Rapid Response Systems: Update and Critique
By David J. Pierson, MD, Editor, Professor, Pulmonary and Critical Care Medicine, Harborview Medical Center, University of Washington, Seattle, is Editor for Critical Care Alert.
In an attempt to enhance what it considered a sluggish nationwide response to the Institute of Medicine's calls for reducing error and improving patient outcomes in hospital care, the Institute for Healthcare Improvement (IHI) initiated in 2004 an ambitious, highly visible, 18-month program. Called the 100,000 Lives Campaign, it was based on the implementation of "evidence based practices" in 6 clinical areas or "Planks" (Table 1).1 Eighteen months later, IHI pronounced the program a resounding success, announcing that 122,300 lives had been saved as a result of the program. Bolstered by this success, IHI promptly extended its goals to the saving of 5 million lives by the end of 2008, using the "planks" of the 100,000 Lives Campaign and other interventions.2
The Six "Planks" in the Institute for Healthcare Improvement's 100,000 Lives Campaign3
While the evidence base supporting several of the "planks" in Table 1 is strong, that for rapid response teams—and the increasing mandate for their establishment in all hospitals—has been called into question.3,4 This brief essay examines the rapid response system (RRS) phenomenon and the current evidence that initiation of such systems in hospitals reduces mortality and events such as cardiac arrest and unanticipated transfers to the ICU.
What Are Rapid Response Teams?
Hospitalized, acutely-ill patients who are not in ICUs are at risk for serious adverse effects such as cardiac arrest, unplanned admission to the ICU, and death. Although these events may occur suddenly and without warning, in many if not most instances they are preceded by warning signs—tachypnea, tachycardia, oxyhemoglobin desaturation, or mental status changes, for example—that could alert staff to intervene and potentially prevent the adverse outcome. The RRS concept emerged from an appreciation of this potential: recognition or suspicion of a patient's critical unmet need would trigger a rapid sequence in which a pre-established team came to the bedside, assessed the patient, and either intervened directly or facilitated such intervention.5
Teams operating within an RRS can have different structures and memberships, and in fact the literature has been somewhat confusing with respect to their composition and terminology. The most commonly described types of team are the medical emergency team (MET), the rapid response team (RRT), and the critical care outreach team (CCO), which are somewhat distinct:
- A MET is usually headed by a physician. It can prescribe therapy, manage the airway, place central lines, and initiate ICU-level care on the ward, in the angiography suite, or wherever the emergency occurs.5
- An RRT is typically headed by an ICU nurse or other non-physician, usually includes a respiratory therapist, and generally has quick access to a designated in-house intensivist or other physician (in teaching hospitals, commonly an ICU fellow or senior resident) when needed.
- As described in the literature, CCO teams encompass one of the above structures but also include a more prospective, proactive component aimed more specifically at prevention rather than response.
The nomenclature used to designate these and other variations of RSS has been a source of confusion, and a consensus group recently made recommendations intended to reduce this.5
A Systematic Review of the Evidence
To examine the strength of the evidence supporting the use of RRSs, Winters and colleagues at Johns Hopkins performed a systematic review of studies that had been published on this topic through mid-2005.4 Using accepted methodology for such analyses, they identified 10,228 abstracts of publications in English that were potentially relevant. Studies were subjected to the formal analysis if they included data on outcomes in both intervention and control groups, and if they reported mortality and/or the incidence of cardiac arrest. Reports of out-of-hospital interventions and of restricted populations within the hospital (such as patients transferred out of an ICU to the general ward) were excluded. The investigators attempted to evaluate potential bias by collecting comparison data on intervention and control groups, and assessing the degree to which the reported measures and outcomes were explicitly defined.
Using the criteria described, 46 studies were evaluated more thoroughly. Of these, 38 were rejected according to the investigators' a priori criteria, and 8 studies were included in the formal review. Six of these 8 were observational in nature, 5 employing historical controls and 1 a concurrent control group. Two other studies were interventional, 1 a cluster-randomized trial within a single institution6 and the other a large, multi-center study also using a cluster-randomized design.7
None of the 8 studies had a control group that Winters et al considered to be truly comparable to the intervention group; half of the studies made some attempt to adjust for differences between intervention and comparison groups, but the attempts varied and their effectiveness could not be determined with certainty. With varying detail and objectivity, the 8 included studies listed the "alert criteria" used to activate the RRS in the intervention group (see Table 2 below).
