Sterility testing (revision 15)

Sterility testing determines whether a sample yields microbial growth when incubated in defined media under defined conditions. A passing result does not establish that a batch is sterile; it establishes that the units tested did not yield growth, which is a statistical statement about the batch rather than a property of it.^[1]

This limitation is intrinsic and well recognised. Testing a small number of units from a large batch has limited power to detect low-level contamination, and for this reason sterility assurance in manufacture rests on process validation and environmental control rather than on end-product testing.^[2]

Sterility is distinct from freedom from pyrogens. A sterile preparation may contain bacterial endotoxin from organisms killed during processing, and the two determinations are not substitutes for one another.^[3]

Two methods are recognised. Membrane filtration passes the sample through a 0.45 μm filter that retains organisms, washes away any inhibitory substance, and incubates the membrane in growth media; it is preferred where the sample can be filtered. Direct inoculation adds the sample to media directly and is used where filtration is impractical.^[1]

Two media are used in parallel: a soybean–casein digest medium incubated at 20–25 °C for fungi and aerobic organisms, and a fluid thioglycollate medium at 30–35 °C for anaerobes and aerobes. Incubation is for not less than fourteen days, which is why a sterility result is never available quickly.

Method suitability must be demonstrated before the test means anything. The sample must be shown not to inhibit growth of specified challenge organisms in the media used; a sample with antimicrobial activity that has not been neutralised will pass the test by suppressing the growth it is meant to reveal.^[1]

The number of units tested is small — commonly ten to twenty depending on batch size — and contamination in a batch is usually sporadic rather than uniform. The probability of detecting a low contamination rate is correspondingly low.^[2]

The figures are illustrative of the shape of the relationship rather than exact, and the exact values follow from the binomial distribution and the sampling plan. The point is that end-product sterility testing detects gross failure reliably and low-level failure poorly, which is why it functions as a final check on a validated process rather than as the basis for release on its own.^[2]

• USP General Chapter 71
• Bacterial endotoxin test
• USP General Chapter 85
• Reconstitution of lyophilised peptides
• Bacteriostatic water