Severe sepsis caused by overwhelming infection is responsible for significant morbidity and mortality among hospitalized patients.1 Early identification and aggressive treatment, particularly in regard to fluid resuscitation and antibiotics, has been shown to improve survival.2
The Surviving Sepsis Campaign has created guidelines for best practices, and federal reimbursement for sepsis care has been tied to resuscitation “bundles” encouraging such guidelines are followed.
Clinical identification of sepsis commonly includes two or more of the systemic inflammatory response syndrome (SIRS) criteria in the presence of a suspected infection.1 A hallmark of severe sepsis is hypoperfusion leading to end-organ damage and cardiovascular collapse (i.e., septic shock).
Objective measures for hypoperfusion allow for risk stratification along the continuum of the disease process. Elevated serum lactate levels (> 4 mmol/L) have traditionally been used as a sign of tissue hypoperfusion and an outcome predictor for septic patients.1,2
This has led to the development of screening tools and alert systems that may improve adherence to treatment guidelines.
An analysis of sepsis incidence and outcomes in prehospital emergency care demonstrated that septic patients are commonly transported by EMS and have a high mortality rate.3
Several studies have shown that a large percentage of septic patients arrive at the ED by ambulance, including those that require critical care.3,4 Characteristics of septic patients transported by EMS included elevated heart and respiratory rates.4
A study has shown that the time to antibiotics was decreased when sepsis was identified by prehospital providers directly in the narrative of their patient care report.5
Another study showed that prehospital placement of an IV catheter and administration of IV fluids decreased mortality in patients with severe sepsis.6
Severe sepsis has a high mortality rate that may be lowered by early identification and specific treatment. This suggests that prehospital processes including pre-arrival notification and immediate therapy similar to those used for ST elevation myocardial infarction and stroke may influence outcomes.
Prehospital screening tools for severe sepsis have been developed to improve awareness.7,8
Most of these screening tools have relied on the suspicion of infection, ≥ 2 SIRS criteria and point-of-care serum lactate levels.
Although one study has found that activation of a sepsis alert protocol utilizing point-of-care lactate levels may decrease mortality,8 it’s now also possible to screen for severe sepsis without costly lactate monitoring devices and potentially time-consuming blood draws.
Capnography, the waveform measurement of exhaled end-tidal carbon dioxide (EtCO2), is easily and frequently used in prehospital care.
The detection equipment is standard for many EMS units. EtCO2 can be measured immediately and noninvasively, making it a simple, clinically useful tool for prehospital providers.
EtCO2 is a continuous variable that’s determined by basal metabolic rate, cardiac output and ventilation. Thus, abnormal levels may reflect derangement in perfusion, metabolism or gas exchange.
EtCO2 also has several prehospital applications including confirmation and monitoring of proper endotracheal tube placement,9 and evaluating return of spontaneous circulation during cardiopulmonary arrest.10
Multiple studies have shown a relationship between EtCO2 and disease severity or mortality in patients with shock,11 diabetic ketoacidosis,12 and metabolic disturbances.13 Additionally, low EtCO2 levels are associated with elevated lactate levels, odds of operative intervention, and mortality in trauma patients.14,15 In these studies, decreased EtCO2 levels reflected acidotic, low-perfusion states.
Severe sepsis is characterized by poor perfusion, leading to a buildup of serum lactate and resulting metabolic acidosis. Both poor perfusion and metabolic acidosis result in lower EtCO2 levels, because patients increase their minute ventilation as a compensatory mechanism in the setting of metabolic acidosis. This increased respiratory rate “blows off ” CO2 and lowers EtCO2.
Poor tissue perfusion also decreases the amount of blood flow to the alveoli of the lungs, reducing the amount of CO2 that can be exhaled. This is demonstrated most dramatically during extremely low-flow states and cardiac arrest.
In 2013, we reported that EtCO2 is inversely proportional to serum lactate levels in ED patients with suspected sepsis, severe sepsis, and septic shock—as lactate levels increased, EtCO2 levels dropped.Furthermore, EtCO2 performed as well as serum lactate as a predictor of mortality in these patients.16
More recently, we described that abnormal prehospital EtCO2 levels were associated with mortality and metabolic acidosis across a wide cohort of patients.17
In this study, all patients had EtCO2 recorded by EMS along with the initial vital signs, and hospital outcomes such as their primary diagnoses were then linked.
Subsequent data analysis of this group showed that low prehospital EtCO2 predicted sepsis in patients who had been diagnosed with an infection of any kind, and that low prehospital EtCO2 predicted mortality among those patients who had ≥ 2 SIRS criteria.
