Rapid Response Team Activation Criteria: Charter, Staffing Model, and Programme Measures

Most pages about rapid response teams answer the question “what is a rapid response team?” and stop. The person who actually needs the answer is standing up, chairing or auditing the programme, and their questions are different: what belongs in the charter, who staffs the team and at what cost, exactly which physiological thresholds and non-physiological triggers appear on the activation card, whether family members can call, what the activation record must capture, and which numbers the board will ask for at the next quality committee. This guide covers that work, and it is honest about the one thing most RRT pages avoid — the evidence that rapid response systems reduce hospital mortality is genuinely contested.

Scope. This is a guide to designing and governing a rapid response programme: charter, staffing model, activation pathway, documentation, activation review and measurement. It is written for patient-safety officers, quality directors, critical-care outreach leads, risk managers and the committee that owns the deteriorating-patient policy. It is not clinical guidance on assessing or treating a deteriorating patient, and no threshold reproduced below is offered as a clinical decision rule — every one of them is a local decision, made locally, and the sources say so explicitly.

Disambiguation. In hospital operations, RRT means rapid response team. In nephrology and critical care the same three letters mean renal replacement therapy (dialysis and its continuous variants), and in some organisations both usages appear in the same policy library. If your documents use RRT for both, that is a real safety-relevant ambiguity in a call script: spell out “rapid response team” in the charter title, the activation card and the switchboard script, and reserve the acronym for body text where the context is unmistakable.

The rapid response system: afferent and efferent limbs

The useful unit of design is not the team. It is the rapid response system — the term the 2006 first consensus conference on medical emergency teams adopted precisely because programmes kept being described by their staffing and evaluated on their outcomes while the part that actually failed sat in between (DeVita et al., Crit Care Med 2006;34(9):2463–78). AHRQ’s Patient Safety Network describes the same construct as two limbs:

• The afferent limb — the criteria for calling, the monitoring that detects them, and the human decision to pick up the phone.
• The efferent limb — the responding team, its composition, competencies, response time and authority to act.

Almost every programme design conversation is really about which limb you are funding. Hospitals overwhelmingly invest in the efferent limb, because that is the part that has staff, a pager and a line item. The evidence, discussed below, points the other way: in the largest randomised trial of the intervention, the team existed, was resourced, and was called to only 30% of the patients who met its own calling criteria and subsequently needed the ICU. That is an afferent-limb failure, and no amount of team seniority fixes it.

The afferent limb usually has two independent triggers running in parallel: a set of single-parameter criteria on a laminated card, and an aggregate track-and-trigger score. If your organisation runs an aggregate score, the two must be reconciled deliberately rather than left to accumulate side by side — see choosing and calibrating an early warning score, which is the same programme viewed from the trigger end. That page’s design decisions (push versus pull escalation, whether the responder sees the trend or just the trigger, what closes the loop) are the handoff into everything on this page.

What accreditation actually requires — and what it does not

This is the single most commonly misstated fact in this territory, and it changes what you have to defend at survey.

The Joint Commission introduced a National Patient Safety Goal in 2008 that required hospitals to implement systems enabling, in the Goal’s own words, “healthcare staff members to directly request additional assistance from a specially trained individual(s) when the patient’s condition appears to be worsening.” That Goal is widely credited with the near-universal adoption of rapid response teams in United States hospitals. In 2010 it was retired from the NPSG chapter and converted into an ordinary accreditation standard in the Provision of Care, Treatment and Services chapter, PC.02.01.19, “The hospital recognizes and responds to changes in a patient’s condition.”

Two consequences follow, and both are worth getting right in writing before a survey:

1. The requirement attaches to written criteria and a response mechanism, not to a team. The standard’s elements of performance require the hospital to have written criteria describing early warning signs of a change or deterioration in a patient’s condition, and to define when additional assistance is sought. Reported guidance from the accreditor is explicit that a hospital is not required to create a rapid response team or medical emergency team to comply. A small hospital that meets the standard through a defined on-call escalation chain, documented and audited, is compliant. What is not compliant is criteria that exist only as ward custom.
2. The 2026 chapter reorganisation does not change this. Effective 1 January 2026, the Joint Commission’s Hospital and Critical Access Hospital accreditation programs replaced the National Patient Safety Goals chapter with a National Performance Goals chapter. That reorganisation consolidated existing above-regulation requirements; PC.02.01.19 sits in the Provision of Care chapter and is unaffected by it. Other accreditation programs (ambulatory, behavioral health, home care, laboratory, nursing care center, office-based surgery) continued under NPSGs for 2026.

