Why medical device data is the best way to fill meaningful use EHRs and conduct comparative effectiveness research (CER)

I will be presenting at the O’Reilly Open Source Convention (OSCon) in Portland at the end of the month. As an avid reader of dozens of O’Reilly’s technical books over the years I was excited when they reached out to ask if I would talk about open source in the healthcare world. While open source isn’t a major force in the healthcare IT ecosystem right now, that will be changing over the coming years and should be able to change the medical world in the same way that open source has improved enterprise IT and made the consumer web possible.

My presentation on Thursday July 28th at OSCon is on why medical device data is the best way to fill meaningful use EHRs and how open source technologies and open architectures can make that possible. I was interviewed by O’Reilly’s Andy Oram about this subject and the podcast is available.

What I told Andy during the interview was that “Meaningful Use” is all about data, not about EHRs — the government needs data for cost comparisons, health professionals need data for treatment research and chart management, and patients need data for choosing the right providers and treatments. EHRs are just a vehicle, not the end goal.

Right now we know that Medicare and Medicaid are paying more 50% of the nation’s healthcare costs but doing so as “fees for services” without regard (for the most part) to what treatments, medications, or tests really work. The evidence-based research that goes into figuring out what works and what doesn’t is the foundation of what has been affectionately known as “Comparative Effectiveness Research” (CER) and is being re-branded as “Patient Centered Research”.

The government needs tons of data for CER, which is designed to inform health-care decisions by providing evidence on the effectiveness, benefits, and harms of different treatment options. The evidence is generated from research studies that compare drugs, medical devices, tests, surgeries, or ways to deliver health care.

In the best cases, researchers review evidence about the benefits and harms of each choice for different groups of people from existing clinical trials, clinical studies, and other research. Then, over time of course, researchers conduct studies that generate new evidence of effectiveness or comparative effectiveness of a test, treatment, procedure, or health-care service.

CER sounds like it’s all about the government and evidence-based medicine to contain healthcare costs but ultimately it’s about providing treatment comparison choices to help make informed decisions. In the end healthcare professionals must deliver tools to the patient that can help the patient and their families select the right treatment options.

Given that CER is very important and is ultimately why the government is giving away billions of dollars in incentives for healthcare IT (and not to prop up the EHR market), it’s important to understand where data comes from so that we know what the best approaches are to collect it.

There are two kinds of data that we normally see: one is structured data which is computable, analyzable, comparable, quantifiable, and reportable; the other is unstructured data which is basically dictations, documents, faxes, letters, and almost everything that is not computable or easily analyzable but is electronic nonetheless.

There are many sources of both unstructured and structured data but and oversimplified view is shown below. In a typical provider environment medical information comes from patients, professionals, labs/diagnostics, and medical devices. The table below shows how unstructured and structured data is “sourced” today.

Patient Health Professional Labs & Diagnostics Medical Devices
Source Self-reported by patient Observations by HCP Computed from specimens Computed real-time from patient
Unstructured Data Yes Yes Yes No
Errors High Medium Low
Time Slow Slow Medium
Reliability Low Medium High
Data size Small Small Large
Availability Common Common Common Uncommon
Structured Data Yes Yes Yes Yes
Errors High Medium Low Low
Time Slow Slow Medium Fast
Reliability Low Medium High High
Discrete size Small Small Small Small
Streaming size Large
Availability Uncommon Common Somewhat Common Uncommon

As is evident by the table above, many of the existing MU incentives in Phase 1 (patient reported and healthcare professional entered especially) promote the wrong kinds of collection: unreliable, slow, and error prone. Accurate, real-time, data is only available from connected medical devices and labs / diagnostics equipment.

Given that meaningful Use and CER advocates are promoting (structured) data collection for reduction of medical errors, analysis of treatments and procedures, and research for new methods it’s important to see that we’re not going to get real gains until the medical device vendors are fully connected and providing data directly into EHRs or clinical data warehouses.

Please check out my O’Reilly interview with Andy Oram about this subject and let me know whether you agree.

Newsletter Sign Up

4 thoughts on “Why medical device data is the best way to fill meaningful use EHRs and conduct comparative effectiveness research (CER)

  1. Pingback: Galen Healthcare Solutions: Allscripts Consultants Enterprise EHR » EHR Unstructured Data Mining

  2. Pingback: EHR Unstructured Data Mining | Galen Healthcare Solutions: Allscripts Consultants Enterprise EHR

  3. Shahid,
    I’m a known evangelist regarding potential value of medical device data integration in EHR but see disturbing “IT-think”  in MDI articles from many well regarded IT sources.  There is failure to recognize 1) role of monitor data in larger assessment process, 2) QA of continuous monitor data done by RNs today and 3) potential  impact of unvalidated  device data “entered” into EHR.   

    In most intensive monitoring (high acuity) units, RNs today “audit” (QA) physiologic monitoring data in paper and electronic worlds.  RNs don’t just copy any number on a monitor and chart it, they evaluate continuous monitoring readings, reject  artifact, outlier and erroneous data, to select values to chart or enter into EHR.  Inaccurate data “entered” into EHR can trigger a series of inappropriate and potentially contra-indicated alerts, warnings or conditional orders in EHR or surveillance systems.  

    The other missing concept is process of in-person assessments  by RNs (the primary reason for hospitalization).  In addition to “auditing” monitor “readings”, RNs  observe, palpate, auscultate, smell and interview patient, as indicated.  They also chart associated characteristic of data values which are critical (e.g. pulse thready, respirations stertuous) which no device can do.  Even if values are accurate and normal, they are not complete picture – vital signs are often a late state indicator of change in patient status.    

    The fact nursing data are not charted real time or codified are different problems.   We’ve had reinforced in  recent reseach studies (e.g. Rothman, Shever) that it’s nursing observation / critical thinking that prevents patient deterioration before they require  “rescue”.   These data need to be viewed in context.   

    We do need to code nursing data and enforce best practices and provide IT that supports near real time charting (using MDI to importing data for RN QA).  However, the concept that a stream of unverified numbers coming from devices “entered” (whose e-signature?) into EHR  is of little value and may be dangerous, yet appears to be what’s being promoted in IT circles. 

     What’s we should be striving for are accurate, timely patient assessments (and near real time charting) of all data types required for clinicians to make safe decisions re patient status, response to treatments and needs for interventions  – data from monitors (with QA!), specimens, RN (coded) observations and findings and patient responses.    

    1. Ann, terrific feedback and thank you for responding.  All your points are quite valid and I was wondering if you’d like to elaborate your thoughts in a guest posting? I’d love to catch up on a call and discuss what you think should be done (because MU phase 3 is doing to be requiring device integration). Thanks.

Add Comment