We welcome the research of Barker and colleagues1 into medication errors arising in hospital settings. They emphasize the importance of dosage errors, the second largest source of errors. Dosage errors can occur, however, not just at the time of drug administration, but also during the calculation and preparation of treatment.
Observing the behavior of subjects as they undertake their tasks has been shown to introduce bias—the Hawthorne effect.2 This may explain why, in 2 units, Barker et al1 found no errors, especially since those units were also being assessed for accreditation during the period of the study. We and our colleagues conducted a study3 to detect errors in the giving of acetylcysteine to patients with paracetamol (acetaminophen) poisoning. In the United Kingdom this drug is given by intravenous infusion of 3 sequential doses that depend on body weight. Our study avoided direct observation: nurses and pharmacists who prepared bags for infusion were asked to prepare as usual. We asked the staff, however, to send us 5 mL of the infusate taken before the infusion was administered and, if possible, 5 mL taken after the infusion had finished. We compared the concentrations of acetylcysteine in the infusate with what should have been present, based on the patients' weights.