What’s in YOUR Medical Record?
By Ken Schafer
If my wife were admitted to the hospital with diabetic ketoacidosis (DKA), I’m pretty sure I wouldn’t want her electronic record to erroneously record a leg amputation (BKA). I’m equally confident that if this documentation mistake were made, I wouldn’t care too much how it happened. I would just want it fixed.
And if incorrect documentation on my diabetic wife resulted in an incorrect treatment course, which resulted in her death? You might end up with a $140 million verdict like this one.
Inga’s post on The Atlantic’s “The Drawbacks of Data-Driven Medicine” (from Big Datty,on 6/12/13) illustrates something that we all know to be true. Our medical records often contain mistakes, and electronic errors perpetuate themselves embarrassingly quickly. But her comments – and the source article – miss two very important points.
Doctors are responsible for the content of the records they create. This is true regardless of the method used to document patient encounters. Blaming the speech recognition system for hearing “DKA” instead of “BKA” makes no more sense than blaming a keyboard for a typographical error. If the physician picked the wrong checkbox on an EHR interface, would that be the fault of the EHR? Of course not.
Speech recognition, keyboarding, and dropdown menus are all methods for data capture. For that matter, so is a more traditional transcription process. But all of these methods have one element in common: the final content should be reviewed and validated by the documenting clinician. Physicians who fail to do this put their patients at risk.
Doctors make mistakes. I know a radiologist who dictated “liver” when he meant “heart.” The transcriptionist dutifully returned the report with the word “liver,” and it was signed by the physician. When the mistake was discovered, the audio was retrieved. The doctor listened to himself dictate the wrong organ, and blamed the transcriptionist. The point? Doctors are people, and people make mistakes, whether they own up to them or not.
That same physician was convinced speech recognition would eliminate transcription errors, and he was right – sort of. What speech recognition systems really do is eliminate transcriptionists, not errors. If radiologists are involved, there will still be errors. There’s no speech recognition system that will hear the word “liver” and change it to “heart.”
In fact, in our DKA:BKA example, the doctor may have had a bad day and actually said BKA to the speech recognition system. No matter what, though, the doctor made a mistake – either in what he said, or in what he saw on the screen and failed to correct.
Those with experience greater than mine often post to HIStalk about the shortcomings of EHRs in terms of the data they contain, with usability and completeness being favorite topics. My concern for our records is more specific. Especially when speech recognition is involved, what metrics do we have in place to make sure that narrative data is recorded accurately? If doctors are responsible for the content of their documents, and we know they make mistakes, how do we monitor and improve the quality of the narrative components of our EHRs?
As the government, physicians, patients, and the free market determine what systems we are to use and how they should work, we should never lose sight of this one truth: no matter what’s in the record, it should be right.
Ken Schafer is executive vice president, industry relations for SpeechCheck.