Wednesday, July 10, 2013

This Is An Important Perspective On The Whole E-Health Enterprise. Vital Reading.

This important report appeared a little while ago.

Researchers: Health IT creates its own 'reality'

June 27, 2013 | By Susan D. Hall
Electronic records create a third "reality" in healthcare--one beyond the patient's physical reality and the clinician's understanding of the issues and treatment--and yet another way to miscommunicate, according to a new study.
What if the physician could take a magic stylus and mark errors and ambiguities for developers to address? That would be an ideal scenario, according to research published online this week in the Journal of the American Medical Informatics Association.
The researchers, from Dartmouth College and the University of Pennsylvania, compiled 45 scenarios of miscommunication involving not just EMRs, but also physician order entry systems, pharmacy technology and other systems. They noted that even different clinicians looking at the same screen might develop different ideas about a given situation. They grouped the problem areas in five categories:
  • Information that's too coarse: Significantly different scenarios are described in the same way. For instance, saying the patient has cancer isn't helpful to oncologists.
  • Information that's too fine: Very granular categories within ICD-10 might suggest a certainty that does not exist. To select a very specific subcategory of several possible cancers might prevent continued consideration of others.
  • Missing reality: Some details are missing. Only lab reports and medications are listed; not symptoms or history.
  • Multiplicity: Differing clinicians and staff have differing opinions of reality. Lab results might present others. Including them all can be misleading or distracting.
  • Looking glass: When information in an electronic health record creates a different or incorrect reality. Incorrect sensor data, for example, which the clinician would reject, in the EHR becomes a reality that never existed.
Scenarios examined included:
  • A pill being ordered for a patient who then vomits it up. Has the patient received the medication? The system would show yes, the medication was administered.
  • A doctor ordering medication, but the order not being approved by the pharmacy. In some systems, the order could simultaneously exist and not exist. That could lead another physician to order the medication and the patient to receive a double dose.
  • In the United States, weight generally is measured in pounds or kilograms, and medication is ordered using the metric system. Some EHRs, however, do not designate the unit of measurement, so a 5 in weight could be significantly different depending on whether it meant pounds or kilograms. The difference in medication dose could be lethal for newborns.
Lots more here:
Here is the abstract referred to:
J Am Med Inform Assoc doi:10.1136/amiajnl-2012-001419

Healthcare information technology's relativity problems: a typology of how patients’ physical reality, clinicians’ mental models, and healthcare information technology differ

  1. Sean W Smith1,
  2. Ross Koppel2


Objective To model inconsistencies or distortions among three realities: patients' physical reality; clinicians' mental models of patients' conditions, laboratories, etc.; representation of that reality in electronic health records (EHR). To serve as a potential tool for quality improvement of EHRs.
Methods Using observations, literature, information technology (IT) logs, vendor and US Food and Drug Administration reports, we constructed scenarios/models of how patients' realities, clinicians' mental models, and EHRs can misalign to produce distortions in comprehension and treatment. We then categorized them according to an emergent typology derived from the cases themselves and refined the categories based on insights gained from the literature of interactive sociotechnical systems analysis, decision support science, and human computer interaction. Typical of grounded theory methods, the categories underwent repeated modifications.
Results We constructed 45 scenarios of misalignment between patients' physical realities, clinicians' mental models, and EHRs. We then identified five general types of misrepresentation in these cases: IT data too narrowly focused; IT data too broadly focused; EHRs miss critical reality; data multiplicities–perhaps contradictory or confusing; distortions from data reflected back and forth across users, sensors, and others. The 45 scenarios are presented, organized by the five types.
Conclusions With humans, there is a physical reality and actors' mental models of that reality. In healthcare, there is another player: the EHR/healthcare IT, which implicitly and explicitly reflects many mental models, facets of reality, and measures thereof that vary in reliability and consistency. EHRs are both microcosms and shapers of medical care. Our typology and scenarios are intended to be useful to healthcare IT designers and implementers in improving EHR systems and reducing the unintended negative consequences of their use.
The abstract is found here:
I am not sure just what is the ideal response to this report - but I have a feeling this paper is very much on the something.
Just what needs to be done to provide solutions to the issues raised will take a long time I suspect - and that it will be a very interesting journey.


Bernard Robertson-Dunn said...


My response is:

Isn't that all a bit obvious? Get the data wrong and things don't work properly.

And has been observed before, if you automate something that is wrong, it just goes wrong faster and has worse consequences.

And while I'm at it - none of those three realities are real. They are all interpretations of reality - that's why they are different and why you have to work very hard to keep them in alignment.

Anonymous said...

Agree with Bernard - 'e-health' should not be blamed for what is really a problem of 'multiplicity'. Every healthcare profesional interacting with a patient (and the cloud of documents and anecdotes around her or him) constructs their own 'reality', influenced by their own discipline, background, bias, opinion of other professionals involved, recent 'framing' events, etc. Various professionals' realities - about the same patient - may diverge widely, particularly for complex cases - hence the rise of 'multi-disciplinary meetings' (MDMs). The success of MDMs is very dependent on the ability of their leadership to have the various realities articulated and reconciled.
As Bernard says, e-health makes it easier to disseminate multiple realities more widely. Ideally an e-record would contain established facts and correct interpretations of them, but in the real world the best we can hope for may be clearly stated observations, preferably in standard terminologies, distinguished from the interpretations placed on them, with the source of each indicated (Dr A's gastroscopy findings with images, Dr B's history & exam, Dr C's serology results, Dr D's formulation of the case, etc). That would be 'very hard work' for clinicians with current tools. Can IT make it easier ?