Justifying cause as core to clinical reasoning

As mentioned in the last post (here), perhaps starting a blog that “presupposes cause as important” might have been a mistake. I am now going to attempt to justify cause as the core of clinical reasoning and clinical epistemology over the next few posts. First, what I mean by justifying cause. By justifying cause I will attempt to show that cause is central (core) to how we practice (all aspects of practice), that knowledge about causal structures is the knowledge required for practicing (making decisions based on a given (and evolving) set of information), and that reasoning about that knowledge involves making inferences which forms the foundation of clinical reasoning. I will not attempt to justify the existence or reality of the law of causation; nor will I, here address the issues of complex, multi-causal structures (will get back to those later :)

dagitty-model

There was a request in the last post for readers to share an example of a clinical scenario where causal reasoning is NOT involved. I was presented with one broad consideration regarding the importance of knowledge about causal relations and structure, but was not presented with an example where causal reasoning was not involved (of course it does not mean one does not exist - though I remain unable to come up with one).

One reader presented the idea that causal inference was not always important in a clinical case since we cannot always track back to the original cause, or know with certainty which causal explanation is the correct one for a particular case. Here we see that causal inference is still core. The causal explanation is still sought after, the best we can do is come up with various alternate explanations and make decision based on which one (or two, or three) we believe in the most (see belief and truth, but eventually I really need to post on what I mean when I say to “know”). That process is still, at it’s core, one of causal inference. Perhaps the better way to say this is that knowing the precise cause is not always important. But even if we accept that knowing the precise cause is not always important in a specific case (more later), that does not remove the necessity of ruling out specific causes that would be important to know (i.e., it is important to know that the cause is not likely to be a condition that is a contraindication for the treatment you may plan), or considering the set of possible causes in the specific case which lead to causal deductive inferences about which interventions to utilize. It is possible the the ultimate cause (original - if we could agree upon when one would stop the regress back in time to attempt to say something was the “original” cause) is not important for the person before you seeking treatment. However, there are more proximal causes that should be identified in order to prescribe the best treatment. So, only if we limit our use of the word “cause” to something we might agree on as the “ultimate” cause could we say that cause is not important in all cases. But I am not willing to do that (to limit the use of the word cause in such a way) because the more proximate causes are causes, they are important and they are what we would base our interventions. For an example of a causal chain see this post on differential diagnosis.

Now let’s try a more systematic approach to how cause is core to all clinical reasoning by taking, in turn, the elements of clinical reasoning and demonstrating their ultimate causal basis. This does not replace other approaches to the description of how the causal reasoning is conducted in various aspects of the clinical scenario. For example in the citation: Christensen, N., & Nordstrom, T. (2013). Facilitating the Teaching and Learning of Clinical Reasoning. In Handbook of Teaching and Learning for Physical Therapists (pp. 183–199). There are several clinical reasoning strategies presented in the model and they are all excellent, valid and important elements for fully understanding clinical reasoning: diagnostic reasoning, narrative reasoning, procedural reasoning, predictive reasoning,…, etc. By saying that cause is core to clinical reasoning I am simply saying that causal inferences are being pursued in all of these strategies. We could attempt to go further to demonstrate how the logical inferential approaches of deduction, abduction and induction are being used in a dynamic inferential interplay in all of these strategies.

Several of these different strategies of reasoning involve a different strata of reality see the post on critical realism), and a different ontology see the post on ontology determines epistemology) (analogical use of language is also helpful). What this means is that we have different approaches to understanding the causal structure of the knowledge at different stratifications. But that does not eliminate the causal structures as being core to the reasoning being done within each strata of knowledge.

To close out this post I will tackle what could be argued as two types of clinical reasoning where causal inferences seem harder to identify. First, issues related to the reliability and the validity of clinical tests and measures. This was presented to me years ago in a class when sharing the idea that cause was core to all clinical reasoning. This student presented that while the results of the clinical test represent something about reality that provides information about possible causes, or effects, and are part of the overall clinical cause effect relationships, that whether a test is reliable or valid had little to do with cause. I disagree. A clinical test or measure is an extension of our basic sensory perception (see basic reliability of sensory perception), and as such there is a causal structure to whatever exists in reality and the clinical test / measure itself. In other words, reality causes the clinical test to report whatever it reports. The nature of that causal relationship will greatly influence the reliability and validity of that clinical test.

Second - I will address the ethical reasoning strategy from the aforementioned chapter (Christensen, N., & Nordstrom, T. (2013)): “Strategy that requires the awareness and resolution of ethical and pragmatic dilemmas in patient care.” The idea here is that this strategy may seem the furthest from causal inferences. True to my Peircian leanings, the awareness of an ethical dilemma involves an abductive causal inference. What is observed is a set of situations and circumstances, becoming aware that these are ethical situations and circumstances proceeds from identifying the causes of those observed situations and circumstances to certain human behaviors. For example, someone was over billed. That is a situation we can observe. We then abductively attempt to explain what caused someone being over billed. If this process of abductive causal inference results in it being clear that someone intentionally over billed a patient then we have an ethical dilemma. Resolution of ethical dilemma’s more clearly relates to cause as it involves an intervention to correct a problem, any intervention is an attempt to cause the intended effect. Clearly, identifying the strategy as ethical reasoning is very important ontologically and educationally. But at the same time the causal core is important to consider generally, and perhaps specifically as it could directly interact with other elements of clinical reasoning within a larger causal structure.

In upcoming posts I will responsively address comments and cases presented to me as possible examples of cause not being core; and I will continue to present examples of the underlying theme of “cause as core to clinical reasoning”, eventually posting on why recognizing cause as core is important to clinical practice, research and education. I will then get back to the task of attempting to formalize a system for generating and using graphical causal models as representations of practice knowledge for use in clinical practice, research and education (as a reminder, my current course on pathology involves students generating and sharing such models about pathology (GitHub repo is here).

As always - collaborators (and authors to the blog) are welcome!

Leave a Comment