Results may have been due to education of ward staff

EDITOR—Proof that a medical emergency team can reduce the incidence of and mortality from unexpected cardiac arrest is eagerly awaited, as such a proposal is intuitive. However, the number of such arrests can be influenced by several factors, including the number of “do not resuscitate” decisions made. Buist et al's paper fails to take this into account, and suffers from other methodological errors too.1

The study used a historical control group and was undertaken in a setting in which there was already a trend towards a reduced incidence of and mortality from cardiac arrest. Moreover, the case mix varied considerably between the two study periods. The authors' definition of cardiac arrest included patients who had not actually experienced an arrest yet excluded four who had been allocated do not resuscitate orders but for whom a call was made. Ward patients who do not have a cardiac arrest have a better outcome than those who do; hence by adopting a loose definition the study denominator has been artificially enhanced, giving a false benefit.

Some patients receive cardiopulmonary resuscitation despite it being futile, and thus the resuscitation status of critically ill patients must be established. However, any increase in do not resuscitate orders inevitably reduces the incidence of and mortality from unexpected cardiac arrests. The introduction of a medical emergency team increases the number of do not resuscitate orders.2 Buist et al report that, in 1999, the medical emergency team made 13 such orders for patients who subsequently died but do not report the overall incidence of these orders in the hospital in either year studied.

The study's design makes it impossible to separate the beneficial impact of the formal education process provided to ward staff from that due to specific interventions by the medical emergency team. The education may have led to better ward care or to more do not resuscitate orders being applied by staff who were not members of the team. Perhaps education alone led to the results that are being attributed to the team.

The cover of the issue of the BMJ containing Buist et al's paper, showing a medical emergency team in action, suggests that the authors have confirmed a beneficial role for the team. This may be correct, as the team may have reduced the number of inappropriate cardiac arrest calls by increasing the incidence of do not resuscitate orders or by improving the education of ward staff. What is not proved is that a medical emergency team can reduce the incidence of and mortality from unexpected cardiac arrest.

Bottom up approach works too

EDITOR—Like Buist et al,1 we have recognised that care preceding admission to the intensive care unit can be improved.2 To do this we chose a combination of a bedside physiology based scoring system,3 increased education of both nurses and doctors in the recognition of critically ill patients, and use of “outreach” nurses with skill in intensive care who can both support patients on the ward and help with their admission to the intensive care unit.

This initiative was backed by a protocol that ensured escalation of care if the patient was not improving and was for resuscitation. This bottom up approach, which contrasts with Buist et al's use of specialists who are “parachuted in,” was welcomed by nurses and junior doctors, who were empowered to call for help. The physiology based scoring system was applied to 2000 surgical patients and identified all patients who went on to die in hospital well before they did so. The clear protocol for responding to these patients set standards that could easily be audited.

Before and after studies such as Buist et al's will always be vulnerable to failure to measure confounders. Were more patients allocated “do not resuscitate” orders with and without involvement of the medical emergency team, and did the intensive care unit change its admission policy? It would have helped if secular trends in similar hospitals had been measured, but conducting a randomised controlled trial of a mixed educational and therapeutic intervention in which one of the treatment outcomes is intensive treatment is extremely difficult and some would argue unethical.

Surely the question is not whether knowledgeable staff should see patients with physiological problems early and involve more senior staff if they are not improving but rather how such a service should be delivered.

Authors' reply

EDITOR—We suspect that Smith and Nolan may be kept waiting some time for absolute proof that medical emergency teams can reduce the incidence of and mortality from cardiac arrest in hospital. As they point out, our study, which was done in a single hospital with a historical control group and an unblinded intervention and analysis, is well short of the gold standard for randomised controlled trials.

We believe that an intervention such as a hospital wide medical emergency team is hard to evaluate by a randomised controlled trial in a single institution. Some of the shortcomings of our study may be addressed by the forthcoming multicentre randomised cluster control study of a medical emergency team intervention. This study (the MERIT study), which is being undertaken under the auspices of the Australian and New Zealand Intensive Care Society Clinical Trials Group, will randomise an intervention by a medical emergency team in clusters to over 30 participating hospitals throughout Australia and New Zealand.

We would accept Smith and Nolan's comment concerning the effect of “do not resuscitate” orders on the incidence of cardiac arrest in hospital. However, we think that if the medical emergency team increased the rates of such orders throughout the hospital and thus reduced the incidence of cardiac arrest this would be a positive effect of the medical emergency team system. As pointed out, we did not measure this; furthermore, we are unaware of any data that support this contention, despite the reference to a study by Parr et al (M J A Parr, personal communication).1

Finally, with respect to the case mix; although the number of planned admissions did increase, the number of emergency admissions also increased, but by a much larger absolute number. In fact, most of the increase in planned admissions arose through better management of the “others” category (mostly direct admissions from private specialists), which reduced dramatically over the study period. As in any acute care hospital, over time our patient population has become older with more severe illness.

We agree with King et al that a top down approach to some of these issues will not greatly alter the nature of the nurses' and junior doctors' interaction with patients. There is still much to do in identifying patients at risk, developing systems for scoring severity of illness of hospital inpatients, and developing education packages for our junior medical and nursing staffs.