Abstract
Sepsis has continued to climb the charts as one of the most frequent principal diagnoses for hospitalizations in the US and one of the most expensive conditions to treat in the nation’s hospitals. It is unsurprising that it warrants additional scrutiny by payers and also remains one of the most frequently denied diagnoses. Challenges arise in sepsis billing due to the variety of definitions of the condition and changing clinical indicators impacting documentation and coding. This article reviews the literature related to the diagnosis, documentation, coding, and billing of sepsis since the more widespread implementation of the Sepsis-3 definition in 2017 to outline the challenges and recommendations discussed by industry leaders. Addressing accurate sepsis diagnosis and reimbursement relies on clear organizational policies, accurate and helpful tools, education and training, and consistent denial management.
Keywords: sepsis, septicemia, coding, denials, CDI
Introduction
Sepsis has been recognized as a leading cause of death and significant financial burden with incidence increasing annually, with some claiming that it may still be underrecognized and underreported.1 This may be due, in part, to the lack of a universal clinical definition, challenges in coding sepsis, and regular denials of sepsis on claims.2 The clinical criteria remain complicated and ambiguous, without clear biological, imaging, or laboratory features to uniquely identify a septic patient.3 Septicemia or severe sepsis with a major complication, Medicare Severity Diagnosis Related Group (MS-DRG) 871, was the most frequently billed MS-DRG in FY 2019, leading many payers to scrutinize sepsis-related claims, reviewing them both prepayment and post-payment.4 There are increasing clinical validation denials due to misaligned sepsis criteria between providers and payers, a lack of clinical indicators and/or documentation, and the additional focus on these types of claims. Denials are expensive and resource-intensive, and inaccurate coding of sepsis in claims data may negatively impact funding and accurate epidemiological representation.5
A Rising Problem
It is clear that the challenge of diagnosing, treating, and billing sepsis are not going away, as the number of sepsis-related admissions and claims continue to rise. Table 1 shows the number of discharges and associated average charges and payments from Medicare inpatient hospitals in 2020, both nationally and within the highest and lowest states. This highlights the increasing number of discharges and the wide range in average submitted charge.
The Healthcare Cost and Utilization Project (HCUP) has further highlighted the burden of sepsis-related admissions and claims amongst all payers. Septicemia was the most frequent principal diagnosis for hospitalizations in the United States in 2018 with 2,218,800 stays representing eight percent of nonmaternal, nonneonatal inpatient stays.6 This number more than doubled in seven years, with just over one million discharges indicated in 2011.7 Sepsis also remained the most expensive condition treated in U.S. hospitals, costing $41.5 billion in aggregate and averaging $18,700 per stay.8 This was up from $38.2 billion (8.8 percent of national costs) in 2017, $23.6 billion (6.2 percent) in 2013, and $20.3 billion (5.2 percent) in 2011.9,10,11 Cost was highest in the West and lower in the Midwest and South.12
Part of this cost may be associated with the high use of the intensive care unit (ICU) for patients diagnosed with sepsis. In 2011, HCUP indicated that MS-DRG 871 was among the most common conditions with the highest proportion of ICU utilization in 29 states at 59 percent. Those without a major complication or comorbidity, MS-DRG 872, were at 27.5 percent.13 A 2012 statistical brief from HCUP that highlighted super-utilizers, those individuals who consume a large share of health care resources, indicated that septicemia was amongst the 10 most common principal diagnoses for these patients across all payers.14 It was also among the top 10 conditions with discharges to post acute care (PAC) in 2013, with MS-DRGs 870-872 accounting for 441,400 discharges to PAC (39.4 percent) with the most going to skilled nursing facilities (53.5 percent).15 Septicemia also accounts for a large portion of readmissions. In 2014, septicemia ranked among the top 20 diagnoses with the highest 7-day (6.7 percent) and 30-day (18.5 percent) readmission rates.16 This was confirmed again in 2018 when hospital stays for septicemia had the highest number of 30-day all-cause readmissions at 314,600 (accounting for 8.3 percent of all readmissions).17
In addition to high cost, septicemia is of particular concern for certain vulnerable populations. According to the 2018 HCUP data, rural areas had the highest rate of stays for septicemia but decreased mean length of stay and cost per stay.18 A 2014 study showed that septicemia was more than 10 times higher among those aged 75+ years than those aged 18 to 44 years, making it the most common diagnosis among that age group (7.6 percent).19 It was also found that among patients readmitted within 30 days of an index stay for septicemia, uninsured patients were more likely than patients with insurance to return within 7 days.20 In 2018, average readmission costs were highest amongst the self-pay/no charge patients and those with Medicaid accounting for 8.6 percent and 9.6 percent of the aggregate, respectively.21,22 As this data shows, sepsis continues to be a prevalent and problematic condition with high costs. It is no surprise that it has warranted increased scrutiny by payers and thus, higher denial rates. It is important that facilities address the diagnosis, documentation, and coding of sepsis to ensure quality patient care and accurate reimbursement.
