What are the implications of electronic prescribing and
robots on purchasing practices?
BPharm MSc MBA MProf
In Hospital Pharmacy Europe November 2009, I discussed some of the factors to be considered when choosing an electronic prescribing system. City Hospitals Sunderland is a 1000 bedded acute hospital in Northern England, and has been operating an integrated electronic prescribing (EP) system for over 8 years. Recently (2009), we have bought an original pack automated dispensing machine, and are integrating its functions into our medicines management systems and the comments I make in this article are in the context of the British NHS hospital system.
Previously, it was outlined how the integration of the software for different modules within the hospital created opportunities for potential efficiencies across the organisation, but that the efficiencies were realised if the opportunity was used to change practices. The same applies for the dispensing robots. Over 3501 have been installed in hospitals across the UK, and they have been installed with a variety of features. At Sunderland, we considered the business case would need to exploit all the advantages with as many features as possible, which were included, for ‘future proofing’ the investment. Our approach is similar in principle to the EP system; namely the greatest integration of functions should realise the greatest efficiencies. For this reason, a machine was purchased with a view to dealing with pharmacy store functions, ward box picking, automated loading from a hopper, automated delivery reconciliation, automated labelling for dispensary items, an out of hours ‘night safe’, and a small fridge unit. Our robot has over 6km of shelving within a “footprint” of 15×8 metres of shelving. It follows that a critical component is the interface software linking EP to the robotic dispensing processes.
Our medication processes within Sunderland Royal Hospital (SRH) mean medical staff are able to sit anywhere in the hospital, and prescribe from the EP screen. Once checked electronically on screen by the pharmacist, the robot can pick the product, label it, and move the item to the appropriate point for onward distribution for the ward. The efficiency of this is obvious, though the full benefit as yet needs to be better quantified. The second point is that as the prescriber is effectively doing the labelling (the value of integration of software), the label will always be 100% (accurate) of what was keyed in. Since the products are barcoded in the robot, there would be an expectation that picking errors would be minimal, and that as the processes are controlled, labelling errors would be reduced because of the automation of the labelling to the picking. In theory, could Sunderland have mitigated the problems of dispensing errors? For some hospital colleagues this might seem that we have built a dispensing system fit for Utopia.
However, Utopia comes at a price! There are implications if this is to work as intended. First, it should be envisaged what is happening in our processes. In a virtual sense, the prescriber is using EP to “reach” into the pharmacy, pick a product, and label it. To do this he has to pick products from his screen. Herein lies the first implication of having integrated EP and robotic dispensing. When the prescriber prescribes in EP, he not only selects a medicine, but its strength, form, and pack size. Consider an intention to prescribe paracetamol. On the doctor’s screen, he will have to select from the following (see figure 1).
Faced with the above, it is possible for an incorrect drug form or pack size to be selected, and once the selection is made, this could go right through the automated pharmacy dispensing processes without being spotted. 60% of the dispensing process is initiated outside the pharmacy at Sunderland. The problem can also be aggravated as a result of confusion between other drugs with similar names across the formulary.
This means that simplification of the prescriber’s screen is paramount to avoiding incorrect selection errors. This means that in its turn, the pharmacy has to become more careful in selecting what it purchases, and fully understand the implications of any actions taken within purchasing practices.
The pharmacy at Sunderland is open 80 hours per week, which leaves 88 hours in the week available for use if this can somehow be utilised. Developments should seek to be able to use the remaining 88 hours in the week in some way. The robot uses some of this time for replenishment of stock through the automated loading hopper at night. However, if there are problems and the overnight fill is not done, then this has to be re-worked into the crowded schedules of the working day. Recently (September 2009) there was recognition of the fragility of the medicines supply system in Britain when a letter was circulated by the Department of Health asking hospitals not to put strain on the supply system by increasing stocks too much during a flu pandemic. It is alleged that the supply system in Britain is partly affected by the strong Euro making parallel imports unprofitable, thereby reducing national stock availability. This leads to an increasing number of “stock-outs”, and variation of product supply. Many NHS hospitals use PASA (Purchasing and Supply Authority) contracts to obtain best prices, but currently do experience more frequent stock unavailability. Routine products normally used within the hospitals become increasingly used by community pharmacies because parallel imports are unavailable. Our experience is there is usually some form of the product available, but not necessarily the contract brand. Leaving the financial issues of the matter aside, this means that currently we often receive products into the pharmacy stores which are not the brands ordered by the pharmacy. Such products will have some or all of the following differences from what was expected:
- Different bar code.
