Health Information Management

Seven savvy tips for coding sepsis and SIRS

JustCoding News: Inpatient, May 23, 2012

Want to receive articles like this one in your inbox? Subscribe to JustCoding News: Inpatient!

Many physicians say that systemic inflammatory response syndrome (SIRS) criteria are insufficient and confusing at best, and don't indicate whether a patient is truly sick, says Robert S. Gold, MD, founder and CEO of DCBA, Inc., in Atlanta.

Some patients—particularly those who are critically ill—may meet necessary criteria for SIRS and truly have sepsis or another severe diagnosis. Others, however, may meet two of four criteria (e.g., heart rate > 90 and respiratory rate > 20)—which technically constitutes a SIRS diagnosis—but not have SIRS.

"Abnormalities in vital signs and abnormalities in laboratory studies can be due to things that are totally unrelated to a patient's infectious process in the body or can be present totally unrelated to an inflammatory process in the body," says Gold. "If there is no inflammatory process, physicians should not call it SIRS because you must have an inflammatory process to get a systemic inflammatory response."

For example, tachycardia with atrial fibrillation and rapid ventricular rate doesn't justify a SIRS diagnosis, says Gold. If a patient has leukocytosis with injection of steroids, this also doesn't imply SIRS. Similarly, tachypnea with tachycardia caused by running does not meet SIRS criteria, he says.

To confuse matters, some patients—particularly those who are immunocompromised—may have sepsis without meeting any criteria, says Gold. "You have to look at the possibility that a patient can be septic and indeed be in septic shock and not have the SIRS criteria met at all," he says. "It's a clinical judgment of the physician in looking at the patient to be able to determine if the patient has a risky infectious process or a risky noninfectious process."

Bridge coder-physician communication gap
So where does all of this information leave coders? Often, between a rock and hard place, says Jennifer E. Avery, CCS, CPC-H, CPC, CPC-I, senior regulatory specialist at HCPro, Inc., in Danvers, Mass. Physicians don't diagnose—or document—consistently, which often leads to inaccurate data capture, she says. Many coders are uneasy coding records in which physicians mention sepsis or SIRS only once or twice without documenting any clear clinical evidence or treatment. They hesitate to code the condition because they know payers may deny the claim, she says.

Coders should take time to more thoroughly review and learn from these records rather than be overwhelmed by them, says Avery.

"Coders have never had to be as clinical as we're challenged to be now," she says. "Coders have the ability to gain some of the pathophysiology knowledge to read the record and be able to abstract what's important."

As with all documentation challenges, coders should emphasize to physicians the importance of capturing patient severity. This includes the following elements related to sepsis and SIRS:

  • The inflammatory condition, whether infectious or noninfectious
  • The causal organism
  • Whether a noninfectious process is contributing to a patient's illness and the specific process

When in doubt, coders should query even though it may seem as if they are questioning physicians' clinical judgment, says Avery.

"I don't think it's really that we're questioning their judgment per se, I think it's more that we're questioning the accuracy of the record," she says. "If the condition is not clearly documented, then we shouldn't be picking it up."

Effective strategies for coding sepsis
In addition, coders should consider the following seven coding tips:

1. Note differences between streptococcal sepsis and streptococcal septicemia. When physicians document streptococcal septicemia, coders should report ICD-9-CM code 038.0 (streptococcal septicemia) only. They should not report code 995.9x (SIRS) as an additional code. Coders also should query physicians to determine whether a patient actually has sepsis instead, in accordance with Official ICD-9-CM Guidelines for Coding and Reporting.

Nonetheless, query with caution, says Avery. "I think we over-query in this area for a condition that's really not there," she says. "I think a lot of physicians over time have become desensitized to it."

Coders should review clinical evidence in the record before querying physicians, she says. They should also be careful when referencing SIRS criteria to avoid backing physicians into a corner to provide diagnoses that may technically satisfy diagnostic criteria without actually being present, she says.

Conversely, if physicians document streptococcal sepsis, coders should report codes 038.0 and 995.91 (SIRS due to infectious process without acute organ dysfunction).

2. Look for links between organ dysfunction/failure and severe sepsis. Severe sepsis (code 995.92) occurs when a patient suffers from sepsis and signs of failure of at least one organ. Physicians must document all organ dysfunctions and failures—including any related treatments (e.g., tracheostomy)—to support the overall diagnosis, says Avery.

However, an acute organ dysfunction must be associated with the sepsis to assign the severe sepsis code, she says.

