Home  |   Subscribe  |   Resources  |   Reprints  |   Writers' Guidelines

August 3, 2009

Toxic Mix — Coding for Sepsis and SIRS
By Robert S. Gold, MD
For The Record
Vol. 21 No. 15 P. 20

Logic says current definitions can lead to over reflection of severity.

When the pharmaceutical company Eli Lilly asked for the 995.9x series of ICD-9-CM codes so it would have a valid code that would provide medical necessity for the use of Xigris, it led to a breakthrough in the confusion between sepsis and septicemia. In their wisdom, the cooperative parties (the AHIMA, the American Hospital Association, the Centers for Medicare & Medicaid Services, and the National Center for Health Statistics) worked with the drug company to develop code sets for infectious and noninfectious sources of systemic inflammatory response syndrome (SIRS). They started out properly, defining the fifth digit of 0 to indicate SIRS without defining whether it was due to an infectious source, the 1 and 2 for infectious sources and the 3 and 4 for noninfectious sources. All was well.

Originally, 995.91 and 995.92 were named “SIRS due to infectious source without” and “with organ dysfunction,” where dysfunction was to indicate organ failure. This was later determined to be less than appropriate. Thus, the definitions were changed to “sepsis” and “severe sepsis,” which would have been a wise decision had the secondary terminology not been retained.

SIRS criteria had been defined by the Society of Critical Care Medicine and the American College of Chest Physicians as the following:

• a fever of more than 38˚C or less than 36˚C;

• a heart rate of more than 90 beats per minute;

• a respiratory rate of more than 20 breaths per minute or a PaCO2 level of less than 32 mm Hg; and

• an abnormal white blood cell count (more than 12,000/µL or less than 4,000/µL or more than 10% bands).

Officials stated that patients who had two of these four criteria in the proper clinical setting may have SIRS. The goal was to demonstrate the start of a potential continuum of biochemical events that could progress to multiple system organ failure and death if appropriate intervention did not take place to support the patient. However, all the work was based on patients in critical care units. Certainly, when the critically ill patient had two of the criteria for SIRS in the face of an infection, the patient likely had sepsis.

Then the governing bodies that establish coding rules and guidelines stated that anyone with an infection, documentation indicating SIRS, and two of the four criteria has sepsis and should be assigned code 995.91 or 995.92. Further rules of sequencing advise that 038.xx must be sequenced first whenever sepsis or severe sepsis is assigned, followed by the 995.9x code and the code for the specific infection. Unfortunately, that may be inappropriate in the clinical circumstances.

When a patient has an infection, it is likely to be accompanied by two of the four criteria indicating SIRS. Much of today’s medical literature agrees that this is only part of the infection and alone does not mean that the patient has sepsis.

Almost all patients who have an infection—whether they be children, adults, inpatients, outpatients, with indwelling catheters or without indwelling catheters—have two of the four SIRS criteria. Regardless, they do not all have sepsis.

But because of the vagaries of the terminology (sepsis, severe sepsis, SIRS, bacteremia), coders have been led to assign inappropriate code sets based on documentation of a “condition” that probably shouldn’t have been documented in the first place. They are told to assign the 995.9x codes in cases in which an infection exists, validated where two of the four criteria are documented, and where SIRS appears in the chart. In turn, that makes them assign 038.xx. As a result, companies that teach hospitals how to maximize Medicare payments or hospitals whose goal is to increase revenues instruct their physicians to document SIRS in every case with an infection that demonstrates two of the four criteria in order to assign the case to a septicemia diagnosis-related group (DRG) despite the fact that the patient is “awake, alert, in no distress” and has no signs or symptoms of sepsis. This is incorrect and must be fixed.

As stated in a 1998 article in Chest: “The practical problems associated with the SIRS criteria are amply demonstrated in the study by Bossink and colleagues. Fully, 95% of unselected patients admitted to a general medical service met the clinical criteria for SIRS (two or more of the following: fever, tachycardia, tachypnea, or leukocytosis). Give any of these patients a positive urine culture and they would meet the consensus definition for sepsis. In a similar study in a South African ICU population of noncranial trauma, 87.8% of 450 patients met SIRS criteria within the first 24 hours of hospital admission. What is the discriminatory power of a categorical definition that includes the vast majority of medical and surgical inpatients? The definition is much too sensitive and insufficiently specific to be of clinical utility. The definition of sepsis also is overly sensitive and does not meet the original expectations of the consensus committee members. The definition of severe sepsis (sepsis criteria with evidence of end-organ injury in one or more systems) comes closer to the traditional notion of the ‘septic patient’ to many practicing clinicians. Even the committee chairperson of the original consensus conference, Roger Bone, MD, FCCP, acknowledged that the initial definitions do not account for the complexities of the variable nature of regulatory and counterregulatory processes that occur in sepsis over time. The framers of the original definitions intended their proposals to be provocative and to be modifiable as new information becomes available which could improve upon or replace the consensus definitions. It has become increasingly clear that SIRS does not delineate a patient population that is particularly useful in the care of patients. It approximates saying that a patient is ‘sick.’ A great deal more basic and clinical research will be necessary to define more tightly a patient group likely to respond to a treatment intervention in a predictable and uniform manner. The term ‘SIRS’ will need to be modified or replaced by groupings which more accurately reflect the nature of the patient’s immune status in the face of critical illness.”

