Healthcare Consulting Services
Healthcare Consulting Services with a Focus on Patient Safety Solutions and Quality Improvement Across the Health Care Continuum. Your Patient Safety Resource Solution.
February 28, 2017
The Copy and Paste ETTO
Our September 15, 2009 Patient Safety Tip of the Week “ETTO’s: Efficiency-Thoroughness Trade-Offs” discussed efficiency-thoroughness trade-offs or ETTO’s. That concept, best associated with Erik Hollnagel (Hollnagel 2009), is well known to everyone in the human factors and safety fields. Basically, the ETTO concept means there are certain procedures and practices that we do which make our work more efficient but at the risk we may compromise thoroughness or safety. Hollnagel, in his book, notes that such things usually go right but occasionally go wrong. A classic example of an ETTO is the “copy and paste” function that we all know well from our word processors and spread sheets. Copy and paste functionality is also widely used in electronic medical records. It allows us to easily input a large amount of text, images, etc. into one part of an EHR without having to type in all the details. This can be a huge timesaver.
But there are times when “copy and paste” can go wrong. For example, we might only copy part of a patient’s medication list, inadvertently leaving off some important medications. (That’s particularly a problem when the medication list is long and may span more than one computer screen or is otherwise truncated). Or we may copy information that is no longer accurate, such as copying an old medication list and not amending it to account for medications since discontinued or those added or those whose dose has been adjusted.
There are also instances where an inaccurate piece of information may get deleted (or more properly amended with appropriate attribution) from the medical record but someone copies and pastes that item from a prior part of the medical record, thus perpetuating the erroneous information. A good example is when the record says a patient is allergic to a certain antibiotic but that later gets amended when a physician realizes there was no true allergy. If the original note noting the “allergy” gets copied and pasted, the patient may be deprived of a most appropriate antibiotic in the future.
And in our many articles on wrong patient/patient identification errors we’ve noted that information sometimes gets copied from the chart of one patient inadvertently into the chart of a different patient.
In 2015 two significant studies highlighted the problems associated with “cut and paste”. A National Institute of Standards and Technology (NIST) study (Lowry 2015) showed that the integrity of information in EHR’s is frequently compromised by how data is used and reused, with “copy and paste” being a major contributor. The second was a comprehensive review of the practice by the ECRI Institute (ECRI 2015). A follow up report from NIST is now available (Lowry 2017).
The ECRI study (ECRI 2015) did a nice job of trying to determine the frequency of copy and paste in the EHR. They note that reported rates vary depending upon the definitions used, the venue, and the method used to detect copy and paste (eg. self-report, chart review, direct observation, etc.). You can get all the details in the ECRI study but a good example is the self-reported frequency found in a study by O’Donnell and colleagues (O'Donnell 2009). They found that 90% of physicians surveyed using an EHR for inpatient documentation used copy/paste to write daily progress notes, and 78% identified themselves as high-frequency users. 81% of copy/paste users frequently copied notes authored by other physicians and 72% copied notes from prior admissions.
The ECRI study also found that nearly all aspects of the medical note have been subject to copy/paste, including chief complaint, history of present illness, past medical history, review of systems, physical exam, medications, lab and radiology results, assessment, plan, etc.
The ECRI study noted there is a paucity of studies documenting the adverse consequences of copy/paste. Most of the examples of adverse outcomes are in individual case studies. They do note that the large study of electronic medical records at the VA (Singh 2013) noted substantial errors related to copy/paste but did not provide details (as discussed in our March 2013 What's New in the Patient Safety World column “Diagnostic Error in Primary Care”).
The ECRI study also noted prior attempts to categorize the risk level of the various copy/paste events. One study (Thielke 2007) gave as an example of “high risk” copying notes from another physician or copying notes greater than 6 months old. A “moderate risk” example was copying from oneself 1 to 6 months prior and a “lesser risk” example was copying from oneself from <1 month prior. In the Thielke study 55% of copy/paste events were in the highest risk category, and 18% and 27% in the moderate and lesser risk categories, respectively. Other studies categorized by whether a full note was copied, whether minor changes were made, or substantial changes made.
The ECRI study also noted that many, if not most, physicians recognize that errors might be made through copy/pasted. The O’Donnell study had found that 25% agreed that copy/paste makes progress notes more likely to lead to a mistake in patient care but only 3% reported committing an error related to confusion caused by a note with copy/pasted text. Physicians also agreed that frequent copy/pasting can result in notes that are less accurate, lengthier, and less organized and felt copy/paste facilitated generation of progress notes that were more likely to contain outdated or inconsistent information. However, in keeping with the concept of ETTO’s, O’Donnell’s study also noted that copy/paste had important benefits like a “more trustworthy” medical note, improved documentation of the patient’s hospital course, documentation for legal purposes, and documentation for billing.
Much as we have seen, the ECRI study noted four problems for the medical chart:
The ECRI study also noted the following factors that contribute to problems related to copy/paste:
The ECRI study has numerous recommendations. Responsibilities for the authors of medical record notes should:
Some items that should not be copied include medical student notes or the history of present illness. Also not to be copied from another provider’s notes are history of present illness, review of systems, physical examination, assessment, and plan. But some sections such as past medical history, family history, and social history might be amenable to a “copy-forward” approach with modifications after the author confirmed the accuracy with the patient.
