Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Editor,—We congratulate Edhouse et al1 on their attempt to try and produce some clarity in the murky waters of diagnosing and treating patients with acute myocardial infarction (AMI), who present with left bundle branch block (LBBB). Unfortunately, we feel that the study has some limitations that could provide for some confusing “take home” messages. There are four points to consider:
The original article, Sgarbossa et al,2 was published with an editorial and generated a number of subsequent letters,3, 4 which were rightly critical of the cohort chosen and the subsequent extrapolation of results. These views are not adequately reviewed and the “spin” in the discussion by Edhouse et al in our opinion, is overly supportive of Sgarbossa's criteria.
The prevalence of AMI in Edhouse's article is 0.5 and is unusually high for patients presenting to accident and emergency with cardiac sounding chest pain. The method section seems to suggest that these patients were derived from a database of patients eligible for thrombolysis, which would not be an appropriate study population. This is an important point that requires clarification by the authors.
In the conclusion, the first sentence rightly points out the need for thrombolysing all patients with LBBB and persisting cardiac pain. The last sentence suggests serial ECGs. It is unclear which side of the fence the authors wish to sit on. The assertion that serial ECGs should be used if “the diagnosis is in doubt” is based upon five patients. We would suggest this recommendation is untenable on the evidence provided.
Schlipak et al5 have published the most robust study thus far (with a more representative prevalence of 28% for AMI) for this group of patients. It is disappointing that this study is not reviewed in more detail by Edhouse et al. The result and conclusions in Schlipak's paper invalidate using any of the Sgarbossa criteria for detection of AMI in the presence of LBBB. They also performed a decision analysis suggesting that an extra 10 to 12 lives per thousand patients could be saved by adhering to the American College of Cardiology/American Heart Association guidelines on the subject.6 These recommend that in the absence of contraindications, thrombolysis should be used in all patients with LBBB who have a clinical presentation of AMI.
Future research strategies using new technologies may herald greater precision and provide the solution for this difficult group of patients. These could include rapid bedside cardiac enzyme analyses, vectorcardiography and/or neural networks. In the meantime we should adhere to the best available evidence.
The authors reply
We welcome the opportunity to discuss the points raised by Mohammed et al. Our data include all patients with acute chest pain and left bundle branch block (LBBB) on presentation or within 12 hours of admission. Thus we have complete data on all patients with LBBB and acute myocardial infarction (AMI) during the study period.
Mohammed states that a 52% prevalence of AMI is unusually high, and claims the prevalence of 28% quoted by Shlipak is a “more representative” figure, but provides no evidence to support this assertion. Hands et al1 found a prevalence of 57%, a figure very similar to our own. We note that Shlipak2 mistakenly attributes a prevalence of 25% to the study by Hands.
The recommendation that all patients with chest pain and LBBB receive thrombolysis is correct and evidence-based, but is also rather simplistic. This is reflected in the significant under-use of thrombolysis both in the UK3 and the United States,4 and the lengthy treatment delays these patients experience.3
At least half the patients with LBBB and chest pain are not infarcting.1 While the overall mortality reduction justifies the administration of thrombolysis to all such patients,5 this does not make the decision to treat an individual patient any easier. It is hard to think of another circumstance in which we expose patients to a significant stroke risk on the strength of a diagnosis of which we are only 50% certain.
Management decisions are further complicated if the presentation is not classic, or if relative contraindications exist. The decision to administer thrombolysis, especially in the elderly population, is often a careful balancing act between potential benefit and complication risk. The manifest reluctance of doctors to expose patients to significant risk without a definite diagnosis is perfectly understandable, and merely reiterating the guidelines does not help the clinician at the sharp end.
We acknowledge the limitations of our small, retrospective study. Nevertheless, the differences between our results and those of Shlipak are remarkable. Ischaemic change evolves over time, even in the presence of LBBB.6 If only the presenting ECG is analysed, evolving changes will be missed and the sensitivity of the predictive criteria7 underestimated, particularly if patients present early in the course of their infarct.
Shlipak noted the ECG criteria infrequently2; in contrast we found at least one of the criteria in 19 of 24 patients with AMI.3 When we analysed only the first ECG, the presence of any of the criteria indicated a diagnosis of AMI with sensitivity = 62.5%. A series of ECGs was available in only 33% of patients, but incorporating even this small number into the analysis increased the sensitivity of the criteria to 79% (specificity 100%).3 We note that in Shlipak's study only the presenting ECG was analysed.2
Our unpublished data on 797 consecutive patients presenting with AMI revealed a median interval between onset of pain and arrival at hospital of 135 minutes, whereas audit data from the United States report a median of 89 minutes.4 A relatively early presentation, coupled with analysis of only the first ECG, may partly explain the low sensitivity found in Shlipak's study.
Our findings strongly support those of Sgarbossa in showing that a significant number of infarcting patients can be identified quickly and counselled confidently regarding their need for thrombolysis.3 This is of immediate practical benefit to clinicians and should facilitate considerable reductions in treatment delays. Thrombolysis should also be recommended when the ECG criteria are absent, along with an individual risk benefit assessment to allow patients to participate in the management decision. Where patients do not receive immediate thrombolysis serial ECGs are essential; if evolving changes indicate a definite infarct, the balance of risk and benefit may change considerably.