A Standardized Definition of Near-Fatal Child Maltreatment: Results of a Multidisciplinary Delphi Process

Abstract

Background:

The 2016 Presidential Commission to Eliminate Child Abuse and Neglect Fatalities identified systematic review of all cases of near-fatal child maltreatment as a necessary step towards prevention of child maltreatment fatalities. A critical barrier to adoption of this recommendation is the lack of a standard definition of “near-fatality” in the context of suspected child maltreatment.

Objective:

To develop a consensus definition of near-fatal child maltreatment to be used in practice, policy, and research.

Participants and Setting:

A multidisciplinary expert panel of 23 individuals from across the U.S. including child abuse pediatricians, pediatric intensivists, pediatric emergency medicine physicians, child welfare administrators, child welfare researchers, and child injury/fatality researchers.

Methods:

A modified Delphi process reflecting an iterative process of 3 rounds of surveys of expert opinion, statistical summary of survey response, and feedback of summary statistics. Consensus was defined as 75% of panelists ranking an element as required (≥80 on a scale of 0–100) to meet a definition of near-fatality (75th% threshold).

Results:

Experts defined near-fatal child maltreatment as life-threatening cardiopulmonary dysfunction directly attributable to suspected abuse or neglect as evidenced by (a) respiratory insufficiency/failure requiring intubation and mechanical ventilation, (b) respiratory insufficiency/failure requiring medications to reverse effects of toxic ingestion, or (c) cardiac arrhythmia with/without cardiopulmonary resuscitation (CPR).

Conclusions:

A consensus definition of near-fatal child maltreatment should be introduced in child protective services processes and in child fatality/near-fatality reviews to improve our ability to identify, review, and respond to trends in near-fatal child maltreatment at local, regional, and national levels.

Background

Every year, between 1,500 and 3,000 children in the United States die as a result of child maltreatment (Berger, Sanders, Rubin, & The Commission to Eliminate Child Abuse and Neglect Fatalities, 2015; Commission to Eliminate Child Abuse and Neglect Fatalities (CECANF), 2016; U.S. Department of Health and Human Services (USDHHS), 2018). Child fatality review teams across the country are tasked with a comprehensive, multidisciplinary review of the circumstances around these deaths (Fraser, Sidebotham, Frederick, Covington, & Mitchell, 2014; Palusci & Covington, 2014). Prevention of child maltreatment fatalities requires that we identify those children at highest risk for this outcome in order to guide effective, targeted prevention efforts. Prior research has identified important characteristics associated with child maltreatment fatalities. For example, unrelated adult caregivers are often recognized as an important “red flag” for severe and fatal child maltreatment.(Schnitzer & Ewigman, 2005) A referral to child protective services (CPS) for non-fatal maltreatment concerns is associated with a five-fold increase in risk for child maltreatment fatality prior to the fifth birthday (Putnam-Hornstein, Cleves, Licht, & Needell, 2013). High concentrations of community poverty are associated with a three-fold risk of child maltreatment fatalities (Farrell, Fleeger, Monuteaux, Wilson, Christian, & Lee, 2017). These risk indicators, however, cannot be used effectively in child maltreatment fatality prevention because of their high prevalence (e.g., many children experience these conditions) and low specificity (e.g., most children experiencing these conditions will not die from child maltreatment) (Kim, Wildeman, Jonson-Reid, & Drake, 2017; Parker, Horowitz, & Rohal, 2015; Semega, Kollar, Creamer, & Mohanty, 2019).

A 2016 final report by the Presidential Commission to Eliminate Child Abuse and Neglect Fatalities (CECANF) identified the inclusion of all cases of near-fatal child maltreatment in the comprehensive child fatality review process as a necessary step towards prevention of child maltreatment fatalities (Berger et al., 2015; CECANF, 2016). CECANF commissioners pointed out that differences between fatal and near-fatal child maltreatment more often reflect the quality and timing of medical care received rather than the child, family, or household conditions in which the maltreatment occurred. The Commission concluded that reviewing cases of near-fatal child maltreatment alongside fatal child maltreatment can significantly increase our ability to detect patterns and trends in child, family, or household conditions that can guide evidence-based prevention efforts.

