• 
• 

Three Part Question

In [ED patients presenting with clinical features of sepsis], are [biomarker panels better than physiological markers] at [diagnosing sepsis?]

Clinical Scenario

A 66y female presents to the ED. She has been unwell for 3 days with SOB, cough and lethargy. Her PMH includes hypertension, for which she is on Lisinopril. She looks unwell with a pulse of 112, RR 26/min, BP 106/45 and O2 sats of 92% on air. She has clinical and radiological signs of a left basal pneumonia.

You begin treatment for a community acquired pneumonia, including oxygen and antibiotics. As well as routine haematology and biochemistry, you are aware that other blood tests may assist in prognostication and treatment planning, but are unsure which blood tests are best in this regard.

Search Strategy

The OVID interface was used to search both the Medline (1966 to present day) and Embase databases. The Cochrane Library was then searched.
Medline & Embase:
((Emergency Room) OR (Emergency Department) OR (Accident and Emergency Department) OR (Accident and Emergency) OR (A&E) OR (ED) OR (ER) OR (emergency medical services))
AND
((sepsis) OR (septicaemia) OR (severe sepsis) OR (septic shock))
AND
((biomarker) OR (biological marker) OR (biologic marker) OR (serum marker) OR (biochemical marker) OR (laboratory marker) OR (surrogate marker))
AND
((early warning score) OR (NEWS) OR (new early warning score) OR (clinical signs) OR (physiology))
Limits – last 10 years, human

Cochrane search term = 'sepsis biomarkers'. Filtered to title/abstract/keyword

Search Outcome

Embase
- search results = 69
- relevant after title and abstract review = 23
- relevant after full text review = 8
- reasons excluded after full text review = 3x ICU population, 1x English full text not available, 5x compares biomarkers to ICU scores (eg, APACHE II) not applicable to ED environment, 5x no comparison to physiological markers, 1x literature review

Medline
- search results = 91
- relevant after title and abstract review = 28
- relevant after full text review = 12
- reasons excluded after full text review = 6x ICU population, 11x compares biomarkers to ICU scores (eg, APACHE II) not applicable in ED environment, 4x no comparison to physiological markers, 1x expert consensus

Cochrane
- search results = 3
- relevant after title and abstract review = 2
- relevant after full text review = 0
- reasons excluded after full text review = 2x included studies looking at ICU populations

