Implementation of Smart Infusion Pumps: A Scoping Review and Case Study Discussion of the Evidence of the Role of the Pharmacist

“Smart” infusion pumps include built in drug error reduction software which uses a drug library. Studies have reported the drug library build should be undertaken by a multidisciplinary team, including a pharmacist; however, the extent or nature of the input required by the pharmacist for greatest benefit is unknown. This review aimed to identify key factors for the implementation of the smart infusion pumps, with a focus on the role of pharmacists and compare this to the experience from a case study. A literature review was conducted using Embase and Ovid Medline, and 13 eligible papers were found. Predominant themes relating to the pharmacist’s role and successful implementation of the smart infusion pumps were determined. Key factors for success included team involvement across the entire process from procurement, set-up through to implementation including risk assessment and device distribution, and training, which were comparable to the case study experience. Few studies described the extent or details of the pharmacist’s responsibilities.


Introduction
The National Patient Safety Agency (NPSA) in England identified that incidents involving injectable medicines represented 62% of all reported incidents leading to death or severe harm [1]. Observational studies conducted at multiple hospitals within the United Kingdom (UK) and the United States (US) have reported error rates of up to approximately 60% [2,3], with recent studies reporting that only up to 1.1% of all the discrepancies and infusion errors witnessed were potentially harmful, and none would have prolonged hospital stay or resulted in long term harm [2,4].
Infusion pumps that control the administration of medication infusions at a set rate were introduced worldwide over 40 years ago. These have evolved vastly over time from manual control into sophisticated automated systems with in-built safety features and ability to interface with other electronic systems [5]. Safety features include a sensor for detecting air within the line; a battery fail safe system, which is a backup power system and ensures that the infusion pumps continue administering medication in the event of an electricity outage; pressure sensors which can detect pressure changes within the lines cause by issues such as blocked lines or veins or empty bags of medication. The newer, so called "smart" infusion pumps include built in drug error reduction software (DERS), which uses a bespoke drug library. DERS is considered a safety feature as it has the capacity to have pre-set limits for specific medications, thereby preventing unsafe infusion rates or errors in dose calculation [5].
Benefits of smart infusion pumps and the DERS are dependent on software, the drug library set-up and limits as well as uptake and use in practice [6]. The uptake of the smart pumps with incorporated

Method
A scoping literature review was conducted using Embase and Ovid Medline (all from inception to 13 July 2020). A review protocol was developed by the authors but not registered. The same search strategy was used for all the databases, using key words as follows: ("smart pump" OR "smart pumps" OR "intravenous smart pumps" OR "smart IV pump" OR "smart pump safety" OR "smart pump technology") AND ("drug libraries" OR "drug library" OR "drug error reduction software") AND ("pharmacist" OR "pharmacy") AND ("implement*" and "develop*"). Articles were restricted to the English language. All duplicates were removed.
Inclusion criteria were any original research studies, narrative papers or abstracts detailing or describing the process of implementation of the smart pumps or the process of creating the drug library. All other studies including those reporting the impact of the smart infusion pumps without details of implementation strategy or drug library set-up were excluded.
Titles and abstracts were screened against the eligibility criteria, and for those which may be eligible, the full texts were reviewed. References of included articles were also screened for other potentially relevant titles. All screening activities were carried out by the lead author, and descriptive information including the study objective, reported intervention, reported study outcomes, study setting, whether standardised concentration infusions were used and details of the healthcare professionals involved in the implementation and development of the drug library was collated into a table.
Themes that featured prominently in the included literature were identified using a deductive thematic approach and compared to the experience in a case study from the perspective of the pharmacy team at a tertiary care academic hospital. The implementation team at the case study site consisted of medical device experts, IT experts, nursing staff, consultants, finance department, education leads within nursing, the medication safety officer and pharmacists. The latter were tasked with the responsibility of creating a drug library that would be suitable for use across the entire organisation.

