Kayaks, Perspective, and Decision Support for Meaningful Use | [node:field-byline] | Healthcare Blogs Skip to content Skip to navigation

Kayaks, Perspective, and Decision Support for Meaningful Use

July 22, 2011
by Joe Bormel
| Reprints
Clinical observations or data we collect over time in our EHRs often conflict with other observations

Take a look at the first photo. You’ll see the Washington Monument clearly to the left of the Lincoln Memorial. The picture below, taken moments later shows, the Washington Monument squarely behind the Lincoln Memorial.

If you had no other knowledge, you might conclude that because the perspective of each photo is almost the same, then one of these structures was moving. Of course, that conclusion would be wrong.



I took these photos from a Kayak, while having a great time paddling around on the Potomac River.

Although the phenomena of the moving monuments is simply the parallax we all learned in high school, the relevance to Meaningful Use is actually very profound.

The clinical observations or data (photos in our analogy) we collect over time in our EHRs often conflict with other observations. Some are clearly collected at different times and may represent an important trend. At other times, many other times in fact, the observations of two care providers about the same patient at the same time don’t agree. One doctor’s assessment is incurable Stage 4 cancer. A second opinion by another doctor indicates it’s “not a death sentence;” it’s both Stage 3 and highly likely curable. Both doctors file their semantically interoperable consultations into the same EHR.

The inter-relationships of the clinical processes, user workflows and applications, as well as their independence, assures that we will be collecting a lot of conflicting observations, interpretations and recommendations.

I work with clients who are pursuing and meeting Meaningful Use criteria. Within their EHRs, conflicts show up in problem lists, generalized order sets for individual conditions, and in patient summaries. However, when a record is viewed by other providers, the patient or institutions, distinctly different insights and perspectives may be drawn. Since the majority of patients have more than one problem on their list, and often these can be thought of as co-morbidities, there’s a need to manage this implied complexity.

For example, if a patient presents with both palpitations and a stroke, or heart failure and lung disease, there’s a high likelihood that the order sets for two distinct problems are incompatible with each other. This often means that sorting out and resolving these disconnects requires considering each patient’s distinct context.

In addition to problems, other recommendations for orders or documentation can be dependent on lab results and physical exam findings. If a patient has, for example, a low platelet count and evidence of internal bleeding, it may be appropriate to hold their blood thinner. And yet, from the perspective of a patient with a history or risk of life threatening clotting, holding their blood thinner may be exactly the wrong thing to do.

As more and more information is collected, both within individual encounters and across a longitudinal view for an individual patient, it will be necessary to sort things out. Where is the Washington Monument relative to the Lincoln Memorial when it’s known that the photos are accurate?

Today, such photos here can be electronically studied and compared to a satellite view showing both from where the photos were taken (shown in red circle below), as well as the actual locations and sizes of the structures on the Washington Mall.

In fact, nothing short of information technology would be capable of reliably and quickly combining all of the images into a single, coherent, useful picture. I think that the era of Meaningful Use will produce the equivalent in healthcare information as the EHR evolves. Developing our technology accordingly is exactly the perspective my team has taken to ensure electronic decision support helps clinicians make the right decisions. It’s a matter of patient safety and quality care.

What do you think?

Joe Bormel, M.D., MPH
CMO & VP, QuadraMed
This post: http://bit.ly/MUKayak
Previous Post: http://bit.ly/ViabilityOfACOs

“The task of the real intellectual
consists of analyzing illusions
in order to discover their causes.”
Arthur Miller


Dr Khan, thank you for your comment. Platforms like yours that focus on immediate problems of co-morbidity and clinical context to determine appropriate and safe order sets have come of age. Like driving without seatbelts, it's quickly becoming clear that information accidents will happen. Spotting them and addressing them before harm comes to patients will intensify the use of CCDs, transcending their basic patient summarization role.

As an aside, the boat houses the I get my Kayaks from defy GPS naviation to get to them, at least according to numerous Yelp reports. The problem is that most GPSs are two dimensional, not three dimensional. One of the boat houses, Jacks, is under two bridges. GPSs understandably get confused and take their users in circles. The second, Thompsons, has an arcane sets of time-dependent one-way streets around it. Again, drivers are routinely taken in circles. Many GPS systems simply consider those roads, or those with weight limits as simply not there.

