Professor of Clinical Pharmacy
Department of Clinical Pharmacy & Toxicology
Leiden University Medical Centre
The landmark report To Err is Human published by the United States Institute of Medicine in 1999 initiated considerable interest in the issue of medication safety, not only in the USA but also in Europe. It made people, both healthcare workers and those not affiliated with healthcare, realise that the morbidity and mortality caused by medical errors, and especially medication errors, is comparable to that caused by traffic accidents.
Attitudes have shifted away from blaming the person who made an error towards a more systematic approach. The question is now not “Why did this person make an error?” but “How could the system let this error occur?” This approach encourages the performance of risk assessments aimed at identifying specific weaknesses in the entire medication distribution chain – from prescribing, through dispensing, to drug administration. A number of studies, essentially from the USA, have shown that both prescription and drug administration are relatively weak links in the chain, in terms of the frequency and seriousness of errors. Therefore, it seems to be a rational approach to improve the processes of prescribing and administration.
Computerised physician order entry
In April 2000, a statement was made by the US Institute of Safe Medication Practices setting out a goal that handwritten prescriptions be eliminated within three years. However, recent surveys indicate that the penetration of computerised physician order entry (CPOE) systems in US hospitals is still only 5%. Such a goal was not set for European hospitals, and it is unknown how widespread the penetration of CPOE systems actually is. Nevertheless, it is recognised that the application of CPOE systems is gaining more interest, and several institutions are committed to implementing such systems in the near future.
Features of current systems
Essentially there are two basic procedures for the application of CPOE systems. In many hospitals medication orders are entered into a computerised system in a centralised manner. The physician writes the medication order manually, and the orders are translated, interpreted and entered in the pharmacy computer system by pharmacists or technicians. Medication checks usually take place, but signalling and interpretation are not performed during prescribing. In contrast, with a decentralised CPOE system, the physician enters the medication order into the computer system, and safety alerts are generated during prescribing, which then need to be interpreted by the prescriber.
These methods rely on different philosophies and role patterns. The first system does not facilitate the pharmacotherapy decision-making process, nor does it act as an order management system. Moreover, it allows errors in transcription and interpretation to occur. In the second approach, with responses during prescribing, the role of the pharmacist is to advise the physician on the clinical relevance of alerts rather than correcting the physician’s prescription. This may be considered a more professional role. On the other hand, pharmacists are better equipped to interpret safety alerts, and the usually high number of safety alerts generated by CPOE systems may induce a less critical proceeding in the prescriber.
A weak point of most current CPOE systems is their low level of integration with other information systems. Linkage of the CPOE system with other components of the hospital information system, such as the biochemical laboratory (eg, organ function, blood levels), the microbiological laboratory (eg, identity and drug sensitivity of microorganisms), diagnosis or – in the future – pharmacogenetics, is of great importance. These data not only may help to judge the clinical relevance of medication safety alerts, but are also helpful in monitoring the effectiveness of the therapy. Future developments may also include linkage of CPOE systems with the total patient data monitoring system. In such a system, individual patient parameters such as blood pressure or blood glucose levels may be archived and help to monitor the efficacy of drug treatment in a structured manner.
The application of a CPOE system as an order management system could have interesting logistic and quality advantages. For example, medication orders could be placed automatically by the central pharmacy of the institute, while orders (and preparation protocols) could be sent to the centralised cytostatic drug compounding department, the IV room or the satellite pharmacy. Also, orders for blood level monitoring of drugs and biochemical markers could be placed automatically and in a protocolised manner at the laboratories. It would also be useful to link up with the outpatient clinic and with computer systems of family practitioners and retail pharmacies.
Barriers for implementation
The major barrier to implementation of a CPOE system in a clinic is its user-friendliness. The opinion of most physicians is that using a CPOE leads to an increase in the time necessary for prescribing medications.
The use of a Windows-based program and the application of standard medication orders and sets of orders (eg, protocols for chemotherapy) may greatly improve the acceptance of such a system. Also, the accessibility of the drug database, limitation of safety alerts to those that are clinical relevant, and integration with other systems in the hospital may improve the ease of its use. Acceptance of a CPOE system is critical to its success, and involvement of clinicians at the developmental stage of a system and during implementation is considered essential. Moreover, performing a user satisfaction survey after implementation may not only be helpful for evaluation but also generate important information for system improvement.
Implementation of CPOE systems has the potential to contribute considerably to the improvement of medication safety in hospitals. The systems currently in use can be further optimised (eg, by supporting pharmacotherapeutic decision-making), but the impact of these activities on the improvement of the safety chain as a whole remains unclear and could be limited. Probably, optimisation of another part of the drug distribution chain is a more rational approach for making a second big step forward.
With regard to this aspect, the availability of systems for digital patient data filing, equipped with barcode-assisted drug administration that can be used by nurses when administering drugs to patients, is a promising future development.