Compiled by Amanda Griffith, Contributing Writer
EHRs provide a wealth of data, but the time it takes to make sense of that information can be overwhelming. Massachusetts General Hospital implemented a solution to efficiently mine unstructured data for clinical decision support.
As doctors and nurses struggle with legacy EHR systems that are big and clunky, their health systems’ leaders are increasingly searching for tools that gather unstructured data and give it some substance to try to add value and usability. The mission: Find a way to capitalize on reams of data coming from all angles — EHRs, HIEs (health information exchanges), and even home-based platforms and devices — and make it meaningful. Through solutions that help piece together a complete portrait of the patient to enable clinicians to analyze data stored in a digital health record, it’s possible to enhance patient care and improve outcomes.
Garry Choy, assistant medical information officer at Massachusetts General Hospital, shares this mission. He has been mining unstructured data for clinical decision support (CDS) for quite some time with QPID (Queriable Patient Inference Dossier) Health, a solution that helps reduce the time and effort required to prepare quality reports and registry forms. Using this platform, Mass General is extracting useful patterns of EHR data from clinical data repositories.
Q: How do you use technology to extract information from the EHR to make it usable?
A: When you think about it, an EHR is essentially a database, an electronic version of the traditional paper file. The difference today is that QPID gives us the ability to search for data rapidly, like a Google for physicians, in a way that’s akin to scanning through patient charts, only faster.
On the most basic level, we use the technology for CDS to help improve patient outcomes and extract meaningful information from our EHR. I can pull out cohorts of patients, such as those treated in our cardiology department, and use that information for metrics and quality reporting. It’s also allowed us to better analyze medical records to further develop meaningful CDS tools such as a readmissions risk predictor dashboard so that physicians can better care for inpatients and reduce readmissions risk.
Q: How does mined data improve clinical decision making?
A: Simply having access to information is really important for decision making. If it’s hard to find, though, it’s possible you could miss it. Or, it could take so long to search for the data you need that you’re wasting time that would be better suited for patient care.
This QPID technology platform represents a huge step forward. Being able to quickly search unstructured data, realistically, and then synthesize that information from the EHR — lab tests, physical findings, assessments and plans — means the treating physician or the next clinician caring for your patient can make the best possible decision. It’s vital to have a safety net so you know you are capturing all the information available.
Other search engines exist, but I haven’t found another that searches as if it understands what the physician is thinking. If I type in “arrhythmia,” this technology thinks like a physician and will bring back search results for “atrial fibrillation” as well — it knows the concept and meaning behind the keywords.
Q: What problems were Mass General looking to solve by implementing this particular solution?
A: The main reason is historic. The technology was actually developed by some of our own physicians. QPID was created as a health intelligence platform incorporating an EHR search engine, a scalable library of HIPAA-compliant search queries, and a programming system for application and query development. It was developed to enable better analytics to be performed on the data stored within the EHR. In essence, QPID is designed to turn static “Big Data” into “smart data.”
The tool became so useful it spread organically through our organization. What began as a CDS tool for research and quality improvement metrics has spread to gastroenterology, anesthesiology, cardiology, and across the entire hospital network. Everyone sees the value in taking advantage of the data stored in the EHR and using it for clinical and administrative purposes.
Q: What have the results of these efforts been to date?
A: With the ability to look at behavior in aggregate, physicians at Mass General can now use the platform for anything medication-related. For instance, clinicians can quickly check what medications their patients are taking. The most commonly searched category terms by internal medicine and primary care physicians are “medication name” and “medication dosage.” Not only are medicine reconciliations very important because of The Joint Commission recommendations, but also it’s just the right thing to do, to track which medications are taken and then reconcile and check doses.
We also use the technology to screen patients scheduled for MRIs. MRI exams are performed so frequently that they have become a high-volume, high-risk event. Every MRI ordered prompts an automatic search of patient records to search for mention of a pacemaker or any other metal device in one’s chest, something that would be extremely harmful if placed in an MRI. Before, we performed this search through a manual check; now we have an added layer of patient safety.
Q: What would you say to other providers considering implementing this technology?
A: Have a good communications plan in place, one where physicians know not only that the technology exists but also what its capabilities are. Once people start using it, it becomes the best way to derive the most value from an EHR. Who wouldn’t want to practice medicine with a tool that makes it more enjoyable and provides faster access to data? It makes life as a clinician, in the year of IT, less burdensome. In a world where many physicians are frustrated with their current EHR technology, it’s great to have an IT solution that looks for, finds, and allows you to extract necessary data.
Also, make sure you work with your organization’s CMIO and CIO offices to ensure coordination with IT is done properly and that data governance, data integration, and data storage are seamless. This ensures there are no unnecessary stresses to the system.
Q: How do you plan on using this technology in the future?
A: One thing I’d like is to make the technology better known to even more physicians and other staff members across the organization. For example, perhaps we could develop a structure that allows researchers to use QPID even more extensively. The company has consistently added features and the ability to perform data analytics in smarter ways that interpret unstructured data, such as physicians’ notes. With the ability to analyze large cohorts of data, it’s a key piece of clinical care and helps us operate better under the accountable care model. Population health matters, so being able to analyze not just one, but multiple, patients is an attractive feature and a great use case. Furthermore, we are ensuring that QPID works as a fully integrated solution with our EPIC EMR rollout across our multiple member hospitals within the Partners Healthcare enterprise.
Many organizations today look at enterprise data warehouses or install the latest integrated EHR. I’d argue it’s important to also research data analytics or data intelligence platforms that make the most use of data already stored in what are essentially somewhat static silos. Ours is one way to go beyond just storing the data by actually meeting the demands of clinicians who want to use it day to day.