Victoria Valencia, MPH, Assistant Director of Healthcare Value
Christopher Moriates, MD, Assistant Dean of Healthcare Value
Dell Medical School at The University of Texas at Austin
With tele-machines beeping, robots rolling by and so many different people rotating in and out of rooms, the hospital environment can be chaotic. Similarly, the data environment of many of our electronic health records (EHRs) can be quite unruly. EHR systems are not always designed for clinicians to input data in a way that can be researched and used in quality improvement.
Just as hospitalists become comfortable working amongst the whir of the wards, data specialists learn how to handle the thicket of data libraries. Report writers and clinicians who request reports often have little to no experience with the workflow and context of the other. It’s no wonder why it often feels like we are speaking different languages and why the reports hospitalists get back don’t always contain the information they had in mind.
There are things we can do to overcome this gap:
- Report writers can join clinicians on the wards. Simple things hospitalists or nurses may take for granted can be the key to obtaining the correct data. For example, a request for a report of all blood labs completed on patients on the hospitalist service seems straightforward. However, when we requested this, the numbers on the reports were coming up short. Only with resident physicians working together directly with the data specialists did we realize that when some of the orders were placed, the specimen types were left blank, and thus not captured. The next time you are requesting a report, why not invite the report writer to come meet you on the hospital floor and watch how you enter your data? This way, a report writer or analyst you work with can get closer to the data you need without so much back and forth.
- Clinicians can strive to be more data savvy. Doctors are responsible for knowing a continuously expanding depth of medical knowledge for clinical care, and certainly not all of us – or even many of us – should become data scientists, too. However, simply spending some time with a data specialist and trying to understand some of their basic methods and language can help tremendously with translation. Our report writers do not need to know the correct antibiotics to choose for hospital- acquired pneumonia, but it would help if they at least recognize that there are different classifications for pneumonia. Physicians do not need to understand how to write an algorithm for filtering data, but shouldn’t they all know a little something about data filters and how they can affect the results (just like a filter for “specimen=blood” could have led to the inaccurate results in the example above)? We can also embed concepts like data analysis and informatics directly into medical school curricula to prepare our future clinicians for a more data-driven clinical environment. Some medical schools are developing these type of initiatives through the American Medical Association (AMA) Accelerating Change in Medical Education program.
- Create regular opportunities for data folks and clinicians to meet and have conversations. When it comes to changing behaviors, the data are not nearly as important as conversations about the data. One of our colleagues from UCSF presented last year on the plenary stage at Hospital Medicine 2016 about how data dashboards were shared with teaching medicine teams during “STAT rounds” – a once-weekly, 15-minute pow-wow immediately following a regular lunchtime conference. Residents and attendings review their team-based data with a colleague and/or data analyst who was involved in building and maintaining the dashboard. The physician teams ask questions about the data and talk through issues they were having with meeting any specific metrics. For data specialists, attending STAT rounds was a great experience. What Victoria, Dell Medical School’s Assistant Director of Healthcare Value, loved it about it was that she, a self-described data nerd and non-clinician, was immersing herself in what it was like to be on the input side of EHR data. She had come into the job thinking, “Why are these data so messy? Why can’t these clinicians enter these data better?” It turns out it is because the system is not really designed to make it easy, and this insight changed how everyone appreciated each other and worked together. For example, resident teams were not sure why they were not scoring better on our metric of a “high-quality after visit summary,” and the data/administration side couldn’t seem to understand why physicians just couldn’t do a better job with providing follow-up information in the chart. At STAT rounds, we discovered that the algorithm only checked one portion of the discharge note for this information and that sometimes residents were placing it in the “post-discharge instructions” section, failing to get credit for something they did in fact do.
Doctors and data specialists nearly always come from different backgrounds with different skillsets, experiences and insights. When we bridge these gaps and understand each other’s worlds, it is amazing how suddenly the data can start to make sense.
Victoria Valencia, MPH has a background in biology and public health and has worked in clinical research and quality improvement for over five years. She is a data analyst and programmer and uses R to analyze and visualize complex data from clinical research databases and electronic health records to help answer questions around quality, value and appropriateness of care provided to patients. She has co-authored several articles that investigate ways to improve the value of care patients receive while in the hospital.