Impact of electronic prescribing on patient safety in hospitals: implications for the UK

Introduction

Government policy is increasingly promoting the use of technology in healthcare. For example, the UK government has launched several initiatives and funding streams to drive technology use within the NHS^[1],
[2]
. Similarly, both the US government and the European Commission have recognised the potential for healthcare information technology (HIT) and encouraged its meaningful use^[3],
[4]
. Despite worldwide advances in HIT, implementation of hospital electronic prescribing (EP) is lagging behind in many countries. In the UK, although the use of EP is ubiquitous in primary care^[5],
[6]
, deployment in secondary care remains slow and patchy^[6],
[7]
. Of 101 English acute trusts responding to a national survey of EP systems, more than two thirds (70 trusts; 69%) had at least one form of EP in use at the time of the survey, with over half of these having more than one system (39 trusts; 56%)^[7]
. However, EP was often used in limited clinical areas and for limited types of prescribing. This may be because processes and the level of care provided are far more complex in secondary care^[6]
. The low uptake is comparable to many other European countries. A study of the use of EP in seven different countries, including the UK, reported that despite the variations in healthcare systems, both the uptake of EP and the extent of systems’ interoperability were low in all seven locations^[8]
. Hospital-wide EP was reported to be least widely used in Germany, France and the UK^[8]
.

While there is no universally agreed definition for EP, definitions usually refer to the ordering or prescribing of medication electronically. For instance, the US eHealth initiative defined EP as “the use of computing devices to enter, modify, review, and output or communicate, drug prescriptions”^[9]
. NHS Connecting for Health defined EP as “the utilisation of electronic systems to facilitate and enhance the communication of a prescription or medicine order, aiding the choice, administration and supply of a medicine through knowledge and decision support and providing a robust audit trail for the entire medicines use process”^[10]
. In the United States, the term ‘computerised provider (or physician, or prescriber) order entry (CPOE)’ tends to be used instead of ‘EP’ in the hospital setting, where the scope of CPOE may include other types of medical orders such as laboratory tests and radiology^[11]
.

Hospital EP systems are therefore far more complex than those used in primary care settings. They can include some, or all, of a range of different functions, and may be implemented in a wide range of organisational contexts and involve different healthcare professionals at different points of care. A report published in 2009 suggested that the UK hospital EP software market comprised four major types of systems^[5]
:

• Pharmacy-based systems (systems emanating from providers of pharmacy stock control solutions);
• Clinical specialty-based systems (e.g. cancer systems, renal medicine and intensive care systems);
• Components or modules of larger hospital information systems (such as part of an integrated electronic health records solution);
• Home-grown software (systems developed locally by in-house informatics teams).

In 2014, a specific classification of commercial EP systems used in UK hospitals was proposed^[12]
. Two broad categories of commercial systems were identified: bespoke systems and packaged applications. Bespoke systems were defined as systems designed to meet particular needs of a single organisation, while packaged applications were standard systems designed to meet requirements of different hospitals, which may then be configured to meet certain requirements of a particular hospital. The authors further divided packaged applications into four subcategories: standalone systems, modules within an integrated hospital information system, speciality systems, and functionalities spread over several modules^[12]
.

EP and CPOE systems vary in functionalities, both those that are possible within the system itself and those that are configured for use in a given organisation. Therefore, EP can range from simple systems that allow basic prescribing to advanced systems that are integrated with clinical (computerised) decision support systems (CDSS) of varying sophistication^[13]
. It has been argued that the greatest benefits are associated with integrated EP systems because they offer potential improvements in communication among patients, prescribers, pharmacists and other stakeholders involved in medicines management^[14]
. Hospital EP is an important item on the UK government health policy agenda. English hospitals are expected to be paperless by 2020^[15]
and therefore the deployment of EP systems is expected to rise rapidly in the near future.

This review aims to summarise the available evidence on the impact of EP on patient safety in the inpatient setting, with a focus on implications for the UK. The evidence for the effects of EP on medication errors, adverse drug events (ADEs), workflow, and healthcare professional communication is also discussed. Potential unintended consequences of EP are highlighted, together with how they can be identified and mitigated. The article concludes with considerations of the evolution of EP in hospitals, especially in relation to advances in HIT and interoperability, and how these developments may be used to benefit patient safety.

