By now, I think we’ve all had the infuriating experience of calling our cable provider or any number of other oversized companies and being asked by a friendly but thinly disguised ‘robot’ voice to explain the reason for your call. It then processes and regurgitates what you’ve just said and asks you for verification. If it was incorrect, this process can continue until you find yourself screaming “REPRESENTATIVE!” into the phone five times before you are either disconnected or miraculously routed to an actual human.

Now, imagine this same technology being used in your medical records. Yes, really. It is called speech recognition (SR) and in the last 10 years, it has single-handedly wiped out an entire occupation made up of highly trained medical transcriptionists. It has also become one of the most common sources of errors in medical records and most people are not even aware of it.

Before I tell you why this should be important to you, let me explain how things worked in the days pre-SR. It used to be, for instance, that a radiologist would read your MRI and while he was looking at all the images on his computer monitor, he would pick up a microphone and dictate his findings. That dictation would then go to the medical transcriptionist, who would transcribe his dictation into a neat, accurate, and grammatically correct medical report. She/he would be trained in medical terminology and anatomy and would know enough to catch any mistakes that the doctor made. If, for example, it was an MRI of the left knee and the doctor stated “right knee” in the report, the transcriptionist would flag that and bring it to the doctor’s attention for correction. If a doctor mentioned “ovaries” in a report done on a male patient, the transcriptionist would catch that and send it back to be fixed. You get the picture.

Now, however, doctors are talking to their computers. They are dictating to software that cannot catch mistakes like a human can. Add to the mix the accents of many doctors, bad grammar, background noise, and what you get is often funny, but always dangerous. Some actual examples: “MRI of the brain shows the testicles to be mildly prominent”, “the patient has 2 teenage children but no other abnormalities”. Funny, yes, but grossly incorrect.

Not only that, but doctors are now being asked to proofread their own reports before signing off on them, something their transcriptionists used to do. Most doctors are not willing to take precious time away from seeing patients, and the result is that they sign the reports electronically and send them off without looking. In fact, you may see at the bottom of some of your test reports a disclaimer stating something to this effect: “this report was produced without benefit of proofreading and may contain typographical errors”. This is a recent effort to get around any medical liability for errors. So your report could be signed and sent off with the recommendation for biopsy of the wrong breast, with comments on organs you don’t possess, or with just plain gibberish.

Bottom line? Voice recognition saves hospitals and clinics money, and despite quality being sacrificed, they’re not going back. The industry-wide decision to go all in on SR software is the single most overlooked area in which medical errors are occurring. So review your medical records carefully. You are entitled by HIPAA to obtain copies of your medical records. Be proactive and do so, especially any diagnostic testing. Read through carefully, and be sure to ask the provider who issued the report to correct any mistakes, and then get a copy of the corrected report. And if you need assistance obtaining the records or translating the medical terminology, contact me. Organizing and correcting your medical records is a service patient advocates offer. This could avoid potential medical errors and protect your health down the line.