The Wolf in the Woods: Taming Medication Management with Technology (A Thought Leadership Roundtable)
“The question is, why aren’t we more focused on medication management? And the answer is that health care tends to focus on the wolf closest to the sled. So, if there are eight wolves gathered around the sled, we’re not focusing on the wolf that’s off in the woods.”
Daniel Barchi, CIO, NewYork-Presbyterian
Introduction
DrFirst recently hosted the second roundtable in a series of thought leadership roundtables focused on how to improve outcomes and processes in medication management. Six members of the College of Healthcare Information Management Executives (CHIME) gathered to share insights and envision future innovations to improve patient care. CHIME President and CEO Russell Branzell moderated the roundtable, and DrFirst Chairman and CEO James Chen contributed to the discussion.
CHIME Members participating:
- Daniel Barchi, Group Senior Vice President and Chief Information Officer, NewYork-Presbyterian
- Ray Gensinger, MD, Chief Information Officer, Hospital Sisters Health System
- John Kravitz, Corporate Chief Information Officer, Geisinger Health
- Stephanie Lahr, MD, Chief Information Officer and Chief Medical Information Officer, Monument Health
- Theresa Meadows, Senior Vice President and Chief Information Officer, Cook Children’s Health Care System
- George Reynolds, MD, CHIME Clinical Informatics Executive Adviser
Locating risk in medication management
Effective medication management optimizes patient health outcomes and minimizes error. Error can creep into the medication management process in a number of ways—through patient error, data processing error, or through insufficient staff capacity for medicine documentation and reconciliation.
One of the thorniest issues for clinicians arises from human error in patient reporting. Whether or not a patient is taking a prescribed medication is often known only to the patient, and if the patient accidentally reports incorrectly on their own adherence to dosage for prescriptions, or omits a medication when giving a list, that human error can cause adverse health outcomes. Patients with chronic conditions may experience illness or even death if they cease their medication, while others may suffer from adverse drug interactions if they fail to report all of the drugs they are taking.
For this reason, providers point out that if the profession could ensure perfect reliability in patient reporting, the challenges of medication management would be significantly reduced. “A lot of what we’re dealing with is a people problem that we haven’t quite identified the technology to solve elegantly,” said Dr. Stephanie Lahr, CIO and CMIO at Monument Health. “If you’re a very compliant patient with a straightforward list of medications, then med reconciliation is a great process. But there is a whole group of patients that are very seldom able to report reliably: nursing home patients and dementia patients, for instance.”
Roundtable participants collectively made an informal estimate that only 10% of patients are highly engaged in their medication adherence, follow their correct dosages, and can produce an accurate list of their current medications at any time when asked. The majority of patients cannot be as accurate for a variety of reasons including mental incapacity, lack of time, lack of motivation, or lack of understanding.
A second challenge arises when patients may report correctly, but do not take their medications as prescribed. Lahr noted that there are definitional issues in what it means when clinicians ask if a patient is “taking” a prescribed medication. “The patient may answer, ‘Well, I take it about half the time at half the dose,’ or ‘I take it when I need it.’ Then, how do you impart that information to whoever is truly reconciling the medication and deciding to increase it or decrease it?” Lahr asked.
In addition to patient noncompliance or reporting error, the medication management process can suffer from human transcription errors, incomplete digital reporting, or data migration errors. Even the names of drugs can be a challenge to accurate drug identification and management: generic and brand names vary so widely that it became necessary for the federal government to create a process for normalizing drug labels with RxNorm.
All these risks are why medication management first became heavily regulated to the point that finding time for full compliance is almost impossible for clinicians. The administrative time crunch caused by medication management threatens the time that doctors and nurses need to actually interact with the patient and provide care. The constant balancing act between the need to manage medication and the need to address other patient issues leads to a situation in which too often, patient care suffers: either medication management consumes too much of the appointment, or the medication management is not as thorough as it should be because it must be completed too quickly.
