evolution, revolution and pathology
Why so complex?
This is the first of several blogs on the subject of pathology data. Others will cover principles and pragmatic tradeoffs.
There are many independent factors associated with diagnostic tests, that when combined, produce an unparalleled level of complexity in data capture, representation and exchange, when compared with most other subdomains of health.
Firstly scope. Pathology and diagnostic testing in Australia covers a wide range of sub-disciplines, including anatomical pathology, autopsy, clinical chemistry, general pathology, blood bank, haematology, histology, immunology, cytology, cytogenetics, microbiology, serology, … And that’s excluding all the diagnostic imaging sub-domains! New diseases are being discovered continuously, and whole new branches of pathology will no doubt arise in the future. Within these sub-disciplines, a raft of different tests are conducted, with a range of different test methodologies. The way tests are named and grouped varies across the sub-disciplines and the identification and grouping of tests change according to the perspective – e.g. clinical versus logistical versus funding.
Tests are often carried out on specimens, sometimes multiple per patient. Sometimes multiple laboratories are required where specialist tests are requested. Usually the primary laboratory on-sends the specimen to a secondary lab to undertake a rarer or more specialised test. An individual test may have multiple phases or a group of tests might vary in the time it takes for the component tests. Such situations can lead to interim or amended results. Tests can vary substantially in their urgency.
These complex workflows call for sophisticated data communications protocols. HL7 messaging is used widely in many countries, including Australia for this purpose and the version 2.x protocols, with their many trigger events, workflow states and acknowledgements, support and reflect much of the workflow complexity.
Tests are frequently carried out for diagnoses, including screening or discounting certain conditions as part of the diagnosis process. They can also be conducted episodically for monitoring an already diagnosed condition, and they can be conducted to determine a patient’s readiness for treatment, such as when checking neutrophil levels prior to chemotherapy.
Not only do we have the complexities of the tests, test workflows, test specimens and test purposes described above, but we have to contend with the complexity of the results!
Some results are subjective, some objective. Some are qualitative, some quantitative. Of the quantitative style tests, the results can often be compound and/or complex (e.g complete blood picture, or ECG waveforms ) and typically include reference ranges. The reference ranges may vary with age, sex, or other patient cohort characteristics. And for a given test type, the reference range may even vary from laboratory to laboratory!
Finally, the way the data is best represented will depend on the purpose to which the data is to be put. The primary purpose is usually to directly inform the clinician who requested the test. This suggests optimising the result data for presentation to the clinician. Test reports are enhanced for this purpose by including formatting instructions – bold, colour, tabular data, etc. where appropriate. Laboratories can differentiate their services from their competitors through report reformatting.
However, as we head ( albeit, slowly ) towards better reuse of pathology data, via shared health records for better shared care, or for registry reporting for better research and better management functions, then computer processability becomes an important determinant of data representation. Consistency of structure, agreed terminology, well designed information models, datatypes all become paramount. But they need to be consistent and coherent with other data standards across the e-health spectrum – no longer determined solely be agreement amongst pathology labs!! Data does not need to be structured, codified, nor typed, for the benefit of the labs!
All of the above discussion leads me to the conclusion that the landscape of pathology communications and pathology data processing is very complex. This, coupled with the fact that we already have significant adoption of electronic communication of pathology results in Australia, further leads me to the conclusion that in such a complex landscape, evolution is better than revolution. But it is not clear to me that for the past 10 years we have had either. This situation could easily be attributed to systemic incompetence – everyone’s collective fault, but no individual’s fault! It reflects poorly on all of us who have had involvement over the past 10 years.
Despite the overall complexity, some things should surely have been achieved.
- It should have been easy to standardise the names of tests!
- It should have been easy to improve consistency and quality of data through accreditation and conformance testing of messages.
- It should have been easy to facilitate electronic reporting to communicable disease, cancer, and other registries via web and messaging portals.
Quite simply, it appears we have had a leadership vacuum. Leaders can make good decisions, bad decisions or no decisions. In my experience, those who make bad, or no decisions mainly do so through ignorance – they can’t, or don’t wish to, understand the problem. Perhaps the complexity of pathology renders such leaders impotent? Hence the leadership vacuum?
Vacuums tend to get filled. And leadership vacuums are often filled through revolt! Therein lies a paradox for us all.
Thanks Eric,
we have only scratched the surface wrt HL7 V2 pathology and the potential workflow capabilities are enormous and there for the taking.
The limitations are in the implementations and not the standard. In a way this is part of the issue, IHE often dumbs down the capabilities to make implementation easier.
Reporting is often also dumbed down because of poor compliance. It is possible to transmit ECG data as raw waveform data in HL7 V2, but its not often supported.
We need a clear direction so that companies can invest in building solid reliable V2 implementations, rather than sit on their hands waiting for Nehta to change the rules. The US has Selected HL7 V2 for pathology and Australia should reaffirm this choice so people can invest with confidence. We certainly need compliance testing of both senders and receivers, as at the moment you cannot send compliant V2.3.1 messages to 95% of PMS systems.
A lot of time and money has been wasted trying to extract business rules from the Pathology Use Case when they are already catered for in excruciating depth by the standard. We need more local standards and handbooks to narrow down the target, but the standard is more than capable. There has been no investment by Nehta in this area, which is a huge missed opportunity.
Adding EN 13606 Archetypes to HL7 V2 gives you excellent metadata and backward compatibility and is one path forward to support more complex data structures than are currently supported. We need the basic compliance fixed first however as adding rich data structures required solid standards support as a baseline.
Great post. The complexities of pathology, and eDiagnostics in general, are hugely complex from a business perspective, let alone the technical perspective. Not only do we see different reference ranges from different labs, but different reference ranges from differing testing equipment within a single lab.
There’s also the issue of safety: how can we (safely) perform longitudinal studies on the results if we can’t statistically normalise the reference ranges? At the moment the clinician is making decisions based on results with different reference ranges which may or may not be safe because we can’t represent the data in a ‘user friendly’ format. I’d be very interested to hear of success stories here.
Just on a quick note, unless I’m mistaken, the Australian Standard for pathology messaging is HL7 v2.3.1 or (preferred) 2.4
There are many issues affecting the delivery of pathology results in a reliable and usable way and you mention many of them. However one you don’t mention is billing, and more specifically “pathology coning”.