The introduction of computerised physician order entry and computerised decision support reduced prescribing errors in adult and paediatric intensive care units at the University Hospital of Ghent
Pieter De Cock
Head of the Pharmacy
Literature strategies to prevent medication errors suggest the introduction of a ward-based clinical pharmacist, the implementation of information technology such as computerised physician order entry (CPOE) with clinical decision support systems (CDSS) and computer-generated dosing alerts or even the combination of both: pharmacist-to-dose computerised request.
When focusing on the implementation of a CPOE system at the intensive care unit (ICU), a comparison of the medication errors between paper-based prescribing and the newly introduced computerised system is useful, not only because these patients are very vulnerable and have a high number of drug therapies in a complex environment but also because every CPOE system is different.
Experience with CPOE at Ghent University Hospital
In our centre, the CPOE system that is used is part of a commercially available intensive care information system (ICIS) (Centricity Critical Care Clinisoft, GE healthcare) with full connections to monitors, ventilators and syringe pumps. Nurses have to register every drug administration and start and stop the time of continuous infusion in a separate ICIS chart module.
For adults, specific clinical decision support (CDS) features were developed: medication order sets for specific patient groups eg, liver transplant patients, predefined prescriptions for patients with significant renal and/or liver insufficiency. Physicians are also notified about clinically significant drug-drug interactions and possibly life-threatening adverse drug reactions (ADRs) (eg, QT- interval changes with erythromycin) at the time of prescribing. Allergy status of the patient is shown by means of a highlighted icon.
Before going live on the paediatric ICU, an adapted version of ICIS for paediatric use was designed. CDS within this paediatric ICIS version consists of an obligation for preselection of the final drug dilution, drug prescriptions per age category with predefined dose (in mg/kg) and interval and an automated drug dose/infusion rate calculation. Continuous fluids, electrolytes and parenteral nutrition components are prescribed in a separate module without predefined prescriptions or dose calculation. Hereunder, we describe our experience with the use of this CPOE system at the adult and paediatric ICU.
Experience at the Adult intensive care unit, Ghent University Hospital
A prospective, controlled trial was conducted at the adult surgical ICU in two paper-based units (PB-Us; total of 14 beds) versus one computerised unit (C-U; 8 beds), 10 months after implementation of the ICIS in the latter unit in 2003. The primary outcome measure was the difference in incidence and severity of medication prescription errors (MPEs) in the C-U versus the PB-U.
During the five week study period 160 patient days divided equally over both C-U and PB-Us and 90 different patients were analysed. Analysis meant review and categorisation of the MPEs into severity levels (Table 1) by a unit-blinded board of intensive care specialists and a clinical pharmacist (not involved in the registration part). Of the 2,510 evaluated medication and fluid prescriptions, comprising 1,286 in the C-U and 1,224 in the PB-U, 44 MPEs occurred in the C-U versus 331 in the PB-U (3.4% versus 7.0%, P <0.001). Overall, the ICIS resulted in a relative reduction of 86.7% for all types of errors associated with medication ordering.
In the C-U, the minor MPEs were mainly wrong pharmaceutical form errors and infusion rate errors. The intercepted MPEs particularly involved double prescriptions, but also problems with trailed zeros (for example, aspirin 3g instead of 0.3g), and problems with continuous infusion prescriptions (for example, propofol or remifentanil infusion being still activated two days post extubation). The non-intercepted potential ADEs were mainly dosing errors or incompleteness of low molecular weight heparin prescriptions. There were only two ADEs (0.15 per 100 orders) that occurred in the C-U (an antibiotic overdose and a vasopressin infusion overdose causing cardiac ischaemia).
In the PB-U, minor MPEs were mainly due to illegible writing, incomplete orders, or abbreviations. The intercepted MPEs were mostly errors of negligence (for example, wrong route of administration) or transcription errors. The ADEs (1 per 100 orders) were mainly dosing errors (for example for antibiotics and anti-epileptic drugs). These findings and examples confirm
the different profile of errors (double prescriptions, missing commas and wrong drug selections versus illegible orders) in a newly used electronic system. Nevertheless, an overall reduction of errors is a fact and the pharmacists’ effort, clarifying and correcting orders (eg, missing or illegible information or abbreviations merely due to “performance deficits”) is reduced to 0%.