Verification note. jointcommission.org serves an automated bot challenge and returns HTTP 403 to direct retrieval, so the standard number, title and element-of-performance content above are recorded here at reported tier: they are consistent across AHRQ PSNet, accreditation-consultancy compliance summaries and Joint Commission-hosted asset titles, but were not read from the primary manual. Confirm the current wording of PC.02.01.19 and its elements of performance in your own organisation’s E-dition subscription before writing it into a policy or a survey binder. The regulatory floor beneath it is the CMS Conditions of Participation, which your accreditor’s deemed status rests on — see how DNV and Joint Commission accreditation differ for how that deeming relationship works. Outside the United States, the equivalent expectation is NICE clinical guideline CG50, whose recommendation 1.4 requires “a clear and explicit statement of the parameters, cut-off points or scores that should trigger a response,” and whose recommendation 1.9 places the thresholds themselves under local ownership.

Choosing a team model: MET, critical care outreach, or RRT

“Rapid response team” is a label attached to at least three materially different staffing models. AHRQ PSNet distinguishes them by personnel and by duties, and the duties column is what actually drives your budget:

Model	Typical personnel	Duties beyond responding
Medical emergency team (MET)	Physicians (critical care or hospitalist) and nurses	Responds to emergencies only
Critical care outreach	Critical care physicians and nurses	Responds to emergencies; follows up patients discharged from the ICU; proactively reviews high-risk ward patients; educates ward staff
Rapid response team (RRT)	Critical care nurse and respiratory therapist, with physician (critical care or hospitalist) backup	Responds to emergencies; follows up post-ICU patients; proactively reviews high-risk ward patients; educates and acts as liaison to ward staff

The choice is a governance decision with three real dimensions:

• Physician-led or nurse-led? A physician-led MET can make and enact treatment decisions at the bedside without a second call. A nurse-led RRT is cheaper, is more often perceived as approachable by ward nurses (which raises activation rates, which is the point), and needs an explicit, written escalation route to a decision-maker. Neither model is endorsed as preferred by the evidence; NICE says directly that no specific service configuration can be recommended as a preferred response strategy.
• Dedicated or collateral duty? A team assembled from staff who are simultaneously carrying an ICU assignment has a response time that varies with ICU acuity, and that variance will show up in your activation-to-arrival distribution as a long tail rather than a shifted mean. If you staff collaterally, measure the tail, not the average.
• Proactive rounding in or out? The outreach and RRT rows above include proactive review of high-risk ward patients and post-ICU follow-up. That work is where much of the plausible benefit sits, and it is the first thing cut when the team is busy. If you intend the team to round, the charter must ring-fence the time; otherwise write it out of the charter honestly rather than claiming a duty nobody performs.

Whichever model you pick, define the relationship to the cardiac arrest (code blue) team explicitly. The two are different interventions: a code team is summoned after cardiopulmonary arrest; a rapid response team is designed to intervene in the window before it, usually on a general medical or surgical ward. Whether the same roster covers both, and what happens when an RRT activation deteriorates into an arrest mid-visit, is a charter question, not a ward-level improvisation. The interface with a facility-wide emergency footing — mass casualty, evacuation, surge — belongs in a separate structure; see hospital incident command system.

Writing the activation criteria

The single-parameter trigger set

Most programmes converge on a similar card, because the underlying physiology is not institution-specific. AHRQ PSNet publishes a representative set — the criteria any staff member may use to call the team:

• Heart rate over 140/min or less than 40/min
• Respiratory rate over 28/min or less than 8/min
• Systolic blood pressure greater than 180 mmHg or less than 90 mmHg
• Oxygen saturation less than 90% despite supplementation
• Acute change in mental status
• Urine output less than 50 cc over 4 hours
• Staff member has significant concern about the patient’s condition

Criteria used at some institutions in addition: chest pain unrelieved by nitroglycerin; threatened airway; seizure; uncontrolled pain.