A critical barrier to meaningful adoption of this recommendation is the lack of a uniformly accepted definition of near-fatal child maltreatment in practice, policy or research. CECANF emphasized the need for better, more consistent data on fatal and near-fatal child maltreatment across all states to guide prevention efforts. Inconsistent definition of the critical event to be reviewed precludes meaningful measurement of the risk for and incidence of near-fatal child maltreatment over time and place. The Child Abuse Prevention and Treatment Act defines near-fatal child maltreatment as “an act that, as certified by a physician, places the child in serious or critical condition”(Child Abuse Prevention and Treatment Act, as amended by P.L. 111–320, the CAPTA Reauthorization Act of 2010). Unfortunately, terms such as “serious or critical condition” were developed by hospitals to provide information to the public, and have variable meaning to individual medical providers who may be making a referral for a concern for suspected maltreatment. Interpretation of this medical information by child welfare workers is likely to be similarly inconsistent.

In efforts to standardize near-fatality determinations, multiple states have defined near-fatality in state law, statue, or policy. Despite good intentions, these definitions fail to establish a shared national standard. Some states do not specify a level of hospital care required to establish a definition of near-fatality. Other states require admission to an intensive care unit (ICU) , or admission to an ICU and mechanical ventilation. Some states include only cases of near-fatal physical abuse while others also count cases of near-fatal neglect such as failure-to-thrive or opioid ingestion. Without a standard definition of near-fatal child maltreatment, we cannot compare cases over time and place and we cannot use these cases to effectively guide efforts to prevent child maltreatment deaths as envisioned by CECANF.

The multi-disciplinary child fatality and near-fatality review process is well-suited to determining whether these tragic cases are best classified as abusive or not. These determinations take into account social, family, and community context and are historically resistant to categorization by administrative data alone. Our goal, therefore, was to establish a clear, objective standard to determine which cases should be selected for inclusion in these comprehensive reviews. Development of standard data elements and definitions for near-fatality in cases of suspected child maltreatment is necessary to advance practice, policy, and research in child abuse and neglect (Leeb, Paulozzi, Melanson, & Simon, 2008). An ideal definition of near-fatality in cases of suspected child maltreatment would not rely on the subjective opinion of individual medical providers, could be easily abstracted from available data, and would be generalizable across child welfare and health care systems. We used a modified Delphi process with a national, multidisciplinary expert panel to develop a consensus definition of near-fatality in the context of child maltreatment that could be pragmatically used in practice, policy, and research across the United States.

Methods

Design

We used a modified Delphi process to develop a consensus definition for near-fatal child maltreatment. The Delphi method is a consensus process that relies on an iterative process of individual surveys of expert opinion, statistical summaries of survey responses, and group feedback of anonymized summary statistics to achieve convergence of expert opinions around a question for which empirical evidence is lacking (Fink, Kosecoff, Chassin, & Brook, 1984; Jones & Hunter, 1995; Murphy et al., 1998).

In a traditional Delphi process, panelists remain anonymous to the group to reduce the influence of real or perceived hierarchy. We chose to introduce two modifications of Delphi methodology in this project to improve consensus building despite a theoretical risk of removing this anonymity. First, experts were invited to provide free-text responses to all survey questions. Second, investigators planned for a conference telephone call after two survey rounds if low consensus was observed based on anticipation of difficulty with consensus on this topic. Each of these modifications introduced the potential for unmasking of otherwise anonymized expert panelists. This potential was reduced, though not eliminated, by removing references to specific professional duties in circulated comments and by discouraging identification of individual panelists during the conference call.

