actually use Windows Vista, what do you do when the User Account Control pops up a message like the one above?

Maybe the first few times you read it and see what the program is all about. After that, you glance at it when it pops up and click “Continue”. After a month or so, it grows to be a real pain in the but-tocks and you tend to ignore, or even get frustrated with it.

A recent Archives of Internal Medicine study titled “Overrides of Medication Alerts in Ambulatory Care” shows that the same concept holds true when doctors prescribe medications through a computer program.

Researchers studied more than 3.5 million electronic prescriptions written using a specific electronic prescribing system in several states between January and September 2006. They tracked 2872 total clinicians, looking to see how the clinicians would respond to “alerts” programmed into the electronic prescribing system. Alerts were programmed for allergies to medications and for potentially dangerous interactions between drugs being prescribed.

Of the 3.5 million prescriptions tracked, roughly half a million total “alerts” were generated by the electronic prescribing system. Half of those alerts were excluded by the researchers because they were “duplicate” alerts that occurred after the prescriber overrode the first alert.

The remaining 233,000 alerts were then studied to show how often prescribers accepted the alerts. Not surprisingly, the alerts were often ignored. “Drug-Drug Interaction” alerts were overridden 91% of the time and “allergy” alerts were overridden 77% of the time.

The researchers concluded that “Given the high override rate of all alerts, it appears that the utility of electronic medication alerts in outpatient practice is grossly inadequate … For active clinicians, most alerts may be more of a nuisance than an asset.”

I had three other thoughts about the study.

First, it proves that the more you computerize patient records, the more that the data will be tracked and analyzed. I really believe that data mining is the major impetus for the electronic medical record initiative pushed by the government.

Second, it shows how forcing clinicians to jump through more and more micromanagement and regulatory hoops in order to practice medicine won’t necessarily have the intended effect. It would be interesting to compare the productivity of the clinicians before and after the electronic medical record system was instituted.

Finally, the study shows that at least some of the “warnings” about drug interactions and drug allergies are theoretical ones and not clinical ones.
There were a quarter million prescriptions written in this study that were against the better judgment of some computer program.
Even though the prescriptions were written in 2006, I still haven’t heard about all the gloom and doom adverse patient outcomes tied to those those hundreds of thousands of transgressions.
Could it be that there were very few adverse outcomes and that the “warnings” were mostly overblown?

Like to see those study results.


  1. Oh my god…this was my afternoon!!!!! We had a TBI with HTN patient discharged on an ARB (Like I have any idea what that means!)

    Anyways, patient has medicare D. Medicare denies the drug. Attending is a physiatrist asks me to call primary as patient was on the drug before injury. Primary docs don’t work on Fridays….Call patient, patient (with the TBI) tells me what drug she was on. It is a different drug. She tells me what pharmacy she got it at. I call that pharmacy, they never heard of her. I call new pharmacy (that told us it was denied) ask them what drug is approved. They give me prior approval number. I talk to robot, robot does not give me option I want. I scream “help”! No help given. I say “prior approval” It beeps. I say “provider” It responds. I stay on hold for another 5 minutes. It finally tells me to call another number. Attending comes in office, looks at me and shakes head. I call the “other” number. I am on hold 15 minutes. Attending comes back and curses in his native language. He calls in new drug and request a homecare nurse to check blood pressure.

    I never got to talk to a human….

  2. Why just yesterday my EMR gave me allergy warnings for my patient. Before the c-section it reminded me that the patient was allergic to PCN even though I was ordering Kefzol. After the section it alerted me when I ordered a morphine PCA pump and percocet for when the pump was discontinued. It warned me based on an allergy to Lortab.

    The patient itches with Lortab and had an unknown reaction to PCN as an infant.

    She is alive and doing well today even though I overrode the allergy warnings 5 times (once for the kefzol, twice for the pca and twice for the percocet.) Each time I had to wait while the computer digested the “Warning has been read and acknowledged” input. It certainly didn’t speed me up

  3. Retail pharmacies have the same problem. The “alerts” get over-ridden so often I sometimes wonder if they are being read at all. I swear the most common alerts are Beer’s List related for elderly patients, or contra-indications for pregnancy in 72 year old little old ladies. I think the study is flawed though. Just because the physician over-rode the alert doesn’t mean he didn’t read it. He may have used clinical judgment and decided it wasn’t a big deal. The cefazolin/penicillin cross-reactivity is real, although very small. Some early studies concerning cephalosporin allergies in PCN allergic patients were flawed because until the 1980’s commercial penicillins were grown using Cephalosporium mold.

