The biopharmaceutical oncology unit at Bretonneau Hospital, Tours, France, provides 25,000 doses of medication a year and aims for delivery within an hour of validation of a patient’s prescription. The unit has developed a computer model to help achieve this
Director of UBCO
Anne De Laguerenne*
Director of Pharmacy
Universitaire De Tours
Centre Henry Kaplan
Rabelais De Tours
The Biopharmaceutical Clinical Oncology Unit (UBCO) at Bretonneau Hospital in Tours, France, provides treatment to patients. The production centre is required to provide 25,000 doses a year for nearly all oncohaematological specialties. In 2007, the unit gained ISO 9001 v 2000 certification. Some patients are allocated to a specific daycare unit at the hospital, and their prescriptions are known in advance. Once a patient is officially admitted to the hospital, a physician validates the prescription before starting production. The production unit aims to deliver the prepared medication within one hour of validation, although in some cases this may in fact take half-an-hour longer. Achieving this goal may be difficult because good preparation time is directly related to the unit’s consistency in its flow of production.
In order to carry out this project we developed computer modelling of charges, with specific settings in the form of heuristic algorithms. Rationalisation of production initiation has aided the shift from approximated manual mode to application-guided mode, with high visibility of production volume.
The production unit (Figure 1) can be likened to a workshop, where the machines are the sterilisers and isolator chambers. The operators are the pharmacy technicians and production orders are prescriptions, comprising a set of jobs to be produced in succession. The difficulty faced by the production unit is in assigning tasks between machines and operators in a way that maximises speed of production. Accordingly, production workload is divided between four workstations (two isolators and two posts) in a way that dynamically delimits four axes of output. The isolation chambers are divided into two parts, with each part associated with one pharmacy technician. Generally, a certain job is assigned to a particular technician who then takes responsibility for preparing both the steriliser and the part of the isolation chamber they will be using.
Figure 2 illustrates a production centre. Using this system allows us to prepare a typical load – any number of preparations between one and 12 for each isolator (with a maximum of six for the side window and six for the side wall; the wall and window allow us to differentiate the two posts). Advance planning by the hospital affords the production unit the time to sterilise preparations so prescription preparation can start as soon as the patient arrives. The time gained allows pharmaceutical validation, preparation and control of products and sterilisation (25 minutes in total). The software used – a Java application – allows variations in planning depending on the availability of each post (on the basis of domestic resources) and time duration and priority of preparation.
Planning preparations are determined by patient arrival times and the handling of their treatment, including the time to prepare each dose. This constitutes a “hypothetical workflow” load, because we have to wait for validations to prepare doses. If preparations are not validated they constitute a “dead volume” in the isolator, also limiting following charges. “All-comer” validations are added to these schedules. Once validated preparations are sterilised, they constitute the “real work volume” (delineated in light green, pink or orange). If anticipated delays become inevitable due to numbers of preparations or late validations, the preparations are delineated in red or dark orange. Among other things, this visual support system allows the critical timing of initiation doses to be determined.
At the start of the day, all available manufacturing data are transferred for implementation of planning. Thus, 70% of the day’s imported production provides guidance for the following day and allows us to prepare in advance or not, according to the gradual arrival of patients.
Procedures differ considerably throughout the week. For example, Monday, at the start of the week, yields very few validated preparations. However, by Wednesday there is a mix of validated and “standby” preparations, allowing anticipation of workload because stable preparations are planned at optimum times during the day. The pattern on Friday tends to resemble that for Wednesday, except that we are required to prepare all doses planned for the weekend, if they are stable. Thus, the number of preparations may increase very sharply.
Despite these differing patterns, by about 11.30am each day we have enough information to allow us to anticipate any problems with achieving targets and accordingly minimise these by adapting the equipment or operator used (for example, by giving an operator additional assistance).
The tests allow us to optimise division of labour depending on needs and on time available, significantly cutting waiting times for all services.
As regards UBCO’s internal organisation, planning avoids “embolisation” of non-emergency preparations. This support gives us more room to work and allows a safety margin for emergencies. The average number of charges since the programme’s initiation is seven carts at one time.
Some volumes of sterilisations may require operators to prepare a maximum load – that is, with all 12 available places occupied without any previous preparations still in the isolator. Such situations tend to occur with certain products, such as carmustine, melphalan and dacarbazine.
An additional problem relates to cases where preparations are delayed or cancelled; here, it may be difficult to define the optimal charge. Utilisation of colour coding in terms of relative urgency allows users to gain a clear understanding of the nature of the incoming workload. This tool allows streamlining of production. An ongoing study will allow evaluation of real impact after several months of use.