How to come up with a differential diagnosis?

[unsung]

---

Members don't see this ad.

Yeah, I know the acronyms- VINDICATE, CITTENS, etc.

They're fine and dandy. But I still feel I'm drudging up stuff out of nowhere 75% of the time, throwing it at a wall, and hoping it sticks. (Saying this as a MSII, for what it's worth.)

There are 2 problems with this method:

1) very likely to miss potential diagnoses (short of yrs & yrs of practical, on the job clinical experience), due to lack of systematic consideration

2) I feel like I'm playing match-maker. Idea "pops" into my head, I try to match the symptoms of the idea (that I memorized) to the symptoms the patient is complaining of.

I mean, if that were all there is to it, couldn't a computer do this a lot better? I'm just playing "match the symptoms".

Isn't there some more systematic way of going about this?

For ex, some "rules", such as- look for systemic signs such as "fever", 1st, or evaluate the CBC before you look at other tests. (Not saying those should be rules at all lol... just tossing out random examples)

The reason I think there MUST be some "best" systematic way of doing this, is 'cuz software that comes up with diagnoses exist. (And... for whatever reasons have never become popular in healthcare, despite performing as well as, or even better than physician diagnoses...)

So a software operates according to certain "rules". Just like there's sort of an automated process to identify a particular G(-) organism (run X test 1st --> if (-) proceed to Y test, etc.).

I just find it very odd that I'm not being taught a "systematic" way of considering a patient's entire set of symptoms, and how to evaluate them.

I find myself trying to figure out "rules" for myself, such as look for signs of inflammation, look for systemic involvement vs localized organ dysfunction, look for signs of infection. I also find myself trying to think pathophys and somehow "fit" all the symptoms into the etiology of a particular disease.

Basically, some way to shape and form the chaos. I just don't like the random "guess and check" method I'm currently stuck with and do not know how to extricate myself from.

Thoughts? Ideas?

P.S. On second thought, a possible "problem" with trying to find a systematic way of going about this is that I find myself trying to "fit" all the patient's symptoms into one specific diagnosis.

Now... this may or may not work for cases on a school exam, where they are trying to test our knowledge of a particular condition. But of course IRL, patients are going to come in with all sorts of comorbidities... so likely oftentimes, symptoms won't "make sense", and will go unexplained, won't "fit together", etc. So... probably woudn't work IRL.

In which case, it may actually be BETTER to stick to a "guess and check" with what "pops" into my head strategy...

???

 

[ArcGurren]

The way I always think is, just throw EVERYTHING you can think of up there and see what sticks. There is no systematic way that I really know of

 

[derspiegel]

Great thread Unsung,
I've wondered about this myself as well.
Don't know the answer, hopefully the future dr. House among medstudents will relieve us🙂

 

[werd]

it's really 1) having the appropriate knowledge base and 2) clinical experience that let you come up with a good differential diagnosis. trying to be very systematic can help if you're stumped or feel like you're missing something, but most of the time your initial management and appropriate workup to define the diagnosis will be based on your clinical knowledge and experiences. forming good differentials is something your residents/attendings help you do during clerkships, and something you don't have much chance of doing successfully until you've learned most ms2 material.

 

[nlax30]

it's really 1) having the appropriate knowledge base and 2) clinical experience that let you come up with a good differential diagnosis. trying to be very systematic can help if you're stumped or feel like you're missing something, but most of the time your initial management and appropriate workup to define the diagnosis will be based on your clinical knowledge and experiences. forming good differentials is something your residents/attendings help you do during clerkships, and something you don't have much chance of doing successfully until you've learned most ms2 material.

This.

It's just going to take time, and usually that whole process becomes a little more clear once in clinical years where you actually put this thought process to practical use and watch how attendings/residents do it.

Some sort of "system" is helpful: Whatever mnemonic you like or by organ system, or location in the body.

 

[BigRedder]

You don't have to think of everything right away. Do a complete history and physical with ROS. Remember the common and emergent diseases associated with each major symptom/body part/system. Test for those and you will generally find your answer. Everything else can be looked up/consulted until it becomes second nature.