Scope and Objectives
The scope of this article is twofold: to provide a comprehensive review of the current challenges related to sepsis documentation, coding, and billing and to provide recommendations on how to address these challenges. A search was conducted of peer-reviewed articles published since 2017 related to the diagnosis and billing of sepsis using EBSCOhost database and Google Scholar.
The search was limited to this time period to highlight the challenges since the Sepsis-3 definitions were announced and had begun adoption amongst providers and payers. Search terms were used to identify sources that addressed sepsis definitions, coding sepsis, clinical documentation integrity (CDI) guidelines concerning sepsis, sepsis query recommendations, and sepsis-related claims denial management. Of the more than 200 articles reviewed, the content of nearly 100 specifically referenced the diagnostic and billing challenges of sepsis and associated recommendations. The research team recognizes that much of the expertise in this area is not evident in peer-reviewed journals. Thus, the search was expanded to professional journals and publications, as well as associated blogs and Q&A forums moderated by coding and CDI experts. This included the Journal of AHIMA, ICD-10 Monitor, Coding Clinic, Centers for Medicare & Medicaid Services (CMS) data tables, and Association of Clinical Documentation Integrity Specialists (ACDIS) publications and forums, among others. After screening the sources that met the criteria of the search, 95 sources were thoroughly evaluated for inclusion in this article. In total, 51 were used to synthesize the challenges of coding and billing for sepsis and recommendations to reduce sepsis-related claims denials.
The Challenges
Definitions and Clinical Indicators
According to the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3), sepsis is “life-threatening organ dysfunction caused by a dysregulated host response to infection.”23 This organ dysfunction and dysregulated host response is what differentiates sepsis from other infections.24 However, this definition still raises some confusion and controversy.25 This may be due to the changing definitions of sepsis since 1992 when the Definitions for Sepsis and Organ Failure and Guidelines for the Use of Innovative Therapies in Sepsis first identified what is now referred to as Sepsis-1, based primarily around systemic inflammatory response syndrome (SIRS) with infection. This committee also introduced the concept of a continuum of severity, establishing a model of sepsis, severe sepsis, and septic shock. Septic shock is defined as “a subset of sepsis in which underlying circulatory and cellular metabolism abnormalities are profound enough to substantially increase mortality.”26 The definition was again revised in 2003, expanding the list of findings that may be seen in septic patients and defining sepsis “as the presence of infection and a wide list of general, inflammatory, perfusion, and hemodynamic parameters.”27 These changing definitions shifted the diagnosis of sepsis from being based on SIRS with defined vital signs and laboratory values suggestive of infection to being based on organ dysfunction.28
The Sepsis-3 definition relies on the measures of organ dysfunction, which can be based on a Sequential Organ Failure Assessment (SOFA) score of two points or more. The SOFA score is a mortality prediction score to grade the severity of organ dysfunction in the presence of infection. It is assumed a normal, healthy patient has a baseline SOFA score of zero.29 This scoring model was developed and validated in an ICU setting; when used outside of the ICU it may underestimate organ dysfunction and sepsis.30 Duke et al. found that “the SOFA score had high specificity but poor sensitivity for clinical sepsis,” stating that “clinical reliance on the SOFA score may delay recognition and treatment of patients during early stages of sepsis.”31 One challenge is that there are variable thresholds for defining organ dysfunction. As Rhee et al. states, “For example, there are multiple definitions for acute kidney injury, which differ from thresholds used in the SOFA score and other ICU-based organ dysfunction scores.”32
