Statistics from Altmetric.com
Editor,—Most fatal events in patients with ischaemic heart disease occur outside hospital and therefore the greatest opportunities for reducing mortality from acute coronary events lie in the prehospital setting1. In their recent paper Soo and colleagues2 published the results of a study to determine whether survival after cardiac arrest outside hospital was influenced by the availability of different grades of ambulance personnel and other health professionals. We are concerned with their conclusions about technician-only crews, and we wish to make some more general comments about their paper.
We feel that the data presented by Soo et aldid not support their statement that “provision of defibrillation plus basic life support by technicians appears to be inadequate compared with the complementary early provision of advanced cardiac life support by paramedics”. Clearly, in the population studied, overall survival was worse with technician-only crews than with paramedic crews. However, as mentioned by Sooet al, technician-only crews dealt with patients whose chances of survival were already prejudiced by several adverse factors—they were less likely to have had a witnessed arrest, bystander cardiopulmonary resuscitation, and an initial rhythm of ventricular fibrillation. It is interesting that among patients with ventricular fibrillation the proportion discharged home alive was higher for technician-only crews than for paramedic crews (10.9%v 10.5%). Viewed from the perspective of survival from ventricular fibrillation (the presenting rhythm most commonly associated with survival) it is thus difficult to conclude that the service provided by technician-only crews was “inadequate” compared with paramedic crews.
The interventions that offer the greatest benefit to victims of cardiac arrest are immediate basic life support and early defibrillation.3 Soo et albriefly mentioned possible strategies aimed at improving the chances of survival, including increasing the number of other resuscitation trained professionals able to provide defibrillation. To optimise access to early defibrillation we believe that the issue of alternative first responders deserves serious consideration. Restoration of circulation and survival depends on the rapidity of defibrillation, regardless of who delivers the shocks, and even small differences in the call to shock time have an influence on survival.4 The fire and police service have already taken on this role in some parts of the UK and others have expressed an interest in supporting the ambulance service as first responders.5
Finally, are Soo et al aware of a similar paper6 from their institution (containing a common set of patients) that concluded that any survival advantages in victims of cardiac arrest associated with paramedic care were short term and diminished over time? We feel this study should have been referenced by the authors.
This letter was shown to the authors, who reply as follows:
Drs Soar and Absalom have highlighted the dangers of interpreting results by just examining factors in isolation. We used multivariate analysis by logistic regression method to take into consideration all factors (including those mentioned by Soar and Absalom) identified in our study that might have contributed to survival chances. This technique is particularly useful when dealing with potential confounders or when assessing interactions between variables. As a result of adjusting for confounders and interactions, the odds ratios we reported do support our conclusions.
We were indeed aware of another paper from our institution1-1 but we considered citation of the latter inappropriate. Sound observational studies require a defined population; this may be the entire population with a specific characteristic (in this case, resuscitation from out-of-hospital cardiac arrest) or a sample taken in some systematic but random fashion from this. The conclusions drawn by Nguyen-Van-Tamet al may well be compatible with the data they reported but their population was neitherentire nor a random sample—such selective populations are a potential source of bias.1-2 We are confident that we identifiedall resuscitation events in Nottinghamshire over a four year period. We chose to analyse and present the complete population, failing to account for just 3% of all patients (as our Utstein style template shows). The claim that the two papers have used “a common set of patients” is clearly wrong. We do not believe that it is possible to make comparisons between our study and that of Nguyen-Van-Tam et al.