To Err is Not Just Human

In the article How to Eliminate EHR-Based Medical Errors, presents a list of computer errors that affect patient care. While many tend to blame poor programming and software glitches, these developmental issues are only partially to blame. The ultimate culprit is a lack of standardization in EHR systems, in general. Sufficiently standardized systems would make it possible for a physician to enter data at one hospital, and when at a different one, use the very same methods.

 

Unfortunately absence of standardized interfaces is ubiquitous in our currently technological society.  For example, how often does one go to one store, say a grocery store, and when trying to pay for goods, has to deal with a card reading device that is entirely different from the adjacent store.  One has to figure each one out independently. If spending too much time shopping, one could easily run into ten different interfaces.  Quite maddening.  But that is the normal world, one would expect better of medicine and medical technology. Or is that expecting too much?

Some examples of the confusions encountered include:

1.  Data base/CPOE’s (computerized physician order entry systems)have inconsistent dosing for medications, or manages to switch the dosing between 2 medications
a.  Solution is national data base for CPOE. The data base can “live in the cloud” and be accessible to all

2.  Artificial intelligence that is artificial but not intelligent, such as incorrect weight adjustment for natural growth
a.  Solution –data base that is accurately age and weight adjusted and that catches irregularities and sends notification.

3.  Failure to inform clinician of critical lab information is a major problem.  The more clinicians are depending on technology, the more they naturally depend on them to flag problems.
a. Solution –warning system of critical lab results---coordinated through a national data base for national CPOE, mentioned above.

4.  Prescription with wrong decimal point for dangerous medications. Deadly !
a.  Solution—standardized CPOE/Pharmacy and automated safety testing.

5. Duplicate patient records.  How difficult is it when Patient Sam Q. Brown enters and registers as Sam Brown.  But the computer is thrown off by the absent Q…..and then a second record is created, one that does not have all the important information that Sam Q needs to have in place.
a. Database error created by expected variance in human nomenclature. The error should be trapped and the medical team prompted to sort it out and reconcile the records. Google does this, why can’t medicine.

6.  OUTPUT is “Gobblygook” at best. A combination of typing, cut and paste, macros, mini-macros, and  “computerize” that makes no sense. The inputs are all given the instruction to create text. There is no instruction set that can be made which will create automatic meaningful text.
a.  One solution may be greater use of voice recognition. For example, Voice Activated Technology (Dragon) allows a clinician to dictate several common sense sentences about what really happened during the encounter.  Siri does pretty well on this too.  Apple, Dragon do it, why not get medicine more fully on board?

7.  Scanned documents sent to wrong patient.
a.  No fail safe mechanism exists for stupidity or incompetence

8.  Action items never were seen or acted on by the clinician.
a.  Programming issues/inappropriately mild alerts can be blamed, often
b.  Often, it is difficult for a provider to comment on action taken because there is an inability to easily add addendum.  Developers should really try to make the systems at least better than paper in this regard.

The fundament problems are lack of standardization, easy navigation, appropriate warnings, and ______________ fill in the blank. Meaningful use money would have been better spent on a national data base and CPOE. Each vendor could then create their own system with built-in universal knowledge, integration, and interoperability as a baseline.