Furthermore, this definition relies on some knowledge of baseline functioning that may not be known.33 Patients may also be evaluated using the quick SOFA (qSOFA) score, but it has also been studied with mixed results. Arberry et al. found that qSOFA led to an overestimate of sepsis, recommending the use of both qSOFA and SOFA to clearly identify organ dysfunction.34 In a study by Anand et al., of the 271,500 patients who had a positive qSOFA score on admission, only one in six had sepsis pointing to low sensitivity for identifying sepsis.35 Salomao et al. found that the qSOFA lacked sensitivity when compared to the previous SIRS criteria and Evans et al., (2021) recommends “against using qSOFA compared with SIRS, NEWS, or MEWS as a single screening tool for sepsis or septic shock.”36,37 Singer et al. clarifies that “neither qSOFA nor SOFA is intended to be a stand-alone definition of sepsis.”38 Similarly, SIRS is meant as a screening test.39 The qSOFA should be used as the screening protocol within the first hour and then SOFA criteria becomes more credible with findings of organ dysfunction as the patient stay progresses.40
In addition to the challenges related to defining and measuring organ dysfunction, the adoption of different sepsis definitions by various entities has added to the confusion. Some professional groups have declined to endorse the new Sepsis-3 standards, including The American College of Emergency Physicians, Infectious Diseases Society of America, and Latin American Sepsis Institute. CMS has shown preference for Sepsis-2 definitions for review and payment and Sepsis-1 measurement criteria over Sepsis-3.41 However, many other third-party payers have adopted Sepsis-3 criteria and recovery audit contractors (RACs) have begun moving towards Sepsis-3. In a member survey conducted by ACDIS (2020), 65 percent indicated that their facility uses Sepsis-2 criteria and only 15 percent use Sepsis-3 criteria. Physicians are not bound to a particular sepsis definition or set of criteria, and many hospitals are either still using Sepsis-2 criteria or unsure which one to use. This can lead to conflicting evaluations of claims and ultimately result in denials.42
Regardless of the definition or clinical criteria, a sepsis diagnosis is dependent upon the patient story and physician judgment, and can be subjective.43,44 Patients present with ailments that may or may not be infectious and organ dysfunction that may or may not be due to infection, and a septic patient may or may not have a positive blood culture and bacteremia may or may not cause sepsis.45,46 The uncertainties in the diagnosis, scoring models, and clinical presentation can lead to unclear documentation that may negatively impact coding and ultimately lead to a denial. The physician’s documentation must make a clear connection between the abnormal clinical findings that support organ dysfunction and the diagnosis of sepsis.47
Coding and CDI
Coding sepsis can often be as challenging as diagnosing it, with CMS guidelines keeping the “SIRS plus infection plus organ dysfunction” definition that does not clearly align with the new clinical definition of sepsis.48 The AHA Coding Clinic Fourth Quarter 2017 states “a code is assigned when the provider documents sepsis and an associated acute organ dysfunction,” but this not always explicitly clear in physician documentation practice.49 The listed conditions that are identified as representing acute organ dysfunction is not exhaustive and if the documentation is unclear it warrants a query.50 When sepsis is coded, it is most commonly (72 percent of the time) code A41.9, Unspecific Sepsis.51
Additional challenges can arise when coding viral sepsis as ICD-10-CM does not provide a specific viral sepsis code, but rather directs A41.89, Other specified sepsis, as the best option.52 According to a random sample of 2021 fee-for-service claims through the Comprehensive Error Rate Testing (CERT) dataset, incorrect coding was indicated for 9.8 percent of MS-DRG 870 claims, 5.9 percent of MS-DRG 871 claims, and 8.5 percent of MS-DRG 872 claims.53 Additional claim errors were noted for no documentation. This leads to an increased number of queries and possible underreporting. In a study done by Arberry et al., it was found that although 59 percent of the charts audited had evidence of sepsis at admission and 52 percent had sepsis documented somewhere in the notes, sepsis was much less likely to be documented on the discharge summaries (10 percent) or coded (17 percent).54 Another study, by Wilhelms et al., concluded “that 55 percent of critically ill patients with severe sepsis were discharged from hospital without ICD codes that are widely used to identify sepsis.”55 Similar studies have shown how poorly sepsis is coded in administrative data.56
