sepsis mortality at Kaiser Pe”>Correspondence

Contents lists available at ScienceDirect

American Journal of Emergency Medicine

journal homepage: locate/ ajem

Decreasing sepsis mortality at Kaiser Permanente Northern California

To the Editor,

In his correspondence, “Does adoption of a regional sepsis protocol reduce mortality?,” Durston [1] asserts that a systematic approach to sepsis identification and management at Kaiser Permanente (KP) did not reduce mortality and may have caused harm. It is useful to have an informed discussion on sepsis performance improvement activities, both for hospitals engaged in them and for researchers studying these efforts. Less helpful, however, is the use of limited data and flawed assumptions in an attempt to discredit these activities.

Sepsis is a major Public health concern [2]. It affects millions of patients each year [3]. It contributes to as many as 1 in 2 hospital deaths in the United States [4]. And it is the single most expensive cause of hospitalization nationally [3,5]. Despite this, ample evidence suggests that sepsis is still underrecognized [5]. As a result, one of the primary objectives of sepsis performance improvement programs has been to maximize the rapid identification of sepsis patients to improve their care [3]. Success in recognizing sepsis at earlier and, ideally, more treatable stages has been associated with improved mortality [6,7]. However, this success is accompanied by a new challenge-to determine how much of the observed mortality reduction is due to Improved care and how much is due to the identification of a sepsis cohort with lower illness severity.

Central to his correspondence, Durston [1] introduces the

misleading term sepsis admission volume-independent mortality (SAVi) and erroneously concludes that increases in so-called SAVi over time mean that sepsis programs are harming patients. In fact, observant readers will see that “SAVi” is highly dependent on sepsis admission volume because the number of sepsis deaths over any period is directly dependent on the number of sepsis admissions. At KP NC over the past 5 years, there has been a systematic focus on early identification, risk stratification, and resuscitation of patients with sepsis at all levels of risk. As a result, the relative fraction of all inpatient admissions and deaths attributed to sepsis in our system increased steadily. Over the same period, the percentage of sepsis deaths relative to all admissions (so-called SAVi) also rose but much more modestly.

This rise is not surprising. Maintaining an unchanged sepsis mortality count during a period in which sepsis admission volume increased by thousands would require no net increase in mortality resulting from the newly identified population-a population known to have mortality at least 3-fold higher than that of the nonsepsis hospital population [8]. Thus, the author’s conclusion that the observed modest increase in the number of deaths due to sepsis in the setting of a large increase in the number of sepsis diagnoses is evidence that “harm” is erroneous. Instead, answering the author’s question would require a much more sophisticated analysis of data

with significantly more detail than simple admission and mortality counts copied from a hospital newsletter.

In fact, since the introduction of a regional sepsis performance improvement program in 2008, we have witnessed steady declines in risk-adjusted mortality among patients with the most Severe forms of sepsis, including those treated with early goal-directed therapy (Figure). Unfortunately, the prior correspondence failed to include these risk-ad- justed sepsis mortality trends or all-cause hospital mortality trends, all of which consistently demonstrate substantial reductions over time.

Drawing heavily on our investments in an extremely large and data-rich electronic medical record, our partnership with the KP Division of Research, and the dedicated efforts of thousands of clinicians, we have spent much of the last 5 years working to understand the impact of sepsis on our hospital population and improving the detection, risk stratification, and early aggressive treatment of patients across the sepsis spectrum. As we have recently witnessed, the evidence guiding sepsis care will almost certainly continue to evolve [9,10]. At KP, we are fully committed to bringing the benefit of that evidence to every patient in need.

Vincent Liu, MD, MS Kaiser Permanente Division of Research 2000 Broadway, Oakland, CA 94612

E-mail address: [email protected]

Figure. Observed and predicted hospital mortality of sepsis patients directly admitted to the intensive care unit from the emergency department between January 2008 and August 2013 in KP NC. The regional sepsis quality improvement program (including early goal-directed therapy implementation) began in 2008 and continued to 2013. The predicted hospital mortality and 95% confidence limits (bars) are based on an electronic adaptation of the Simplified Acute Physiology Score, version 3 with original North American coefficients and is based on data from multiple health information systems, including the KP HealthConnect (KPHC) electronic medical record. Observed mortality rates are based on patient data from KPHC; the staged roll-out of KPHC began in 2008 and was fully deployed to all 21 hospitals by mid-2010.

0735-6757/(C) 2014

Alan Whippy, MD Medical Director of Quality and Safety The Permanente medical group

1800 Harrison St, Oakland, CA 94612 E-mail address: [email protected]

John W. Morehouse, MD The Permanente Medical Group Kaiser Foundation Hospital East Bay Department of Emergency Medicine

275 West MacArthur, Oakland, CA 94611 E-mail address: [email protected]

http://dx.doi.org/10.1016/j.ajem.2014.07.024

References

1. Durston W. Does adoption of a regional sepsis protocol reduce mortality? Am J Emerg Med 2014;32:280-1.
2. Angus DC. The lingering consequences of sepsis: a hidden public health disaster? JAMA 2010;304:1833-4.
3. Dellinger RP, Levy MM, Rhodes A, Annane D, Gerlach H, Opal SM, et al. Surviving Sepsis Campaign: International Guidelines for Management of Severe Sepsis and Septic Shock: 2012. Crit Care Med 2013;41:580-637.
4. Liu V, Escobar GJ, Soule J, Greene JD, Whippy A, Angus DC, et al. Hospital deaths in patients with sepsis from 2 independent cohorts. JAMA 2014.
5. Elixhauser A, Friedman B, Stranges E. HCUP Statistical Brief #122. October 2011. Rockville, MD: Agency for Healthcare Research and Quality; 2009 [Available at: http://www.hcup- us.ahrq.gov/reports/statbriefs/sb122.pdf. 2011. Accessed March 1, 2014].
6. Miller III RR, Dong L, Nelson NC, Brown SM, Kuttler KG, Probst DR, et al. Multicenter implementation of a severe sepsis and septic shock treatment bundle. Am J Respir Crit Care Med 2013;188:77-82.
7. Levy MM, Dellinger RP, Townsend SR, Linde-Zwirble WT, Marshall JC, Bion J, et al. The Surviving Sepsis Campaign: results of an international guideline-based performance improvement program targeting severe sepsis. Intensive Care Med 2010;36:222-31.
8. Liu V, Morehouse JW, Soule J, Whippy A, Escobar GJ. Fluid volume, lactate values, and

   mortality in sepsis patients with intermediate lactate values. Ann Am Thorac Soc 2013.

