EDITOR’S NOTE: This is the final installment in a two-part series on the enigma of sepsis. You can read Part I here.
As I thought about writing this article, I was at first going to propose that we figure out a way to marry the old sepsis definition and coding schema with the new sepsis definition, but I came to realize that I agree with Sepsis-3. I realized that the factors that made me identify a patient as SICK was organ dysfunction. You can be sick from a pneumonia or a UTI, but when you start getting encephalopathic or experiencing acute kidney or liver dysfunction/failure or have a Type 2 MI with elevated troponin, that is when the provider can recognize you have crossed the threshold to SICK and septic. Your providers probably are like me – their gut is also telling them that the underlying infection has caused a systemic cascade, manifesting in organ dysfunction, which might not have risen to the level of “failure.”
In 2016, the Third International Consensus Definitions for Sepsis and Septic Shock, Sepsis-3, in JAMA defined sepsis as a “life-threatening organ dysfunction caused by a dysregulated host response to infection.” They proffered an increase in the Sequential Organ Failure Assessment (SOFA) score of ≥ 2 as the grounds to make the diagnosis of organ dysfunction. Since several of the variables are dependent on laboratory tests, which may not be immediately available (e.g., platelet count, serum bilirubin, and serum creatinine), a new measure with predictive validity was sought. The qSOFA (“q” for quick) score was derived to allow application at the non-ICU bedside, with the intent to identify patients at risk of dying. The variables are altered mentation, systolic BP ≤ 100 mm Hg, and RR ≥ 22/min; demonstrating two out of three variables is positive.
(Sidebar: When I teach this to residents, I always point out that the first derangement is elevation of respiratory rate. Respiratory rates as recorded are notoriously inaccurate unless they are machine-derived. I would strongly suggest counting the respirations yourself.)
Sepsis-3 goes on to define septic shock as a subset of sepsis in which “profound circulatory, cellular, and metabolic abnormalities increase mortality risk.” The authors then conclude with recommendations for ICD coding. Those of us reading this subsection who are familiar with coding guidelines found ourselves scratching our heads in utter amazement. Their recommendation is to code R65.20, Severe sepsis without septic shock for the terminology of “sepsis,” and R65.21, Severe sepsis with septic shock for the diagnosis of “septic shock.” This is not compliant. Documentation must match the ICD-10-CM code, or at least be able to index to it. “Sepsis” cannot be capriciously coded as “severe sepsis,” although sepsis with sepsis-related organ dysfunction can compliantly be coded as severe sepsis. Semantically, Sepsis-3 abolished “severe sepsis,” but practically speaking, it eliminated recognition of cases of early sepsis prior to onset of organ dysfunction. This has set up our collective clinician discomfort at diagnosing, documenting, and coding sepsis in the Sepsis-3 world. We feel like the sepsis continuum isn’t continuous or complete anymore.
The addition of a comorbid condition or complication (CC) or major CC (MCC) does not change DRGs 871 and 870, there being no “with CC” or “with MCC” variants. As far as I can tell, if sepsis establishes your DRG, the addition of severe sepsis alone (without accompanying organ failure codes) does not change the relative weight or severity of illness/risk of mortality (SOI/ROM). However, if the sepsis is not present on admission, and the underlying infection is establishing the DRG, the addition of severe sepsis on top of sepsis may change the SOI/ROM, which could be significant. This is another reason why the elimination of the diagnosis of “severe sepsis” (that is, only having sepsis and septic shock as diagnostic options) is unacceptable unless they change the coding implications to reflect the clinical reality.
The last piece of this puzzle is the SEP-1 Centers for Medicare & Medicaid Services (CMS) core measure. They still define sepsis as 2/4 SIRS plus suspected infection, and severe sepsis as having an elevated lactate or (sepsis-related) organ dysfunction. There are specific, time-dependent actions one must take for a patient with severe sepsis or septic shock, but the included population is inpatients age 18 and over with an ICD-10-CM principal or secondary diagnosis of sepsis, severe sepsis, or septic shock. The intent is “to facilitate the efficient, effective, and timely delivery of high-quality sepsis care,” thereby reducing use of resources and decreasing mortality. Most facilities have implemented some type of sepsis order set or bundle to ensure compliance with the core measure, and the outcome has been absolute mortality reduction. It is odd to have the definitions for the core measure not correlate with the definitions in the most current guidelines.
The criteria as put forth by the Third International Consensus have not been universally embraced by the medical community yet. Here is the apparent dilemma:
- If we continue to use Sepsis-2, we will have three categories: sepsis, severe sepsis, and septic shock. The auditors will have a field day with denials.
- If we transition to Sepsis-3, as it stands, we will compliantly only be able to code two choices if they are documented purely as sepsis or septic shock. This will fundamentally change the composition of the sepsis spectrum pie chart and wreak havoc with SEP-1.
Factors include the following:
- Auditors will use whatever criteria fit their narrative. It is likely they will jump on the Sepsis-3 bandwagon and find opportunities to downgrade the sepsis DRGs, MS-DRGs 871/870, to the underlying infection if there is no R65.20 or R65.21 or organ dysfunction as a secondary diagnosis.
Any denials stemming from an admission prior to Feb. 23, 2016 referencing Sepsis-3 should be forcefully rejected on the grounds that Sepsis-2 criteria were all we had available to use at the time. Providers are expected to be omniscient, not prescient.
