The search identified 1477 articles. Following deduplication and exclusion of four non-English papers, 1301 articles underwent title and abstract screening. The titles and abstracts for 130 of these were screened by the second researcher (MM). Based on title and abstract review, 171 articles underwent full-text screening; of these, 34 were also screened by TM. The inter-reviewer agreement was deemed almost perfect for the title and abstract screening (Cohen’s kappa = 0.948) and moderate for the full-text screening (Cohen’s kappa = 0.637, 9). Six papers were reviewed by BDF and their inclusion or exclusion agreed by consensus. The screening process is summarised in Fig. 1. Twenty-five studies met the inclusion criteria. The full data extraction table can be found in Additional file 2: Table S2. The MMAT quality assessment was completed by two reviewers for 13 studies, with strong inter-reviewer agreement (Cohen’s kappa = 0.816). Once all the studies were assessed, ten were rated 100%, four as 75%, ten as 50%, and one as 25% (see Additional file 3: Table S3). Fourteen of 25 studies were therefore deemed high quality (75–100% score).
Of the 25 studies, seven were from the UK [10,11,12,13,14,15,16], four from the US [17,18,19,20], four from the Netherlands [21,22,23,24], three from France [25,26,27], two from Australia [28, 29], two from Saudi Arabia [30, 31] and one each from Denmark [32], Spain [33] and Iran [34]. Nineteen studied commercial systems, one a home-grown system [25], one both a commercial and a home grown [15] system and for four studies it was not possible to establish the system type [11, 30, 31, 34]. Nineteen studies referred to their electronic systems as CPOE, and among these, three specified that the system was for prescribing medication only [22,23,24]. The remaining six papers studied ePMA systems, although in two cases these were referred to as eP [12, 14]. Sixteen papers studied CPOE systems with electronic medication administration and three studied CPOE without electronic medication administration. The six papers that studied ePMA systems included one exploring a mix of ePMA systems and one that studied a standalone eP system. The included studies used a range of data collection methods and study designs, mainly cross-sectional. Most applied quantitative methods (n = 14) such as surveys, ten applied qualitative approaches including focus groups, interviews and observations, and one used mixed-methods [26]. Across the 25 studies, nurses were included in 18, doctors in 17 and pharmacists in 9 studies. Four key themes were derived from the studies’ findings: communication, time taken to complete tasks, clinical workflow, and workarounds.
Communication
Twelve papers highlighted the impact of eP systems on HCPs’ communication among professions. Two reported a positive impact on HCPs [17, 20], two reported no significant difference [27, 29], three reported a negative impact [22, 23, 26] and five reported a preference for verbal communication over electronic [11, 15, 21, 25, 31]. Two specifically reported a positive impact on doctor-nurse communication since introduction of eP [17, 20]. In one of these, interviewed doctors perceived that communication with colleagues and nurses improved through better documentation [17]. Similarly, in the second study, nurses reported adequate communication with doctors when using eP [20]. Furthermore, in this study and another qualitative study it was found that communicating orders electronically risked miscommunication between HCPs as there were no bedside systems to enter medication orders [22, 23]. The doctor therefore had to rely on their memory or write a brief note on paper to remind them to prescribe medication later [22, 23]. In another study, it was reported that eP systems benefited the doctor-pharmacy and nurse-pharmacy workflows but hindered doctor-nurse workflows as the unidirectional nature of medical dominance in the ordering phase caused nurses difficulties in their workflow [23].
Two studies revealed that both medical and nursing staff preferred verbal communication rather than communication through an eP system [21, 31]; this view was further supported in interviews conducted with doctors and pharmacists [15]. Nurses reported that they always supplemented eP communication with a phone call to confirm medication orders, which they perceived to add to their workload [21, 31]. Three studies focused on the impact of eP on pharmacists and their communication with doctors [11, 15, 25]. It was found that pharmacists’ interventions were well accepted by doctors when communicated both electronically and orally [25], but with a significantly higher acceptance rate for those communicated orally. This study also suggested that pharmacists preferred oral communication in situations requiring a rapid modification to medication [25]. These two studies also indicated that there was an increase in communication between doctor and pharmacist as eP introduced a new ‘technical’ expert role for pharmacists [11, 15], suggested to have evolved due to suboptimal doctors’ training [15].
Two studies reported no significant impact of eP systems on HCPs’ communication [27, 29]. One identified common rounds, briefings and opportunistic exchanges as opportunities for medical and nursing staff to exchange patient-related information, and then compared the impact of these on communication between an eP site and a paper-based site; no statistically significant difference in cooperative activities was identified [27]. Similarly, a controlled before-and-after time and motion study found that an electronic system was not associated with any significant change in the proportion of time medical and nursing staff spent in professional communication with each other [29].
Time taken to complete tasks
Six papers focused on the impact of eP on time taken while completing particular medication-related tasks [13, 16, 18, 19, 28, 29]; the majority adopted uncontrolled before-and-after study designs (n = 4), one a controlled before-and-after design [29] and one was a longitudinal qualitative study [28]. Of the former, one focused on the impact of eP on nurses’ medication-related activities [16]. This study suggested that eP did not significantly affect the length of time spent on a medication administration round but altered the distribution of tasks with a doubling of the time spent on documentation [16]. The duration of the medication round appeared to decrease post-eP but data collection ceased before the reason for this change could be fully explored [16]. In another study in which HCPs were interviewed at four time points, doctors and nurses perceived that prescribing and medication administration took longer post-eP compared to paper medication charts. Six months post-eP, participants perceived that they had become more efficient in using the system but the time taken for medication administration had not returned to pre-eP durations as the process now included additional steps such as double signing for each dose administered [28].