Experts

A core premise of this project was that a meaningful and practical consensus definition of near-fatal child maltreatment required opinions of experts in pediatrics and child welfare from across the United States representing expertise in practice, policy, and research. With this in mind, we recruited an expert panel of 23 individuals including child abuse pediatricians (n=5), pediatric intensivists (n=4), pediatric emergency medicine physicians (n=4), child welfare administrators (n=4), child welfare researchers (n=3), and child injury/fatality researchers (n=3). Several panelists represented expertise in 2 or more areas. Median time in professional roles for the panelists was 21 years (IQR 8–24 years). Panelists were offered a gift card of $50 for each Delphi round completed. The study was reviewed and approved by the Institutional Review Board of the University of Utah.

Survey development

Initial survey development occurred over three phases. First, the authors drew on our experiences in practice, research, and policy to identify broad themes in cases of suspected child maltreatment including physical abuse, neglect, and drug exposure. Second, one author (KAC) conducted a rapid review of PubMed limited to all English language publications with an abstract associated with the term “near-fatal” in infants and/or preschool aged children to determine whether a commonly accepted definition of pediatric near-fatality could be identified in the medical literature.(Hartling et al., 2008) This review identified 104 eligible publications including 103 case reports, case series, and clinical trials in disease processes such as asthma, cardiac disease, or anaphylaxis and a single case series related to near-fatality in child abuse (Pierce et al., 2017). Each abstract was reviewed to extract clinical characteristics associated with the designation of near-fatality. Full test review was completed for 48 manuscripts for which sufficient detail could not be ascertained from the abstract. No accepted standard definition of pediatric near-fatality was identified in this process. Common themes associated with the designation of near-fatality included intensive care unit (ICU) admission, intubation/mechanical ventilation, cardiopulmonary resuscitation, multiple organ failure, and/or use of medications to sustain or restart vital functions (e.g., epinephrine for anaphylaxis and naloxone for opioid overdose). In the third and final phase, a survey tool informed by these first two steps was drafted, piloted, and revised over several iterations prior to distribution (supplemental file).

Participants were asked to indicate the degree to which each of six characteristics was required for a definition of near-fatality in the context of suspected child maltreatment using a sliding scale (0–100) (“not required” (0), “not sure” (50) or “required” (100)) The six characteristics included physical injury, hospitalization, ICU admission, intubation/mechanical ventilation, cardiopulmonary resuscitation, and use of medications to sustain/restart vital functions. The survey then prompted panelists to review 20 clinical vignettes of suspected child maltreatment and indicate whether each case should be defined as a near-fatality using a similarly anchored sliding scale if CPS investigation supported the medical diagnosis of child maltreatment (“not a near fatality”(0), “not sure” (50) or “near fatality” (100)). Each case provided participants with a “real world” composite of different combinations of the six clinical characteristics listed above.

Survey, Summary, Feedback, and Iteration

The survey was distributed to panelists via Research Electronic Data Capture (REDCap), a secure web platform for building and managing online databases and surveys maintained by the University of Utah Center for Clinical and Translational Science. Summary statistics for each characteristic and vignette were calculated following each survey to describe group opinion. Summary statistics and anonymized comments were distributed to each panelist to provide insight into the range of perspectives related to definitions of near-fatal child maltreatment. Following each round, the survey was revised to exclude elements with high levels of consensus in order to minimize participant burden. One case was removed in Round 3 based on conference call concerns that accurate ascertainment of out-of-hospital emergency response across all cases would be too variable for use in a standard definition of near-fatal child maltreatment. Finally, round 3 surveys allowed panelists first to rate a case as “severe/egregious” and then to rate the same case as “near-fatal.” This addition grew from free-text comments and conference call discussions that highlighted a shared desire to identify cases with potential lethality (e.g. abusive head trauma or multiple fractures), but recognition that not all of these cases should be defined as a near-fatality (e.g., no hospital admission or ICU admission required). After each revision, the survey was redistributed to begin the subsequent cycle of survey, summary and feedback (Figure).

Figure.

