Prescription errors are frequent on intensive care units. The perception of prescribing as a low status task rather than an essential element of therapy, perceived time pressure and distractions may all be contributory factors.
The authors altered practice on our tertiary paediatric intensive care unit in two stages: formal consultant review of prescription charts on daily ward rounds and requesting re-write for any errors was introduced with the aim of raising the status and visibility of prescription as a task. Subsequently, a dedicated prescription desk was provided and prescription elsewhere was not permitted. Staff were not permitted to interrupt a prescriber at this desk.
The authors termed these combined interventions “zero-tolerance prescription” (ZTP) following a similar approach in Cardiff. The authors undertook an observational study of the impact of these on prescription error rates over 6 months in a tertiary paediatric intensive care unit.
Methods Prescription and administration errors have been recorded prospectively on a daily basis by our ward pharmacist against 44 criteria. These include “clinical errors” (dosage, route of administration, frequency) and non-clinical errors (signature illegible, unapproved names or abbreviations etc). Total errors adjusted for ICU occupancy (errors per occupied PICU day) are presented for three periods: (A) baseline, (B) consultant checking prescription charts and (C) full ZTP. Comparisons are made between mean error rates with t-tests.
Results (A) Baseline mean error rate over 12 weeks was 1.8 errors per occupied PICU bed day (95% CI 1.5 to 2.1), (B) In the 20 weeks following formalised consultant checking of charts, this was reduced to 1.4, errors per occupied PICU bed day (1.1–1.6) (p=0.0035 vs A). (C) Following the introduction of the full ZTP, protocol error rate was 1.1 (0.8–1.3) (p=0.001 vs A, and p<0.05 vs B) over a 10-week period. This constitutes relative risk 0.59 for error. Infusion prescriptions errors were most improved A) 0.3 day (0.2–0.4) per occupied PICU bed vs C) 0.1 (0–0.2), p=0.02 (relative risk 0.45).
Comment In this unblinded study, the ZTP package was associated with a significant and prolonged reduction in errors. The impact of these changes is likely to be highly influenced by local factors but merit consideration on PICU.