by Colin Banas, Chief Medical Officer, DrFirst
November 2022 – Physicians know this scenario well: A patient shows up with a piece of paper scribbled with lists of medications or hands over a paper bag full of old pill bottles. These scraps of information are better than nothing. Many patients have only vague recollections or no details at all about their prescriptions.
This is more than a complex puzzle for clinicians to piece together. If medication history is assembled incorrectly, the consequences can be severe. Unfortunately, getting it wrong is quite common. Seven of every ten patients admitted to the hospital through the emergency department (ED) have mistakes on their home medication lists. And 85% of inpatient medication errors originate from medication histories collected during admission. With 1 in 100 medication errors resulting in an adverse drug event, mistakes and omissions in medication history can cause serious harm. That is why The Joint Commission added medication reconciliation as a National Patient Safety Goal in 2005. Yet, nearly 20 years later, we are still talking about the problem instead of solving it.
We urgently need to treat medication reconciliation as more than a check-the-box task to avoid errors and improve care in other ways. For example, easy access to patients’ complete medication history means clinicians can make informed decisions that prevent adverse drug events and readmissions. But let’s go one step further and examine patient prescription fill records to identify those who aren’t taking their medications as prescribed. With this information, we can address barriers to compliance, such as costs or side effects, by recommending a therapeutically equivalent alternative. Then, by sorting data by specific populations of high-risk patients with chronic health conditions, we can identify those who could benefit from clinical interventions to help them better manage their conditions.
Sounds good, right? Yet, it’s more complicated than it sounds.
Vague Memories and Interoperability Issues Get in the Way
Relying on people’s memory for what medications they take, including specifics about the strength, dose, and other details, is not a recipe for success—especially in an emergency setting when people may be in pain or anxious. To gather this information, clinical staff often need to make numerous calls to local pharmacies, other providers, and family members, then manually enter it into the patient record. This is time-consuming and can introduce keyboard errors.
In addition, gaps are common when medication history is imported from an external data feed. For example, if fill information comes from payers, it may omit prescriptions purchased with cash or a coupon. For patients using small, independent pharmacies (often the case in rural areas), the standard data feed may not include their records.
Language barriers are also frequent issues for computer systems, not just for people. Incoming data often uses different terminology or vocabulary than the receiving electronic health record (EHR) system, which causes information to arrive in a free-text paragraph rather than discrete fields. This means the EHR can’t use the data to trigger critical safety alerts; clinical staff must manually enter what’s missing or incomplete, another time-consuming and error-prone task.
Accurate Medication History Data Can Prevent Adverse Drug Events
If manual entry is inefficient and can introduce errors, and interoperability issues can lead to information gaps, what is the path forward? The solution is to take technology one step further, beyond collecting records to interpreting records for electronic systems and clinicians. This is the domain of machine learning (ML) and artificial intelligence (AI).
ML and AI are generally discussed in terms of “moonshot” projects. Yet they have very practical day-to-day capabilities, such as efficiently and thoroughly interpreting records. In addition to gathering more comprehensive information from multiple sources, new AI can clean the data and fill in missing prescription instructions, known as sigs, which avoids manual entry. With so many inefficiencies to overcome, many healthcare providers are embracing this digital process to augment in-person patient interviews and improve the accuracy of imported data, including:
- Cone Health, a North Carolina-based health system with more than 100 care locations and nearly 2,500 beds, wanted to improve its medication history process to improve safety and efficiency. Pharmacy technicians had access to an external medication history source within their Epic EHR, but information was often missing. As a result, they needed to call pharmacies to gather and confirm a patient’s medication list, then enter it manually.
The health system changed to a solution that provides a more complete source of medication history and properly translates the incoming records. Today, they receive data for 93% of patients and 98% of high-risk patients over age 65. AI translates that data and populates it into the appropriate fields to make it available in the clinical workflow, so technicians can spend less time manually gathering and entering this information.
- Michigan-based Covenant HealthCare turned to an AI-powered solution to gather comprehensive medication history data from multiple sources (including local and independent pharmacies). The AI cleans the data, translates sigs into preferred language, and imports it into discrete fields in its Epic EHR. Covenant reduced medication errors by 89% and improved productivity by 33%, resulting in a $309,000 savings in labor costs in the ED pharmacy, plus $6.7 million in annual savings from avoided errors.
Prescription Fill Insights Help Improve Adherence
Just as data gives physicians an understanding of patients’ medication history before prescribing, it can also shine a light on patients’ adherence after they leave the doctor’s office or are discharged from the hospital.
As the former Surgeon General C. Everett Koop said, “Drugs don’t work in patients who don’t take them.” That was 37 years ago, and the problem of medication non-adherence is still widespread today. Up to 25% of new prescriptions are never filled; medications are not taken as prescribed 50% of the time. The costs to patients and the healthcare system are considerable, with 125,000 avoidable deaths each year and an annual cost of $290 billion attributed to non-adherence.
When clinicians have a complete view of prescription fill history—including adherence scores and trend maps sortable by drug class—they can ask questions to discover what’s preventing patients from taking medications as prescribed. Common reasons include cost, concerns about side effects, lack of understanding about a drug’s importance, or simple forgetfulness. These are all challenges that clinicians can help troubleshoot.
- Sibley Memorial Hospital in Washington, D.C., has ED pharmacists receive data around missed prescription fills, which they review with patients to facilitate constructive conversations and create plans to keep patients on their medications.
- South Shore Health in Massachusetts focuses on medication adherence for high-risk patients. Visual timelines give clinicians a quick view of adherence gaps, remaining refills, and the ability to sort by drug class to identify which patients could be helped by clinical interventions.
Up-to-date prescription-fill data can also shed light on potential opioid abuse. A visual interface can provide a clear view of patients crossing state lines for opioid prescriptions from multiple providers and those with several medications that represent unsafe daily morphine milligram equivalent (MME) levels.
Data Scaled to Specific Populations Identifies High-Risk Patients
Prescriptions are often critical to preventing complications and achieving positive health outcomes, especially for some chronic health conditions. When scaled to populations of patients, medication history and adherence data can help care managers identify those at risk of adverse outcomes and intervene early to get patients back on track.
- Stillwater Medical Center in Oklahoma is working to improve medication adherence and avoid unnecessary readmissions for its patients with diabetes. Data was key to the project’s success: with insight into medication use following discharge, care managers identified high-risk patients, allowing them to conduct clinical interventions and discussions with those who weren’t adhering to prescribed therapies.
- The University of Maryland School of Pharmacy adopted medication history tools to power a population health improvement initiative. As a result, the organization drove first-fill rate increases of 10% for congestive heart failure patients and 15% for patients with chronic obstructive pulmonary disorder.
Better Medication History Data Leads to Better Patient Outcomes
Building a comprehensive medication history for every patient is essential, but it requires ready access to complete and accurate data. It’s also vital that this data be available within established clinical workflows to alleviate the burden of numerous system logins and multiple mouse clicks. New technologies powered by AI and ML are doing just that by giving clinicians access to more complete medication history data. To reduce the time and effort spent gathering and entering this information and establish a solid foundation for effective population health programs, it’s time to scrap our reliance on patients’ scribbled notes in favor of accurate digital records.
About the Author
Colin Banas, M.D., M.H.A., is Chief Medical Officer at DrFirst and former Chief Medical Information Officer for Virginia Commonwealth University Health System.
Leave a Reply