In the September 2017 issue of JAMA Rhee et al. (1) presented the results of a study of sepsis incidence in the United States using clinical vs. administrative data. The article highlights the implications of using administrative data to measure clinical conditions. In clinical data from 409 hospitals, sepsis was present in 6% of adult hospitalizations, and in contrast to claims-based analyses, neither the incidence of sepsis nor the combined outcome of death or discharge to hospice changed significantly between 2009-2014. The findings also suggest that EHR-based clinical data provide more objective estimates than claims-based data for sepsis surveillance.
This is not a surprise for many coders or physicians who understand documentation and coding. An important fact about coding inpatient diagnoses was not mentioned by the article. Uncertain conditions are still required to be coded when the treating physician has a particular degree of suspicion of a diagnosis. For example, a condition stated as “suspected”, “possible” or “likely” is codified in a similar fashion to a confirmed condition using the 10th revision of the International Statistical Classification of Diseases (ICD-10). Such convention of coding is necessary to avoid that hospital admissions be coded as non-specific symptoms (such as fever or hypotension), and to allow a vehicle for reimbursement based on hospital resource consumption. This same convention does not apply to outpatient services, where only confirmed conditions can be coded. As awareness of sepsis has increased over the last decade following studies of sepsis outcomes using Early Goal Directed Therapy, it is not surprising that many more records had to be coded as sepsis. In fact, most population-based studies have used administrative data and identified cases of sepsis via ICD codes, leading to a discrepancy between true incidence and suspected incidence. Although there are some good merits for the creation of a sepsis process measure (SEP1), it is clear that many of the precipitating factors for the perceived sepsis “epidemic” were misunderstood.
However, the larger problem with sepsis incidence discrepancies stems from evolving definitions. These discrepancies are likely to persist, given the controversies surrounding the new Sepsis and Septic Shock Definitions, published by the Sepsis-3 Task Force in February of 2016 (2,3). While Sepsis-3 may help eliminate inappropriate diagnoses of sepsis in patients with Systemic Inflammatory Response Syndrome (SIRS) who do not have infection, early diagnosis of sepsis may be missed, as the definitions no longer include sepsis without organ dysfunction. Nonetheless, for these patients, even though their condition may be less severe, a risk of morbidity and mortality persists, despite being excluded from the latest sepsis definition. Sterling et al. (3) reported that 57% of patients meeting old definition for septic shock (Sepsis-1) did not meet Sepsis-3 criteria. Although Sepsis-3 criteria identified a group of patients with increased organ failure and higher mortality, those patients who met the old criteria and not Sepsis-3 criteria still demonstrated significant organ failure and 14% mortality rate.
Using mortality as an outcome, it is now clear that SIRS is less predictive than SOFA. This was noted by a recent Chinese study that found that the AUROC of systemic inflammatory response syndrome (SIRS) is significantly smaller than that of Sequential Organ Failure Assessment (SOFA) (0.55 [95% confidence interval, 0.46–0.64] vs. 0.69 (95% confidence interval, 0.61–0.77], P = 0.008) to predict 28-day mortality rates of infected patients. However, the same study found that 5.9% of infected patients were diagnosed as sepsis according to sepsis-1 but not to sepsis-3 (4) which is not a negligible rate. In addition to avoiding death, clinicians who see sepsis patients are concerned about identifying septic patients and treating them appropriately to avoid adverse morbidity outcomes. Besides prolonged stays in the ICU, morbidity outcomes were not used to derive with the Sepsis-3 definitions. In order to improve the accuracy of any future sepsis definition, incorporating morbidity outcomes will be important.
Rhee’s article (1) provides glimpses that such a methodology for defining an outcome could potentially be accomplished. Calculating such a morbidity score or index could be done using the same analytical methods leveraging Electronic Medical Record (EMR) data. How about the future of diagnosing sepsis? The advent of more sophisticated machine learning techniques, coupled with more access to biomarker data may pave the way for even better definitions. Taneja et al. (5) used machine learning techniques to assess the predictive power of combining multiple biomarker measurements from a single blood sample with EMR data for the identification of patients in the early to peak phase of sepsis in a large community hospital setting. Combining biomarkers and EMR data achieved an area under the receiver operating characteristic (ROC) curve (AUC) of 0.81, while EMR data alone achieved an AUC of 0.75. Furthermore, a single measurement of six biomarkers (IL-6, nCD64, IL-1ra, PCT, MCP1, and G-CSF) yielded the same predictive power as collecting an additional 16 hours of EMR data (AUC of 0.80), suggesting that the biomarkers may be useful for identifying these patients earlier. Ultimately, advances in biomarker science and machine learning will offer better recognition and definition tools. In the meantime, it will be important to acknowledge the limitations of definitions, including Sepsis 3.
Sam Antonios MD, FACP, SFHM, CPE, CCDS
Chief Medical Officer - Via Christi Hospitals
1. Rhee C, Dantes R, Epstein L, Murphy DJ, Seymour CW, Iwashyna TJ, Kadri SS, Angus DC, Danner RL, Fiore AE, Jernigan JA, Martin GS, Septimus E, Warren DK, Karcz A, Chan C, Menchaca JT, Wang R, Gruber S, Klompas M, for the CDC Prevention Epicenter Program. Incidence and Trends of Sepsis in US Hospitals Using Clinical vs Claims Data, 2009-2014. JAMA. 2017;318(13):1241–1249. doi:10.1001/jama.2017.13836
2. Rhee C, Klompas M. New Sepsis and Septic Shock Definitions: Clinical Implications and Controversies. Infect Dis Clin North Am. 2017 Sep;31(3):397-413. doi: 10.1016/j.idc.2017.05.001. Epub 2017 Jul 5.
3. Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, Bellomo R, Bernard GR, Chiche J, Coopersmith CM, Hotchkiss RS, Levy MM, Marshall JC, Martin GS, Opal SM, Rubenfeld GD, van der Poll T, Vincent J, Angus DC. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA. 2016;315(8):801–810. doi:10.1001/jama.2016.0287
4. Sterling, Sarah A. MD1; Puskarich, Michael A. MD1; Glass, Andrew F. MD1; Guirgis, Faheem MD2; Jones, Alan E. MD1. The Impact of the Sepsis-3 Septic Shock Definition on Previously Defined Septic Shock Patients. Crit Care Med. 2017 Sep;45(9):1436-1442. doi: 10.1097/CCM.0000000000002512.
5. Cheng B, Li Z, Wang J, et al. Comparison of the Performance Between Sepsis-1 and Sepsis-3 in ICUs in China: A Retrospective Multicenter Study. Shock (Augusta, Ga). 2017;48(3):301-306. doi:10.1097/SHK.0000000000000868.
6. Taneja I, Reddy B, Damhorst G, Dave Zhao S, Hassan U, Price Z, Jensen T, Ghonge T, Patel M, Wachspress S, Winter J, Rappleye M, Smith G, Healey R, Ajmal M, Khan M, Patel J, Rawal H, Sarwar R, Soni S, Anwaruddin S, Davis B, Kumar J, White K, Bashir R, Zhu R. Combining Biomarkers with EMR Data to Identify Patients in Different Phases of Sepsis. Sci Rep. 2017 Sep 7;7(1):10800. doi: 10.1038/s41598-017-09766-1.