Information from these studies suggested that an EtCO2 measurement of ≤ 25 mmHg is strongly correlated with lactate levels > 4 mmol/L in the setting of suspected sepsis, and therefore may be useful as an objective measure for tissue hypoperfusion.
The Orange County Study
After a successful pilot study performed in the ED, the Orange County, Fla., EMS system developed a sepsis alert protocol utilizing low EtCO2 as an equivalent to elevated serum lactate for identification of hypoperfusion. (See Figure 1, above.)
This has allowed our EMS agencies discretion to utilize their existing equipment to perform fast, noninvasive screening for severe sepsis rather than purchase approved lactate monitors.
Prior to this protocol release, we had provided minimal mandatory education on the identification or treatment of sepsis, and had no distinct mechanism in place to alert the receiving hospitals of a potentially septic patient.
The current protocol and its subsequent training materials provide pre-arrival notification and the initiation of fluid resuscitation.
In order to successfully deploy the severe sepsis screen and sepsis alert protocol, we educated our providers about infection, the SIRS criteria, sepsis and severe sepsis. This required a distinctly different approach from “symptoms-based” treatment to “disease-based” treatment.
The algorithm focused on suspecting infection—especially in the case of altered mental status, immunocompromised, and elderly patients—then looking for SIRS criteria and evidence of hypoperfusion. (See Figure 2, below.)
One of our biggest challenges was to educate providers to maintain a high index of suspicion even in non-febrile patients. Another was to ensure understanding that evidence for hypoperfusion was the true discriminating factor between a patient who might be ill and one who was seriously ill.
We frequently used cases of patients with nonthreatening viral illnesses who had ≥ 2 SIRS criteria to show the value of this discrimination. For example, a patient with an upper respiratory illness, a fever and tachycardia, but who had an EtCO2 of 35 mmHg, does not meet sepsis alert criteria, and may just have a cold or flu. However, that same patient with an EtCO2 of 21 mmHg meets sepsis alert criteria (the ≤ 25 mmHg threshold in our protocol), and may have severe pneumonia or sepsis.
Further information was supplied to our personnel on the high morbidity and mortality of sepsis and the important role EMS may play in detection. Training was accomplished through multiple mediums including a brief, mandated online course, case reviews with individual crews, lectures at local conferences, and continuous quality review of sepsis alert activations.
We also worked in tandem with our receiving hospitals to ensure that a sepsis alert notification would initiate a tangible response at the ED.
The Orange County sepsis protocol development has been a team approach including both prehospital and hospital partners. Prior to protocol implementation, representatives from area hospitals and local emergency physicians were consulted.
The Orange County, Fla., EMS system developed a sepsis alert protocol utilizing low EtCO2 as an equivalent to elevated serum lactate levels.
Pilot data was shared, and the sepsis alert protocol was disseminated so the receiving facilities and emergency physicians would understand what was meant by a sepsis alert notification.
Simultaneously, specific teams within the hospitals developed their own internal response systems. These teams typically included administrative, ED and critical care representatives
These responses vary, but generally provide for immediate offload and patient evaluation in the ED to expedite resuscitation, drawing of blood cultures and the administration of antibiotics.
After protocol rollout, data on the sepsis alert activations was shared with the hospitals at our monthly quality council meetings.
Screening Tool Evaluation
Prospective evaluation of the screening tool took place for the first year after deployment to determine whether we were identifying the appropriate patients. This was conducted as a prospective cohort study among the patients transported by agencies in the Orange County EMS System to our regional hospitals.18
Inclusion criteria consisted of any case where a sepsis alert was activated by prehospital personnel. Despite the training provided, variation in protocol compliance existed and was the basis for our comparison groups.
Our “protocol compliant” group was formed by sepsis alerts activated after EMS personnel had followed all diagnostic criteria of the protocol. In 100% of these patients, EMS personnel suspected infection, patients had ≥ 2 SIRS criteria, and initial EtCO2 values were ≤ 25 mmHg. This was strictly a protocol-driven process—EMS personnel activated sepsis alert notifications based on their determination that patients met all criteria.
Our “protocol noncompliant” group was made of patients whereby EMS personnel activated a sepsis alert but did not follow all diagnostic criteria of the protocol.
In the protocol noncompliant group, EMS personnel suspected infection in all patients, but they either didn’t have ≥ 2 SIRS criteria (5% of the cases), or EtCO2 values were > 25 mmHg (95% of the cases). This was a paramedic discretionary process, where sepsis alert notifications were activated despite the fact that not all protocol criteria were met.
