This study was designed to investigate the impact of CPOE, without decision support, on MEs in the critical care setting. The data collected were viewed in terms of proportion of errors, patient outcomes arising from the error and types of error.
The proportion of MEs reduced following the introduction of CPOE. There was also some evidence that a learning curve occurred with CPOE, as the proportion of errors appeared to decline over time. This learning curve could have included improvements made to the system in light of experience, although it is conceivable that the ME rate may have reduced by itself over time. The error rates found were less than those reported in a recent study of prescription errors in UK critical care units [12]. There was no difference in the mean APACHE II score in the HWP and CPOE periods, indicating that it is unlikely that severity of illness differed substantially in the monitored periods.
It was decided to separate the recording of non-intercepted and intercepted errors (where an error was spotted and corrected before having an impact on the patient). The intercepted errors were scored on the basis of what might have occurred if the patient received the medication as prescribed. There was a demonstrated benefit on patient outcome scores with CPOE prescribing when the intercepted errors were combined with the non-intercepted errors. It was reassuring to note that no patients suffered permanent harm or death as a result of any non-intercepted error. Three errors, which all occurred with CPOE, could have led to permanent harm or death had they been administered as prescribed. This CPOE system lacks the ability to effectively deal with drugs with variable dosage regimens such as vancomycin, gentamicin and warfarin. In addition, our impression is that prescribers often prescribed too quickly and made mistakes when using pull-down menus, as seen with the diamorphine error. A lack of product knowledge probably led to the amphotericin error. Prescribers need to develop a thorough, systematic approach to prescribing, similar to that which they employ for diagnosis. This aspect of our findings is in accordance with a recent study that identified that a CPOE system frequently increased the probability of prescribing errors [13].
Most of the errors were defined as 'minor' in outcome and, as such, did not cause the patient harm but, in some cases, may have lead to an increase in monitoring but with no change in vital signs. There were four errors, however, that caused either patient harm or increased monitoring and 34 intercepted errors that could have potentially caused harm had they been administered. The fact that these MEs were rectified before they harmed the patient underlines the value of daily prescription review by an experienced clinical pharmacist [14, 15]. In contrast to other views [8], it was decided not to regard abbreviated drug names as errors, because this would have distorted the results in favour of CPOE. In justification of this treatment of the results, no abbreviated drug name led to a patient receiving the wrong drug, but it is regarded as poor prescribing practice as defined by our own prescribing guidelines and national guidelines [16]. CPOE effectively eradicated the use of abbreviations.
The study was not designed or powered to identify differences in the types of errors under the two systems. Future work should be designed to focus on these differences. Omission of key prescription details such as dose, units, frequency and signatures appeared to be much reduced with CPOE, as the computer program did not permit drug entry with missing key data entry fields. Dose errors were still prevalent with CPOE, however, as a result of physicians choosing the wrong drug template, selecting from multiple options, or as a consequence of constructing their own drug prescriptions using pull down menus.
There were also many missed prescribers' signatures with CPOE. This did not affect the patient but, in these cases, nurses administered medication without a legally valid physician order. Although an absent 'signature' with CPOE was regarded as an error, the audit facility of the Clinical Information System did record who prescribed the drug. There were several cases where necessary drugs were not prescribed with CPOE; this was probably not related specifically to the prescribing system.
The categories described were specific to the setting and systems, thus a general taxonomy of medication errors [17] was not used as it was considered that this did not adequately characterise the errors. The categories used here specifically describe the event and general taxonomies were considered to be too broad to provide a specific and useful description of the episode.
During the data collection period, key staff such as consultants, senior nurses and the pharmacist remained the same, so this did not influence the results. Pharmacist attendance at ward rounds has been associated with a reduction in adverse events [15]. In this study the pharmacist attended the ward round throughout the study. No other significant organisational changes occurred during the study period. The only possible changes were the junior medical staff who did change during the study and this may have affected the results. Ideally, the impact of this could be minimised by sampling over a longer period and more frequently, but this was beyond the scope and resources of this study. Alternatively, we could have statistically adjusted for experience level, although this is a difficult issue and has not been attempted by other researchers. Furthermore, the MEs recorded were all proactively identified from the daily pharmacist prescription chart review, and thus did not rely on the notoriously low reporting of multi-disciplinary adverse incident reports. Patient outcome was assessed by the pharmacist and clinical director, who were not blinded to the prescribing system; this could have introduced the potential for bias in the results and is a limitation of the study.
Medical errors are among the leading causes of death in the United States. In its highly publicised report, the Institute of Medicine estimates that between 44,000 and 98,000 Americans die as a result of medical errors each year, with the majority of these errors being preventable [18]. MEs are the leading type of medical error [3]. Previously, in a setting that included general wards and ICUs, a similar type of CPOE was associated with a halving of the rate of non-intercepted MEs [19]; ours is the first study identified that investigates the impact of CPOE on MEs solely in an adult ICU. CPOE is already the subject of considerable interest [20] and has already shown benefits in paediatric medicine [21–23]. A systematic review of the impact of clinical decision support systems (CDSS) [6] has demonstrated statistically significant improvements in antibiotic-associated MEs or adverse drug events and an improvement in theophylline-associated MEs, while several studies have shown non-significant results. CDSS is worthy of future study in the adult ICU in order to build on the experience gained from the limited CDSS system used in a mixed ICU and general ward setting [19].