New Visions and Current Evidence for Safety in Anesthesia

To the Editor We enjoyed the recent discussion article by Grigg and Roesler1 and agree with the general logic of the argument that human factors are important for safety in anesthesia. However, there are a number of statements in their article that seem to us a little like “tilting at windmills” since the authors appear to omit much of the real progress that has been made in the very areas they themselves are advocating. For example, Grigg and Roesler1 state that we do not know the true prevalence of medication errors in anesthesia.1 However, a good estimate of this is possible, and neither do we need to know the absolute “true” rate to detect improvements using data collected in the same consistent way. One of the most accurate ways to collect large data sets on the rate of medication error is to use prospective, facilitated incident reporting, in which a report is made for every anesthetic, an accurate denominator is known, and dedicated personnel follow up if such a report is not made. Five such studies have now been conducted around the world, in New Zealand, the United States, South Africa, and China, comprising a total of 402 reported medication administration errors in 82,881 anesthetic cases.2–6 These studies yield a pooled rate of error of 0.49% per case and a narrow 95% confidence interval of 0.44%–0.53% (Figure).Figure.: Forest plot of prospective, facilitated incident reporting studies showing the number of medication administration errors per total anesthetics in each study (95% CI), and the pooled estimate of the rate of error at 0.49% (95% CI, 0.44%–0.53%) per case, based on 402 errors in 82,881 anesthetic cases. CI indicates confidence interval.Grigg and Roesler1 also claim that there are no alarms on the medication side of the anesthesia workspace for bolus administrations of drug, and that it is too early to know whether barcode-based checking systems that do exist are durable solutions. Our group began using a barcode-based system in early 1998 that includes an auditory announcement of the identity of the drug before each bolus administration.7 The system has been in continuous and increasingly wide use ever since, and incorporates many of the human factor safety principles mentioned by Grigg and Roesler.1 The system now includes auditory and visual alerts for known allergies, an auditory and visual prompt if the prophylactic antibiotic has not been given within a certain time, and a reminder for subsequent doses of antibiotic after 4 hours. It also generates a complete anesthetic record. Using facilitated incident reporting, with a control group, in 74,478 anesthetic cases, we have shown that the system can reduce medication errors in anesthesia by 35% (from errors in 0.049% of bolus administrations to 0.032%; P = .002).7 In a smaller scale randomized observational study of 1075 cases, we have also shown that the system can reduce errors in medication administration and record keeping by 21% (from 11.6 errors per 100 cases to 9.1 errors; P = .045).8 We believe that these results are cogent evidence that such approaches work, and 20 years of experience with the system would also suggest that such solutions are durable. We agree with Grigg and Roesler1 that the future of safety in anesthesia, and health care generally, relies on automated data collection, and we look forward to the further evolutionary advances this allows. Craig S. Webster, MSc, PhDAlan F. Merry, MBChB, FANZCADepartment of AnaesthesiologyUniversity of AucklandAuckland, New Zealand[email protected]