COMMENTS ON: "BET 3: Evaluation of intra-aortic balloon support in cardiogenic shock"

Maria Cristina Acconcia(a), MD, Flavia Chiarotti(b), DStat, Francesco Romeo(c), MD, Quintilio Caretta(d*), MD.

(a)Department of Cardiovascular Disease, University of Rome - La Sapienza, Rome, Italy (b)Department of Cell Biology and Neuroscience, Italian National Institute of Health, Rome, Italy (c)Department of Cardiovascular Disease, University of Rome - Tor Vergata, Rome, Italy (d) Department of Clinical and Experimental Medicine, University of Florence, Florence, Italy

*Corresponding author at: Quintilio Caretta, MD, Clinical and Experimental Medicine, University of Florence, Largo Brambilla, 3 - 50134 Florence, Italy. Tel: 0039-3487809379; Fax: 0039-06-20904008;E-mail: qcaretta@unifi.it.

In the meta-analysis "The outcome of intra-aortic balloon pump support in acute myocardial infarction complicated by cardiogenic shock according to the type of revascularization" by Romeo et al (2013), we assessed the impact of intra-aortic balloon pump (IABP) on in-hospital mortality, safety end points (stroke, severe bleeding) and long-term survival, using risk ratio (RR) and risk difference (RD) estimates(1). We found that IABP support did not significantly affected the risk of death in patients who did not undergo reperfusion, while it reduced significantly in the Thrombolysis (TT) subgroup and significantly increased in the percutaneous coronary intervention (PCI) subgroup the in-hospital mortality.

Humphrey et al (2013) recently performed a short-cut review to establish whether IABP improves mortality in cardiogenic shock after acute myocardial infarction (2) including our meta-analysis(1) . In Table 3 they stated that the use of the observational studies and the different mortality rates among the three subgroups (83.9% for the no reperfusion subgroup, 66.9% for the TT subgroup, and 38.4% for the PCI subgroup), suggested a weakness of the study. They concluded that "the role of IABP support in patients with cardiogenic shock from myocardial infarction remains unclear, without evidence of clear confirmed benefit compared to conventional therapy, especially when PCI is available". With respect to these remarks we would like to make some comments. First, the statement on the observational studies is formally correct, but in the scientific literature the evidence of IABP support in cardiogenic shock is mainly based on registry data, due to feasibility; indeed, in our meta-analysis the inhospital mortality was analyzed on 14186 patients from 16 studies, 13 observational, including 13526 patients, and 3 randomised controlled trials, contributing with 22, 40, and 598 patients, respectively. Furthermore in our meta-analysis we adopted the more conservative random effect model to take into account heterogeneity among studies. In adjunct Benson and Hartz (2000) performed meta-analyses of randomised clinical trials and observational studies and found that treatment effect estimates from observational studies reported after 1984 were similar to those obtained in randomised controlled trials(3). Also, in their meta-analyses based on randomised clinical trials and observational studies on identical clinical topics, Concato et al (2000) found that the average results of well-designed observational studies (with either a cohort or a case-control design) were markedly similar to those of the randomised controlled trials(4). Finally, an integrated approach is advisable because "Discarding observational evidence when randomised trials are available is missing an opportunity. Conversely, abandoning plans for randomised trials in favour of quick and dirty observational designs is poor science"(5). Second, we think that the authors did not take into appropriate account the role of reperfusion strategies as confounding factor in the meta- analysis. Indeed, from a clinical point of view it is quite different to support patients affected by cardiogenic shock with IABP alone, or with IABP in combination with TT or PCI. This is clearly demonstrated in our meta-analysis by the fact that the three groups of patients who did not receive IABP support (control groups) had significantly different in- hospital mortality rates due to the clinical treatment (83.9%, 66.9%, 38.4%, for no reperfusion, TT and PCI, respectively) (see Figure 4 in Romeo et al, 2013)(1). Thus the actual impact of IABP support could be assessed only if the subgroups were stratified according to clinical treatment. And this must not be interpreted as a gap, but as correct application of the statistical method. Finally, we performed a trial sequential analysis (TSA) using TSA program (The Copenhagen Trial Unit, Center for Clinical Intervention Research CTU, Denmark; version 0.9 beta; available at www.ctu.dk/tsa)(6,7) to settle any further doubt. TSA provides the required information size, a threshold for a statistical significant treatment effect and a threshold for futility. We calculated the relative risk reduction (RRR) both for RR and RD, using the event proportion observed in the control group (i.e. the basal risk) and the actual difference in risks between the experimental and control group observed in our meta-analysis. TSA performed on TT and on PCI subgroups demonstrated that the sample sizes were adequate to verify the hypotheses. The required number of participants was reached in both subgroups of patients (TT: RR, n=1646 and RD, n=1287, RRR=26.6%; PCI: RR, n=2671 and RD, n=2523, RRR=-18.2%). The monitoring boundaries constructed to detect significance were crossed by the z-curves. Thus, our meta- analysis can be considered as conclusive in contrast with the final statement by Humphrey et al (2013)(2).

References 1. Romeo F, Acconcia MC, Sergi D, et al. The outcome of intra-aortic balloon pump support in acute myocardial infarction complicated by cardiogenic shock according to the type of revascularization: A comprehensive meta-analysis. Am Heart J 2013:165:679-92. 2. BET 3: Evaluation of intra-aortic balloon support in cardiogenic shock. Emerg Med J 2013;30:1063-4. 3. Benson K, Hartz AJ. A comparison of observational studies and randomized, controlled trials. N Engl J Med 2000;342:1878-86. 4. Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies, and the hierarchy of research designs. N Engl J Med 2000;342:1887-92. 5. Ioannidis JPA, Haidich A-B, Lau J. Any casualties in the clash of randomised and observational evidence? No--recent comparisons have studied selected questions, but we do need more data. 2001;322:879-80. 6. Thorlund K, Engstr?m J, Wetterslev J, et al. User manual for trial sequential analysis (TSA). Copenhagen Trial Unit, Centre for Clinical Intervention Research, Copenhagen, Denmark. 2011. p. 1-115. Available from www.ctu.dk/tsa 7. Wetterslev J, Thorlund K, Brok J, et al. Estimating required information size by quantifying diversity in a random-effects meta- analysis. BMC Medical Research Methodology 2009;9:86.