Criteria used to activate the rapid response systems in the 8 included studies (data from reference 5)
|Alert Criterion||Number of studies
In which it was reported
Change in mental status
Desaturation (pulse oximetry)
“Worry” (staff concern)
Seven studies (5 observational, 2 cluster-randomized) reported hospital mortality data. The relative risk for mortality with the use of RRSs among the 5 observational studies was 0.87, with 95% confidence interval 0.73-1.04. The corresponding relative risk in the 2 cluster-randomized trials was 0.76 (95% CI, 0.39-1.48), and the p value for heterogeneity in these studies was 0.01, suggesting a high degree of heterogeneity. Neither result for a reduction in hospital mortality was statistically significant, with the 95% CI in each instance surpassing 1.
Five of the 8 included studies (4 observational, 1 cluster-randomized) reported cardiac arrest data. For the observational studies the pooled relative risk for cardiac arrest in the hospital was 0.70 (95% CI, 0.56-0.92) in comparison with controls, which was statistically significant (albeit with a high degree of heterogeneity at p <0.01). In the cluster-randomized study reporting cardiac arrest data the relative risk was 0.94 (95% CI, 0.79-1.13) compared with control.
With respect to unanticipated transfer to the ICU from a general inpatient floor, 3 observational studies that reported these data had a pooled relative risk for such transfer of 0.84 (95% CI, 0.55-1.26) with the RRS. In the multicenter cluster-randomized trial,7 the relative risk for unanticipated ICU admission with a RRS was 1.04 (95% CI, 0.89-1.21) vs control.
Winters et al concluded from their systematic review of published studies that the evidence supporting the use of RRS is "weak to moderate" with respect to their primary objectives—reducing hospital mortality and cardiac arrest rates and the incidence of unanticipated admission to an ICU.5 The authors could not find strong support for considering RRSs to be standard of care, or for mandating their institution in all hospitals.
Conclusions and Current Unknowns
When deficiencies of care exist, improvement in the quality of care will lead to better outcomes. The introduction of RRSs has been one means for improving the quality of care, targeting serious adverse events such as cardiac arrest, unplanned ICU admission, and death. Insofar as preventable adverse events can be detected and effective interventions brought to bear to head them off, initiation of an RRS should be expected to decrease their incidence. However, as the systematic review by Winters et al5 shows, objective, rigorous evidence documenting such a decrease is scant. The more rigorous the study design (and thus the higher on the hierarchy of evidence quality), and the farther removed from anecdotal, purely observational reports, the less compelling is the support from the current literature. Also lacking is demonstration that an RRS is necessarily the best way to bring about the desired improvements. Table 3 summarizes some of the current unknowns in this era.
Remaining Unknowns about Rapid Response Systems in Hospitals
To a certain extent this discussion is academic, since establishment of RRSs (especially in the form of RRTs) is being thrust upon hospitals by the Joint Commission and other forces. However, it is unlikely that the final word on RRSs is in from the standpoint of evidence-based medicine, and future studies will probably shed additional light on their benefits and costs as well as on how they or future modifications should best be employed.
- Berwick DM, et al. The 100,000 Lives Campaign: Setting a goal and a deadline for improving health quality. JAMA. 2006;295:324-327.
- http://www.ihi.org (assessed 12/5/07)
- Wachter RM, Pronovost PJ. The 100,000 Lives Campaign: A scientific and policy review. J Comm J Qual Patient Saf. 2006;32(1):621-627.
- Winters BD, et al. Rapid response systems: A systematic review. Crit Care Med. 2007;35(5):1238-1243.
- DeVita MA, et al. Findings of the first consensus conference on medical emergency teams. Crit Care Med. 2006;34(9):2463-2478.
- Priestley G, et al. Introducing critical care outreach: A ward-randomised trial of phased introduction in a general hospital. Intensive Care Med. 2004;30:1398-1404.
- Hillman K, et al. Introduction of the medical emergency team (MET) system: A cluster-randomised controlled trial. Lancet. 2005;365:2091-2097.