The ECRI study also has numerous recommendations for those who design EHR’s.
With the ECRI study (ECRI 2015) and the first NIST study (Lowry 2015) as background, a follow up report from NIST was just published (Lowry 2017). The researchers in the current study collected data while observing clinicians (nurses and physicians) interacting with the EHR during their routine tasks and then follow-up interviews were conducted. Specifically, they looked at how practices met with four ECRI recommendations:
The current NIST report focused on improving EHR systems with the intent of:
The NIST report recommends that a mechanism for copy/paste should be available in the EHR but that there should be a mechanism where the material to be copied should be visually enhanced so that the copier does not inadvertently copy only part of the information, leaving key information uncopied. There should be a mechanism facilitating verification that “the copied information was read consciously and edited by the clinical provider which would promote the attribution of the source of the information.” Moreover, there should be a display of the “chain of custody” of the information, providing appropriate attribution.
They recommend certain elements be prevented from being copied: demographic information, dates, and any information should be blocked from entry into a blood bank information system. Demographic data should be autopopulated by the EHR and copying demographic information from one chart to another should never be allowed.
The report also has some recommendations about vital sign documentation, including date and time stamping of not only when the vital signs were taken and recorded but also when it was signed, revised and retrieved. (They also recommend including how the vital signs were taken.) Allergies can be copied and pasted but should have a clear “chain of custody” for attribution. They recommend that ensuring that surgical notes be copied in toto since context might be lost if only part of a note is copied.
The report notes that copying a medication list may actually be preferable to using drop down menus (because of the known vulnerabilities to inaccurate selection from drop down lists). But it specifies that copy/paste should never be allowed for ordering new medications (so that the provider is forced to consciously think about the order). And any copied medications should have a clear “chain of custody” for attribution.
They note that the discharge summary is one place where copy/paste can improve efficiency but stress the need for a “chain of custody” for attribution.
And then a point we have stressed over and over: there must be a mechanism to ensure a provider using copy/paste between two systems (eg. copying information from a radiology system into an EHR) the “EHR system must keep the clinician oriented as to which patient’s record they are accessing at any given point in the process”, again with a clear “chain of custody”.
Regarding the recommendation to “Ensure adequate staff training and education regarding the appropriate and safe use of ‘copy and paste’” the current NIST study confirmed the importance of training for copy and paste functionality, noting that training raised awareness of the error-prone nature of copy/paste. Moreover, they found during task performance that participants learned instructions better by watching an instructional video than reading printed material.
And don’t forget that the electronic medical record is not the only healthcare IT system vulnerable to copy and paste errors. In our June 17, 2008 Patient Safety Tip of the Week “Technology Workarounds Defeat Safety Intent” we noted that in our very first barcoding implementation we saw an instance where the label on the medication could not scanned so the nurse simply cut and pasted the bar code information from the computer, totally bypassing the safety feature of a barcoding system!
And in our April 15, 2014 Patient Safety Tip of the Week “Specimen Identification Mixups” we also noted the caveat in laboratory information systems to never allow two patient records to be open at the same time so that copy/paste can never get a report into the wrong chart.
Copy and paste is a great computer tool. We couldn’t do Patient Safety Tip of the Week without it, since we often copy information from previous columns. But if we are not careful, mistakes will occur. Whenever we copy information we have to consciously verify the accuracy of all the copied information. For example, links to other columns or to our references may have expired or changed so we have to verify the current links.
So “copy and paste” and healthcare IT in general are classical ETTO’s. Technology has greatly changed the way we practice medicine and in most respects these changes have been very positive. Nevertheless, technology introduces its own set of unanticipated consequences and errors so we need to remain vigilant at all times and try to design our information systems to anticipate, mitigate, and minimize errors.
See some of our other Patient Safety Tip of the Week columns dealing with unintended consequences of technology and other healthcare IT issues:
Hollnagel E. The ETTO Principle: Efficiency-Thoroughness Trade-Off. Why Things That Go Right Sometimes Go Wrong. Burlington, VT: Ashgate Publishing Company, 2009
Lowry SZ, Ramaiah M, Patterson ES, et al. NISTIR 7804-1. Technical Evaluation, Testing, and Validation of the Usability of Electronic Health Records: Empirically Based Use Cases for Validating Safety-Enhanced Usability and Guidelines for Standardization. National Institute of Standards and Technology 2015; October 2015
ECRI Institute. Copy/Paste: Prevalence, Problems, and Best Practices. Health Technology Assessment Information Service Special Report. ECRI Institute 2015; October 2015
Lowry SZ, Ramaiah M, Prettyman SS, et al. NISTIR 8166. Examining the ‘Copy and Paste’ Function in the Use of Electronic Health Records. National Institute of Standards and Technology. January 2017
O'Donnell HC, Kaushal R, Barron Y, et al. Physicians' attitudes towards copy and
pasting in electronic note writing. J Gen Intern Med 2009; 24(1): 63-68
Singh H, Giardina TD, Meyer AND, et al. Types and Origins of Diagnostic Errors in Primary Care Settings. JAMA Intern Med 2013; 173(6): 418-425 published online February 25, 2013
Thielke S, Hammond K, Helbig S. Copying and pasting of examinations within the electronic medical record. Int J Med Inform 2007; 76(Suppl 1): S122-128.
