Clinical Decision Making: Part I

I practice in a university clinic which functions partly as a tertiary referral center, which means we get referrals from other specialists. I also get many referrals for second opinions. Sometimes the entire cause for the patient’s desire for a second opinion, it seems to me, is the simple fact that they did not understand the reasoning of the previous specialist. They were given a diagnosis and a course of treatment, but not an explanation of how their doctor arrived at those conclusions.

I am not being judgmental – different practices are under different pressures and time constraints, and it can be very difficult to gauge a patient’s understanding. Often the physician and the patient are proceeding based upon differing assumptions and narratives that are not expressly stated. The doctor may think they have explained the situation entirely, but simply did not confront misleading assumptions they were not aware their patient had.

This is part of the advantage of engaging the public about health issues and confronting pseudoscience, myths, and misconceptions – you develop a deep awareness of how the general public thinks about medicine.

Carl Sagan warned against scientists becoming a “priesthood” with inaccessible knowledge and jargon, and this pertains to the applied science of medicine as well (perhaps especially). We need to engage with the public so that there is a general understanding of not only the findings of science but how science works. In the context of medicine, this means understanding clinical decision making – how doctors think.

Toward that end I plan on writing a short series of posts that explore various angles of clinical thought. These are talks I have had with medical students, residents, and my patients, to make my thought process as transparent as possible. I have found it to be extremely useful and a worthwhile investment in time.

The Dr. House syndrome

Patients often come to their doctor’s visits with a hidden narrative, a narrative I call the Dr. House approach. This is a narrative the public has largely learned from watching doctors on TV. It is also a perfectly reasonable narrative, and is not wrong as far as it goes. It’s just very narrow and incomplete.

The narrative is this – a patient has an illness (a certain set of symptoms), and that illness is caused by a specific pathological process or disease. The job of the physician is to be smart enough to make the correct specific diagnosis, which will then lead directly to a cure and resolution of the illness.

When this sequence of events does not occur or fails for some reason it is solely due to the fact that the physician simply did not make the proper diagnosis, so they should try again, order more tests, or refer the patient to someone with greater knowledge or more appropriate expertise.

On TV at least, Dr. House always eventually makes the correct diagnosis and saves the day.

Sometimes this is actually how medicine works. In fact doctors love telling each other stories about the complex or difficult case that they eventually diagnosed, especially when that further led to a cure, or at least an effective treatment. Often these stories are formally told as case reports or presented at conferences. These are fascinating stories because they are the exception, rather than the rule, of our day-to-day practice.

Other clinical pathways

While this is one possible pathway that diagnosis and treatment may take, it is not the only legitimate pathway. Here there is much variability by specialty. For most specialties, however, most patients will present with common ailments and diseases (common things are common – it may seem obvious but we do have to constantly remind ourselves of this – in fact this one point is worthy of its own post).

Diagnosing common conditions often is a matter of making a specific diagnosis and treating it. Patients have hypertension, diabetes, multiple sclerosis, etc.

What I am discussing here are those patients with something other than discrete, common, and easy to establish diagnoses. How do we approach such cases?

Ruling out

It is often as important to know what a patient does not have as knowing what they do have. We order some tests to rule out diagnoses that are possible, serious, and/or treatable. (An algorithm as to which diseases to consider and which tests to order can be the topic of a future installment in this series.)

It was surprising to me when I first realized how upset some patients can get when you inform them that they do not have a horrible diagnosis. I actually had one patient cry when I gave them the “good news” that their test was negative. That’s because I had yet to understand the narrative under which they were working. They were thinking diagnosis leads to treatment leads to cure – no diagnosis, not treatment.

I have since learned that I have to explain to patients as I am giving them their results that it is a good thing when certain tests come back negative. They do not want to have the diagnoses I am ruling out.

Further I explain that it is sometimes better to have no diagnosis than to have a bad diagnosis. Patients in whom all the tests come back negative (depending on the clinical situation) often do better than those with a specific diagnosis.

Further still, it is often not necessary to have a specific diagnosis in order to treat an illness.  (Again, this is very specialty-specific – surgeons do need a specific diagnosis most of the time, whereas neurologists often have to proceed without one.)

To treat without a diagnosis

To summarize everything above, an alternate narrative to “diagnose, treat, cure” is this: Rule out everything bad, understand the phenomenology as best as you can, then treat symptomatically. This may seem less satisfying, but it is often the preferable situation (again – you do not want to have a horrible diagnosis just to have a diagnosis).