Treat that list as a starting point for local decision, not a standard to be adopted verbatim. The numbers on it are wide by design, and every one of them interacts with your own case mix — a set of thresholds calibrated for a general medical ward will behave differently on an oncology unit, a post-surgical floor or an obstetric service. NICE CG50 recommendation 1.9 puts this plainly: trigger thresholds should be set locally and reviewed regularly to optimise sensitivity and specificity. A charter that reproduces a published card without recording who reviewed it against local data, and when it is next due for review, has skipped the only step that makes it defensible.

The staff-concern criterion is not optional

The last line — “staff member has significant concern about the patient’s condition” — is the most important line on the card and the one most often quietly undermined in practice. It exists because deterioration does not reliably announce itself through a threshold crossing, and because the nurse at the bedside is frequently the earliest detector.

NICE CG50 recommendation 1.8 makes the same point structurally: a response should be triggered by either a physiological track-and-trigger score or clinical concern. Two design implications follow:

• A concern-based activation must carry the same standing as a threshold-based one. If the review process treats “criteria not met” as a defect, the criterion is dead within a quarter and your activation rate will fall for reasons that have nothing to do with patient acuity.
• The charter should say, in writing, that no staff member requires permission to activate. The single most common suppressed activation is a nurse who was told to call the intern first. If your escalation policy requires a hierarchical step before the RRT can be called, you have converted a parallel safety net into a serial one and added the delay you built the team to remove.

The accountability framework matters here for the same reason. Where an escalation is not made and harm follows, the analysis has to separate a system that discouraged the call from an individual choice — the distinction the Just Culture algorithm is built to make, and one of the few places where getting the framework right measurably changes the next quarter’s activation rate.

The aggregate-score trigger and the handoff from the early warning score

If your organisation also runs an aggregate track-and-trigger score, the score becomes a second afferent pathway and the two must be reconciled in one policy rather than maintained as separate documents. The cleanest structure treats the aggregate score as the routine, continuous detector and the single-parameter card as the catastrophic-value override, with both terminating in the same response.

The Royal College of Physicians’ NEWS2 Chart 4 shows what a graded terminus looks like in practice: an aggregate total of 5 or more is the urgent response threshold, requiring urgent assessment by a clinician or team with core competencies in the care of the acutely ill; a total of 7 or more is the emergency response threshold, requiring emergency assessment by a team with critical-care competencies including advanced airway skills, with consideration of transfer to level 2 or 3 care. NICE CG50 recommendation 1.10 describes the same graded structure in three bands, with the medium band requiring a simultaneous call to personnel with core competencies in acute illness and the high band an emergency call to a team with critical-care and resuscitation competencies.

Note what that implies for your efferent limb: a graded afferent limb needs a graded response, and the two competency tiers are not the same team. A programme with one team and one pager is answering a three-band trigger with a one-band response. Recommendation 1.11 adds a further bypass — a patient identified as a clinical emergency goes straight past the graded system, with cardiac arrest handled as the high-score group. Full detail on calibrating those numbers, the alarm-burden arithmetic and the counting rules is on the early warning score implementation guide; this page picks up where its escalation call ends.

Criteria that need an explicit local decision

Six questions that a published card will not answer for you, and that a charter should:

• Which populations are in scope? CG50’s own scope excludes children, patients already in critical care areas, and patients in the final stages of a terminal illness. Paediatric and obstetric inpatients are conventionally covered by dedicated criteria and, often, dedicated teams. State which criteria apply to which population rather than letting a general adult card be applied by default.
• How do treatment-limitation decisions interact with activation? An existing do-not-attempt-resuscitation order is not a do-not-call order, and conflating the two is a recurrent source of avoidable harm. The charter should say explicitly that activation is appropriate regardless of resuscitation status, and that the team’s role in that circumstance includes symptom management and clarifying the plan.
• Do the criteria apply in procedural, imaging and outpatient areas? Deterioration in radiology, endoscopy or a hospital-based clinic has different response geography. If those areas are in scope, the response time commitment has to be achievable there.
• Does an activation ever require a specific parameter to be re-measured first? Any “confirm before calling” clause is a delay. If you include one, know why.
• What is the response-time commitment, and is it a target or a standard? Publish a number. A charter that says “immediately” cannot be audited.
• Who is authorised to stand the team down? Ambiguity here produces both premature departures and teams held at the bedside without a role.

Family and patient activation: Condition H and its descendants

Family-activated escalation is not a generic concept, and it should not be written into a charter as one. It has a named origin, a named programme and a documented rationale, and attributing it correctly is both accurate and a good deal more persuasive to a board than “best practice.”

