No One Is Perfect, Not Even Computers

My last post described how a precisely regimented dosage of intravenous medication delivered to me over six hours by a state-of-the art computer actually depended on the existence (and the survival for 6 hours) of a handwritten yellow Stickie hanging on my IV pole. I write this post as a recipient, certainly not a victim, since no harm occurred, of a “care error” caused by a computer.

After my first infusion I grumbled to my physician that it had taken 6 hours, and that the package stuffer the nurse gave me recommended about a 2 hour infusion for someone my weight and age. He was surprised but responded, “Those nurses are really good. They probably have more information about the drug. I would go with what they say.” So I called the Head Nurse in the Infusion Center. She told me that the infusion rates come from the computer. “How does the computer know them?”, I asked. She responded, “The Hospital Pharmacy Committee puts them in.” I called the Chief Pharmacist, noted the difference between the package insert and the computer recommendations, and asked him to review the information because I would sure like to spend just 2 hours off my boat rather than 6 for the next treatment. He contacted me a couple of days later to tell me that that medication infusion rate had been entered into the computer several years ago and was based on data from the one manufacturer of the medication. “There are now three manufacturers and two different concentrations. Each one has different infusion rates. Yours could go in over 2 hours. I will take care of updating the computer’s recommendations for your medication before the next treatment.”

The Institute of Medicine describes a medical error as “following a wrong plan of care or not completing a correct plan of care”. My computer-associated medical error was caused by “failure to update reference information”.

Do computers cause errors? The FDA maintains a data-base that categorizes voluntarily submitted adverse events associated with thousands of medical devices. Only five of the categories have the word “computer” in it. The Huffington Post made a considerable effort to analyze the most recent year’s findings in this data base and found 237 incidents that were related to health information technology. (1)

Six deaths were associated with computer adverse events. Except for two hospital-wide computer system crashes which delayed medications (both in 2006 in Cerner installations) the events included well recognized causes of NON-computer medical errors: delay in sending an x-ray image to another facility, a physician missed reading a significant “addendum note” on another physician’s progress note, an incorrect patient identification on an xray film, and an incorrect mixing of a chemotherapy solution.

Forty-three injuries associated with computer adverse events were reported. Many of these involved incorrect manufacturing of intravenous solutions rather than incorrect computer-directed delivery as well as incorrect dates, patient identification, or study type in radiology filing systems (PACs). These radiology “errors” became “injuries” when the errors were not perceived by human readers using multiple display screens, multiple screens, screen short-cuts, etc.

Studies to date of computer errors in clinical care have by and large identified the computer/human interface as the most frequent cause of error: transcription errors, misreading of displays, mis-navigation among screens, ignoring alerts, overriding warnings or alerts, failing to update reference and resource information. It is comforting to know that very few of these have led to harm because most of these are recognized as errors by trained clinicians before harm occurs. There is little data currently to suggest that we are just seeing the “tip of a gigantic iceberg.” Even the harshest critic of UK’s attempt to implement a nationwide EHR has been focused on the business plans, difficulties of implementation, and cost. (2)

Several years ago a banker spent a day shadowing both an internist and a surgeon at our hospital as part of a Doctor-For-A-Day program and summed up his impression as: “A doctor’s job seems to be a day-long search for credible data.” As long as we have well-trained clinicians providing our care, the help that computers will give them, and us as patients, in finding credible data will far out weigh any of the “new-found errors” (3) that will surely emerge as the new technology is implemented. The greatest threat to medical safety from computers will come from our trust in them, thinking that they are always right. A vigilant, skeptical clinician, and patient, is still the best defense against any subsequent harm from “computer error”.


  1. The Health Care Blog, “Do EHRs Kill People”, June 11, 2010, Margalit Gur-Arie
  2. http://hcrenewal.blogspot.com/ – Health Care Renewal – a blog “addressing threats to health care’s core values, especially those stemming from concentration and abuse of power."
  3. National Research Council, January 2009.

Herbert Mathewson blogs at HUB's LIST, a compilation of medical fun facts gleaned from a variety of medical journals, newspapers, other public and professional sources, and an occasional private communication.

Categories: Uncategorized

Tagged as: ,

17 replies »

  1. We need to continue to embrace the use of computers in health care for quick data access as you state. However, we need to find a way to keep the human compassion element which seems to elude us every time technology is added to an application whether it’s health care or airport screening.

