Balance of probabilities

There are a number of ways to ensure that the drugs given to patients are sterile.

You would hope that hospitals were the last place in which you would pick up an infection, but this is not always the case as sterility is based on probabilities not certainties. The term 'sterile' is generally defined as something that is free from viable micro-organisms, and sterilisation methods involve either removing or inactivating these organisms.

Sterilisation is an exponential process – the longer the process lasts the fewer micro-organisms survive – but it is asymptotic with respect to time (figure 1). Taking a log scale, the slope of the curve is a measure of the resistance of the organism to that particular process.

The D value is expressed as the amount of treatment that gives one log reduction in the number of survivors, i.e. a 90% reduction in the survivors (figure 2). It is fundamental to many of the calculations in terms of sterility levels and the sterility assurance levels set. If the line is extrapolated to point A (equivalent to a population of 10-⁶), this gives a D value for the probability of finding a survivor.

However, it doesn't guarantee that in any pack used there is not going to be a survivor. It is a matter of probabilities: as not every item can be tested, the legislators resort to a probabilistic approach. Sterility of an item in a population of items can't be guaranteed. Sterility of the population is defined in terms of the probability of existence of a non-sterile item in the population – expressed as Sterility Assurance Level, or SAL – and it is generally accepted that sterility is defined as the probability of finding an item with a single viable organism present as equal to or less than 10-⁶, i.e. one in a million.

This definition has been encompassed in the EU standard EN556: sterility assurance level for medical devices labelled sterile. It is intended for devices sterilised by steam, irradiation or ethylene oxide (EtO), and for an item to be labelled 'sterile' an SAL of 10-⁶ or better should be achieved by a validated sterilisation process. However, certain medical devices in the US can meet a lower SAL of 10-³ if they are non-invasive, such as surgeons' gowns. Table 1 lists the different sterilisation processes compared with disinfection.

Heat-based sterilisation processes can involve moist or dry heat. Gamma (γ) radiation is generated by bombarding ⁵⁹Co with neutrons, which produces an unstable ⁶⁰Co atom. This releases gamma radiation at 1.17MeV and 1.33MeV and a beta particle. Gamma radiation is highly penetrative, but has slow dose delivery so the process time is in hours.

However, the dose uniformity, i.e. the sterilisation across the container and the products within, is good (figure 3). Produced using ⁶⁰Co, γ irradiation of the sample is carried out from both sides. The total dose depends on the target density and thickness, and the minimum needs to be above the SAL. However, the maximum, at the edges, is often an order of magnitude higher. Typically, if the dose is 25KGy (1 KGy = 1 joule/kg) in the centre, it may be 50-75 KGy at the maxima, and this needs to be addressed in qualification/validation procedures.

With electron beam irradiation, the sample is irradiated with a high-energy electron beam. There is poor penetration (figure 4), but a rapid dose delivery compared with g irradiation, so the process time is in seconds. The dose uniformity is also poor.

Again, for sterilisation the minimum must be above the SAL, but in this case there are significant differences in the minimum and maxima. With a level of 25Krad for the minimum, the maxima could be as high as 100Krad, and again, this needs to be addressed in q/v procedures.

Ethylene oxide processing is a chemical process carried out in three stages: precondition, sterilise and degas or aeration. Batches of palletised product move through the plant for each stage of processing, although companies are now starting to compact the processes into one.

There are a number of processes that can be used to ensure sterility, but whichever is used it is important that the requirements for the validation and routine control of the process are met, because the process is based on probabilities and not certainties.