Farzad Mostashari , in a meeting with the National Quality Forum in late April, says, “In 2016, it’s going to be rare to find a doctor without EHRs.” He went on to say we need to “… find people who have done it, who understand deeply what it takes and not just the challenges but how to overcome those challenges.” He finished by reminding us to “put patients at the center.” Before we can populate the EHR, we must capture this information in a form that is usable in the EMR.
This is fantastic news for all of us healthcare analytics fans. The proliferation of EMRs signals the industry movement toward installing tools that can capture discrete data that is the cornerstone of a good business intelligence strategy. Putting tools to capture this key data in everyone’s hands is a mandatory first step towards using this data to improve outcomes, save costs, and put the patients at the center.
EMRs are a giant step from paper, but they are only useful if the data captured can be reused. I urge everyone in the industry to take it just a tiny bit further. While you are training yourself and your teams to use these EMR systems, spend a little time on standardization and vernacular. Train your staff to use specific terms that are searchable and uniform. The benefits of this are hundredfold.
This is important for research. To keep them flexible, EMRs enable a lot of free-form text entries. This is a double edged sword. This flexibility enables the healthcare team to capture notes and details about patient-specific conditions. Unfortunately, if these notes are not consistent, this flexibility also encourages the entry of terms that are nearly impossible to search and group.
Recently, I did analysis at a large ED on the east coast. I was reviewing the Complaint field and I found entries including SOB, SB, hard to breathe, shortness of breath, short breath, breathing difficult, and Lungs hurt. All of these mean the patient is complaining of shortness of breath. Unfortunately, when a database engine encounters each of these terms, it assumes they are all different. In a scenario where we want to report on the number and types ED visits, it would be preferable to group all these terms under a single condition and report seven instances of SOB as opposed to a single instance using each term in the system.
There are some translation tools that use ICD, SNOMED, and UMLS to try to standardize language. These are useful and get the data a little bit more standardized. These tools are not nearly as useful as training and oversight. Just a few minutes spent during the roll-out of the EMR and a couple instances of reviewing the data and some remedial training will show a far larger return on the time invested. This approach also appeals to my motto “don’t use technology to fix a training problem.”
