“Errors” were defined as any kind of mistake in the process of ordering or administering a drug or an adverse drug event, which includes harm or injury to a patient related to a drug, whether or not it was caused by an error.
How else could they conduct this study? It would be completely meaningless if observers had to second-guess after an observed harm or injury whether or not it was an "error" or just bad luck. It makes no sense to catalogue interactions which could have caused harm (but may or may not have done so) while ignoring the ones which did cause harm (but weren't obviously errors).
Then the headline shouldn't call them all errors, if they do not have evidence that they were in fact errors. "Medication errors or adverse reactions found in 1 out of 2 surgeries" is a less sexy but actually accurate title.
There's a method known as "chart review" that's meant for exactly this purpose - essentially reviewing the evidence available to see if the same decision would have been made. It's expensive, but doable.
“Errors” were defined as any kind of mistake in the process of ordering or administering a drug or an adverse drug event, which includes harm or injury to a patient related to a drug, whether or not it was caused by an error.