Migraine is an excellent example of this. Patients present with a set of symptoms that we recognize and categorize as a migraine. There is a great deal of variability, however, and many patients may present with an atypical migraine, or migraine-like syndrome. There is no way to objectively make this diagnosis, no positive diagnostic tests.

The diagnosis is partly based on understanding the phenomenology – what kind of process is happening. It is also partly based upon ruling out other things that can cause similar symptoms (like a brain tumor). We then treat based upon a categorical diagnosis – the kind of process that is likely happening. Response to treatment then becomes the best confirmation we can get that the diagnosis is probably correct.

This may all sound very unsatisfying, but this just exposes yet another difference between public and expert narratives. Doctors learn to think in terms of what is likely to happen. How is this patient’s story going to play out? After years of practice you develop a perspective on this, and you realize that it is not a terrible thing when all the tests come back negative. Those patients may do quite well.

Another way to think about the above approach is this: The diagnostic workup is focused on finding active processes and pathology, especially those that are treatable. We then treat what we find, in addition to treating symptoms to improve quality of life. At no point in this process do we need to establish a specific diagnosis.

For example, I see patients with nerve damage (neuropathy). First I establish that the patient’s symptoms are due to nerve damage and characterize that as best possible. Then I look for any active process or anything treatable that might contribute to poor nerve health. I may find that the patient has a low vitamin B12 level. For many patients I still have no idea if the low B12 is causing or contributing to their neuropathy. Except in advanced cases, this is difficult to determine, and I won’t get a nerve biopsy just to confirm this. Instead I will just replace the B12.

Categorical and vague diagnosis

Another important distinction to make is that not all diagnoses are created equal. This too will be the topic of a future post, but for now suffice to say that there are hierarchies of diagnosis, and different types of diagnosis. Not all diagnostic labels refer to a specific and distinctive pathological entity. Sometimes we can determine what type of process is happening (an inflammatory process, for example) but not get more specific than that.

In other words, making a diagnosis is not a black or white, all or nothing proposition.

This approach to some patients has led, in my opinion, to the creation of vague diagnostic categories. This is because we need something to label patients in whom we have ruled out everything specific. These categorical diagnoses can be useful, but also a trap if you treat them as if they were a specific pathological diagnosis.

Fibromyalgia is the poster child of this phenomenon, in my opinion. It is a convenient label for anyone with muscle and connective tissue pain (myofacial pain) that is otherwise undiagnosed. Sometimes clinicians will call such a label a “garbage pail diagnosis” – we throw everything in that we cannot label more specifically.

This can be a useful placeholder, it can be useful to put the diagnosis into its proper category in terms of which kind of process is happening, and it is convenient short hand. In this case fibromyalgia may also be a specific disease, with specific findings, but it is used as a catch-all for any myofacial pain syndrome.

I prefer to use the term “myofacial pain syndrome” because it does not imply more specificity than actually exists. There is a bias, however, toward labels that sound like specific diseases.

Don’t be complacent

All of this is not to suggest that we should be complacent with less than complete knowledge. This is a tricky balance that also benefits from experience. We have to understand the limits of our knowledge and technology, and understand how to approach and treat patients in whom we will not make a specific diagnosis. But this should not lead to complacency – accepting too quickly that we don’t have a diagnosis.

The most important factors in determining how aggressive our posture should be toward making a more specific diagnosis is whether or not we have found specific pathology, and how the patient is doing clinically. If the patient is stable or improving, and/or responding to treatment, it may not be necessary or advantageous to continue ordering more diagnostic tests. If the patient is progressively getting worse, however, this suggests the need for more information.

Also, sometimes tests come back negative and sometimes they come back with abnormal but non-specific findings. For example, an MRI scan of the brain may find a “lesion” (a deliberately vague word) but not show specific features that would allow for a diagnosis of what the lesion is, exactly. We now have possible evidence of a pathological process, and this warrants a more aggressive workup than if the MRI scan came back pristinely normal.


The goal of this series of posts is to help physicians and patients communicate more effectively by exploring and understand the various assumptions and narratives that each bring to the clinic visit. This is an exercise in metacognition – thinking about thinking, in this case thinking about how doctors think.

As you can see, it quickly gets complicated. I have raised many points here, each of which deserve further exploration. The comments are likely to raise points and questions that I can also explore further in future posts.

In approaching this series the most challenging aspect was figuring out how to divide up the various topics. They all are interrelated in a web of decision-making. Hopefully the result was not too disjointed, but I will attempt to keep referring back in future posts and hopefully it will all come together in the end.

Other entries in this series

Part II
Part III

Posted in: Science and Medicine

Leave a Comment (32) ↓