Print “The Copy and Paste ETTO”
To get "Patient Safety Tip of the Week " emailed to you, click here and enter "subscribe" in the subject field.
If you don't see the search term you expected to see here, its probably because that tip already went to our Tip of the Week Archive. We do a new tip every week. Click here to search the entire site or you can Go to Tip of the Week Archive a patient safety resource solution loaded with tips, tools, and techniques you can use in your patient safety and quality improvement initiatives. Or it may have moved to our What's New Archive.
Click here to see the consulting services and patient safety solutions that we provide.
Click on the "Contact Us" button at the left to send us your comments on our "What's New in the Patient Safety World" columns.
To get "Patient Safety Tip of the Week "emailed to you, click here and enter "subscribe" in the subject field.
It’s been over 6 years since we last discussed adverse events related to loading doses of various medications (see our December 14, 2010 Patient Safety Tip of the Week “NPSA (UK): Preventing Fatalities from Medication Loading Doses”).
But a recent AHRQ Web M&M brought the issue of problematic loading doses to our attention again (Mucksavage 2017). A patient with a known seizure disorder had a subtherapeutic serum phenytoin level. A neurologist verbally recommended that she be re-loaded with phenytoin. The ED physician ordered the correct loading dose of intravenous (IV) phenytoin, to be administered every 8 hours for 3 doses. But there was miscommunication and the patient was not switched back to a regular maintenance dose after the 3 loading doses. So the patient continued to receive IV phenytoin every 8 hours. And after 3 days developed signs and symptoms of phenytoin intoxication. It was only then that the error was recognized.
Phenytoin, of course, happens to be one of the medications most commonly involved in loading dose errors. In our December 14, 2010 Patient Safety Tip of the Week “NPSA (UK): Preventing Fatalities from Medication Loading Doses”) it was one of the 4 medications most commonly involved in such events in the report by the UK National Patient Safety Agency (NPSA 2010), the other 3 being warfarin, amiodarone, and digoxin. In that NPSA report 11% the loading dose was either repeated or continued incorrectly as the maintenance dose as was done in the above case. The NPSA report also noted that handovers and transitions of care were particularly vulnerable to missed communications regarding loading/maintenance doses.
Perhaps one of the unspoken issues about the safety of loading doses is asking the question “Is a loading dose of this medication really needed at this time?”. Over the years we’ve seen numerous instances where there was probably overaggressive treatment that led to problems. To repeat from our December 14, 2010 Patient Safety Tip of the Week “NPSA (UK): Preventing Fatalities from Medication Loading Doses”: for phenytoin the answer depends on the clinical circumstances. Obviously for true status epilepticus an intravenous loading dose is appropriately indicated. But for a patient who has had a single seizure (unless that seizure happens to occur during neurosurgery) and is now back to their usual cognitive baseline, an intravenous loading dose is probably not necessary. The gray zone would be in the patient who has had a flurry of several seizures but does not meet the definition of status epilepticus.
The rate of administration of IV phenytoin (and IV fosphenytoin) is the critical factor in producing hypotension, bradycardia or cardiovascular collapse. The rate is not to exceed 50 mg/min. in adults (1-3 mg/kg/minute in neonates) and should be by a slow IV push, not an infusion. That means that for a typical loading dose in an adult (1000-1500 mg), the physician would need to spend 20-30 minutes administering the drug. Over the years we’ve seen corners cut and either the rate would be accelerated or an IV infusion would be used. Some of that is because many neurologists have not seen significant cardiac side effects from IV phenytoin. But those of us old enough to remember giving IV phenytoin to cardiac patients (once upon a time it was used more frequently as a second- or third-line antiarrhythmic agent) recall watching blood pressures bottom out as we increased rates of the IV push. A recent study showed that 39% of patients receiving an IV loading dose of fosphenytoin had hypotension as an adverse effect (Clark 2016). An FDA review (FDA 2010) notes that the majority of cardiovascular deaths (for both IV phenytoin and fosphenytoin) occurred in adults and at recommended doses. Most had pre-existing cardiovascular disease.
Another key issue with phenytoin is use of the wrong dilution technique. It is supposed to be given in normal saline, not glucose solutions. Note also that intravenous phenytoin has been associated with the “purple glove syndrome” (FDA 2010), a rare but serious condition. That was actually the primary reason for the FDA safety review.
The Pennsylvania Patient Safety Authority (PPSA) published one of its advisories on loading doses in 2012 (Carson 2012). They found 580 events related to loading doses over an 8-year period in their Pennsylvania Patient Safety Reporting System (PA-PSRS). Over 70 medications were involved. Vancomycin was the drug most frequently involved, accounting for 14.8% of reports. Ten of the top 20 medications in the PPSA report were also in the top 20 in the UK NPSA study. These were amiodarone, caffeine citrate, clopidogrel, digoxin, gentamycin, heparin, magnesium sulfate, morphine, phenytoin, and vancomycin. Interestingly, warfarin did not make the PPSA top 20 list (warfarin was the most frequently involved drug in the UK NPSA study). Phenytoin was the only drug in the top 5 list in both studies, perhaps not surprisingly given that it is used extensively.
