Despite $40 billion in federal spending on EHRs during the last decade, physicians struggle to use data to improve outcomes for patients in the real world.
While Seattle Children’s Hospital was able to use its EHR to access clinical history for an individual patient, staff had no way to use the EHR data to assess clinical outcomes across patients for clinical questions without months of data mining and analysis.
“Take a simple example from our Bellevue Clinic & Surgery Center,” said Dr. Greg Latham, attending anesthesiologist at Seattle Children’s Hospital, and coauthor of the study, titled “In Pursuit of an Opioid-Free Pediatric Ambulatory Surgery Center: A Quality Improvement Initiative,” recently published in the journal Anesthesia & Analgesia. “A few years ago, we changed our anesthesia protocol for a common surgery by adding a pre-operative pain medication.
“Without the ability to look across our patients, we were unable to quantify the effectiveness of our change in protocol – i.e., did patients experience less pain after surgery with the new medication? Ultimately, answering this simple-sounding question entailed a nine-month process of extracting and analyzing data from our EHR – requiring extensive investment of scarce physician, analyst and bio-statistician resources.”
Improvement in medicine historically has moved very slowly, with the time lag from research discoveries to implementation (bench to bedside) estimated at 17 years. More recently, “improvement science” has become a valid methodology to improve processes. It uses rapid-change ideas, such as the Plan-Do-Study-Act cycle, to inform improvements.
“While simple in theory, improvement cycles require data harvesting and statistical analyses,” Latham explained. “Since it often takes nine to 12 months for data and analyses requests to be fulfilled, an improvement cycle would easily take a few years to complete. We wanted to bring practical improvements in care to our patients much faster, so we needed a new way to leverage our data to improve and manage care across patients.”
Health IT vendor MDmetrix offered Seattle Children’s Hospital the opportunity to get actionable answers to the hospital’s outcomes questions in just minutes, instead of months or years, Latham said. MDmetrix’s AI-based Mission Control platform also enabled the hospital to interactively monitor and evaluate clinical metrics, so that hospital teams could continuously assess and adapt treatment protocols and workflows, he added.
“The goal was to be able to ask and answer clinical-performance questions – self-serve, on the fly, without needing to be a technical super-user,” he said. “We needed for our physicians to be able to quickly ask a question such as, ‘Which of our protocols for Procedure X is best serving our patients on Metric Y?’ without having to call in scarce IT or other resources. We needed technology that would enable us to visualize, assess and manage clinical performance of our protocols and our team across patients.”
In practical terms, there are no straight lines in medicine, Latham said. So, there always is a background level of data “noise” due to natural variation when one looks at clinical outcomes, he added. The hospital knew it needed an artificial intelligence-based system that could deliver data in an actionable form to clinicians by parsing out “signals” in the data so staff could distinguish real improvement from mere change, Latham explained.
MEETING THE CHALLENGE
The hospital used the Mission Control platform to provide daily monitoring and assessment of anesthesia protocols across key metrics, including pain scores, administration of pain-rescue medications, post-operative nausea and vomiting, PACU length of stay, and readmissions. The platform’s AI technology made it easy for staff to distinguish actual data signals from noise, Latham said.
“For example, the AI gave us confidence that enough time had elapsed for us to be able to assess the value of each protocol change given our volume of surgeries so that we could ensure meaningful results for our patients,” he explained. “With the ability to ask our own clinical questions on the fly, we could quickly and easily interrogate our data whenever we wanted to ask more nuanced questions or to explore signals in our data.”
“The AI gave us confidence that enough time had elapsed for us to be able to assess the value of each protocol change, given our volume of surgeries, so that we could ensure meaningful results for our patients.”
Dr. Greg Latham, Seattle Children’s Hospital
Using the platform, staff was able to accelerate improvement work at a pace far beyond traditional methodologies, he added. Specifically, staff was able to quickly assess protocols for tonsillectomies, and then staff scaled up its improvement project across a wide range of outpatient procedures.
“In essence, our clinicians were empowered to improve protocols in weeks – a review process that previously would have spanned years,” Latham said. “By enabling us to continuously monitor and evaluate outcomes across patients, the platform has allowed us to implement a truly adaptive clinical-management system that fully leverages our real-world data.”
