Annual Scientific Meeting - 2001

THE IMPLICATIONS OF USING ‘NON-APPROVED’ DEVICES OR PROGRAMS

J B Glen, Glen Pharma Ltd, Cheshire. (jbg@glenpharma.com)

If using a “non-approved” device or program to deliver a drug by Target Controlled Infusion (TCI), consideration needs to be given to legal aspects related to medical device and medicines regulation and product liability; and practical aspects which influence the performance and capabilities of different systems.

The regulatory approval of infusion systems for the administration of drugs by Target Controlled Infusion (TCI) is complex.  Before a pharmaceutical company can promote the use of TCI as a mode of administration for a particular drug, there is a need to obtain approvals from both device and medicines regulatory authorities.  While prototype TCI systems have been used for research in anaesthesia for a number of years, the ‘Diprifusor’ TCI system developed by AstraZeneca, in association with a number of infusion pump manufacturers, was the first TCI system to gain regulatory approval[1]. The approval of the Diprifusor system for the administration of the iv anaesthetic Diprivan (propofol), required a number of separate but interrelated submission documents. These included  information supplied to a Notified Body on the device, to demonstrate compliance with “essential requirements” as required by the European Council Directive concerning medical devices (93/42/EEC) to support CE marking of the device.  In addition, a submission to the Medicines Control Agency (MCA) was required to support changes to Diprivan prescribing information, including the provision of guidance on appropriate blood propofol target concentration settings to induce and maintain anaesthesia with a Diprifusor TCI system.  The linkage between the two is critical, as the guidance on target settings will depend on the delivery performance of the pump at a particular target setting and the latter is dependent on the particular pharmacokinetic (PK) model incorporated in the pump software.

Pumps incorporating the Diprifusor TCI module are integrated systems and do not require connection to an external computer.  On the other hand, there are a number of “non-approved” programs which may be obtained from their originators or in some cases simply downloaded from the internet (http://pkpd.icon.palo-alto.med.va.gov).  Software programs such as RUGLOOP (Michel Struys, Ghent), STANPUMP (Steven Shafer, Stanford) and STELPUMP (Johan Coetzee, Stellenbosch) can be loaded into a laptop computer and used to control a computer compatible infusion pump.  These programs, when used in this way, are considered to be medical devices and hence fall within the scope of medical device legislation. The originators of these programs indicate clearly that these are experimental programs and responsibility for their use lies with the individual user or their institution. As they are not CE marked, they fall into the category of “devices intended for clinical investigation” and should only be used in clinical investigations following local ethics committee approval and notification of the proposed clinical investigation to the Medical Devices Agency (MDA).

If a prototype TCI device is developed by one legal entity for use in patients in that same legal entity it is not considered to be “placed on the market” and device regulations do not apply. The regulations do apply if the originator is seen as the manufacturer and the device is transferred to another legal entity.  A third option, as follows, may be relevant to some situations encountered in anaesthesia research with TCI systems:  If a healthcare establishment commissions a separate establishment to develop a device or to modify an existing device, and takes full responsibility for the design changes, the commissioning party may be considered the manufacturer of an “in-house” device.  If this device was intended only for research or treatment of the second entity’s own patients, the device would fall outside of  the scope of the UK Medical Device Regulations and there would be no requirement to notify the Competent Authorities of the trial. The feasibility of this latter approach is thus dependent on the willingness of the commissioning health care authority to accept the responsibilities of the “manufacturer”, including that of product liability.

In practical terms, experimental programs and devices as described above can provide an informed clinician with a flexible research tool. A good understanding of pharmacokinetics is required to allow the user to select the most appropriate PK model for a particular drug and a particular patient. Most systems can be used for a range of different drugs and, in addition to controlling the estimated blood concentration, it may also be possible to control the estimated effect-site concentration in the brain [2].  In view of differences in infusion rate control algorithms, pump communication protocols, and possibly the PK model selected for a particular drug, the delivery performance of these systems will not be standardised as with Diprifusor TCI.  As with all TCI systems, there will be a need to titrate the target concentration setting as required, to account for PK and pharmacodynamic variability and thus achieve the effect desired in a particular patient.

By virtue of the need for connection of the infusion pump to an external computer via a serial port and cable, communication errors resulting from electrocautery or other extraneous interference may be more likely to occur with the experimental systems. These systems have not been subjected to a formal and rigorous process of software validation or failure mode and effect analysis and software “bugs” may occasionally appear.  However, as these programs have evolved over many years, it is expected that any errors detected in earlier versions have been corrected in current versions. A key difference between these experimental systems and the Diprifusor TCI system is that the latter has a second microprocessor which, using information from the pump motor encoder, cross checks the performance of the controlling model, thus providing an important safety feature [3].

[1] Glen JB. The development of ‘Diprifusor’: a TCI system for propofol. Anaesthesia 1998; 53, Supplement 1: 13-21

[2] Jacobs JR, Williams EA. Algorithm to control “Effect Compartment” drug concentrations in pharmacokinetic model-driven drug delivery. IEEE Transactionson Biomed Engineering 1993; 40: 993-999

[3] Gray JM, Kenny GNC.  Development of the technology for ‘Diprifusor’ TCI systems.  Anaesthesia 1998; 53, Supplement 1:22-27