Moreover for patients with renal failure, a three-fold significant reduction of serious MPEs in the C-U versus the PB-U was observed. In the PB-U, 91% of these serious MPEs were due to dosing errors, which is significantly higher than the proportion of dosing errors in the C-U (41%, P <0.001).
Experience at the Paediatric intensive care unit, Ghent University Hospital
A comparable trial was conducted at the paediatric ICU (6 bed unit) before (2007) and after implementation (2009) of the adapted paediatric version of ICIS in combination with a web-based drug information database (information on maximum dose, clinically significant drug-drug interactions, medication protocols and drug administration).
During a 4-week period before and after implementation, medication errors were registered (results not yet published). MPEs were detected through chart review. Primary outcome measures included the incidence and severity of medication errors. Severity of errors was evaluated by a clinical pharmacist, independent from the ward and familiar with medication error evaluation, using the same classification as mentioned above (Table 1). Drug prescriptions were evaluated in 86 patients (37 patients before implementation of CPOE; 49 patients after implementation of CPOE/115 patient days before implementation of CPOE; 208 patient days after implementation of CPOE). Age distribution and length of stay were not significantly different between the two populations.
In summary, preliminary results show that implementation of CPOE resulted in a relative MPE reduction of 32.6% (p=0.001). Overall, also a reduction in severity of MPE was notified after ICIS implementation (p<0.05) with a significant decrease in potential ADEs (p=0.02; 41 before CPOE implementation versus 0 after CPOE implementation).
Main categories of MPE before implementation of CPOE included: illegible or incomplete orders (minor MPE) and dosing errors (potential ADE and ADE). These results are consistent with the results of the study at our adult intensive care unit as described above.
After CPOE implementation, dosing errors were reduced (p<0.001) and the remaining were almost completely due to incorrect infusion rates of electrolytes and total parenteral nutrition components. This is possibly due to the fact that prescriptions for these agents cannot be predefined in the ICIS system and orders have to be renewed every day. Misselection of age category was another reason for dosing errors and can be classified as a newly CPOE-introduced medication error. Another CPOE-associated prescription error sometimes occurred when an intravenous to oral switch was conducted. In ICIS, every order automatically starts on the time of input and sometimes it was forgotten to lengthen the dosing interval. This resulted in the first orally administered dose given too early in respect to the last intravenous dose.
Introduction of CPOE with a certain degree of CDS significantly reduced incidence and severity of MPE at our adult and paediatric ICU. As described by van Rosse et al several investigations in the intensive care environment came to the same conclusions. The actual effect of CPOE on avoidance of adverse drug events at the ICU is less clear but this is possibly due to an insufficient power of most studies to measure a difference in ADE incidence (estimated frequency of ADE occurrence in hospitalised patients: 5-7/100 hospitalisations).[7,8]
Infants and children admitted to the ICU are even more prone to medication errors than adults because most drugs are prescribed in mg/kg with a risk of exceeding the maximum (adult) dose, changing drug and nutritional requirements per age category and limited line access. To our experience, it is therefore advisable to adopt information technology systems for this specific patient group to assure and enhance paediatric
On the other hand the introduction of new types of medication errors with CPOE is a matter of concern in our studies. This was already described by several previous investigators and highlights the need for stringent evaluation of every single CPOE system. Furthermore, more trials in the complex ICU setting are warranted in an adequate controlled design. Finally, the use of uniform terminology for medication errors, (potential) adverse drug events and severity grades among the different new trials is to be advised to allow adequate comparison of trial results.
1. Leape LL et al. JAMA 1999;282(3):267-270.
2. Kaushal R et al. Arch Intern Med 2003;163(12):1409-1416.
3. Vincent WR et al. J Am Med Inform Assoc 2008.
4. Colpaert K et al. Best Pract Res Clin Anaesthesiol
5. Colpaert K et al. Critical Care 2006;10(1).
6. van RF et al. Pediatrics 2009;123(4):1184-1190.
7. Zegers M et al. Qual Saf Health Care 2009;18(4):297-302.
8. Bates DW et al. JAMA 1995;274(1):29-34.
9. Spencer DC et al. Am J Health Syst Pharm 2005;62(4):416-419