Sepsis raises challenges for CDI professionals as well. According to a 2019 survey by ACDIS, sepsis was the top queried diagnosis for clinical validation.57 Clinical validation is a process by which documentation is evaluated to ensure that the medical record demonstrates enough clinical support for all documented diagnoses as mandated by the False Claims Act.58 A clinical validation query request should be sent whenever there is a lack of clinical support for sepsis within the documentation. Options for specifying sepsis as “ruled in” or “ruled out” should be included in the query choices, in addition to the clinical indicators which support the diagnosis of sepsis during the current encounter.
Quality documentation ensures that anyone reading the medical record after discharge should come to the same conclusion as the providers in regards to the diagnosis of sepsis. Some speculate that the diagnosis of sepsis may be overused by the providers because the result of not treating a case of potential sepsis could be fatal for the patient. “Therefore, anyone who submits a clinical validation query for sepsis should consider the option of prophylactic or empiric treatment as a valid choice for those cases with little to no clinical support in the medical record.”59 A note from the ACDIS editor in 2019 also indicated that sepsis remained a top search term for the organization’s website, highlighting the challenges and uncertainty that come with its documentation.60
Denials and Payment Reviews
Despite these findings of potential underreporting, MS-DRGs 871 and 872 remain increasingly frequently billed DRGs leading to enhanced scrutiny by payers and challenges in compliance and reimbursement.61 These uncertainties make sepsis a regularly and controversially denied diagnosis.62 In a September 2022 Q&A forum, ACDIS indicated that nearly 70 percent of respondents to interviews of CDI professionals reported that sepsis was one of their top denied diagnoses.63 Most denials are due to a lack of documentation or clear clinical indicators; even when coded based on a definite physician diagnosis, missing clinical indicators can lead to a denial.64,65 This may be due, in part, to payers capitalizing on the gaps between Sepsis-2 and Sepsis-3 clinical criteria.66
It is important that those involved in the revenue cycle check insurance contracts for clear definitions and requirements. The record may not show evidence of impaired homeostasis as evidenced by altered mental status from baseline, hyperglycemia, hypotension, oliguria, coagulopathy, thrombocytopenia, ileus, acute hepatic failure, elevated lactate and capillary mottling, or acute respiratory distress syndrome, among others.67 The payer may indicate “there is no evidence that the patient’s symptoms were due to any localized infection” or may acknowledge the condition as documented, but not “think this was a valid diagnosis.”68 The record may lack laboratory findings that support SOFA indicators or clinical evidence of SOFA criteria. Denials have included reasonings such as “no mention of toxic in appearance” or “no positive blood culture.” The treatment plan is also considered and may result in a denial if it does not reflect the greater levels of monitoring and intervention required to treat sepsis.69 Continued issues in the documentation and coding of sepsis will only lead to increased denials and audits.70
Recommendations
In order to accurately diagnosis and bill for sepsis, facilities must identify their sepsis issues and their impact on the facility. This includes identifying the prevalence of sepsis and sepsis-related issues (such as denials) in their patient population by considering community demographics, their payer mix, claims and quality reporting data, and benchmarks. Costs can be identified through various reports and payer contracts, and current medical definitions should be clarified. Once the problem has been clarified, there are strategies to improve sepsis diagnosis, documentation, coding, and billing. A consistent, system-wide approach should be used utilizing a sepsis team that may include medicine and nursing, laboratory and pharmacy, billing and admitting, coding, CDI, and quality assurance.71
Policies and Tools
