Adverse drug events are an epidemic patient safety problem, affecting 5% to 40% of hospitalized patients and 12% to 17% in the 30 days after discharge, ACP Member Jeffrey L. Schnipper, MD, MPH, told attendees at Internal Medicine Meeting 2018. There are two major kinds of medication errors that lead to such events: history errors and reconciliation errors, said Dr. Schnipper, who is director of clinical research on the hospitalist service at Brigham and Women's Hospital and an associate professor at Harvard Medical School in Boston.
“These errors are happening every single day, in your hospitals and in mine,” he said.
With history errors, the sources of information are inaccurate, out of date, or unavailable, or there's no time to access available sources. With reconciliation errors, the orders are wrong despite the history being correct, and this problem often occurs because hospital clinicians do not access the patient's preadmission medication list at the time orders are written, or a clerical error occurs, he said.
“[Reconciliation errors are] a common problem at discharge, because at that point you are reconciling what they came in on with what they're currently on,” Dr. Schnipper said. “Maybe you forget to restart the aspirin you held. Maybe you forget to turn the short-acting med back to the long-acting med, or the statin that's on formulary at your hospital back to the statin that's actually covered by their insurance.”
A structured medication reconciliation process can help solve these problems, Dr. Schnipper told his audience, involving verification, clarification, and reconciliation. In the first step, clinicians should develop an accurate list of the patient's medications. In the second step, medication, dose, and frequency should be verified when orders are written. In the third step, discrepancies between the patient's medication list and medication orders should be identified and corrected, changes should be documented, and the accurate list should be communicated to the patient, caregivers, and receiving clinicians.
“Why is this hard?” Dr. Schnipper asked. “I would say that med rec is a microcosm of almost everything that's wrong with our health care system as a whole.” That includes information systems that don't talk to each other, fragmentation many transitions of care, lack of communication between patients and clinicians and among clinicians, lack of engagement by patients in their own health care, polypharmacy, and frequent changes in regimens, he said.
There's also the lack of value placed on medication reconciliation. “This is not a billable thing that you can do. No one's paying you or your institution to do med rec. It's just really important for patients day to day,” he said.
Dr. Schnipper offered solutions from two studies on which he was the principal investigator: the Multi-Center Medication Reconciliation Quality Improvement Study (MARQUIS) 1, which involved five sites, and MARQUIS 2, which involved 18.
The first component of the MARQUIS intervention was a “med rec bundle,” a list of things Dr. Schnipper said every patient needs: a best possible medication history (BPMH), reconciliation of medications at discharge, patient and caregiver counseling, and forwarding of the medication information to the next clinician.
“To do this work ideally you would do some risk stratification,” he said. “Your highest-risk patients should get a more intensive version of that bundle, while your average-risk patients should get a standard version of that bundle.”
In addition, clinicians received training in how to take a BPMH and perform discharge counseling, and efforts were made to improve access to preadmission medication information, Dr. Schnipper said, ideally by having patients keep an up-to-date list of their medications with them at all times. “If every patient showed up in your emergency department with an accurate list in their wallet, that would make everything a whole lot easier,” he said.
The MARQUIS intervention also included such interventions as implementing or improving health information technology (IT), using social marketing or advertising techniques to explain the importance of the issue to patients and clinicians, and engaging community resources.
In MARQUIS 1, one site out of five did not implement any of the medication reconciliation interventions due to lack of time, Dr. Schnipper said. “This is reality any time we need to do QI work in multiple institutions,” he noted. Of the remaining four sites, three saw improvements in their rate of unintentional discrepancies, for an 8% reduction per month over baseline trends when compared with control units. While results from MARQUIS 2 had not yet been published at the time of the meeting, preliminary data are promising, Dr. Schnipper said.
MARQUIS 2 built on the findings of MARQUIS 1 by refining the medication reconciliation toolkit and disseminating it to additional sites. Both studies were funded by grants from the Agency for Healthcare Research and Quality. Both studies involved a mentoring component, Dr. Schnipper said, where site leaders were paired with mentors to help coach them through the implementation of the project toolkit. Mentoring involved monthly phone calls, site visits, and individualized guidance and feedback.
In both studies, hiring additional pharmacy staff and training existing staff to assist with medication reconciliation and counseling at discharge was one of the most effective intervention components, along with clearly defining each clinician's roles and responsibilities.
“There's no need for your triage nurse in the ED to take a history, for the physician in the ED to take a medication history, for the intern on the floor to take a medication history, and for your nurse to take a medication history,” Dr. Schnipper said. “It should be done once, it should be done early, it should be done correctly, and everyone else should be quickly verifying it.”
The least effective component, perhaps surprisingly, was attempting to improve the IT surrounding medication reconciliation by implementing an electronic health record (EHR). This was due in large part to the fact that this “is a huge effort that pulls resources and time and effort away from a focus on medication safety...All other quality improvement interventions in your hospital basically come to a grinding halt the year before you implement the EHR and probably for at least six months after you implement the EHR,” Dr. Schnipper said.
Dr. Schnipper noted that the success of the MARQUIS intervention depended heavily on the level of institutional support, which in turn depended on whether medication reconciliation efforts aligned well with institutional priorities. “The successful efforts were where executive sponsors from the C-suite...saw that this might reduce readmissions, reduce length of stay, reduce adverse drug events, so therefore put their weight behind it,” he said. He also stressed that administrative support at the hospital level is required for clinician training, project management, and ongoing data collection.
The presence of concurrent QI interventions could either help or hinder medication reconciliation efforts, Dr. Schnipper noted. “They were a barrier if they competed with time, attention, and resources from the med rec QI efforts, but they could be a facilitator if they were complementary to the med rec interventions,” he said.
Training clinicians to take a BPMH was not the most effective component of MARQUIS 1 but was more successful in MARQUIS 2 due to lessons learned, Dr. Schnipper said.
“We realized that it's insufficient to just teach people how to take a good medication history and assume then they're competent at it. Think of this as like [advanced cardiac life support]. You need to watch somebody actually do the chest compressions, do the sequence of the drugs. You have to prove that they actually have competence in doing this particular skill.” He noted that peer-to-peer training, for example, physicians training physicians or pharmacists training pharmacists, is probably the preferred method because it allows learners to observe modeling of best practices.
The most important implication of the MARQUIS findings for clinicians, Dr. Schnipper said, is to be a clinical champion. “Spread the word that med rec is not just a regulatory requirement. This is about medication safety,” he said. “At the end of the day, you as ordering providers are responsible for making sure that those med lists and those orders are correct.”
He stressed that internists don't need to take on the whole med rec process themselves but do need to take some responsibility for its overall quality and should know when to get help from other clinicians. “Med rec errors can undo a lot of otherwise excellent care,” he warned. “You can spend an hour deciding what dose of Lasix to send your [congestive heart failure] patient home with, but if the patient doesn't understand to take that new dose instead of their old dose, that undoes all the good that you just did.”
Clinician educators can also advance med rec improvement by helping to change the curriculum, Dr. Schnipper said. “I certainly never learned how to take a good medication history when I was a medical student or a resident. I now know how, and I'm happy to teach other people, and we do ‘train-the-trainer’ workshops,” he said. “But this needs to be just a regular part of medical education.”
He noted that he often tells his teams that one of every 11 patients is admitted to the hospital because of a medication problem, “either a side effect, or the medication working too well, the medication not working well enough, they're taking the med incorrectly, they're not taking it at all,” he said. “This is just part of good history taking. We need to teach that skill, and get that message out there.”