We compared patient-level costs and the number of
peripheral intravenous catheters used per patient with
arithmetic means and the independent sample t test; we
used bootstrapping with 1000 replications to calculate
95% CI for costs.16 We collected resource-use data as how
many catheters a patient had fi tted and removed, and the
costs associated with the insertion setting (eg, by staff in
the operating theatre vs the medical and surgical wards).
We compared overall treatment time (all intravenous
catheters) between groups with the median, quartile,
range, and Mann-Whitney test. We deemed p values less
than 0·05 to be signifi cant.