Part of SNOMED CT PaLM Mapping Best Practice
Background information
SNOMED CT PaLM benefits
Pathology and laboratory medicine (PaLM) tests are crucial to the diagnosis and management of disease. More than a billion lab tests are performed by the NHS each year, and lab test reportables account for more than a third of all the SNOMED CT recorded in GP records each year. Consequently, the reporting terminology is fundamentally important to patient care.
The Pathology Standards Programme was set up at the former NHS Digital under the CCIO7 initiative, with a remit to deliver a terminology to replace Read PBCL, as this outdated terminology generates less reproducible and comparable lab test result data than is needed. Whilst Read PBCL is a national terminology, it does not provide a common language representing the same meaning across all labs. Through extensive consultation and collaboration with the professional community, the new terminology was iteratively developed, with the first release of SNOMED CT PaLM in 2023.
In terms of data quality, SNOMED CT PaLM reportables are a considerable upgrade on Read PBCL as they provide a common language that can be interpreted the same way across labs, conform to well-defined editorial principles, and carry modelled components that logically define each concept. This empowers machine processing, in turn strengthening clinical decision support and data analysis. SNOMED CT PaLM also allows for the creation of new reportables that can be released in a short space of time, whereas Read PBCL does not. The granularity of SNOMED CT PaLM offers the means to represent multiple reportables carried in complex microbiology reports, which are currently only represented using text strings.
The government has the stated aim of using "AI tools to streamline public services, eliminate delays through improved data sharing, and reduce costs." AI systems are only as good as the data they are trained on, and as pathology reporting data constitutes such a high percentage of NHS data, mapping to SNOMED CT PaLM will help provide better data quality in a significant area of healthcare diagnostics, creating a virtuous circle with quality data allowing AI systems to make better decisions.
Strategic drivers
SNOMED CT PaLM is part of the DAPB4101 Pathology and Laboratory Medicine Reporting Information Standard that aligns to:
- NHS England - Transformation Directorate Data saves lives
- NHS England - Diagnostics: Recovery and Renewal
- DHSC - The future of healthcare: our vision for digital, data and technology in health and care policy
Laboratory landscape
There are currently 122 NHS labs in England, administered by 27 NHS Pathology Networks, who support lab to GP reporting. There are also a small number of privately run labs who provide this service. Due to several factors, labs do not currently use a single, standardised terminology to represent reportables 'natively' in their laboratory information management systems (LIMS). There are multiple LIMS providers, and multiple legacy LIMS instances. Consequently, there is significant variance in reportable source data across the estate.
Historical mapping
20 years ago, to support interoperability between labs and GP practices via the PMIP EDIFACT messaging integration, labs were required to manually map their local reportables to Read PBCL codes. This labour-intensive process was repeated by every lab in England, resulting in variation in the type of source data used to map, and the methods employed, thereby impacting data quality. The ambiguity found in many Read PBCL codes made things worse, because different labs interpreted them differently. Additionally, there was no best practice around governance, assurance, and maintenance, resulting in a lack of source information and questions over the reliability of maps.
Why mapping is required
The use of SNOMED CT PaLM as a standardised terminology used natively in LIMS, thereby negating the need for mapping, is recognised by NHS England as a long-term goal. However, analysis shows this is unlikely to happen in the short to medium-term as it would involve systematic change requiring extensive clinical and technical assurance, plus implementation of system upgrades across the entire estate, which would be labour intensive, time consuming, and costly to the NHS. In recent years, many labs have commissioned new LIMS, thereby tying them into long-term contractual agreements around system capabilities. Any change to these agreements would therefore require re-negotiation, incurring cost.
NHS England considers SNOMED CT PaLM to be a 'reference' terminology rather than an 'interface' terminology. This means it is less suited to the 'user-friendly' (character-limited and human-readable) descriptions that healthcare professionals prefer to use.
DAPB4101 mandates SNOMED CT as an interoperability solution, that is, it supports the sharing of pathology test results between healthcare settings, rather than mandating that it is used at source. Consequently, SNOMED CT PaLM reportables do not have to be generated and flow natively from LIMS, if maps from local reportables to SNOMED CT PaLM are subsequently used to transform the data when sharing the report to the GP system. This can be achieved using 'middleware', as per the process that supports PMIP EDIFACT.
Now that SNOMED CT PaLM is available, a similar mapping effort to that involved in mapping to Read PBCL is required. NHS England is seeking to help make this process as straightforward, robust, efficient, and economical as possible, using automation where appropriate, thereby easing the burden on lab staff.
Mapping effort and benefit ratio
NHS England’s SNOMED CT Code Usage in Primary Care data provides a clear illustration of the value of mapping even a small number of SNOMED CT PaLM Reportables. Mapping just the 20 most used reportables in lab to GP reporting would account for 50% of the pathology and laboratory medicine lab test result data recorded in primary care. Increasing this to the top 300 reportables would account for 99% of the data. Consequently, offering even the most basic mapping support would return huge value.
Last edited: 22 May 2025 3:04 pm