Previous Abstract | Next Abstract
Printable Version
A138
October 14, 2006
9:00 AM - 11:00 AM
Room Hall E, Area E
Prospective Assessment of a New Anesthetic Drug Administration System Designed To Improve Safety
Alan F. Merry, F.A.N.Z.C.A., Craig S. Webster, Ph.D., Louise Larsson, R.N., Jennifer Weller, F.A.N.Z.C.A., Chris M. Frampton, Ph.D.
Department of Anaesthesiology, University of Auckland, Auckland, New Zealand
Introduction: Concern about drug administration errors led the FDA to mandate barcodes on all drug ampules in 2004. Drug errors have been reported in 0.75% of anesthetics in two tertiary hospitals in New Zealand1 and 0.68% in a university hospital in the USA2. In our hospital (A) we progressively introduced from August 1998, a new barcode-based system for administration of drugs and automated record keeping in anesthesia3, aiming to reduce error. We conducted facilitated anonymous incident reporting at hospitals A and B to evaluate the impact of this intervention.

Methods: With ethics approval we collected data to November 03; at A from February 98 and at B from June 99 (including previously reported data to August 991). Anesthesiologists returned a form for every anesthetic, indicating if a drug error or near-miss occurred, with more detail if the response was yes. Case numbers and administrations per case were calculated for 16 casemix categories by randomly sampling 10 records from hospital databases for each category. Errors per administration and per 'standard anesthetic' were derived (= total administrations/overall average administrations per anesthetic). Statistical comparisons used Chi square tests.

Results: 59273 forms were collected from 74478 anesthetics (response rate: 79.6%); 364 errors were reported (0.49% of anesthetics) from 733942 drug administrations for an average of 9.9 drug administrations per case (8.7 with conventional means and 16.9 with the new system, reflecting a mixed vs a predominantly major casemix respectively). Allowing for casemix, error rates were significantly lower with the new system (Table 1). The computerized announcement of the scanned drug name prevented 11 errors.

Conclusion: There were fewer errors per drug administration with the new system than conventional methods. Limitations: this was not a randomized controlled trial (RCT); reporting was voluntary; user compliance with the system was variable (possibly reducing its effectiveness); we measured errors rather than harm. Strengths: large numbers of incident forms; high compliance of participants; facilitation of reporting by asking for negative and positive responses. Every drug administration provides an independent opportunity for error, so estimating errors per bolus is sound, but estimating a 'standardized anesthetic' rate ruled out artificial statistical significance from inflation of the denominator. Demonstration of efficacy for any safety initiative in medicine is notoriously difficult, but these data go a long way towards this goal. A large RCT is planned.

1. Webster, CS, The frequency and nature of drug administration error during anaesthesia. Anaesth Intensive Care, 2001. 29: p. 494-500.

2. Bowdle, A, et al., Anesthesia drug administration errors in a university hospital. ASA Meeting Abstracts, 2003: p. A-1358.

3. Merry, AF, Webster, CS, et al, A new, safety-oriented, integrated drug administration and automated anesthesia record system. Anesth Analg, 2001. 93: p. 385-90.[table1]

Anesthesiology 2006; 105: A138
Table 1. Drug administration errors per bolus, and per standardized anesthetic (administrations divided by 9.9; see Methods. P ≤ 0.0002 for each comparison).
AnestheticsAdministrationsErrorsErrors per administrationErrors per standard anesthetic
New system10816183214600.033%0.32%
Conventional636625507283040.055%0.54%