  4. Somehow, this reminds me of one of those movie trailer voiceovers:

    In a world where data is manipulated and mined to further corporate greed.. to abuse sick people and the physicians who treat them. A new day is dawning: panic… pain… torture… anguish… despair… denial. Just when you thought EMRs might provide some relief. Coming soon to a medical facility near you: ICD-10 – The Nightmare

  5. Out of ignorant curiosity, what is the potential for these medical records to be subpoenaed and then used against the provider? The thought of facing questioning about why I bypassed an automated warning on one of the computer systems that I work on is intimidating enough when I’m just risking a paycheck. I wouldn’t be willing to stand up to legal questioning about bypassing a forgotten warning on a network that I worked on a year or two ago and don’t remember clearly……. Can these records be retrieved by lawyers?

    Just looking for clarification.
    Thank you.

  6. William is correct. This is the real problem of computer generated warnings. They put you on notice should some untoward effect actually happen. What is your defense? “I didn’t know?” This just proves GIGO, garbage in/garbage out. When you see a record of an allergy that isn’t a true allergy, it has to be corrected. The real benefit of EMR is not management of care, but access to a complete picture of the patients previous care.

  7. We carry pocket alarms connected to pts tele monitors so that at all times we can hear if the pt’s monitor is alarming. Problem is that they alarm so frequently for non-life threatening events that everyone just ignores them. Honestly I don’t even hear them anymore. I can’t tell you how many times someone says doesn’t that noise bug you? What noise? Hospital spent alot of money on this system. I hit over ride on the pyxsis every time I work. It’s a pain.

  8. Pingback: Overrides of Medication Alerts in Ambulatory Care « The ACUTE CARE Blog: Non-Urban Emergency Medicine

  9. I am renewing metformin and glipizide.
    1)Warning one: duplicate therapy, risk of hypoglycemia. Sure. Click.
    2)Warning two: no creatinine within 60 days. Who orders a creatinine on every diabetic every 60 days? Since when would insurance pay for that? Sure. Click.
    3)Warning three: Are you sure? STFU. Click.

  10. The You can’t be too safe people do not Understand that they are making things more dangerous for patients.

    They view the human decision as the weak point in patient care. They think that by eliminating individual decisions, they will prevent bad decisions. this is a much worse decision than all of the individual decisions combined.

    They think that constant alarms mean that everyone will be constantly alert. Constant alarms encourage us to ignore the alarms.

    The expansion of top secret classification means that almost everything is secret, so nobody pays special attention to the true secrets. the result is that the real secrets are less secure.

    Companies are insisting that computer passwords have upper case letters, lower case letters, numbers, and symbols in them. This will result in people writing the passwords down, so that they can remember. What could be less secure than a password that is written down? Then they require that you change your password every couple of months.

    The truth is that these risk managers do not understand safety or risk management. They are the most dangerous people to be making decisions about safety. Maybe they need to watch Dr. Strangelove – the original version of Fail Safe.

  11. Long, long ago, in the era of PPOs– before HMOs– my dad told me that we only prescribe drugs out on market for at least five years. As a heavy reader of what’s new I felt very frustrated. Rheumatology was a field where a lot of basic research was then going on but back then we offered little more than NSAIDs– But boy, Ibuprofen 1gm per day for 60 days!!!!!!!

    Today we wouldn’t do that. But I watched the whole Vioxx panel sessions and was most struck by the patients who insisted they would die if it got black boxed. Three years later I attended a lecture (in a fancy French Restaurant) where Merck invited, as the salesman told me, “a top name in rheumatologist” (actually a payed hack) lecturing on “The Difference Between Rheumatoid and Osteo-arthritis.” All the guy did was rehash the Vioxx data and advised to give aspirin half an hour after NSAIDs in patients on preventive aspirin because platelets would become “sticky” again after that time. nothing new on who could and who couldn’t take Vioxx. I was very upset and ate nothing. The others at my table hadn’t heard the lecture but loved the meal.

    The point of that story is that clearly for some Vioxx is poison, for others it is life saving– so much for the generic “best practices.” All those TZD studies and meta-studies make a clinician wonder: what the hell? The answer came from the chemist who first developed the TZDs. He said: “Well, the patient should know that the drug is a joint venture between doctor and patient and that they are both taking chances with their lives.”–Duuuhhhh!

