The Need for a Standard Outcome for Clinical Deterioration in Children’s Hospitals

Unrecognized clinical deterioration is a common and significant source of preventable harm to hospitalized children.¹  Yet, unlike other sources of preventable harm (eg, patient falls), clinical deterioration outside of the ICU lacks a clear, “gold standard” outcome to guide prevention efforts. Instead, children’s hospitals rely on locally defined criteria, including adaptations of published outcome measures (“outcomes,” unless specified). For example, some hospitals develop internal definitions for emergent or preventable transfers to the ICU on the basis of the timeline of the transfer and care interventions required. Unfortunately, this reliance on context-specific outcomes severely limits multicenter learning, which is crucial for identifying effective and generalizable interventions for harm prevention.

In this article, we compare common outcomes for evaluating clinical deterioration outside of the ICU. We highlight sources of variation and vulnerability, with an aim of underscoring the need for a standard outcome for harm prevention. Of note, establishing a generalizable approach to preventing harm from unrecognized clinical deterioration will require alignment on both a standard outcome and a “bundle” of evidence-based structures, processes, and tools for achieving harm prevention. In this report, we will focus primarily on the need for a standard outcome as the most immediate priority. However, we will also comment on potential bundle elements aimed at detecting and managing clinical deterioration outside of the ICU (eg, pediatric early warning scores^2–4 ). A broader review of these structures and processes can be found in a recent publication by our team.¹ 

Ultimately, at the core of this discussion is a simple but important goal: To keep hospitalized children safe from preventable harm from unrecognized clinical deterioration. This framing acknowledges that children may become increasingly unwell during an acute hospitalization, yet it also expects health care teams to recognize and respond to signs of critical illness, and to escalate care to the ICU when appropriate. If these expectations are not met, hospitalized children unfortunately suffer the consequences, for delayed recognition and escalation of deterioration to the ICU (even by a few hours) is associated with increased risk of in-hospital mortality and prolonged length of stay.^1,5–7 

We will use terms in this article that may be less familiar to clinicians who infrequently interact with critically ill children and the hospital systems designed to care for them. We hope the content of this article will be easily interpreted by any reader, but we acknowledge that a fair amount of technical jargon is required to accurately describe this unique set of outcomes. To aid in understanding the terms and technical concepts used in this field, Fig 1 provides a broad overview of the language used to describe events in which hospitalized children experience clinical deterioration outside of the ICU.

The following outcomes are commonly described in the literature and operational practice. Each outcome has unique criteria, but unplanned ICU transfer is a universal requirement. Of note, “unplanned ICU transfer” is sometimes cited as a distinct outcome of clinical deterioration.⁸  We have excluded it here because it is impacted significantly by hospital setting (eg, varying thresholds for ICU escalation) and is not validated for relevant patient outcomes.

We also note that the outcomes described below are agnostic of patient admission source. That is, these outcomes apply to all patients admitted to a ward unit, regardless of whether they were admitted from the emergency department, or if they were directly admitted from a specialty clinic or from home. These outcomes also apply to all hospital settings caring for children (eg, freestanding children’s hospital, children’s hospital within an adult hospital system, etc), although there are important contextual challenges to consider in applying these outcomes in settings where PICU and/or emergency resources may be scarce.

The Children’s Hospital Association (CHA) defines a qualifying code outside of the ICU as an event of cardiopulmonary arrest (CPA) or acute respiratory compromise (ARC) (Fig 1). Using the illustrative patient case cited in the introductory paragraph, this event of shock would not meet criteria for a qualifying code outside of the ICU by the CHA’s criteria. This is because the patient in this example required neither chest compressions nor cardioversion outside of the ICU (ie, the event would not qualify as a CPA); nor did they receive any airway interventions outside of the ICU (ie, the event would not qualify as an ARC). Thus, by failing to meet either set of criteria (CPA, ARC), this event would not be considered a CHA qualifying code outside of the ICU.

In reviewing this outcome, it is important to acknowledge that the CHA’s criteria have the advantage of aligning with the broader resuscitation literature, which commonly uses CPA and ARC nomenclature.^9–11  However, events requiring cardiopulmonary resuscitation (CPR), emergent intubation and/or bag mask ventilation outside of ICU settings are relatively uncommon in North American children’s hospitals. This limits the ability of this outcome to power analyses necessary to evaluate harm prevention-related improvement work. This outcome also lacks sensitivity to subacute signs and nonrespiratory etiologies of deterioration (eg, shock, such as in the case example). This limits opportunities for system learning. Additionally, this measure often requires manual review of ICU transfers to identify qualifying events, which could invite bias from incomplete electronic health record (EHR) data and/or limited interrater agreement.