The primary outcome was diagnosis of sepsis and severe sepsis upon hospital admission. We also measured mortality, patient disposition described as discharge, hospital admission or ICU admission, ICD-9 codes, and the relationship between EtCO2 and bicarbonate (HCO3), anion gap and lactate levels.
Overall, there were 330 prehospital sepsis alerts activated over a one-year period. Of the 330 alerts, 183 (55%) were protocol compliant and 147 (45%) were protocol noncompliant.18 The mean patient age was 70 years, 169 (51%) were male, 286 (97%) were admitted to the hospital, 100 (34%) were admitted to the ICU, and 25 (8%) died during hospital admission.18
Among the patients with complete hospital records, 187 (63%) were diagnosed with sepsis and 87 (29%) were diagnosed with severe sepsis.18
The protocol compliant sepsis alert patients were significantly older (72 vs. 67 years old, p = 0.014) and more likely to be admitted to the hospital (99% vs. 93%, p = 0.014) and to the ICU (41% vs. 25%, p = 0.006).18
A significantly higher percentage of the protocol compliant sepsis alerts were diagnosed with sepsis (78% vs. 43%, p < 0.001) and severe sepsis (47% vs. 7%, p < 0.001), and these patients also had a higher mortality rate (11% vs. 5%, p = 0.036).18
To better establish the effectiveness of the sepsis alert protocol, comparisons were performed between the protocol compliant and protocol noncompliant groups. For those sepsis alerts that followed the protocol, the sensitivity for predicting sepsis was 69% (95% CI 62–75%), the specificity was 67% (95% CI 57–75%), the positive predictive value was 78% (95% CI 70–84%) and the negative predictive value was 99% (95% CI 92–100%).18
The sensitivity for predicting severe sepsis was 90% (95% CI 81–95%), the specificity was 58% (95% CI 52–65%), the positive predictive value was 47% (95% CI 39–97%) and the negative predictive value was 93% (95% CI 87–97%).18
The sensitivity for predicting mortality was 76% (95% CI 54–90%), the specificity was 46% (95% CI 40–52%), the positive predictive value was 11% (95% CI 7–18%) and the negative predictive value was 95% (95% CI 90–98%).18
Protocol & Process Validation
This study demonstrates that it’s feasible for prehospital providers to utilize a sepsis screening tool that incorporates EtCO2 as an objective measure for hypoperfusion, and that it’s highly sensitive for predicting severe sepsis.
Appropriate use of the protocol identified patients who were more likely to be diagnosed with sepsis or severe sepsis, required ICU admission, and had a higher mortality rate. Additionally, decreased prehospital EtCO2 levels correlated with metabolic acidosis and increased in-hospital lactate levels across all patients.
The most common cause for noncompliance with the protocol was the criteria for EtCO2 (139, or 95%, of the group met all criteria except for EtCO2 ≤ 25 mmHg). In fact, the average EtCO2 was significantly higher in the protocol noncompliant group.
Interestingly, low EtCO2 correlated with elevated lactate levels and predicted sepsis, severe sepsis, and mortality across all patients, suggesting that it serves as an objective measure for hypoperfusion.
This may explain why sepsis alerts that followed the protocol were significantly more likely to be diagnosed with severe sepsis than those in the protocol noncompliant group (47% vs. 7%)—objective evidence for hypoperfusion discriminated between them.
It was important for us to develop a sensitive screening tool that could also avoid so-called “alert fatigue” in the EDs. The discrimination between potentially septic patients and those with severe sepsis is one way to achieve this.
There are several limitations to this study. The comparison groups were due to poor adherence to the initial criteria for sepsis alerts. However, it’s not atypical for large EMS systems to have lag time with new protocol compliance, and performance improved over the course of the study.
Another limitation to this study is that patients enrolled were all sepsis alert activations, and we’re unaware how many potentially septic patients weren’t recognized by providers.
Ideally, all patients diagnosed with severe sepsis in the hospital would be linked back to EMS transport to determine the actual incidence of appropriate protocol usage.
The overall admission rate of 97% and ICU admission rate of 34% is evidence that regardless of protocol compliance, the sepsis alert activations represented seriously ill patients.
That nearly half of the protocol noncompliant group was still diagnosed with sepsis is also telling about the discretion of our providers. However, the screening process of the protocol was meant to narrow this group down to those with severe sepsis who require immediate action on behalf of the receiving hospitals. Utilizing EtCO2 levels of ≤ 25 mmHg as an objective measure of hypoperfusion provided that discrimination.
