At Inteliquet, Jim Reddy works with organizations as a consultant for workflow and data analysis. He has been working with hospitals and physician groups for a number of medical specialties for the past 28 years, focusing on clinical operations, data quality and metrics. He has overseen multiple EMR installations, organized countless clinical team and interdepartmental workflow reviews, and developed reporting methods for projects such as the Oncology Care Model.
“Every two years, the amount of written data in the world since the beginning of human history doubles…”
Since I spoke those words at a Community Oncology Alliance presentation in 2013, the amount of written data in the world has risen by 300 percent.
In healthcare, the ability to ingest and understand this ever-growing mountain of information is critical to improve patient outcomes, advance science, institute reforms, manage costs, and set effective policy. Until now, the only “Big Data” available to clinicians, researchers, insurance providers, and policy makers has been billing and claims data. However, in recent years the widespread use of Electronic Medical Record (EMR) systems have begun to give us the amounts of clinical patient data needed for large-scale data mining. This data can be stored within the system in varying ways – each with its own mining challenges.
EMR data can be, broadly speaking, broken down into two buckets: structured and unstructured.
- Structured data is information entered into pre-defined fields and includes phrases in pick-lists and drop-down boxes, numeric fields for entering patient vitals, and most electronically captured lab results.
- Unstructured data, which is roughly 90 percent of all data within EMR systems today, includes physician progress notes, pathology and radiology imaging reports, or any scanned documents.
While structured data is far easier to ingest, unstructured data holds the most potential to improve clinical outcomes. But it requires either human curation (a person reading the information and manually extracting the needed data) or computer curation, often referred to as either Artificial Intelligence (AI) or Natural Language Processing (NLP).
Sheer volumes preclude human curation on the scale needed for the vast number of documents that healthcare organizations create on a daily basis across the nation. Using computer technology to mine through unstructured data remains a significant undertaking. At Inteliquet we employ the most advanced NLP processes available and also use other techniques such as Optical Character Recognition (OCR) to parse and extract as much clinical data as possible from progress notes, imaging reports, molecular lab reports, etc., and are constantly improving upon our methods to pull everything we can from these sources. We are then able to combine this information with a client’s structured EMR data elements and provide them with a more complete database that allows them to quickly and accurately query for research purposes and patient trends. But this technology is still young and the amount of data gleaned from unstructured data sources is still relatively small compared to the amount of unstructured data elements available.
Though AI continues to rapidly advance, at this point in time the healthcare industry is still dependent primarily upon structured data elements for large-scale extraction and analysis for things like improving patient outcomes and payment reform, which is edging towards a quality-based payment system which will require ongoing data submissions. What this means for healthcare organizations is that the amount you get paid for caring for your patients will depend in great part upon the completeness and quality of the data you report on.
No one has felt the changes brought about by EMRs and the increasing demand to capture structured clinical data more than physicians. EMR adoption has dramatically changed how physicians interact with their patients over the past decade. According to a study published in The Annals of Family Medicine, physicians now spend twice as much time interacting with computers as they do face-to-face with their patients. This has been cited as one of the chief contributors to the current crisis in physician burnout (Mayo Clinic Proceedings): 48% of all physicians polled show at least one sign of burnout. This is actually down from a high of 54% in 2014, but still at twice the rate of American workers in general. Nurses and other staff are just as overloaded in their own data-entry responsibilities. Consequently, the challenge for most healthcare managers is to figure out how to make the entire clinical team efficient at EMR documentation.
First Things First: Identify What Needs Documenting in Structured Fields
The first step is to identify the critical data elements to be captured as structured data in your EMR. Clinicians know the data elements most critical to the care of their patients, but make sure to take the time to list out all structured data elements that your EMR is capable of accepting. This will give you a starting point for discussion as you begin to prioritize which structured data elements are most important in the documentation process.
Include the Entire Team in the Process.
If you do it right, this will evolve into a weekly team huddle for ongoing coordination of work. The time needed for this get-together will quickly reduce as your team becomes more proficient at it.
Involving all team members is critical because this is the best way to identify gaps in documentation, whether the right member is entering the right information, etc. We all know how precious time is in today’s medical practice, but taking less than one hour a week to review and hone workflow will make the other 49 hours much more productive.
Everyone Works to the Top of Their License.
I first heard this term used by a friend of mine during a workflow analysis meeting and it means that the key tasks that make up the majority of each team member’s day should be ones that only someone of their level of training can complete. Start with the team member with the highest level of training and responsibility (in this case the Physician) and list all the structured data elements that only that person can determine. Prioritize those elements that are considered most critical. Then move on to the team member who is next in line, and so on.
For example, going through this process with an Oncology team may result in the following list of the most important data elements each member is responsible for entering into the EMR:
- Physician: Diagnosis, stage, ECOG status, treatment regimen, treatment intent, disease status
- RN: Chemo administration, triage data, supportive diagnoses, chemo and pain prescription refills
- Medical Assistant: Vitals, medication list, patient-reported symptoms, smoking status, injections
Having a physician wrestle with entering smoking status – when it can be done instead by the Medical Assistant – is an inefficient use of the physician’s time. By shifting some of these responsibilities among the team, you might be surprised by how much time this exercise can create to allow each person to “work to the top of their license.”
Evaluate Your Current EMR Data for Gaps.
Understanding the state of your current EMR data will not only give you a base to gauge improvement, it will highlight those areas of your workflow that are most in need of revision. For instance, if in the above example you find that staging rates are very low among your physicians, you can focus on workflow changes that allow your physicians to get this data recorded more regularly, whether it’s through process improvement that frees up the time to do so, or finding quicker ways for the physician to add this data into the record through their note template, for example.
Evaluating your EMR data can be challenging without a data analyst who has access to the underlying tables of your EMR and can query the data in any number of ways, otherwise you are limited to using the canned reports that come with your EMR. A major strength of Inteliquet’s software is that it puts powerful querying capabilities at the fingertips of the end user, allowing you to understand your data much more fully than what the EMR is capable of doing on its own.
Reevaluate the Data Capture Rate for Critical Data Elements at Regular Intervals.
Do this on a quarterly basis at a minimum. The more frequently it is reevaluated, the easier the evaluation process becomes.
Use Your EMR To Its Full Functionality.
Many organizations fail to review the use of their EMR after the initial implementation and go-live. The data demands of today are not what they were five years or even one year ago. Establish templates that help providers and other clinicians get notes done as quickly as possible.
At my last practice, our provider note templates underwent constant revision as we discovered quicker ways to help physicians efficiently complete their notes. For instance, we developed a method for Medical Assistants to quickly enter patient-reported symptoms into the EMR at each visit. This data could be pulled into the physician’s note and populate a patient’s review of systems, if the physician so desired, significantly reducing the provider’s data-entry requirements for their progress notes. Some EMRs have also developed patient portals that allow patients to enter in select information on their own, which could cut down on that which is typically entered by the Medical Assistant.
Review, Revise, Repeat.
Workflow improvement is a lifelong process.
Medical providers and their staffs already face enormous challenges in documentation. As the needs and requirements for clinical data increase, it becomes more important that your clinical team functions as a well-oiled documentation machine to minimize the impact that time-to-document can have on patient care, while ensuring that your data is prepared for future payment reform and as a tool for improving patient outcomes and healthcare policy. Ongoing review of your data completeness and workflow processes is critical to achieving this goal.