As highlighted by Chou et al[1] it is becoming more evident that extracorporeal membrane oxygenation (ECMO) during cardiopulmonary resuscitation (CPR) is feasible and compares well against conventional CPR. As technologies such as ECMO[2] and echocardiography (Echo),[3] previously limited to intensive care units and cardiology suites, become increasingly available in the Emergency Department (ED), it is important that we fully utilize the information and support they can provide to carefully select cardiac arrest patients for advanced ED resuscitation.

The percentage of patients who leave hospital alive following CPR varies from 0% to 20% and has not significantly improved in the last 30 years.[4] The recently published CHEER trial (mechanical CPR, Hypothermia, ECMO and Early Reperfusion), a single center, prospective, observational study from Australia, assessed the CHEER protocol, developed for selected patients with refractory in-hospital and out-of-hospital cardiac arrest.[5] The protocol involved mechanical CPR, induction of intra-arrest therapeutic hypothermia, early commencement of veno-arterial ECMO, and early coronary angiography for patients with suspected coronary artery occlusion. ECMO was established in 24 (92%) of 26 eligible patients, with a median time from collapse until initiation of ECMO of 56 min. Return of spontaneous circulation was achieved in 25 (96%) patients. Survival to hospital discharge with full neurological recovery occurred in 14/26 (54%) patients. Another study from the United States recently reported similar survival rates; 13 of 24 (54%) patients survived to hospital discharge with an ECMO based CPR protocol. Seven of these patients were discharged without any neurological deficit.[6] While these early results show promise for this form of advanced ED resuscitation, caution is required before rolling out this technology for all cardiac arrest patients.

Can our health care systems afford the increased requirement of intensive care bed-hours that such a policy would lead to? Can we select which patients are most likely to benefit from ED-ECMO?

A meta-analysis of predictors of survival from out-of-hospital cardiac arrest in 2010 found that survival to hospital discharge was more likely among those witnessed by a bystander or emergency medical services (EMS), those who received bystander CPR, were found in a shockable rhythm (VF/VT), or achieved return of spontaneous circulation (ROSC).[4] A further meta-analysis in 2012 looked at the of detection of cardiac activity on echo to predict survival during cardiac arrest. Pooled data showed that as a predictor of ROSC during cardiac arrest, echo had a pooled sensitivity of 91.6%, and specificity was 80.0%;[7] promising, but not independently predictive of survival.

Protocols including ED-ECMO are feasible and may be associated with a relatively high survival rate. The introduction of such protocols to emergency medicine should be encouraged, but must involve careful patient selection, optimizing survival benefit. This may involve bedside echo in the ED, as well as other demographic and clinically derived predictors of survival.

1. Chou T.-H. An observational study of extracorporeal CPR for in- hospital cardiac arrest secondary to myocardial infarction. Emerg Med J 2014;31: 441-7.

2. Shinar Z, Bellezzo J, Paradis N, et al. Emergency department initiation of cardiopulmonary bypass: a case report and review of the literature. J Emerg Med. 2012;43(1):83-6.

3. Hayhurst C, Lebus C, Atkinson PR, et al. An evaluation of echo in life support (ELS): is it feasible? What does it add? Emerg Med J. 2011 Feb;28(2):119-21.

4. Stub D, Bernard S, Pellegrino V, et al. Refractory cardiac arrest treated with mechanical CPR, hypothermia, ECMO and early reperfusion (the CHEER trial). Resuscitation 2014. DOI: http://dx.doi.org/10.1016/j.resuscitation.2014.09.010.

5. Sasson C, Rogers MA, Dahl J, Kellermann AL. Predictors of survival from out-of-hospital cardiac arrest: a systematic review and meta- analysis. ?Circ Cardiovasc Qual Outcomes 2010;3:63-81.

6. Peigh G, Pitcher H, Cavarocchi N, Hirose H. Saving Life And Brain With Extracorporeal Cardiopulmonary Resuscitation (E-Cpr) Chest. 2014;146(4_MeetingAbstracts):722A. doi:10.1378/chest.1990723.

7. Blyth L, Atkinson P, Gadd K, Lang E. Bedside Focused Echocardiography as Predictor of Survival in Cardiac Arrest Patients: A Systematic Review. Acad Emerg Med 2012;19: 1119-1126.