Sources and selection criteria

The authors aimed to identify systematic and narrative reviews, including reviews of reviews, which examined the effects of EP on the safety of the medicines management process in the inpatient setting. A literature search was conducted using PubMed/MEDLINE and EMBASE for articles published between January 2000 and December 2015. The search terms used were ‘electronic prescribing’, or ‘CPOE’, or ‘computerised provider order entry’, or ‘computerized provider order entry’. For inclusion, reviews had to be published in English and focus on EP or CPOE use in the hospital inpatient setting. Reviews focused on settings other than inpatients and those limited to a specific patient population, such as paediatrics or critical care patients, were excluded. Reviews that examined evidence for the effects of EP on the following outcomes were included: medication errors, ADEs, workflow, healthcare professional communication, and unintended consequences of EP systems use. Inclusion and exclusion criteria are detailed in ‘Table 1: Inclusion and exclusion criteria’.

Discussion

The search yielded 879 potential publications, of which 103 were duplicates. Initial review of title and abstract resulted in 41 publications for full text review, of which 31 were subsequently excluded (see figure 1 for a depiction of the screening and selection process).

Of the ten reviews that were included^{[16],
[17],
[18],
[19],
[20],
[21],
[22],
[23],
[24],
[25]}
, nine evaluated the effects on error and/or ADE rates^{[16],
[17],
[18],
[19],
[20],
[21],
[22],
[24],
[25]}
, two reported effects on workflow^[22],
[23]
, and one reported unintended consequences of systems use^[22]
. The majority of the original studies included in the reviews identified were from the United States; only a few original studies were conducted in the UK. Details of all ten included reviews are summarised in ‘Table 2: Summary of reviews that examined the evidence for the effects of EP/CPOE with, or without, CDSS on the safety of the medicines management process’.

Medication errors and ADEs

Nine reviews examined the effects on error rates and/or ADEs and suggest that inpatient EP use is associated with benefits in reduced medication errors and, to a lesser extent, ADEs. However, the evidence is limited by the reviews’ varied inclusion and exclusion criteria, variation in definitions of errors and ADEs, relatively weak study designs of the included studies, and a lack of contextual data about the systems implemented. Included studies were also heterogeneous in design, setting and methods of evaluation. Moreover, most systematic reviews were based on a preponderance of US studies, with only four UK studies reported in six papers^{[26],
[27],
[28],
[29],
[30],
[31]}
(see ‘Table 3: Characteristics of the studies that evaluated the introduction of EP, in comparison with paper medication orders, in UK hospitals’).

All four UK research studies identified in the reviews evaluated pilot introductions of EP systems in specific clinical areas, and revealed mixed findings in relation to the effect on medication errors. All used uncontrolled before and after study designs. Evans et al. conducted a study in an intensive care unit (ICU) to compare the accuracy, completeness and time requirements of EP versus handwritten prescriptions^[26]
. They found that although electronic prescriptions were more complete (100% vs. 47% respectively), errors in prescribing were not reduced by EP. The percentages of correct medication orders for intravenous fluids, intravenous infusions and intermittent IV drugs were 64%, 47.5% and 90% for handwritten prescriptions, compared with 48%, 32% and 90% for electronic charts, respectively.

Mitchell et al. evaluated a pilot EP and electronic medication administration record (eMAR) system in surgical wards, operating theatres and the recovery ward of a teaching hospital^[27]
. A total of 4,927 computerised prescriptions were written during the pilot. An audit of handwritten medication administration records identified significant omissions of patient and drug information. Omissions were less frequent with eMAR, with all mandated fields being 100% complete. An audit of electronic drug orders identified 143 errors, giving an error rate of 2.9% of medication orders. The highest error rates were seen in the first week of the project (6.4%), half of which were caused by the selection of the wrong formulation of the required drug. The authors identified errors specific to the use of EP/eMAR in 1.2% of electronic orders (57 of 4,927).

