» Technical Bulletin – Reporting Legionella results
28 November, 2017
In March 2017 UKAS issued communication to all accredited laboratories involved in Legionella testing of waters (appended). The letter set out the expectations for the reporting of results in the volume tested. The letter made specific reference to excluding the extrapolation of counts to give a result in a litre volume in cases where a smaller volume is received.
Feedback from ISO/IEC 17025 assessments since the circulation of the letter has been varied with some laboratories reluctant to implement the terms of the letter because of concern over customer expectations. Some laboratories have proposed to present the accredited test result for the volume tested alongside a disclaimed extrapolation count for a litre volume. This is not an acceptable solution as it creates an inappropriate association between the accredited result and the invalid extrapolation.
There is no issue with a test report identifying that a test is outside of the scope of accreditation, or is a deviation from / modification of an accredited method, but extrapolating from an accredited result gives no indication to the end user of the implications for uncertainty associated with the result.
When a laboratory receives a water sample of less than one litre it has no knowledge of the incidence of legionella in the wider water system. It should never be the case that “not detected” or “less than” results are extrapolated as “Not Detected in one litre” when a smaller volume has been analysed. “Not detected” or “less than” results can only be presented for the volume tested and in the case of reporting ND, the limit of detection should also be specified clearly.
In the case of positive enumeration results when receiving less than one litre of sample, calculating “upwards” to a litre result is technically invalid because the distribution of organisms in that larger volume is unknown. It is an extrapolation rather than calculation as there is no way of demonstrating equivalence of the result for a litre if the laboratory did not have a litre in the first instance.
It is recognised that HSE technical guidance series HSG 274 state action levels “per litre” and, in agreement with the customer, through documented contract review, a laboratory may need to provide a statement against these even though a smaller volume has been received and analysed. To comply with the requirements of ISO/IEC 17025, such statements must be clearly presented on a test report as additional comment given to assist in the interpretation of the result and indicate that associated uncertainty is unknown and therefore the likelihood of any extrapolated value being a reasonable estimate is also unknown.
In cases where the laboratory receives a given volume and then sub-samples from this, e.g. receives 1 litre and samples 200mL for centrifugation, calculation of a count in the received volume may be acceptable if the laboratory has procedures to ensure homogeneity of the sample when sub-sampling. In such cases the laboratory has control over the entire sample. Appropriate validation data will need to be available to support this position, enabling a reliable estimate of uncertainty to be established.
Appendix: Text of Customer Letter dated 24 March 2017
Detection and enumeration of Legionella in water – reporting under accreditation
This letter has been sent to you because our records show we accredit your organisation for Legionella testing activities. We wish to bring to your attention a reporting practice come across during our assessments that does not meet requirements for accreditation but that laboratories perceive to be common.
The majority of UKAS accredited laboratories use ISO 11731:1998/BS 6068-4.12:1998 as their reference method. Section 10 of this standard (Expression of results), states the following:
Report the confirmed presence (or absence) of Legionella pneumophila and the presumptive presence (or absence) of other Legionella species. Report absence as “not detected” in the volume examined. Ideally report the serogroup of all isolates of Legionella pneumophila”.
Section 11 (f) of the standard reiterates this. ISO 11731-2:2008/BS 6068-4.18:2004 makes similar statements.
UKAS’ expectation is that the reporting of Legionella test results by accredited laboratories is as defined in these standards when they are given as the reference method. Additionally, extrapolation of a microbiological test result from the volume of water examined to a larger volume is not statistically valid (see, for example, MoDW – part 3 – Practices and procedures for laboratories; Section 8.3) and will not meet ISO/IEC 17025 requirements for any purposes that can be envisaged.
Examples of appropriate reporting formats are:
Not detected in 1 litre
xxxx cfu in 1 litre
Not detected in 200ml
xxxx cfu in 200ml
In support of the above, a robust contract review is vital to ensure that the volume analysed by the laboratory is suitable and appropriate to meet customer needs.
Reporting of Legionella test results in conformance with the above and associated contract review will be a focus of planned UKAS assessments in the coming year.
Please contact me if have any queries on the above.
Download printable version of this Technical Bulletin