Conflict of Interest:

None declared

The authors state that current evidence supports the use of a negative D-dimer to rule out a suspicion of pulmonary embolism (PE) in pregnancy. They show in Table 2 5 relevant studies to support their conclusion. However, the study of Damodaram was the only study that included patients with a clinical suspicion of venous thromboembolism, thus including pulmonary embolism. All other 4 studies included either healthy pregnant patients or patients with a clinical suspicion of deep vein thrombosis (DVT). Evidence found in healthy patients cannot be extrapolated to pregnant symptomatic patients. Performance of the D-dimer test should be tested in pregnant patients with a clinical suspicion of PE before any recommendation can be made. Since PE is still one of the leading causes of maternal death, ruling out this diagnosis accurately is imperative.

Furthermore, in non-pregnant patients, use of D-dimer test as a stand -alone test is not supported. The authors state that the Wells score can be a useful tool in calculating pretest probability, as its categories remain relevant. However, this statement is not supported by evidence. The title of this BET may be interpreted as if a D-dimer test can be used as a stand-alone test. In non-pregnant patients with a high suspicion of PE (Wells score >4 points), the presence of PE is found to be 10%, despite a negative D-dimer (Wells, Thromb Haemost 2000). Therefore, the most important conclusion should be that there's a lack of good-quality evidence concerning the management of a clinical suspicion of PE in pregnancy.

Conflict of Interest:

None declared

Dear Editor,

The "Sepsis Six" resuscitation bundle is now common in UK hospitals, and endorsed by the College of Emergency Medicine, SIGN guidelines, and the Royal College of Physicians.1-3 However, the evidence behind it remains limited and must be re-evaluated as new data emerges - especially since the results of the PRoCESS and ARiSE trials on Early Goal Directed Therapy (EGDT) were published this year.4,5

The paper supporting Sepsis Six was originally published in 2010 in this Journal, and remains (as far as we are aware) the only published evidence on Sepsis Six.6 We believe this trial is internally flawed, and unable to provide convincing evidence of the benefit of Sepsis Six.

We have two main criticisms of this sepsis 6 trial - firstly, it is inconsistent. Given that two large randomised controlled trials this year have shown EGDT to have negligible effect on mortality, it is hard to interpret the results of this trial - which showed a relative risk reduction of 91% (mortality dropped from 90.3% without EGDT to 7.5% with it). Also, the subgroup of patients who were shocked shows internal inconsistencies. The worst performing group - with a staggering 97.1% mortality were the shocked patients who received Sepsis 6 - yet if they received EGDT alone, their mortality was 8%. However, the combination of Sepsis 6 and EGDT in shocked patients performed best of all - with a 4.7% mortality!

It seems inconceivable that in septic shock Sepsis 6 would increase mortality to 97.1% if performed alone, but reduce it to 4.7% if performed in combination with EGDT (which we know from RCT data has a negligible effect).

The only possible explanation is that of confounding effects - younger, fitter people were receiving EGDT and Sepsis 6- who had a low mortality rate and survived, and less fit people did not receive it This leads to our second criticism - lack of co-variate analysis. Although the authors comment on the difference in age between some of the cohorts, they did not perform a multivariate logistic regression (or similar) that could have included age, MEWS score, source of infection. Given the complexity and difference between the subgroups, it is impossible to compare them without covariate statistical analysis.

It is very hard to draw causality from observational trials (and the authors do not try here), but given the significant problems with this trial, we do not feel it can be interpreted as supporting evidence for Sepsis 6 either.

Although we do not aim to be unnecessary critical of what may be seen as 'common sense practice', we have three main concerns. Firstly, the claim that sepsis 6 in unlikely to be harmful is not a sufficient basis for it's continued use. 'Common sense' medicine has repeatedly been proven to be harmful on numerous occasions. Secondly, emergency departments and wards are busy and under huge time pressure, and protocols that might tie up staff for an hour or more must be scrutinised. Thirdly, sepsis 6 is likely to lead to the greater use of stat doses of antibiotics in any febrile patient, given the impetus to administer them within the first hour for any unconfirmed sepsis diagnosis. We do not know the long-term impact of this on antibiotic resistance - which is rapidly increasing.

Sepsis 6 means well, but is unproven, and should be treated as such.