One quantitative study explored the impact of a closed-loop eP system on staff time [13]. Nurses’ medication rounds took less time post-implementation but more time was required for medication-related tasks outside the medication round. Prescribing and pharmacists’ reviews took more time post-implementation. However, it was highlighted that since all medication charts were electronic, they were always accessible, so there was an increase in the number of charts available for pharmacists’ review, which could have contributed to the increase in time taken [13]. Conversely, another study found that time taken for a pharmacist to verify a medication order reduced compared to hand written orders [18]. This was because the eP system allowed some steps in the ordering process to be eliminated, contributing to time saved during the pharmacists’ review [18, 19]. Another study found that the time taken to communicate orders from the prescriber to the pharmacy and the time taken to dispense and administer medication to the patient improved [19]. In contrast, in a controlled study, the proportion of time taken for medical and nursing staff to complete medication-related tasks did not change relative to control wards [29].
Impact on clinical workflow
Three papers concluded that nurses perceived the introduction of eP to positively impact their workflow [30, 33, 34]. However, in one of these, it was reported that nurses who believed they received substandard training for eP were less satisfied with their workflow than those who perceived their training to be fair or good [30]. In another study, nurses rated their post-eP workload as good or very good in comparison to doctors who rated theirs as fair or poor [33].
Three papers suggested a negative impact on nursing workflow following implementation of eP [10, 24, 32]. For example, in a qualitative study, nurses reported being hesitant to adopt the system at the start, feared letting go of familiar aspects of their job and expressed resistance to computers becoming a more substantial part of their role [10]. In a questionnaire study, responses of nurses switching from two different paper-based processes to eP reported that they would prefer to continue using the eP system, although the study also suggested that nurses believed that the new system did not support their work processes [24]. Another study also reported that eP did not support a ‘collaborative working environment’, as doctors and nurses were less likely to negotiate and discuss patient treatments together [32].
Doctors saw advantages in having the ability to enter orders within and outside the hospital, allowing easy access to legible patient information, but perceived that entering electronic orders took more time compared to the paper-based system [12, 28]. These perceptions support previous research that highlighted a longer duration for medications to be prescribed electronically [13, 33]. Doctors also expressed their frustrations by describing a new eP system as being time-consuming, as it impacted their perceptions of the system’s suitability and usability [28]. The notion of becoming over-dependent on the technology was suggested, but doctors perceived that having access to information improved clinical decision making [33]. However, the extra steps needed to obtain the information from the system were seen as a burden and an increase in workload [17, 28, 33]. Doctors had more negative responses towards the eP system compared to nurses and pharmacists [12].
Two papers presented pharmacists’ perception of the impact of eP systems on their workflow [11, 14]. In a small UK study, hospital pharmacists all highlighted that more clinical screening was being completed away from the ward and in one hospital the role of pharmacy technicians had changed to become more ward-based before the roll out of eP to support maintaining medication stock and dispensing items for the ward [14]. Since the pharmacists were relieved of conducting these tasks post-eP, they had more clinical input on the wards by attending ward rounds [14]. Five of seven hospital pharmacists interviewed believed the amount of time pharmacists spent on the ward had not changed and four reported that pharmacists visited all the patients daily whether they had a wireless or fixed device system [14]. This contradicts the findings from another UK study that suggested pharmacists conducted their work away from the patient due to the lack of available computers in patient areas following the introduction of an electronic system [11]. As eP systems offer the flexibility to complete remote screening, pharmacists in both studies were concerned about reduced patient contact and denying patients opportunities to ask questions [11, 14]. Furthermore, during focus groups, pharmacists who had been using an eP system for 8 months reported that reduced patient contact had resulted in poorer relationships with patients [11]. Pharmacists at three hospitals reported their pharmacy workload had increased while their pharmacy workforce remained the same [14]. In most cases, only one extra staff member (pharmacist, technician or system manager) was recruited to help implement and support the system [14].
Workarounds
Two papers explored the introduction of workarounds in the context of eP [22, 28]. A number of workarounds were identified at each stage of the medication use process [22]. At the point of prescribing it was highlighted that often the computer terminal was not near the patient, thus, the review and prescribing of medication took place away from the patient and was reliant on the prescriber’s memory [22]. An example of a nursing workaround introduced following eP is nurses administering medication without an electronic prescription if the doctor was busy and not able to prescribe the medication at the patient’s bedside [22]. In this situation, the nurse would start to administer the medication based on the doctor’s verbal or paper-based order and either handwrite the order onto the medication record card (instead of affixing a label) or call the doctor to remind them to prescribe the medication [22]. In a paper-based environment, a handwritten order would satisfy the prescription requirements, but with the electronic system used in this study, in which nurses administered against paper records, additional steps were required to produce a valid prescription such as an electronic order and to print a prescription label for nurses to administer against. In a separate qualitative study, it was found that 6 months after eP implementation, workarounds were adopted to overcome limitations of slow computers. Nurses no longer took computers to the bedside and some nurses viewed this workaround to be less safe, as medication details and patient identification were no longer being checked immediately prior to medications being administered [28].