   The ProCESS Investigators. A randomized trial of protocol-based care for early septic shock. N Engl J Med 2014.

9. Lilly CM. The ProCESS Trial-a new era of sepsis management. N Engl J Med 2014.

   Author’s reply re: Sepsis mortality at Kaiser Permanente Northern California?,??

   

   To the Editor,

   I appreciate the opportunity to respond to Morehouse et al criticisms of my correspondence [1] in which I presented data showing that the sharp reduction in sepsis mortality claimed by Kaiser Permanente (KP) after adoption of a regional early goal- directed therapy (EGDT) sepsis protocol [2-5], modeled after the protocol published by Rivers et al [6], was more likely to be due to a change in the definition of sepsis than to real improvements in sepsis care. I noted that adoption of the protocol was actually associated with an increase in the number of patients dying of sepsis per 1000 hospital admissions for all causes, suggesting possible harm from some elements of the protocol.

   The “limited data” to which Morehouse et al refer include data from 27424 patients admitted with a sepsis-related diagnosis from January 2006 to February 2010. Kaiser Permanente denied me access to more recent data.

   ? Source(s) of funding: None.

   ?? Name of organization and date of assembly if the article has been presented: Not applicable.

   Morehouse et al claim that the term sepsis admission volume- independent mortality (SAV-I) is misleading. Kaiser Permanente’s claims of reduced sepsis mortality are based entirely upon a decrease in the percentage of patients admitted with a sepsis-related diagnosis who died, a rate which I termed sepsis admission volume-dependent mortality (SAV-D) because it can be factitiously reduced by inflating the denominator, the number of patients admitted with a sepsis- related diagnosis. I coined the term SAV-I to denote the number of patients dying of sepsis per 1000 hospital admissions for all causes. I noted that SAV-I is less dependent than SAV-D on sepsis admission volume because sepsis admissions account for a small fraction of total hospital admissions. To demonstrate that Morehouse et al are mistaken in claiming that SAV-I is “highly dependent on sepsis admission volume,” I have calculated the number of patients dying of sepsis per 1000 hospital admissions for all causes other than sepsis, which I term sepsis admission volume-independent mortality, revised (SAV-I^R), and I have added SAV-I^R results to my original Table. Sepsis admission volume-independent mortality, revised is entirely indepen- dent of sepsis admission volume because sepsis admission volume does not make up any part of the numerator or denominator in the SAV-I^R calculation. As shown in the Table, the differences between SAV-I and SAV-I^R are so small that the terms can essentially be used interchangeably.

   Morehouse et al argue that a rise in SAV-I mortality is still consistent with their claims that sepsis care improved dramatically after adoption of the EGDT protocol. This could only be true if either a sepsis outbreak coincidentally struck the region just at the same time the EGDT protocol was adopted, or KP physicians were failing to correctly attribute inpatient deaths to sepsis in the pre-EGDT epoch. No other Northern California hospitals reported an outbreak of fatal sepsis cases or other spike in mortality from any “newly identified population” during this same period, and Morehouse et al present no evidence that, before adoption of the EGDT protocol, KP physicians were diagnosing patients who died of sepsis as dying of other causes.

   Morehouse et al appear to be using the term newly identified population, to refer to the fact that as part of the rollout of the EGDT protocol, KP physicians were encouraged to use “systemic inflamma- tory response syndrome (SIRS)” criteria, which include relatively minor abnormalities in vital signs and white blood count, to make a diagnosis of sepsis [2]. Previously, most physicians had traditionally reserved the term sepsis, to refer to patients who were critically ill due to life-threatening infections [7]. It is reasonable for Morehouse et al to argue that the use of SIRS criteria may lead to earlier recognition and treatment of patients with sepsis, but they cannot reasonably claim in the same correspondence that earlier identification and treatment explain an increase in SAV-I mortality. Patients admitted with suspected sepsis based on SIRS criteria have in-hospital mortality rates of less than 1% [8], not rates “at least 3-fold higher than that of the nonsepsis hospital population.”

   Morehouse et al submit a graph, which they claim demonstrates that mortality was much lower than predicted in “the most severe forms of sepsis” after adoption of their EGDT protocol. They present no supporting data for the graph, no supporting references, and only a cryptic description of their methods. They reference the Protocolized Care for Early Septic Shock (ProCESS) trial [9] later in their correspon- dence, but they fail to note that the ProCESS trial showed a trend toward increased mortality in patients treated by EGDT protocol as compared with usual care, with the greatest difference being in the sickest patients (38.2% mortality in the EGDT group vs 26.4% in the usual care group in patients with serum lactate N 5.3 mmol/L; P =

   .05). The ProCESS trial results cast further doubt on KP’s claims of marked reductions in sepsis mortality after the adoption of an almost identical EGDT protocol. They are quite consistent, though, with my conclusion that the KP EGDT protocol may have had a net deleterious effect on sepsis mortality.