Legend: Iterative cycles of survey, summary and feedback used to establish a multidisciplinary expert consensus definition of near-fatality in child maltreatment

Consensus definition development

There is no universally recognized threshold to define consensus within a Delphi process. Based on the study goal of identifying critical elements in the identification of near fatality, we set a conservative a priori definition of consensus, requiring that 75% of panelists rank an element as required (≥80 on a scale of 0–100) to meet a definition of near fatality (75^th% threshold), an approach that has been used by our team and others in prior research (Campbell, Olson, & Keenan, 2015; Murphy et al., 1998). Survey questions specific to clinical characteristics (e.g. intensive care, intubation/mechanical ventilation) were used to create a broad definition of near-fatal child maltreatment. Survey questions describing case vignettes were used to refine this general definition based on specific clinical presentations (e.g. physical abuse, failure to thrive, drug exposure).

Results

All twenty-three (100%) panelists participated in the first 2 Delphi rounds and 22 (95.7%) participated in the final round. Fourteen (60.9%) panelists participated in a 90-minute conference call between the second and third survey.

More than 75% of panelists identified physical injury and hospitalization as required components of near-fatal child maltreatment. ICU admission, intubation/mechanical ventilation, or cardiopulmonary resuscitation were not required by our expert panel. Panel review of case vignettes highlighted the insufficiency of these broad categories and provided better understanding of expert opinions regarding near-fatal child maltreatment under various clinical conditions (Table).

Table 1.

Iterative cycles of survey, summary and feedback used to establish a multidisciplinary expert consensus definition of near-fatality in child maltreatment.

	Case			Characteristics					Ranking
	Case	Age	Case description1	ICU	ETT (any)	ETT (ext)	CPR	MEDS	75th%2	95% CI3
Physical Abuse	1a	6mo	1 FX	no	no	no	no	no	10	1–29
	1b	6mo	3 FX	no	no	no	no	no	29	10–50
	1c	2mo	1 FX + RSV-related ICU care	yes	yes	yes	yes	yes	35	11–50
	2a	6mo	AHT	no	no	no	no	no	52	25–95
	2b	6mo	AHT + ICU	yes	no	no	no	no	64	34–94
	2c	6mo	AHT + ICU + ETT(12hr)	yes	yes	no	no	no	100	99–100
	2d	6mo	AHT + ICU + ETT(72hr)	yes	yes	yes	no	no	100	99–100
	2e	6mo	AHT + ICU + ETT + CPR + MEDS	yes	yes	yes	yes	yes	100	100–100
Drug Exposure	3a	2yo	Opioid O/D	no	no	no	no	no	25	3–39
	3b	2yo	Opioid O/D + Naloxone	no	no	no	no	yes	99	90–100
	3c	2yo	Opioid O/D + Naloxone + ICU + ETT	yes	yes	no	no	yes	99	90–100
	3d	2yo	Opioid O/D + Naloxone + ICU + ETT + CPR	yes	yes	yes	yes	yes	100	99–100
	3e	2wko	Opioid O/D	no	no	no	no	no	33	5–36
	4a	birth	Fetal exposure + prematurity + NOWS	yes	yes	yes	yes	yes	50	25–99
Neglect	5a	12mo	FTT	no	no	no	no	no	40	12–58
	5b	12mo	FTT + refeeding syndrome	no	no	no	no	no	79	62–100
	5c	12mo	FTT + refeeding syndrome + ICU (arrhythmia)	yes	no	no	no	no	98	85–100
	6a	5yo	PNA + ICU + ETT + CPR + MEDS	yes	yes	yes	yes	yes	100	87–100

Abbreviations:

Abusive head trauma (AHT), Intensive care (ICU), Intubation and mechanical ventilation (ETT), for any period of time (ANY) or for extended (≥72 hours) period (EXT), Cardiopulmonary resuscitation (CPR), Medications to restart or sustain cardiac or respiratory function (MEDS), Opioid overdose (Opioid O/D); Neonatal opioid withdrawal syndrome (NOWS), Fracture (FX), Failure to Thrive (FTT).

^1.

