Use and impact of technology-assisted workflow (TAWF) systems for drug compounding in pharmacy practice: a scoping literature review

Study type

This is a scoping literature review.

Data sources and selection of studies

A search was conducted on studies describing or evaluating the use of TAWF published from January 1, 2015 to July 31, 2021. Two databases were searched (i.e., PubMed and Embase), followed by a search on Google Scholar.

In PubMed, the following search strategies were used: (“compounded” [All Fields] OR “compounder” [All Fields] OR “compounders” [All Fields] OR “compounding” [All Fields] OR “compoundings” [All Fields]) AND (“workflow” [MeSH Terms] OR “workflow” [All Fields] OR “workflows” [All Fields] OR “workflow s” [All Fields]) AND (“pharmacie” [All Fields] OR “pharmacies” [MeSH Terms] OR “pharmacies” [All Fields] OR “pharmacy” [MeSH Terms] OR “pharmacy” [All Fields] OR “pharmacy s” [All Fields])

In Embase, a search was carried out with two key words “drug formulation” and “workflow” in the advanced search tool.

In Google Scholar, the following search strategy was used: “drug compounding pharmacy workflow”. Manual selection was made on the first pages of search results.

For each search strategy, articles were selected based on of title, then abstract, then full text. The selection was made by two research assistants and a pharmacist. Differences were resolved by consensus.

Articles describing or evaluating the use of TAWF in hospital pharmacy practice were included. Abstracts, thesis, descriptive articles without original data and articles not related to pharmacy hospital practice, were excluded. In addition, articles describing or evaluating a technology (e.g., balance, camera or robot) without sufficient details about the characteristics of TAWFs were excluded.

Data extraction and analysis

Data relating to TAWFs was extracted from the articles included. The impact of the TAWF was assessed on four parameters when applicable: error capture rate, preparation time, validation time by the pharmacist and costs generated or saved. The impact was rated as positive, negative or uninterpretable. Descriptive statistics were produced.

Profile of studies

A majority of the studies were from the United States (12/17). The studies evaluated commercial products (16/17) or locally developed products (1/17). TAWFs all included preparation assistance software (17/17), barcode reader (17/17), photo or video taking (17/17), some included gravimetric systems (9/17), and others used robots (2/17). Most of the studies used technology for parenteral preparations (15/17), one used technology for oral preparations only (1/17), and one used technology for both types of preparations (1/17). Most of the articles selected prepared these drugs for adults (10/17), the others were intended for children (4/17) or mixed (3/17).

Four studies assessed staff satisfaction (4/17). The perceptions that emerged were that TAWFs were perceived to be more secure and accurate than the non-TAWF approach. Three studies evaluated the accuracy of TAWFs (3/17).

Eight studies declared conflicts of interest, either in fees paid to authors (5/8) or funding by the company that marketed the TAWF (3/8).

Description of the use of TAWF

Seven descriptive studies described the systems implemented in their pharmacy (Table 1). One study compared two TAWFs. Three parameters were evaluated among these studies: error detection rate (n=6), preparation time (n=2), and costs generated or saved (n=2).

Impact of TAWF

Ten studies evaluated the pre-post impact of the implementation of a TAWF (Table 2). Four parameters were evaluated: error capture rate (n=10), preparation time (n=6), validation time (n=6), and costs (n=6).

Impact of TAWF on error detection rate

Four studies evaluated error detection and all showing an increased rate with the TAWF systems [25, 30, 32, 33]. The error detection rate varied from 0.096 to 1.44% pre implementation vs. 1.94–42.02% post implementation.

The types of errors reported are mostly a dilution error, wrong vehicle or diluent, wrong bag type or wrong drug.

However, the results were difficult to interpret because the length of the evaluation period or the number of preparations audited differed greatly between the pre and post implementation phase of TAWF. For example, Bucci et al. [33] reported a significant increase in error detection rate, but the number of preparations was smaller in the post implementation phase (0.47% (n=116,686 preparations) vs. 42.02% (n=5,195 preparations), p<0.0001). The large difference is explained by the fact that the system alerts on all types of errors, including errors within the tolerance threshold. Considering that the expected error rate is often low, it is preferable to target similar and sufficiently long observation periods.

A systematic review by Batson et al. [36] highlighted the impact of automated compounding technology for the preparation of chemotherapy. They demonstrated that compounding technologies improved accuracy in dose preparations and reduced dose errors compared with manual compounding. Flanagin et al. [37], reported a low proportion of error detected with the use of a TAWF, both with one that was not integrated with patients electronic medical files and with one that was integrated (0.72 and 0.88%, respectively).