"Just because a physician identifies a patient has acute respiratory failure or acute renal failure [doesn't mean] that you can jump to a conclusion that it's sepsis," Avery says. Coders should query physicians when documentation is unclear regarding whether acute organ dysfunction is related to sepsis or another medical condition, she says.

3. Know how to apply sequencing guidelines. "[Applying sequencing guidelines is] really clear if a patient comes in with some type of localized infection and then develops sepsis while they're in the hospital," says Avery.

However, sequencing isn't as clear when patients appear to be admitted for ¬sepsis, organ failure, localized infection, or something else, she says. Consider the following suggestions:

  • Sepsis or severe sepsis is POA and meets the definition of a principal diagnosis. Assign a code for the systemic infection (e.g., 038.xx or 112.5 [disseminated or systemic candidiasis]) first, followed by 995.91 or 995.92.
  • The reason for admission is sepsis, severe sepsis, or SIRS and a localized infection (e.g., pneumonia or cellulitis). Assign a code for the systemic infection (e.g., 038.xx or 112.5) first, followed by 995.91 or 995.92, and then a code for the localized infection.
  • A patient is admitted with a localized infection (e.g., pneumonia), but sepsis or SIRS doesn't develop until after admission. Assign a code for the localized infection as the principal diagnosis. Also assign a code for the systemic infection (e.g., 038.xx or 112.5) and code 995.91 or 995.92 as secondary diagnoses.
  • Sepsis or severe sepsis is not POA but ¬develops during the encounter. Assign a code for the systemic infection and code 995.9x both as secondary diagnoses. When signs or symptoms of sepsis are POA but physicians don't document the condition until after admission, the record may justify a query to determine whether sepsis was POA. Official ICD-9-CM Guidelines for Coding and Reporting instruct coders to assign "Y" for conditions diagnosed during an admission that were clearly present but not diagnosed until after admission occurred.

4. Wait for the discharge summary. If a physician documents a diagnosis as probable, suspected, likely, questionable, possible, or still to be ruled out at the time of discharge, coders can report the condition as if it existed or was established. Physicians might document possible sepsis or probable sepsis in the record. If the condition is also documented in the discharge summary, coders can report it as if it existed, says Avery. If it's not included in the discharge summary, clinical evidence in the record might justify a query to possibly confirm the diagnosis, she says.

5. Note unique aspects of coding newborn ¬sepsis. When a physician documents newborn sepsis, coders should report code 771.81 (septicemia [sepsis] of newborn) with a secondary code from category 041.x (bacterial infection in conditions classified elsewhere and of unspecified site) to identify the organism. Coders shouldn't report a code from category 038, nor should they assign code 995.91, says Avery. Conversely, if a newborn has any associated acute organ dysfunction, report 995.92.

Remember that the 770 code series is reserved for conditions that follow the birth process and are directly related to it. These conditions must occur within the first 28 days of life. For example, coders should report 038.x when a baby develops sepsis from bacterial superinfection of a viral pneumonia caused by his 2-year-old sibling.

6. Encourage physicians to stop documenting urosepsis. This vague term currently maps to code 599.0 (UTI, site not specified) in ICD-9-CM. However, in ICD-10-CM, urosepsis is not a codeable term. The Alphabetic Index instructs coders to "code to the condition," and it doesn't provide a default code.

Start encouraging physicians to document greater specificity now, says Avery.

7. Don't make assumptions when coding post-procedural sepsis. "You cannot make an assumption that just because the patient has some type of post-procedure infection that develops into sepsis that [the procedure and sepsis] are related," says Avery. "Physicians must clearly document the cause-and-effect relationship."

If a localized infection is post-procedural and related to an operation, assign a code for the complication (e.g., 998.59, other postoperative infection, or 674.3, other complications of obstetrical surgical wounds) first, followed by the appropriate sepsis codes (i.e., 995.91 or 995.92). Coders should report additional codes for any acute organ dysfunction or failure in cases of severe sepsis. Refer to Coding Clinic, Fourth Quarter 2011, pp. 151-153 for more information.

Editor's note: This information was originally presented during HCPro's audio conference "Sepsis Coding: Learn Documentation Improvement Techniques to Ensure Accurate Coding." This article originally appeared in the May issue of Briefings on Coding Compliance Strategies. Email your questions to Senior Managing Editor Andrea Kraynak, CPC, at

Want to receive articles like this one in your inbox? Subscribe to JustCoding News: Inpatient!

Most Popular