Recognizing that you need to consider loading doses and maintenance doses as part of a package, the PPSA report categorized the events as follows:
Missed or omitted loading dose accounted for 25.5% of their reports. Patient transfer, either within or between facilities, was a major contributing factor. For example, the loading dose is often ordered in the ED but the patient transferred prior to receiving it and then only a maintenance dose is given at the destination. Wrong loading doses were often related to the fact that dose calculations (based, for example, on patient weight) are needed for many of the involved drugs. One specific contributing factor the PPSA noted was that sometimes the pharmacy would deliver a loading dose and maintenance dose to the unit at the same time and staff on the unit would incorrectly select the maintenance dose. Loading dose given multiple times accounted for 7% of reports in both the PPSA and UK NPSA studies. Again, the patient transfer process was a contributing factor in almost 20% of these cases. One factor we’d speculate about would be non-integration of electronic medical records between the ED and the main hospital. In the early days of electronic medical records and CPOE we often saw implementation take place in a piecemeal fashion and ED’s often lagged behind or even had different IT systems. So a loading dose given in the ED might not be recorded in the inpatient IT system, creating the opportunity for the double loading dose error.
Fortunately, two of the drugs most often mentioned in the UK NPSA study have likely dropped significantly in rank of medications associated with loading dose errors. First is digoxin. Its dropoff has nothing to do with the risks of loading doses but rather with the significant reduction in the use of digoxin over the last 2 decades.
Second is warfarin. Whereas typical practice years ago was to give a loading dose of warfarin, wait a few days and then resume warfarin at a maintenance dose based upon the INR result, such practice has largely changed. A systematic review in 2010 (Heneghan 2010) concluded that there is no advantage to loading patients with a 10 mg. dose compared to starting with 5 mg. daily and they discouraged use of the 10 mg. dose, particularly in elderly patients. The 10 mg. dose may or may not get the patient to a therapeutic INR faster (depending on which study you read) but may also be associated with early overanticoagulation and there is even some theoretical concern that the loading may actually promote the early hypercoagulability sometimes seen during warfarin initiation. But there are still some who recommend a higher initial dose. More recent recommendations (Witt 2016) are:
Interestingly, both the CHEST guideline for VTE therapy (Kearon 2016) and the AHA/ACC/HFS guideline for management of atrial fibrillation (January 2014) are silent on loading doses of warfarin. Warfarin loading doses are also becoming less frequent because so many patients are instead being started on novel oral anticoagulants (NOAC’s) instead of warfarin. In fact, that new VTE guideline suggests use of non-vitamin K antagonist oral anticoagulants (NOACs) over warfarin for initial and long-term treatment of VTE in patients without cancer (see our February 2016 What's New in the Patient Safety World column “Updated VTE Guidelines from ACCP”).
Another drug that has often been associated with loading dose errors is acetylcysteine, used for the treatment of acetaminophen poisoning (Hayes 2008). The dosing regimen is complex, consisting of a loading dose followed by 2 maintenance doses, each with different infusion rates.
As we noted above, transitions of care may be particularly vulnerable to errors related to loading doses. We’ve noted that mistakes commonly occur with ED-to-inpatient transitions where loading doses are ordered in the ED and the patient is admitted to the hospital. Sometimes the loading dose is assumed to have been given in the ED when, in fact, it was not. Other times the loading dose order is assumed to be the maintenance dose order and very high doses are continued on a daily basis. But another very vulnerable scenario is the LTC-to-ED-to-LTC scenario where a long-term care (LTC) patient is seen in the ED but sent back to the LTC facility. Often the notes accompanying the patient back to the LTC are insufficiently clear regarding the recommendations for maintenance therapy and patients may end up getting high daily doses of the medication. Don’t rely on just the written notes in such scenarios. A verbal communication with the LTC facility to clarify dosing of that medication can go a long way to avoiding errors.
Lastly, don’t forget the most common transition of care: ED-to-PCP or ED-to-SCP. We remain puzzled in this age of electronic medical records how frequently the physicians responsible for the patient on a daily basis (either the primary care physician or the specialist managing a particular problem) are not made aware that their patient was even in the ED! It’s rare enough that the PCP gets notified of the ED visit and sometimes the PCP is notified but the specialist who is the prescriber of the medication at issue does not get notified. Patients after an ED visit are often confused about how to take their medications and may go home thinking they should continue taking the higher dose given in the ED.
The following is a list of the previous NPSA and PPSA recommendations supplemented with some of our own recommendations:
Technological solutions are obvious potential means to avoid such errors. But does CPOE actually reduce the chance of errors with loading doses or could it paradoxically increase that risk? We’ve seen some pretty “clunky” IT systems that are not particularly user-friendly when it comes to ordering medications. That is especially so when the order is a complex one in which different doses of a drug are being given on different days, as is the case with loading doses followed by maintenance doses. Considerable confusion may occur when entering such orders, whether directly entered by the physician or entered by a nurse or pharmacist. We have seen instances where the loading dose of a drug gets continued every day or ones where the patient gets both the loading dose and maintenance dose on the same days. Theoretically, use of standardized order sets or protocols (whether paper or electronic) may help avoid such errors but such have not specifically been studied for drugs with loading doses.
The other problem, of course, is that clinical decision support tools that can make CPOE and pharmacy computer systems safer are still suboptimally used. Many current systems do not provide dose range alerts that would flag a relatively high dose of a medication for verification. Also, most current systems do not require an indication field be filled out for each drug. A good system would require input of the indication, with a check box or drop down list where “loading dose” could be indicated (and the system programmed to not continue loading doses beyond the specified time period).
