0
Editorial |

Septic Shock Surveillance: Critically Important but Not Straightforward FREE TO VIEW

Michael W. Sjoding, MD; Robert C. Hyzy, MD, FCCP
Author and Funding Information

FUNDING/SUPPORT: This study was supported by grants to M. W. S. from the NIH [Grants T32HL007749].

FINANCIAL/NONFINANCIAL DISCLOSURES: None declared.

aDivision of Pulmonary and Critical Care, Department of Internal Medicine, University of Michigan, Ann Arbor, MI

bCenter for Healthcare Outcomes & Policy, Institute of Healthcare Policy and Innovation, University of Michigan, Ann Arbor, MI

CORRESPONDENCE TO: Robert C. Hyzy, MD, 3916 Taubman Center, Ann Arbor, MI 48109


Copyright 2016, American College of Chest Physicians. All Rights Reserved.


Chest. 2017;151(2):247-248. doi:10.1016/j.chest.2016.08.1433
Text Size: A A A
Published online

Inpatient care for patients with sepsis has placed a substantial and growing burden on health-care systems. Between 1980 and 2000, there was a threefold increase in sepsis incidence. Sepsis is now the most costly hospital condition and the most common diagnosis among patients receiving intensive care. Sepsis also causes significant morbidity and mortality and contributes to one in every two to three hospital deaths. Because of its health-care burden, high mortality, and evidence showing that an early sepsis treatment bundle significantly improves patient outcomes, the Centers for Medicare and Medicaid Services have adopted sepsis care as part of its core inpatient performance measurement program.

FOR RELATED ARTICLE SEE PAGE 278

However, many large sepsis epidemiologic studies, and the sepsis performance measure itself, are based entirely on administrative billing data to identify sepsis cases. Concern has been raised that shifts in how inpatient admissions have been billed over time may have distorted the findings of multiple studies showing that sepsis incidence has increased whereas case fatality rates have declined. Nationally, if financial pressure to improve reimbursement leads to sepsis upcoding among patients who do not meet criteria, sepsis incidence would falsely appear to increase and mortality to decline, even if true sepsis incidence and outcomes were unchanged.

To further explore these issues, in this issue of CHEST, Kadri and colleagues analyzed 6.5 million adult hospitalizations from 2005 to 2014 at 27 academic hospitals to calculated trends in septic shock incidence and mortality over time. Because their data set contained detailed records of each hospitalization, including treatments administered during the hospital stay, the authors developed a non-ICD-9-CM (International Classification of Diseases, Ninth Revision, Clinical Modification)-code-based clinical algorithm for identifying septic shock cases. In addition to a blood culture order and antibiotic administration, their clinical surveillance definition required that patients receive vasopressor support for two or more consecutive days. The authors then compared septic shock incidence and mortality derived from this clinical definition with those derived from the septic shock ICD-9-CM billing code. Finally, they validated each definition by performing chart reviews on a subset of 1,000 patients to calculate sensitivity and specificity of each.

Using their clinical definition, the authors found the incidence of septic shock increased over the study period; however, the rise was much lower than estimates based on the ICD-9-CM code (4.9% vs 19.8% increase per year, respectively). They also found that septic shock mortality declined, but the decline was also much more modest than ICD-9-CM code estimates (0.6% vs 1.2% per year, respectively). Validation of each definition showed the clinical definition had substantially higher sensitivity than the ICD-9-CM code definition (74.8% vs 48.3%, respectively), whereas the ICD-9-CM code definition had slightly higher specificity (97.2% and 98.9%, respectively). Each had comparable positive predictive value (83% and 89%, respectively).

Importantly, the very high specificity of the ICD-9-CM codes suggests the rise in septic shock incidence was not caused by an increase in coding of patients who had not met criteria for septic shock. The results suggested the sensitivity of septic shock coding had actually improved over the study period. In particular, the proportion of patients identified by the clinical definition who also had a septic shock ICD-9-CM code increased from 30% to 56%. This suggests there was improved recognition, documentation, and subsequent coding of septic shock over the study period.

