A few weeks ago, I was asked by my partner to defrost some fish for dinner. Out of curiosity I asked, “What sort of fish is it?” to which she replied, “I don’t know, it’s just fish”. In a household kitchen, not having perfect information probably isn’t going to cause many issues; but under different circumstances–as in healthcare–it could cause significant, life-threatening problems. 

Orion Health recently released my white paper The Value of Semantic Interoperability for Healthcare. The paper discusses how, amongst other things, computer systems that can infer meaning from standardised clinical terms–condensing broad sets of information into easy-to-access terminologies–can greatly benefit healthcare.  As the paper is quite technical, I thought it would be worthwhile expanding on an important topic in that ‘easy-to-access’ way; which is where fish comes in.

Words can mean different things to different people and the word “fish” is a good example of this. Your local takeaway store will, more than likely, think of fish in the terms of whether it tastes better grilled or battered and which is more popular. When a marine biologist thinks about fish they possibly think of types and sub-types, colours, observed migration paths, and physiological and anatomical structures. Then there’s me who thinks about fish simply as food.

The language we use, and the words we choose, convey meaning and emotion in a way that is unrivalled in the animal kingdom. Yet humans are fallible and can easily be misunderstood and misinterpreted, which can lead to issues – especially within healthcare.

The aquatic example shows how a simple, well-known word can carry numerous different, potentially abstract meanings in a variety of contexts. In the demanding healthcare environment, conveying unambiguous meaning is essential. Because of this, it’s important that internal IT systems aggregate and present terminologies in a standardised way that is familiar to all clinicians. Collecting this data in such a way allows for precise, more comprehensive analytics and quicker, more accurate diagnoses in the future.

Healthcare professionals can use a range of terms to describe illnesses or injuries, for instance, heart attacks. What we call a “heart attack” could be described as exactly that or, under a different term, as a myocardial infarction, a myocardial infarct, an MI, or even a cardiac infarction. All these terms refer to the exact same thing, yet they’re all represented differently. Further to this though, what is the context of this condition? Does the patient have a personal history of heart problems? Is there a family history of this condition? Is this happening right now? Etc.

Let’s imagine that you’re a healthcare provider in a clinical diagnostic scenario. You might conduct diagnostic tests and procedures with your observations which are then analysed as evidence to conclude whether a disorder is likely to be present or absent. You also might not have enough evidence either way and therefore consider it as a differential diagnosis. Once the evidence is clear and the diagnosis is confirmed, you would want a clear representation of that diagnosis, and its context, for other clinicians to unambiguously know what that could mean within their specialty – you don’t want them to have to fish for meaning.

Some more ambiguous terms that are based on classification systems could be compared to a generalised, statistical “net”, meaning everything that’s grouped in that net must fit under the same category. This isn’t typically supportive of efficient clinical workflow or high clinical data quality with relevant granular detail.

Healthcare professionals often don’t have time to remember and reference codes or their exact descriptions. You want to quickly document what you know, in a way that’s familiar to you. Common clinical abbreviations, acronyms, and synonyms should ideally be usable and the ideal computer system should be able to parse what you mean.

Lucky for us, we already have something that does that: SNOMED CT. SNOMED CT is an international healthcare domain ontology standard that represents both the common clinical terms and short hand that we use, and the various details about the clinical condition, procedure, item etc. that clinicians know. It is also valuable for clinical auditing, population health monitoring, and decision support.

There’s great value in capturing clinical data at a level of detail that conveys rich clinical meaning, using SNOMED CT, and enabling:

  • Easier, more intuitive clinical system use–clinicians can focus more on clinical care which minimises change management efforts
  • Minimised data duplication and re-entering through data aggregation and reuse
  • Improved clinical data quality and accuracy for auditing, reporting and resourcing analysis
  • Advanced clinical decision support tools–this includes improved machine learning models that can support clinicians’ workflows
  • Advanced population health analytics
  • Cost savings

If you’re a healthcare professional, don’t just be forced into a general statistical net. Think about it. What do you want from the time spent on your clinical system? What do you expect from your data? If you want to capture what you mean in a way that can save time while also capturing clinical detail, catch it hook-line and sinker.