  2. Great Post Indeed Mr. Herbert
    This only shows that Humans are far more intelligent and complex than Techonolgy. We created Technology, not the other way around. Surely, It has helped in a lot of ways anyone could have ever imagine but that doesn’t mean we have to solely rely on them. Computers can store massive datas in seconds but once it’s reprogrammed everything will disapper. Humans are not like those. Computers dont have hearts to see and feel things like we do.

  3. Great Post Indeed Mr. Herbert
    This only shows that Humans are far more intelligent and complex than Techonolgy. We created Technology, not the other way around. Surely, It has helped in a lot of ways anyone could have ever imagine but that doesn’t mean we have to solely rely on them. Computers can store massive datas in seconds but once it’s reprogrammed everything will disapper. Humans are not like those. Computers dont have hearts to see and feel things like we do.

  4. I actually won’t consider this as a computer error but a human error. We have to take in consideration that computers are just tools for us to make our lives easier. A computers may have a memory loss sometimes, I agree, but in this case it was not the computer’s fault that it was not updated. It is definitely a shortcoming on the part of the pharmacy Committee of the hospital.
    It is just so sad to know that people in health constitutions are so irresponsible and careless wherein a single mistake can cause death in some cases.

  5. Although, I agree with your discussion. I still feel that as professionals instead of focusing on the ‘business’ aspect of healthcare, we should direct our efforts in making our clinical skills more sound. Below is a nice website, where you can start.
    1. Central searchable repository of a pharmacists ‘curb-side’ notes.
    2. Transition between hospital and retail settings.
    3. For access from anywhere, when I do not have direct access to my references/websites.

  6. Great post, good comments. Improving the delivery of health care is a continuous process. Disrupting systems – even bad ones – often increases errors because the old ‘work arounds’ no longer work and the new ones have yet to be designed. We are making progress. How many errors are attributable to paper charts and the old ways? Definitely agree that those that are only running after the meaningful use dollars are crazy – it’s hard work to get an EHR configured properly and redesign your care processes to take advantage of all the things an EHR can do. If you think its just making a paper chart electronic you are sadly mistaken. Thanks for posting!

  7. Indeed true, reports show a large percentage of medical errors through electronic system. However, with advancement in technology day by day, we do hear of the computers coming up with features of listening to our problems and reacting to them easing out many processes. At the end, we do need human check for technology as well. This would any day be quicker and better than not having such kind of artificial intelligence.

  8. Honestly i prefer my learn as much about my medicines myself. This has actually lead me to become a pharmacist for this very reason.

  9. I realize that drugs have their place, but if you don’t invest in your health today you may be forced to pay for sickness tomorrow.
    According to the AMA (American Medical Association) PROPERLY perscribed medications kill over 100,000 people per year (not to mention the many errors that cost lives).
    That equals 273 people per day dying due to medications.

  10. Why is this? “an alarming percentage of them have a tendency to pronounce the problem unsolvable and revert to the familiar. Un-flippin-believable.”
    The arrogant HIT geeks with important titles and limited education and HIT vendors hellbent on profits are not interested in correcting the interface failures (if they even know how), nor are they interested in improving the systems’ usability. They do not understand how the effective clinician operates effectively, nor, in my experience, are they interested in soing what we need. Repulsive

  11. I’ve said this in this space before and I’ll say it again. Doctors were the smartest kids in high school, went to the best colleges, got the best grades in all the toughest college classes (P-chem, anyone?), aced the MCATs, endured med school, rotations, and internships. They were the best and brightest — literally the smartest guys and gals in the room.
    Yet, it never ceases to amaze me how, when presented with a problem like correcting a flawed computer interface, an alarming percentage of them have a tendency to pronounce the problem unsolvable and revert to the familiar. Un-flippin-believable.

  12. In the vast majority of computer-human interactions, it is the human who makes the mistake, and it is the programmer of the software and hardware, another human, who may have made the program hard to use or poor in validating consistency of data. As a result, the real problem and a hard one is one of designing human-computer interfaces that minimize the chances for human error. Far too often, this aspect of design is minimized with the results being a higher probability for mistakes than what could be achieved with a better design.

  13. Actually I am not surprised that the doctor deferred to the nurse’s judgment when it came to the medication. More often than not, the nurses are the ones with the practical experience administrating the medicine and know a lot more about it than the doctors do. The doctors just understand the desired effect, not necessarily the means of delivery.

