Final post on why cause is at the core of clinical reasoning

This should be the final post on the theme “cause as core to clinical reasoning”, and it will address why recognizing cause as core is important to clinical practice, research and education. Of course if additional questions come up that require a response additional posts might pop up.

In the last post there was a reference to an excellent resource on clinical reasoning: 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).

This reference discusses the clinical reasoning strategies model, and outlines several reasoning strategies. The last post started with one of these strategies (ethical reasoning) based on it being (in my opinion) the most difficult to connect to causal reasoning (causal inferences based on an underlying causal structure that represents our understanding of the world, our knowledge so to speak). This post will go through all of the strategies, in turn, to attempt to demonstrate their ultimate connection to causal reasoning.

Diagnostic reasoning: attempts to connect a set of signs and symptoms to their underlying cause,which is the diagnosis (pathological classification). Involves abduction, and for additional examples on the causal structure search the blog for the series on differential diagnosis (here is a post on the differential diagnosis abductive challenge). How well we can diagnostically reason with the information available to us has do entirely with the underlying real causal structure. What information is needed in order to make a diagnosis is completely related to what I have started calling the abductive adjustment set.

Narrative reasoning: “Strategy that requires the establishment of and understanding of the “person” inside the patient.” This is similar to diagnostic reasoning methodologically, but on a different stratification of reality (see the post on critical realism). The therapist is getting answers to questions which provide information about the underlying cause of those answers. If someone tells you their are “vegan” then people typically ask if they are vegan for health reasons or for animal rights reasons (here as in the past on the blog I am using “or” in its inclusive or logical form, so both are possibly true). The idea here is that there is typically a cause for the outward behavior of eating a vegan diet. The two best explanations (though not the only explanations) are health reasons or animal rights. If it is for health reasons you then assume other behavioral characteristics about this person - you probably would be surprised if they were a smoker. However, if it is for animal rights, you may be less surprised by the person being a smoker as you may assume a lower causal association between animal rights as a cause and being against smoking as an effect; as compared to being concerned about health as a cause and being against smoking as an effect. The only point is that we are working with a causal structure of behaviors and these provide us insights into the person’s values, beliefs, attitudes, etc, which all are important in the course of clinical care for a whole myriad of reasons. But at the core we are considering the causal structure of such an understanding of the person inside the patient.

Procedural reasoning: “Strategy that requires choice in administration of interventions.” The choice of interventions is predicated on the understanding of the causal structure of the problems being faced by the patient. The intervention is then, in it’s own right, a cause in the causal framework. This was addressed in the post on graphical causal models and intervention.

Interactive reasoning: “…means of approach and interaction with the patient…” Reading someone’s behavior during interpersonal communications is an abductive process to identify the possible causes of their particular behaviors, and reactive (interventional) in your next behavior. If I enter a room and shout “Hi, I’m your PT and am going to get you moving!!!!” and the patient simply looks up with lips turned down, eyes caring, brow furrowed, I should interpret those outward signs as being caused by the patient’s distaste for how I presented myself. My intervention is to attempt an approach of interacting that might be more appealing to the patient. But ultimately this interplay of interacting is, at its core, a matter of interpretation behaviors, attempting to discern their underlying cause, specifically asking people whether we are right about our perceptions, in a highly dynamic inferential process based on our understanding of causal structure at a stratification of reality of interpersonal communications. Clinicians that are skilled in this, can read behaviors, identify their cause and adapt their approach based on those causes are in a great position to establish a rapport with a patient.

Collaborative reasoning: a working relationship with the patient. The examination and evaluation of a patient involves figuring out the causal structure of that patient’s problems. Explaining this to the patient involves working with them to verify the causal structure and prioritizing the possible interventions based on that underlying causal structure. It also includes abductively responding to the patient’s feedback and attempting to know them better (based on narrative reasoning, and having a rapport based on interactive reasoning). In some ways collaborative reasoning, like reasoning about teaching, involves more of an acceptance of causal reasoning and a “meta - causal reasoning” process than actually using causal reasoning in the collaboration.

Reasoning about teaching: Reasoning about teaching - like collaborative reasoning - is more of a “meta-causal reasoning” process. To learn about other clinical reasoning strategies (diagnostic, procedural, narrative, predictive, etc) students are taught about the essential cause - effect relationships either implicitly or explicitly. They learn about causal structures (again either implicitly or explicitly) relevant for clinical practice. Students can be taught to build causal models based on the knowledge they are gaining in PT school - see this post, and see the GitHub repo for the DPT pathology class I am teaching at UMass Lowell this semester here.

Predictive reasoning: predictive (prognostic) reasoning is based on understanding the causal structure of a system and predicting what will happen given certain states of causal variables (conditions, factors, etc) in the future based on given assumptions and constraints. For more about predictive reasoning - see the post on this topic here.

Ethical reasoning was covered in the last post here.

With causal reasoning at the core of each of these strategies of reasoning it should be clear now that causal reasoning is core to clinical reasoning. Causal reasoning involves knowledge about an underlying causal structure, which is depicted as a causal model and is an approach to demonstrating the knowledge used in practice.

Recognizing cause as core provides bridge for communication between research scholarship, practice and teaching. This is directly discussed in the early post on graphical causal models here. Researchers are developing causal models - either through basic research, clinical trials, systematic reviews, practice guidelines, or simply generating models for testing and organizing them for use in practice. Clinicians are using the causal models in practice - they are reasoning through them to make decisions. Students are learning about them - both about how they are generated through research and how they are used in practice. It allows us to develop syntax and semantics for developing, testing, adapting, sharing, collaborating, reporting and using a system of causal models as representations of causal structure as representations of knowledge for practice.

The blog has already started to organize methods (syntax and semantics) to this end and the hope is that with additional interest and training collaborative teams should start developing causal models that depict the current state of knowledge in the profession.

Leave a Comment