Automated dispensing systems are beginning to make an impact in United Kingdom hospitals in the wake of the Audit Commission recommendations. At a meeting in the Stirling Royal Infirmary on 9 October, the audience heard how different types of automated systems have been implemented in several hospitals in England. Christine Clark reports
Train staff fully for maximum benefit from robotic dispensing
Mr Swanson: installation requires a lot of preparation
Robotic dispensing using the Pack-Picker (Swisslog) system went live at the Royal Liverpool and Broadgreen University Hospitals (RLBUH) in September, Derek Swanson, the hospital’s deputy director of pharmacy, announced.
After a detailed evaluation of the available products, the Pack-Picker was chosen because it most closely met the needs of the department. An important feature was the option to site the input unit on one floor and the picking and delivery units on the floor above. The Pack-Picker uses a random storage system in “honeycombs” of diagonally arranged cells. One of the advantages of this arrangement is that round packs can also be stored easily. Cells of different sizes are available and each cell can hold more than one item.
At RLBUH, the robot has been installed so that goods can be loaded into it from a lower ground floor stock distribution area. At the ground floor dispensary level there are five linked storage modules equipped with five picking heads. The robot, which can hold a total of 20,000 packs, dumps picked items on to an internal conveyor that then delivers them to one of five labelling stations. A roller table then takes completed robot-dispensed items on to the next point where non-robot items are added and the final check is done.
At the pre-implementation stage pack dimensions had to be checked for compatibility with the robot. Refrigerated items were excluded because there is no facility for these in the Pack-Picker. Bottles larger than 300ml, because of the risk of breakage on the conveyor, and Controlled Drugs were also excluded. Only packs with EAN barcodes can be stored in the Pack-Picker. A policy decision was made to give priority to high-usage lines. Application of these criteria resulted in list of 1,200 products that were suitable for use in the robot and 300 that “gave cause for concern,” said Helen Metcalfe, information management and technology technician, RLBUH.
Installation of the system required a great deal of preparatory work which must not be underestimated, said Mr Swanson. Part of the vision for the purchase was not just to buy a robot but also to improve workflow in the dispensary. Originally, the dispensary was cluttered and staff had to walk to and fro many times to collect items. The remodelled dispensary has made a simple circular workflow possible. “Even with the noise of the conveyor, the dispensary is a lot quieter because people are not bustling round,” said Mr Swanson.
Training of staff is critical to effective implementation of a robot and adequate time must be allowed for this to be done properly, emphasised Mr Swanson. For example, the technique for loading of an item involves allowing the machine to scan and check the dimensions of the product. If this is done incorrectly — by getting a finger in the way or moving the item during scanning — the machine will miscalculate and may be unable to load the product into a suitable cell.
Another issue has been the new procedure for dispensing mixed prescriptions. Staff working with the robot are now expected to handle robot items only and pass the incomplete prescription on to other staff for completion. “Staff need to understand that this is a more efficient way of working,” said Mr Swanson.
Turning to the lessons learnt, Mr Swanson said that good preparation is paramount and ideally this should include talking to colleagues with experience of robots. When the machine is put into use a loading strategy is also essential because this in itself is a time-consuming exercise. At RLBUH the majority of items were loaded during a weekend and the remainder during the next few weeks, when the robot was operational. The impact on the workforce needs to be taken into account. At the beginning of the project many of the staff thought that a robot would be a mobile unit and were concerned that it would get in their way — the concept of a static robot had to be explained.
Mr Swanson concluded that the real benefits of a robotic system are reaped when the re-engineering that the robot makes possible is also undertaken.
Automation does reduce dispensing errors
Pharmacies could become people-free zones in future if automated systems were to be fully exploited, according to Neil Caldwell, deputy chief pharmacist, Wirral Hospital NHS Trust. A Rowa Speedcase dispensing robot has now been in operation since January 2001 at the Wirral Hospital (PJ, 5/12 January 2002, p10) and has produced a number of significant benefits, he said.
One of the expected benefits has been a reduction in the dispensing error rate. During the period 1994–2000, the most common dispensing errors involved picking the wrong strength of a product or the wrong product. Some of the errors in the “wrong product” category were accounted for by look-alike and sound-alike names, for example, confusion between droperidol and domperidone or prochlorperazine and procyclidine. After the implementation of automated dispensing the error rate fell by 51 per cent, from 15.7 per 100,000 items to 7.7 per 100,000 items, explained Mr Caldwell.
Dispensing turn-round time has also been reduced. The number of items dispensed per technician-hour has risen from 10–12 to more than 15. The biggest impact has been the opportunity to re-engineer medicines management processes as a result of the release of 31 per cent of technician time. Technicians have now been redeployed as medicines managers at ward level. “One of the best things we have done is to increase the number of technicians available for ward activities,” said Mr Caldwell. A knock on effect of this development was to simplify the ordering process, he added.
The amount of space devoted to storage on the acute site is less as a result of the installation of the robot. This is largely because items can be stacked in the robot from ankle-height up to a height of two to three metres, since access for people is no longer a consideration. “There is massive potential to improve the use of space,” said Mr Caldwell.