Parshuram’s SCD outcome¹²  expands on the CHA’s criteria by including fluid resuscitation, vasopressor initiation, and extracorporeal membrane oxygen initiation (Fig 1). SCD events also include criteria met in the 1 hour after ICU transfer. Using the case example above, this event of shock would qualify as an SCD because the patient required intubation within 1 hour after ICU transfer.

As a result of its expanded timeline and criteria, the SCD outcome could drive greater event capture needed to power improvement analyses and drive system learning. However, as with CHA codes, SCD events do not comprehensively address more rare potential etiologies of acute deterioration, such as neurologic causes. The SCD outcome is also complex and often requires time-intensive, manual review of ICU transfers for event identification, which could invite similar biases to those discussed for CHA qualifying codes.

Developed by Brady and colleagues, ETs (formerly “UNSAFE” transfers¹³ ) involve specific interventions in the 1 hour before or after ICU transfer^5,7,13  (Fig 1). The cited case example above would qualify as an ET by all criteria (ie, 60 mL/kg intravenous fluid resuscitation, vasopressor initiation, and intubation within 1 hour after ICU transfer).

Importantly, ETs have been validated as a predictor of patient outcomes, including length of stay and mortality.^5,7  ETs also occur at increased frequency relative to more critical deterioration outcomes (eg, CPAs), and thus have greater utility for improvement analyses. ET criteria may also capture early signs of deterioration, such as evolving shock (as in the example case). This outcome also has interoperability, because ETs can be identified through automated data capture in the EHR,^7,14  a process that can minimize bias.

However, the ET outcome also has several limitations. First, it does not comprehensively address critical interventions provided to deteriorating patients, such as CPR. But because CPR is often coincidental with other ET criteria, this limitation rarely affects event identification. However, some patients present more nuanced challenges. For example, tracheostomy patients have an invasive airway in place, so escalation of their respiratory support would not qualify as an ET. Furthermore, ET criteria do not account for best practice guidelines (eg, fluid resuscitation or early vasopressor initiation for sepsis). Thus, automated event identification may require additional screening to understand which events reflect guideline-concordant care.

Developed by Bonafide and colleagues,¹⁵  CDEs involve a unique set of critical interventions within 12 hours after ICU transfer (Fig 1). The cited case example above would qualify handily as a CDE because of intubation and vasopressor initiation within 1 hour after ICU transfer.

Although less commonly used in current practice, the CDE outcome has several advantages. Like ETs, CDEs have been validated as a predictor of patient outcomes (eg, mortality)¹⁵  and can be identified through automated data capture in the EHR, a process that can minimize bias. However, CDEs account for a much broader time horizon. This timing can drive event capture, improvement analyses, and system learning. However, this timing can also invite bias from factors associated with ICU-based care. This is an important distinction, because the primary goal for any outcome is to guide evaluation of ward-based deterioration events. Furthermore, ICU transfers that fail to qualify as CDEs are challenging to interpret; because if patients do not require critical interventions within 12 hours after ICU transfer, did they really require ICU admission at all? This is an important question to consider, because overutilization of ICU resources can have unintended consequences for both patients and health systems, including increased health care costs and potential scarcity of ICU beds.

The outcomes above share a common vulnerability with regard to generalizability: Patient scope. Although there is consensus that admitted patients on ward units should be in scope, the definitions of “ward patient” and “ward unit” lack fidelity. For example, in some hospitals, chronic respiratory failure with tracheostomy and ventilator dependence can be managed on wards if patients are on baseline support settings; however, at other hospitals, these patients require automatic admission to the ICU. Institutional policy differences like this greatly impact the denominator of patients at risk for deterioration on a ward unit. They may also bias events because of population differences in acuity and complexity. Non-ICU ward units caring for specialty patient populations present a similar challenge. For example, in some hospitals, ET criteria include patients transferring to the ICU from a cardiology ward unit; however, other hospitals may consider the cardiology ward out of scope, particularly if those patients are transferring to a distinct cardiac ICU.

Developing effective interventions for harm prevention requires multicenter collaboration. To drive collaborative efforts to prevent unrecognized clinical deterioration outside of the ICU, pediatric safety and quality leaders must first align on a standard outcome. This outcome must have clearly defined clinical and operational criteria, including unit- and patient-level inclusion and exclusion parameters. It must allow for interoperability, with a pragmatic event identification process that is generalizable across settings.