Loading doses can even be a problem when they are not loading doses! One of the medication error slides we like to show is a hand-written prescription for 300 mg of an anticonvulsant that was intended to be taken at bedtime. However, the “S” in “qHS” was missing. The pharmacist interpreted the “qH” as “q4” and assumed that “q4h” was intended. While the pharmacist recognized this would be a very large dose of this anticonvulsant, he also assumed that this was likely a loading dose. It was thus dispensed with the directions to “take every 4 hours” and the patient presented to the ER several days later with anticonvulsant toxicity. We like this particular example because it demonstrates several cognitive biases:
Loading doses are an error-prone facet of the medication process, particularly at transitions of care, that have been underrecognized but have the potential to cause significant patient harm. You should consider adding an initiative on loading doses to your medication safety program. At a minimum you should try to get a handle on how often and for which drugs loading doses are being used in your organization.
Mucksavage JJ, Tesoro EP. Cases & Commentaries. Hazards of Loading Doses. AHRQ Web M&M. Published January 2017
NPSA (UK). Rapid Response Report. Preventing Fatalities from medication loading doses. November 2010
Rapid Response Report
Clark SL, Leloux MR, Dierkhising RA, et al. IV fosphenytoin in obese patients
Dosing strategies, safety, and efficacy. Neurology Clinical Practice 2016; Published online before print November 4, 2016
FDA (US Food and Drug Administration). Joint Meeting of the Peripheral and Central Nervous System Drugs Advisory Committee and the Drug Safety and Risk Management Advisory Committee. November 3, 2010
Carson SL, Gaunt MJ. Events associated with the prescribing, dispensing, and administering of medication loading doses. PA-PSRS Patient Saf Advis 2012; 9: 82-88 September 2012
Heneghan C, Tyndel S, Bankhead C, et al. Optimal loading dose for the initiation of warfarin: a systematic review. BMC Cardiovascular Disorders 2010; 10:18
Witt DM, Clark NP, Kaatz S, et al. Guidance for the practical management of warfarin therapy in the treatment of venous thromboembolism. J Thromb Thrombolysis 2016; 41: 187-205
Kearon C, Akl EA, Ornelas J, et al. Antithrombotic Therapy for VTE Disease: CHEST Guideline. Chest 2016; January 2016
January CT, Wann LS, Alpert JS, et al. 2014 AHA/ACC/HRS Guideline for the Management of Patients with Atrial Fibrillation: Executive Summary. A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and the Heart Rhythm Society. Journal of the American College of Cardiology 2014; 64(21): e1-e76
Hayes BD, Klein-Schwartz W, Doyon S. Frequency of medication errors with intravenous acetylcysteine for acetaminophen overdose. Ann Pharmacother 2008; 42(6): 766-70
Print “March 2017 Loading Doses Again”
In our May 27, 2014 Patient Safety Tip of the Week “A Gap in ePrescribing: Stopping Medications” we highlighted what we consider to be a major flaw in current e-prescribing systems, namely that they do not put the same emphasis on stopping medications as they do on starting them.
In that column we noted a case report in the Medical Journal of Australia (Tong 2014) in which discontinuation of one medication led to excessive levels of a different medication because there had been a drug-drug interaction. And, while there were clearly some communication issues that contributed to that adverse event, we commented about the gaps in our information technology systems that might have prevented it. Even though most regions are developing HIE’s or RHIO’s that integrate health information from multiple sources, those resources are often not routinely accessed by physicians and may not yet be integrated into the EMR’s and e-prescribing systems in physician offices. Moreover, e-prescribing integration with community pharmacies typically covers medications ordered but seldom covers discontinuation of medications.
We also highlighted a critical issue: stopping a medication is much different than starting one. The clinical decision support built into our EMR’s and e-prescribing systems generally is pretty good at identifying potentially serious drug-drug interactions and generating alerts at the time a medication is prescribed. That presumes the alerts are turned on and the “severity” threshold for the particular alert is enabled. (To avoid alert fatigue we usually recommend that only the more serious alerts are enabled.)
But stopping a medication is much different. Most systems are not programmed to generate any alerts at the time you discontinue a medication. Hence, even if your system would have generated a drug-drug interaction alert when you first prescribed a medication, it would not likely generate an alert later when you discontinue that medication. Moreover, starting a medication requires an active process – you either write a prescription, enter one into a computer, or call the pharmacy. Whereas discontinuing a medication is often more passive – you may just tell the patient over the phone to stop it when the patient calls about a potential side effect. You don’t call the pharmacy to stop it. And, if there was no associated office visit, you might forget to update the patient’s medication list in your EMR (or paper records) until the patient’s next office visit.
Another problem is that a patient may continue to get medications that you thought you had stopped. A study done in a large multispecialty group practice in Massachusetts (Allen 2012) showed that among targeted medications that were electronically discontinued (on the practice’s EMR) 1.5% were subsequently dispensed by a pharmacy at least once. And this was just at the practice’s internal pharmacy. How often this happened at community pharmacies was not known. Moreover, when they did manual chart reviews of selected high-risk medications that had been discontinued they found that 12% of cases were associated with potential harm.
