An elderly man was admitted with an acute abdomen and free air visible under the diaphragm on CXR. He was fluid resuscitated before undergoing emergency laparotomy, where a perforated duodenal ulcer was oversewn. He was admitted to ICU postoperatively, extubated the next morning and deemed fit for discharge to the surgical ward later that day. Due to a lack of surgical beds, he was eventually discharged from ICU at 22:30. Eight hours post discharge, he was urgently re-referred to ICU after being found moribund on the ward. Before he could be seen and assessed he suffered an unrecoverable asystolic arrest. Review of his observation charts showed that there had been a clear deterioration in recorded observations, including hypotension for the two preceeding hours. However, the Early Warning Score had been calculated incorrectly, and no escalation had occurred.
What evidence is there that rapid response systems are effective in preventing patient deterioration and improving outcomes?
The majority of in-hospital cardiac arrests are preceded by a potentially reversible deterioration, with many patients having abnormalities in simple vital sign measurements for many hours before clinical deterioration (1). The extent to which these events are preventable remains unknown, but it has been speculated that early identification of the signs of deterioration on general hospital wards followed by immediate treatment to prevent further deterioration and reverse the effects of hypoxia and ischaemia may prevent progression to cardiopulmonary arrest in many cases. This rationale, coupled with the concentration of expertise and resource to deal with seriously ill patients within critical care areas, has led to the widespread development of what are generically described as rapid response systems (RRS) (2). Several forms of RRS have been described, including critical care outreach teams, rapid response teams and medical emergency teams. Although differing in composition and activation criteria, all of these systems incorporate a mechanism of identifying the seriously ill patient early and triggering a rapid response to them from a team experienced in the care of the critically ill.
These systems have now been widely implemented in the UK, and have come to be seen by many as standard of care. However, although the concept of early identification and treatment of physiological instability makes intuitive sense, the effectiveness of RSSs remains uncertain with only limited evidence that they improve outcome. Several studies have suggested a reduction in the incidence of unplanned ICU admission, cardiac arrest and deaths, but these studies have been limited by the use of historical controls and lack of randomisation (3). To date only two large randomised controlled trials (RCT) into the effect of rapid response systems have been published.
In the first of these studies, Priestley and colleagues randomised wards within a single UK centre to the introduction of a nurse-led critical care outreach service, with this intervention being introduced to all wards in sequence. Although there was a (non-significant) trend to increased length of hospital stay in this study, outreach intervention significantly reduced in-hospital mortality compared to controls (odds ratio 0.52; 95% CI 0.32-0.85)4. However, in the largest RCT investigating the impact of rapid response systems, the MERIT study, 23 Australian hospitals were cluster randomised to continue functioning as usual or to introduce a medical emergency (MET) system. The primary outcome measure in this study was a composite of cardiac arrest, unexpected death and unplanned ICU admission in the 6-month period following MET activation. Although the introduction of the MET greatly increased emergency team calling incidence, there was no significant difference in the composite primary outcome measure between control and MET hospitals in this study (5.86 vs 5.31 per 1000 admissions, p=0.64) (5).
A recent systematic review of the impact of rapid response systems which included both the Priestley and MERIT studies, together with 6 observational studies, concluded that there was only weak evidence that rapid response systems are associated with a reduction in hospital mortality (pooled relative risk 0.76, 95% CI 0.39-1.48) or cardiac arrest rates (pooled relative risk 0.94, 95% CI 0.79-1.13)3. However, the authors of this review highlighted that limitations in the quality of the original studies, heterogeneity between studies and the wide confidence intervals limited their ability to conclude that rapid response systems are effective interventions, and that further large RCTs are required to clarify their efficacy before they are adopted as a standard of care. In particular, the optimal composition of the responding team, the identification of appropriate triggers to activate the team and strategies for increasing the utilisation of the team have yet to be resolved.
A factor that may have compounded the failure to identify the clinical deterioration in the patient described in this case is the overnight discharge (22:30). In a retrospective study of more than 79,000 consecutive patients, Priestap and colleagues found that patients discharged from ICU between 21:00-06:59 had a higher hospital mortality (9.2% vs 12.2%, p<0.001) and a higher ICU readmission rate (5.2% vs 6.3%, p=0.001), even after adjustment for illness severity and age (6). Night time discharge occurred in ~10% of patients in this study, and intriguingly, only half of these discharges were followed by a subsequent ICU admission within 2 hours, suggesting that bed pressure was not the sole reason for the discharges. The NICE guidelines relating to the care of acutely ill patients in hospital published in 2007 suggest that the transfer of patients from a critical care area to a general ward should be avoided between 22:00 and 07:00 if at all possible, and, that if the transfer of patients occurs between these times, it should be recorded as an adverse event.
The cause of the clinical deterioration in this case remains unclear, but progression to cardiopulmonary arrest appears to have occurred over a period of several hours, with clearly charted deterioration in vital signs over this time period. Whether events could have been prevented by the earlier identification of deterioration with subsequent triggering of intervention to prevent further deterioration and reverse the effects of hypoxia and ischaemia is unknown. Although there is no formal RRS in the hospital in which this case occurred, there appears to be limited evidence that these systems significantly improve patient outcome. The use of early warning charts is routine in the hospital in question and the observations recorded should have triggered an urgent referral to the hospital at night team several hours prior to his actual referral. The fact that this did not occur represents a clear systems failure, and highlights the need for the education of all medical and nursing staff in the recognition of critically ill patients together with clear guidance of when and how to trigger urgent review.
Patients discharged from ICU at night have a higher hospital mortality and ICU readmission rate than those discharged during the day. The transfer of patients should therefore be avoided between 22:00 and 07:00 if at all possible.
1. Hillman, K.M. et al. Antecedents to hospital deaths. Intern Med J 31, 343-8 (2001).
2. Devita, M.A. et al. Findings of the first consensus conference on medical emergency teams. Crit Care Med 34, 2463-78 (2006).
3. Winters, B.D. et al. Rapid response systems: a systematic review. Crit Care Med 35, 1238-43 (2007).
4. Priestley, G. et al. Introducing Critical Care Outreach: a ward-randomised trial of phased introduction in a general hospital. Intensive Care Med 30, 1398-404 (2004).
5. Hillman, K. et al. Introduction of the medical emergency team (MET) system: a cluster-randomised controlled trial. Lancet 365, 2091-7 (2005).
6. Priestap, F.A. & Martin, C.M. Impact of intensive care unit discharge time on patient outcome. Crit Care Med 34, 2946-51 (2006).