Adverse drug events caused by medication discrepancies are estimated to affect as many as 40 percent of hospitalized patients and up to 17 percent of patients in the 30 days following discharge. The Society for Hospital Medicine’s MARQUIS Med Rec Collaborative is helping hospitals reduce unintentional medication discrepancies by identifying medication reconciliation process gaps, developing better ways to prescribe and document medications, and measuring improvement over time.
Medication discrepancies are unexplained differences in medication regimens when patients make transitions. Patients might be on one regimen before the hospital, a different regimen in the hospital and another after they leave, not for medical reasons but because providers make mistakes in the process.
Jeff Schnipper, M.D., M.P.H., a physician at Brigham & Women’s Hospital and professor of medicine at Harvard Medical School, is leading the MARQUIS effort. In a recent interview, he said that problems with “med-rec” are a microcosm of all the problems in our healthcare system.
“We have fragmented healthcare systems that don't talk to each other, IT systems that don’t talk to each other, providers who don’t talk to each other, and patients who may have limited health literacy or ability to understand their own medications,” he said, “so it is not so easy to know what meds a patient was on before they came to the hospital.”
MARQUIS is an acronym for Multi-Center Medication Reconciliation Quality Improvement Study, which began five years ago with five hospitals. “That worked pretty well, but we had inconsistent results,” Schnipper said. “Then we did a second study, Marquis II with 18 hospitals, and we are just submitting those results for publication now. At this point, we know it works and we know how to customize it for individual sites and to scale it up widely. We just need to get this out to as many hospitals as possible. So the collaborative has been up and running for two years. We have been doing cohorts of five hospitals at a time and we are ready to scale up to hundreds.”
There are some tools in the EHR that help, Schnipper notes, “but none of the large EHR vendors have very good med-rec modules. At discharge, the EHRs are getting better, but I would still say the EHRs in their native state don’t do as good a job as they should of pulling in the data from all the pre-admission medications and organizing it in a way to make discrepancies clear.”
One vendor collaborating with MARQUIS is DrFirst, which makes solutions that put patient medication history front and center for care teams. “They came across my tools three or four years ago and saw the potential for collaboration,” Schnipper said, adding that he has no formal or monetary relationship with the company. “They have a great series of products, and a lot of what their primary products do is help serve up the medication information in a really effective way to that most complicated step of med-rec, which is taking a best possible medication history,” he explained.
DrFirst approached Schnipper and asked if he could customize the toolkit for their customers. “It becomes a win-win because we are trying to train people on how to take a medication history. It makes sense to create a customized version of our toolkit to train them using the particular set of tools they already have, which happens to be an excellent set of tools and work on recruiting more hospitals to join our collaborative. They have a large number of hospitals they already work with. By definition, these are hospitals that have already decided to take this issue seriously enough that they want to invest in DrFirst products, so it made a lot of sense for this collaboration to happen.”
Schnipper said Marquis I and Marquis II used mentored implementations, in which every site was assigned a mentor who would talk to them every month, and do extensive hand-holding and take them through the current process, the problems with it, which tools they want to adapt, make iterative refinements, and help with data collection. “Instead, in the collaborative I am the mentor for everyone, and we hold monthly office hours for everybody. All the sites get on and talk about how they are doing, and we share best practices and give advice. We have a nice online system to share all of our materials, and there is a wiki and they can ask questions of each other.”
A 14-Month Process
Schnipper described the MARQUIS process. The clinical champions are usually a pharmacist and a hospitalist, leading a team of health IT, nursing and other stakeholders.
“We start with making sure they have the right people on their quality improvement team. You need all the stakeholders involved because med-rec is by definition a team sport,” he said.
Second, they need to understand their current care processes, so the collaborative has them do some process mapping to help figure out where the gaps are. “We like them to understand their processes and outcomes at baseline, and do process mapping so they can start collecting data on discrepancies. It reveals how high the discrepancy rates are. The average patient probably has three errors in their medication list. You are talking baseline error rates of 15 percent of all admission and discharge orders are wrong. That blows away by orders of magnitude other types of medication errors that we spend a lot more money on fixing, such as administration errors. If you don’t know what meds the patient was taking when they came in, you are going to write the wrong order at discharge. It is not rocket science. I do not care how smart a clinician you are. It is such a basic piece of blocking and tackling.”
Then the hospital identifies which interventions they want to work on. “We used to be less prescriptive about that, but we are now more prescriptive,” he said.
One of the key issues is having a dedicated group of trained people taking the best possible medication history, ideally in the emergency room before they are admitted upstairs, he said. “It looks like the best group of people to do this in terms of quality and efficiency is pharmacy technicians. They can be trained to do this very effectively and we now have created a course where they can be trained to do this very well. We help sites make the case for hiring new pharmacy techs if they don’t have them, and then going through training and verifying their competency in doing this work.”
The rest of the 14-month period involves getting interventions up and running, iteratively refining them, and continuing to measure discrepancy rates to see what is working, spreading to larger areas of the hospital and then work to sustain their gains.
One key ingredient is doing some kind of risk stratification of the patients, and then for the highest risk patients having a pharmacist intervention at discharge time to make sure meds are reconciled correctly at discharge, he added.
Another component is getting better access to pre-admission medication sources. “DrFirst does a good job of pulling data from pharmacy and pharmacy benefit managers, but you need to grease the wheel with local primary care practices, if they are on a different EHR platform, and do social marketing to get the message out to patients that you need to keep an accurate and up-to-date list of meds with you all the time, so that when you come into the emergency room it is in their wallet. That would make everyone else’s job a whole lot easier.”
Hospital Motivation to Join Collaborative
What motivations do hospitals have to participate in the MARQUIS Collaborative? Schnipper noted that the Leapfrog Group is now measuring medical discrepancies in the same way, “which is not surprising because I helped them design the measure. This year for the first time, there is going to be public reporting not just that you have done the measurement, but what your results are. We are hoping that will lead to some motivation.”
He added that most hospitals are motivated because they realize these kinds of discrepancies prolong length of stay or lead to increased readmission rates or both and that there are financial gains to be made.
“We have an ROI calculator. They can plug in their own estimate of discrepancy rates and we can estimate how much money they are losing by prolonged length of stay from adverse drug events that happen in the hospital, and how much money they would save by hiring a few pharmacy techs,” Schnipper said. “The other thing that motivates them is if they have had some major event happen, where a patient had a life-threatening or fatal drug event because of a medication reconciliation error. Adverse drug events are happening all the time, and they don’t even know they are due to a med-rec error because no one took a gold standard medication history in the first place to figure it out.”