Cases:

1. Infant with fracture(s) concerning for physical abuse admitted to general pediatrics ward. In case 1c, the infant is admitted to intensive care based on life-threatening bronchiolitis; a fracture concerning for abuse is an incidental finding identified during admission and unrelated to critical illness.
2. Infant with subdural hemorrhage and associated findings concerning for abusive head trauma (AHT) receives escalating levels of care based on injury severity.
3. Toddler (3a–3c) or non-mobile infant (3d) presenting with opioid toxidrome receives escalating levels of care based on severity of overdose.
4. Neonate with extreme prematurity and opioid withdrawal syndrome (NOWS) attributed to untreated maternal substance use disorder receives extensive care in the neonatal intensive care unit.
5. Infant with failure to thrive attributed to nutritional neglect receives escalating care based on severity of malnutrition and refeeding syndrome.
6. Child with complex chronic medical needs develops aspirations pneumonia (PNA) and experiences life-threatening complications attributed to parental refusal to seek care.

^2.75^tth percentile of expert panel rankings for the given scenario on a scale of 0 (not a near-fatality) to 100 (definitely a near-fatality). Shaded cells meet the a priori definition of near-fatality as a 75^th percentile expert ranking >80.

^3.95% confidence interval around 75^th percentile ranking. Shaded cells reflect scenarios not meeting threshold for near-fatality, but with wide variability in expert opinion suggesting a lack of consensus around these cases.

Child physical abuse with isolated fracture(s) (e.g., skeletal injury in the absence of additional findings) did not meet the 75^th% consensus threshold for near-fatal child maltreatment in any case. Child physical abuse with abusive head trauma (AHT) met this threshold if the child required both (a) ICU admission and (b) intubation/mechanical ventilation for any period of time. In cases of drug exposure, opioid overdose (accidental or intentional) met consensus threshold for near-fatality when naloxone was required to restart respiratory efforts, even when ICU admission or intubation/mechanical ventilation was not required. Fetal exposure to opioids associated with neonatal opioid withdrawal syndrome and complications of extreme prematurity did not meet consensus thresholds. Failure to thrive due to nutritional neglect did meet the threshold for near-fatal child maltreatment when ICU admission for monitoring of cardiac arrhythmias associated with refeeding syndrome without the requirement for intubation and mechanical ventilation. Finally, expert consensus defined medical neglect as near-fatal child maltreatment when it resulted in ICU admission and cardiopulmonary failure, even in a setting where underlying complex chronic conditions contributed to a child’s vulnerability to near-fatal complications.

Several case vignettes highlighted areas where consensus was limited as evidenced by broad confidence intervals around the 75^th%. All cases of AHT, regardless of severity of clinical presentation, were ranked by at least two panelists as near-fatal (cases 2a–2c). In comments, panelists noted that AHT was the “most common cause of fatal abuse in children” and that “any abusive head trauma could result in fatality.” Similarly, every case involving ICU admission for care of a condition attributable to maltreatment was ranked as a near-fatal event by at least two panelists regardless of need for intubation and mechanical ventilation (cases 2b–2e, 3c–3d, 4a 5c, and 6a). In this scenario, however, dissenting panelists argued that “using ICU admission as criteria [to define near-fatality] is fraught with problems due to inconsistent PICU practices across hospitals” Failure to thrive and refeeding syndrome attributed to nutritional neglect (Cases 5b–c) was considered near-fatal by many but achieved the 75^th% consensus threshold for near-fatality only in the context of a concern for cardiac dysfunction and ICU admission. Finally, fetal exposure to opioids associated with prematurity and poor neonatal outcomes requiring extensive ICU interventions did not meet the consensus threshold for near-fatality but had wide variability in panelist response. Panelists debated the direct association between prenatal exposures and neonatal outcomes from a medical perspective, noting that “prematurity, and not the opioids, are the cause for her need for intubation and mechanical ventilation” and “there is [not] enough evidence, in this case or in the extant literature, to attribute this child’s preterm birth to maternal opioid use.” Other panelists identified wide variability in child welfare response to fetal exposure, remarking that “some states do not consider substance exposure to be maltreatment.”

