Surgery centers could mark 2020 as the year to focus on quality improvement projects to prevent medication errors. Accreditation agencies have published information that can help.
The Accreditation Association for Ambulatory Health Care (AAAHC) recently released a benchmarking study about medication errors and medication reconciliation.1 AAAHC notes that medication errors cost $21 billion in annual healthcare costs. There are 3.5 million physician office visits and 1 million ED visits that result from medication errors.
The AAAHC report authors noted many ambulatory healthcare organizations struggle with documentation as well as updating and verifying medication records, all of which contributes to patient complications and overall costs.
The goal is for surgery centers to document patients’ allergies and sensitivities in medical charts consistently. The study revealed that 86% of charts included this documentation, according to Belle Lerner, MA, assistant director of the AAAHC Institute for Quality Improvement. About 65% of charts included documentation of patients taking the medication as prescribed, Lerner adds.
“The organizations voluntarily participate and submit 15 to 25 cases, and those are chart-reviewed,” she explains. “Another key finding is in regard to changes of medications prior to a procedure. Eighty-one percent documented the number of days in which the patient was instructed to stop the medication, prior to the procedure, and 90% indicated the patient was instructed to resume any stopped medication post-procedure.”
When a patient does not resume a critical medication, such as an antiplatelet or anticoagulant, repercussions can be life-threatening. The study revealed ambulatory facilities performed well with single-source medication documentation, with 99% achieving that goal, reports Naomi Kuznets, PhD, vice president and senior director of AAAHC Institute for Quality Improvement. “We were very happy to see levels of 99% for single-source documentation,” Kuznets says. “The thing we were more concerned about was documentation of the number of days to stop medication prior to the procedure.”
As surgery centers focus on preventing medication errors, one technique is to use a machine learning system that sends alerts when potential mistakes are identified. A recent study revealed that an automated system (MedAware) for identifying prescription errors can generate many clinically valid alerts that otherwise might be missed. More than 80% of the alerts were valid, while nearly 63% were considered of medium or high clinical value.2
“In this study, we assessed a novel solution to identifying prescription errors, mainly — but not only — in hospital settings,” says Ronen Rozenblum, PhD, MPH, director of the unit for innovative healthcare practice & technology, Brigham and Women’s Hospital in Boston, and assistant professor of medicine at Harvard. “Based on data, I believe this addresses a huge need in the area of prescription safety. We’re talking about $20 billion annually in problems with prescription errors.”
Current solutions to the problem mostly focus on rule-based, decision-support systems, working based on the indication and contraindication of medications. “Looking at the big picture, we’re entering an era where we’ll see more healthcare organizations that use data and analytic tools to improve quality of care,” he says. “We’ve heard of this big solution for a while, but we’re reaching a tipping point of why using big data tools can do this.”
The machine learning system works with big data analytics, identifying three outliers:
- Clinical outliers: Patients receive the wrong medication;
- Time-dependent outliers: Patients are prescribed the correct medication for a specific point in time, but later when the patient’s clinical situation changes the medication no longer is what is best for the patient;
- Dose outliers: The medication is correct, but the dosage is wrong for this particular patient.
Investigators evaluated more than 700,000 patients who logged at least one outpatient visit with a provider affiliated with Brigham and Women’s Hospital or Massachusetts General Hospital within a two-year period between 2009 and 2013.2
The software system developed its algorithms through random selection and analysis of half the total patient population. The system generated thousands of alerts, including all three outlier categories. Most alerts were related to time-dependent outliers. The authors reviewed 300 alerts in depth. The machine learning system proved more accurate than the regular electronic alert system.
“Our study was retrospective, and we assessed whether our own homegrown system had flagged the alerts,” Rozenblum explains. “We found that 68% of the alerts had not been generated by the existing system.”
Rozenblum and colleagues also examined the economic cost of the healthcare system failing to prevent medication errors. They found there was a potential cost savings of $60.67 per alert, mainly because of the prevention of adverse drug events.
“We had a health economist on our team assess the potential cost savings, and we found it saved $1.3 million in healthcare costs just for this cohort,” Rozenblum says. “Until now, we didn’t have significant evidence-based data that show the clinical data and economic data for machine learning. Based on our study, it’s a promising solution for healthcare that could save lives and also save money.”
Surgery centers already are trying to prevent medication errors, Kuznets observes, but not all organizations ensure patients receive sufficient instructions about stopping and starting medications before and after procedures.
“The chances are there will be some patients that end up in the hospital because they started their medication too early and bled,” Kuznets says. “Or, they failed to start their medication on time to counter issues related to blood clotting, and ended up hospitalized.”
Sometimes, surveyors are asked what to do about situations in which a physician’s office has patient information about allergies, but that information is not included in the surgery center’s records, Kuznets reports.
The answer is for surgery centers to check medications and allergies on the day of surgery instead of relying on the pre-op visit information. This is good practice regardless of whether the physician’s information was shared with the surgery center, Kuznets offers.
“Things can change between when the pre-op evaluation is done and you get it within the center,” Kuznets explains. “You should find out whether the patient discontinued the medication they were supposed to in a certain period.”
AAAHC is focusing on medication reconciliation because this was identified as a high deficiency area in a quality roadmap report several years ago, Lerner notes.
“There are organizations that struggle with appropriate medication reconciliation. Although some do a pretty good job, we would like to continue monitoring it,” Lerner adds. “We have several tools available for our organizations that struggle with medication reconciliation.
AAAHC offers a medication reconciliation form as well as a patient safety toolkit and an e-learning module that helps organizations institute best practices for medication reconciliation. (Editor’s Note: A blank medication reconciliation form can be viewed at: http://bit.ly/2unEBJY. The toolkit and module are available for purchase at: http://bit.ly/2tGofw3.)
AAAHC’s medication information includes best practices shared from the Agency for Healthcare Research and Quality (AHRQ). For instance, AHRQ provides a patient safety primer on medication reconciliation, which was updated in September 2019. The primer provides evidence-based data about interventions to prevent medication errors and lists challenges in achieving safety improvements via medication reconciliation. (Editor’s Note: Much more information is available online at: http://bit.ly/2G4BZ6i.)
Health systems likely will be the early adopters of big data analytics, but surgery centers and other providers may follow their lead soon, Rozenblum predicts. “The benefits outweigh the obstacles,” he says. “It can work if you’re really solving a big problem that’s a real threat to patients’ lives, and you’re also saving money for an organization.”
Market forces and a change in culture could hasten this adoption. Just as healthcare systems have evolved over the past decade by incorporating electronic medical record (EMR) systems, they likely will begin using machine learning systems to prevent medication errors and other adverse events.
“Fifteen years ago, not a lot had EMRs, and now a majority of health systems have EMRs,” Rozenblum notes. “Now, we’re moving from this generation of a rule-based system to a more sophisticated system that uses big data and predictive analytics. It’s the next big jump, and we’re one of the first to have evidence around that.”
- Accreditation Association for Ambulatory Health Care. AAAHC publishes medication reconciliation benchmarking study findings. Published Dec. 12, 2019. Available at: http://bit.ly/2TMNA28. Accessed Jan. 22, 2020.
- Rozenblum R, Rodriguez-Monguio R, Volk LA, et al. Using a machine learning system to identify and prevent medication prescribing errors: A clinical and cost analysis evaluation. Jt Comm J Qual Patient Saf 2020;46:3-10.