The biggest challenge in diagnosing septic arthritis is to think of it. Once you think of it, there is a diagnostic process that you activate. The second biggest challenge is that that diagnostic process we learn in training is flawed. We have no alternatives right now, so I will share with you my thoughts on how we should use our judgment as clinicians.


Triggers to consider septic arthritis essentially are the same triggers as infection anywhere. Celsus’ cardinal signs of inflammation are dolor (pain), calor (warmth), rubor (redness), tumor (swelling). Some authors have written that fluor (flow) should be the 5th cardinal sign and in this case the effusion is one of the most important clinical signs.


The detection of effusion at the bedside has been covered elsewhere but includes inspection, palpation, and milking. Smaller volumes of fluid are harder to detect. Ultrasound can be utilized to improve our accuracy.


The clinician generates hypotheses at the bedside and then circles back to revisit risk factors. This is an underappreciated process that deserves more attention. The following risk factors have high specificity:



-joint surgery (mainly recent)

-joint prosthesis (at any time)


There are others but those are the main ones. You might think of it as “compromised immune system” and “compromised joint.” If it helps you in remembering this, recall that the synovium is a vascular tissue. It lacks a basement membrane (which tends to filter out bacteria in other tissues, like the cornea). The joint relies on the immune system to protect us from bacterial translocation. Recall also that bacteria love to find crevices to hide in. A disrupted joint, or a compromised immune system creates the setting for septic arthritis.


They don’t all have to have those high risk features though. Often the clinician is left with enough suspicion to pursue a risk stratification process before considering a tap. In low risk patients that process can involve inflammatory markers (traditionally WBC, CRP, ESR, with emerging roles for procalcitonin and perhaps other markers) to lower your suspicion back below the threshold of further testing. I recommend you use as many of these as you need to reassure you in a low risk patient that there is no infection. All of them have limitations but if the results are normal, you are greatly reassured.


For patients where doubt continues to exist, the diagnostic pathway for septic arthritis ends in arthrocentesis, though its results may be less clear than we realize. Common practice is to use synovial blood counts (often 50K) as disease-defining, which does not actually work very well. Synovial WBC cutoffs at all levels will both miss and overcall septic arthritis. Even if you use the cutoff as 25K, it still misses as many as 1/4 of all cases, one prospective study said it would miss more than 1/3 of all cases (Margaretten JAMA2007). Meanwhile there will continue to be patients with WBC above 50K who have other causes (gout for example).


The gram stain sensitivity is as low as 40% (Ross Infect Dis Clin North Am2017). If you have strong suspicion based on risk factors, drain the joint and have them follow up in 24 hours, giving antibiotics while awaiting culture results.



-Septic arthritis is considered when the patient has an inflamed joint (effusion, warmth, pain).

-Inflammatory markers, if all normal, can lower your concern enough in a low risk patient to rule out disease.

-Risk factors (compromise of joint, compromise of immune system) may be more important than the other tests.

-Synovial WBC has significant limitations in sensitivity and specificity, and is not a disease-defining reference standard.


Selected References:


Septic Arthritis of Native Joints.

Ross JJ.

Infect Dis Clin North Am. 2017 Jun;31(2):203-218. doi: 10.1016/j.idc.2017.01.001. Epub 2017 Mar 30.


Does this adult patient have septic arthritis?

Margaretten ME, Kohlwes J, Moore D, Bent S.

JAMA. 2007 Apr 4;297(13):1478-88. Review.

PMID: 17405973



Anecdote 1:Don’t call it a seizure, start CPR

Resident: Hey, remember that patient in room 7 with the heart attack? Now he’s having a seizure!

You: That is not a seizure, let’s start CPR!

(the patient survives, and receives emergent catheterization opening a blocked vessel)


Cardiac arrest is often accompanied by jerking movements. The cartoons of the 1940s understood this. Bugs Bunny would typically do a few good jerking leg kicks prior to faking death with Elmer Fudd, Yosemite Sam, etc. If the setting is more appropriate for cardiac arrest, check pulses first.