  14. Corpuscle Connie, this entire article has nothing to do with the inaccuracy of computers, but negligence on the part of people responsible for inputting data. Everything described in this article, everything feared, is all human error and inexperience, and will be overcome quickly. You mean to tell me people can’t misread a piece of paper? Lose a medication inset? The only risk here is people who want to doubt the power of a computer. A properly maintained and updated database will not lie, and in this case it wasn’t.
    Investments in proper health IT solutions and people are pivotal for it to work, just like anything else in this world. That said, it isn’t as if this was a bug, this was purely human error from what I can see. Also, the response from the nurse was inappropriate. Upon realizing the discrepancy, the nurse should’ve contacted someone. This all goes back to training. Don’t blame the tool, blame the people using it improperly.

  15. Dear Herbert,
    I agree with you that technology is not perfect but got to say that it is the result of human work so it is bound to have some kind of errors.
    I understand that these errors are highly unacceptable especially when it to comes to dealing with patients.
    An large scale EHR implementation is surely a big challenge and it depends on certain factors which need to be dealt with in an appropriate way.
    I believe the following are the most crucial factors in a large scale quality EHR implementation:
    1. Use of the right tools & Services :
    Each specialty EHR has its own set of challenges or requirements which I believe is overlooked by in most EHR vendors in a effort to merely follows federal guidelines. This is resulting in low usability to the practitioners, thus less ROI, finally redundancy of the EHR solution in place.
    2. Looking at the profitability of the EHR investment
    I think ROI is very important factor that should be duly considered when look achieve ‘meaning use’ out of a EHR solution. Though one may get vendors providing ‘meaning use’ at a lower cost, their ROI / savings through the use of their EHR might be pretty low when compared to costlier initial investment.
    3.Having the right ( in terms of appropriate knowledge and experience) support function
    Also the introduction of REC’s through the HITECH act. is a great way to avail of quality EHR solutions at competitive prices. The stiff competition among not only these REC’s but also among EHR vendors ( to become a preferred vendor of a given REC) will result in lot of positives to medical practioners.
    4. Creating the right infrastructure for implementation:
    Looking at the funding provided to the REC’s, the staggered grant allocation system also promises to be an unbiased way of allocating funds. It will also help in the concept of REC’s helping out each with their own unique business models. It can be one of the possible answers to the
    ’safe vendor challenge’ as discussed by many critics.
    Do you all agree with me?

  16. Your report points to another reason why EMRs are dangerous.Intense cognitive multitasking is exactly what is forced upon the users of the poorly usable EMRs. Mistakes are promoted. The device is only as good as the data that has been programmed in to it. Decision support??? I am smarter than the idiots who copied the guidelines that oft do not apply to a particular patient. One letter missclicked and disaster can occur.
    It is the cognitive overload of reading screens of what MIMD calls “legible gibberish” which is why the EMRs cause as many, if not more, mistakes as they correct.
    Doctors and hospital administrators are advised by this consultant and practicing MD.Rsist the temptation to be snookered by the financial aid from the government. Do not take the bait.
    The NY Times covered this 2 years ago:
    Multitasking Can Make You Lose … Um … Focus
    …“Although doing many things at the same time — reading an article
    while listening to music, switching to check e-mail messages and
    talking on the phone — can be a way of making tasks more fun and
    energizing, “you have to keep in mind that you sacrifice focus when you
    do this,” said Edward M. Hallowell, a psychiatrist and author of
    “CrazyBusy: Overstretched, Overbooked, and About to Snap!” (Ballantine,
    2006). “Multitasking is shifting focus from one task to another in
    rapid succession. It gives the illusion that we’re simultaneously
    tasking, but we’re really not. It’s like playing tennis with three
    …Dr. Hallowell has termed this effort to multitask “attention deficit
    trait.” Unlike attention deficit disorder, which he has studied for
    years and has a neurological basis, attention deficit trait “springs
    entirely from the environment,” he wrote in a 2005 Harvard Business
    Review article, “Overloaded Circuits: Why Smart People Underperform.”
    “’As our minds fill with noise — feckless synaptic events signifying
    nothing — the brain gradually loses its capacity to attend fully and
    gradually to anything,” he wrote. Desperately trying to keep up with a
    multitude of jobs, we “feel a constant low level of panic and guilt.’”