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 “” 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