Similarly, the world of clinical decision support will prove to be anything but 2 or 3 dimensional. Our metaphors will quickly break down.

Thank you for an interesting post and a great perspective. Using the concept of fusion to think about bringing various viewpoints together to create a more useful picture of a patient's health is very novel. Translated into informatics speak we need to look at all of the patient's data including labs, meds, radiology, clinical history, family history etc. to determine the best course of action. As you point out, current systems have the challenge of not being able to use this patient information to remove or at least identify contraindications, duplications and redundancies. Unfortunately, I feel that we missed a great opportunity with MU Stage I. By removing most of the Clinical Decision Support requirements in Stage I we allowed the health system to punt on adding these capabilities even though these capabilities are currently available.

At DiagnosisOne, we have been working on exactly the problem you described for the past decade. We recognized that often times patients' full histories were not being factored into their care. Our experience is that much can be done to help providers bring patient-specific context, pulling in all known patient data, synthesizing that information, and delivering only those alerts and order sets that are the most relevant and most appropriate to the point of care in real-time. With the explosion of electronically available and codified information, it is now critical that providers are aware of the challenges you described.

Thanks again, and I look forward to reading more from your perspective.

You are correct.

The speed and quality issues are real. That said, we had been moving too slowly, prior to ARRA. Electronic semantic interoperability, which the CCD addresses for example were not on a path to mainstream care delivery.

As pointed out in this post, when information is more available than it is today, we will have new challenges. Good challenges. Challenges that come with progress.

Dr. Bormel,
Your "perspective" is very interesting. It gives me pause to think about my hospital and meeting MU criteria, as well as our existing goals to increase quality and patient safety.

Mr. Land's comment concerning, what I think most of us feel is correct, the perceived fact that MU is pushing the envelope for healthcare IT to capture the metrics necessary to qualify not only correct, it's disturbing. Stage 1 MU appears to be relatively reasonable, at least for healthcare providers in a stable financial environment. But Stage 2 is not.

Although consideration is being given to perhaps changing the Stage 2 criteria somewhat and moving the deadline out to qualify, it causes me concern that the original concept was made in the first place. I keep reverting to the old saying, "What were they thinking?" Why set criteria if the ability (technology) does not exist to conform to it? Why is there so little apparent real world thinking behind setting these goals?

Dr. Kahn indicated that his firm has been working for a decade on context challenges for providers, and that considerable progress has been made. However, I infer that much more needs to be done and that providers need to be more aware of what is available . . . and what is not. Again, his organization has been working on this for a decade!

You state that the era of MU will produce an evolution in the EHR. But if the requirements to meet it are not attainable, this seems to me to be counter-productive. Reasonable, attainable goals are the only way to move forward, particularly if we expect the majority of healthcare providers nationwide to do so.

Are my conclusions incorrect? I look forward to your perspective on my questions.

Doc Benjamin

Thanks for your comment, Rich. Seems like the old adage, "this is a good problem to have" applies here.

Dr. Joe,

Your excellent analogy jumped me to 2 considerations. First, if one remembers the PCAST ·path forward' report )http://www.whitehouse.gov/sites/default/files/microsites/ostp/pcast-health-it-report.pdf, one will recognize that a fundamental in the PCAST vision is a Google-like ability for automated web crawlers and such hi-tech tools to sift through EHRs and extract pertinent data. While those tools are getting better and better at context and may well be able to pull data bits very reliably, I don't think they are at the point where clinicians and patients can base treatment decisions on those data points given the complexity of reasons which your blog describes. Misinterpretation of accurate data will lead to invalid conclusions, much like the brain will conclude the incorrect spatial relationship of the monuments in your photos.

Second, MU itself is pushing the envelope in the capability of capturing the metrics required for MU, namely, automating the extraction of data for the clinical quality measures (e-measures) on what percentage of patients with Symptom X got Treatment Y while being (automatically) able to remove from numerator and denominator those patients with Symptom X but who also had Condition Z that contraindicated Treatment Y. In this one and some of your earlier blogs, I recall that you had broached that subject: MU does not require the workflow necessary to support the documentation that would be essential to enable an automated feature to consider Condition Z when counting patients with Symptom X and Treatment Y.