Rethinking Rapid Response Teams

Current debate in the medical community centers on the benefits of rapid response teams (RRTs), hospital-based teams composed of clinicians with intensive care unit (ICU)–level clinical expertise. These teams rapidly respond when the condition of patients being cared for outside of the ICU suddenly deteriorates, and such patients often require transfer to ICUs.¹ Those on one side of the debate suggest that RRTs save lives; this assertion is supported by common sense, numerous anecdotal reports, and some observational studies.² Those on the other side of the debate suggest that preventing, recognizing, and treating deteriorating patients is common sense. How best to achieve this remains elusive based on systematic reviews,³ which have failed to show benefit of RRTs but note that RRT studies were often of poor quality and clinicians often failed to call an RRT when they should have, leading to uncertainty in the estimates of benefit. Proponents favor further research, encouraging hospitals to experiment with strategies such as RRTs, enhanced nurse staffing, or hospitalists who would respond to deteriorating patients, stressing prevention rather than recovery from deterioration. Those on both sides of the debate are united in their frustration that patients are needlessly experiencing morbidity and agree that preventing patients' health from deteriorating is the optimal solution.

The debate obscures a more fundamental question: why are RRTs needed in the first place? The answer seems to be simple. An RRT is needed when the condition of a patient who is receiving care in a medical/surgical unit deteriorates or requires ICU-level expertise to avoid further deterioration or even death. There are 2 reasons patients deteriorate. First, some deteriorate despite adequate clinical care. These patients would benefit from having an organized system to identify and treat patients whose conditions worsen, such as an RRT or code team. Second, patients deteriorate because of inadequate care; in other words, the level of care (eg, clinician training, staffing) provided to the patient in the inpatient unit is inadequate for the patient's condition. Even though empirical evidence regarding the proportion of RRT calls caused by each of these reasons is lacking, the philosophy of RRTs is premised on the idea that current care is inadequate; therefore, introducing ICU-level care will benefit the patient. If current care is adequate, an RRT is not likely to make a difference.

Underlying inadequate care is that patients have been admitted to a unit that provides inadequate care. A triage error or inability to admit or transfer a patient to the preferred unit is the main driver⁴ of patient misplacement. Underlying the triage error is the way patient flow is managed or mismanaged. Every physician and nurse would prefer that patients are cared for in a unit that can provide the appropriate level of care, where sufficient physician, nurse, and monitoring resources are available. Physicians commonly request that their patients remain in the ICU or are admitted to a specific nursing unit, often with monitored beds, believing care is better in some units than others.

Intensive care units and monitored beds are scarce resources, demand for these resources periodically exceeds supply, and patients are often not admitted to these preferred units.⁵ This situation is especially problematic in hospitals without critical care physicians who use clearly defined protocols to coordinate the use of monitored beds. A common although often erroneous solution is to add more ICU and monitored beds. Even if the cost of adding a bed (about $1 million capital for a regular inpatient bed⁵ ) is ignored, experience suggests that adding more beds does not solve this problem. Eventually, demand for these beds will again exceed capacity.⁵

Why, then, is there a seemingly insufficient number of ICU and monitored beds? Why don't hospitals define which patients should use ICU and monitored beds? One reason is that mismanaged patient flow in the form of artificial peaks does not allow compliance with any such definition. Despite average US hospital occupancy of 66% to 67%,⁶ hospitals are periodically overcrowded. The key word is “periodically.” Rarely are particular hospital units overcrowded 100% of the time, in which case more beds would be needed. Rather, these hospitals are typically overcrowded on certain weekdays (usually Tuesdays, Wednesdays, or Thursdays) and at particular times on those days (usually during the middle of the day).

These peak time–based bed needs are well known in hospitals, but their occurrence appears to be largely unpredictable. During these times, everything goes wrong: ambulances are diverted, patients are boarded in emergency departments, patients are often prematurely discharged from the ICU to make room for more ill patients or elective surgical cases, nurses are overloaded and stressed,⁵ and patient discharges take place prematurely, resulting in patient readmissions.⁷ On days like these, hospital clinicians and managers face an unlikable dilemma: to admit a patient to a nonpreferred unit or to board the patient in the emergency department or the postanesthesia care unit until a bed in the preferred unit becomes available. During these times, proper patient placement is an exercise in wishful thinking and the definition of a preferred bed becomes “the one that is available.” Because of these artificial peaks in scheduled admissions, US hospitals ration ICU beds, monitored beds, and even regular-floor beds every day. Surprisingly, this situation is much easier in managing natural (unpredictable) fluctuations in the demand. In this case, queuing theory, widely used in other industries although rarely used in health care,^{5 ,8} is helpful, providing an accurate way of matching random demand to fixed capacity.⁹

During these times of stress, patients are at risk of being admitted to the wrong type of unit and of receiving an inadequate level of care. These patients have an increased likelihood of needing an RRT. It is these patients, who have experienced a triage error, that RRTs “save.”

High-stress days are largely caused by the way hospitals schedule their elective admissions.⁵ When there is an unnecessary peak in elective surgical admissions, many of whom require ICU care, ICU beds and other monitored beds are occupied, resulting in many patients being admitted to an incorrect unit and receiving an inadequate level of care, manifested as either inadequate nurse or physician training or an insufficient nurse-patient or physician-patient staffing ratio.¹⁰

Consequently, RRTs are being launched to rescue these patients if they experience signs of clinical deterioration. Yet it seems perverse to measure the success of RRTs by counting the number of saved lives that were put at risk by triage errors, driven by ineffective management of patient flow. Ironically, if number of saved lives is the appropriate metric for RRT effectiveness, even more lives could be “saved” by closing the ICUs and monitored beds, decreasing the number of floor nurses by half, or admitting patients anywhere in the hospital and sending an RRT to rescue them. This would be like rewarding a pilot for ultimately landing safely after he or she had made a dangerous decision to take off during inclement weather.

Researchers should seek to identify and mitigate risks borne by patients admitted to an incorrect hospital unit. For the majority of patients whose condition deteriorates while receiving inadequate care in an improper unit, efforts should be made to ensure that they receive adequate care in the proper unit, to move away from taking credit for rescuing patients who experience triage errors, to focus on patient flow, and to provide each patient with the right care at the right time, not more and not less.