The authors note that when a physician discontinues a medication on an EMR he/she often (erroneously) assumes that such information is being transmitted to the pharmacy. Such is seldom the case with today’s EMR systems. Further, many pharmacies today have sophisticated systems that let you know, as a patient, that you have a refill waiting for you at the pharmacy. Patients may erroneously presume that their physician restarted that medication.
And our February 28, 2017 Patient Safety Tip of the Week “The Copy and Paste ETTO” reminds us how the copy/paste function in today’s healthcare IT systems can lead to erroneous medication lists that might result in a patient being inappropriately restarted on a medication that had actually been discontinued.
Fischer and Rose (Fischer 2017) in a recent JAMA viewpoint article point out that outside of the VA health system and a few integrated health systems, most US healthcare has little or no capability for e-discontinuation. They note that it is common for patients with chronic diseases to have prescription orders that may be refillable for up to a year. And that the only way a physician can notify the pharmacy of a medication discontinuation is to phone the pharmacy, which is time-consuming and non-reimbursable, so few physicians do this when stopping a patient’s medication. They note that a standard for e-discontinuation, called CancelRx, has been available for 2 decades as part of the SCRIPT standard for e-prescribing. They further note that MACRA includes rules that require EHR’s to include the ability to cancel prescriptions and other features of the SCRIPT 10.6 standard. But its use by pharmacies is not mandated and there is no way to ensure that pharmacies can receive and process such messages.
Fischer and Rose note that this failure to integrate e-discontinuation is occurring against a background where numerous online pharmacies, chain pharmacies, and community pharmacies are contacting patients by multiple means (phone, email, smartphone apps, etc.) to remind them to refill their medications.
From the case discussed earlier, it’s important when stopping a medication to look at all the other medications a patient is taking and assess the likelihood that blood levels or some physiologic effect may be altered when you stop this one medication. Having clinical decision support tools available to help us spot a drug-drug interaction that will disappear when stopping a drug would be very helpful. Also when stopping a medication we always need to decide whether the medication can simply be stopped all at once or whether it needs to be tapered to prevent a withdrawal syndrome. Having a clinical decision support tool to alert us when to taper would also be helpful. And, just as we do when starting a drug, we need to tell the patient what symptoms or signs to watch out for and what to do if they occur.
There is another very important point we need to add to the process of e-discontinuation. Just as we have advocated for inclusion of the indication for new prescriptions, it is important that we always somehow record why we have discontinued a medication. How often have you suggested a medication and your patient says “yes, I was on that medication once" but can’t tell you why they were taking it or why it was stopped. Was it simply not effective (for whatever indication it was prescribed, which may not even be the reason you are now recommending it) or was it stopped because of some unwanted effect? And was the unwanted effect an allergic response, idiosyncratic response, an anticipated side effect, or simply a dose-related side effect. It’s very important to have details available about the reasons for discontinuation. Also, as we noted above, medications are often discontinued at times when a physician or other prescriber may not have access to the EHR or e-prescribing system. Often they get a phone call from a patient and tell them over the phone to stop the medication and then forget to record that in the patient record.
Too bad we’re not as good at stopping medications as we are with starting them. Time to focus on all aspects of the medication use process – including how to properly discontinue them. Many articles have been written about the medication use process including the following stages or phases: deciding about treatment, ordering or prescribing, transcribing, preparing/dispensing, administering, and monitoring (hospitals or pharmacies might also add at the beginning selection, procurement, and storage). We could not find any that include discontinuation as a stage or phase of the medication use process. It’s time to add that.
And don’t forget some of our past columns on deprescribing:
Tong EY, Kowalski M, Yip GS, Dooley MJ. Impact of drug interactions when medications are stopped: the often forgotten risks. Med J Aust 2014; 200 (6): 345-346
Allen AS, Sequist TD. Pharmacy Dispensing of Electronically Discontinued Medications. Ann Intern Med 2012; 157(10): 700-705
Fischer S, Rose A. Responsible e-Prescribing Needs e-Discontinuation. JAMA 2017; 317(5): 469-470
We’ve done numerous columns on the need to reduce the number of unnecessary CT scans. Not only do unnecessary scans expose patients to ionizing radiation but they add expense and often uncover incidental findings that trigger the “diagnostic cascade” or “investigation momentum” where one test leads to another and on and on…. Of course, there is probably an excess performance of CT scans for most body parts but CT scans of the head are high on the list of CT scans that are overused. So there have been numerous attempts to develop ways of minimizing the ordering of unnecessary head CT scans.
One of the ways we’ve attempted to optimize the use of head CT scans is using clinical decision rules. There’s no shortage of clinical decision rules guiding the ordering of CT scans in patients with minor head trauma. We have the Canadian CT Head Rule (Stiell 2001), the New Orleans Head CT Rule (Haydel 2000), and the NICE guideline (NICE 2014) in adults. And for children we have CHIP (Smits 2007), CATCH (Osmond 2010), and the NICE guideline (NICE 2014).
