1. Introduction
In the Netherlands, every year over 10% of all citizens are admitted to a hospital (www.cbs.nl; accessed April 2013). Pre- scription of one or more drugs is an important aspect of almost all hospital treatment regimens. Medication-related processes (prescribing, ordering, preparation, dispensing and administration) thus can be considered as regular activities affecting a substantial part of the hospitalized population. During this process, errors can be made. Medication errors (MEs) are reported to occur in a mean of 5.7% (range 0–49%) of all episodes of medication administration [1]. Prescription errors form a considerable part of these errors. A recent study showed that in 43.8% of prescriptions for hospitalized patients one or more MEs occurred [2]. Medication errors can lead to adverse drug events (ADEs), which are generally considered to be preventable (pADEs). Studies report an equally wide range for ADEs, namely 2–52 ADEs per 100 patients [3]. In turn ADEs may lead to excess length of stay, reported to vary from 2.9 days [4] and 3.2 days [5] to 6.2 days [6] in recent studies. The costs associated with ADEs in these studies were 970 euro, 3420 US dollar and 2507 euro, respectively [4–6].
Thus, every possible improvement in medication-related processes is desirable with regard to health gains, patient well- being and safety. One of the possibilities is to improve the medication prescribing process. Traditionally, medication pre- scribing systems are paper based, which rely on hand-written prescribing and transcribing activities of various healthcare professionals. Computerized physician order entry (CPOE) sys- tems and CPOE with a Clinical Decision Support System (CDSS) are considered to be a useful alternative to enhance patient safety. These systems are becoming increasingly com- mon in the Netherlands and worldwide and a growing body of knowledge on their clinical effectiveness is presented in (inter)national journals. Recently, a review on the quality of systematic reviews on the impact of CPOE on clinical out- comes was carried out [7]. Of the more recent reviews with the highest quality, Shamliyan et al. [8] reported a signifi- cant impact of CPOE on medication errors in most studies. Heterogeneity prevented the construction of pooled relative risk estimates. Another review with relatively high quality looked at both MEs and pADEs and concluded that electronic