Tuesday, October 8, 2013

Pattern recognition vs problem based approach to veterinary cases: we need both

Just like Mr B and Phil, pattern recognition and problem based analysis may seem poles apart on the diagnostic continuum, but a recent paper in JVME argues that we must use both.
“In an ideal world, every patient would display unambiguous signs of disease confirming to classical textbook descriptions, and the clinician’s pharmacy would be an assembly of rational and efficacious therapeutic agents that would collectively address all the diseases of the animal kingdom. Unfortunately, the ideal world is not the real world, and a series of limitations relating to all aspects of diagnosis and therapy make veterinary medicine (as with human medicine) a “science of uncertainty’” – Stephen May, 2013.

As a veterinary student I operated under the assumption that one qualified I would be a clinician working in the ideal world, as described by May. I believed that once I graduated and was in full command of all veterinary knowledge, I would be able to manage cases and determine the single correct diagnosis and treatment plan. It turns out that companion animals don’t read textbooks. They certainly don’t follow them.

The type of clinical reasoning taught in veterinary schools is one which assumes certainty – when that can lead one up the (wrong) garden path in practice.

It may be the wrong garden path, but if I unleashed my guinea pigs here there would be a lot more path and a lot less garden.
In a recent article in the Journal of Veterinary Medical Education, Royal Veterinary College Deputy Principal Stephen May highlights one of the major flaws of veterinary curricula: the mixed messages it sends about clinical reasoning. Students are taught problem-based learning and discouraged from relying on pattern recognition, when pattern recognition clearly has merit.

The article makes an important distinction between scientific reasoning (with its emphasis on objectivity, presupposition that the observer stands outside the process, solving scientifically framed questions with appropriate technology, and where data requirements are saturated) and clinical reasoning (more subjective, where the clinician is part of the process and outcome, problems are not neatly framed, datasets are incomplete and ultimately the clinician must act in the absence of complete knowledge or data). Unlikely scientific reasoning which begins with a hypothesis, clinical reasoning starts with a problem and works forward, on the basis of probability, using inductive logic, to determine the most likely answer.

May discusses two major forms of reasoning: type 1 being consistent with a pattern recognition approach (this is intuitive, rapid and a not always conscious response to cues); with type 2 being consistent with a problem based, analytical, first-principles approach. It is much more demanding time wise – we tend to use this kind of thinking when presented with an unusual condition, an uncommon presentation of a common condition, or when something tells us this just isn’t right. Type 2 reasoning requires a large working memory.

He argues persuasively that type 1 and type 2 reasoning are complimentary and should be used together. We may arrive at an answer via type 1 reasoning but should be prepared to cross check where any doubt is present.

“It is essential that experts and novices alike be aware of the need and have the capacity to resort to analytical reasoning when cases are difficult,” he writes.

But he adds that it is dangerous to assume that type 2 reasoning is free of bias or free from error. It requires some accurate framing of the problem, and because it employs rules it is prone to error in cases where those rules don’t apply.

So sound clinical reasoning involves weighing up probabilities, cross-checking our gut feeling with the available dataset and the rules we know we can apply, and being aware of gaps in our knowledge and limitations in our thinking.

Sir William Osler, cited by May, expresses this beautifully:
[Medicine is the] practice of an art which consists largely in balancing possibilities. It is a science of uncertainty and an art of probability…Absolutely diagnoses are unsafe and made at the expense of conscience.
To study the phenomenon of disease without books it to sail an unchartered sea, while to study books without patients is not to go to sea at all.
It is wonderful to see that clinical reasoning has become a subject of study unto itself. Thinking through problems is central to what veterinarians do. One of my colleagues, a retired equine practitioner, told me “vets are decision making machines. You could make 1000 decisions in a day.” The mechanisms of decision making therefore impact on clinical outcomes and patient wellbeing and are worthy of our attention in their own right.


May SA (2013) Clinical reasoning and case-based decision making: the fundamental challenge to veterinary educators. Journal of Veterinary Medical Education 40(3):200-209.