 

[Meerkatology]

it just comes naturally as you build up your knowledge base, and get more practice using it.

closest thing to a systematic way would be to think through all the anatomical stuff that can cause the sx. I.e. patient comes in with LLQ pain. you go through all the structures there that can cause it.

knowing whats common and whats uncommon helps too. not for tests, but clinically.

a patient comes with in bp of 220/110 and headaches.
if its the test you say pheochromocytoma.
if its the real world you say the patient wasn't taking his bp meds over the last week.


and things like von hippel lindel disease or tuberous sclerosis doesn't actual exist in real life.
As there are only 3 differentials in internal medicine
CHF, pneumonia, wait did i already say CHF?

2 dfferentials in peds
Viral URI, viral gastrointeritis.

1 differential in surgery
cholecystitis

all other medical conditions you learned about are made up, like santa claus

 

[Knicks]

it just comes naturally as you build up your knowledge base, and get more practice using it.

closest thing to a systematic way would be to think through all the anatomical stuff that can cause the sx. I.e. patient comes in with LLQ pain. you go through all the structures there that can cause it.

knowing whats common and whats uncommon helps too. not for tests, but clinically.

a patient comes with in bp of 220/110 and headaches.
if its the test you say pheochromocytoma.
if its the real world you say the patient wasn't taking his bp meds over the last week.


and things like von hippel lindel disease or tuberous sclerosis doesn't actual exist in real life.
As there are only 3 differentials in internal medicine
CHF, pneumonia, wait did i already say CHF?

2 dfferentials in peds
Viral URI, viral gastrointeritis.

1 differential in surgery
cholecystitis

all other medical conditions you learned about are made up, like santa claus

I like your style.

 

[p30doc]

Great thread Unsung,
I've wondered about this myself as well.
Don't know the answer, hopefully the future dr. House among medstudents will relieve us🙂

it is pretty simple

 

[chiz2kul]

it just comes naturally as you build up your knowledge base, and get more practice using it.

closest thing to a systematic way would be to think through all the anatomical stuff that can cause the sx. I.e. patient comes in with LLQ pain. you go through all the structures there that can cause it.

knowing whats common and whats uncommon helps too. not for tests, but clinically.

a patient comes with in bp of 220/110 and headaches.
if its the test you say pheochromocytoma.
if its the real world you say the patient wasn't taking his bp meds over the last week.


and things like von hippel lindel disease or tuberous sclerosis doesn't actual exist in real life.
As there are only 3 differentials in internal medicine
CHF, pneumonia, wait did i already say CHF?

2 dfferentials in peds
Viral URI, viral gastrointeritis.

1 differential in surgery
cholecystitis

all other medical conditions you learned about are made up, like santa claus

nice

 

[unsung]

it just comes naturally as you build up your knowledge base, and get more practice using it.

closest thing to a systematic way would be to think through all the anatomical stuff that can cause the sx. I.e. patient comes in with LLQ pain. you go through all the structures there that can cause it.

knowing whats common and whats uncommon helps too. not for tests, but clinically.

a patient comes with in bp of 220/110 and headaches.
if its the test you say pheochromocytoma.
if its the real world you say the patient wasn't taking his bp meds over the last week.


and things like von hippel lindel disease or tuberous sclerosis doesn't actual exist in real life.
As there are only 3 differentials in internal medicine
CHF, pneumonia, wait did i already say CHF?

2 dfferentials in peds
Viral URI, viral gastrointeritis.

1 differential in surgery
cholecystitis

all other medical conditions you learned about are made up, like santa claus

That's funny. 👍

 

[nima123]

The key is to think simple and be systematic.

I faced this challenge big time. I even went as far as reading a book called How Doctors Think: clinical judgment and the practice of medicine by Kathryn Montgomery (The other book with this title is by Groopman and is more of a money maker for him than substantial info).

I think that the practice of coming up with a list of things that could explain a presentation (DDx) may be a good learning exercise for some people (not me!) but is not really that useful in real life, because it takes too much time and doesn't add much to a proper workup. Even internists who preach DDX practically give up on it after a while.

Here's my approach:
When you take a history from a patient, identify what kind of a problem you're dealing with (e.g. old guy with SOB --> most likely some heart or lung dz), train your ears for selective hearing, patients give you a lot of BS but your job is to tease out the useful clues so that you can focus your questions and your exam. Then figure out if you will need any tests to confirm your suspicions. Only when you do this and still don't figure it out, would you wanna sit down and go through a list to see what you may have missed.

Regardless how you do it, it's a complex process and takes years to master, so don't try to be a super star.

And House can suck it!