Identification of this outcome must also reflect the input of multiple clinical and operational stakeholder groups, including clinicians who provide patient care outside of ICU environments; critical care clinicians who are familiar with responding to urgent and emergent patient events outside of the ICU (such as through participation in a code blue response team); and safety and quality leaders who can speak to generalizable principles for harm prevention. Finally, this outcome must feel meaningful to patients and families, and it should be validated for what matters most: safe patient outcomes.

Over the last few decades, the pediatric health services literature has seen the publication of a wide diversity of structures, processes, and tools aimed at detecting and managing clinical deterioration outside of the ICU.¹  Commonly referred to as elements of a rapid response system,^1,16  these elements range from semiautomated risk prediction systems in the EHR (eg, BedsidePEWS¹² ) to unit safety huddles aimed at driving shared situation awareness of patients at risk for clinical deterioration (eg, “Watcher” patients).^6,17  These systems also include a variety of ICU care models that have been developed to respond to urgent and emergent events (eg, medical emergency and code blue teams)^9,18  and, more recently, to proactively monitor ward patients with increased risk of deterioration through ICU surveillance.¹⁶  A majority of these rapid response system elements have been investigated (to some degree) for their impact on patient care and outcomes, largely in single-center studies.

Yet, the field of scholarship surrounding pediatric clinical deterioration has overall struggled to produce robust, generalizable evidence as to which structures, processes, and tools are necessary and sufficient for preventing harm from unrecognized clinical deterioration. This is a critical and costly gap in knowledge for both patients and health system leaders, who, in current state, face the challenge of allocating precious hospital resources without sufficient evidence or guidelines to aid in their decision-making. The lack of a standard outcome for evaluating clinical deterioration outside of the ICU is a major barrier to the scholarship needed to address these gaps in knowledge. As such, a standard outcome must first be identified before the right harm prevention bundle can be determined.

Although every source of preventable harm carries a unique risk to patients and their families, it is unfortunately the case that no coordinated safety or quality initiative currently exists targeting prevention of harm from unrecognized clinical deterioration in hospitalized children. Indeed, although ad hoc scholarly efforts have addressed this topic within the health services literature, no safety or quality efforts exist at a national or multicenter level to address this type of preventable harm. Frankly, given the severity of outcomes associated with unrecognized clinical deterioration (ie, significant morbidity and death), this gap is startling. One can only imagine how families would react if they were to understand that, at many children’s hospitals, there are more standardized processes in place to prevent their child from suffering a fall than a cardiac arrest.

It is imperative that pediatric safety and quality leaders apply the same harm prevention principles to deteriorating patients as they do to other sources of harm such as patient falls, particularly given the significant risk of morbidity and mortality that hospitalized patients face from unrecognized clinical deterioration. However, safety and quality leaders must first identify a standard approach to evaluating patient outcomes; for only by standardizing outcomes can scholarship be pursued to readily determine what structures, processes, and tools are effective at preventing this type of harm in a generalizable fashion. In this way, standardizing outcomes can also help determine high-value care strategies, an important factor to consider given the associated high costs of clinical deterioration programming (eg, EHR data capture and display platforms, ICU response teams, etc).

We believe that a stepwise approach is required to comprehensively address and prevent this type of harm (Fig 2). Most immediately, formal, rigorous scholarship should be pursued using consensus methods and with the stakeholder groups listed above to identify a standard outcome, which is the first step needed to advance successful harm prevention efforts. A crucial goal of this immediate phase of work will be to identify a strategic plan for partnering with the right complement of national leaders and subject matter experts in patient safety, harm prevention, and clinical deterioration programming to propose consensus recommendations, and to establish a multicenter collaborative that can drive future evidence-based policy. National collaboratives such as the Children’s Hospitals’ Solutions for Patient Safety network¹⁹  and various resuscitation science networks represent important stakeholder forums for convening and facilitating this type of work. The American Academy of Pediatrics and other professional societies are also important partners that can help guide policy creation and dissemination.

Among other potential safety threats affecting hospitalized children, unrecognized clinical deterioration is a common and significant source of preventable harm. However, unlike other sources of harm, unrecognized clinical deterioration lacks any coordinated safety or quality initiative targeting identification and spread of evidence-based interventions for harm reduction. Identification of a standard outcome for clinical deterioration outside of the ICU is a critical step needed to foster the multicenter collaboration necessary for this important harm prevention work.

ARC

acute respiratory compromise

CDE

critical deterioration event

CHA

Children’s Hospital Association

CPA

cardiopulmonary arrest

CPR

cardiopulmonary resuscitation

EHR

electronic health record

ET

emergency transfer

SCD

significant clinical deterioration