Journal Logo

Diagnosis Deconstructed

Diagnosis Deconstructed: Fever with a Source in the N of 1

Morchi, Ravi MD

Author Information
doi: 10.1097/01.EEM.0000446058.27113.99

    Chart review. Looking back at her labs, I can see where it went wrong and what could have saved her. Just seven days after those labs were sent, she would present with severe sepsis requiring transfer to a tertiary center. And soon after, her life would end.

    First visit. She claimed she had a fever at home to 100.4°F axillary. No temperature elevation in the ED. Clinicians believed she had an infection localized to the urine, but her urine was not all that convincing on my retrospective review. Pyuria, yes. But nitrite negative and no bacteria. Urine cultures would brew in the lab and show no growth days later. We think it is a harmless thing to add a urinalysis to a subjectively or objectively febrile patient for source localization. Unless it returns somewhere between negative and positive.

    The question at the bedside should be, what is my pretest suspicion of a UTI? Does she have symptoms of cystitis? CVA percussion tenderness for pyelonephritis? We can be misled by what actually represents sterile pyuria or glomerulonephritis if we put our weight on urinalyses ordered reflexively and built on poor clinical foundation. It is difficult to look back at our clinical assessment in an unbiased way once the results return.

    CBC. The total white count was normal, and she was discharged immediately after. Had they waited a little longer or had the lab been a little quicker, they would have seen the differential: 40% bands. It is unfortunate that things unfolded the way they did, but as clinicians we cannot always control the turnover of labs. What we can control is our interpretation of them. What did they think of the results that were available?

    Platelet count of 44. I find new thrombocytopenia to suggest an intravascular inflammatory response in a potentially febrile patient. Endothelium may be injured and platelets huddled together over the top of the exposed collagen somewhere in the arterial or venous tree, eventually linked together by mature fibrin strands. Circulating platelet counts then look low on an antecubital vein blood draw. The source of her marginalized and clumped platelets? Bacteremia is first on the list. But we should not forget parasitemia in the case of the recent traveler and microangiopathic hemolytic processes.

    PMHx. She had DiGeorge syndrome and truncus arteriosus repaired years ago with a bio-prosthetic valve and conduit connecting the right ventricle to the pulmonary artery. A percutaneous valved stent had been deployed within the conduit in the catheterization laboratory years later because of ensuing stenosis. A 3 of 6 systolic murmur across her precordium was her baseline.

    My guess: they thought her low platelets went along with her valved stent. Artificial valves known to shear RBCs and even consume platelets clumping over top their leaflets. But it would have been clear that this was a new phenomenon, not one ascribed to her valve, if they had prior platelet counts available.

    Duke's criteria for endocarditis. The two major criteria will not be met in most stable ED patients. Blood culture results will not be available to us. Many perfusing candidates are not getting emergent formal echocardiograms to find a characteristic vegetation. So we are left with accumulating minor criteria to make the diagnosis of probable endocarditis. And the most common of those criteria are fever and a predisposing condition.

    Could she have had even more? If we look closely and weigh the diagnosis highly in our mind, then we will inspect her palms and soles, her fingers and toes. Embolic or immunologic findings only found if we make a concerted effort, if we anticipate their existence. But do we actually look? Do we as a field teach the art of looking? Of all the minor criteria that could have changed the course of her life, were some present under her socks? Unfortunately, her ED extremity exam was documented no differently from others: checkboxes and auto-population stating “warm, no cyanosis, no edema.”

    Suspicion would mean looking with a purpose. It means a deeper palpation while in the right lateral decubitus position for that spleen tip. It means appreciating urine RBCs and WBCs without bacteria as glomerulonephritis or sterile pyuria of a systemic bacterial process, not a UTI.

    Seven days later, her second visit was for severe sepsis with lesions scattered over her pulmonary parenchyma. A subspecialty consultant would proclaim a historic unlikelihood of infection in this particular kind of prosthetic valved stent.

    “Four percent.” But is that logic valid? Is it worthwhile for our consultant colleagues to invoke percentages and epidemiological characteristics of a general population when we are faced with an N of 1 defined by clear objective physical abnormalities? The single patient before us does not have a three or four percent chance of valved stent infection. She does not even have a 25 or 60 percent chance. She either has an infected prosthesis or she does not. It is either 100 percent or 0.

    The real question for us as first-line defenders is a much more specific one: What is the likelihood of an infected artificial pulmonary conduit in a 29-year-old with fever, tachycardia, tachypnea, leukocytosis, a 3/6 systolic murmur over the RV outflow tract, sterile pyuria, and scattered pulmonary lesions suspicious for septic pulmonary emboli? What is the chance of an infection in this particular patient?

    Broad-spectrum antibiotics were initiated on her second visit, and she was admitted to the ICU. Blood cultures grew Staphylococcus aureus. Shortly afterwards, she was transferred to the home institution tertiary center where her care had been handled since childhood. She died despite maximal medical efforts.

    What could have saved her? I do not endorse admitting all well-appearing patients for endocarditis on the first visit simply because they have a predisposing condition and a self-reported, low-grade fever prior to arrival. I think such a recommendation in her case would be driven by my retrospective bias and a lack of appreciation for the subjective assessment of doctors caring for her on that day. If the thrombocytopenia is not recognized as new, the bandemia is not available or not seen, and clinicians proceed with discharging her, at the least she should have blood cultures sitting in our lab and a reliable outpatient follow-up mechanism in place for fever without a clinically sound source.

    Click and Connect!Access the links in EMN by reading this issue on our website or in our iPad app, both available


    © 2014 by Lippincott Williams & Wilkins