A recent study looked at the appropriateness of head CT scans for minor head trauma using the Canadian CT Head Rule (CCHR) as the guideline (Klang 2017). The authors retrospective reviewed 955 head CT scans and found 10.9% were not indicated according to the CCHR. However, for patients under the age of 65, 37.3% of scans ordered were not indicated according to that rule. Looking at factors associated with inappropriate ordering of head CT scans they found that neurologists (present company, of course, excluded!) were 3.5 times more likely to order them. Surgeons were statistically less likely to order. They did not find any significant difference by seniority of the ordering physician. Also, regarding injury mechanism, four-wheel motor vehicle accidents and being hit on the head with an object were associated with higher rates of non-indicated CT scans. Interestingly, motor vehicle accident as a pedestrian and two-wheel vehicle driver were associated with lower rates of non-indicated CT scans. The study did confirm that the CCHR had 100% sensitivity and 100% negative predictive value for either brain hemorrhage or fractures. The authors do note that its possible those cases where a neurologist was involved may have been more complicated and perhaps could have had other indications for CT scanning. The authors suggest that interventions to reduce the frequency of non-indicated head CT scanning might include targeted education of staff members, protocol implementation, and implementation of computerized decision rules.
A previous study had compared compare the cost-effectiveness of using selective CT strategies with that of performing CT in all patients with minor head injury (Smits 2010). Five strategies were evaluated (1) CT performed in all patients with minor head injury (2) selectively according to the New Orleans criteria (NOC) (3) selectively according to the Canadian CT head rule (CCHR) (4) selectively according to the CT in head injury patients (CHIP) rule or (5) in no patients. A Markov model was used to analyze long-term costs and effectiveness. Results showed that performing CT selectively according to the CCHR or the CHIP rule could lead to substantial U.S. cost savings ($120 million and $71 million, respectively), and the CCHR was the most cost-effective at reference-case analysis. When the prediction rule had lower than 97% sensitivity for the identification of patients who required neurosurgery, performing CT in all patients was cost-effective. The CHIP rule was most likely to be cost-effective. The authors concluded that selecting patients with minor head injury for CT renders cost savings and may be cost-effective, provided the sensitivity for the identification of patients who require neurosurgery is extremely high. But uncertainty regarding long-term functional outcomes after minor head injury could justify the routine use of CT in all patients with these injuries.
But there are numerous scenarios where these clinical decision rules cannot be applied. For example, all those rules basically do not apply to patients who are on anticoagulants. We’ve discussed CT scanning in patients on anticoagulants (see the full list of prior columns below). But two other scenarios not covered are: (1) the patient who is intoxicated and (2) the patient first presenting to the ED beyond 24 hours.
Regarding the alcohol-intoxicated patient presenting to the ED with altered mental status, a recent study provides some reassuring evidence about timing of CT scanning (Granata 2017). The authors did a retrospective review of patients presenting to the emergency department (ED) with altered mental status and alcohol intoxication who had CT scanning at varying times after presentation. Of the 5943 patients included in the study none of those scanned in less than 3 hours had intracranial findings on imaging requiring neurosurgery, whereas 1 patient with a deferred CT scan required a neurosurgical intervention (which was not emergently performed). The authors conclude that CT scanning of alcohol-intoxicated patients with altered mental status is of low clinical value and that deferring CT imaging while monitoring improving clinical status appears to be a safe practice.
Another recent study looked at CT scanning in those patients with head injury presenting more than 24 hours after the injury (Marincowitz 2016). They compared how the NICE guideline (NICE 2014) predicted intracranial injuries in those patients presenting within or after 24 hours from the injury. They found that 8.4% of CT scans had traumatic abnormalities in those presenting within 24 hours and 9.9% in those presenting after 24 hours. The sensitivity of the guidelines for intracranial injuries was 98% for those presenting within 24 h and 70% for those presenting after 24 h of injury. The presence of a guideline indication did predict significant injury and this was unaffected by time of presentation. The authors conclude that existing guidelines appear to predict traumatic CT abnormalities irrespective of timing of presentation but that their sole use in patients presenting after 24 hours may result in significant injuries not being identified.
Some of our previous columns on head trauma in the anticoagulated patient:
April 16, 2007 “Falls With Injury”
July 17, 2007 “Falls in Patients on Coumadin or Heparin or Other Anticoagulants”
June 5, 2012 “Minor Head Trauma in the Anticoagulated Patient”.
July 8, 2014 “Update: Minor Head Trauma in the Anticoagulated Patient”
Stiell IG, Wells GA, Vandemheen K, et al for the CCC Study Group. The Canadian CT Head Rule for patients with minor head injury. Lancet 2001; 357: 1391–96
Haydel MJ, Preston CA, Mills TJ, et al. Indications for Computed Tomography in Patients with Minor Head Injury. N Engl J Med 2000; 343: 100-5 (New Orleans Head CT Rule)
NICE (UK National Institute for Health and Care Excellence). Head injury: assessment and early management. Clinical guideline [CG176] Published date: January 2014
NICE imaging algorithm
Smits M, Dippel DWJ, Steyerberg EW, et al. Predicting Intracranial Traumatic Findings on Computed Tomography in Patients with Minor Head Injury: The CHIP Prediction Rule. Ann Intern Med. 2007; 146: 397-405
Osmond MH, Klassen TP, Wells GA, et al. CATCH: a clinical decision rule for the use of computed tomography in children with minor head injury. Can. Med. Assoc. J., Feb 2010; early release published February 8, 2010 doi:10.1503/cmaj.091421
Klang E, Beytelman A, Greenberg D, et al. Overuse of Head CT Examinations for the Investigation of Minor Head Trauma: Analysis of Contributing Factors. J Amer Coll Rad 2017; 14(2): 171-176 Published online: November 8, 201
Smits M, Dippel DWJ, Nederkoorn PJ. Minor Head Injury: CT-based Strategies for Management—A Cost-effectiveness Analysis. Radiology 2010; 254: 532-540
Granata RT, Castillo EM, Vilke GM. Safety of deferred CT imaging of intoxicated patients presenting with possible traumatic brain injury. Am J Emerg Med 2017; 35(1): 51-54
Marincowitz C, Allgar V, Townend W. CT Head Imaging in Patients With Head Injury Who Present After 24 h of Injury. A Retrospective Cohort Study. Emerg Med J 2016; 33(8): 538-542
We’ve done many columns on handoffs/handovers and are advocates of using structured tools to facilitate such. There is certainly no shortage of structured tools and formats for handoffs (see our February 14, 2012 Patient Safety Tip of the Week “Handoffs – More Than Battle of the Mnemonics” and the many columns listed below).