Documentation vs. reconciliation
To tackle the time crunch for clinicians, healthcare organizations have split the medication management process into two major phases: documentation and reconciliation. Medication documentation is the process of interviewing the patient to try to confirm an accurate list of that patient’s medicines and the dosage and frequency at which the patient is taking them. Medication reconciliation is the decision-making act of approving or changing those medications—an act which must be performed by a qualified physician.
Many roundtable participants discussed the addition to their organizational staff of pharmacy technicians (“pharm techs”) or pharmacy students who now serve as medication documentation staff in order to free up the time of the doctors and nurses to provide other care.
“We have pharmacy techs who at least get the meds list cleaned up and get them in a state where a doctor could actually reconcile the meds,” said Theresa Meadows, CIO of Cook Children’s Health Care System. But Meadows pointed out that there is still a deeper problem for physicians.
“We’re a multi-specialty hospital, so many of our physicians—especially surgeons—are not comfortable reconciling meds that they didn’t write orders for,” Meadows said.
Ownership in a specialized, multi-provider system
Roundtable participants agreed that many physicians feel they are not necessarily qualified to make medication recommendations that fall outside their specialty. “Even if the techs have gotten it perfect, the doctors want nothing to do with making a decision about a medication that falls outside their scope of practice,” said Dr. Ray Gensinger, CIO of Hospital Sisters Health System.
As complex as the medication management process can be for the inpatient setting, it becomes even more challenging in the ambulatory setting, where patients have less time to see their doctors and other staff members. And as each patient collects an increasing number of providers across a range of healthcare specialties, the challenge of medication management ownership intensifies.
“With each added provider, the complexity and reality of trying to get to the truth gets more complicated. And I think that’s what we’re going to have to overcome, because we’re going to end up with more and more providers caring for patients,” said Gensinger. “When you get to the discharge, how do we get the pool of providers to weigh in on medication reconciliation?”
Data overload and the need for data curation
Twenty-five years of digitization of medical records has led to the current patient electronic health record (EHR), which has produced clinical benefits for patients, but also liabilities in the form of data overload and a heavy administrative burden. Clinicians need to find the right data to make important decisions, but instead they are overwhelmed with years of irrelevant data in the records. Roundtable participants reported that physicians now resist the thought of any change that adds more data, because there is already too much patient data to assimilate or use in helpful ways. Sometimes, too much data can even cause unnecessary interventions, as when the rise of constant fetal heart-rate monitoring during labor led to false-positive predictions of fetal compromise and unnecessary c-sections. Though data used wisely can significantly improve patient outcomes, the need for data curation is acute.
On the other end of the scale, insufficient data becomes a challenge when lack of interoperability prevents data sharing from one health system to another. Data must be migrated from incompatible systems or outdated legacy systems, creating enormous technical projects that consume budget and resources that might better be used for innovation. Sometimes, solutions intended to transmit data from multiple systems contain only 40-80% of a patient’s records, creating incomplete records that do not lead to optimal care.
Leaders have patched together solutions such as creating one clinical archive to amass all past records rather than trying to maintain 15 (or 150) legacy applications. In other healthcare systems, staff members have been hired to create short abstracts of full records in order to condense data into a more manageable form.
Now, with patient-collected data such as Fitbit-monitoring information also swamping the physician’s data horizon, it is clear that effective medication management solutions must arrive now in order to allow the transformative potential of data to revolutionize healthcare.
Non-technological solutions
Leaders suggested solutions in three general areas: more human power, changes in policy, and advanced technology.
One improvement might lie in creating a new career in the clinical setting for specialists who are more expert than pharmacy techs, but not as high-level (or expensive to hire) as pharmacists. These pharmaceutical specialists might be better able to understand the decisions that lie ahead for the physician and thus complete medical documentation in a more useful fashion that highlights potential issues for the physician in advance.
Another potential source of outcome improvement might occur from changing policy so medical reconciliation is required at fewer trigger points. By choosing those points more carefully, with the knowledge of clinicians as a reference, regulatory policy could significantly ease the burden for a number of healthcare practices.