Conflict of Interest:

None declared

The case Dr. Alzetta describes is similar to the ones I described. Although these cases are rare in any one location and undocumented especially after death I believe that taken nationally they are of significant numbers. The evidence lies in a paper written to discover the cause of the dramatic increase in asthma deaths in the sixties by Speizer, Doll et al. They studied all the deaths in England and Wales for six consecutive months using Death Certificates from the Registrar General from 1st Oct. 1966-31st March 1967 in which asthma was the underlying cause. They wrote to the doctors and hospitals concerned for full details of the cases. Most of the cases surprisingly occurred in persons with mild asthma only 59% had ever been admitted to hospital. Death was sudden and unexpected in 80% of cases. In 25% death occurred in less than one hour and only 29% survived more than 24 hrs. That death was commonly sudden is confirmed by the fact that 59% of deaths (109 out of 184) were certified by coroners. In 39% of cases (67 out of 171) the practitioner had not regarded the patient as suffering from severe asthma in the terminal episode. This paper is very important as : 1) it demonstrates the only way that one can collect and study these cases in any number. 2) It is the only record we have of the numbers of deaths due to asthma in that era and the dramatic increase at that time. The reason being that asthma deaths were included with all other types of respiratory diseases until several years late. It is very important that this work is repeated today to discover the prevelance today.

Conflict of Interest:

None declared

Arbon and colleagues [1] helpfully demonstrate the differences between major incident triage as taught on courses such as MIMMS [2], and that practised ¡§in the real world¡¨. I have two concerns however.

Firstly, the authors incorrectly suggest that the role of ¡§scanning and sweeping¡¨ before undertaking triage has been de-emphasised in current teaching. In the MIMMS course, the mnemonic ¡§CSCATTT¡¨ places the ¡§A¡¨ of assessment ahead of ¡§T¡¨ for triage and this is firmly reinforced in both the manual and the course. An estimate of the number and type of casualties, hazards and so on, before triage, is emphasised strongly in current teaching.

Secondly, the authors conclude that a complex process of paramedic decision making expanding beyond physiological parameters is preferable to ¡§theoretical models¡K such as triage sieve and sort¡¨. When time is available, triage can and does involve elements such as mechanism and anatomy of injury, and the experience of senior clinicians. This is described in MIMMS as secondary triage, and is concerned with subsequent patient disposition. However, triage sieve (and the American ¡§START¡¨) are both simple, rapid, repeatable, objective primary triage systems that are performed in the initial few minutes when immediate life-threats must be identified and prioritised without delay.

Physiological primary trauma systems are based on the best evidence available, despite the difficulty establishing robust evidence in mass casualty incidents. Until there is evidence that other factors predict imminent mortality better than physiology, it would be dangerous to add subjective elements to the critical process of initial life-saving triage.

Whether one approaches triage ethically from a utilitarian or an egalitarian position, the strongest imperative is simply to save the most lives.[3] To do this, triage should accurately predict which patients need the earliest life-saving intervention. But it should also minimise any subjective bias on the part of the practitioner, especially bias according to gender, race, age, disability, social usefulness or past merit. If practitioners move ¡§beyond¡¨ triage systems which focus solely on immediate life-threat, and instead apply ¡§gut-feeling¡¨ or subjective experience there is a danger that triage will be both inaccurate and unjust.

It is important to appreciate the differences between what is taught and what is practised, and understand the reasons for those differences, but we must not assume that departure from what is taught is a sign of good or mature practice. Effort should be made instead to explain the ethical and clinical importance of accurate triage so that practitioners understand the need to follow standardised and simple protocols that save the most lives, and research must continue to improve our methods of primary triage.

References

1. Arbon P, Zeitz K, Ranse J et al. The reality of multiple casualty triage: putting triage theory into practice at the scene of multiple casualty vehicular accidents. Emerg Med J. 2008;25:230-4.

2. Advanced Life Support Group (2002). Major incident medical management and support: the practical approach at the scene (2nd ed). London: BMJ, 2002.

3. Winslow GR. Triage and justice. London: University of California Press, 1982.

Editor, I read the recent publication by Mann et al. with a great interest. Mann et al. concluded that " There is a significant risk of harm with false-positive diagnoses and potential delays in appropriate treatment [1]." I agree that there are several problem in diagnosis of swine flu. Several problems can lead to the failure of using any scoring system or algorithm for diagnosis [2-3]. On the other hands, although several new diagnostic tools can be availble, the problem of false positive can be seen. How to manage and weight for risk and benefit on using simple method with possible false negative and new modern tools with possible high cost and false positive should be the topic to be discussed.

References

1. Mann C, Wood D, Davies P. An evaluation of the UK National Pandemic Flu Service swine flu algorithm in hospitalised children, and comparison with the UK National Institute for Health and Clinical Excellence fever guideline. Emerg Med J. 2010 Sep 3. [Epub ahead of print]

2. Wiwanitkit V. Scoring system for diagnosis of swine flu. Heart Lung. 2010 Jul-Aug;39(4):345-6.

3. Wiwanitkit V.Swine flu: the present pandemic infectious disease. Kulak Burun Bogaz Ihtis Derg. 2009 Mar-Apr;19(2):57-61.

Conflict of Interest:

no