The definition of an error varied from one study to another (wrong dose errors, wrong drug/diluent errors, wrong concentration errors …). Also, TAWF systems allowed the detection of errors that were not detectable by human (e.g., too small discrepancy to detect) and possibly not clinically significant. Further studies should use a detailed definition of error including all types encountered and their level of clinical significance.

Impact of TAWF on preparation and validation time

Six studies evaluated the impact of TAWF on preparation and validation time [25, 27, 28, 31], [32], [33]. Four studies have documented a reduction in preparation time of 9.2–34.8% [27, 29, 30, 34] and five in validation time [23, 27, 29, 34, 36]. One study had documented an increase in preparation time and in validation time of 13.2% [33]. With the use of TAWF systems, the mean preparation time per dose varied (i.e., 51 s [33] and 180 s [36]). The data presented was difficult to interpret given the variability in what was being measured (i.e., TAWF start-up time, preparation time per dose, validation time per dose, combined time, type of preparation). The measurement of the workload involving a TAWF system must take into account the start-up time required for the system implementation (e.g., installation of tubing, calibration, preparations, other requirements related to the operation of the TAWF). The higher the number of preparations, the less the weight of these steps in the calculation of the average preparation time will be. There should be a minimum threshold for the number of preparations for a TAWF to be considered cost-effective in a hospital. This threshold is currently unknown. With the current studies available, it is not possible to rule on the effect of TAWF in pharmacies on the working time.

Impact of TAWF on costs

Six studies assessed the impact of a TAWF on costs [27], [28], [29, 31, 34, 37]. All the studies concluded on a reduction in costs, linked either to a reduction in production costs (i.e., reduced preparation or validation time) or to a reduction in costs associated with preparation errors. Calculated savings were also difficult to interpret given the data presented. In Batson’s systematic review, the compounding technologies were associated with reductions in annual costs compared with manual compounding [36].

Staff satisfaction

Three studies assessed the satisfaction of pharmacy staff exposed to the use of a TAWF [28, 33, 34]. In general, respondents perceive TAWFs to be more secure given the increased traceability of the process and the technologies used to facilitate error detection. Regarding the speed of the systems used in Bucci et al. [33] study, staff members agreed that the volumetric method was faster than the gravimetric TAWF method. In Marzal-Alfaro 2020 [34], all employees would like the speed of compounding to be increased. This perception is not, however, a reflection of reality since four studies have shown a reduction in preparation time and five a reduction in validation time.

Future research

It is often difficult and costly to properly assess a change in practice. A well-designed study should include a starting hypothesis, a calculation of an adequate sample size to answer the main outcome being assessed. In most of the studies consulted, it is rather a convenience sample that does not provide a robust answer to the questions asked. Hospitals should be encouraged to come together and come up with a solid study design for evaluating the integration of a new technology before generalizing it within a regulatory authority. Such evidence should also be published independently. The evaluation of the impact on productivity, risks, quality and costs remains relevant, but we must ensure to measure the effect of technology on all the tasks entrusted and not in isolation.

Strengths

This study targeted different TAWFs concerning all types of preparations (IV and oral, hazardous and non-hazardous) with a focus on gravimetric systems. It highlighted the components of a TAWF and the main outcomes assessed. It also makes it possible to identify the limits of the studies and the improvements to be made to the estimates for the next studies. The data presented might also help pharmacists identify their local assessment criteria for a possible acquisition of a TAWF system.

Limitations

The search strategy was limited to the period 2015–2021 due to the rapid evolution of technologies. Previous publications were not considered to avoid previous software versions or technologies. It should also be noted that only studies written in English were included in this review. A broader search strategy would provide a broader perspective on TAWF and their evolution. Six of the 17 studies selected were descriptive and did not assess the impact of implementing TAWFs. Studies that assessed the impact of TAWF used a pre/post design and were neither controlled nor randomized. In addition, there was positive publication bias in favor of technologies (e.g., negative studies are not published). TAWF change over time and the results should of these studies should be interpreted with caution considering the evaluated version of a given TAWF. It should be noted that the different studies included often had a different definition of medication preparation (e.g., Achey’s study was the only one to consider loading up the medication on a cart as part of the medication preparation process [31]). This had an impact on the comparison and analysis of the temporal impact of TAWFs. There is a learning curve when implementing a new system, so the duration of the post implementation period must be considered. Most studies did not have a comparable duration of study for the pre and the post phase. Finally, several of the studies published included (8/17) a declaration of conflicts of interest involving TAWF manufacturers.