Condition H — “Condition Help” — is the programme at UPMC Shadyside in Pittsburgh that established the model. It was created following the death of Josie King, an 18-month-old who died at the Johns Hopkins Hospital in Baltimore in 2001 after her mother, Sorrel King, raised concerns about her condition and was reassured that her vital signs were satisfactory. Sorrel King and a UPMC colleague developed Condition H, piloted it on a single medical-surgical unit, and it was subsequently extended hospital-wide. The Josie King Foundation continues to publish the programme materials, and patient- and family-activated rapid response has since been adopted widely under a variety of local names.

What makes the model work is that its triggers are not physiological. UPMC’s published criteria for a Condition Help call are:

• “An emergency when patients or support persons cannot get the attention of hospital staff”
• “A communication breakdown with the care team in how care is given”
• “Uncertainty over what needs to be done”

Patients and support persons dial a facility-specific number and a rapid response team is dispatched to the room. Note that two of those three triggers describe a communication failure rather than a clinical change. That is the design insight: a family activation is often the only sensor the system has for a breakdown in the afferent limb itself.

Implementation decisions that determine whether it works rather than sits dormant:

• How does the family learn it exists? A line in an admission booklet produces near-zero activations. Programmes that generate real call volume put it on the whiteboard in every room, on the bedside phone, and into the admission conversation as a spoken sentence.
• Who answers, and is it the same team? Some organisations route family calls to the clinical rapid response team; others to a patient-relations or nurse-manager pathway with escalation into the clinical team. The choice determines what the call can actually resolve.
• Is a “trivial” call a defect? No. A programme whose review process treats communication-driven calls as inappropriate will extinguish the behaviour it just spent a year building. Count them, categorise them, and route the communication-failure themes into the ward’s own improvement work.
• Language access. A hotline that functions only in English is not available to a substantial share of the families most likely to need it.

The programme charter: what the governing document must contain

A rapid response programme is easy to run informally and impossible to defend that way. The charter is the artefact a surveyor, a plaintiff’s expert or a new quality director will ask for. At minimum it should specify:

1. Purpose and scope — populations and care areas covered, and those explicitly excluded, with the reason.
2. Governance — the committee that owns the policy, its reporting line to the medical executive committee and the board quality committee, meeting frequency, and the named accountable executive. Programmes without a named owner drift.
3. Team composition and coverage — roles, required competencies and credentials, hours of coverage, and what happens outside them. If coverage is not 24/7, say so and state the alternative pathway.
4. Activation criteria — the full card, including the staff-concern criterion and the family-activation route, plus the date of last review and the review interval.
5. Activation mechanism — the exact number or call type, who may call, and an explicit statement that no permission is required.
6. Response-time commitment — a published figure, with the measurement definition attached (see below).
7. Team authority — what the team may order, initiate or transfer on its own authority without the attending physician present, and how and when the attending is notified. This is the clause most often left vague and most often the reason a team hesitates.
8. Documentation standard — the required record, and where it lives.
9. Review process — which activations get reviewed, by whom, on what cadence.
10. Measures and reporting — the measure set, definitions, and where each is reported.
11. Education and competency — for the team and, separately, for ward staff on when to call. CG50 recommendation 1.7 requires not only that staff be trained but that they be assessed to ensure they can demonstrate the relevant competencies.

The activation pathway, step by step

Write the pathway as a sequence with an owner and a clock at each step; that is what makes it auditable.

1. Detection. A criterion is met, an aggregate score triggers, or a staff member, patient or family member is concerned.
2. Activation. The caller places the call. Clock starts. No permission step.
3. Dispatch. The switchboard or automated system pages the defined roster. Log the dispatch timestamp separately from the activation timestamp — the gap between them is a fixable system problem, not a clinical one, and programmes routinely discover minutes hiding there.
4. Arrival. First responding team member at the bedside. Clock stops for the activation-to-arrival measure.
5. Structured handover. A defined format for what the bedside nurse conveys: the situation, which criteria are met and which parameters are driving them, the trajectory rather than the single value, what has already been done, and what the caller is worried about. A defined format shortens the call and removes ambiguity.
6. Assessment and intervention. Within the team’s charter-defined authority.
7. Disposition decision. Remain on the ward with a plan; remain with increased monitoring; transfer to a higher level of care; escalate to the attending or to critical care. CG50 recommendation 1.13 sets a useful benchmark for the last of these: a critical-care admission decision should involve both the consultant caring for the patient and the critical-care consultant.
8. Notification. The attending physician is informed, on a defined timescale, regardless of outcome.
9. Documentation and loop closure. Findings and plan return to the record and to the ward team. An activation that produces no documented outcome cannot be audited and frequently cannot be shown to have happened at all.
10. Handback. Explicit statement of who owns the patient afterwards, and whether the team will return to review. Ambiguity here is a recurring source of repeat deterioration going unnoticed.