- Providers are not reliably documenting organ dysfunction currently. In clinical documentation improvement (CDI), we work hard to transition folks from “altered mentation” to “encephalopathy” to capture the severity of illness the provider was trying to convey. When you notice the creatinine has bumped, or the INR is high, what you are thinking is that there is acute renal or liver dysfunction. No one ever diagnoses critical illness polyneuropathy or myopathy, and no one thinks of ileus as an organ dysfunction.
So what should we do? The issues are what constitutes sepsis and how to diagnose it, how to document it, and then how to compliantly code it.
- Your facility/organization/system should have a discussion about sepsis and set up internal practice guidelines. The goal should be to recognize sepsis as early as possible, but truthfully. The objective is to save lives, not to maximize revenue.
- Convene your infectious disease physicians, critical care specialists, hospitalists, emergency physicians, and any other providers who might want to participate in hashing out the internal practice guidelines.
- If your medical staff does not buy into the Sepsis-3 criteria, have a written guideline indicating what they believe constitutes sepsis (especially without organ dysfunction) and how to diagnose it. Your providers may state that they believe the criteria from Sepsis-2 are still valid. I advise the guidelines to specify that the patient should be sick – just meeting SIRS criteria is not sufficient. However, if at some point the Sepsis-3 criteria become the prevailing definition for a reasonable, prudent clinician in your locale, you will have to revisit this policy.
- Be sure that they understand that no matter how they choose to define sepsis, they will still be held to any core measures set by CMS.
- Your providers must understand that the coding guidelines state that “a provider’s statement that a patient has a condition is sufficient.” If they think a patient is septic, they should document it and supply their clinical evidence to support their diagnosis.
- It is better to consider sepsis early and rule it out than to miss it initially and have unwelcomed outcomes. You can make an uncertain diagnosis (rule-out, possible, probable, suspected) and then figure it out. If it gets ruled out, it isn’t coded. Diagnose, evolve, resolve, recap.
If the diagnosis was sepsis, the record should be consistent (i.e., not one provider only documenting the diagnosis of UTI and another documenting sepsis due to UTI), and the diagnosis should appear in the discharge summary. It is quite suspect to have a principal diagnosis that is missing from the discharge summary.
- Empower clinical documentation integrity specialists to query for clinical validity if they do not find support for a sepsis diagnosis. It is easier to get it right on the front end than to appeal a denial on the back end.
- The qSOFA is meant to provide simple bedside criteria to identify infected patients with the potential for poor outcomes. This is done to prognosticate and direct transfer to the intensive care setting.
- R41.82, Altered mental status, unspecified or R40.4, Transient alteration of awareness are suboptimal ICD-10-CM codes, even if they support qSOFA. Try to educate your physicians to recognize, document, and code encephalopathy.
- A blood pressure of less than 100 mm Hg may be relative or actual hypotension, depending on the patient’s baseline and the absolute number. Precision of language gets you the most specific codes.
- There are many ways to express elevated lactic acid as a consequence of tissue hypoperfusion (“hyperlactatemia” is the term found in Sepsis-3). Your providers either need to document “lactic acidosis” or “excessive lacticemia” to index to E87.2, Acidosis. This latter term is definitely not doctor-friendly. You are not going to be able to train your docs to spew that verbiage. One option is to figure out a way to make an acronym expansion to get there.
- If you are using Sepsis-3, I posit that it is possible to have “organ dysfunction” without frank organ failure. This may result in some “other specified diseases of (insert organ here)” instead of the organ failure code. If an auditor claims there is no sepsis because there is no organ failure code, but your physician documented “organ dysfunction,” fight it.
- Leverage your electronic medical record (EMR).
- If your providers insist on uniting Sepsis-2 and Sepsis-3, you could have three check-off boxes available to them:
□ Sepsis (per Sepsis-2 criteria, no organ dysfunction noted),
□ Sepsis (per Sepsis-3 criteria, with organ dysfunction consistent with severe sepsis by Sepsis-2 criteria), and
□ Septic shock.
Let them tick off the one that describes the patient’s condition.
- Give them a list of organ dysfunctions to choose from to support their diagnosis. They will find more often than not that the patient did meet Sepsis-3 criteria.
- Make sure your sepsis bundle is helpful and guides them to the necessary SEP-1 measures and to get the mandatory data points.
- Be aware of exclusion criteria for SEP-1. If it is in the patient’s best interest to be transitioned to comfort care, it is in the provider’s best interest to do it within three hours of presentation of severe sepsis, and within six hours of presentation of septic shock.
My advice is to have your providers go with their gut. When a patient is SICK from a presumed or confirmed infection, recognize sepsis and document: Sepsis with organ dysfunction. This provides clarity for anyone reading it – the patient meets Sepsis-3 criteria, and the coder can compliantly code R65.20, demonstrating severity. List and support the organ dysfunction, and don’t forget encephalopathy, acute kidney injury, lactic acidosis from tissue hypoperfusion, hyperbilirubinemia or coagulopathy indicating acute liver dysfunction or failure, Type 2 myocardial infarction, hypoxemia or acute respiratory failure, and critical illness polyneuropathy or myopathy.
Sepsis is a very serious condition, and it is vital that we diagnose it and treat it in a timely fashion. In terms of coding to get credit for having taken care of it, my expectation is that we will not be left in limbo forever. I have it on good authority (Jim Kennedy, MD, CCS, CCDS, CDIP) that Coding Clinic will eventually render an official opinion on how to handle sepsis and Sepsis-3. Until then, my final recommendation is the same as it is for any topic:
Do the right thing for the patient, document so you make the patient appear as sick in the EMR as he or she looks in real life, and let the quality metrics and reimbursement fall where they may.