- Different packaging dimensions.
- Different pack size (quantity of tablets).
- Different “robustness” of the packaging.
- Perhaps a different MEPA score (Medication Error
- Potentia l Assessment).
Within the automated systems in place, these can cause problems at various points in the dispensing processes. The pharmacy process starts with loading then storage, then selection, then labelling, then dispatch.
If the product has the wrong barcode (ie, unrecognised) then the product will not load. Pharmacy systems usually have a multiple barcode facility, and use this facility when the need arises (eg, if the pack contents were different). However, at Sunderland, this is not necessarily an acceptable option, as the software files containing the drug dictionary are linked to other modules across the hospital. This is a consequence of an integrated EP system. Alternative new products may require more drug dictionary items to be built, so that the drugs received can be used. However, the effect this has on the prescribers screen is potentially enormous, as instead of having the list of products as indicated on the PARACETAMOL example, added to all those product lines would be multiple entries of all of the above from a variety of different suppliers multiple brands. From the prescribers’ perspective, they would have no way of knowing which were the products they should choose from the list presented to them. Consequently the risk of an incorrect selection error increases significantly. Therefore the pharmacy aim is to keep the prescribers’ selection list to a minimum to minimise this problem. This is not easy. For example, to prescribe methotrexate, the doctor has to go through three screens of products to select the correct preparation.
Another implication is the automated labeller. In theory, once the product is loaded, the dimensions are known, and the auto-labeller would adapt the label to it. However, there are limits to the smallest size of packs that can be labelled, and the defaults can cause serious problems in the dispensary within the way the dispensing processes work at Sunderland. Many hospitals with robots will have the capability to accommodate these un-preferred packs, but when linked to an integrated EP system, the crucial point to remember is that there is a lack of flexibility to accommodate the non-preferred products. This is a discipline imposed by the drug product files being used by the EP software and other hospital software modules that links into these files (eg, files that flag allergy warnings that work in real time).
What this means is that we have to develop an agreed procedure with our suppliers when contracted lines are unavailable. Our preference would be to have a default brand, but that yet needs to be negotiated with the various wholesalers. It is a process we are only just starting to develop at Sunderland. It means the wholesalers have to be aware of the hospitals buying patterns, and they have to stock our default brand. This is a significantly different practice from the “nearest cheapest alternative” that tends to be the current norm in these situations. The consequences to the pharmacy and the hospital of this practice could be significant in terms of medication errors, therefore the problem needs to be managed differently.
One of the consequences of having an agreed second choice product with a wholesaler, is that much more information needs to be shared with wholesalers about the hospital needs, to ensure continuity of supply. It forces the hospital to consider developing longer term partnerships with its wholesalers. The motivation for this is to give the business commitment to the wholesaler to try to ensure “as cheap” acquisition prices for medicines when placed against the current methods. There is a further consequence for those agreeing to a PASA contract to supply hospitals. If there are problems for any reasons, and the second choice product is used by the hospital as outlined above, if the price difference is small, there is little incentive for the hospital to go back to the original contract brand. In this scenario, the secondary supplier of the drug could become the dominant supplier nationwide. The business view would be, that is what markets are all about, and those companies which provide what the customer wants at an acceptable price should be the ones to benefit. However, it means companies supplying into the UK market may have to increase stockholding compared to current levels of activity. In their turn, in future, the suppliers may seek to offset the cost of extra stock by offering less discount to the NHS hospitals service.
Much more work needs to be done to quantify the potential impact on supplying into the NHS, but there is an interesting tension between the economics of national supply and currency values, and the operational requirements of hospitals that progressively seek automation to make their processes more efficient. It will be interesting to see how the pharmaceutical supply market changes as more hospitals go down the automation/EP route in the coming years.
1. Personal communication from ARX ltd.