Failure to rescue patients in hospital is often due to a systems failure and breakdown in care at a number of levels which we have described as a “chain of survival” [40]. Safe care of deteriorating patients depends on robust and reliable recording of vital signs, recognition of abnormalities, reporting of patient deterioration as soon as detected, an appropriate and timely response, and more often than not a repeat cycle to check whether interventions have had the desired effects (Fig. 37.2). All elements of the chain need to function seamlessly in order to provide reliable and safe care [41]. The following sections will describe the elements of the chain of survival, the reasons for failure and possible mediating mechanisms.

Deterioration of patients is often clear in hindsight from the characteristic changes in vital signs [42, 43] or pathology results [44]. The majority of patients admitted to Intensive Care Units or suffering cardio-pulmonary arrests demonstrate signs of deterioration for a minimum of 6 h prior to the “event” [45]. In the majority of patients failure to rescue is therefore not due to a failure to record vital signs but failure to recognize the trend in the patient status. It is unclear how many patients have cardiac arrests without physiological abnormalities purely due to the fact that no observations or no complete set of observations were recorded in the hours prior to the event. In general terms, a full set of vital signs could comprise respiratory rate, oxygen saturations, blood pressure, heart rate, temperature, level of consciousness, and possibly urine output. The most powerful parameter predicting patient deterioration, and at the same time the most often missed vital sign, is the respiratory rate [46]. Respiratory rate (RR) changes with thoracic cage and lung conditions, metabolic acidosis, infection, fever, etc. RR is measured manually and not electronically like other key measured parameters and is more time consuming. The optimal frequency of observations for acutely unwell patients is not clear from the literature. A report about “Standardising the assessment of acute-illness severity in the NHS” by the Royal College of Physicians [47] recommended at least 4 h vital signs on general wards. In Dutch hospitals the frequency is often less [48]. In many other systems vital signs might only be assessed by healthcare providers once or twice per day and consist of blood pressure, heart rate, and temperature only, thus potentially missing opportunities to capture deterioration through a full set of vital signs.

Standardization of vital sign recordings might improve the number of opportunities for intervention. Standardization of vital sign recordings and analysis of abnormality is described in the literature as Medical Emergency Criteria [49] (Table 37.1. Medical Emergency Team criteria) and as Early Warning Scores [50] (Table 37.2. Modified Early Warning Score). Triggers of abnormal physiological signs are complemented by nurse concerns as an important safety net for those patients who do not or not yet exhibit gross abnormalities [51].

Validation of Early Warning Scores has been undertaken predominantly in acutely unwell medical patients [47] and to a lesser extent in surgical patients [52]. Standardization can be anchored in clinical teams through training using a common model to describe severity of illness [48]. Automated monitoring can also improve monitoring of post-surgical patients [53–55].

Perception of “illness” and mental models of providers about the disease severity can have a major influence on behavior and decisions of healthcare professionals. In the words of Peter Senge [56]: “Mental models are deeply held internal images of how the world works, images that limit us to familiar ways of thinking and acting. Very often, we are not consciously aware of our mental models or the effects they have on our behavior.” Mental models are subtle but powerful. Subtle, because we usually are unaware of their effect. Powerful, because they determine what we pay attention to, and therefore what we do. For example, if a young patient “looks well” with red cheeks and a smile despite a systolic blood pressure of 70 mg than the nursing staff is much less likely to trip the alarm than in an elderly patient who has been unwell for several days with the same vital signs. The perception that young patients are usually well and can’t really be that ill remains an ongoing recognized risk and a form of normalized deviance [57]. Recognition of physiological abnormalities is often in the context of what is expected: it is easier to spot “abnormal” in a patient in whom staff expect this abnormality. For example, in a post-operative patient hypotension might be expected; a patient with chronic obstructive pulmonary disease might post-operatively be more short of breath because of metabolic acidosis or volume overload but his or her respiratory rate will be interpreted in the context of their previous condition. Furthermore, we know that elderly patients’ physiological response to acute illness is attenuated [58]. This might make it more difficult for staff to classify changes in vital signs as “critical” and requiring further action. Age might, however, not be the defining factor for prognosis. Crucial to the understanding of acute physiology is the underlying degree of frailty. Frailty is a syndrome with measurable metrics [59] based on pathophysiological modeling and epidemiological data from large cohorts of aging patients (Fig. 37.3).

Increased levels of frailty are associated with higher mortality, higher levels of complications after surgery, and higher mortality after admission to ICU. The majority of patients with physiological deterioration and those experiencing failure to rescue are frail [60] (Fig. 37.4).

Reporting on patient abnormalities or staff concerns are an important function of communication between professional groups. Real or perceived hierarchy plays a major role in acting on available warning signs [61]. Professionals might hesitate to discuss abnormalities if they fear and lack psychological safety or have a non-supportive recipient of the information. In the context of activation of RRS nurses might be hesitant to call a physician if they fear that the physician will not take their concerns seriously or will be short on the phone because of real or perceived pressures of work. The failure to report can be “simple forgetfulness” when workflow pressures and conflicting priorities over-ride the need to escalate care. It can be a conscientious decision that the reporting of abnormalities is not a priority for the patient or workflow. Nursing staff might judge abnormalities to be within the expected range for a given patient or hope that they are transient and resolve without further intervention.

Failure to treat can be the consequence of a failure to record or recognize or equally a failure despite recording and recognizing. Correct treatment will depend on the clinical competencies (i.e., knowledge, skills, and attitudes) of the treating clinician and their mental model of the patient’s disease and situation [62]. Reliability of treatment can be enhanced by using “care bundles” [63] and by making available a RRT with critical care skills [64].