This study adds to the growing evidence that EtCO2 can be used as a surrogate marker of hypoperfusion and metabolic disturbances in the prehospital environment. Of note, the screening portion of this sepsis alert protocol is entirely noninvasive, suggesting that with proper training it may also be applicable for non-ALS providers, possibly as part the decision-making process for requesting ALS transport.
Further studies are necessary to evaluate if this protocol reduces hospital mortality, and whether any improvement in care is due to pre-arrival notification, specific interventions, or both. Also of interest is the trending of EtCO2 during resuscitation, and how it may relate to lactate clearance as a prognostic indicator.
1. Dellinger RP, Cartlet JM, Masur H, et al. Surviving sepsis campaign for management of severe sepsis and septic shock. Crit Care Med.2004;32:858–873.
2. Yealy DM, Kellum JA, Huang DT, et al. A randomized trial of protocol-based care for early septic shock. N Engl J Med. 2014;370(18):1683–1693.
3. Seymour CW, Band RA, Cooke CR, et al. Out-of-hospital characteristics and care of patients with severe sepsis: A cohort study. J of Crit Care.2010;25(4):553–562.
4. Seymour CW, Rea TD, Khan JM, et al. Severe sepsis in pre-hospital emergency care: Analysis of incidence, care, and outcome. Am J Respir Crit Care Med. 2012;186(12):1264–1271.
5. Studnek JR, Artho MR, Garner CL, et al. The impact of emergency medical services on the ED care of severe sepsis. American J of Emerg Med.2012;30(1):51–56.
6. Seymour CW, Cooke CR, Mikkelsen ME, et al. Out-of-hospital fluid in severe sepsis: Effect on early resuscitation in the emergency department.Prehosp Emerg Care. 2010;14(2):145–152.
7. Jansen TC, Bommel JV, Mulder PG, et al. The prognostic value of blood lactate levels relative to that of vital signs in the prehospital setting: A pilot study. Crit Care. 2008;12(6):R160.
8. Guerra WF, Mayfield TR, Meyers MS, et al. Early detection and treatment of patients with severe sepsis by prehospital personnel. J Emerg Med.2013;44(6):1116–1125.
9. Silvestri S, Ralls GA, Krauss B, et al. The effectiveness of out-of-hospital use of continuous end-tidal carbon dioxide monitoring on the rate of unrecognized misplaced intubation within a regional emergency medical services system. Ann Emerg Med. 2005;45(5):497–503.
10. Garnett AR, Ornato JP, Gonzalez ER, et al. End-tidal carbon dioxide monitoring during cardiopulmonary resuscitation. JAMA. 1987;257(4):512–515.
11. Kheng CP, Rahman NH. The use of end-tidal carbon dioxide monitoring in patients with hypotension in the emergency department. Int J Emerg Med. 2012;5(1):31.
12. Fearon DM, Steele DW. End-tidal carbon dioxide predicts the presence and severity of acidosis in children with diabetes. Acad Emerg Med.2002;9(12): 1373–1378.
13. Kartal M, Eray O, Rinnert S, et al. EtCO2: A predictive tool for excluding metabolic disturbances in nonintubated patients. Am J Emerg Med.2011;29(1):65–69.
14. Caputo ND, Fraser RM, Paliga A, et al. Nasal cannula end-tidal CO2 correlates with serum lactate levels and odds of operative intervention in penetrating trauma patients: A prospective cohort study. J Trauma Acute Care Surg. 2012;73(5):1202–1207.
15. Deakin CD, Sado DM, Coats TJ, et al. Prehospital end-tidal carbon dioxide concentration and outcome in major trauma. J Trauma. 2004;57(1):65–68.
16. Hunter CL, Silvestri S, Dean M, et al. End-tidal carbon dioxide is associated with lactate levels and mortality in emergency department patients with suspected sepsis. Am J Emerg Med. 2013;31(1):64–71.
17. Hunter CL, Silvestri S, Ralls G, et al. The sixth vital sign: Prehospital end-tidal carbon dioxide predicts in-hospital mortality and metabolic disturbances. Am J Emerg Med. 2014;32(2):160–165.
18. Hunter CL, Silvestri S, Ralls G, et al. A prehospital screening tool utilizing end-tidal carbon dioxide predicts sepsis and severe sepsis. Am J Emerg Med. 2016;34(5):813–819.