Total after removing duplications = 12

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Galtung et al 2022 Germany	N=312 Adult patients presenting to ED with clinical suspicion of acute infection, with at least one vital sign change	Prospective, observational cohort study	Predicting in-hospital mortality	IMX-SEV-2 had an area under the receiver operating characteristic (AUROC) of 0.84 in predicting in-hospital mortality. This was higher than lactate, qSOFA and NEWS2 individually	- Weaknesses of observational studies: correlation not causation, confounding factors - Of 312 patients, only 22 died – low statistical power - Patients still hospitalised after 28-day analysis were not included in end-point analysis (reason given to reduce effects of unrelated mortality)
			Predicting 72-hour multiorgan failure	IMX-SEV-2 had an AUROC of 0.76, which was not significantly different to that of the comparators listed above
Tong-Minh et al 2021 The Netherlands	Patients with sepsis. Not defined.	Systematic review of observational studies.	1-month mortality	A combination of the MEDS score combined with several different biomarkers gave an AUC for predicting 1-month mortality of 0.731 to 0.891.	- Reviewed papers investigating combinations of biomarkers with clinical scoring systems, rather than comparing biomarkers to physiological scores - Didn’t include studies with other endpoints (ICU admission, long-term mortality) - Few studies used a pre-defined cut-off value for biomarkers – difficult to translate to clinical practice - included studies had high risk of bias due to small cohorts - heterogeneity - no meta-analysis - included studies that either didn't explicitly detail study population, or didn't exclusively examine and ED population.
Ishikawa et al 2021 Japan	N = 161 Consecutive adults aged 20 or above, predicted to be hospitalised for at least 48 hours by physician in charge. With diagnosis of SIRS according to ACCP/SCCM criteria.	Prospective observational cohort study	Predicting early multiple organ dysfunction (MOD) at day 2	Procalcitonin (PCT) was the best predictor of MOD compared to other biomarkers (IL-6, CRP, WBC, IL-8, IL-10, TNFa). A combination of qSOFA score + PCT had a higher AUC (0.814) in predicting early MOD than PCT or qSOFA alone.	- small sample size - bloods taken 6 hours after admission (may have already had antibiotics and other treatments) - the study also measures biomarkers taken in the days after ED admission. Much of the data analysis focuses on that + combines day 0 and day 1 labs etc which is unhelpful when looking at ED investigations alone. - qSOFA may be unreliable in patients with cognitive dysfunction prior to becoming septic
Hausfater et al 2021 International	N= 1517 Adults whose initial evaluation included a complete blood count with differential (CBC-DIFF). Then categorised according to sepsis-2/sepsis-3 criteria into non-SIRS (case controls), SIRS, infection, sepsis, severe sepsis, and septic shock.	Prospective observational cohort study. Blinded	Sepsis detection (using sepsis-3 definitions)	When abnormal, monocyte distribution width (MDW) increased the odds of sepsis by 7.6 (5.1-11.3, CI 95%)	- overall sepsis prevalence reported was higher than usual prevalence – selection bias towards sepsis patients at inclusion - Conflict of interest - main author received payments from several in vitro diagnostics companies. Three authors are employees of one of these companies, Beckman Coulter Inc, which sponsored the study.
Ruangsomboon et al 2020 Thailand	n=250 Adults greater than 75 years old Presenting with suspicion of sepsis, which was defined as having a blood culture taken and IV antibiotics given Then retrospectively reviewed and classified into either non-sepsis, sepsis, and septic shock, based on Sepsis-3 definitions	Prospective observational cohort study	Diagnosis of sepsis	Prespsin (AUC 0.792), PCT (0.751) and CRP (0.767) had similar diagnostic utility and prognostic accuracy for sepsis. Comparatively, NEWS had an AUC of 0.705, and qSOFA of 0.705. The highest diagnostic accuracy (0.847) was from combining presepsin, PCT, and qSOFA	- one centre - time course of when biomarkers not taken (did not look at serial measures) - observational methodology
			Diagnosis of septic shock	Highest diagnostic accuracy (0.819) was from combination of presepsin, PCT, and qSOFA score
			30-day mortality	Highest prognostic accuracy (0.780) for 30-day mortality was from combination of presepsin, PCT, and qSOFA score
Saeed et al 2019 International	n=1175 derivation, n= 896 validation patients Adults with clinical suspicion of infection (presenting symptoms, vital signs, blood culture request or lab findings during ED assessment)	Observational derivation patient cohort study	28-day mortality	MR-proADM had the strongest association with 28-day mortality in the derivation cohort, with an AUC of 0.88. This was greater than qSOFA (0.75), NEWS (0.720) and other early warning scores, as well as PCT, lactate, and CRP. MR-proADM also had the strongest association (0.89) with 28-day mortality in the validation cohort.	- potentially subjective inclusion criteria - validation cohort and derivation cohort were statistically different (derivation were - older, longer length of hospitalisation, higher prevalence of respiratory infection)
			Hospitalisation	MR-proADM had the strongest association with hospitalisation decisions across both cohorts too, greater than all measured early warning scores and other biomarkers (PCT, lactate, CRP)
Gonzalez del Castillo et al 2019 Spain	N = 684 Adults, clinical suspicion of infection as judged by treating physician based on usual clinical practice (vital signs, symptoms, request for a blood culture, overall lab findings during standard ED assessment)	Prospective observational cohort study	Hospitalisation and need for antibiotics.	MR-proADM had strongest association with hospitalisation (AUC 0.79) and need for antibiotics (0.77), compared to other biomarkers and clinical scores.	- observational methodology - antibiotic administration times may have been influenced by ED waiting times
			ICU admission	mR-proADM had strongest association with ICU admission. Subgroup with high MR-proADM and low NEWS had significantly higher rates of ICU admission
			28-day mortality	mR-proADM had strongest association with 28-day mortality (AUC 0.84)
Zhang Q, Chun-Sheng L 2019 China	N=301 Adults with SIRS or sepsis according to ACCP/SCCM criteria.	Prospective observational cohort study	28-day mortality	When measured on day 1 of admission (presentation to ED), the MEDS score had a larger AUC in predicting 28-day mortality from sepsis (0.809, p<0.001) than any of these biomarkers. The largest AUC was found using vWF/ADAMTS-13 + MEDS together (0.856)	- Excluded psychiatric patients, no reason given. - single centre - small sample size - non-randomised
Shankar-Hari et al 2018 UK	N = 272 Priori sampling method Aged 16 and above SIRS criteria met Clinical suspicion of sepsis (blood cultures taken, antibiotics started) With no clinical suspicion of severe sepsis or septic shock	Prospective observational cohort study	Severe sepsis (SOFA score of 2 or more at 24 hours)	No biomarkers had clinically relevant predictive ability of severe sepsis	- Non-randomised sampling method - small sample size - excluded SIRS negative patients with sepsis who could have progressed to develop sepsis
			Critical care admission	No biomarkers had clinically relevant predictive ability of critical care admission
			Death within 72 hours	No biomarkers had clinically relevant predictive ability of death within 72 hours
Liu et al 2013 China	N=859 Consecutive patients who fulfilled the criteria for sepsis as defined by the American College of Chest Physicians/Society of Critical Care Medicine (ACCP/SCCM)	Prospective observational cohort study	Diagnosing sepsis	AUC of presepsin was significantly higher than that of PCT (0.820 vs 0.724, p <0.01) in diagnosing sepsis	- how were healthy controls enrolled? - single-centre - sepsis was established on clinical features, imaging and blood tests, but not confirmed by blood culture
			Severe sepsis prediction	AUC of presepsin was 0.840 (higher than MEDS score or PCT). Presepsin + MEDS higher than using MEDS alone
			Septic shock prediction	AUC of presepsin alone was not statistically significant. Using presepsin and MEDS together was significant 0.924, p<0.01
			28-day mortality	MEDS better at predicting mortality than presepsin, was even better when using MEDS and presepsin together
Yin et al 2013 China	N = 680 Consecutively enrolled adults with suspected infection (history, examination and lab tests) with 2+ criteria of SIRS	Prospective observational cohort study	Predicting severe sepsis	Plasma sTM levels in severe sepsis were higher than those with sepsis (p<0.001) sTM was an independent predictor of severe sepsis (OR 1.11). AUC of sTM + MEDS was 0.909, higher than sTM or MEDS alone	-Single centre study -sTM naturally elevated in renal dysfunction. Not accounted for in this study. -Sepsis not confirmed by blood culture
			30-day mortality	Non-survivors had higher plasma sTM levels (p<0.001) sTM was an independent predictor of 30-day mortality (OR 1.059). AUC of sTM and MEDS was 0.805, higher than sTM or MEDS alone
Li et al 2018 China	N= 821 Adults with sepsis through diagnosis with sepsis-3	Retrospective observational cohort study	Mortality at 28-days	C-index of TIPS (0.772) was higher than MEDS (0.666) and qSOFA (0.669) in predicting mortality at 28-days.	- excluded pregnant women, cardiac or respiratory arrest or used vasoactive drugs before admission - retrospective so counts on electronic data being correctly recorded - telephone call to identify adverse outcomes from patients or relatives. Risk of recall bias. -‘best’ biomarkers were chosen to calculate TIPS, what does this mean? Were they the blood tests taken at ED admission?
			Mechanical ventilation, consciousness disorder and admission to ICU	Elevated TIPS independent predictor of mechanical ventilation, AICU