Facilities need to clarify which sepsis definitions and criteria are being used to ensure consistency between providers.72 Even if it is decided to use different criteria than the payer, having clear definitions and documentation guidelines can support an argument for denial by highlighting standardized internal criteria.73,74 Institutional definitions provide consistency and clarity for providers, coders, and CDI staff in documentation, education, clinical validation, and the query process.75 Including a works cited with facility definitions helps verify that the criteria is from the industry, not just an individual facility.76
A consensus statement should define sepsis and severe sepsis, and provide guidance regarding documentation of “early sepsis” or “meets sepsis criteria.”77 Facilities should review and update sepsis screening policies, intake forms, and treatment protocols utilizing a multidisciplinary group of medical professionals to ensure they meet the designated definitions.78 Working together to clarify and update the definitions, policies, and tools encourages conversation across multiple departments and helps the facility establish common standards and consistency.79
Diagnostic criteria should be established for the organization and communicated to clinical staff for accurate recognition and reporting of sepsis based on infection and organ dysfunction.80 Facility-defined clinical indicators including sepsis screening criteria, definitions, and SOFA criteria should be clearly communicated and readily available to clinical staff .81 This could be incorporated into a template, such as a discharge summary template, to provide clinicians with an aide and encourage thorough documentation. This can then be used by the coders as a clear indication of sepsis.82 Arberry et al. found that implementing such a template improved sepsis documentation by 28 percent.83 Organizations such as ACDIS provide forms and tools such as query templates. Some electronic health record vendors, such as Epic, provide dropdown menus that guide more specific documentation and/or scoring tools such as a SOFA score calculator.84 Facilities may also consider creating a care process model like the one created by a workgroup at Intermountain Healthcare that provides algorithms for diagnosing and treating sepsis and septic shocks, outlines important definitions and clinical characteristics, and includes an adult sepsis bundle worksheet and tidal volume conversion tables.85 Such algorithms can clearly show the clinical indicators that led to the diagnosis of sepsis. Once facility-wide definitions and criteria are decided upon, a process should be put in place to regularly re-assess and update the criteria to ensure compliance with the latest clinical guidance and coding guidelines.86
Policies and procedures related to the coding and billing of sepsis should also be reviewed. It should be clear when coding and CDI queries are needed to establish sepsis. Coding and CDI staff should work together to ensure clinical indicators clearly support a diagnosis of sepsis, and query when the evidence is insufficient.87 For example, the ICD-10-CM official guidelines recommend querying about a negative or inconclusive blood culture, documentation of “urosepsis,” and any indication of acute organ dysfunction if it is not clear that it is related to sepsis. Competing etiologies can weaken the validity of sepsis and should be outlined in a query to ensure the provider has all key pieces of information.88 Query templates may be helpful to ensure they include the relevant clinical indicators to support the diagnosis of sepsis.89 Organizations may want to implement pre-bill reviews for sepsis stays.90 The facility should establish clear escalation policies and the need for a second-level review for sepsis claims and denials. Rebutting a denial should be supported by a comprehensive appeal letter. A standard appeal template may be helpful, particularly for repeated denials related to a disagreement between sepsis definitions and clinical criteria.91
