How algorithm driven medicine can affect patient care

Whenever someone is scheduled for an operation, the assigned nurse is required to fill out a “pre-op checklist” to ensure that all safety and quality metrics are being adhered to. Before the patient is allowed to be wheeled into the OR we make sure the surgical site is marked, the consents are signed, all necessary equipment is available, etc. One of the most important metrics involves the peri-operative administration of IV antibiotics. SCIP guidelines mandate that the prophylactic antibiotic is given within an hour of incision time to optimize outcomes. This has been drilled into the heads of physicians, health care providers, and ancillary staff to such an extent that it occasionally causes total brain shutdown.

Let me explain. For most elective surgeries I give a single dose of antibiotics just before I cut. For elective colon surgery, the antibiotics are continued for 24 hours post-op. This is accepted standard of care. You don’t want to give antibiotics inappropriately or continue them indefinitely.

But what about a patient with gangrenous cholecystitis or acute appendicitis? What if, in my clinical judgment, I want to start the patient on antibiotics right away (i.e. several hours before anticipated incision time) and then continue them for greater than 24 hours post-op, depending on what the clinical status warrants? I should be able to do that right?

Well, you’d be surprised. You see, at two different, unaffiliated hospitals I cover, the surgeons have seen that decision-making capability removed from their power. If a young patient comes in with acute appendicitis and I feel that it would be prudent to continue the Zosyn an extra couple of days, an automatic stop order is triggered in the department of pharmacy and the antibiotic is stopped after 24 hours, no matter what. Unless the surgeon specifically writes “please do not stop this antibiotic after 24 hours; it is being administered for therapeutic purposes, not prophylaxis,” the antibiotic will not be sent to the patient’s floor for administration. As a result, patients end up being treated sub-optimally, and potentially harmed, due to an over-emphasis on “protocol” and “quality care metrics.”

Similarly, the 60-minute timeline for pre-operative antibiotic administration can be problematic. I have had patients come into the ER with appendicitis or cholecystitis and, in my pre-op orders, write for Zosyn or whatever, to be started ASAP, no matter what time the operation is scheduled. Not too long ago, I admitted a gallbladder over the phone at 2am. I gave the nurse admitting orders which included one for a broad spectrum antibiotic.

When I saw the patient in the morning, I added her on to the OR schedule. By the time a room opened up, it was about 10:30am. The OR nurse asked me if I wanted to give an antibiotic for the case. I told her that the patient was already on antibiotics as part of her admit orders for treatment. The nurse shook her hand. It had never been given; the floor nurse held it so that it wasn’t administered until 60 minutes before the scheduled OR time, just like the algorithm dictates — despite the fact it had been ordered nearly 8 hours prior to the case, not for peri-op prophylaxis, but for treatment of an established pathology. And there it was, the cefotetan, hanging on her IV stand. Now nothing bad happened, but here you have a situation where health care providers are so terrified of violating Quality Assurance Protocol that they end up withholding necessary treatment. It’s just astounding.

As surgeons, we have bitched and moaned. You would think that these issues would be quickly rectified. But no. It is the responsibility of the surgeon to write qualifying statements for therapeutic antibiotics because the default mode is to override a licensed physician’s clinical judgment. This is what I’m talking about when I say that blind allegiance to a top-down, systems analysis-driven algorithm can turn everyone involved in health care into a bunch of mindless drones.

Jeffrey Parks is a general surgeon who blogs at Buckeye Surgeon.

Submit a guest post and be heard on social media’s leading physician voice.

Comments are moderated before they are published. Please read the comment policy.

  • decarlo2

    This article strikes an interesting discussion surrounding the issues of quality in health care.  It appears as though there is a disconnect between the protocol checklist, which is established by insurance companies to void liability, and hospital/patient care.  Moreover, Dr. Parks highlights the need for the re-evaluation of the algorithm SOP that at times fails to provide the quality he sees fit for patients.  Although procedure is procedure, it doesn’t make much sense unless we establish new guidelines for benefit patients, Insurers and hospitals to raise the overall standard of care.  For a good resource on the issue of patient care quality, refer to the following link:

  • ninguem

    The surgeon is no longer captain of the ship. You have to be a team player.

    Well……until something goes wrong. Then it’s all your fault.

  • Kathryn

    I suspect that the algorithm protects surgeons more times than it causes trouble.  In the long run, I don’t think it hurts to have physicians think about how  to describe the need for the medication.  “For pre-existing infection” might be one way.  If the staff is not following protocol when they interrupt an ordered course of medication, that’s not the fault of the algorithm. 

  • Margalit Gur-Arie

    It’s a different philosophy that drives these algorithms. On a population level, working within the algorithm will have better statistical outcomes then working with no algorithm. On an individual level, it’s quite another story.
    The best of both worlds would be to have the algorithm and make physician override part of the algorithm, but I guess that’s too complicated for algorithm makers to consider.