Fergus Hamilton Bryony Lewis

1. SIGN. Care of deteriorating patients: Guideline 139. 2014. Available at: http://www.sign.ac.uk/pdf/SIGN139.pdf. Accessed October 14, 2014.

2. RCP. Acute Care Toolkit 9: Sepsis.; 2014.

3. College of Emergency Medicine. Clinical Standards for Emergency Departments. 2013. Available at: http://www.collemergencymed.ac.uk/Shop- Floor/Clinical Standards. Accessed October 14, 2014.

4. The ARISE Investigators and the ANZICS Clinical Trials Group. Goal-Directed Resuscitation for Patients with Early Septic Shock. N. Engl. J. Med. 2014:141001063014008. doi:10.1056/NEJMoa1404380.

5. 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):1683-93. doi:10.1056/NEJMoa1401602.

6. Daniels R, Nutbeam T, McNamara G, Galvin C. The sepsis six and the severe sepsis resuscitation bundle: a prospective observational cohort study. Emerg. Med. J. 2011;28(6):507-12. doi:10.1136/emj.2010.095067.

Conflict of Interest:

None declared

Wright and colleagues1 discuss some interesting issues around blood cultures and sepsis outcome measurement. We agree that evaluating and tracking sepsis associated mortality over time is important. It would facilitate assessment of the impact of changes in practice, both intended effects of sepsis improvement interventions and unintended consequences of other initiatives, e.g. restrictive antibiotic policies, as well as epidemiological changes in populations. An ideal outcome measure would be "simple, objective, clinically meaningful, resistant to ascertainment bias, and, ideally, suitable for automation using data routinely stored in electronic health records"2. Unfortunately, there is no such perfect measure to track sepsis associated mortality.

Analysis of recorded International Classification of Diseases, Tenth Revision (ICD-10) codes is frequently utilised for tracking trends in healthcare, including mortality3. At first, this seems appealing as ICD- 10 codes should be simple and clinically meaningful and are available from routinely collected data. However, inconsistency in assignment of codes and changes in coding patterns make this susceptible to ascertainment bias thereby compromising the validity and reliability of applying this as a standalone measure2. Where remuneration and penalisation in healthcare is linked to coding, there is also potential for gaming and coding trends to emerge which do not reflect a real change in the population2,3. Additionally, due to the presence of a range of appropriate codes that could be applied to patients with sepsis, analysis of coding is complex4. Both primary and secondary discharge codes require interrogation and organ -specific infection codes, such as community-acquired pneumonia (ICD- 10 code J18.9), do not indicate whether the patient had sepsis, thereby reducing the sensitivity of this as a measure. All-cause mortality is also utilised within the literature. It is simple, objective and routinely collected, however, it lacks sensitivity to detect the impact of improvement interventions in clinical practice.

We agree with Wright et al that blood cultures should only be drawn where appropriate. As advocated as part of the Sepsis Six5, when patients fulfil two or more Systemic Inflammatory Response Syndrome (SIRS) criteria (see Box 1) and have a suspected or confirmed source of infection blood cultures should be obtained7.

Box 1: SIRS criteria6

SIRS present when two or more of the following measured:

* Temperature <36oC or >38oC

* Heart Rate >90 beats/min

* Respiratory Rate >20 breaths/min OR PaCO2 <4.3kPa

* White blood cell count <4000/?L or >12000/?L

One concern which we share with Wright et al is that blood cultures are often taken from patients who do not meet SIRS criteria, as found in an audit of patients with cellulitis presenting to an Acute Medical Unit8, but this is not encouraged if the Sepsis Six is implemented correctly. Such snapshot audits are limited to the clinical observations available at the time of data collection and this is likely to result in an underestimation of the prevalence of sepsis at any time during a hospital admission among patients who have a blood culture taken. With the exception of immunocompromised patients or those with suspected endocarditis, blood cultures should only be taken from patients who have two or more SIRS criteria otherwise there is potential to waste resources and cause patient harm7.

Wright et al reported a rise in blood culture sampling within NHS Forth Valley. This should not necessarily be interpreted as a negative development. The Scottish Trauma Audit Group9 reported that 39% (n=251) of patients fulfilling criteria for severe sepsis in the emergency department had no blood cultures sampled during the first 48 hours following presentation. From this baseline the number of blood cultures being drawn should rise initially if the Sepsis Six is being implemented reliably. Interestingly the beginning of NHS Forth Valley's rise in blood culture sampling appears to pre-date the beginning of the Scottish Patient Safety Programme (SPSP) national Sepsis Collaborative in January 2012. This apparently crude rate may reflect an already increasing awareness of sepsis but may also have been influenced by the increasing number of inpatient discharges in NHS Forth Valley in the period 2010-201310.