Payer contracts should specify which definition of sepsis and diagnostic criteria the payer is using. As changes are made to payment policies and contracts with payers, these changes need to be communicated to providers and revenue cycle staff. Such contracts should outline the desired clinical information, facts, and context needed to support a diagnosis of sepsis.92 For example, if an organization decides to use the SIRS criteria, this should be reflected in the contract so if denials arise due to a conflict between the SIRS criteria and Sepsis-3 criteria, the facility to can refer back to the contract in their appeal letter.93 Compare payer denial activity to contract terms and identify issues. If sepsis claims are continuing to be denied even when agreed upon criteria are being used, there could be an error occurring such as a disconnect with a third-party reviewer. Track contact requirements to ensure they are being met and follow up as needed.94
Education and Training
Clinical staff may need updated training on documentation requirements. They should be encouraged to document a clear clinical picture including signs and symptoms to indicate both the infection and organ dysfunction, tying together observations, indicators, and logic.95,96 An infection without organ failure may sometimes be described as “septic”, but does not actually meet the definition of “sepsis.”97 Documentation needs to identify “sepsis,” “severe sepsis,” or “septic shock,” be clear and consistent, identify the source of infection, and clarify that the sepsis is related to the source infection. Clarifying the relationships can be evident with words like “causing” or “caused by,” “associated with” or “related to,” “from,” “due to,” or “with.”98
The documentation must include a statement like “dysregulated host response to infection” and the association organ dysfunction, such as hypotension, renal failure, encephalopathy, etc.99 General phrases such as “multiple system organ failure” should be avoided.100 Clarifying a clear relationship is important. Rather than stating “elevated creatinine in the setting of sepsis, hypotension,” a statement such as “acute kidney injury, likely due to sepsis with hypotension” shows a stronger link between the conditions. Documentation should also clarify if the infection was or may have been related to a recent surgery or device and if it was present on admission.101
Regardless if the sepsis resolves during an inpatient stay, discharge summary documentation should highlight that sepsis was a condition on admission through consistent documentation by all providers.102,103 If using the Sepsis-3 definition, clinical staff should be encouraged to include the SOFA and/or qSOFA score on every chart where sepsis is a diagnostic consideration.
In addition to clarifying what providers need to include in sepsis documentation, training may also need to be provided on what to avoid or use caution with when documenting. Providers should avoid sepsis-adjacent phrases, such as “urosepsis,” “sepsis-like,” “meets sepsis criteria,” or “sepsis syndrome.”104 The term “urosepsis” is vague and nonspecific; if the patient has sepsis due to a urinary tract infection, that must be clarified in the documentation.105 Even terms such as “septicemia,” “SIRS,” and “septic, toxic” can be problematic because they may describe infection but do not definitively clarify sepsis with organ dysfunction. Documenting a “history of sepsis” does not indicate the condition on the current stay, even if it resolved following admission.106 Although R78.81 Bacteremia will still group to MS-DRGS 870-872, documenting “bacteremia” alone can also be problematic. Providers may use these terms interchangeably, but by the coding definition bacteremia does not usually meet medical necessity for an inpatient admission as it implies the patient is asymptomatic.107 If sepsis is present, documentation should indicate it with statements like “sepsis due to e-colic bacteremia” or “sepsis with positive blood cultures.”108,109
In addition to education and training for clinical staff, a targeted effort should be made to ensure all coding and CDI staff are adequately trained on the new definitions and clinical criteria. Documentation of SIRS with an infection is no longer enough to code sepsis; coders needs to look for sepsis and organ dysfunction.110 Arberry et al. found that additional training for coders improved their ability to capture a sepsis code by 21 percent.111 CDI staff need to look for the progression of infectious symptoms throughout the patient stay and for contradictory notes between the emergency department and attending physician.112 Does the treatment provided support the diagnosis of sepsis? According to DeFilippis, “the best practice for CDI is to know each of the 3 standards and try to ensure that all clinical indicators pre