The modest decline in septic shock mortality in this study does not necessarily contradict other recent non-claims-based analyses describing more dramatic declines in overall sepsis mortality during the same time period., The authors’ strict septic shock definition required that patients receive vasopressor agents for at least two consecutive days, which enriched the population with severely ill patients. Baseline mortality among these patients was > 50%, well above rates typically reported in epidemiologic studies and clinical trials. However, the minimal decline in mortality in this population suggests that much work still needs to be done to improve outcomes in this subgroup of patients with sepsis.

The most disconcerting result of the study was the presence of only moderate overlap in patients between definitions. Among the 6.5 million hospitalizations, 44,651 patients with septic shock were identified by both definitions, whereas 54,661 patients were only identified by the clinical definition and 37,669 were only identified by the ICD-9-CM code. For health-care systems interested in tracking hospital performance and outcomes in patients with septic shock, this is troubling news. Patients identified by either definition are likely to have had septic shock, based on their equivalently high positive predictive values > 80%. This means that both definitions missed a number of true positive cases. Acknowledging this concern, the authors combined definitions and re-estimated septic shock incidence to be rising by 8.1% per year and mortality to be declining by 1.2% per year.

Since October 1, 2015, the Centers for Medicare and Medicaid Services have required hospitals to collect data on their adherence to an early management bundle for sepsis and septic shock. The bundle’s primary requirement is for patients to have blood cultures performed, lactate measured, and broad-spectrum antibiotics administered within 3 hours of presentation. If a patient has septic shock, an appropriate fluid bolus should also be administered (30 mL/kg). At present, only patients with an International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) diagnosis code for sepsis or septic shock are evaluated. Unfortunately, this study shows how a significant number of patients with septic shock may be misclassified by septic shock administrative billing codes as not having septic shock and true 3-hour bundle adherence may not be accurately reflected in these data.

Ultimately, inpatient hospital performance measures are likely to use electronic health record data to evaluated measure performance, instead of billing data and manual chart review, as these systems become more advanced and interoperable. This study illustrates how even well-intentioned electronic health record-based definitions, while a substantial step forward, can still misclassify significant numbers of patients when the prevalence of the condition is low. Given its high mortality and costs, efforts to develop highly accurate, flexible methods for identifying patients with septic shock are still warranted to ensure these patients are appropriately treated and hospitals are reliably assessed for treating these patients.

References

Martin G.S. .Mannino D.M. .Eaton S. .Moss M. . The epidemiology of sepsis in the United States from 1979 through 2000. New Engl J Med. 2003;348:1546-1554 [PubMed]journal. [CrossRef] [PubMed]
 
Torio C, Andrews R. National inpatient hospital costs: the most expensive conditions by payer, 2011.http://www.hcup-us.ahrq.gov/reports/statbriefs/sb204-Most-Expensive-Hospital-Conditions.pdf. Accessed August 2, 2016.
 
Sjoding M.W. .Prescott H.C. .Wunsch H. .Iwashyna T.J. .Cooke C.R. . Longitudinal changes in ICU admissions among elderly patients in the United States. Crit Care Med. 2016;44:1353-1360 [PubMed]journal. [CrossRef] [PubMed]
 
Liu V. .Escobar G.J. .Greene J.D. .et al Hospital deaths in patients with sepsis from 2 independent cohorts. JAMA. 2014;312:90-92 [PubMed]journal. [CrossRef] [PubMed]
 
Levy M.M. .Rhodes A. .Phillips G.S. .et al Surviving sepsis campaign: association between performance metrics and outcomes in a 7.5-year study. Crit Care Med. 2015;43:3-12 [PubMed]journal. [PubMed]
 
Cooke C.R. .Iwashyna T.J. . Sepsis mandates: improving inpatient care while advancing quality improvement. JAMA. 2014;312:1397-1398 [PubMed]journal. [CrossRef] [PubMed]
 
Rhee C. .Gohil S. .Klompas M. . Regulatory mandates for sepsis care–reasons for caution. N Engl J Med. 2014;370:1673-1676 [PubMed]journal. [CrossRef] [PubMed]
 