Documentation: what the activation record must capture

The activation record is the programme’s only evidence base. Design it as a structured form, not a free-text note, because every measure below depends on a field being populated the same way every time.

• Activation timestamp, dispatch timestamp, arrival timestamp — three separate fields, not one.
• Activating role (bedside nurse, charge nurse, physician, therapist, patient or family) and unit.
• Trigger type: which specific criterion or criteria, aggregate score value, or concern-only.
• Vital signs at activation and at team departure.
• Interventions performed, and whether any exceeded the team’s charter authority.
• Disposition, coded to a fixed list.
• Whether an arrest occurred during or within a defined window after the activation.
• Whether the attending was notified, and when.
• Whether this was a repeat activation for the same patient within the defined window.
• A free-text field for the caller’s own account of what prompted the call — the most useful qualitative data the programme generates, and the input that most often surfaces afferent-limb defects.

Reviewing activations

Review is where a rapid response programme stops being a service and becomes a safety intervention. Three review streams, on different cadences, answer different questions:

• Every activation, screened. A short standard screen against defined criteria: was the response time met, was a criterion met well before the call, did the patient arrest, was a repeat activation required, was there a documentation gap. Most activations exit at this stage.
• Selected activations, reviewed in depth. Screen-positive cases plus every activation followed by arrest, unplanned ICU transfer or death. This is the stream that finds afferent-limb failure, and it should specifically look for the pattern MERIT quantified: criteria documented in the record as met, hours before anyone called.
• Aggregate learning. Themes, not cases, into the ward’s improvement work and the morbidity and mortality conference. Where an activation failure meets the definition of a reviewable serious event, it enters the sentinel event pathway with its own comprehensive systematic analysis and corrective action plan requirements.

Two protections are worth understanding before you design the review. First, in the United States, review conducted within a patient safety evaluation system for reporting to a listed Patient Safety Organization may attract federal privilege and confidentiality protections — see patient safety organization work-product privilege for what does and does not qualify, because the boundary is narrower than most committees assume. Second, run improvement changes to the pathway as deliberate PDSA cycles with a stated measure rather than as policy amendments, and if you intend to publish the result, structure the work to the SQUIRE 2.0 reporting guideline from the beginning rather than reconstructing it afterwards.

The measures a board will ask about

The measure set below is the conventional one for this programme type, and each measure exists because it detects a specific failure. Define the denominator explicitly and keep it stable — the most common reporting error in this territory is a rate whose denominator quietly changed from admissions to discharges to patient-days between two board meetings.

1. Activation rate per 1,000 admissions (or discharges)

The core utilisation measure, and the one most likely to be misread. A low activation rate is the danger signal, not the good news: it usually means the afferent limb is suppressed, not that patients are stable. MERIT gives a real anchor for the magnitude of the effect a functioning programme has on this number — introducing a medical emergency team raised the overall emergency-team calling rate from 3.1 to 8.7 calls per 1,000 admissions (p=0.0001). Treat a rate that does not move materially after implementation as evidence the programme has not actually started.

2. Proportion of cardiopulmonary arrests occurring outside the ICU

This is the measure with the strongest evidential support behind it, and the reason is worth explaining to a board. Chan et al.’s meta-analysis of 18 studies covering nearly 1.3 million admissions found that implementing an RRT in adults was associated with a 33.8% reduction in cardiopulmonary arrests outside the ICU (RR 0.66; 95% CI 0.54–0.80) — while finding no associated reduction in overall hospital mortality. Non-ICU arrest is therefore the outcome the intervention most plausibly moves, and the one your programme should be held to. Report it as a rate per 1,000 admissions and as a proportion of all arrests, because the two move independently.