Further, according to Critical Care and Intensive Care Medicine, SIRS is being separated from sepsis, and there is a movement within the profession to differentiate between these conditions through the use of C-reactive protein and procalcitonin assay where rising titers of procalcitonin will define worsening sepsis, indicating progress toward septic shock. Neopterin assay (a chemical from those macrophages) may be helpful, as well as testing for endotoxin itself.

Current literature states that the four criteria are inherent in infections and should not be documented separately by physicians. It goes on to say that SIRS, in infectious processes, is inherent in the process, but SIRS due to a noninfectious source needs documentation to pick up the presence of that condition. General coding guidelines recommend that coders should not assign a code for a condition that is uniformly inherent in a disease process—and SIRS is uniformly present (except if the patient is immunocompromised) in simple bacterial, viral, and fungal infections.

Based on this information, it may be wise to change the terminology and guidance of assigning codes to SIRS. To specifically rely on the identification of SIRS criteria to label a patient population with sepsis is faulty and not truly representative of a patient’s true illness. As a result, the data will be faulty, and the industry will continue to inappropriately pay billions of dollars into the program to support healthy patients who have been coded to sepsis at the expense of sicker patients with other conditions (because Medicare is a budget-neutral system).

As recovery audit contractors begin to closely monitor billing practices, healthcare organizations that follow guidelines established using faulty logic can expect to pay back billions of dollars for overcoding. That’s the danger of following the current sepsis rules.

How can the rules governing the coding of sepsis be corrected to more fairly reflect a patient’s condition? Change 995.91 and 995.92 to sepsis and severe sepsis and eliminate alternate terminology that refers to SIRS. Have notification that SIRS is inherent in infectious processes and should not be coded separately. Leave 995.93 and 995.94 alone, but consider that multisystem organ failure related to infection be an acceptable term to define 995.92 and related to noninfection be an acceptable term to define 995.94, although the physician should name the failed organs.

Although the 038.xx series is for septicemia, it is somewhat understandable that sepsis patients should be assigned that series of codes. After all, for decades, there was not a code for sepsis and, regardless of how many people said sepsis and septicemia are two different animals, all of the cases were assigned to septicemia DRGs. In order to avoid losing the integrity of trackable data, that data stream should probably be maintained.

But septicemia is the infection of the bloodstream with organisms; it doesn’t matter what kind of organism, whether it be bacteria, fungi, protozoans, or viruses. The term “septi” refers to infectious organisms. Strictly speaking, it does not refer to the incidental presence of these organisms in the bloodstream due to contamination from an outside source such as a transrectal prostate biopsy, periodontal surgery, or any procedure that is likely to cause a transient presence of organisms in the bloodstream without true infection. In most cases, these organisms will cause a single episode of symptoms and, in the immunocompetent patient, they will be totally cleared from the bloodstream. Septicemia is caused by the actual organism in the bloodstream, whether or not it can be retrieved by blood culture techniques.

Sepsis is an immune response caused by cell walls of organisms or chemicals released by organisms in an infectious source somewhere in the body. This leads to endotoxin, exotoxin, or some other chemical circulating in the bloodstream that stimulates macrophages to release chemicals known as proinflammatory cytokines. These chemicals cause the physiologic processes that produce the four criteria of SIRS, which can lead to organ failure and death. Other mechanisms of body injury can cause the macrophages to release the same chemicals—these being noninfectious causes of the systemic inflammatory response, such as necrotizing pancreatitis, body burns, or anaphylactic reactions. There is no implication that there are organisms in the bloodstream.

Perhaps in the future, sepsis codes will be allowed to stand on their own without assigning a code for septicemia when that condition does not exist.

— Robert S. Gold, MD, is CEO of DCBA, Inc and a member of the For The Record editorial advisory board.