In doing this, there was a critical ingredient that goes beyond what’s captured in staff’s academic papers. To accomplish this kind of transformative change, staff needed real engagement and buy-in across the clinical team. The MDmetrix system enabled every one of the hospital’s clinicians to have direct access to data.
“Because each of our clinicians could interrogate the data in their own way – making sure for themselves that our changes were real ‘improvements’ for our patients, we were able to effortlessly enlist a broad-based team in effectively transforming care,” Latham stated. “Today, all of the anesthesiologists, nurse anesthetists and surgeons at our surgery center are using MDmetrix – and we look to all of these stakeholders to bring improvements to our clinical system.”
Seattle Children’s Hospital was able to reduce the use of opioids. With 5% of adolescents who undergo surgery becoming persistent opioid users, staff felt strongly that they needed to find ways to reduce patients’ opioid exposure.
“Between October 2018 and March 2019, we successfully implemented an opioid-free anesthesia protocol for outpatient tonsillectomy surgery,” Latham noted. “Post-operative nausea and vomiting, the most common side effect of anesthesia, dropped from 3.5% to 0.5%. Most important, our patients who had opioid-free anesthesia were comfortable – same pain scores – and there was no difference in the 30-day return-to-surgery rate for bleeding complications.”
It cannot be overemphasized that prior to this study the thought of an opioid-free anesthetic for tonsillectomy was not even a consideration, he added.
“We were able to implement evidence-based medicine literature that described safe and efficacious use of dexmedetomidine and ketorolac (two non-opioid analgesic medications) published in 2010 and 2014,” Latham said. “MDmetrix allowed us to reduce the time to complete a clinical PDSA improvement cycle to just 12 weeks – a huge improvement when you realize that previous protocol iterations were measured in years. The platform enabled us to leverage the real-world data routinely collected in the EHR and use that data to understand the comparative effectiveness of our current and historical protocols.”
Building on this success, staff used the AI platform to drive protocol improvements across all outpatient surgeries. With the new improvement cycles, staff was able to reduce intraoperative opioid administration from 84% to 8%. Post-operative morphine administration dropped from 11% to 6%. To date, more than 6,000 patients have successfully had outpatient surgery with opioid-free anesthesia.
“By offering opioid-free anesthesia, we have reduced the risk profile for our patients, and we have been able to extend the type of surgeries that can be safely performed in an outpatient setting,” Latham explained. “Finally, because surgery is a critical gateway to opioid misuse, we believe that this shift to opioid-free surgery is a tremendous breakthrough for patient care, benefiting our patients long after they leave the hospital.”
In addition to the clinical results, there were operational results: for example, increasing capacity and reducing cost.
“For the opioid reduction initiative, it is worth noting that although reducing cost was not the primary objective, the surgery center managed to achieve an 85% reduction in their analgesic medication costs,” Latham said. “In addition, patient and parent satisfaction scores increased.”
While staff was working on improving anesthesia protocols, they also used MDmetrix to take a careful look at resource utilization. Fortunately, the same visualizations used to evaluate treatment processes also could be used to assess workflow processes.
“Using the platform to look across our outpatient surgeries, we were able to surface inefficiencies in our case workflows,” Latham explained. “By applying the same improvement philosophy to our surgery workflows that we used to improve our anesthesia protocols, we were able to leverage insights to add two to three cases per day to our ENT OR without any increase in staffing hours. This improvement meant that we could serve more patients, improving access to care in our community, while also bringing in more revenue to fund our mission.”
ADVICE FOR OTHERS
Hospitals have heavily invested in EHR technology to capture digitized patient data. They all should be leveraging this digitized data to improve patient outcomes, patient safety and operational efficiencies, Latham advised.
“Physicians need to be able to ask nuanced data questions themselves, without waiting months for manually intensive data mining that isn’t scalable,” he concluded. “Healthcare institutions need to invest not just in creating and storing data, but also in putting tools directly into their clinicians’ hands that empower them to leverage that data to improve patient care and optimize critical resources.”