Lessons Learned

Watch Out for This Four-Letter Word

Sifting through safety reports at your facility in search of significant issues is daunting. CHPSO shares the challenge: we have about 870,000 reports and more arrive daily.

This challenge is also an opportunity, as we aggregate information from hundreds of providers, and can observe patterns and commonalities in risks and causes that may not be obvious at a single site.

This brings us to the four-letter word of danger. We have found a large number of events in which “auto” signals a significant cause. Sometimes the word is part of a larger word, such as autocorrect, auto-cancel, auto-fax or auto-substitute.

Whether alone, or as part of a compound word, this is a signal that the automatic actions of the system may have surprised someone, acted in an unexpected way, or been overly relied upon. In short, events bearing this four-letter word are a window into the perils of automation. In many of these cases, if a human were actively involved, the error could have been caught.

Some examples:

An auto order associated with starting epidural analgesia discontinued the narcotics given to a chronic pain patient. No one questioned that automatic order.

Mis-transcription from the EHR to the pharmacy computer resulted in autofill of a syringe to 20X the ordered dose, which was then placed in the automated dispensing cabinet. The nurse pulled the dose out of the cabinet and administered it to the patient.

When dispensing a 4 mg vial, “3” instead of “3 mg” was entered in the system. Since units were not specified, the system automatically interpreted the 3 to be “3 vials”, dispensing four times the intended dose.

Antibiotic was given for extended period, as auto-stop policy was approved but to everyone’s surprise not successfully turned on in the system.

A substitute drug was dispensed by automatic therapeutic substitution, but the original drug not discontinued. Both were given.

The hospital had autostop on perioperative antibiotics. The pre-op orders had to be altered post-op, and the EHR failed to put the post-op order on autostop. It turned out that the autostop rule as implemented only applied to antibiotics ordered pre-operatively.

Changes in a patient’s “admit” status caused the system to auto-verify unverified orders.

Blood products ordered late in the month but given the next month were supposed to be automatically carried over to the new month, but the system dropped some orders.

Lab technician manually corrected blood count without realizing that the instrument had automatically corrected the blood count, resulting in overcorrection.

About 7 am, an hour after the 6 am medications were given, the drug order was changed to “daily.” The system automatically added a dose at 9 am that day and cleared the MAR of the prior medication as that had been “discontinued,” resulting in a double-dose.

Physician accidentally ordered a duplicate CBC. The lab tech was notified and deleted a CBC order. Unbeknownst to her, the system had already detected and deleted one of the duplicate CBC orders, so no CBC was run.

A procedure was ordered, but then automatically canceled by the EHR as the procedure did not exist in the system (the hospital does provide the procedure).

Suggested search strategy:

Whole text search of the narrative portions of the reports,  WHERE Word Like “auto*” And Word Not In (“automotive”,”automobile”,”autologous”,”autograft”,”autos”,”autozone”) And Word Not Like “autop*” And Word Not Like “autotransf*” And Word Not Like “autoclav*” And Word Not Like “autoim*”. CHPSO has over 1200 cases meeting this criterion.

Commands