Written by: Katherine Rourke
As long as there have been EMRs, there’s been endless debate over what system designs are most appropriate. Unfortunately, no matter how heated a threaded discussion gets, it’s unlikely to solve big problems.
Now, however, we may have a chance to build a consensus on what NOT to do in building out EMRs. A new report from NIST has painstakingly analyzed which EMR design factors have an impact on usability (PDF), including one subset which seems likely to cause patient harm.
The section on design problems which may cause patient harm is (unfortunately) rather long, so I’ll only provide some of the highlights, but you can download the whole PDF by clicking on the link above. (The “potential for harm” section begins on page 66.)
One major area NIST addresses is patient identification errors. For example, if EMR displays don’t have headers with two patient identifiers, lock out or control multiple accesses to records, or fail to provide full patient identification with integrated apps like imaging, the wrong actions could be performed on the wrong patient.
Another major concern NIST identifies is data accuracy errors. There’s lots of ways EMR design foster data errors, the report notes, including when information is truncated on the display, when accurate information isn’t displayed unless users refresh the data, when discontinued meds aren’t eliminated and when changes in status aren’t displayed accurately.
NIST also identifies data availability errors as a big issue. Among other concerns, clinicians can easily make mistakes if they can’t easily see all the information they need to understand doses without additional navigation; if complex doses aren’t easily understandable without extra navigation; and if information accurately updated in one place shows up accurately and efficiently within other areas or integrated software.
As you can imagine, NIST has a lot more to say here. The report also includes analyses of how mode errors, interpretation errors, errors when physicians are forced to remember data, lack of system feedback when clinicians make inappropriate actions for the context and other tricky designs cause errors that can harm patients.
While I’m not a clinician, so bear this in mind, my feeling is that everyone here ought to read this report. Lots o’ valuable insights here!