3. Activation-to-arrival time

Report the distribution, not the mean. A collaterally-staffed team has a long right tail that a mean conceals entirely; the operationally useful figures are the median, the 90th percentile, and the count of activations exceeding the charter commitment. Because dispatch is logged separately, this measure decomposes into caller-to-dispatch and dispatch-to-bedside, which point at different owners.

4. Criteria met but not called

The hardest measure to produce and the most informative. It requires sampling records for documented criteria that were met without a corresponding activation. MERIT’s finding is the benchmark to hold against: in the trial’s MET hospitals, the team was called for only 30% of patients who fulfilled the calling criteria and were subsequently admitted to the ICU. A programme that has never measured this does not know whether its team is underused, and underuse is the dominant failure mode.

5. Disposition mix and repeat activations

The proportion of activations resulting in ICU transfer, and the proportion of patients with a second activation within 24 or 48 hours. A rising repeat rate suggests the handback step is failing. A very high ICU-transfer proportion suggests the team is being called too late; a very low one is not automatically good news, and should be read alongside measure 4.

6. Failure to rescue

The outcome measure that frames the whole intervention. The concept comes from Silber et al. (Med Care 1992;30(7):615–29), who defined it as death following an adverse occurrence — separating a hospital’s rate of complications from its ability to recognise and respond to them once they happen. That separation is exactly the capability a rapid response system is meant to supply, which makes failure to rescue the natural programme-level outcome measure even though attribution to the RRT specifically is weak. Report it as context, not as the team’s scorecard.

7. Family-activation volume, separately reported

Report family activations as their own count with their own trigger categorisation. Aggregating them into the main rate hides the only signal you have about whether the family pathway is alive.

For the reporting cadence and structure these measures sit inside, and how they relate to the accreditor’s wider goal set, see National Patient Safety Goals; broader context for the programme sits on the patient safety pillar.

What the evidence actually shows

Rapid response teams are, in AHRQ PSNet’s framing, the intervention that came to exemplify the tension between deploying a conceptually attractive safety intervention on anecdotal evidence and waiting for rigorous evaluation. Any programme document that asserts an RRT saves lives is overstating the literature. Here is what the literature actually says.

• MERIT (Hillman et al., Lancet 2005;365(9477):2091–97) — the largest randomised evidence, and negative on its primary outcome. Twenty-three Australian hospitals were randomised, 11 continuing as usual and 12 introducing a MET system, with a six-month study period. The composite primary outcome (cardiac arrest, unexpected death, or unplanned ICU admission) was 5.86 per 1,000 admissions in control hospitals versus 5.31 in MET hospitals (p=0.640). Individual outcomes were likewise not significantly different: cardiac arrests 1.64 versus 1.31 (p=0.736); unplanned ICU admissions 4.68 versus 4.19 (p=0.599); unexpected deaths 1.18 versus 1.06 (p=0.752). The authors’ interpretation: the MET system “greatly increases emergency team calling, but does not substantially affect the incidence of cardiac arrest, unplanned ICU admissions, or unexpected death.” Two secondary findings matter more to a programme lead than the headline: calling rose from 3.1 to 8.7 per 1,000 admissions, and the team was called to only 30% of patients who met criteria and subsequently needed the ICU. Cardiac arrests and unexpected deaths also fell from baseline in both arms, which complicates attribution further.
• Chan et al. (Arch Intern Med 2010;170(1):18–26) — 18 studies, nearly 1.3 million admissions. In adults: a 33.8% reduction in non-ICU cardiopulmonary arrest (RR 0.66; 95% CI 0.54–0.80), but no reduction in hospital mortality (RR 0.96; 95% CI 0.84–1.09). In children: a 37.7% reduction in non-ICU arrest (RR 0.62; 95% CI 0.46–0.84) and a 21.4% reduction in hospital mortality (RR 0.79; 95% CI 0.63–0.98) — but the paediatric mortality estimate was not robust to sensitivity analyses. The authors also noted that studies frequently reported deaths prevented out of proportion to the reduction in arrests, which raises questions about the mechanism. Their conclusion: “although RRTs have broad appeal, robust evidence to support their effectiveness in reducing hospital mortality is lacking.”
• Rocha et al. (Rev Bras Ter Intensiva 2018;30(3):366–75) — a later meta-analysis reaching a more favourable pooled estimate: mortality RR 0.85 (95% CI 0.76–0.94) and cardiac arrest RR 0.65 (95% CI 0.49–0.87). The authors themselves graded the evidence as low quality for both outcomes, because of high heterogeneity and risk of bias in the primary studies. Cite it as a genuine counterweight, not as a resolution.