This is probably not substantially different from convulsive syncope but I will describe that separately:


Anecdote 2: Don’t call it a seizure, refer to cardiology

“we never figured out why the Brugada syndrome patient had a seizure” –someone who should know better


Blood bank studies show that about 10% of syncope events are accompanied by a convulsion. These can look like epileptic generalized tonic clonic seizures. The differences are:

Pre-ictal: what were the setting, the symptoms, and the signs, before the event. An aura suggests a seizure. Standing in church suggests syncope. Alcohol withdrawal suggests seizure, etc.

Ictal: convulsive syncope is less rhythmic, less symmetrical, and less sustained

Post-ictal: Convulsive syncope patients regain normal arousal within a minute, seizure patients take 10-15 minutes


It isn’t a seizure until you have a diagnosis. It is a convulsion. Apply an appropriate differential that includes convulsive syncope.


Ancedote 3: Don’t call it a seizure, check the temperature

A middle aged patient is waiting to be seen for generalized weakness. The triage RN rushes them out of the waiting room because of a “seizure.” He never lost consciousness and was awake the entire time. Temp is 103 oral. You ultimately diagnose sepsis from pyelonephritis.


Rigors can cause tremendous shaking and can make us worry about seizure. Obviously a seizure can raise the temperature so judgment is required. But don’t automatically assume that a convulsion from sepsis is a seizure. Rigors happen when the temperature is rising, so recheck the temperature.


Anecdote 4: Don’t call it a seizure, educate the family

A patient is here for opioids. The doctor said no. She has a history of developing “seizures” when she does not get narcotics. Now she is screaming loudly and, wait for it…. The RN runs to you announcing a seizure and asks if you will give Ativan. Okay, I admit it, I often give Ativan if I am not sure. One time I went to the patient and said “really, you are having a seizure? Can I see the tongue biting, show me the tongue” and she proceeded to show me her teeth, biting the tongue.


This is tough. Don’t expect the family to understand the difference between seizures and psychogenic convulsions (also known as pseudoseizure). Educate them on the potential for psychogenic causes (but don’t prematurely rule out epilepsy either, unless it is abundantly obvious)



-Our terminology can box us in. Don’t call it a seizure unless you are committing to an epileptic etiology.

-Ask about circumstances before, during, and after the event to identify possible convulsive syncope


We speak of physical diagnosis as if the sign and the suggested diagnosis always match. Often they do, for example when we see acromial step off and suspect anterior shoulder dislocation.


But more often there is ambiguity. Most bedside information is non-specific. For example, tachycardia can mean a lot of different things. Later when we look at everything at once, there is the chance to “put it all together.”


But when will you do that? It is necessary at times to give uninterrupted concentrated thinking to a patient’s symptoms and signs. For example, how about the chronic headache patient who saw the chiropractor for neck pain? When you put it all together you might think of vertebral dissection.


Some call this a “cognitive pause,” others just call it medical decision-making, and others focus on the disruptive effect of interruptions.


Many ED groups routinely expect charting to be done at the end of the shift, after the patients are gone.


But charting is a chance to think critically, to put it all together. And sometimes when we do that early in the visit, it can prompt a “lightbulb” moment where we realize the need to check something else.


I advocate real time charting. The act of creating a chart requires thinking. Why not do that while the patient is still in the ED? I think at the end of the visit just before discharge is okay but even better would be right after seeing the patient. The recitation of the symptoms and signs are most accurate at that time and the “cognitive pause” of thinking through the whole presentation then can happen early, when it can change the workup.


It is simply impossible to do that on all patients on all shifts. But this is something we should try to do. The bottom line is that most historical and physical findings are ambiguous. There needs to be an explicit stage after information collection, which is information “integration.”


Take Home Points

-Complex patients require a “cognitive pause”

-It is hard to do that on a busy shift but real-time charting makes it more possible



Check out Mark Jaben EP Monthly April 2013

To Reduce Medical Errors, Take a Cognitive Pause