Skip to content Skip to navigation

At AMIA 2015: Not Your Mother’s EHR

December 23, 2015
by Chuck Appleby
| Reprints
What were some of the key insights shared with participants attending AMIA 2015 in San Francisco, around EHR optimization and value?

Attending the “Informatics Year-in-Review” presentation at the AMIA [American Medical Informatics Association] 2015 Annual Symposium in San Francisco was a little like watching one of those history-of-the-world-in-five-minutes shows.  Beneath the breathtaking volume of research highlighted, however, was the clear implication that medical informaticists are bent on staying relevant in an industry quickly moving to value-based care. In her presentation, “Clinical & Consumer Informatics,” Patti Brennan, R.N., Ph.D, of the University of Wisconsin-Madison, shared many nuggets on how the medical informatics community is embracing the issue of EHR [electronic health record] optimization for accountable care.  Here are a few :

EHRs do more for process than for outcomes.

“EHRs result in better care management but do not improve patient outcomes,” said Brennan, “so we have to think about integrating the technology into the process.” She cited research that showed use of EHRs resulted in

  • improved case management;
  • reduced in-hospital mortality (but not much else);
  • better interception of wrong patient orders;
  • reduced inappropriate haloperidol orders by 80% (those not consistent with an alert);
  • more appropriate ED admit/discharge decisions;
  • and improvement on 11 of 12 quality measures.

Meaningful use is really useful, but not helpful (yet).

The MU incentive program has resulted in better and more accessible electronic health records and the next step is to ensure they demonstrate an impact on outcomes. “The tools work as they promise,” said Brennan.  MU has improved provider performance in the sense that use of specific EHR functions was associated with higher performance on healthcare process metrics such as condition-specific, best-practice alerts and order sets), but not outcomes. Interestingly, pediatric hospitals lag in adoption of key MU features.

Extracting meaningful info from EHRs

“When you use the right models—new tools like natural language processing, information extraction and online learning—we can extract the right information out of an EHR. However, EHRs will require additional technology work in order to ensure we’re not just pulling data out but are able to draw conclusions about patients and their health status,” said Brennan.

It’s not your mother’s EHR.

FHIR is a data-exchange standard that enables a piece of information from one EHR to integrate itself into another EHR completely, going beyond the limitations of data messaging.

“EHRs can now become platforms for population health, identifying risk and facilitating predictive analytics. They can be engines to help the business side, become more proactive. This is really exciting,” said Brennan, adding that evidence shows EHRs can help identify clinical-trials participants via relevance-based method reasoning, promote open communication with patients, improve medication adherence and predict future health.

Novel EHR structures

Three points here, noted Brennan, about EHR customizability:  One, EHRs are not simply shrink-wrapped products, but can offer innovative access to data such as large image files in the cloud. Two, EHRs can provide the platform for specialty electronic records. Three, using innovative data architecture, grids can be overlaid on EHRs that allow visualization and image management.

HIEs promote care coordination, seem to save money, & reduce test ordering.

“HIEs are proving to be as good as they promised,” said Brennan, noting that they have cut costs 11 percent in Korea and have led to improved care coordination and reduced imaging in EDs in the United States. However, the Veterans Administration has found, not surprisingly, that HIEs are only good as the data they exchange.

Clinical Decision Support systems WORK…if they are used and built to standards.

“This theme comes through over and over in all the research: the tools are having an effect but not showing effectiveness,” said Brennan. “If clinicians are not using them, they are worthless.”

  • Clinicians who used CDS gave twice as much care, but patients did not lose weight, for example.
  • Value is in the computational support, not knowledge access.
  • CDS does not improve outcomes.
  • Case studies reveal limits of CDS and the lack of standards to guide vendors or purchasers.

Assisted cognition, cognitive informatics and situation awareness

“All EHRs have individual users who think in unique ways. It’s possible to augment someone’s thinking rather than merely present data to them. If you build your CDS alerting system for a drug/drug interaction, for example, you can save a clinician or pharmacist time and make them more efficient. However, health systems can individually target CDS to help a clinician in the ER in the things they don’t do well and leave alone the things they do well. Don’t try to replace human judgment,” said Brennan. 


“We know poor implementation of EHRs can lead to serious errors, especially when inappropriately integrated into care,” Brennan said. “It’s even worse if you force people to use a bad tool.”

Research among nurses shows that attitude—not skill or EHR features—is the most predictive factor of future “cognitive burden” of EHR implementation.

Providers using the same EHR developed personalized patterns of use and that’s ok. Health systems need to accommodate these differences to best optimize EHR adoption.

Infrastructure readiness