Both these solutions have weaknesses, however. Adding more expert documenters adds more staff who will still be subject to the same problems of data overload and data error. While changing federal policy may reduce some of the administrative burden, policy cannot completely solve the data issue at the heart of the medication management conundrum.
The sheer mass of data overload eventually led the roundtable participants to collectively envision a future in which data can raise healthcare to a whole new level because the data is curated by artificial intelligence (AI).
Making medication management the case study for digital transformation with AI
As roundtable participants pointed out, one of the challenges in radically improving medication management is overcoming the inertia caused by 25 years of struggling with the status quo without moving the dial. For many providers, too many years without satisfactory solutions in medication management have created a general lack of hope that such game-changing improvement might be possible. But artificial intelligence and machine learning are becoming increasingly common in healthcare technology, as shown in the 2020 CHIME Digital Health Most Wired Survey. And with the rise of AI, the world of what’s possible is opening up to a level of sophistication only previously imagined in science fiction. The potential of more advanced technology to solve decades-old challenges is limitless, and the only question is how long such a transformation will take.
“Health care as a whole needs to pick some narrow lanes where we go all in on automation,” Barchi said. “And we’re not going to go all in on automation for fields like obstetrics or surgery, because they are too multifactorial. But pharmacy is really data-driven, so maybe this is the best choice for massive automation to raise healthcare up to the next level.”
Many leaders see the adoption of AI and machine learning as crucial to the future of healthcare. “It’s essential,” Gensinger said. “We have to embrace it. Because the data is going to come from everywhere. We’re going to be doing more continuous data gathering than ever before. And we need the tools to help find the needles in the haystacks.”
Leaders point out that the expertise of the pharmacist is in creating a decision tree, and because that decision tree rests on data, an AI algorithm could be designed to create very complex decision trees that would stand in for the pharmacist at the doctor’s elbow.
Lahr raised the example of issues in pharmacogenomics such as when antidepressant A may be most effective for a certain patient’s metabolic profile, but antidepressant A cannot be co-prescribed with the medication that would be best for that individual’s high blood pressure. In that case, Barchi suggested, the automated AI decision tree (informed with all the power of the pharmacist’s book) would be equipped to deal with masses of data and “if-thens” in a way that clinicians with limited time cannot.
“If care is being based on a chronic condition or multiple chronic conditions, we look at those because there’s a prevalence,” said John Kravitz, CIO for Geisinger Health. “The perfect application for a technology to sit on top of is the reorganization and reprioritization of the medications.”
Understanding that genomics will drive much of future medicine is key. “We know that in the future, medicine is going to be increasingly protocol-based,” Barchi said. “We’re going to know your metabolic profile. We’re going to know your genomics. We’re going to have enough information to guide you. And the job of the outstanding doctors and nurses will be managing the process and catching the outliers.”
The transformation is coming sooner than the public or even many in the healthcare system realize. Dr. George Reynolds, CHIME Clinical Informatics Executive Adviser, predicted the imminent sea change. “It’s not coming 50 years in the future. It’s coming much sooner.”
Conclusion
Medication management may be the most suitable challenge to use as a case study for the transformative power of artificial intelligence across the healthcare system. The data-driven nature of pharmaceutical case management would be a perfect foundation to allow algorithms to reduce the current heavy burden that data overload places on clinicians. In addition, machine learning has the potential to help minimize the impact of patient error in medication reporting, because ML’s unlimited data-processing capacity would allow the efficient curation of massive amounts of data from continuous-monitoring devices that could track actual patterns of adherence.
Allowing technology to reduce the burden of data management would be the natural solution to what for over two decades has been seen as an intractable, insoluble challenge. In this case, the “wolf in the woods” may be just the right wolf for healthcare to put under harness and charge with leading the sled.
This thought leadership roundtable was written by Rosslyn Elliott, CHIME Editor, and brought to you by DrFirst.
RETURN TO CHIME MEDIA