AHRQ PSNet’s own summary of the state of play is the fair one: the best available evidence indicates that rapid response systems slightly reduce unexpected cardiac arrests in ward patients but do not affect overall in-hospital mortality; the reasons for inconsistent effects are complex and in some cases relate to local practice and culture that leave the team underused.

How to write this into a business case honestly. Do not claim a mortality benefit. Claim what is supported: a consistent reduction in cardiopulmonary arrest outside the ICU; a mechanism whose failure mode is measurable and fixable (underutilisation); a structural requirement under PC.02.01.19 that has to be met by something; and a detection capability for afferent-limb failures that no other system in the hospital provides. That case is defensible, and it survives the first person on the committee who has read Chan et al. The case that does not survive is the one built on 100,000 Lives Campaign enthusiasm alone.

Common failure modes

• Funding the efferent limb and neglecting the afferent one. The team is staffed, credentialed and idle. Underutilisation, not team quality, is the dominant documented failure.
• A hierarchical permission step before activation. Converts a parallel safety net into a serial one.
• Treating “criteria not met” as an inappropriate call. Kills the concern criterion, which was the point.
• A graded trigger with an ungraded response. Three trigger bands, one team, one pager.
• Copying a published criteria card without local review. Defensible only if you record who reviewed it and when it is next due.
• Unstated team authority. A team that has to phone for permission to act has reintroduced the delay it exists to remove.
• Reporting the mean response time. Conceals exactly the tail you need to see.
• A family-activation pathway announced only in the admission booklet. Produces a compliance artefact with no call volume.
• No handback clause. Repeat deterioration goes unnoticed.
• An unstable denominator. Rates that quietly switch between admissions, discharges and patient-days destroy every trend line the board is looking at.

Frequently asked questions

What is a rapid response team?

A team of clinicians summoned to the bedside when a hospitalised patient shows signs of clinical deterioration, with the goal of intervening before cardiac arrest, unplanned ICU transfer or death. It is the responding — efferent — component of a broader rapid response system that also includes the calling criteria and detection arrangements that trigger it. In hospital operations the acronym RRT means rapid response team; note that in nephrology the same acronym means renal replacement therapy.

What is the difference between a rapid response team and a code blue team?

Timing and purpose. A cardiac arrest or code blue team is summoned after cardiopulmonary arrest has occurred. A rapid response team is designed to intervene in the deterioration window before arrest, typically on a general medical or surgical ward. Some hospitals staff both from one roster; whether yours does, and what happens if an RRT activation deteriorates into an arrest, should be stated in the charter rather than improvised.

Who is on a rapid response team?

It depends on the model. A physician-led medical emergency team pairs critical-care or hospitalist physicians with nurses. A rapid response team in the narrower sense is typically a critical-care nurse and a respiratory therapist with physician backup. Critical care outreach teams pair critical-care physicians and nurses and add proactive ward review and post-ICU follow-up. No configuration is endorsed as preferable by the evidence, and NICE states directly that no specific service configuration can be recommended as a preferred response strategy.

What are the activation criteria for a rapid response team?

A representative published set covers heart rate above 140 or below 40/min; respiratory rate above 28 or below 8/min; systolic blood pressure above 180 or below 90 mmHg; oxygen saturation below 90% despite supplementation; acute change in mental status; urine output under 50 cc in 4 hours; and significant staff concern. Some institutions add chest pain unrelieved by nitroglycerin, threatened airway, seizure and uncontrolled pain. These are a starting point for local calibration, not a standard — NICE CG50 recommendation 1.9 requires thresholds to be set locally and reviewed regularly.

Can a family member call a rapid response team?

At many hospitals, yes. The model was established as Condition H (Condition Help) at UPMC Shadyside, created after the death of Josie King at the Johns Hopkins Hospital in 2001 and developed by her mother Sorrel King with a UPMC colleague. UPMC’s published triggers are an emergency when a patient or support person cannot get staff attention, a communication breakdown with the care team, and uncertainty over what needs to be done — two of the three being communication failures rather than clinical changes.

Does the Joint Commission require a rapid response team?