Wright et al suggest that the measurement of mortality among patients with a positive blood culture would be more useful than blood cultures alone. However, the proportion of patients with a positive blood culture is a small subset of the wider sepsis population with one cohort reporting that only 21% of patients with sepsis had a bacteraemia11. Therefore, to restrict the measurement of sepsis associated mortality to this population would significantly underestimate the true incidence of sepsis. Additionally, the presence of a positive blood culture has not been demonstrated to be independently associated with higher mortality once adjusted for age, sex and comorbidity11,12. These considerations limit the clinical meaningfulness of this as a measure, and negate Wright et al's concern about "treating culture-negative sepsis with broad spectrum antibiotics". This concern is not consistent with the objective of the Sepsis Six, which is to ensure that patients with clinical signs of sepsis receive prompt, appropriate and timely antibiotic treatment.

The proxy outcome measure for the SPSP national Sepsis Collaborative examines the thirty day mortality among adult inpatients in acute hospitals who have had a blood culture sampled. Utilising the existing Scottish Hospital Standardised Mortality Ratio (HSMR) model, the data will be adjusted for changes in clinical activity and acuity of care. However, even crudely adjusted (e.g. for age, sex and occupied bed days or admissions) blood culture mortality is a superior outcome measure compared to the other currently proposed options. We accept that it is not perfect, and are undertaking further work to validate its scope and applicability, including investigating combining blood culture sampling with other routinely available data. Improving the selection of patients for blood culture sampling, as discussed above, would improve the performance of this measure. However, even as it stands, it is simple, relatively objective, clinically meaningful and more resistant to ascertainment bias than most other options, and is suitable for automation using data routinely stored in electronic health records.

References

1. Wright DJ, Beckett DJ, Cooke B. Use of sepsis 6 raises some interesting questions. Emerg Med J Published Online First: 01.09.2014

2. Rhee C, Gohil S, Klompas M. Regulatory Mandates for Sepsis Care - Reasons for Caution. N Engl J Med 2014; 370: 1673-1676

3. Vaughan Sarrazin MS, Rosenthal GE. Finding Pure and Simple Truths with Administrative Data. JAMA 2012; 307(13):1433-1435

4. MacPherson D, Griffiths C, Williams M, Baker A, Klodawski E, Jacobson B, Donaldson L. Sepsis-associated mortality in England: an analysis of multiple cause of death data from 2001-2010. BMJ 2013; 3(8)e002586

5. Daniels R, Nutbeam T, McNamara G, Galvin C. The sepsis six and the severe sepsis resuscitation bundle: a prospective observational cohort study. Emerg Med J 2011; 28:507-512

6. Jones GR, Lowes JA. The systematic inflammatory response syndrome as a predictor of bacteremia and outcome from sepsis. Q J Med 1996; 89:7

7. Coburn B, Morris Am, Tomlinson G, Detsky AS. Does This Adult Patient With Suspected Bacteremia require blood cultures? JAMA 2012; 308:5

8. Marwick C, Rae N, Irvine N, Davey P. Prospective study of severity assessment and management of acute medical admissions with skin and soft tissue infection. J Antimicrob Chemother 2012; 67:1016-9.

9. The Scottish Trauma Audit Group. Sepsis Management in Scotland. 2010.

10. http://www.isdscotland.org/Health-Topics/Hospital-Care/Inpatient- and-Day-Case-Activity/ [last accessed 22.09.14]

11. Marwick CA, Guthrie B, Pringle JE. Identifying which septic patients have increased mortality risk using severity scores: a cohort study. BMC Anesthesiol 2014; 14:1

12. Phua J, Ngerng WJ, See KC, Tay CK, Kiong T, Lim HF, Chew MY, Yip HS, Tan A, Khalizah J, Capistrano R, Lee KH, Mukhopadhyay A. Characteristics and outcomes of culture-negative versus culture-positive severe sepsis. Critical Care 2014; 17:R202

Conflict of Interest:

None declared