Comment(s)

There are a number of small studies providing limited, low-quality evidence about utility of biomarkers for emergency department patients with suspected sepsis. The best diagnostic and predictive accuracy for sepsis was generally found when combining biomarkers with clinical scores.

Clinical Bottom Line

Further prospective studies are required to ascertain if biomarker use, combined with clinical scores or not, can be used diagnostically or prognostically in Emergency Department patients with suspected sepsis.

References

1. Galtung et al Prospective validation of a transcriptomic severity classifier among patients with suspected acute infection and sepsis in the emergency department European Journal of Emergency Medicine April 2022
2. Tong-Minh et al Predicting mortality in adult patients with sepsis in the emergency department by using combinations of biomarkers and clinical scoring systems: a systematic review BMC Emergency Medicine June 2021
3. Ishikawa et al Risk prediction of biomarkers for early multiple organ dysfunction in critically ill patients. BMC Emergency Medicine November 2021
4. Hausfater et al Monocyte distribution width (MDW) performance as an early sepsis indicator in the emergency department: comparison with CRP and procalcitonin in a multicentre international European prospective study. Critical Care June 2021
5. Ruangsomboon et al Diagnostic and prognostic utility of presepsin for sepsis in very elderly patients in the emergency department Clinica Chimica Acta September 2020
6. Saeed et al The early identification of disease progression in patients with suspected infection presenting to the emergency department: A multi-centre derivation and validation study. Critical Care February 2019
7. Gonzalez del Castillo et al Biomarkers and clinical scores to identify patient populations at risk of delayed antibiotic administration or intensive care admission Critical Care October 2019
8. Zhang Q, Chun-Sheng L Risk stratification and prognostic evaluation of endothelial cell specific molecule1, von Willebrand factor, and a disintegrin like and metalloprotease with thrombospondin type 1 motif for sepsis in t Experimental and Therapeutic Medicine April 2015
9. Shankar-Hari et al Early PREdiction of sepsis using leukocyte surface biomarkers: the ExPRES-sepsis cohort study. Intensive Care Medicine October 2018
10. Liu et al Diagnostic value and prognostic evaluation of Presepsin for sepsis in an emergency department Critical Care July 2013
11. Yin et al The role of soluble thrombomodulin in the risk stratification and prognosis evaluation of septic patients in the emergency department. Thrombosis Research October 2013
12. Li et al Evaluation of a novel prognostic score based on thrombosis and inflammation in patients with sepsis: A retrospective cohort study. Clinical Chemistry and Laboratory Medicine May 2018