The thesis objective is to analyze smart infusion pump data from the Children's
Hospital of Eastern Ontario to learn about pump usage. Trends in compliance and
hard limit events were shown to be potentially associated with drug library updates,
and can be used to assess library changes. In a case study for the drug gentamicin,
selection errors were the cause of more than 50% of hard limit events. A human
computer interaction issue was identified as the problem, where gentamicin has two
library entries that are named in a manner that may be confusing. Decision trees
were used to
determine factors that are associated with hard limits. Results showed
that higher hard limit event rates were associated with the NICU and Emergency
profiles, and with patients who weighed over 48 kg; these factors can be used to
target training and research. This thesis contributes to research in medication errors
and patient safety.