Reported guidance says no. The applicable standard, PC.02.01.19, requires the hospital to recognise and respond to changes in a patient’s condition, including written criteria describing early warning signs and when to seek further assistance — a hospital may meet it through another defined mechanism. The 2008 National Patient Safety Goal that drove near-universal RRT adoption was retired from the NPSG chapter in 2010 and converted into this ordinary standard. Confirm the current wording in your own E-dition; jointcommission.org blocks automated retrieval, so this is reported rather than primary-verified here.

Do rapid response teams reduce mortality?

The evidence does not support that claim confidently. The largest randomised trial, MERIT, found no significant effect on cardiac arrest, unplanned ICU admission or unexpected death. Chan et al.’s meta-analysis found a 33.8% reduction in non-ICU cardiopulmonary arrest in adults but no reduction in hospital mortality. A later meta-analysis (Rocha et al., 2018) did find a pooled mortality benefit but graded its own evidence as low quality. The defensible claim is reduced non-ICU arrest, not reduced mortality.

What is a good rapid response team activation rate?

There is no universal target, and a benchmark imported from another organisation’s case mix will mislead you. What the literature does supply is a direction and a magnitude: introducing a MET in the MERIT trial raised emergency-team calling from 3.1 to 8.7 per 1,000 admissions. A low or flat rate should be investigated as suppressed activation, not celebrated as stability, and should always be read alongside a criteria-met-but-not-called audit.

What is the afferent limb of a rapid response system?

The detection and calling side: the criteria for activation, the monitoring or scoring that surfaces them, and the decision to call. The efferent limb is the responding team. The distinction matters because the documented dominant failure mode sits in the afferent limb — teams that exist and are not called — while most investment goes to the efferent one.

How does an early warning score relate to the rapid response team?

The score is the afferent limb’s continuous trigger; the team is the efferent limb’s response. A triggering aggregate score should terminate in a defined response tier, and the graded response bands in NEWS2 Chart 4 and NICE CG50 recommendation 1.10 imply more than one tier of responder. See early warning score implementation for score selection, threshold calibration and alarm burden; this guide covers what happens once the call is made.

Sources

• Agency for Healthcare Research and Quality, Patient Safety Network primer, Rapid Response Systems — source for the afferent/efferent construct, the three-model table, the representative calling-criteria set, the 2008 National Patient Safety Goal wording, and the summary of effectiveness evidence.
• Hillman K, Chen J, Cretikos M, et al. Introduction of the medical emergency team (MET) system: a cluster-randomised controlled trial. Lancet 2005;365(9477):2091–97.
• Chan PS, Jain R, Nallmothu BK, Berg RA, Sasson C. Rapid response teams: a systematic review and meta-analysis. Arch Intern Med 2010;170(1):18–26.
• Rocha HAL, Alcântara ACC, Rocha SGMO, Toscano CM. Effectiveness of rapid response teams in reducing intrahospital cardiac arrests and deaths: a systematic review and meta-analysis. Rev Bras Ter Intensiva 2018;30(3):366–75.
• DeVita MA, Bellomo R, Hillman K, et al. Findings of the first consensus conference on medical emergency teams. Crit Care Med 2006;34(9):2463–78.
• Silber JH, Williams SV, Krakauer H, Schwartz JS. Hospital and patient characteristics associated with death after surgery: a study of adverse occurrence and failure to rescue. Med Care 1992;30(7):615–29.
• UPMC, Condition Help (Condition H) — the published activation triggers quoted above. Programme history and materials: Josie King Foundation, Condition Help.
• NICE, clinical guideline CG50, Acutely ill adults in hospital: recognising and responding to deterioration, published 25 July 2007; a January 2020 surveillance review found no new evidence affecting the recommendations. Recommendations 1.4, 1.7–1.13 are cited above.
• Royal College of Physicians, National Early Warning Score (NEWS) 2, 19 December 2017, including Chart 4, Clinical response to the NEWS trigger thresholds.

Joint Commission standard content on this page (PC.02.01.19, its elements of performance, the 2010 conversion of the 2008 National Patient Safety Goal, and the January 2026 replacement of the NPSG chapter with National Performance Goals in the hospital and critical access hospital programs) is recorded at reported tier: jointcommission.org returns HTTP 403 to automated retrieval and the primary manual was not read directly. Verify against your organisation’s E-dition before relying on it. No patient outcome figure on this page is attributed to any named institution other than as published in the cited studies, and no threshold quoted here is offered as a clinical decision rule.