Another UK study compared the impact of EP (without decision support) with handwritten prescribing on the frequency, type and outcome of medication errors in an ICU^[28]
. The authors found that medication errors were significantly lower with EP (117 errors in 2,429 prescriptions; 4.8%) than with handwritten prescriptions (69 errors in 1,036 prescriptions, 6.7%; P <0.04). The prevalence of errors reduced with time following EP introduction (P <0.001).

Finally, in a study conducted in a surgical ward of a UK teaching hospital, researchers assessed the impact of a closed-loop EP and automated dispensing system on prescribing errors, administration errors and staff time^{[29],
[30],
[31]}
. The system reduced prescribing errors (3.8% of medication orders pre-intervention and 2.0% after intervention, P <0.001), medication administration errors (7.0% pre-intervention and 4.3% after intervention, P =0.005), and increased confirmation of patient identity before administration.

Effect on workflow and healthcare professionals’ communications

Two reviews examined effects on workflow^[22],
[23]
and one examined effects on time^[25]
. One UK study reported effects on workflow^[26]
and found that the mean time taken to prescribe one medication order was 20 seconds for medication orders and 55 seconds for EP^[26]
. There were no data about communication with patients or their involvement in the medicines management process.

Findings of the identified reviews suggest that EP is associated with significant workflow changes but the effects of these changes are mixed. It is therefore hard to reach a definitive conclusion as to whether these changes result in an overall positive or negative effect on patient care. The reviews also indicate that although order entry times seem to be increased by computerisation, the overall time spent on care may be similar or reduced. This evidence is, however, extrapolated from international literature, predominantly US studies where medication-related workflow is different to the UK. Moreover, many studies looked at subsets of clinical workflow and not the overall workflow across a whole day. Another major limitation is the lack of studies looking at medicines-related workflows specifically, as US CPOE systems are generally used for ordering more than just medicines.

Unintended consequences of EP and how can they be mitigated

One review examined unintended consequences and new safety issues of EP use, and concluded that they limit the system’s safety benefits^[22]
. However, there was no UK evidence included. The authors also highlighted that there is no standardised approach to avert alert fatigue^[22]
.

Overall consequences of changing from paper to EP

International evidence shows that EP may improve the safety of inpatient medicines management processes, reduce medication errors and, to a lesser extent, reduce ADEs. However, unintended consequences, including new errors, may occur. Evidence on the effects of EP on workflow is limited as most studies evaluated CPOE systems in general and did not specifically explore medicine-related workflow. The application of this evidence to practice in the UK is difficult as most of the existing evidence is from the United States where prescribing and administration practices, and workflow, are very different to those in the UK^[32]
. For instance, there are differences in medication documentation, labelling and supply practices, as well as variation in policies governing patients’ own medication use^[32]
. Therefore, the benefits and negative consequences of changing from paper to EP may be different in the UK setting.

Implications for practice

There are potential patient safety benefits from EP use, but the realisation of these benefits is dependent on successful implementation and utilisation. EP adoption results in significant changes to existing practice^[24]
. Healthcare representatives from all clinical areas should therefore be a part of the EP adoption and evaluation team to include all perspectives and ensure that the new system meets the hospital’s specific needs^[5],
[33]
. Building the right teams and equipping staff with project management and IT skills are important for success^[5]
.

As illustrated earlier, most studies in this area have been conducted in the United States. While robust UK studies are under way^[34]
, it is important to capture the impact before, during and after a wide range of EP implementations wherever possible, to provide further UK evidence.

While there is much promotion of the benefits of EP, various unintended negative consequences of EP systems have been described in the literature^{[35],
[36],
[37]}
. Clinical unintended adverse consequences resulting from CPOE implementation have been classified into nine types^[38]
. These comprise: more or new work for clinicians, unfavourable workflow, ‘never-ending system demands’, problems related to paper persistence, untoward changes in communication patterns and practices, negative emotions, generation of new kinds of errors, unexpected changes in the power structure, and overdependence on the technology^[38]
.