Discussion

Based on strict expert consensus, this study defines near-fatal child maltreatment as life-threatening cardiopulmonary dysfunction directly attributable to conditions resulting from suspected abuse or neglect as evidenced by (a) respiratory insufficiency/failure requiring intubation and mechanical ventilation, (b) respiratory insufficiency/failure requiring medications to reverse effects of toxic ingestion, or (c) cardiac arrhythmia with or without cardiopulmonary resuscitation (CPR). While this definition of near-fatal child maltreatment is not strictly limited to cases of physical abuse, there does have to be evidence of critical physical injury—or physical harm—attributable to the maltreatment. It is important to note that our panelists expressed high levels of concern for all cases of suspected AHT even if they did not meet near-fatal criteria due to the potential lethality of such cases and the need for targeted AHT prevention efforts.

This definition of near-fatality does not include newborns with neonatal ICU admission for conditions associated with fetal exposure to drugs and/or alcohol due to lack of medical consensus regarding the association between fetal exposure and neonatal outcomes, as well as lack of child welfare consensus related to regulatory oversight of these cases. Furthermore, while ICU admission is frequently used as an indicator of near-fatal events, variability in institutional practices around ICU admission threatens the ability to compare rates of near-fatal child maltreatment across locations and over time.

Future research on the use of this definition is needed. Validity across medical, child welfare, and child death review systems should be investigated to assure that the definition has acceptable specificity and sensitivity for cases interpreted as “near-fatal” by key stakeholders. Implementation in practice, with checklists that can be used to communicate concern for near-fatalities across sectors, should be developed and tested. Research measures, which can capture near-fatalities within administrative claims data using diagnostic, procedural, and pharmacy codes, should be developed and validated with medical record review. Finally, the results of near-fatality reviews using these measures should be monitored at local, state, and federal levels to assure that the goal of prevention of severe and fatal child maltreatment is being supported.

Our findings must be considered in the context of certain limitations. These conclusions reflect consensus opinions of 23 experts, and as such do not reflect the full range of opinions of the many professionals who care for these at-risk populations. Prior research has identified a panel of 23 as sufficient to achieve a stable consensus, although this has not been studied in panels representing such a wide diversity of experience and opinion (Akins, Tolson, & Cole, 2005). Given the difficult nature of defining “near-fatality,” it is perhaps not surprising that we identified several areas of limited consensus. This was addressed with the addition of a conference call to the Delphi panel process as well as review of comments provided by the panelists. Additionally, our findings are limited in anticipating all potential cases of near-fatal child maltreatment. A brief survey cannot tease apart the complexity of care required by every seriously ill child. Our results may not capture fine distinctions in expert opinion between brief episodes of respiratory insufficiency or cardiac dysfunction and sustained periods of life-threatening cardiorespiratory failure. Ultimately, however, areas of limited consensus or potential misclassification required decision-making by the authors in arriving at a definition for near-fatal child maltreatment. This definition has been reviewed by our expert panelists with feedback incorporated into this final report.

Conclusion

Despite limitations, we believe that this consensus definition of near-fatal child maltreatment can be executed as a standard checklist process during CPS intake procedures, with updates at case closure should clinical conditions change between initial reporting and case closure. We recommend that states adopt this standard definition of near-fatal child maltreatment in child fatality and near-fatality review processes. Having such procedures in place will allow local, regional, and national comparison of near-fatal child maltreatment, detect emerging trends in near-fatal child maltreatment, and identify common risk indicators in these cases. Such information can be used to guide and evaluate efforts to prevent the most severe cases of child maltreatment at a population level.

What is known

Systematic review of near-fatal child maltreatment is critical to child maltreatment fatality prevention. Existing guidelines allow medical providers to identify near-fatalities on a case-by-case basis. A standard definition of near-fatal child maltreatment is needed to guide practice, policy and research.

What this study adds

Based on multidisciplinary expert panel consensus, we propose a standard definition of near-fatal child maltreatment that can be used to improve our ability to identify, review, and respond to trends in near-fatal child maltreatment at local, regional, and national levels.