Results
A total of 44 papers were identified using the search criteria, which reduced to 26 when duplicates were removed and limited to "English language". Of these, thirteen papers met the eligibility criteria. The rest were not eligible as these did not specifically describe steps taken to build the drug library or describe the process of implementation of smart pumps. From screening the references of the included and excluded studies, no further studies were found.
All the included articles involved pharmacists in the development of the drug library (Table 1); nine involved nursing staff and ten had at least one physician involved with the development of the drug library. Four of the studies had implemented smart infusion pumps throughout the organisations [16][17][18][19]. Six of the studies had implemented smart infusion pumps only within paediatric intensive care or in a paediatric setting only [10,[20][21][22][23][24]. Five of the studies had developed drug libraries for other specific areas, including intensive care units, ambulatory settings, for home infusions, for epidurals only or for anti-infectives only [12,[25][26][27][28].
Key factors for success, as identified from themes of the literature, included team involvement across the entire process from procurement, set-up through to implementation including risk assessment and device distribution, and training. Each key theme is discussed further with reference to our case study.

Procurement
Two studies discussed the importance of ensuring that all the correct parties are involved in the procurement of the smart infusion pumps [20,27]. One study included representation from pharmacy, as well as clinicians, nursing, biomedical engineering, purchasing and human factor experts [27] and the other included nursing, pharmacy, clinical engineering and physicians [20]. This enabled all parties to contribute to the selection of the pump to be purchased, and to have an awareness of any strengths and limitations of the smart pumps, which could be factored into the drug library build [20,27]. This is similar to the case study, where procurement was led by the medical physics team and involved nurses and a medication safety pharmacist.
Our experience shows similarities to the literature. The implementation of the smart infusion pumps was led by the medical physics team during an organisation-wide project to replace the older infusion pumps, and all relevant parties, including nursing staff, physicians, the medication safety officer and pharmacists were invited to the procurement presentations during the tendering process. As with all technology, smart infusion pumps were noted to have some limitations; however, the limitations vary with different manufacturers. Knowledge of the limitations was of utmost importance, as this had a direct effect on how the drug library was built. Procurement of the smart infusion pumps was based on a scoring system and considered the potential impact of change from the infusion pumps already used (as recommended by end-users, mainly nursing staff), aesthetics of the pumps including size and display of infusion details to take into account human factors, as well as cost associated with the purchase of new pumps.

Risk Assessment
Two studies used a human factors approach prior to implementing the smart infusion pumps using models such as "failure mode and effect analysis" (FMEA) [21,28]. The identification of risk points in the different stages of implementation should be carried out by a multidisciplinary team, including pharmacists as demonstrated by Manrique et al. [21] and Wetterneck et al. [18]. This is an important step as appropriate action can be taken to manage the identified risks during the implementation stage. In our case, whilst a formal human factors assessment was not conducted at the case study site, a hazard workshop was conducted, and a clinical safety case developed [19].

Device Distribution
Unlike the case study where the intention was to introduce smart infusion devices across all clinical areas for all medicines, the majority of the studies described smart infusion pump implementation in specific units, such as intensive care settings where high-risk infusions are often administered or for a particular set of drugs, such as epidurals or anti-infectives [10,[20][21][22][23][24].
This required decisions about the drug library build and visibility. With approval and input from relevant clinical governance committees within the organisation, a single drug library was created, with each of the medication infusions configured to appear in either all care setting areas or in specific ones only. This approach was considered the most suitable one, due to the flow of the patients within the organisation who may have infusions in situ whilst being transferred from one care setting to the next-for example, from a theatre recovery setting to the critical care unit. Another benefit of this was that all care area entries could be accessed by any care area in case of any outlying or transitional patients-for example, requiring initiation of critical care infusions in a ward environment whilst awaiting transfer to a critical care unit or vice versa.
The approach and time taken varied considerably. For example, Howett et al. described the creation of a drug library was often the rate limiting step in the implementation of smart infusion pumps [22]. Manrique-Rodriguez et al. described creating the drug library for a paediatric intensive care unit with a multidisciplinary team consisting of a clinical pharmacist, a physician and the chief nurse of the unit over a period of seven months. Literature reviews were undertaken to determine standard concentrations and accurate limits for intravenous infusions for high-risk drugs [10,11]. Another group standardised the process for managing smart pump drug libraries within a specific geographical area by forming a working group comprising pharmacists, nurses and industrial engineers, with the main tasks being to evaluate the variability of different drug libraries being used and advice on standardised dosing limits, alerts, policies and best safety practices [17].
The case study site used a mixture of the two approaches. In order to ensure that the drug library was of a robust build, the strengths and limitations of the specific DERS were considered using a risk based approach and focussing on high-risk medicines. A multidisciplinary clinical advisory group, consisting of medical device experts, consultants, pharmacists and nursing staff from across the organisation provided input on the drug library development, taking into consideration any professional practice, governance and safety implications. Principles and processes were established to ensure a standardised approach for the drug library build and maintenance. These principles were applied to mitigate some of the foreseen errors.
The creation of the drug library required practices from across the organisation to be standardised to a certain extent-i.e., the same infusion should be administered in the same way in order to have the same limits for the drug for the limits to have meaning. In the creation of the drug library, all the different disciplines were approached and questions were raised when issues in practice were uncovered or when standardisation could not be achieved. For the latter specific drug library entries were made for the different uses, and careful consideration was applied when considering which care setting was selected to ensure mis-selection errors were minimised. Clinical pharmacists from differing specialties were asked to work within their multidisciplinary teams, consisting of nursing staff and clinicians from specialist areas to help standardise infusion practice across the organisation.
The pre-existing infusion pumps required the rate to be set manually in millilitres/hour (mL/h). The drug library set-up allowed the display to default to rate (mL/h) or dose (mg/kg/min). The clinical advisory group was consulted about this as it required a change in work practices and introduced a potential risk in misinterpreting the display value. The group advised that, for safety reasons, the dosing units and rate display should match the prescription as much as possible.
In our experience, the drug library creation took place over a period of approximately 18 months and was led by the medication safety team within the pharmacy, with input from a multidisciplinary clinical advisory group and oversight from the hospital medication safety committee. This was a longer time frame than was originally anticipated due to the standardisation process and checking/governance processes implemented each time the drug library was amended. None of the literature stated exactly what governance processes had been implemented for any of the sites that used drug libraries.