New medication errors can occur when prescribers pick from a drug list (drop-down menu) or while filling free-text fields in an electronic prescription. This could potentially be minimised by reducing the size of drop-down lists, minimising free-text prescribing, and building well-designed, pre-defined ‘order sentences’ and ‘care sets’ into the system. Order sentences are defined as ‘a complete pre-written medication order that includes dose, dose form (when necessary), route of administration, frequency and, if appropriate, a PRN (pro re nata) flag and reason’^[39]
. Care sets allow clinicians to launch a combination of related orders (e.g. medication, laboratory tests and/or other examinations) for a specific clinical situation.

Another consequence of CPOE that may contribute to errors is overriding alerts, especially those warning of severe drug interactions or significant allergies. Alert fatigue is well documented in the literature^{[40],
[41],
[42]}
. There is no standardised approach to avert alert fatigue^[22]
, although tailoring alerts has been suggested as a solution^[43],
[44]
. The integration of patient, illness and medicine information, as well as the development of an alert hierarchy to generate, at most, one clinically relevant alert per prescription, have been suggested as measures to reduce alert fatigue^[45]
.

Future opportunities for digitalised health in the UK

Integration and interoperability of healthcare

Widespread use of EP in hospitals, particularly in the context of integrated systems, offers an opportunity to improve patient safety and quality of care. First, the rise of integrated approaches to the delivery of healthcare, and HIT interoperability, may enhance care coordination through improved access and exchange of clinical data^[46]
. Increased access to medication-related information by healthcare professionals and patients, as well as increased access to clinical data at the point of prescribing, would be expected to improve efficiency and reduce clinical risks^[47]
. Interoperability may also offer an opportunity for improved exchange of clinical data between primary and secondary care. Second, there is potential for healthcare providers and patients to assume different roles in the delivery of healthcare. Connectivity could lead to time savings^[47]
and thus provide healthcare professionals with an opportunity to focus on more clinical roles. Moreover, integration and interoperability supports team-based care approaches. Similar to healthcare providers, patients will be able to access and communicate information with their caregivers efficiently, which may support patient involvement in their own safety^[48]
. Finally, interoperability may help monitor policies both locally and nationally, and facilitate secondary use of data for audit and feedback, which may contribute to improved patient care^[49]
.

Inpatients’ involvement with their medication in the context of EP

It is increasingly recognised that it is essential to involve patients with their medication^[50],
[51]
. Recognition has also been given to the importance of patient involvement in enhancing safety^[51],
[52]
. Such patient involvement can increase satisfaction, improve health outcomes and reduce the likelihood of avoidable harm^[53]
. Patient safety activities relating to inpatient medication include — but are not limited to — viewing their inpatient medication records, prompting staff to avoid dose omissions, providing information to aid handover between shifts and professional groups, and raising queries with prescribers, pharmacists or nursing staff.

While EP has the potential to facilitate patient and carer involvement with inpatient medication — for example, by the creation of patient-friendly interfaces — research in this area suggests that it currently acts as a barrier^{[54],
[55],
[56]}
. Few EP systems include a patient interface^[55]
and qualitative studies^[54],
[56]
suggest that patients have less access to their electronic medication records because a healthcare professional login is required. In addition, where electronic records are not available at a patient’s bedside, conversations about the patient’s medication appear to be less likely to happen in front of them and healthcare professionals may be unable to answer their queries^[54],
[56]
.

Limitations of this review

The objective of this review is to provide a broad overview, and therefore, a simple search strategy was employed. Some relevant reviews and individual UK studies may have been missed because of limitations in the search strategy; for instance, the use of other search terms may have yielded more results. Papers where the full text was not obtainable or not written in English were excluded and a search of the grey literature was not conducted. The search strategy was conducted by one researcher only and the quality of the included reviews was not assessed.

Conclusions

Safety in hospitals could be improved by EP use. However, the extent to which the existing evidence is applicable to UK settings is not yet clear. Moreover, benefits are likely to be dependent on how systems are implemented and used in practice. Further evaluations of the effects of EP systems on the safety of the medicines management process in UK hospitals are required to inform both policymakers and end users. There may be great potential to improve patient safety, and quality of care, through greater integration and interoperability of HIT in the UK. This could be achieved through improved access and exchange of clinical data, and supporting a team-based care approach. EP should ideally be used to facilitate patient and carer involvement with inpatient medication, rather than acting as a barrier.