Kadri S.S. .Rhee C. .Strich J.R. .et al Estimating ten-year trends in septic shock incidence and mortality in United States academic medical centers using clinical data. Chest. 2017;151:278-285 [PubMed]journal
 
Kaukonen K.M. .Bailey M. .Suzuki S. .Pilcher D. .Bellomo R. . Mortality related to severe sepsis and septic shock among critically ill patients in Australia and New Zealand, 2000-2012. JAMA. 2014;311:1308-1316 [PubMed]journal. [CrossRef] [PubMed]
 
Stevenson E.K. .Rubenstein A.R. .Radin G.T. .Wiener R.S. .Walkey A.J. . Two decades of mortality trends among patients with severe sepsis: a comparative meta-analysis*. Crit Care Med. 2014;42:625-631 [PubMed]journal. [CrossRef] [PubMed]
 
The Joint Commission. Specifications manual for national hospital inpatient quality measures.https://www.jointcommission.org/specifications_manual_for_national_hospital_inpatient_quality_measures.aspx. Accessed August 2, 2016.
 

Figures

Tables

References

Martin G.S. .Mannino D.M. .Eaton S. .Moss M. . The epidemiology of sepsis in the United States from 1979 through 2000. New Engl J Med. 2003;348:1546-1554 [PubMed]journal. [CrossRef] [PubMed]
 
Torio C, Andrews R. National inpatient hospital costs: the most expensive conditions by payer, 2011.http://www.hcup-us.ahrq.gov/reports/statbriefs/sb204-Most-Expensive-Hospital-Conditions.pdf. Accessed August 2, 2016.
 
Sjoding M.W. .Prescott H.C. .Wunsch H. .Iwashyna T.J. .Cooke C.R. . Longitudinal changes in ICU admissions among elderly patients in the United States. Crit Care Med. 2016;44:1353-1360 [PubMed]journal. [CrossRef] [PubMed]
 
Liu V. .Escobar G.J. .Greene J.D. .et al Hospital deaths in patients with sepsis from 2 independent cohorts. JAMA. 2014;312:90-92 [PubMed]journal. [CrossRef] [PubMed]
 
Levy M.M. .Rhodes A. .Phillips G.S. .et al Surviving sepsis campaign: association between performance metrics and outcomes in a 7.5-year study. Crit Care Med. 2015;43:3-12 [PubMed]journal. [PubMed]
 
Cooke C.R. .Iwashyna T.J. . Sepsis mandates: improving inpatient care while advancing quality improvement. JAMA. 2014;312:1397-1398 [PubMed]journal. [CrossRef] [PubMed]
 
Rhee C. .Gohil S. .Klompas M. . Regulatory mandates for sepsis care–reasons for caution. N Engl J Med. 2014;370:1673-1676 [PubMed]journal. [CrossRef] [PubMed]
 
Kadri S.S. .Rhee C. .Strich J.R. .et al Estimating ten-year trends in septic shock incidence and mortality in United States academic medical centers using clinical data. Chest. 2017;151:278-285 [PubMed]journal
 
Kaukonen K.M. .Bailey M. .Suzuki S. .Pilcher D. .Bellomo R. . Mortality related to severe sepsis and septic shock among critically ill patients in Australia and New Zealand, 2000-2012. JAMA. 2014;311:1308-1316 [PubMed]journal. [CrossRef] [PubMed]
 
Stevenson E.K. .Rubenstein A.R. .Radin G.T. .Wiener R.S. .Walkey A.J. . Two decades of mortality trends among patients with severe sepsis: a comparative meta-analysis*. Crit Care Med. 2014;42:625-631 [PubMed]journal. [CrossRef] [PubMed]
 
The Joint Commission. Specifications manual for national hospital inpatient quality measures.https://www.jointcommission.org/specifications_manual_for_national_hospital_inpatient_quality_measures.aspx. Accessed August 2, 2016.
 
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Find Similar Articles
CHEST Journal Articles
PubMed Articles
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543