    We are entering an era where your genes– not “best practices”– decide what you can and can’t take. THERE IS NO GENERIC PHARMA “N” SUBJECT. The Genome Project is giving us a heads up that soon genes will be the essence of “preventive care.” So, given that health care is making dividends for thousands of investors from the inevitable getting sick of insured people, how can you load the dice for the house with your genome analysis? There is no choice but to end the predatory carnivore days of the HMO investors. We need the ultimate “open” (named ID) genetic data to link to outcome. Just as the rich tailor their suits, we’ll have to stop the get rich bias schemes of one size fits all of Pharma Inc and make the statistical link that leads to the molecular mechanism link that brings us into the medicine of tomorrow. Now physicians trained in physiology and a bit of biochemistry are going by the advice of the visiting drug salesman with the jazz ballet degree and the nice legs in the short dress that is sent to throw at them the “molecular” pharmacology babble of the pill’s maker. We are helpless when all the various DNAs and RNAs are brought up that made this drug’s case at phase I and the testing on Andean Indians somewhere in South America in phase II. Reagan’s only crime in my book was to allow federally funded research to be patented by the academics and Pharma industry so we pay twice for the drugs. Now you getting sick is big bucks for a lot of people who sit on their asses waiting for their dividends. That’s even worse than the jungle. Health Care, therefore, has to be a highly regulated vocation instead of a long chain of money makers. The internist at the bottom, for example, per MEDICAL ECONOMICS, makes less per hour of patient care than does a plumber. he/she is not the problem. The problem is that doctors are not payed to study but Madison Avenue is payed big bucks to dupe them in the bloom of their post-residency ignorance. So, a doctor seeing a WARNING only gets scared but doesn’t know why, doesn’t know what to ask the Pharma parrot and doesn’t know how to say no the the patient insisting: I saw it advertised on TV last night and I am sure it’s what I need. If Health Care has so much money for stupid ads, why don’t it pool the money for REAL further physician continuing education so they don’t panic in ignorance at all the “CAUTION”s?

  12. Donald Green MD on

    The real problem here is that the EMR industry thinks doctors are a bunch of electronic dolts and incapable of learning or taking some time to assure accuracy. It should be regular practice for physicians or other providers to have knowledge of interactions or have access to competent programs such as on UpToDate. We do not have to succumb to an annoying spoon fed type of homogenized medical care.

  13. Dr. Green, as you know, the boards are taken on laptops with time limit, so it is hard to imagine MDs as computer incompetent. The real issue is that PHARMA is molecular and medicine is physiological. There is a mismatch between the physical exam and the rationale from Rx. As for lab values, radiologically they are static anatomy and the morpho-chemical values are really the responsibility of the MD. So, if your lab mixes up specimens and your patient who is hyperkalemic proves to be hyperglycemic, according to the lab results, your *CLINICAL* picture should tell you that the labs are a mix-up. So in the end, what you trust is exactly what my dad trusted 70 years ago: physical exam and history. Is that not closing the barn door after the horse got out?
    If we’re going to be getting preventive medicine we need to bring science to the practice long before signs and symptoms and that takes time to learn, for which the MD should be paid, given that for his patient care time he is paid less per hour than a plumber. Everyone makes off like a bandit except the patient and the doctor. It’s time to take out the profit motive because the only two ends that contact directly– patient and doctor– are the ones screwed by the profit motive of the rest of the path.
    MY idea is not so radical, it was proposed in the 70s and all that’s left of it is the baloney CME credits required annually and the decade re-exam. Medical education should be systematic, like school, and time spent should be time paid since a doctor who understands what’s new is not paid more than a doctor whose year of graduation from residency program shows by his practice of medicine. Physicians want to be as good at physical diagnosis as at molecular diagnosis. The time it takes to achieve that should be work time, not home time– isn’t the time spent with parents, wife and kids short enough already? Don’t you agree? If you do, then I’m sure you agree that the scrib sheet in the pills package–more a legal than a medical document– should be far more explicit and not just giving lists of side effects just so that no one can say “you never told me.” I mean, headache, constipation, cough, etc are pretty hard to use as guides. A lot more pharmacology to explain why some do and some don’t get side effects would be a lot more useful, n’est pas?

Leave A Reply