Training
The literature from the search did not mention whether training the end-users on how to use smart infusion pumps should involve a pharmacist or not. A quality improvement study showed that continuous education on the use of smart infusion pumps can increase the compliance of its use [29].
In the case study, a nurse was appointed to lead on education and training of the smart infusion pumps. In our opinion, it was imperative that the end-users were all trained appropriately as all workflows relating to setting up an infusion were to change due to the use of the drug library. Education of the end-users should be based on their speciality, as certain drugs may need more information being input than others, and re-education should be encouraged if needed. End-users should also have dedicated time in a "play" environment where they can experiment with the pumps using the drug library and ask questions that may arise. This may help reduce workarounds being used later on and will help to ensure that the pumps are being used in a standardised way. From our experience, a pharmacist should be involved in the training or be involved in setting up training tools as well as they have valuable knowledge specifically about the drug library and how that will work. Ensuring the correct training tools are used for each specific clinical setting can minimise errors after the pumps are implemented and can help keep patients safe.

Discussion
There are few publications that describe the key factors for implementation of smart infusion pumps or detail the composition of teams and the process for drug library set-up. However, the limited literature and our experience shows that involvement of a multidisciplinary group, including pharmacists, is essential from the purchasing process through to implementation to facilitate a clear strategy for drug library build and subsequent adoption and use.
The majority of the literature states that the drug library was created with a multidisciplinary team consisting of consultants, nursing staff and pharmacists. However, a pharmacist within in team was constant in all the included studies. Our experience was that the drug library build was led by the pharmacy that had overall responsibility for it and had input from different healthcare professionals. Pharmacists, as medicines experts, are best placed to create the drug library, as they have the working knowledge of how these medicines are used, the evidence behind their use, and the concentrations required. Advising on how medications should be prepared and administered is inherent to a pharmacist's role and therefore they have the necessary expertise and competencies.
Having input from end-users (nursing staff) gives important input into how the drug library may change the workflow for how to administer medications for certain medications. For some medications, the workflow may become more complicated, and for other medications the workflow may become easier. Consultant input from all specialities is also important, as the way the drug library may be set-up for a particular drug may require a change in how the drug is prescribed-e.g., instead of the rate (mL/h) being present within the prescription, the dose (mg/kg/min) may need to be prescribed. Furthermore, consultants were in a position to say when the drug may be used over the normal limits, and this helped to design dosing limits.
There is evidence that infusions should be standardised as much as possible. Our experience is that this is very difficult to do with certain drugs, as some units use a higher concentration due to the required response. Some of the drugs were standardised within our organisation; however, there was a wide variety of drugs that could be used in a number of concentrations depending on the clinical situation of the patient and the rate would greatly vary depending on the final volume to be infused. The lack of standardisation for the use of an opiate drug was one of the reasons that an incident occurred whereby a patient received an overdose of opioids. In this example, if the concentration had been standardised for use across the organisation, the concentration which was incorrectly input manually would not have been required, and the correct bolus dose would have been administered. Furthermore, standardised infusions across a geographical area or a nation would provide a safety feature in terms of the end-users being familiar with the concentrations and how to administer them. If standardisation did occur, the pharmaceutical industry may be inclined to form premade infusions at those concentrations meaning that the nursing staff would not need to manipulate the drugs during the reconstitution phase of infusion preparation to obtain required concentrations.
