Death, disability, and injury caused by fragmented rule-based knowledge

Surprisingly, there is limited, quality research into the state of fragmented rule-based knowledge and the way it is applied in practice. One area of fragmented rule-based knowledge is the world of healthcare. Of course, this is a subject that matters to everybody.

Let’s set aside for the moment that fragmented rule-based knowledge is ubiquitous and counter-productive. ?Because the way humans use rule-based knowledge is not measured, no one can quantify the volume of rules. This lack of a scientific approach to measurement means healthcare organisations are unable to quantify the volume of rules needed for human decision-making.

Insights from other sectors demonstrate that the entire system of rule-based knowledge covering what is written down and the way in which it is enacted in practice, is not fit for purpose. This means under-identification and under-reporting is common place. From a risk perspective, human decision-making that should be rule-based is highly prone to errors.

In the absence of scientific-based measurements it is worth attempting to size the volume of rules in healthcare to start understanding the magnitude of the risks caused by human decision-making. The NHS in the UK is deploying a global standard called SNOMED CT, which is a structured clinical vocabulary, each with a unique code. The NHS version of SNOMED CT consists of circa 500,000 codes, which includes health conditions and medicines.

There has to be a correlation between a SNOMED CT code and the number of rules needed to ensure the code selected is correct. In the absence of science, let us assume the average is ten rules per code. Thus, the volume of rules would be five million. This may well be conservatively low, but at least it is a starting point for consideration.

On the basis these five million rules evolve then ideally the latest rules need to be instantly available to those on the front-line making decisions. As rule-based knowledge is not measured the reality is nothing like the ideal. When mistakes are made the de facto counter measure is to improve training. This is the equivalent of kicking the problem into the long grass as addressing the five million rules cannot be solved by training. How anyone expects human decision-making to have instant and exact recollection of the latest rules when the volumes are so high defies common sense. But the situation is much worse as fragmented rule-based knowledge is ubiquitous.

There are plenty of disbelieving or indeed cynical retorts when trying to highlight the problems with fragmented rule-based knowledge. It does matter far more than most people realise.

In 1999 the USA Institute of Medicine published a now famous study called “To Err is Human”. It’s findings suggested that in the USA more than a million injuries and as many as 98,000 deaths each year are attributed to medical errors.

In 2016, researchers at Johns Hopkins estimated that in the USA as many as 250,000 people die each year due to medical negligence. This would make medical errors one of the top causes of death in the USA.

In 2019, the World Health Organisation found that healthcare is likely to be one of the ten leading causes of death and disability in the world. They estimated that in high-income countries, one in every ten patients is harmed while receiving hospital care, with nearly 50% of them being preventable. Globally they found as many as four in ten patients are harmed in primary and outpatient healthcare.

Rule-based knowledge is about a step-by-step sequence of small decisions. Each small decision may seem innocuous, but collectively the outcome of the way the decisions flow matters. As shown, we need to focus on these small decisions, as they can have a significant impact.

An updated version of my newsletters are available as a podcast https://tinyurl.com/pxdhtt5v .?