One problem we’ve always encountered is that some patients may get “shortchanged” in such handoffs depending upon time limitations and the order of prioritization, i.e. patients discussed earlier in the handoff tend to be discussed in more detail. In our January 29, 2013 Patient Safety Tip of the Week “A Flurry of Activity on Handoffs” we noted a handoff study (Cohen 2012) which showed that in intensive care unit attending-to-attending handoffs at the end of the week, patients discussed earlier had a disproportionate amount of time allocated. This finding was irrespective of the severity or complexity of the patient’s case. Cases earliest in the handoff sessions had about 50% more time in discussion that those discussed toward the end of the handoff.
Now a new study has looked at the impact of two structured handoff tools on multidisciplinary rounds (Abraham 2016). The researchers compared multidisciplinary rounds (MDR’s) in two comparable MICU’s in an academic medical center. One group used a traditional SOAP format (like we’ve used in progress notes for many years). The other used HAND-IT (Abraham 2012), a tool organized by body systems. They then recorded MDR’s and analyzed them for total duration, duration for individual patients, and interruptions/distractions unrelated to the patient being discussed. They did not find that the order of patient presentation impacted time spent on the patient or communication breakdowns. However, for the problem-based (SOAP) tool, there was a significant linear relationship between the time spent on discussing a patient and the number of communication breakdowns. This effect was much less when using the HAND-IT tool. They note that the HAND-IT tool “required more effort and time to gather and document information, but it reduced the time spent and additional effort during rounds to address the information gaps.”
So add one more tool and mnemonic to your handoff toolkit!
Read about many other handoff issues (in both healthcare and other industries) in some of our previous columns:
May 15, 2007 “Communication, Hearback and Other Lessons from Aviation”
May 22, 2007 “More on TeamSTEPPS™”
August 28, 2007 “Lessons Learned from Transportation Accidents”
December 11, 2007 “Communication…Communication…Communication”
February 26, 2008 “Nightmares….The Hospital at Night”
September 30, 2008 “Hot Topic: Handoffs”
November 18, 2008 “Ticket to Ride: Checklist, Form, or Decision Scorecard?”
December 2008 “Another Good Paper on Handoffs”.
June 30, 2009 “iSoBAR: Australian Clinical Handoffs/Handovers”
April 25, 2009 “Interruptions, Distractions, Inattention…Oops!”
April 13, 2010 “Update on Handoffs”
July 12, 2011 “Psst! Pass it on…How a kid’s game can mold good handoffs”
July 19, 2011 “Communication Across Professions”
November 2011 “Restricted Housestaff Work Hours and Patient Handoffs”
December 2011 “AORN Perioperative Handoff Toolkit”
February 14, 2012 “Handoffs – More Than Battle of the Mnemonics”
March 2012 “More on Perioperative Handoffs”
June 2012 “I-PASS Results and Resources Now Available”
August 2012 “New Joint Commission Tools for Improving Handoffs”
August 2012 “Review of Postoperative Handoffs”
January 29, 2013 “A Flurry of Activity on Handoffs”
December 10, 2013 “Better Handoffs, Better Results”
February 11, 2014 “Another Perioperative Handoff Tool: SWITCH”
March 2014 “The “Reverse” Perioperative Handoff: ICU to OR”
September 9, 2014 “The Handback”
December 2014 “I-PASS Passes the Test”
January 6, 2015 “Yet Another Handoff: The Intraoperative Handoff”
Cohen MD, Ilan R, Garrett L, et al. The Earlier the Longer: Disproportionate Time Allocated to Patients Discussed Early in Attending Physician Handoff Sessions. Arch Intern Med 2012; 172(22): 1762-1764
Abraham J, Kannampallil TG, Patel VL, et al. Impact of Structured Rounding Tools on Time Allocation During Multidisciplinary Rounds: An Observational Study. JMIR Hum Factors 2016; 3(2): e29
Abraham J, Kannampallil T, Patel B, et al. Ensuring Patient Safety in Care Transitions: An Empirical Evaluation of a Handoff Intervention Tool. AMIA Annu Symp Proc 2012; 2012: 17-26
Print “March 2017 Loading Doses Again”
Go to the "Whats New Archive"
To get "What's New in the Patient Safety World"emailed to you,click here and enter "subscribe" in the subject field.
To get "What's New in the Patient Safety World"emailed to you,click here and enter "subscribe" in the subject field.