One paper specified how long the drug library took to create (7 months for a stand-alone paediatric intensive care unit) [10,11]. From our experience, it is essential to have realistic time expectations for a dedicated pharmacist to create a drug library, with collaboration from the multidisciplinary team. Creating a comprehensive robust drug library which may be useful in preventing errors will take time, and from our experience took more than a year. In the literature, one of the reasons why the compliance rates to using the drug library was low was due to the drug library not being updated frequently [29]. The time required for maintenance of the drug library should be thought about when putting in a financial bid to procure smart infusion pumps to ensure that the drug library is maintained, and therefore remains useful. Unfortunately, we were not able to measure the compliance of the use of drug libraries post implementation.
As with all introduction of new technology, some unintended consequences or errors may occur [13]. Studies showed that using smart infusion pumps alone will not prevent all infusion errors; however, using a closed loop system where the smart pumps and the electronic prescribing systems communicate with one another may prevent a larger proportion of errors [4,16,30]. This was demonstrated in a study by Gerhart et al. whereby the smart infusion pumps were used as an integrated system with the electronic health records, and demonstrated a high compliance rate of 97% and 782 significant errors were prevented [16]. Ohashi et al. also described some of the negative effects of implementing smart infusion pumps. Some of these were lower compliance rates of using smart pumps, soft alerts being over-ridden, not all errors being intercepted, or the possibility of using the wrong drug library [5]. A continuous quality improvement study showed that compliance to using the drug library varied between different clinical areas, and constant education, audits and drug library refinement after implementation led to increased compliance [29]. This all shows that the workflows for using the smart infusion pumps are vastly different from traditional pumps as these did not incorporate drug libraries.
We used a scoping review to gain an overview of the evidence of the role of the pharmacist in implementing smart infusion pumps. The literature search showed pharmacists were key stakeholders in the development of the drug library and in the implementation of the smart infusion pumps, as having a pharmacist during the development and implementation was a constant [10,12,[16][17][18][20][21][22][23][24][25][26][27][28]31].
The presumed reason for this being that pharmacists are recognised as medication experts and would be in the best position to create a drug library. Our experience has shown that pharmacists have a pivotal role in ensuring an accurate and safe drug library is built, which would be of utmost importance when it comes to patient safety. A limitation of the scoping review approach is that it cannot not provide answers to specific questions. For example, in this review the evidence of the exact role and impact of pharmacists' involvement in specific areas such as training cannot be synthesised.
The main learning points from the literature and our experience was that the drug library build should be of robust quality, and appropriate governance processes should be followed. A multidisciplinary team should be involved in all stages of the drug library build from procurement to actually testing and implementing the pumps. Infusion pumps have been used for years; however, drug libraries have not, and this changes the workflow for the administration of medication. From our perspective, as the pharmacist had the main role of creating a drug library, the training should also include the pharmacist/the team responsible for the build as they may be able to tailor the training to specific care areas and drugs frequently used in these care areas. Other points to take into account are to ensure that enough time is given to build a comprehensive drug library, and that drugs which are used across many specialities are standardised. The standardisation process of the drugs may need to go through organisation-wide committees, such as the Drug and Therapeutics Committee. Some drugs may not be able to be standardised due to the clinical need, and these drugs should have appropriate risk assessments conducted to minimise errors from occurring.

Conclusions
With the implementation of any new health technology, the organisations should involve the key stakeholders. In the case of smart infusion pumps, pharmacists have a role at all stages-from the tendering and procurement process through to implementation and evaluation. Standardisation of intravenous infusions should be considered a prerequisite to building a drug library for smart infusion pumps, and sufficient time and resources should be allocated for the creation of the drug library in order for it to be accurate and safe.