  • Karen Sibert MD

    The beta-blocker SCIP guidelines are another good example.  As an anesthesiologist, I’ll be “dinged” by my hospital for not giving a BB on the day of surgery even when the patient’s heart rate is in the 50′s, because the “rules” say only a pulse rate less than 50 is a valid reason.  And look at the recent experience with tight periop glucose control–not until it was proved that low glucose was killing more people than tight control was helping did the rule get changed.  Many of these so-called quality measures are implemented in haste with very little testing; only later do we learn about all the unintended consequences.

  • Bruce Ramshaw

    I don’t think the problem is with the specific algorithms (none work as intended).  The problem is that any algorithm mandated like this is a simple system solution applied to a complex system/complex problem.  The fact that these mandates don’t work and often have unintended negative consequences is absolutely predicted by the principles of complex systems science.  The solution is not better algorithms, but a change in the structure of the systems for how we deliver care.  Based on the principles of complex systems science, I believe that designing care teams for definable patient groups, who provide care throughout the entire cycle of care for each patient, could improve the quality and value of patient care.

  • Anonymous

    Aaargh!!!! Speaking from a nursing perspective, with a few exceptions (e.g. ACLS) algorithms and clinical pathways should be GUIDES,and not followed slavishly as if they were written in stone,  Unfortunately, what we’re seeing is that they’re being substituted for critical thinking and clinical judgment that should be based on the patient’s situation. Algorithms/clinical pathways are micro-managing at its best (or worst).

  • Gaspere (Gus) Geraci

    Exceptions to the protocol should be as easily checked off as the “standard.” This prevents miscommunication and facilitates everyone’s job. For every checkbox protocol, there should be a checkbox that says, “Off standard protocol” and if needed for the bean counters/quality folks, an explanation of the why’s. And they should be checkboxes too. With today’s technology, it’s called a drop-down box, folks!

  • S Silverstein

    Listen folks, it’s time to stop making excuses for IT designer, vendor and implementer stupidity.  There is nothing new in these revelations about IT getting in the way of care due to fundamental design error that makes physicians work harder to practice care attuned to the patient.

    On workarounds as suggested:

    The rule is, as I posted at regarding mission-hostile user experiences presented by IT – and there is nothing to debate, nothing to discuss on this issue:

    “You should not have to work around something that is not in the way”.

  • Anonymous

    There is no standard patient and protocols can be very dangerous.  They are guides, not replacements.  Unfortunately they have replaced thinking to the detriment of patient care. 
    A good example is the requirement that we write an indication for use of any medication in the hospital before the order is filled.  What a brilliant idea!  If done properly, we could tract efficacy and appropriateness.  But what many hospitals do, in order to satisfy the pushback from physicians too lazy or stupid to think this through, is to populate it automatically.  For example, you want to give Imipenem for a sinusitis with complications.  Well that should be the indication.  Not “infection”. 
    I headed the Center for Pediatric Quality at my hospital for 10 years.  Was kicked out of the job despite >$2 million in grants and awards.  They didn’t want real quality.  They wanted to check off check lists.  I am happier in my own world doing it my own way. 

  • Anonymous

    This kind of complaint is not new and I cringe every time I see one.  Both those who blindly implement protocols and those who find them wanting are missing the point.  Neither of them is skilled at the design and management of a process.  Those who write simplistic processes need to acquire the skills which are constantly being honed in high quality medical facilities across the country.  The Instititute for Healthcare Improvement in Boston is one place to learn these skills.  There are a variety of others.  Patients deserve predictable processes that are carefully and thoughtfully designed.  This author describes a situation whcih fails elementary tests of reliable processes.  Instead of complaining he should be helping to design processes that work.

    • S Silverstein

      Maybe, as a surgeon, he doesn’t have the time.  Surgeons have schedules that are a lot busier and demanding than other docs, and than other non-medical personnel.

      • civis isus

        he doesn’t have time to say “hey, this protocol application needs some fine tuning to allow for documented variation requests”? doesn’t seem to involve all that much work.

        After all, aren’t healers mostly interested in doing more healing, rather than husbanding their valuable procedure-repeating (& billing) time? Or maybe you don’t actually want to answer that

  • Denise

    As a layperson, I can’t help but ask about “admitted a gallbladder.” Did a person come with it?

  • Anand Philip

    I see this as a breakdown of evidence based hospital policy, more than the result of algorithms themselves. It is not very difficult to have protocols that do not hamper physician’s decision making. It is easier for a physician to forget to STOP anti biotics after a certain number of doses than remember to continue them in a fraction of patients. In India, anti-biotic over use has rendered hospital acquired infections almost completely susceptible to antibiotics.