Retrospective Diagnoses: Discharging Sepsis Patients?

Dr. Ronald Hirsch recently had me read an article (Prevalence, Characteristics, and Outcomes of Emergency Department Discharge Among Patients with Sepsis) regarding discharging patients with sepsis from the emergency department (ED). Investigators identified subjects who had had cultures obtained and received antibiotics, presumably for an infection. The charts were then analyzed to see if Sequential Organ Failure Assessment (SOFA) criteria were met. If so, the authors concluded that the patient had sepsis. They then evaluated the patient’s disposition.

Their conclusion was that 16 percent of patients they had deemed as having sepsis were discharged from the ED, and those patients were more likely to be younger and less ill – and their localized infection was most likely to be a urinary tract infection (UTI).

I was happy to see that the study design used Sepsis-3 criteria, as opposed to Systemic Inflammatory Response Syndrome (SIRS), but the flaw in the study is that sepsis is a clinical diagnosis. Just having organ dysfunction concurrently with an infection does not de facto indicate sepsis. The provider must believe that the organ dysfunction is from sepsis. If a patient has acute kidney injury from dehydration or hypoxia, believed to be due to the localized infection of the lung (i.e., pneumonia), then the organ dysfunction is not considered to be sepsis-related and is thus not sepsis-defined.

Clinicians are best positioned to draw the conclusion of sepsis association at the time of the patient encounter. Retrospectively, it may not be clear what the etiology of the organ dysfunction is, unless the documentation specifies it. And if providers consistently produced that degree of documentation, I would be out of a job.

In my emergency medicine career, I am sure I missed cases of sepsis, but I am equally sure that I never knowingly or intentionally discharged a patient home with what I believed to be sepsis. Barring an unusually or unreasonably long ED stay, if the patient turned around promptly after a bolus of fluids and an initial dose of antibiotics and their organ dysfunction resolved, I would rescind my original impression of sepsis and downgrade that patient to a localized infection. In that case, I might send the patient home from the ED with continued antibiotics and close follow-up.

I used to think that even organic brain syndrome (OBS) was not an appropriate disposition for a patient who had true sepsis. I have since decided that in certain limited circumstances, a patient could be dispositioned to OBS with sepsis. The acceptable scenario would include a patient without underlying comorbidities, with reasonable hemodynamics, an uncomplicated source, and rapid resolution of what I was interpreting as sepsis-related organ dysfunction, like altered mental status or acute kidney injury (AKI).

This article illustrates why coders and clinical documentation integrity specialists (CDISs) are not allowed to presume a patient has a condition based on clinical criteria. They need the clinician to assert clinical significance. If there is an infiltrate on a chest X-ray, but no diagnosis of pneumonia…a query must be generated. Say that creatinine or troponin levels are elevated; is there a clinical diagnosis that corresponds?

Even another clinician, like the investigators of this study, or me, when I am performing a chart review, is not able to conclude sepsis after the fact. Being “Sick” with a capital “S” is inherent in the diagnosis of sepsis, and that takes clinical acumen leveraged in real time. Retrospectively, we may be able to intuit that there potentially was sepsis, but it would take inquiry of the treating physician to establish it as a definitive diagnosis.

If a CDIS questions the possibility of sepsis, an early query is advisable. It is not a coding and billing issue; it is for optimizing the patient’s medical care. Posthumous diagnosis of sepsis may get the patient in the correct DRG, but it is better to recognize and treat it aggressively, averting death.

Uncomplicated UTIs rarely meet medical necessity for admission. If there is organ dysfunction, that would be considered to be complicating the UTI. If treatment resolves the dysfunction, you are back to just a UTI. Discharge might be the appropriate disposition for a UTI, but not for true sepsis.

(If you read the original study, I also recommend you read the invited commentary, Elucidating the Spectrum of Disease Severity Encompassed by Sepsis, by Rhee and Klompas. It affirms my skepticism.)

Programming Note: Listen to Dr. Remer every Tuesday on Talk Ten Tuesdays when she cohosts the broadcast with Chuck Buck at 10 Eastern.