The workflow in the dispensary is organised such that incoming prescriptions and orders are checked by a pharmacist for clinical accuracy and appropriateness and then passed to a technician for dispensing. There are four “drop points” in the dispensary at which the dispensing technicians work. When a technician generates a label, the robot, which is situated in a separate room, is activated, the item picked and carried along a conveyor to the point where the label was generated. Non-robot items, for example antibiotics that need to be reconstituted, are assembled on a bench and the completed prescription is passed to a checking bench for the final check. The workflow is simple and staff do not need to leave their seats frequently, explained Mr Caldwell.
Any suggestion that dispensing errors would be completely removed by automation is not necessarily true, warned Mr Caldwell. A prescription for sulindac tablets 200mg twice daily would involve entering the short code for sulindac in the dispensary system, ie, SUL20T. This brings up a choice of sulindac 200mg or sulpiride 200mg and if the wrong product is chosen the robot will pick it. It is therefore important to stress that there is a potential problem with the introduction of automation, he said. The real danger is that this might not be identified at the final check stage. Whereas a sulpiride label on a sulindac pack is likely to be identified by the final checker, the right label on the right pack of the wrong medicines for the patient (for example, a sulpiride label on a sulpiride pack when sulindac was prescribed) is more likely to be overlooked.
For the future Mr Caldwell predicted a scenario in which doctors would prescribe electronically and pharmacists would perform the clinical check at the bedside. The prescription would then be conveyed electronically to the pharmacy dispensing system, which could be organised in such a way as to deliver the dispensed items direct to the patient’s bed. Further refinements would include linking the prescribing systems to primary care and refilling the robot by a wholesaler. Electronics and informatics should not be used to mirror existing processes but to support implementation of the best possible processes, he said.
Future challenges at Wirral hospitals include the installation of an interface between the electronic prescribing system and the robotic dispensing system so that technicians no longer have to type the (electronically generated) prescription details into the dispensary system. Looking further ahead, Mr Caldwell predicted that the dispensing robot could be developed to include “in-line labelling” and then the product could be checked electronically against the prescription using barcode technology. It could then be delivered to the patient’s bedside. Another option might be for wholesalers to stock dispensary robots and only charge the hospital when the product is issued.
In conclusion, Mr Caldwell pointed out that a robot has never applied for another job and has no aspirations to do any job other than that for which it has been designed.
Robot reduces walking time for technicians
Dr Fitzpatrick: space efficiency was important to us
Efficiency and capacity were the two critical considerations that led to automation of the dispensary at the Royal Wolverhampton Hospital (RWH), explained Dr Ray Fitzpatrick, clinical director of pharmacy, Royal Wolverhampton Hospitals.
The hospital will gain approximately 2,000 beds when a new heart-lung centre is opened in 2004 and the pharmacy department is currently small, busy and unable to accommodate additional personnel. At present it handles more than 800 items per day. “We needed something that would improve the working environment and improve the workflow,” said Dr Fitzpatrick.
The Consis (Baxter Healthcare) machine was selected because it turned out to have the highest storage capacity per square metre — 3,333 items per square metre — of all the available devices. Conventional shelving can accommodate about 1,600 items per square metre and random access systems 1,100 to 1,500 items per square metre, explained Dr Fitzpatrick. “Space efficiency was important to us,” he added. The machine uses channel storage (rather than random access) and gravity-fed automated picking. Two linked modules, each with a footprint of 1.20 by 1.35 metres, and a total capacity of 11,000 items have been installed at RWH. The device is equipped with two picking heads: one that can pick single items only and one that can pick multiple items. Picked items are dropped on to a conveyor and moved to one of four workstations. Another important feature of the system selected was the integrated accuracy-checking system that is built into the Consis system.
The original sequence of tasks was a clinical check by a pharmacist followed by label production, picking and assembly, all carried out by assistant technical officers. The final technical check was then performed by an accredited checking technician. “It was a good system but worked in a poor environment,” said Dr Fitzpatrick. Now, after the introduction of the Consis robot, the generation of the label triggers picking and an automated accuracy check (based on barcode scanning to match the dispensed items to the on-screen prescription) is performed at the assembly stage, before the final manual check.
Dr Fitzpatrick said that it was necessary to analyse product usage critically before planning the allocation of the channels in the robot. For example, 4,500 items are listed in the dispensary computer at RWH but only 1,400 are in regular use. “Over a 12-month period there are about 900 lines that are really active,” he said. He had calculated that 700 lines represented 90 per cent of activity — paracetamol tablets alone accounted for 7 to 8 per cent of activity. A graph of cumulative activity against the number of lines showed that about 300 items accounted for 80 per cent of activity. “Adding extra lines does not add extra productivity,” he stressed. Once this information was available the robot could be configured to fit the product range.
Loading the robot is a manual process but a trolley can be taken directly to the rear of the machine. Each product is identified by barcode scanning, and then the channels available are listed on a screen. The operator selects the channel to be filled using the touch screen, a shutter covering the row opens and a laser light identifies the channel selected. The quantity loaded is entered via the touch screen. “It is a lot quicker than loading conventional shelves because there is so much less walking around,” said Dr Fitzpatrick.
Assessing the initial impact on his department, Dr Fitzpatrick said that the expected benefits of speed and efficiency had been achieved. The robot can pick an item in less than 10 seconds and the dispensary had been designed to incorporate the minimum of conveyors so that the advantage of this was not lost. Walking time has also been reduced. A pre-installation study using pedometers had shown that technical staff were walking an average of three miles per person a day in the course of routine dispensary work.
“The rate-limiting steps are now the clinical check at the beginning and the technical check at the end,” he concluded.