While electronic health records (EHRs) are intended to improve
patient care, some risks are increased over the old paper record.
Charting On Wrong Patient In EHR (COWPIE) is one such risk. A
recent study examining one component of that, Computerized
Physician Order Entry (CPOE) has some interesting findings. And,
as the authors state, increasing automation, while reducing the
opportunities for human error, also reduces the opportunities for
humans to identify an error and intervene.
In the children’s hospital studied, automated surveillance
identified 644 probable CPOE COWPIEs. Only four had been reported
to risk management, presumably events in which the wrong patient
received the medication. The great majority of these events
appeared to qualify as near misses, with the errant order rapidly
cancelled and replaced with the correct order for the correct
Risk factors included:
Age: infants and newborns were much more likely (2.9 and 3.6
times, respectively) to have wrong-patient orders.
Last name: two-letter overlap 4.4 times more likely.
Location: patients in nearby rooms 2.8 times more likely.
Day of week: Friday 2 times more likely than Monday.
Hour of day: midnight to 6 am 1.7 times more likely than 6 pm
More physicians ordering for the patient: 1.4 times more
It may be easier to pull the wrong patient’s chart in an
electronic versus paper format, when only a few charts were at
hand. Now all the charts in the system are readily available and
may be pulled up by error. A standard approach should be taken to
assuring that the correct chart is used when entering information
Hospitals may be interested in using this study’s surveillance
method to identify their risks. In the study, they identified
potential CPOE COWPIEs: if a provider 1) ordered a drug on a
patient, 2) cancelled the order within 120 minutes, and 3) then
reordered the same drug on a different patient within 5 minutes
of cancellation, it is presumed to be an error. When the authors
performed chart reviews on a subset of these automatically
identified “errors”, they found that at least 60 percent and
perhaps as many as 100 percent of the charts confirmed the error
(documentation of the reasons for the provider’s actions often